Nederlandse Ziekenhuisfarmaciedagen, 9 en 10 november 2017

Prediction of clinically relevant adverse drug events in surgical patients

J.M. Bos ^a*, G.A. Kalkman ^a, H. Groenewoud ^b, P.M.L.A. van den Bemt ^c, P.A.G.M. de Smet ^b, J.E. Nagtegaal ^d, A. Wieringa ^e, G.J. van der Wilt ^b and C. Kramers ^f

^a Canisius Wilhelmina Ziekenhuis, Nijmegen.

^b RadboudUMC, Nijmegen.

^c ErasmusMC, Nijmegen.

^d Meander MC, Amersfoort.

^e Isala klinieken, Zwolle.

^f Radboudumc, Nijmegen.

* Correspondence: jm.bos@cwz.nl.

Background

Risk stratification of hospital patients for adverse drug events (ADE) would enable targeting patients who may benefit from interventions aimed at reducing drug-related morbidity. It can support clinicians and hospital pharmacists in selecting patients to deliver a more efficient health care service. This study aimed to develop and test a prediction model that helps to identify patients at the day of hospital admission, who are at increased risk of developing a clinically relevant, preventable adverse drug event during their stay on a surgical ward.

Methods

Data of the P-REVIEW study were used. This study was designed to assess the impact of a multifaceted educational intervention on clinically relevant, preventable adverse drug events in surgical patients [1]. Thirty-nine variables were evaluated in a univariate and multivariate logistic regression analysis, respectively. Model performance was expressed in the Area Under the Receiver Operating Characteristics (AUROC). Bootstrapping was used for model validation.

Results

6780 admissions of patients at surgical wards were included during the pre-intervention period of the PREVIEW trial. 102 Patients experienced a clinically relevant, adverse drug event during their hospital stay that was deemed potentially preventable. The prediction model comprised five variables: age, number of biochemical tests ordered, heparin/LMWH in therapeutic dose, use of opioids and use of cardiovascular drugs. The AUROC was 0.86 (95% CI 0.83-0.88). At a cut-off point for an increased risk of developing an ADE of 1.6%, the model had a sensitivity of 80.4% and a specificity of 73.4%. The positive and negative predictive value were 4.5% and 99.6% , respectively. The bootstrap procedure did not significantly affect model parameters.

Conclusion

The combined use of a limited set of easily ascertainable patient characteristics can help physicians and pharmacists to identify, at the time of admission, surgical patients who are at increased risk of developing ADEs during their hospital stay. This may serve as a basis to take extra precautions to safeguard medication safety in those patients.

Literatuur

1. Bos JM, van den Bemt PM, Kievit W, Pot JL, Nagtegaal JE, Wieringa A, et al. A multifaceted intervention to reduce drug-related complications in surgical patients. British Journal of Clinical Pharmacology. 2017;83(3):664-77.

 

A 25% higher vancomycin maintenance dose is required in neutropenic hematologic patients

D. Bury ^a*, R. ter Heine ^b, E.M.W. van de Garde ^c, R.J. Grouls ^a and M.J. Deenen ^a

^a Catharina Ziekenhuis Eindhoven, Eindhoven.

^b RadboudUMC Nijmegen, Nijmegen.

^c St. Antonius Ziekenhuis Nieuwegein, Nieuwegein.

* Correspondence: d.bury@students.uu.nl.

Background

Neutropenic hematology patients with severe sepsis and neutropenic patients with a predisposition for penicillin resistance are indicated for treatment with a glycopeptide antibiotic like vancomycin. There is accumulating evidence that this specific group of patients requires higher dosages vancomycin and that the increased required dose is likely associated with the presence of neutropenia. To prevent longlasting sub-therapeutic exposure and inefficacy of treatment, it is of utmost importance to obtain adequate exposure from the first dose onwards, with as little as dose titration necessary. We, therefore, aimed to quantify the effect of neutropenia on the pharmacokinetics of vancomycin.

Methods

We selected patients within the period from 2010-2016 that were treated with vancomycin for at least 2 days and from whom at least 1 vancomycin concentration was available. To be able to compare pharmacokinetics of vancomycin between various patient populations, patients were divided in three matched cohorts, i.e. 1) patients with a hematologic malignancy, 2) patients with a solid tumor, and 3) patients not known with cancer. Data, including general patient characteristics, vancomycin dose, and peak and trough plasma concentrations were retrospectively collected by analysis of the Electronic Healthcare Record System (EHRS) and by the use of the Santeon Farmadatabase (SFD). Pharmacokinetic analysis was performed by means of non-linear mixed effects modelling, with the software package NONMEM V7.3. Presence of neutropenia was investigated as a binary covariate on clearance and volume of distribution of vancomycin.

Results

A total of 116 patients were included (39 hematologic patients, 39 solid tumor patients and 38 pts not known with cancer). In total, 742 paired time-concentration observations were available for the pharmacokinetic analysis. When investigating the presence of neutropenia as a covariate for clearance of vancomycin, we found that neutropenia relevantly and significantly (P = 0.018) increased the clearance of vancomycin in neutropenic patients by 23% (95% confidence interval 3%-48%). The presence of neutropenia did not significantly impact volume of distribution (P = 0.20).

Conclusion

The clearance of vancomycin is increased in patients with neutropenia by 23%. For rapid pharmacokinetic target attainment, the maintenance dose of vancomycin should be increased by 25% in hematologic patients. Since the volume of distribution seems unaffected, no adjustment in loading dose is indicated. These dose adjustments do not rule out the necessity of further dose individualization by means of therapeutic drug monitoring.

 

Intrathecal morphine/bupivacaine for laparoscopic segmental colonic resection. A randomized controlled trial

M.V. Koning ^a*, E.J. Ruijgrok ^b, A.J.W. Teunissen ^b, E. van der Harst ^b and R.J. Stolker ^c

^a Erasmus MC, Rotterdam.

^b Maasstad Ziekenhuis, Rotterdam.

^c ErasmusMC, Rotterdam.

* Correspondence: markkoning66@hotmail.com.

Background

Pain after laparoscopic surgery is intense, but relatively short-lived when compared to open surgery. Analgesia should be tailored accordingly. Post-operative pain management after laparoscopic colonic resections remains controversial. Enhanced Recovery After Surgery (ERAS) program-guidelines recommend to limit opioid use by administration of multimodal analgesics, which includes regional anesthesia techniques and Patient Controlled Intravenous Analgesia (PCIA). In this randomized controlled trial, we compared the intrathecal administration of an admixture of morphine/bupivacaine with the standard analgetic procedure within an ERAS-program. The goal of the study was to investigate whether intrathecal morphine/bupivacaine would lead to an enhanced recovery and limited need of systemic opioids after laparoscopic colonic surgery as compared to the standard procedure.

Methods

A single-center, double-blinded randomized controlled trial was performed. Patients scheduled for laparoscopic segmental intestinal resections were considered. Exclusion criteria were patients in whom contra-indications were present for spinal anesthesia, conversion to open surgery and gastric and rectal surgery. A sterilised intrathecal admixture of morphine/bupivacaine was formulated and produced by the hospital pharmacy. The intervention group received a single shot intrathecal morphine/bupivacaine (12.5 mg/300 mcg), just prior to the laparoscopic surgery. The control group received a sham procedure and a bolus of piritramide (0.1 mg/kg). Both groups received standardized general anesthesia and a PCIA-pump as postoperative analgesia. A decrease in days to “fit-for-discharge” was the primary outcome parameter.

Results

56 patients were enrolled. Patients in the intervention group had an earlier “fit-for-discharge” rate (median of 3 versus 4 days, P = 0.044). Furthermore, there was a significant decrease in opioid-use and lower pain scores on the first postoperative day in the intervention group. Besides a higher incidence of pruritus in the intervention group (41% versus 8%), there were no differences in adverse events, time-to-mobilisation, fluid administration or patient satisfaction.

Conclusion

This RCT shows that intrathecal morphine is a more effective method of postoperative analgesia in laparoscopic surgery than intravenous opioids. These results have led to the admission of the morphine/bupivacaine intrathecal product into the hospital formulary. Studies to investigate the use of intrathecal morphine/bupivacaine in other indications are currently performed.

 

The effect of an improved collaboration between secondary and primary care on drug-related problems post-discharge

S. Daliri ^a*, J.G. Hugtenburg ^b, B.J.F. van den Bemt ^c, M.C. van Buul-Gast ^d and F. Karapinar-Carkit ^e

^a OLVG en AMC, Amsterdam.

^b VU medisch centrum, Amsterdam.

^c Sint Maartenskliniek, Nijmegen.

^d BovenIJ ziekenhuis, Amsterdam.

^e OLVG ziekenhuis, Amsterdam.

* Correspondence: s.daliri@olvg.nl.

Background

A hospital discharge is a critical period with respect to patient safety. Lack of communication, inadequate transfer of information between secondary and primary care and the patient can result in drug-related problems (DRPs). The objective of this study was to investigate the effect of an improved collaboration between secondary and primary care on the amount of DRPs post-discharge.

Methods

A prospective controlled multicenter study was conducted in two hospitals (OLVG and BovenIJ) and fifty community pharmacies in Amsterdam. Patients discharged from the internal medicine-, neurology-, cardiology- and pulmonology departments were included if at least one medication change was initiated during admission. Usual care patients received routine care: medication reconciliation at hospital admission and discharge by pharmacy technicians and pharmaceutical consultants (specialized pharmacy technicians). Intervention patients additionally received teach back at hospital discharge to assess whether the patient could specify which medication had changed during admission. Also, primary care providers (i.e. community pharmacy, general practitioner, home healthcare nurses) received a medication overview listing (reasons for) all in-hospital medication changes. Finally, the patient’s community pharmacist performed a home visit ≤5 days post-discharge.

Four weeks post-discharge DRPs (adverse drug events, practical problems, doubts and concerns on the effectiveness and safety of medication) were assessed during a structured telephone interview (primary outcome). Also, knowledge regarding all medication changes implemented in the hospital and satisfaction with medication use were measured. Data were analysed by means of descriptive statistics. The independent T test was used for continuous variables and the Chi square test for frequencies. Ordinal logistic regression analyses were performed adjusting for confounders. P-values and adjusted odds ratios (ORs) with 95% confidence intervals (CIs) were calculated.

Results

In total 456 patients were included (usual care: n = 234, intervention: n = 222). DRPs decreased significantly from 66% in the usual care group to 52% in the intervention group (P < 0.01, adjusted OR 0.57 (95% CI 0.37-0.89)), primarily due to decreased post-discharge adverse drug events (25% versus 16%; P < 0.05).

Furthermore knowledge regarding all medication changes implemented in the hospital improved significantly in the intervention group (30% versus 42%; P < 0.05). In total, 82% of the patients in the intervention group reported that they were satisfied with their medication compared to 68% in the usual care group (P < 0.01).

Conclusion

This study highlights the importance of an improved collaboration between primary and secondary healthcare providers to ensure continuity of care and to reduce patient harm due to medication. DRPs were significantly reduced.

 

Effect of CYP3A4*22, CYP3A5*3 and CYP3A combined genotypes on tamoxifen metabolism

A.B. Sanchez Spitman ^a*, D.J.A.R. Moes ^a, A.J. Gelderblom ^a, V.O. Dezentjé ^b, J.J. Swen ^a and H.J. Guchelaar ^a

 

^a LUMC, Leiden.

^b Reinier de Graaf, Delft.

* Correspondence: a.b.sanchez_spitman@lumc.nl.

Background

Tamoxifen is the cornerstone of endocrine therapy for breast cancer. Recently, the decreased activity CYP3A4*22 allele and the loss of function CYP3A5*3 allele, have been described as potential factors that could help to explain the inter-patient variability in tamoxifen metabolism. The aim of this study is to analyze the impact of CYP3A4*22, CYP3A5*3 and CYP3A combined genotypes in addition to CYP2D6 genotype on tamoxifen metabolism.

Methods

DNA from 667 women participating in the CYPTAM study (NTR1509) was genotyped for CYP2D6, CYP3A4*22 and CYP3A5*3. Tamoxifen and metabolite concentrations were measured in serum, and metabolic ratios were calculated. The impact of CYP3A4*22, CYP3A5*3 and CYP3A combined genotypes was examined by multiple linear regression analysis. A two-sided Student’s t-test was used for comparisons between mean concentration levels and metabolic ratios.

Results

Among CYP3A4*22 carriers, a tendency towards higher endoxifen levels was observed (P = 0.088), and significantly higher concentration levels of tamoxifen, N-desmethyl-tamoxifen and 4-hydroxy-tamoxifen were found. The Metabolic ratio tamoxifen/N-desmethyl-tamoxifen was significantly higher in CYP3A4*22 individuals (0.59 versus 0.52, P < 0.001). At the same time, CYP3A4*22 genotype contributed to improving the explained inter-variability between patients (R2 of the (log-transformed) metabolic ratio tamoxifen/N-desmethyl-tamoxifen improved from 21.8 % to 23.9 %, P < 0.001).

CYP3A5*3 carriers did not show any significant difference in tamoxifen and its metabolites mean concentration levels. However, CYP3A5*3 genotype marginally improved the explained variability of the (log transformed) metabolic ratio 4-hydroxy-tamoxifen/endoxifen (from 44.9 % to 46.2 %, P < 0.038). In line with these results, CYP3A combined genotypes did not significantly contribute to improving the explained variability between patients.

Conclusion

Our data demonstrate that CYP3A genotype slightly contributes to explaining the variability between patients in tamoxifen metabolism. However, the effect is small, and therefore, it is unlikely to have any significant clinically relevance for the efficacy of tamoxifen.

 

Highly variable absorption of clavulanic acid during the day: a population pharmacokinetic analysis

F. de Velde, B.C.M. de Winter *, B.C.P. Koch, T. van Gelder and J.W. Mouton

Erasmus MC, Rotterdam.

* Correspondence: f.develde@erasmusmc.nl.

Background

Clavulanic acid is a beta-lactamase inhibitor that is combined with beta-lactam antibiotics such as amoxicillin to target beta-lactamase-producing strains. Surprisingly, little information is available with respect to its pharmacokinetic profile and interindividual exposure. A population pharmacokinetic model for oral clavulanic acid is currently not available in the literature. The objectives of this study were to estimate the clavulanic acid exposure of oral amoxicillin/clavulanic acid dosing regimens, to investigate variability using a population pharmacokinetic model and to explore target attainment using Monte Carlo simulations for both fAUC and %fT>Ct (% of dosing interval that the unbound concentration exceeds a threshold concentration), because the PD target is still unknown.

Methods

Two groups of healthy male volunteers received amoxicillin/clavulanic acid tablets at the start of a standard meal on two separate days one week apart. One group (n = 14) received 875/125 mg every12 hours and 500/125 mg every 8 hours and the other group (n = 15) 500/125 mg every12 hours and 250/125 mg every 8 hours. 1479 blood samples were collected until 8-12 hours after administration. Concentrations were analysed using non-compartmental (WinNonLin 7.0) and population pharmacokinetic methods (NONMEM 7.2: FOCE+I). Five thousand Monte Carlo simulations for 125 mg every 8 hours and 125 mg every12 hours were performed using the final model in NONMEM.

Results

The non-compartmental analysis showed a median C_max of 2.21 mg/L (0.21-4.35) and an AUC_0-8 of 4.99 mg*h/L (0.44-8.31). In 40/58 daily concentration-time profiles, C_max and AUC_0-8 of the morning dose were higher than later doses. The final population model included a lag time (0.447 h), first-order absorption (3.99/h at 8:00 hours, between-subject variability 52.8%, between-occasion variability 48.5%), one distribution compartment (33.0 L, BSV 23.9%) and first-order elimination (24.6 L/h, BSV 26.7%). Bioavailability (fixed at 1 at 8:00 hours, BOV 28.2%) and absorption rate decreased over the day. For 97.5% of the simulated population after 125 mg every 12 hours or every 8 hours, %fT > Ct at 0.5 mg/L was 8.33% (every 12 hours) and 15.2% (every 8 hours), at 1 mg/L 0% (ev ery 12 hours + every 8 hours) and fAUC_0-24 3.61 (every 12 hours) and 5.56 (every 8 hours) mg*h/L.

Conclusion

Clavulanic acid absorption in healthy volunteers is highly variable. Bioavailability and absorption rate decrease over the day. The consequences of the variable concentrations for under-dosing and adverse events should be further studied for multiple day dosing and dosing regimens should be optimized. Studies to determine the PK/PD index and PD target are urgently needed.

 

Pharmacokinetics and pharmacodynamics of infliximab in severe sarcoidosis

H.A. Crommelin ^a*, V.H.M. Deneer ^a, A.D.M. Vorselaars ^b, C.H.M. van Moorsel ^b, J.H. Proost ^c and J.C. Grutters ^b

^a UMCU, Utrecht.

^b St Antonius Ziekenhuis, Nieuwegein.

^c Rijksuniversiteit Groningen.

* Correspondence: h.a.crommelin@umcutrecht.nl.

Background

Sarcoidosis is a multi-organ disease characterised by inflammation and non-caseating granulomas. In severe sarcoidosis, infliximab is an effective off-label third-line therapeutic. The aim of this study was to describe pharmacokinetics and concentration-response relationship of infliximab in severe sarcoidosis.

Methods

Sarcoidosis patients treated with infliximab (n = 68) 5 mg/kg at week 0, 2 and every four weeks thereafter, were followed-up for up to two years. Serum samples were collected at every dose just before infusion and one hour after infusion during the first six months and thereafter every three months. Response on clinical and inflammatory parameters and a composite response score were determined after the first six months of treatment. A pharmacokinetic model was developed using NONMEM.

Results

The median infliximab serum trough level was 17.9 mg/L. A two-compartment, population pharmacokinetic model best described the serum infliximab concentration-time data. Population estimates (typical value (relative standard error)) in the final covariate model were clearance (CL) 0.276 L/day (3.2%), volume of central compartment (V1) 3.16 L (1.6%), intercompartmental clearance (Q) 0.177 L/day (21%) and volume of peripheral compartment (V2) 1.49 L (11%). Interindividual variability for CL, V1 and V2 were 23.5%, 10.9% and 75.4%, respectively. Covariate analyses showed that V1 increased with baseline body surface area and CL decreased with positive antibodies toward infliximab status, high baseline serum albumin and low body weight. No association with inflammatory activity, genotype, ethnicity or treatment phase was found. No association between infliximab area-under-the-curve (AUC) and response on clinical and inflammatory parameters or composite response score was found.

Conclusion

Baseline body surface area, body weight, serum albumin and antibodies toward infliximab status were found to influence infliximab pharmacokinetics in severe sarcoidosis, in congruence with other immune-mediated diseases. Concentrations were high. No concentration-effect relationship was found. These findings could be used to further optimize infliximab therapy in severe sarcoidosis.

 

Phenobarbital and lidocaine pharmacokinetics during therapeutic hypothermia after perinatal asphyxia: evaluation of existing models 

L.M.A. Favié ^ab*, A.D.R. Huitema ^ac, C.M.A. Rademaker ^a, T.R. de Haan ^d, Y.A. Bijleveld ^e, F.Groenendaal ^bf and the PharmaCool Study Group

^a University Medical Center Utrecht.

^b Wilhelmina Children's Hospital, University Medical Center Utrecht.

^c Netherlands Cancer Institute, Amsterdam.

^d Emma Children’s Hospital, Academic Medical Center, Amsterdam.

^e Academic Medical Center, Amsterdam.

^f Brain Center Rudolf Magnus, University Medical Center Utrecht.

* Correspondence: lfavie@umcutrecht.nl.

Background

Pharmacokinetics of drugs may be affected by therapeutic hypothermia in full term neonates with perinatal asphyxia. Little information is available on the pharmacokinetics of drugs frequently administered during therapeutic hypothermia and evidence based dosing guidelines are needed.

Objective

In the prospective, nationwide PharmaCool study (NTR2529) pharmacokinetics of the anti-epileptic drugs phenobarbital and lidocaine were studied. Aim of this study was to evaluate pharmacokinetics under therapeutic hypothermia as published previously (phenobarbital: Clin Pharmacokinet 2012 51:671-679; lidocaine: Arch Dis Child Fetal Neonatal Ed 2013; 98 F341-F345).

Methods

Term neonates treated with therapeutic hypothermia following perinatal asphyxia were eligible for inclusion. Phenobarbital was administered as a first line drug in single or repeated injections. Lidocaine, a second line drug, was administered as continous infusion according to local or national protocols. A maximum of four plasma samples were obtained during the first 120 hours after birth during both hypothermia and normothermia. Plasma concentrations were analyzed using a validated LC-MS/MS method. The lidocaine metabolite monoethylglycylxylidine (MEGX) was also measured. Pharmacokinetic analyses were performed using NONMEM (version 7.3).

Results

Phenobarbital (n=113): plasma levels ranged from 3.10 to 52.6 mg/L (figure 1). In 49 patients only a loading dose was given, 64 patients required additional phenobarbital dosing. Population parameters scaled to 3.5 kg were 10.40 ml/h for clearance and 3,610 ml for volume of distribution. No significant effect of therapeutic hypothermia on clearance of phenobarbital could be identified. Lidocaine (n=20): lidocaine plasma levels ranged from 0.1 to 5.5 mg/L (figure 2). Population parameters for lidocaine scaled to 3.0 kg were 1.96 L/h for clearance (Cl) and 13.2 L for volume of distribution (Vd). Clearance of lidocaine and MEGX was reduced under therapeutic hypothermia by 23% and 19%, but this effect was not statistically significant.

Conclusion

This data confirms the previous findings. Hypothermia has no influence on phenobarbital pharmacokinetics in this patient population. Dosing adjustment of phenobarbital during therapeutic hypothermia is not necessary. Although the effect of therapeutic hypothermia on the pharmacokinetics of lidocaine and MEGX was not statistically significant in this study, it is in line with the previously identified magnitude of the effect. This project was funded by the Netherlands Organization for Health Research and Development ZonMw Priority Medicines for Children (40-41500-98-9002).

Figure 1. observed phenobarbital plasma concentrations (single dose only). Dotted lines indicate the therapeutic window (10-40 mg/L).

Figure 2. Observed lidocaine plasma levels. Plasma concentrations above the dotted line (9 mg/L) are considered toxic and should be avoided.

 

New population pharmacokinetic model that predicts the individual starting dose of tacrolimus following pediatric renal transplantation

L.M. Andrews ^a*, D.A. Hesselink ^a, T. Van Gelder ^a, E.A.M. Cornelissen ^b, S.N. De Wildt ^b, B.C.P. Koch ^a, K. Cransberg ^a and B.C.M. De Winter ^a

^a Erasmus MC, Rotterdam.

^b Radboudumc, Nijmegen.

* Correspondence: l.andrews@erasmusmc.nl.

Background

Multiple clinical, demographic and genetic factors affect the pharmacokinetics (PK) of tacrolimus in children, yet in daily practice the starting dose is based solely on bodyweight. TDM limits the time a patient is exposed to concentrations outside the target range, but it can take two weeks to reach the target tacrolimus concentration. The aim of this study was to improve the starting dose of tacrolimus after pediatric renal transplantation.

Methods

Clinical, demographic, PK and genetic data were collected for the first six weeks after renal transplantation. All children were treated with basiliximab, tacrolimus, mycophenolic acid and glucocorticoids. Every child had at least one tacrolimus PK profile performed over 4h. A population PK analysis was conducted using NONMEM. Demographic, clinical and genetic parameters were evaluated as covariates for all PK parameters containing interpatient variability. The final model was internally and externally validated using visual predictive checks (VPC). Simulations were performed to determine the ideal starting dose.

Results

46 children with a median age of 9.1 years (range 2.4-17.9) were included. Population PK was best described by a two-compartment model with allometric scaling for bodyweight. Clearance (50.5 L/h) increased in CYP3A5 expressers, patients with an increase in eGFR, decrease in hematocrit and recipients of a kidney from a deceased donor. Together these covariates explained 41% of the variability in CL. The model was externally validated using an independent dataset of 23 patients. From the significant covariates, CYP3A5, bodyweight and donor type were useful to adjust the starting dose to reach the target predose level. For each combination of these covariates a new starting dose was calculated to reach a target level of 12.5 ug/L.

Conclusion

During the first 6 weeks after transplantation, the tacrolimus weight-normalized starting dose should be higher in patients with a lower bodyweight, who express CYP3A5 and those who receive a kidney from a deceased donor. Using these parameters an individualized guideline for the initial dosage was developed.

 

High interindividual variability in urinary fosfomycin concentrations in healthy female volunteers

R.A. Wijma *, B.C.P. Koch, T. van Gelder and J.W. Mouton

Erasmus MC, Rotterdam

* Correspondence: r.wijma@erasmusmc.nl.

Background

Fosfomycin is increasingly being prescribed for the treatment of uncomplicated urinary tract infections in an era of emerging drug resistance. Surprisingly, little is known of the urinary concentrations of fosfomycin and its interindividual variation after the standard single 3 gram oral dose. We aimed to gain more insight into urinary fosfomycin pharmacokinetics to evaluate its effectiveness.

Methods

Three grams of fosfomycin trometamol was administered to 40 healthy female volunteers with an estimated mean glomerular filtration rate of > 90 mL/min/1.73m². Urine samples were collected from every urination during 48 hours, and then twice daily for up to 7 days. Time, volume and pH were recorded. Urinary fosfomycin concentrations were quantified with UPLC-MS/MS. Effectiveness was evaluated based on urinary concentrations and the target MIC of E. coli, the most common uropathogen.

Result

A high interindividual variability was found. Peak concentration was 1982.0 ± 1257.4 mg/L, urinary half-life 12.4 ± 5.7 hours and excretion rate over 48 hours 29.9 ± 7.1 mg/h. Recovery was 44.5 ± 12.6% after 48 h and 47.0 ± 10.4% after 7 days and 95% of this total amount was excreted within 42 hours of which the main part was excreted during the first 6 hours. Concentrations remained above the EUCAST breakpoint of 32 mg/L in 100% of the volunteers over the first 24 h, 67.5% for 48 h and 30% for 72 h. A high urinary output was associated with low urinary concentrations and consequently reduced time > MIC, AUC_0-7days/MIC and C_max/MIC values.

Conclusion

Considerable interindividual variability observed in the pharmacokinetics of fosfomycin signifies a risk for inadequate drug exposure in a significant proportion of the population. The current dosing regimen should therefore be reevaluated.

 

Pharmacokinetics and safety of intravenous Sildenafil in newborns with congenital diaphragmatic hernia

B.C.P. Koch ^a*, S. Cochius ^a, F. Kipfmueller ^b, D. Tibboel ^a, B. de Winter ^a and K. Allegaert ^a

^a Erasmus MC, Rotterdam.

^b University of Bonn, Bonn.

* Correspondence: b.koch@erasmusmc.nl.

Background

Sildenafil is often used in patients with congenital diaphragmatic who suffer from severe pulmonary hypertension attributable to altered development of the pulmonary vasculature. Hypertension is often resistant to standard vasodilator therapy. Sildenadil can be used although dosing, concentration and effects are not well researched for this indication. We developed a pharmacokinetic model of intravenous sildenafil and its metabolite and we evaluate the safety of sildenafil in patients with congenital diaphragmatic hernia (CDH).

Methods

A retrospective study was conducted in 2 centres in Germany and The Netherlands between November 2013 and September 2015. 23 patients were treated with a loading dose of intravenous sildenafil, followed by a continuous infusion. 64 samples were taken at different time intervals and analyzed in Erasmus MC by means of a validated LC-MS/MS method. NONMEM was used to develop an pharmacokinetic model. Vital signs were monitored during treatment.

Results

Patients had a mean O/E LHR of 37.1 (17-67). At start of sildenafil the mean postnatal age was 0.9 days (0-8.5 days). Most patients received a standard loading dose of 0.4mg/kg in 3 hours, followed by a continuous infusion of 1.6mg/kg/day. The base model included two-compartment deposition of sildenafil and desmethylsildenafil. Covariate analysis included the use of ECMO, postconceptional age, postnatal age, urea, ALAT and weight. Only postnatal age influenced sildenafil and desmethylsildenafil clearance. Modelled sildenafil levels were 200mg/l (93-262) after 3 hours and increased to 312 mg/l after 12 hours. For patients receiving the standard loading dose this was 212mg/l (190-262) after 3 hours (n = 11). At time point zero Oxygentation Index (OI) was 31 (3-129) and Vasoactive Inotropic Score (VIS) was 47.9 (5.6-98). Hypotension was present in 17% of the patients and 2 patients were on ECMO. In only one patient hypotension was documented after 1 hour of infusion, another patient developed hypotension after 12 hours. However, to evaluate the effect on OI and VIS numbers were too small. 57% needed ECMO during treatment, mortality was 26%.

Conclusion

Sildenafil and desmethylsildenafil in CDH patients can be described in a two-compartment model. Only postnatal age increases its clearance. A loading dose of 0.4mg/kg in 3 hours followed by continuous infusion seems to achieve adequate sildenafil plasma levels and is well tolerated in CDH patients. To evaluate a clinical effect on pulmonary hypertension in this patient group more research is needed.

 

A Pharmacodynamic model of midazolam-induced sedation in terminally ill adult patients

L.G.W. Franken ^a*, B.C.M. de Winter ^a, A.D. Masman ^a, M. van Dijk ^a, F.P.M. Baar ^b, D. Tibboel ^a, B.C.P. Koch ^a, T. van Gelder ^a and R.A.A. Mathot ^c

^a Erasmus MC, Rotterdam.

^b Palliative Care centre, Laurens Cadenza, Rotterdam.

^c Academic Medical Center, Amsterdam.

* Correspondence: l.franken@erasmusmc.nl.

Background

Midazolam is the drug of choice for palliative sedation and is titrated to achieve the desired level of sedation. Because of large inter-individual variability (IIV) the time it takes to achieve adequate sedation varies widely. This IIV can partly be explained by differences in pharmacokinetics (PK) which have been linked to renal function and inflammatory status. Clinical outcome will however also be affected by pharmacodynamic (PD) variability. The objective of this study was therefore to evaluate the pharmacodynamic effect of midazolam in terminally ill patients, to assess the extent of IIV and to find clinically relevant covariates.

Methods

A population pharmacodynamic analysis using nonlinear mixed effect models was performed with data from 43 terminally ill patients. PK profiles were predicted by a previously described population pharmacokinetic model and depth of sedation was measured using the Ramsay sedation score. The laplacian likelihood method was applied to estimate the probability of a specific Ramsay score given a certain midazolam concentration. The effect of the midazolam metabolites was evaluated with an additive interaction model. Patient and disease characteristics, concomitant medication and the time of day were evaluated as possible covariates using a forward inclusion, backward elimination method with P-values of 0.5 and 0.001.The final model was evaluated using a visual predictive check (VPC).

Results

The effect of midazolam on the sedation level was best described with an Emax equation and a differential odds model. The EC50 value was 70.3 ug/L for a Ramsay score of 3 (drowsy/asleep responding only to commands) and 118.3 ug/L for a Ramsay score of 6 (asleep without any response). Neither of the midazolam metabolites showed an additive effect on the sedation level. Co-medication with haloperidol was the only significant covariate found. The VPC of the final model showed good model predictability.

Conclusion

We were able to accurately describe the pharmacodynamic effect of midazolam in terminally ill patients using a differential odds model with categorical data (the Ramsay sedation scale). As expected there was large variability in the overall response to midazolam. The lack of an additional effect of the metabolites may be because 1-hydroxy-midazolam is a formation limited metabolite and the treatment period may have been too short for accumulation of 1-hydroxy-midazolamglucuronide. We did find that the use of haloperidol was associated with a lower probability of sedation. This might be a result of confounding by indication as haloperidol is used to treat delirium and agitation.

 

Clinical validation of a Dried Blood Spot method for determination of risperidone, aripiprazole, pipamperone and their major metabolites

S.M. Kloosterboer *, B.C.P. Koch, B.C.M. de Winter, T. van Gelder and B. Dierckx

Erasmus MC Rotterdam, Rotterdam.

* Correspondence: s.kloosterboer@erasmusmc.nl.

Background

Risperidone, aripiprazole and pipamperone are the three most frequently prescribed antipsychotic drugs in children in The Netherlands. Therapeutic Drug Monitoring (TDM) might reduce the side effects and improve patient outcome. The Dried Blood Spot (DBS) method offers a minimally invasive and patient friendly sampling method for TDM, especially suitable for TDM in children. The aim of this study was to validate the use of the DBS method for risperidone, aripiprazole, pipamperone and its major metabolites 9-OH risperidone and dehydroaropiprazole in a clinical setting.

Methods

Paired DBS and serum samples were obtained from adult patients treated with one of the drugs under study. Samples were analysed by a previously validated UHPLC-MS/MS method. Agreement between two methods was evaluated using Deming Regression and Bland-Altman plots. Estimated Plasma Concentrations (EPC) were calculated from DBS concentrations. Sensitivity was calculated by dividing the number of identical interpretations (‘ULOQ’) by the total number of interpretations.

Results

Mean plasma levels were 8.5 µg/l for risperidone (n = 26), 33.2 µg/l for 9-OH risperidone (n = 26), 156.1 µg/l for aripiprazole (n = 14), 60.5 µg/l for dehydro-aripiprazole (n = 14) and 103.0 µg/l for pipamperone (n = 16). With Deming regression analysis, for all analytes after correction to EPC, the 95% confidence interval (CI) of the slope included 1 and the 95% CI of the Y-intercept included 0. Bland-Altman analysis showed the following mean bias (SD): risperidone 0.1 µg/l (3.9), 9-OH risperidone -1.0 µg/l (16.5), aripiprazole 0.0 µg/l (36.3), dehydroaripiprazole 0.0 µg/l (15.3), pipamperone 0.0 (42.9). As can be seen a large SD was found. However, the sensitivity of the EPCs for clinical interpretation was good with 97% for risperidone, 98% for 9-OH risperidone, 97% for aripiprazole, 99% for dehydroaripiprazole, and 91% for pipamperone.

Conclusion

Our Deming regression and sensitivity analysis demonstrate that DBS is a valid alternative for conventional venous sampling. However, this is the first study in which Bland Altman plots are added to the interpretation of DBS and plasma level comparison for antipsychotics. As can be seen, the deviation is quite large and more patients might be needed to yield a smaller variance in bias.

 

The effect of rs5758550, a putative CYP2D6 enhancer, on CYP2D6*2 phenotype and formation of endoxifen in breast cancer patients using tamoxifen

A.B. Sanchez Spitman ^a*, D.J.A.R. Moes ^a, A.J. Gelderblom ^a, V.O. Dezentjé ^b, J.J. Swen ^a and H.J. Guchelaar ^a

^a LUMC, Leiden.

^b Reinier de Graaf, Delft.

* Correspondence: a.b.sanchez_spitman@lumc.nl.

Background

CYP2D6*2 is currently considered a normal activity allele. However, recently it was suggested that CYP2D6*2 only results in normal activity in the presence of a so-called enhancer single nucleotide polymorphism (SNP) rs5758550. The aim of this study is to investigate the role of the rs5758550 enhancer in breast cancer patients treated with tamoxifen at 3 levels: population-wide, specific-group level and focused analysis of subject with endoxifen levels below the threshold.

Methods

Blood samples from the 667 female patients enrolled in the CYPTAM study (NTR1509) were used for genotyping CYP2D6 and rs5758550. Trough levels of amoxifen and its metabolites were determined with LC-MSMS. CYP2D6*2 carriers without the SNP rs5758550 were reclassified, as follows: extensive metabolizers (EM) as heterozygous extensive metabolizers (hetEM), and hetEM as intermediate metabolizers. The association between CYP2D6 phenotypes and concentration levels before and after this reclassification was analysed. A third analysis comparing CYP2D6*2 with and without the endoxifen threshold describied by Madlensky (5,97 ng/ml) was intended.

Results

In total, fourteen women were reclassified: 9 EM were reclassified as hetEM and 5 hetEM as IM. Population-wide level: After correction for absence of the enhancer, the correlation between CYP2D6 phenotypes and tamoxifen and its metabolites, did not improve. R2 for endoxifen concentration before the reclassification was 0.290, whereas after the reclassification turned was 0.279. Specific group-level: No statistically significant difference in mean concentration levels of tamoxifen and its metabolites (endoxifen, NDM-tamoxifen and 4-hydroxy-tamoxifen) among the CYP2D6*2 carriers, both with and without the described enhancer, and within each separated CYP2D6 phenotype (p ≥ 0.05).

Focused analysis of subject with endoxifen levels below the threshold: All CYP2D6*2 individuals without the SNP rs5758550 present in out cohort had an endoxifen concentration above the 15.8 nM threshold, with 16.4 nM being the lowest level present.

Conclusion

The rs5758550 enhancer does not contribute to improving the prediction of endoxifen levels in breast cancer patients treated with tamoxifen.

 

A multifaceted intervention to reduce guideline non-adherence among prescribing physicians at surgical wards

J.M. Bos ^a*, S. Natsch ^b, P.M.L.A. van den Bemt ^c, J.L.W. Pot ^d, J.E. Nagtegaal ^e, A. Wieringa ^f, G.J. van der Wilt ^b, P.A.G.M. de Smet ^b and C. Kramers ^b

^a Canisius Wilhelmina Ziekenhuis, Nijmegen.

^b RadboudUMC, Nijmegen.

^c ErasmusMC, Nijmegen.

^d UMC Utrecht, Utrecht.

^e Meander MC, Amersfoort.

^f Isala klinieken, Zwolle.

* Correspondence: jm.bos@cwz.nl.

Background

The P-REVIEW study, a prospective, multicentre, open intervention study, has shown that an approach of education of the prescriber combined with audit and feedback by the hospital pharmacist can lead to a reduction of drug-related complications among patients at surgical wards [1]. In this sub-study we also determined whether such an approach improves adherence of prescribing physicians to key pharmacotherapeutic guidelines by prescribers.

Methods

An educational program covering pain management, antithrombotics, fluid and electrolyte management, prescribing in the case of renal insufficiency, application of radiographic contrast agents and surgical antibiotic prophylaxis was presented to all prescribers on the participating wards. National and local hospital guidelines were part of this program. Hospital pharmacists performed medication safety consultations, combining medication review of high-risk patients and visits to ward physicians. Primary outcome of the study was the proportion of patients in which the physician did not adhere to one or more of the included guidelines (overall non-adherence). The difference in proportion of patients in which the physician did not adhere to the guidelines was expressed in odds ratios with 95% confidence intervals. In order to correct for possible confounding, multivariable logistic regression analysis was performed.

Results

1435 Admissions of 1378 patients who were admitted to surgical, urological or orthopaedic wards during the usual care period and 1195 admissions of 1090 patients during the intervention period were included. Overall non-adherence decreased significantly during the intervention period (21.8% (193/886)) compared to the usual care period (30.5% (332/1089)). The adjusted odds ratio (OR) was 0.61 (95% CI 0.49-0.76).

Conclusion

This study shows that education and support of the prescribing physician with respect to high-risk patients in surgical departments can improve guideline adherence among prescribing physicians at surgical wards.

Literatuur

1. Bos JM, van den Bemt PM, Kievit W, Pot JL, Nagtegaal JE, Wieringa A, Van der Westerlaken MML, Van der Wilt GJ, De Smet PAGM, Kramers C. A multifaceted intervention to reduce drug-related complications in surgical patients. Br J Clin Pharmacol. 2017;83(3):664-677.

Figure 3. Forest plot of nonadherence of prescribers to pharmacotherapeutic measures based on prevailing guidelines.

 

Potential harmful excipient exposure using common drug formulations in a paediatric hospital

A.C. van der Vossen *, S. Buljac, A.G. Vulto and L.M. Hanff

Erasmus MC, Rotterdam.

* Correspondence: a.vandervossen@erasmusmc.nl.

Background

Some excipients can be harmful to paediatric patients, often as a result of accumulation of toxic substances from immature metabolism. The European Medicines Agency (EMA) has published recommendations advising maximum daily doses of excipients which are considered to be safe. However, in daily clinical practice, the administration of excipients to paediatric patients is not monitored. Our objective was to assess the extent of exposure to potentially harmful excipients, studying actual dosages and drug formulations at the paediatric wards. Four commonly used excipients with known risks were selected; ethanol, propylene glycol (PG), benzyl alcohol and propyl paraben. With the results, drug formulations were identified that need substitution or adaptation of the composition.

Methods

Information on the composition and the quantities of excipients in the liquid formulations was retrieved from the summary of the product characteristics or via direct communication with the marketing authorisation holder or manufacturer. A dataset containing all prescriptions from February 2017 was obtained from the electronic prescribing systems of the Sophia Children’s Hospital. The daily administered amounts of excipients were calculated for each individual patient and compared with the recommended values for safe exposure.

Results

A total of 383 unique patients were admitted to the Sophia Children’s Hospital in February 2017. Surpassing of daily exposure limits occurred in 22/33 (67%) of NICU patients, 18/77 (23%) of PICU patients, and 22/311 (7%) of medium care patients. Exposure often continued over multiple days (median 6, range 1-15 days), and was most frequent with the use of caffeine oral liquid (16 patients, PG), nystatine suspension (8 patients, ethanol) and alprostadil infusion (7 patients, ethanol). In total 15 products were identified that caused excipient exposure above the recommended values. For propylene glycol, the highest doses were observed for diclofenac, fenytoin and lorazepam IV fluid, and itraconazole oral liquid. No benzyl alcohol administration above the safety limit was observed during our study period.

Conclusion

In clinical practice, a large proportion of paediatric patients is exposed to potential harmful excipients in significant amounts, and during a prolonged period of time. This is attributed to some commonly used drug formulations, which need to be improved to deliver safer drug therapy. Exceedance of recommended values was most prevalent in neonates.

 

Switching to a biosimilar infliximab: results of clinical monitoring in IBD patients

L. Binkhorst ^ab*, R.L. West ^c, R. Stuyt ^b, E.M. Westerman ^c and A. Sobels ^ab

^a Apotheek Haagse Ziekenhuizen, Den Haag.

^b HagaZiekenhuis, Den Haag.

^c Franciscus Gasthuis, Rotterdam.

* Correspondence: L.Binkhorst@hagaziekenhuis.nl.

Background

Infliximab (Remicade^TM) is an effective treatment option for patients with inflammatory bowel disease (IBD). Currently, a biosimilar of Remicade^TM is available (CT-P13; Remsima^TM, Inflectra^TM). Switching patients from the originator biological to a biosimilar is possible, but only under appropriate clinical monitoring. Patients at two general hospitals were switched to the biosimilar infliximab and were monitored carefully. Here we present the results of the clinical monitoring, and assess the feasibility, efficacy and safety of switching IBD patients from originator infliximab to the biosimilar in real life clinical practice.

Methods

All IBD patients who were treated with Remicade^TM were asked to switch to the biosimilar infliximab. The physician assessed disease activity (Global Physician’s Assessment) and disease-specific measures were evaluated (serum C-reactive protein, Harvey-Bradshaw Index or Simple Clinical Colitis Activity Index and faecal calprotectin). Adverse effects were recorded and serum trough concentrations of infliximab were measured. All parameters were assessed at baseline (t = 0) and after two infusions with biosimilar infliximab (± week 16).

Results

A total of 197 IBD patients switched to the biosimilar infliximab (77%). No difference in disease activity (as observed by the physician) or changes in disease-specific measures were seen after switching to the biosimilar. Apart from one infusion-related reaction, no other serious or unexpected adverse reactions were reported. Serum trough concentrations did not differ after switching (median, 4.1 µg/mL [range, 0.03-22 µg/mL] versus 4.6 µg/mL [range, 0.03-22 µg/mL]; P = 0.08, n = 98). A remarkable observation was the high percentage (~43%) of patients with infliximab trough concentrations below the suggested threshold of 3 µg/mL. Most of these patients maintained clinical remission. Therefore, one can doubt about the accuracy of the suggested therapeutic range for infliximab in IBD patients.

Conclusion

We conclude that switching IBD patients from originator-infliximab to biosimilar infliximab is feasible and safe in real-life clinical practice. The results support the role of therapeutic drug monitoring (TDM) in infliximab-treated patients.

 

The impact of the introduction of bortezomib on dialysis independence in multiple myeloma patients with renal impairment: a nationwide Dutch population-based study

B.E. Oortgiesen *, M.H. Hemmelder, R.E. Kibbelaar, N.J.G.M. Veeger, E.N. van Roon and M. Hoogendoorn

Medisch Centrum Leeuwarden.

* Correspondence: Berdien.Oortgiesen@znb.nl.

Background

After its introduction, evidence accumulated that bortezomib has a positive effect on the recovery of renal function in multiple myeloma (MM) patients with renal impairment (RI). In 2010, this led to an update of the Dutch and international guidelines recommending bortezomib as first-line treatment in patients with RI¹. The aim of this study was to determine the effect of the revised guideline on achieving dialysis independence in dialysis-dependent MM patients.

Methods

Patients on renal replacement therapy are registered in the Dutch registry Renine. We selected all MM patients registered between January 2002 and January 2016. As the guideline was published in The Netherlands on March 29, 2010. We divided the cohort into 2 groups: a pre- and post-guideline cohort.

Results

A total of 700 patients were included in the study (422 pre-guideline and 278 post-guideline). Since the introduction of bortezomib as the first-line treatment, 19% of the patients became dialysis independent within 4 years in the post-guideline cohort compared to 11% in the pre-guideline cohort. This occurred predominantly in the first year of dialysis treatment (HR_adj. = 2.3 [95% CI 1.4 – 3.7], Figure 4), during which more patients became dialysis independent more rapidly in the post-guideline cohort; 5% of the patients became independent after 4.6 months and 1.8 months for the pre- and post-guideline periods, respectively. In patients reaching dialysis independence, no significant difference in relapse between the two cohorts was observed (HR = 0.84 [95% CI 0.31 – 2.2]). Age < 75 years (vs ≥ 75 years) and multiple myeloma nephropathy (MCN/LCDD) as the primary renal disease (vs AL amyloidosis) were significantly associated with achieving dialysis independence (HR_adj. = 2.1 [95% CI 1.0 – 4.2] and HR_adj. = 5.7 [95% CI 2.5 – 13.2], respectively).

Conclusion

In this nationwide population-based study of dialysis-dependent MM patients in The Netherlands, the number of patients becoming dialysis independent increased more than two-fold in the first year of dialysis treatment after the introduction of the revised guideline establishing bortezomib as a first-line treatment in MM patients with RI. Age < 75 years and multiple myeloma nephropathy without amyloidosis as the primary renal disease were significantly associated with achieving dialysis independence.

Figure 4. Percentage of patients achieving dialysis independence within 4 years after starting dialysis.

 

Big data analyses for continuous evaluation of pharmacotherapy: A proof of principle with doxapram in preterm infants

R.B. Flint ^a*, W. van Weteringen ^a, S. Voller ^b, J.A. Poppe ^a, B.C.P. Koch ^a, R. de Groot ^c, D. Tibboel ^a, D.M. Burger ^c, C.A.J. Knibbe ^b, I.K.M. Reiss ^a and S.H.P. Simons ^a

^a Erasmus Medical Center, Rotterdam.

^b Leiden Academic Center for Drug Research.

^c Radboudumc, Nijmegen.

* Correspondence: r.flint@erasmusmc.nl.

Background

Drug effect evaluation is often based on human interpretation of a selection of patient data. While clinical use of ‘big data’ remains mostly limited to the traditional ‘snapshot’ assessment of a patient’s health status, analysis of continuous data brings new opportunities to improve care. Continuous quantitative analyses of high frequency patient data could not only allow proper timing of interventions, but also monitor and evaluate the effects of these interventions. We aim to evaluate the usefulness and applicability of high frequency physiological data for analyses of pharmacotherapy.

Methods

As a proof of principle, the effects of doxapram, a respiratory stimulant, on the oxygenation in preterm infants were studied. Second-to-second physiological data on arterial oxygen saturation (SpO₂), respiratory rate and heart rate were collected from bedside monitors from 12 hours before to 36 hours after start of doxapram loading dose followed by continuous maintenance dose in seven preterm infants. The effect of doxapram was determined comparing the mean of each parameter during for each hour four hours before start of doxapram in comparison with the first four hours after start, as well as with a four-hour time window 24 hours after the start of doxapram, using a paired T-test. Besides physiological data, plasma concentrations of doxapram and keto-doxapram were measured.

Results

Arterial oxygen saturation increased immediately after the start of doxapram treatment alongside an increase in heart rate, indicated by decreased variation of heart rate at the lower percentiles. The respiratory rate remained unaffected. The number of saturation dips and the time below a saturation of 80%, as well as the area under the 80%-saturation-time curve (AUC), were significantly lower during four hours after the start of doxapram compared to four hours before start, p = 0.016, p = 0.014, p = 0.011, respectively. The AUC under 90% oxygen saturation also significantly improved after start of doxapram. The improvement of respiratory parameters was sustained with maintenance dosage, comparing four hours before, with the timeframe 24 - 28 hours after start of doxapram (p < 0.05). Plasma concentrations of doxapram and keto-doxapram were quantified.

Conclusion

Using high-frequency monitoring data, we showed detailed effects of pharmacotherapy over time. We could objectively determine the respiratory condition and the effects of doxapram treatment in preterm infants, by-passing discussions on the definition of apnea and interpretation of subjective parameters. This type of analysis might help to develop individualized drug treatments with tailored dose adjustments based on a closed-loop algorithm.

Figure 5. Averages per hour of measurements from 12 hours before until 36 hours after the start of doxapram treatment for 7 patients.

 

The effect of combining acetylsalicylic acid with NSAIDs on the incidence of acute myocardial infarction

K. Lancee ^a*, B. van den Bemt^a, A. de Boer ^b, A. Lalmohamed ^c and F. de Vries ^d

^a Sint Maartenskliniek, Ubbergen.

^b Universiteit Utrecht.

^c UMC Utrecht.

^d Maastricht UMC+.

* Correspondence: k.lancee@maartenskliniek.nl.

Background

Non-steroidal anti-inflammatory drugs provide significant health benefits in the treatment of pain and inflammation. However, the use of most NSAIDs is also associated with cardiovascular adverse events like acute myocardial infarction (AMI). Naproxen has been shown to be the safest NSAID in patients with increased cardiovascular risk. However, in vitro studies postulate that naproxen and acetylsalicylic acid (ASA) have a pharmacodynamics interaction possibly leading to decreased effect of ASA and therefore theoretically increasing the risk of AMI. However, clinical data supporting the effect of this interaction on the risk of AMI are absent. This study therefore aims to assess the impact of the use of ASA on the association between the use of different NSAIDs and the risk of AMI in patients with osteoarthritis.

Methods

Data for this study were obtained from the primary care database Clinical Practice Research Datalink CPRD. All individuals aged ≥18 years who had undergone primary total joint replacement surgery between 1 January 1987 and 31 October 2012 were selected. Patient were divided into four categories: ASA-users not using NSAIDs, NSAID users (not using ASA), ASA+NSAID users and a control group. Total follow-up time was divided into one-month-intervals. Exposure to NSAIDs and ASA was determined in the three months before the start of each interval. The effect of the NSAIDs on cardiovascular events was further stratified by the type of NSAID on compound level and cumulative daily defined dosages (DDDs) in the previous year.

Results

A total of 619.775 person years were recorded. The incidence of AMIs in the control group was 3.82/1000py. In a fully adjusted model, treatment with diclofenac and celecoxib led to highest significant hazard ratios for AMI, compared to the control group (diclofenac HR 1.56 (95% CI 1.34-1.81; celecoxib HR 1.82 (95% CI 1.27-1.81)). On the other hand, use of naproxen did not significantly affect the incidence rate of AMI (HR 1.29 (95% CI 1.06-1.56)). This result was consistent in the ASA+NSAID group: diclofenac HR 1.34 (95% CI 1.05-1.70), celecoxib HR 2.91 (95% CI 1.84-4.58), naproxen HR 0.59 (95% CI 0.30-1.19). Effect of cumulative dose in the previous year was assessed for ibuprofen, diclofenac and celecoxib plus ASA. A dose-effect relationship was seen.

Conclusion

The combination of naproxen and ASA did not lead to an increased risk of AMI compared to ASA users or a control group, whereas the combinations diclofenac or celecoxib plus ASA significantly increased the risk of AMI.

 

The introduction of biosimilar rituximab in clinical practice

J. Bongers ^a*, A. Sobels ^b and E.M. Westerman ^a

^a Franciscus Gasthuis & Vlietland / FarmaXL, Rotterdam.

^b Apotheek Haagse Ziekenhuizen - HagaZiekenhuis / FarmaXL, Den Haag.

* Correspondence: j.bongers@franciscus.nl.

Background

The biological rituximab is widely used for treating various types of autoimmune disease and cancer (e.g. rheumatoid arthritis, idiopathic thrombocytopenic purpura and non-Hodgkin’s lymphoma). For these indications rituximab is an effective treatment, but it is also highly expensive. The introduction of rituximab biosimilars can reduce healthcare costs. Switching from an originator to a biosimilar is possible according to the Medicines Evaluation Board (CBG-MEB) and Dutch Federation of Medical Specialists, provided that adequate clinical monitoring takes place. We evaluated the controlled switching of patients from originator MabThera® to the biosimilar Truxima® in Farma-XL* affiliated hospitals Franciscus Gasthuis & Vlietland and HagaZiekenhuis.

Methods

In collaboration with haematologists and rheumatologists, Farma-XL created a monitoring plan to evaluate the introduction of biosimilar rituximab. Rituximab-naïve patients, as well as switching patients were eligible for inclusion in the monitoring plan, which includes data until mid-August 2017. Follow-up lasted from the first administration of the biosimilar until the second administration or discontinuation. Switching patients were identified by using prescription data generated by the electronic prescribing system (Chipsoft HiX) and Cytostatica Management System (CMS). A switching patient was defined as a patient who had an administration of biosimilar rituximab less than one year after the last administration of the biological. Using the electronic patient data registry, the following data were collected: medical indication, former infusion reactions on the originator, infusion reactions after the first and second biosimilar administration, treatment of infusion reactions and other noteworthy side effects that could be attributed to the administration of rituximab.

Results

101 patients were included in the monitoring plan. Of these patients 26 (25,7%) were rituximab-naïve and 75 (74,3%) switched from the originator to the biosimilar. 15 of the switching patients (20,0%) formerly experienced an infusion reaction on the originator. Two of the switching patients (2,6%) and six of the rituximab-naïve patients (23,1%) experienced an infusion reaction during their first administration of the biosimilar. This is less than reported in clinical trials. None of the patients who experienced infusion reactions discontinued their medication. Reactions were successfully treated by reducing infusion flow or administration of either clemastine and prednisolone, or both. 62 patients were observed during a second administration of the biosimilar. None of these patients experienced infusion reactions.

Conclusion

Switching of rituximab originator to biosimilar in clinical practice did not raise any specific concerns with regard to infusion reactions or side effects.

 

The effect of a medication reconciliation program in two Intensive Care Units in The Netherlands

B.E. Bosma ^a*, N.G.M. Hunfeld ^b, R.A.M. Quax ^b, E. Meuwese ^bc, M.J. Kranenburg ^ad, P.H.G.J. Melief ^a, J. van Bommel ^b and P.M.L.A. van den Bemt ^b

^a Haga Teaching Hospital, Den Haag.

^b Erasmus Medical Centre, Rotterdam.

^c HMC Haaglanden, Den Haag.

^d Gelre Hospitals, Apeldoorn.

* Correspondence: l.bosma@hagaziekenhuis.nl.

Background

Medication errors occur frequently in the intensive care unit (ICU) and during care transitions.The critical illness at the time of ICU admission makes that chronic medications used at home are often temporarily stopped. Unfortunately, when the patient improves, the restart of this medication is easily forgotten. In addition, drugs initiated at the ICU for short time use, are often inadvertently continued after ICU and hospital discharge. Medication reconciliation could be useful to prevent such errors. Therefore, the aim of this study was to determine the effect of medication reconciliation at the ICU.

Methods

A prospective 8-month intervention study with a pre- and post-phase was performed in Haga Teaching Hospital in The Hague (2013) and Erasmus University Medical Center in Rotterdam (2014). The intervention consisted of medication reconciliation by pharmacists both at the moment of ICU admission and prior to ICU discharge.Patients were included when they used at least one medication at home and when the ICU length of stay exceeded 24 hours. Medication transfer errors (MTE) were collected and the severity of potential harm of these MTEs (pADE=0; 0.01; 0.1; 0.4; 0.6) was scored. Primary outcome measures were the proportions of patients with ≥ 1 MTE at ICU admission and after ICU discharge. Secondary outcome measures were the proportions of patients with a pADE score ≥ 0.01 due to these MTEs, the severity of the pADEs and a cost-benefit ratio. Odds ratio and 95% confidence intervals were calculated, by using a multivariate logistic regression analysis.

Results

In the pre-intervention phase 266 patients were included and 212 in the post-intervention phase.The proportion of patients with ≥1 MTE at ICU admission was reduced from 45.1% (n = 119) to 14.6% (n = 31) (OR_adj = 0.18 [95% CI 0.11-0.30]) and after ICU discharge from 73.9% (n = 150) to 41.2% (n = 73) (OR_adj 0.24 [95% CI 0.15-0.37]).The proportion of patients with a pADE ≥ 0.01 at ICU admission was reduced from 34.8% (n = 92) to 8.0% (n = 17) (OR_adj = 0.13 [95% CI 0.07-0.24]) and after ICU discharge from 69.5% (n = 141) to 36.2% (n = 64) (OR_adj = 0.26 [95% CI 0.17-0.40]).The pADE reduction resulted in a positive cost benefit ratio of 2.48, leading to a potential net cost benefit of €103,- per patient.

Conclusion

Medication reconciliation by pharmacists at ICU transfers is an effective safety intervention, leading to a significant decrease in the number of errors and a cost effective reduction of potential adverse drug events.

 

Clinical and financial impact of pharmacist interventions during patient rounds in two Intensive Care Units in The Netherlands

B.E. Bosma ^a*, P.M.L.A. van den Bemt ^b, P.H.G.J. Melief ^a, J. van Bommel ^b and N.G.M. Hunfeld ^b

^a Haga Teaching Hospital, The Hague.

^b Erasmus Medical Centre, Rotterdam.

* Correspondence: l.bosma@hagaziekenhuis.nl.

Background

The risk of prescribing errors and related adverse drug events (ADEs) on the intensive care unit (ICU) is high. A clinical pharmacy service can reduce ADEs and lower overall costs. However most clinical pharmacist studies on ICUs are single center and carried out in North America or the UK, making these findings not generalizable as such for the European setting. This study looked into the clinical and financial impact of interventions made by pharmacists during patient rounds in two different ICU settings in The Netherlands.

Methods

An interventional, prospective study was performed in Haga Hospital, a general teaching hospital (GTH) and in Erasmus MC, a university hospital (UH) in The Netherlands. The intervention was a proactive ICU pharmacist intervention method. This method consisted of an ICU standardized medication review by an ICU trained pharmacist. During the following patient round the collected potential interventions were presented to the attending intensivists. All interventions were collected and categorized based on type of intervention, type of medication involved and whether the intervention was accepted by the attending intensivist. Clinical relevance and a prevented potential ADE score (pADE: 0; 0.01[very low]; 0.1; 0.4; 0.6 [high]) were assessed by two assessors, one intensivist and one hospital pharmacist with ICU experience. Main outcome measures were the proportion of accepted interventions, the number of accepted interventions per patient, the clinical relevance of the accepted interventions and the number of pADE per patient. A preliminary cost-benefit ratio was performed based on cost reduction, cost avoidance and costs of labour of the pharmacist. Cost reduction was measured through the stop interventions and drug dose reducing interventions. Cost avoidance was measured through the interventions which reduced patient harm by the prevention of pADE.

Results

In the GTH 160 patients and in the UH 174 patients were included. 332 and 280 interventions were analysed. Acceptance of the interventions was 67.3% (GTH) and 61.8% (UH). The number of the accepted interventions was 1.2 (GTH) and 0.9 (UH) per patient, and the interventions were mostly scored as clinically relevant, resulting in 0.16 and 0.11 prevented pADE per patient. The preliminary cost benefit ratio was positive in both hospitals leading to a €119 (GTH) and €136 (UH) benefit per accepted intervention.

Conclusion

This clinical pharmacy service in two ICUs resulted in high numbers of accepted and clinically relevant interventions. Our model appeared to be cost effective in both ICU settings.

 

Development of a model to predict qtc interval prolongation in patients who use one or more qt-prolonging drugs

A.N. Bindraban ^a, J. Rolvink ^a*, F.A. Berger ^b, P.M.L.A. van den Bemt ^b, A.F.M. Kuijper ^c, R.T.M. van der Hoeven ^a, A.K. Mantel-Teeuwisse ^d and M.L. Becker ^a

^a Pharmacy Foundation of Haarlem Hospitals, Haarlem.

^b Erasmus MC, Rotterdam.

^c Spaarne Gasthuis, Haarlem.

^d Universiteit Utrecht.

* Correspondence: jrolvink@sahz.nl.

Background

 

Several drugs can induce QTc interval prolongation, which could lead to sudden cardiac death. Healthcare providers receive warnings for interactions between two QT-prolonging drugs. It’s debatable whether these signals are clinically relevant and how these warnings should be managed. Our aim is to develop a model to predict QTc interval prolongation in patients who use one or more QT-prolonging drugs.

Methods

This retrospective cohort study included all ECGs of adult in- and outpatients, who used one or more QT prolonging drugs when their ECG was recorded. ECGs with a QRS interval > 120ms were excluded. Independent risk factors were determined using backwards conditional stepwise multivariate logistic regression. The continuous variables were categorized based on their odds ratios. Significant variables (P≤0.050) were considered independent risk factors. Risk scores were assigned based on their regression coefficient, and the area under the ROC-curve was determined.

Results

Of the 12,949 ECGs included, 880 (6.9%) ECGs showed QTc prolongation (QTc interval >500ms, corrected with Bazett’s formula). The variables analysed were serum electrolytes, age, eGFR (MDRD), serum ALAT, number of QT prolonging drugs in use, maximum QTc measured in the past year, gender, diabetes mellitus, and use of antiarrhythmics, loop diuretics, thyroid hormones, beta-blockers, verapamil and/or diltiazem and acetylsalicylic acid. The independent risk factors and risk score, are age > 70 years (P = 0.009), risk score 1; use of antiarrhythmics (P < 0.001), 1; use of loop diuretics (P < 0.001), 2; eGFR ≤ 60 ml/min/1.73m² (P < 0.001), 2; serum calcium level ≤ 2.14 mmol/L (P < 0.001), 2; serum potassium level 3.0 to 3.4 mmol/L (P < 0.001), 3; maximum QTc measured between 480ms and 500ms (P < 0.001), 3; serum potassium level ≤ 2.9 mmol/L (P < 0.001), 7; and a maximal QTc measured of >500ms (P < 0.001), 7. The area under the ROC curve is 0.728. In a simplified model the maximum QTc measured, serum calcium and magnesium levels were excluded. For this model, the risk factors and risk scores are age >70 years (P = 0.012), 1; use of beta-blockers (P = 0.030), 1; eGFR ≤ 60 ml/min/1.73m² (P < 0.001), 2; use of antiarrhythmics (P < 0.001), 3; use of loop diuretics (P < 0.001), 4; and serum potassium level between 3.0-3.4 mmol/L (P < 0.001), 4 and ≤ 2.9 mmol/L (P < 0.001), 8. The area under the ROC curve is 0.658.

Conclusion

Patients with higher risk scores have a higher chance of developing QTc interval prolongation. This model can be used to identify patients at risk for a prolonged QTc interval and to improve medication surveillance.

 

The nation-wide impact of implementing a preemptive pharmacogenetic panel approach to guide drug prescribing in primary care in The Netherlands

P.C.D. Bank *, J.J. Swen and H.J. Guchelaar

LUMC, Leiden.

* Correspondence: p.c.d.bank@lumc.nl.

Background

Drugs prescribed in primary care are typically not considered of high risk. However, with a high amount of prescriptions each year in combination with a high frequency of actionable phenotypes among the general population pharmacogenetics (PGx) may also have a high impact in primary care in terms of absolute number of patients. In this study we set out to make an estimate of the occurrence of gene-drug pairs among the Dutch population using a panel of eight genes and estimate the number of patients that would have required an intervention in 2016 based on their predicted phenotype.

Methods

The guidelines of the Dutch Pharmacogenetics Working Group (DWPG) were reviewed for drugs with a clinical relevant gene-drug pair and an actionable recommendation for a least one phenotype. Dispensing data in the period from January 1^st up to December 31^st 2016 were obtained from the Foundation of pharmaceutical Statistics. The frequencies of the genetic predicted phenotypes of the panel of CYP2C9, CYP2C19, CYP2D6, CYP3A5, DPYD, SLCO-1B1, TPMT and VKORC1 obtained in the “Implementation of Pharmacogenetics into Primary Care Project” were used as a representation of the national population.

Results

In 23.6% of all incident prescriptions (n = 3.628.597) of 45 drugs in 2016 an actionable gene-drug interaction was present according to the guidelines of the DPWG. In 5.4% incident prescriptions a dose-adjustment or switch of therapy was required as per the DPWG guidelines and in 18.2% of patients the clinical monitoring of therapy should have been intensified.

Conclusion

Based on the presented data it can be concluded that gene-drug pairs are common in primary care and if the genotypes of the selected panel would have been known at start of therapy this would have resulted in a change in drug or dose in one in every nineteen patients.

 

Adherence to the Dutch Pharmacogenetics Working Group guidelines in the PREPARE study

C.H. van der Wouden *, J.J. Swen and H.J. Guchelaar

LUMC, Leiden.

* Correspondence: c.h.van_der_wouden@lumc.nl.

Background

The PREemptive Pharmacogenomic testing for Preventing Adverse drug REactions (PREPARE) study aims to implement pre-emptive pharmacogenetics (PGx) testing of a panel of pharmacogenes in seven European countries to guide pharmacotherapy. The primary objective is to quantify the collective clinical utility of the panel. An important prerequisite to achieving this objective is adherence to the Dutch Pharmacogenetics Working Group (DPWG) guidelines. Here we provide preliminary results regarding DPWG guideline adherence.

Methods

PREPARE is a multi-center, open, randomized, cross-over study (n=8,100). Countries have been randomised to start with either PGx-guided prescribing (study arm) or standard of care (control arm) for 18 months. After this period, the opposite strategy will be implemented for 18 months. Patients of any ethnicity who are at least 18 years old and receive a first prescription for a drug for which a DPWG guideline is available, as part of routine care, are eligible to participate. For patients within the study arm, their results will be: 1) recorded in the (electronic) medical record and 2) provided to the patient in the form of a plastic card. Genetic results and DPWG guidelines can be used by physicians and pharmacists (HCPs) to guide the dose and drug selection for the drug of inclusion. HCPs are given PGx test results but are not forced to adhere to the DPWG guidelines. When a PGx result is actionable, HCPs report whether they adhered to the DPWG guidelines (responded by yes/no). When responded with yes, the effect on the pharmacotherapy and the reason for adherence is asked. When responded with no, the reason for non-adherence is asked.

Results

Currently, 258 patients have been enrolled in the control arm and 386 patients in the intervention arm. Among patients, for whom data is available in the intervention arm (n = 178), 30.3% (n = 54) have an actionable PGx result regarding the drug of inclusion. DPWG adherence was reported by HCPs after acting upon 29 actionable PGx results. For 79.3% (n = 23) prescriptions, HCPs chose to adhere to the relevant guideline and reported: “I thought it was useful for preventing ADEs”(n = 23) as the reason for adherence. 20.7% (n = 6) chose not to adhere and reported: “I felt it would decrease the efficacy of the therapy”(n = 3).

Conclusion

Current DPWG guideline adherence is relatively high, when compared to a previous study where HCP adherence to hypertension practice guidelines was 70.4%. This suggests HCPs have understood, adopted and believe in PGx guided prescribing and dispensing.

 

Clinical validation of clinical rules: evaluation and optimization

B. Sezgi *, E. Engel-Dettmers and M.A. Damhof

Ziekenhuisgroep Twente, Hengelo.

* Correspondence: bsk.sezgi@gmail.com.

Background

Medication surveillance alerts may result in alert fatigue for both prescribers and pharmacists. Only relevant alerts are shown to prescribers and pharmacists in the Ziekenhuisgroep Twente (ZGT) to prevent alert fatigue. A clinical decision support system (CDSS) is in place using clinical rules to generate relevant alerts for prescribers and pharmacists. This may contribute to a reduction in medication errors and improvement of patient safety. The majority of CDSS-alerts are generated overnight and shown on a worklist for the attending pharmacist. Since June 2016 several CDSS-alerts are also shown in the clinical physician order entry system (CPOE). The CDSS is technically validated using ‘dummy-patients’ in a test setting. The objective of this study was to perform a clinical validation of the clinical decision support system.

Methods

All CDSS-alerts were retrospectively compared in March and October 2016 to evaluate the optimization of the clinical rules in use. CDSS-alerts were screened for an intervention by the pharmacy, the relevance and nature of the intervention and follow-up in the electronic medical record.

Results

In the ZGT 62 clinical rules are in use. The CDSS generated 3519 and 2610 alerts in March and October respectively, with an average of 54,3% unique alerts. This is a decrease of 23% in total alerts (after a correction for admitted patients). Pharmacy interventions were determined necessary in 23,7% of the alerts. In approximately 20% of those alerts, pharmacy interventions were not carried out. An increase in pharmacy interventions was seen between March and October (88,5% versus 74,7%). The compliance for follow up by prescribers was overall circa 55%. After introduction of CDSS-alerts in the CPOE the total alerts generated for those rules with CDSS-alerts was decreased with 24,2%.

Conclusion

The introduction of CDSS-alerts in CPOE resulted in a reduction of 24,2% generated alerts for pharmacists. Use of clinical rules results in 23,7% of the alerts in pharmacy interventions towards the prescriber. There was a high compliance in follow up by prescribers. Besides a technical validation it is important to perform a clinical validation of the CDSS. Periodically an evaluation of the CDSS is necessary for optimization of the clinical rules in use.

 

How many patients are at risk for adverse drug events on admission in your hospital?

S.E.H. Detert Oude Weme, A.G. Otten-Helmers, J.J.W. Ros, P.E. Spies and H.J.M. Van Kan *

Gelre ziekenhuizen, Apeldoorn.

* Correspondence: hjm.vankan@gelre.nl.

Background

Over the past decade many efforts have focused on the identification of risk factors associated with adverse drug events (ADEs). These include several specific risk related medication (RRM) categories, that may lead to medication related hospital admissions or otherwise compromise patient safety. In our hospital the departments of clinical pharmacy, geriatrics an internal medicine collaborate to improve patient safety by selecting patients for medication review and make pharmacotherapeutic recommendations to their healthcare providers. For this purpose we developed an evidence based risk classification for use on admission to identify patients who are prone for ADEs or who are admitted due to a possible ADE. The objective of this study was to gain insight on the number of patients eligible for medication review when applying this classification.

Methods

A 4-week prospective cohort study (GelreADRisk) was performed in all patients admitted to our large non-academic teaching hospital for at least 24 hours. The computerized physician order entry system was checked for prescribed RRMs on admission and/or possibly medication related symptoms as reason for admission. Possible ADEs were identified according to the trigger list from the draft multidisciplinary guideline for polypharmacy in hospitalized elderly patients (MPHE). With this information the number of patients eligible for potential medication review by an internist, a geriatrician or a hospital pharmacist, was determined according to a flowchart based on predefined patient categories:

• possible ADE + age up to 70 yrs (internist);

• possible ADE + age ≥ 70 yrs (geriatrician);

• no possible ADE identified, but ≥ 5 RRMs (hospital pharmacist).

Results

A total of 1626 consecutively admitted patients were screened on admission. One or more possible ADEs were detected in 263 (16%) patients. The results of this study in terms of estimated workload (number of patients) for the involved professionals is given in table 1. For hospital pharmacists the estimated average daily workload when a threshold of 4, 3, 2 and 1 RRM is used, would be 10.9, 19.3, 30 and 41.2 patients respectively. The top-3 possible ADE categories found in this study were collapse/hypotension/ dizziness/confusion/sedation, electrolyte imbalance/dehydration and bleeding/elevated INR. The RRM groups that patients used most frequently (top-3) on admission were anticoagulants/antiplatelet drugs, diuretics and ACE/ATII inhibitors.

Conclusion

The results of this study give a first insight on the number of patients eligible for potential medication review according to the proposed classification. A substantial workload for all involved healthcare professionals could be demonstrated.

 

Age-related and sex-related differences in use and toxicity of fluoropyrimidine-oxaliplatin therapy for colon cancer

S.S.S. Jankie ^a*, L. Binkhorst ^a, P. van der Zee ^b, L.E. Visser ^a and A. Sobels ^a

^a Apotheek Haagse Ziekenhuizen - HagaZiekenhuis, Den haag.

^b IJsselland Ziekenhuis, Capelle aan den IJssel.

* Correspondence: s.s.s.jankie@students.uu.nl.

Background

A fluoropyrimidine plus oxaliplatin is the standard of care for the adjuvant treatment of high risk colon cancer. Both capecitabine and 5-fluorouracil can be used as chemotherapy backbone, as the clinical efficacy is equal. There has been some speculation that fluoropyrimidine-related toxicity differs between men and women [1]. In addition, elderly patients may experience more treatment-related toxicity than younger patients. In this study, we evaluated ‘real world’ data from three general hospitals to examine sex-related and age-related toxicity differences in patients treated with a fluoropyrimidine (plus oxaliplatin) for colon cancer.

Methods

Data were collected from adult (age ≥ 18) patients who received adjuvant fluoropyrimidine-based chemotherapy for high risk (stage II-III) colon cancer at three hospitals (HagaZiekenhuis, the Hague; Reinier de Graaf Gasthuis, Delft; and IJsselland Ziekenhuis, Capelle aan den IJssel) between between 2009 and 2016. The collected data included patient demographics, tumor characteristics, stage, fluoropyrimidine therapy (monotherapy or combination, dose, number of cycles), toxicity, comorbidity and clinical outcome.

Results

Data were available for a total of 226 patients; 126 (56%) males and 100 (44%) females. Sixty-five patients (29%) were older than 70 years (at start of treatment). Eighty-one percent of the patients received therapy with capecitabine, nineteen percent of the patients with 5-fluorouracil. Older patients received fluoropyrimidine monotherapy more often than younger patients (52% vs. 4%; p = 0.01 (χ² test)). Although not statistically significant, women received oxaliplatin less frequently compared to men (20% vs. 16%; p = n.s.). Only 41 patients (18%) completed all cycles of therapy (fluoropyrimidine plus oxaliplatin). Women were more likely to discontinue therapy than men (88% vs. 76%; p = 0.033 (χ² test)). Therapy was discontinued because of toxicity or disease progression. The most commonly observed toxicities included hand-foot syndrome, gastro-intestinal disturbances, neuropathy and neutropenia. Toxicity was reported more often in women than in men, however, neuropathy was observed more frequently in men (59% vs. 53%). Hand-foot syndrome was observed more often in older patients than younger patients (35% vs. 17%), while younger patients suffered more frequently from neuropathy (65% vs. 35%).

Conclusion

Capecitabine was the most commonly used fluoropyrimidine. Elderly patients and women received less frequently standard of care therapy with oxaliplatin. Only a small number of patients completed their whole treatment course, usually because of premature oxaliplatin discontinuation. Older patients suffered more from toxicity, except neuropathy. Women suffered more than men from toxicity.

Literatuur

1. Chansky et al., Cancer, 2005.

 

Validation of an automated delirium prediction model with the Delirium Observation Screening scale

R.H. Creemers ^ab*, B.C.M. Jansen ^ab, C. Mestres Gonzalvo ^c, B.P.C. van Oijen ^b, B. Winkens ^c, P.H.M. van der Kuy ^b

^a Universiteit Maastricht.

^b Zuyderland MC, Sittard.

^c Elkerliek, Helmond.

* Correspondence: r.creemers@zuyderland.nl.

Background

Delirium is a rapidly altered mental state characterized by fluctuating consciousness, attentional deficit, psychotic features and altered behaviour. It is a common postoperative complication within geriatric patients. Delirium has negative consequences such as higher health care related costs and increase in morbidity and mortality. In 30-40 % of the cases it can be prevented. To facilitate early recognition, the Delirium Observation Screening (DOS) scale is used. In 2013 a delirium prediction model, the ‘Delirium-model’(DEMO) has been developed retrospectively. This model assesses the risk of developing delirium based on electronically available data. In 2017 the DEMO-analysis was prospectively validated using the clinical diagnosis delirium. As delirium is a clinical diagnosis which is thought to be underreported, we aimed to prospectively validate the delirium prediction model using available DOS-scores.

Methods

Daily, all hospitalized patients ≥ 60 years are analysed using the DEMO-analysis. The DEMO-score predicts whether a patient is at risk of developing delirium after hospital admission. A random sample of 500 elective admissions and 500 acute admissions to the hospital was included. After inclusion DOS-scores were used to check if the patient had developed a delirium in order to classify patients as True Positive, True Negative, False Positive or False Negative. Analysis were performed to validate the model: DOS-score ≥ 3 at day 1, day 3 and day 5 compared to DEMO-analysis at admission.

Results

The data was collected during a 10-month period starting in January 2016. At random a group of 1000 patients was selected of which 500 elective admissions and 500 acute admissions. A total of 314 patients had at least one reported DOS-score during admission: 143 (28.6%) in the elective group and 171 (34.2%) in the acute group. The sensitivity of predicting delirium according to DOSS is high at 78.4% in the elective group and 88.2% in de acute group at day 5 (table 2).

Conclusion

The DEMO has a high predictive value with a high sensitivity. On the other hand, specificity is low. One explanation is that DOS-scores are only documented when there is a suspected higher risk of developing delirium at time of admission. This has in particular an effect on the True Negative value, and thus, the specificity since only patients with a high risk of developing delirium have documented DOS-scores. Further research is indicated to investigate whether the absence of DOS-scores corresponds to the absence of the clinical diagnosis delirium in patients.

 

Bleeding risks are high in elderly persons who use vitamin K antagonists combined with platelet aggregation inhibitors

N. van Rein ^ab*, U. Heide-Jørgensen ^b, W.M. Lijfering ^a, O.M. Dekkers ^ab, H.T. Sørensen ^b and S.C. Cannegieter ^a

^a Leiden University Medical Center.

^b Aarhus University Hospital.

* Correspondence: n.van_rein@lumc.nl.

Background

Patients with atrial fibrillation generally require treatment with vitamin K antagonists (VKAs) and at times with additional platelet aggregation inhibitors. Data are scarce on bleeding rates in high-risk groups receiving combination therapy, such as the elderly or patients with a high CHA₂DS₂-VASc score.

Methods

We conducted a nationwide cohort study of Danish atrial fibrillation patients aged 50 years or older. Treatments were ascertained from a prescription database. These included no anticoagulant treatment and treatment with VKAs, aspirin, clopidogrel, and combinations of anticoagulant drugs. Incidence rates (IRs) of major bleeding and hazard ratios were estimated, overall and stratified by treatment modality, age, CHA₂DS₂-VASc score, and comorbidity.

Results

We identified 216,109 patients with atrial fibrillation. Median age was 75 years and 48% were women. Over a total follow-up period of 854,914 patient-years (py), 24,414 major bleeds occurred [incidence rate (IR) 2.9/100 pys, 95% confidence interval (CI) 2.8-2.9/100 pys]. Compared with VKA monotherapy, adjusted hazard ratios of major bleeding were 1.52 (95% CI 1.37-1.69) for dual antiplatelet therapy, 1.78 (95% CI 1.71-1.86) for therapy with a VKA and an antiplatelet drug, and 3.73 (95% CI 3.23-4.31) for triple therapy. Subgroup analyses showed similar patterns. The IR for major bleeding was 11.9/100 pys among triple-therapy patients. Very high major bleeding rates occured among patients over 90 years (IR 50.0/100 pys, 95% CI 24.4-91.8) and in patients with a CHA₂DS₂-VASc score over 6 (IR 20.0/100 pys, 95% CI 10.2-35.7).

Conclusion

Patients with atrial fibrillation on triple therapy experienced high rates of major bleeding compared with patients on dual therapy or monotherapy. The exceptionally high bleeding rates observed in patients on triple therapy over the age of 90 years or with a CHA₂DS₂-VASc score over 6 suggest that such therapy should be carefully considered in these patients.

 

Medication review in elderly patients by a pharmacist during hospitalisation

A. Keyany *, J. van Citteren, M.A.J.P.J Lenoir-van Berkel and K.A. Simons-Sanders

Elisabeth-TweeSteden Ziekenhuis, Ziekenhuisapotheek Midden-Brabant, Tilburg.

* Correspondence: a.keyany@etz.nl.

Background

Recent report showed that elderly patients are at risk for medication related hospital admissions. Polypharmacy is a growing concern in the elderly population and has been associated with an increased risk of adverse drug events, drug-interactions, non-adherence and reduced functional capacity. There is some evidence that clinical medication reviews by a pharmacist in close contact with the physicians may lead to reduced drug related harms. In this pilot we studied the nature, type and frequency of potential drug related problems (pDRPs) in elderly patients and the effect of medication review by pharmacist during hospitalisation on DRPs.

Methods

Patients 85 years of age and older, who were admitted for at least 24 hours, were selected on a daily basis in the ETZ location Elizabeth. Patients in the intensive care unit and palliative patients were excluded. Medication verification was performed by pharmacy technicians. A pharmacist analysed the medication profile using the STRIP method and according to the STOP- and START criteria. All relevant data (current medication, laboratory results, case history, pDRPs) was entered into a specially developed software system. Identified pDRPs were discussed with the physicians.

Results

Between 1 December 2016 to 10 March 2017, 264 medication reviews were carried out. We found 566 pDRPs, with an average of 3 DRPs per patient. The majority (66%) of the DRPs were related to medication at admission and 34% medication started during the hospital admission. Overtreatment (26%) and undertreatment (36%) were the most common pDRPs. In total 403 pDRPs were discussed with physicians; 37% of the pDRPs were resolved during admission and/or at discharge.

Conclusion

Sixty-six percent of the pDRPs were related to medication at admission. At discharge 36% of the pDRPs were resolved. Since the majority of the pDRPs are related to medication at admission, these pDRPs should be communicated with the general practitioner and community pharmacist.

 

Medication reconciliation at care transitions: a longitudinal prospective study

S. Daliri ^ab*, M. Bouhnouf ^b and F. Karapinar-Carkit ^b

^a AMC, Amsterdam.

^b OLVG ziekenhuis, Amsterdam.

* Correspondence: s.daliri@olvg.nl.

Background

A widely recognized method to reduce medication errors and promote patient safety at care transitions is medication reconciliation (MR). In hospitals, generally the MR process stops at hospital discharge. Little is known of the effect of MR in the healthcare continuum. The aim of this study was to investigate the effect of MR in the healthcare continuum at admission, discharge and post-discharge.

Methods

A longitudinal multicenter study was conducted in two Dutch hospitals in Amsterdam (OLVG and BovenIJ hospital). Patients discharged from the internal medicine-, neurology-, cardiology- and pulmonology departments were included if at least one medication change was initiated during admission. These patients received MR at admission, discharge and post-discharge during a home visit. At admission and discharge pharmacy technicians and pharmaceutical consultants (specialized pharmacy technicians) performed the MR. Community pharmacists of the patients (n=50) performed the MR post-discharge. MR interventions were classified as:

• discrepancies (unintentional differences between actual and documented medication use, e.g. omissions, dose discrepancies);

• optimisations (adhering to guidelines, e.g. discontinue hypnotics without an indication, add laxative to opioid use, adjust medication based on laboratory values);

• and patient handling interventions (improvement of medication use by the patient, e.g. advise on how to use inhaler drugs, explanations regarding side effects).

The primary outcome was the number and type of interventions per patient during MR at these transition moments. Data were analysed by means of descriptive statistics.

Results

In total 197 patients were included. At hospital admission, 2.0 (SD: ± 2.1) interventions per patient were identified, these were primarily discrepancies (1.8/patient) as a result of omissions and dosage discrepancies. At hospital discharge, 2.7 (SD: ± 2.4) interventions/patient were recorded. These were mainly optimisations (1.0/patient, e.g. discontinuing medication without an indication such as hypnotics, laxatives, proton pump inhibitors and adjustments based on kidney function) and patient handling interventions (1.2/patient). Post-discharge 5.6 (SD: ± 3.9) interventions per patients were identified. Mainly patient handling interventions (4.4/patient) were identified as for example the indication of many medication was unclear for patients or there were questions on side effects. Also, discrepancies (0.9/patient) were identified, mainly due to the use of OTC drugs and because patients did not implement the medication changes of the hospital accurately. Optimisations were needed for patients who experienced side effects.

Conclusion

This study shows that continuous medication reconciliation is needed. Medication reconciliation identifies different interventions at the several transition points in the healthcare continuum.

 

Availability and quality of patients' home medication information derived from the ‘LSP', the Dutch national ICT infrastructure linking health care providers' information systems, at admission to the hospital

M. Duisenberg- v Essenberg ^a*, B. Maat ^a and P.M.L.A. Van den Bemt ^b

^a ETZ Tilburg.

^b Erasmus MC, Rotterdam.

* Correspondence: m.duisenberg@etz.nl.

Background

According to Dutch medication safety regulations, a health care provider must have the patient’s up-to-date home medications available when prescribing, administering and dispensing medication. In The Netherlands a national ICT infrastructure linking health care providers’ information systems has been made available per 2015 as a source for patients’ home medication information. It is called ‘Landelijk Schakelpunt’, or ‘LSP’. The Elisabeth-TweeSteden (ETZ) hospital pharmacy started using the LSP in the medication reconciliation process on March 1^st 2015. In a pilot study patients’ LSP information seemed inavailable quite often and of suboptimal quality. Therefore the objective of this study was to determine the percentage of admitted patients for whom home medication information was available using the LSP infrastructure as source, and to assess the quality of obtained information.

Methods

A prospective cohort study was performed including all elective and emergency patients for whom the LSP was queried at admission between 01-03-2015 and 01-06-2015 and between 01-12-2015 and 01-03-2016. Per patient it was documented whether or not the LSP provided home medication data. A random sample of patients was drawn to assess the quality of LSP information by comparing LSP information to the patients’ community pharmacy ‘AMO’ (actual home medication overview), i.e. the source considered the gold standard.

Results

In the first 3-month period, the LSP was queried for 10,944 patients, resulting in home medication information for 54% of the patients (5,945 hits) Half a year later, in the second 3-month period, this percentage was higher: 65% (4,948 hits for 7,633 patients). The assessment of the quality of the LSP information is presented in table 1. Most prominent result was that 31% (n = 101) of home medications that were not in use anymore according to the LSP, were present on the community pharmacy AMO. Also striking was the percentage of medications that was present on the LSP overview, but not on the AMO (16%), and vice versa (18%). Examples of missing medications were irregularly dosed coumarines and ‘if needed’ medications, e.g. nitroglycerin spray and epinephrine auto-injectors.

Conclusion

The LSP as a source for home medication information is not available for a large part of the ETZ patient population. Additionally, the quality of the information is suboptimal, mainly because medications are missing. This leads to patient safety risks that users should be aware of. Independent research on the quality of ICT infrastructures and software should continuously be performed to guarantee patient safety.

Table 1. Quality of LSP information.

 

Use of generic drugs in HIV: acceptance, adherence, quality of life and cost-effectiveness: the FUEL project

P. Oosterhof ^a*, M. van Luin ^b, J. Verdijk ^c, J. Gisolf ^b, A. Colbers ^c, R. van Crevel ^c and D. Burger ^c

^a OLVG, Amsterdam.

^b Rijnstate ziekenhuis, Arnhem.

^c Radboudumc, Nijmegen.

* Correspondence: p.oosterhof@olvg.nl.

Background

Combination antiretroviral therapy (cART) is used by an increasing number of HIV-infected individuals, for increasing lengths of time. As cART makes up a large proportion of chronic HIV-care costs, introduction of generic ART could lead to significant cost reduction. Generic substitution of cART has gained limited interest in The Netherlands so far. In our opinion, generic substitution in HIV-infected patients should be implemented with close monitoring, since life-long patient adherence is essential for continuous viral suppression. The objectives of the FUEL-project are to study (1)patient acceptance of generic HIV-medication substitution; (2)whether switching from branded products to generics influences adherence; (3)quality of life in HIV-infected patients switching to generic cART, and (4)cost-effectiveness of generic HIV-substitution.

Methods

All HIV-infected patients on branded nevirapine-extended-release (Viramune MVA®) and/or abacavir+lamivudine (Kivexa®) in the HIV treatment centers of Radboudumc and Rijnstate were included (February-August 2017). Patients were proposed to switch to generic products by both prescribers and HIV-nurses (“opt-out”). An electronic validated questionnaire was sent to study patient acceptance, self-reported drug adherence and quality of life (SF12). If patients refused to switch to generic, the reason was documented, but the patient continued participation in the study. After 3 and 12 months questionnaires will be repeated. This first analysis reports on the acceptance rates of the generic switch, baseline data for patients’ views on their current medication, and initial costs savings.

Results

A total of 36 patients were included in the study; demographic, disease and treatment characteristics of the respondents are shown in table 4. 35 of 36 patients (97%) agreed to switch to generic; the patient that refused gave as reason a bad previous experience. In the first 3 months of follow-up, three out of 35 (= 9%) patients switched back to branded medication due to perceived side-effects (all abacavir+lamivudine). Baseline results of questionnaires (n = 27) showed a high acceptance for the satisfaction about the HIV treatment: all patients scored maximal or sub-maximal satisfaction for 6 out of 10 questions. During this 3 months of follow-up drug expenses were 4670 € lower than in the 3 months prior to switch.

Conclusion

A pro-active policy for substituting HIV-infected patients on branded HIV-medication to generic was highly successful. First data show a high patient acceptance of treatment at baseline. Only 9% of patients switched back to branded medication within 3 months after substitution. Patients infected with HIV appear to be open for changing branded medication to generic.

 

Case series of compound heterozygous DPYD variant allele carriers and investigation of follow-up methods

C.A.T.C. Lunenburg ^a*, L.M. Henricks ^b, A.B.P. van Kuilenburg ^c, A.H.J. Mathijssen ^d, J.H.M. Schellens ^b,e, H. Gelderblom ^a, H.J. Guchelaar ^a and J.J. Swen ^a

^a Leiden University Medical Center.

^b The Netherlands Cancer Institute, Amsterdam.

^c Academic Medical Center, University of Amsterdam.

^d Erasmus MC Cancer Institute, Rotterdam.

^e Utrecht University.

* Correspondence: c.a.t.c.lunenburg@lumc.nl.

Background

Fluoropyrimidines are the cornerstone of cancer therapy. Up to 30% of treated patients experience severe toxicity (CTCAE grade ≥ 3). Prospective genotyping of variants in DPYD, the gene encoding the key metabolic enzyme dihydropyrimidine dehydrogenase (DPD), is recommended prior to treatment with fluoropyrimidines by the Dutch Medical Oncology guidelines, and increasingly applied in clinical practice in order to reduce toxicity. Currently, dosing guidelines from the Dutch Pharmacogenetics Working Group (DPWG) are available for four variants (DPYD*2A, DPYD*13, c.1236G>A, c.2846A>T). These guidelines can easily be applied in patients who are heterozygous carrier of a single variant. However, little is known about (dose reductions for) compound heterozygous patients (two different variants). The objective of our study is to draw attention to this subpopulation, explore databases for information regarding frequency and phasing of DPYD compound heterozygosity and to investigate suitability of (genetic) follow-up methods.

Methods

Several databases were searched for compound heterozygous patients including phasing details. In addition, seven patient cases were described. In four cases, extra material could be collected to execute additional genetic assays to identify phasing of the variants. Assays were Droplet Digital PCR, PacBio sequencing and a home-brew technique. Also, for all cases DPD enzyme activity was determined.

Results

Databases indicated that the frequency of compound heterozygous patients for four DPYD variants is ~0.2%. With additional genotyping assays it was possible to obtain phasing information for four patient cases. In other cases, incorrect material was available or extra material could not be collected. Three patients carried the variants in trans (on different alleles), one in cis (on a single allele). With phasing details known, one could determine a retrospective dosing advice. These were 25% or 50% depending on the DPYD variants and phasing details. The levels of DPD enzyme activity did not always match with the dosing guidelines.

Conclusion

Currently, it is not possible to apply DPWG guidelines for compound heterozygous DPYD variant allele carriers without phasing details of the variants, thereby exposing this subpopulation to an increased risk for severe fluoropyrimidine-induced toxicity. Of the additional genotyping assays to determine phasing, PacBio sequencing is the most sophisticated technique. When phasing is known, DPWG guidelines can be applied. However, DPD phenotyping may also be a suitable method to decide on dose recommendations for compound heterozygous patients.

 

The impact of dose and simultaneous use of acid reducing agents on the effectiveness of vemurafenib in metastatic BRAF V600 mutated melanoma: a retrospective cohort study

L.M. Knapen ^a*, R.H.T. Koornstra ^b, J.H.M. Driessen ^a, B. van Vlijmen ^c, S. Croes ^a, S. Schalkwijk ^c, A. Colbers ^c, W.R. Gerritsen ^c, D.M. Burger ^c, F. de Vries ^a and N.P. van Erp ^c

^a Maastricht Universitair Medisch Centrum+.

^b Rijnstate Ziekenhuis, Arnhem.

^c Radboud universitair medisch centrum, Nijmegen.

* Correspondence: lotte.knapen@mumc.nl.

Background

The current recommended fixed dose regimen for vemurafenib may result in both over- and underexposure which could affect treatment outcome. Some patients might be overexposed and benefit from a dose reduction. Other patients might be underexposed potentially due to simultaneous use of acid reducing agents (ARAs). The aims of this study were to determine the association between disease progression of metastatic BRAF V600 mutated melanoma and 1) dose reductions of vemurafenib, and 2) simultaneous use of vemurafenib and ARAs.

Methods

A retrospective cohort study was conducted using data from electronic patient records and pharmacy dispensing records of a Dutch academic hospital (March 2012-March 2016). Patients using vemurafenib as first-line treatment for melanoma were included. Cox regression analysis was used to estimate risk of disease progression with full-dose vemurafenib versus reduced dose and simultaneous use of vemurafenib and ARAs versus vemurafenib alone. Analyses were adjusted for age and sex.

Results

In total, 112 patients were included of which 55% developed disease progression on vemurafenib. The median progression-free survival was 6.0 months (95% CI 5.0-6.9). There was no increased risk of disease progression among patients requiring a dose reduction of vemurafenib, or with the simultaneous use of vemurafenib and ARAs versus vemurafenib alone. However, a tendency for disease progression was observed among patients who used full-dose vemurafenib and ARAs versus full-dose vemurafenib alone (adjusted Hazard Ratio (HRa) 2.37; 95% CI 0.97-5.76), which became statistically significant in a sensitivity analysis (HRa 4.56; 95% CI 1.51-13.75).

Conclusion

There was no association between the use of vemurafenib in a reduced dose or the simultaneous use of vemurafenib and ARAs and the risk of disease progression. However, patients using full-dose vemurafenib simultaneously with ARAs might have an increased risk of disease progression. Prescribing of ARAs to this group of patients should be reconsidered.

 

Pharmacokinetics of favipiravir during continuous venovenous hemofiltration in a critically ill patient with influenza

L.M.A. Favié ^a*, J. Murk ^b, A. Meijer ^c, A.L. Nijstad ^a, E.M. Van Maarseveen ^a and M.A. Sikma ^a

^a UMC Utrecht.

^b St. Elisabeth Hospital Tilburg.

^c National Institute for Public Health and the Environment (RIVM), Bilthoven.

* Correspondence: L.M.A.Favie@umcutrecht.nl.

Background

Favipiravir is a novel antiviral drug approved for influenza treatment in Japan. Little is known about the pharmacokinetics of favipiravir in critically ill patients, let alone patients receiving renal replacement therapy. This case report describes the pharmacokinetics of favipiravir in a 62 year-old patient with a severe influenza infection while receiving continuous venovenous hemofiltration (CVVH).

Methods

The patient was admitted suffering from respiratory insufficiency due to a severe community acquired pneumonia. Mechanical ventilation was started together with antibiotic treatment and oseltamivir, based on a throat swap positive for influenza virus. Microbiological investigations on bronchoalveolar lavage (BAL) fluid revealed an infection with influenza virus A(H1N1)pdm09. On day four, the patient developed acute kidney injury necessitating CVVH. Since there was no improvement in the patient’s clinical condition, experimental treatment with favipiravir was added; half of the recommended dose was given because the effects of CVVH on clearance were unknown. On day six, a sputum sample was still positive for influenza A(H1N1)pdm09 virus. There was a steady decline in the patient’s clinical condition with ongoing septic shock and multi organ failure. Eleven days after admission, supportive therapy was stopped and the patient died. Multiple blood samples were collected before and after administration of favipiravir on day three of treatment. Additionally, CVVH samples were collected five hours after administration: pre-filter, post-filter and ultrafiltrate. The samples were analyzed and PK parameters were estimated.

Results

Favipiravir PK parameters (table 1) show that the maximum plasma concentration in our patient during steady state was much lower than in healthy volunteers and that the elimination rate was about three times higher than in healthy volunteers. After both the loading dose and maintenance dose, similar concentrations of favipiravir pre-filter, post-filter and in the ultrafiltrate were measured. Membrane passage was represented by the sieving coefficient (S), found to be 1.1, indicating free membrane passage of favipiravir. However, the total amount of fluid extracted using CVVH is low compared to the patient’s apparent volume of distribution for favipiravir. After each administration, < 1% of the administered dose is removed via the ultrafiltrate.

Conclusion

Pharmacokinetics of favipiravir in a critically ill patient differs from healthy volunteers. An increased elimination rate and high distribution volume may lead to suboptimal blood concentrations. Although favipiravir freely passes the CVVH membrane, this might not result in a clinically relevant contribution to total clearance. Therefore, a priori dose reduction of favipiravir does not appear to be necessary.

Table 1. Pharmacokinetic parameters of favipiravir.

 

Assessing adherence to etanercept in RA patients using etanercept serum trough concentrations and patient self-report

E. Vogelzang ^a, R.C.F. Hebing ^a, M.T. Nurmohamed ^a, A.W.R. van Kuijk^aa, J.W.F. Kruijff ^b, M.J. 'l Ami ^a, C.L.M. Krieckaert ^a and G. Wolbink ^a

^a Reade, Amsterdam.

^b Faculty of Science, Universiteit van Amsterdam, faculty of science, Amsterdam.

Correspondence: research@reade.nl.

Background

The EULAR recommendations for the management of rheumatoid arthritis (RA) update 2013 contained the following research question: “How good is patient adherence to biological agents and can lack of adherence be related to loss of efficacy?”. Limited studies are published in which adherence of RA patients treated with biological disease-modifying anti rheumatic drugs (bDMARDs) has been assessed. Previous studies regarding adherence of RA patients treated with etanercept did not take into account that patients are instructed to temporarily discontinue bDMARDs therapy during e.g. a serious infection. In addition, etanercept concentrations have never been utilized in determining adherence to etanercept in RA patients.

Objective

The aim of our study was to determine the percentage of non-adherent RA patients treated with etanercept, using etanercept concentrations and patient self-report, and to assess the relationship between adherence and clinical outcome during 52 weeks.

Methods

Non-adherence was defined as an etanercept trough concentration <0.1ug/mL at least once and no valid/medical reason to miss an etanercept dose. In this retrospective cohort study patients visited our clinic at baseline and 4, 16, 28, 40 and 52 weeks after initiation of etanercept treatment. At baseline and following visits disease activity score of 28 joints (DAS28) was calculated and blood was drawn to measure etanercept concentrations. During each visit patients were asked if they missed an etanercept dose, at which date and for which reason. Remission was defined as DAS28 < 2.6 at least for two consecutive visits. Low disease activity (LDA) was defined as DAS28 < 3.2 during a minimal of two consecutive visits.

Results

In total 292 patients were included. Mean age was 53 years (SD 12.7), 82% was women and median disease duration was 8 years (IQR 3-16). In total 24 patients had an etanercept concentration <0.1ug/mL. Most patients had a medical reason to miss an etanercept dose, see figure 6. Only ten patients (3.4%) were non-adherent during the follow-up of 52 weeks, see figure 7. Of the adherent patients 82 out of 282 (29%) reached LDA versus 1 out of 10 non-adherent patients. A total of 127 of 282 (45%) adherent patients achieved MDA versus 3 out of 10 non-adherent patients.

Conclusion

Most patients had a medical reason to miss an etanercept dose. The percentage of patients who are non-adherent to etanercept therapy is very low (3.4%).

Figure 6: Non-adherence reasons.

Figure 7. Cumulative percentage non-adherence.

 

Preliminary results from the PREPARE study: enrolled patients are interested in pharmacogenomics

C.H. van der Wouden *, J.J. Swen and H.J. Guchelaar

LUMC, Leiden.

* Correspondence: c.h.van_der_wouden@lumc.nl.

Background

The PREemptive Pharmacogenomic testing for Preventing Adverse drug REactions (PREPARE) study aims to implement pre-emptive pharmacogenetics (PGx) testing of a panel of pharmacogenes in seven European countries to guide pharmacotherapy. The primary objective is to quantify the collective clinical utility of the panel. Secondary outcomes include cost-effectiveness and interest and knowledge in PGx. We hypothesize that PGx guided prescribing will result in a 30% reduction in clinically relevant adverse drug events. Here we provide preliminary results regarding patients’ interest and knowledge in PGx at baseline.

Methods

PREPARE is a multi-center, open, randomized, cross-over implementation study (n = 8,100). Countries have been randomised to start with either PGx-guided prescribing (study arm) or standard of care (control arm) for 18 months. After this period, the opposite strategy will be implemented for 18 months. Patients of any ethnicity who are at least 18 years old and receive a first prescription for a drug for which a DPWG guideline is available, as part of routine care, are eligible to participate. For patients within the study arm, their results will be: 1) recorded in the (electronic) medical record and 2) provided to the patient in the form of a plastic card. Genetic results and DPWG guidelines can be used by physicians and pharmacists (HCPs) to guide the dose and drug selection for the drug of inclusion. Four questions regarding PGx interest and knowledge were collected at baseline and answered on a Likert-like scale (ranging from Disagree (1 and 2) to Agree (3 and 4)).

Results

Currently, 258 patients have been enrolled in the control arm and 386 patients in the intervention arm. Baseline data on PGx interest and knowledge is available for 536 patients. 19.8% (n = 106) reported being “familiar with PGx”, whereas 73.5% (n = 394) reported they would “like to learn more about PGx”. 85.4% (n = 458) believed “pharmacogenetics testing contributes to reducing the frequency and severity of ADEs”. 86.8% (n = 465) reported they would “like to be tested on their PGx profile before a drug prescription”.

Conclusion

Even though patient familiarity with PGx is low, patient interest and belief in PGx guided prescribing is high. PGx interest may be higher in this sample than in the general population.

 

A nationwide cross-sectional survey on knowledge, experience and attitudes of pharmacy students towards pharmacogenetics

P.C.D. Bank *, J.J. Swen and H.J. Guchelaar

LUMC, Leiden.

* Correspondence: p.c.d.bank@lumc.nl.

Background

Pharmacy students are our future healthcare professionals and are bound to come in to contact with pharmacogenetics (PGx) in their career. Currently it is unknown whether they feel informed about PGx and feel adequate to handle PGx data. The goal of this study was to benchmark knowledge and attitude of pharmacy students towards PGx and PGx-testing and compare the results to their previously surveyed practicing colleagues.

Methods

All pharmacy students in The Netherlands were invited by e-mail to participate in a web-based questionnaire consisting of 28 questions using survey tool NetQ. Statistical analysis was performed using multivariate regression using SPSS software.

Results

Out of the 824 invited students, 148 individuals (18.0%) completed the questionnaire. All responders believed in the concept of PGx and had high expectations towards PGx testing to prevent wrong drug therapy, improve efficacy and reduce toxicity of pharmacotherapy. The majority of the students (96.6%) had received some form of education on PGx, but only 12.8% felt adequately informed and 27.7% felt qualified to receive, interpret PGx-test results and advise patients. Responses of the students were compared to results of practicing pharmacists in The Netherlands. Differences between previously surveyed practicing pharmacists (n = 667) and pharmacy students were mainly observed in the use of information, feeling qualified to recommend PGx-testing to predict efficacy of pharmacotherapy and the possible negative psychological impact of unfavorable PGx test results.

Conclusion

Despite having followed education on PGx as part of their curriculum pharmacy students do not feel informed about PGx(-testing). More hands-on experience with PGx in the curriculum is required to fill the perceived knowledge gap among future pharmacists.

 

The use of whole-exome sequencing for pharmacogenetics

M. van der Lee ^a*, S.Y. Anvar ^b, M. Kriek ^b, G.W.E. Santen ^b, H.J. Guchelaar ^a and J.J. Swen ^a

^a LUMC, Dep. Clinical pharmacy and toxicology, Leiden.

^b LUMC, Dep. Human Genetics, Leiden.

* Correspondence: m.van_der_lee@lumc.nl.

Background

Whole Exome Sequencing (WES) is a widely used diagnostic tool in clinical genetics. It provides information on the presence of genetic variations within the entire protein coding sequences of an individual. This offers the opportunity to retrieve pharmacogenetic variants from WES data. However, there is no current bioinformatics tool to identify pharmacogenetics variants from WES data. To date, WES data from over 2000 individuals, most of whom form child-parent trio’s, are available. The objective of this study is to build a bioinformatics pipeline in order to extract pharmacogenetic information from existing WES data.

Methods

We created a bioinformatics pipeline to extract pharmacogenetic variants for the entire panel of genetics variants for which a gene-drug interaction is described by the DPWG (Dutch Pharmacogenetic Working Group). First, genetic variants we identified based on the human reference genome GRCh37. By having access to child-parent trio data, we could then resolve the phasing of identified variants and therefore separate variants of paternal or maternal origin for improved genotyping.The panel of variants extracted from the WES data included a total of 45 variants in 13 genes which are linked to 43 drugs. Variants resulting in a non-extensive metabolizer phenotype are classified as actionable.

Results

In this pilot study, WES data from 166 individuals were processed. For all genetic variants, the observed allele frequencies were in concordance with the literature. Notably, 91% of the individuals carried at least one actionable phenotype, with 16.8% carrying 4 or more. Most actionable phenotypes were observed in CYP2B6 (57%), whereas the least was observed in UGT1A1 with only 2% carrying an actionable phenotype. Six variants were located in intronic or promoter regions and, therefore, were not covered by the WES. This led to a lack of information to screen for CYP2C19*17, CYP3A5*3, UGT1A1, HLA-A and HLA-B. Moreover, 1.2% of identified variants could not be phased, leading to unclear genotypes. For 3 individuals this meant that the phenotype for CYP2D6 could not be called.

Conclusion

Using the bioinformatics pipeline it is possible to extract pharmacogenetic data from routinely collected WES data. In a next step, pharmacogenetic information for all 2000 individuals will be extracted and linked to hospital prescribing data to assess clinical impact. Additionally, the genotypes and phenotypes will be added to the electronic health care records of the patients to aid pharmacogenetics based prescribing.

 

Empiric antibiotic treatment in patients hospitalized with non-severe community-acquired pneumonia in The Netherlands

I. van Heijl ^a*, V.A. Schweitzer ^b, C.H. van Werkhoven ^b, P.D. van der Linden ^a, C.H.E. Boel ^b, J.W. Dorigo-Zetsma ^a and M.J.M. Bonten ^b

^a Tergooi, Hilversum.

^b UMCU, Utrecht.

* Correspondence: Ivanheijl@tergooi.nl.

Background

According to guideline recommendations empiric antibiotic treatment of community-acquired pneumonia (CAP) is based on disease severity. For the categorization of CAP as mild, moderate-severe or severe, the Dutch national guideline provides three classification systems: pragmatic classification (ambulatory care = mild, non-ICU ward = moderate-severe, or ICU admission = severe), the Pneumonia Severity Index (PSI) score, or the CURB-65 score. Recommended treatments are: amoxicillin or doxycycline for mild, penicillin or amoxicillin for moderate-severe and moxifloxacin/levofloxacin, or penicillin plus ciprofloxacin, or a cephalosporin plus erythromycin for severe CAP. We quantified guideline adherence for the empiric antibiotic treatment of CAP according to the three classification systems in CAP patients hospitalized to non-ICU wards in 2015-2016.

Methods

Data were used from the ongoing CAP-PACT trial (ClinicalTrials.gov Identifier: NCT02604628), performed in 11 Dutch hospitals since 2015. The study investigates the effects of antibiotic stewardship interventions in CAP patients admitted to non-ICU wards. Inclusion criteria are age ≥18 years, admission to a non-ICU ward and empiric treatment for CAP. Empiric treatment was defined as antimicrobial therapy given on the day of admission and was compared to the guideline recommendation. Over- or under-treatment is defined as a regimen with a broader or narrower spectrum, respectively, compared to the guideline-recommended treatment.

Results

From November 2015 until October 2016 1,174 patients were included, with a mean age of 70 years (sd 15.5) and 626 (53.3%) were male. 30-day mortality was 6.4%. Amoxicillin/penicillin monotherapy was the most frequently prescribed empirical antibiotic therapy (table 6). When comparing the three classification methods, the empiric treatment was guideline-adherent in 344 (29.3%) using the pragmatic classification, 394 (33.6%) with the PSI-score and 422 (35.9%) with the CURB-65 score. Under-treatment occurred in 90 (7.7%) based on the PSI and 131 (11.2%) with the CURB-65 score. Over-treatment occurred 830 (70.7%) based on the pragmatic, 690 (58.8%) with the PSI and 621 (52.9%) with the CURB-65 score.

Conclusion

Among patients hospitalized with CAP at a non-ICU ward, adherence to guideline recommendations for empiric antibiotic therapy ranges, depending on the severity classification system used, from 29.3% to 35.9%. 82.6% to 100.0% of non-adherence included over-treatment with broad-spectrum antibiotics. These findings suggest that there is a need for Antibiotic Stewardship programs to increase guideline adherence.

Table 6. Empiric antibiotic treatment in hospitalized patients with CAP.

 

Differences and accuracy of medication dispensing records

E.B. Uitvlugt ^a*, W.L. Chung ^a, B.J.F. Van den Bemt ^b, P.M.L.A. Van den Bemt ^c and F. Karapinar- Çarkit ^a

^a OLVG, Amsterdam.

^b Sint Maartenskliniek, Nijmegen.

^c ErasmusMC, Rotterdam.

* Correspondence: e.uitvlugt@olvg.nl.

Background

Medication reconciliation is an important tool to reduce medication errors. Electronic sources are implemented in many regions of the world to obtain up-to-date medication dispensing records (MDRs). The Netherlands is using LSP (literally “National Switching Point”) as an electronic source. Little is known about the reliability and accuracy of LSP compared to the data of the community pharmacy. The aim of this study is to assess differences between MDRs from two LSP sources (from different prescribing systems) and MDRs obtained by fax from community pharmacies.

Methods

The information of all three sources (two LSP sources and fax) was considered to be the best possible dispensing record (BPDR). Each individual source was compared to the BPDR regarding the drug (including dose, frequency and active stop date) and allergies. The BPDR was also compared with patient’s actual medication use during medication reconciliation. Descriptive analysis was used.

Results

90 patients were included and 1053 medications and 45 allergies were analysed. The fax contained the least differences (10.3%), followed by LSP source 1 (14.0%) and LSP source 2 (18.1%). During medication reconciliation 15.8% of the items on the BPDR did not reflect patient’s actual medication use. More than 80% of the patients had at least one discrepancy in their MDRs.

Conclusion

The fax from the community pharmacy has the least differences compared to the MDRs from the two LSP sources. The MDRs cannot be considered as complete and additional information from the patient is still needed to get insight in a patient’s actual medication use.

 

Costs of ownership of ready- to- administer pre-filled sterilized syringes in a Dutch hospital: a cost minimization analysis

K.H.M. Larmene- Beld ^a*, J. Spronk ^a, K. Taxis ^b and M.J. Postma ^b

^a Isala, afdeling klinische farmacie, Zwolle.

^b University of Groningen. Unit of PharmacoTherapy, -Epidemiology & -Economics.

* Correspondence: k.h.m.beld@isala.nl.

Background

Preparation errors occur frequently during conventional multiple step preparation of parenteral drugs at the bedside, causing potential adverse drug events (ADEs), which can be a burden to the patient and involves high costs for the national healthcare system. The use of ready-to-administer (RTA) pre-filled sterilized syringes (PFSS) produced by the hospital pharmacy can prevent a significant part of preparation errors and reduces the risk of bacteremia due to contamination of the intravenous fluid. This research aims to compare the total cost of the conventional preparation methods (CPM) with the PFSS method.

Methods

In the analysis, costs related to the preparation of the drugs, bacteremia due to contamination, ADEs as a result of medication errors and wastage of syringes were taken into account. Annual costs in a general Dutch hospital were consistently calculated. Three scenarios were investigated: (i) all preparations CPM (864.246 administrations per year); (ii) all preparations as PFSS; and (iii) 200.000 PFSS and the remaining part CPM (reflecting a transition state as currently present). Deterministic and probabilistic analyses are performed.

Results

The first scenario shows higher annual costs at €10.862.609 compared to the second scenario. The current situation (third scenario) already shows savings of €2.420.545 compared to the old situation (first scenario). Sensitivity analyses revealed that cost savings of PFSS were mainly the result of decreased risks of medication errors and contamination of intravenous fluids. Extrapolating these results nationwide indicates potential savings over €300 million if only PFSS were used.

Conclusion

The use of PFSS prepared at the hospital pharmacy yielded cost-savings compared to conventional preparation at the bedside in the Dutch hospital.

 

Towards effective and safe antiretroviral therapy during pregnancy 

S.J. Schalkwijk *, R. Greupink, A. Colbers, R. Heine, F.G.M. Russel and D.M. Burger

Radboud university medical center, Nijmegen.

* Correspondence: stein.j.schalkwijk@radboudumc.nl.

Background

The development of effective and safe antiretroviral therapy for the prevention of HIV mother-to-child transmission is not without challenges. Pregnant women are generally excluded from pre-marketing clinical trials. Consequently, no or limited data are available to make well-grounded treatment recommendations for pregnant HIV-positive women. Our aim was to study the pharmacokinetics and safety of antiretroviral agents in pregnant women and their unborn children and the translation of these findings to the clinic.

Methods

Clinical and preclinical studies. Specific focus was given to the combined use of clinical and preclinical (in vitro and ex vivo) data by means of modeling and simulation.

Results

In clinical trials the exposure during pregnancy was reduced for maraviroc (28%), efavirenz (8%), and rilpivirine (45%). Abacavir exposure was not affected by pregnancy and etravirine exposure was increased, by 34%. Using modeling and simulation, we predicted that lowered-dose efavirenz is adequate during pregnancy, potentially reducing costs and increasing universal access to treatment. Also, we predicted that fetal exposure to darunavir is around the HIV EC₅₀ with standard dosing, and that higher maternal doses could provide fetal exposures higher than the EC₅₀. Additionally, in an ex vivo human cotyledon perfusion model we observed that dolutegravir placental transfer was about 60%, indicating sufficient fetal exposure for prophylaxis.

Conclusion

Overall, the exposure observed during pregnancy following standard adult dosing was adequate and safe, except for rilpivirine (higher doses are explored). Modeling and simulation facilitated certain treatment recommendations that would not have been feasible otherwise, such as the impact of maternal dosing on fetal exposure and implications for MTCT.

 

Drug interactions between tyrosine kinase inhibitors and St. John's Wort unnoticed in clinical practice

M.S.S. Sjak Shie *, A. Sobels, P. de Wolf and L. Binkhorst

Apotheek Haagse Ziekenhuizen - HagaZiekenhuis, Den Haag.

* Correspondence: M.SjakShie@hagaziekenhuis.nl.

Background

Many tyrosine-kinase inhibitors (TKIs) have been introduced in (hemato-)oncology recently. Most TKIs are substrate for CYP3A4 and P-glycoprotein (P-gp). These pharmacokinetic characteristics make TKIs susceptible for clinical relevant drug interactions as with St. John’s wort (SJW). SJW constituents induce CYP3A4 and P-gp expression via PXR. This can result in lower TKI exposure, leading to treatment failure. SJW (composed of dried flower leafs of hypericum perforatum) is available as over the counter herbal medicine for symptoms of depression and anxiety. Depression is more prevalent in patients with cancer than in the general population but data on the use of SJW in patients treated with TKIs are scarce. We evaluated whether TKI-treated patients who use SJW would be identified as patients at risk for clinical relevant pharmacokinetic drug interactions in our clinic.

Methods

First we checked whether our medication reconciliation protocol identifies the use of SJW. We also checked whether SJW is listed in medication lists. Secondly, we searched for literature to identify which TKI interacts with SJW (PubMed). In addition, we checked which TKIs are substrate for CYP3A4 and/or P-gp using databases from KNMP Kennisbank, Farmacotherapeutisch Kompas, and MICROMEDEX and if an interaction was mentioned (databases and http://www.drugs.com/). At last we checked whether our electronic prescribing system (HIX 6.0 standard content) generates alerts when TKI are prescribed in combination with SJW.

Results

According to the medication history database of HIX and Pharmacom SJW is not used in our population. Our medication reconciliation protocol does specifically address herbal medicine, particularly SJW. Data on pharmacokinetic interactions are shown in Figure 1. The effects of SJW on imatinib pharmacokinetics have been studied and there is strong evidence for a significant pharmacokinetic drug interaction. However, data for other TKIs are lacking. Electronic prescribing in HIX results in an alert in the case of imatinib in combination with SJW. For other TKIs, like nilotinib and osimertinib, no alert is generated although a significant pharmacokinetic interaction can be expected.

Conclusion

We conclude that most TKIs are probably at risk for a clinical relevant drug interaction when used in combination with SJW, which may remain unrecognized in our setting. Creating more awareness among patients, pharmacy technicians, pharmacists and oncologists about a potential drug interaction between TKIs and SJW, and tuning our electronic prescribing system to yield an alert when co-prescribed, can improve medication safety and treatment efficacy for patients treated with TKIs.

 

The emergency department pharmacist for improvement of under- and overtreatment with medication: an interim analysis

R.N. Rahman ^a*, B. Nikolik ^b, S.C.E. Klein Nagelvoort-Schuit ^c, M.J.A. Janssen ^b, F. Karapinar-Çarkit ^b and P.M.L.A. van den Bemt ^a

^a Department of Hospital Pharmacy, Erasmus MC, Rotterdam.

^b Department of Hospital Pharmacy, OLVG, Amsterdam.

^c Department of Emergency Medicine, Erasmus MC, Rotterdam.

* Correspondence: r.rahman@erasmusmc.nl.

Background

Medication related problems (MRPs) can result in a substantial number of hospital admissions in The Netherlands. An important risk factor is polypharmacy, and associated under- and overtreatment with medication. Medication reviews performed by emergency department (ED)-pharmacists may improve under- and overtreatment with medication in patients admitted due to MRPs.

Objective

To study the effect of medication reviews by ED-pharmacists on under- and overtreatment in ED-patients who are hospitalised due to MRPs.

Method

A prospective multicenter controlled intervention study was carried out. Patients admitted to the hospital after visiting the ED due to an MRP were included. Control patients were patients acutely admitted for another reason than an MRP; they received routine care. For intervention patients, ED-pharmacists assisted in recognizing MRPs and performed medication reviews. Primary outcome was the mean number of interventions in medication regarding under- and overtreatment (using the PCNE classification) after medication review performed by ED-pharmacists and by routine care. The difference between groups was analysed by the two sample t-test.

Results

64 intervention and 71 control patients were included. The mean number of interventions per patient regarding under- and overtreatment in control patients was 0.6 (SD ± 0.9) versus 2.3 (SD ± 2.1) in intervention patients (p = < 0.001).

Conclusion

This interim analysis showed that medication reviews performed by ED-pharmacists contribute to additional interventions regarding under- and overtreatment in patients hospitalised due to a medication related problem. The final study results are needed for more robust conclusions.

 

Follow-up study: Drug interactions in parenteral chemotherapy

J.E. van Citteren *, A. Keyany, M.A.J.P.J Lenoir-van Berkel and K.A. Simons-Sanders

ZiekenhuisApotheek Midden-Brabant, Tilburg.

* Correspondence: j.vancitteren@etz.nl.

Background

Parenteral chemotherapy can cause potential dangerous drug interactions with other medication in oncology patients. In literature an incidence of 13-40% is described. The purpose of this study is to investigate the incidence and clinical relevance of potential drug interactions between home medication and parenteral chemotherapy at Elisabeth-Tweesteden hospital (ETZ).

Methods

Interactions between parenteral chemotherapy and home medication of oncology patients were studied in the outpatient pharmacy at ETZ. Since medication monitoring was not possible in the hospital drug prescription system, the parenteral chemotherapy was entered in outpatient pharmacy drug prescription system (Pharmacom). In this study, patients with a first chemotherapy treatment were included if they have given consent to share their chemotherapy treatment with their community pharmacy. Furthermore, patients must also have given permission to their pharmacy to share information about their medication with other care institutions. If both consents were present then chemotherapy medication were entered and possible drug interactions with home medication detected. Potential drug interactions were determined and checked by a pharmacist for clinical relevance according to national guideline (G-standaard).

Results

In a seven month period between January and August in 2017 a total number of 211 patients with a first parenteral chemotherapy were included. Patients used on average 5 (standard deviation of 3.5) different medications at home. A total of 29 potential drug interactions were found with a maximum of one interaction per patient. Of these interactions, 16 potential drug interactions were clinically relevant. For 14 interactions contact was taken with national thrombosis center and in 2 interactions a doctor was informed. For an overview of these results, see table 1.

Conclusion

An incidence of 13.7% potential drug interactions in 211 patients between their first parenteral chemotherapy and home medication was observed. Of these 29 potential interactions, 16 (7.6%) were clinically relevant and action had to be taken.

 

Development of a clinical prediction model for an INR ≥ 4.5 in hospitalized patients using vitamin K antagonists

A.R. Dreijer ^ab*, J.S. Biedermann ^ab, J. Diepstraten ^b, A.D. Lindemans ^a, M.J.H.A. Kruip ^a, P.M.L.A. Van den Bemt ^a and Y. Vergouwe^{3 a}

^a Erasmus MC, Rotterdam.

^b Reinier de Graaf ziekenhuis, Delft.

* Correspondence: a.dreijer@rdgg.nl.

Background

Vitamin K antagonists (VKAs) are frequently used medications in the prevention and treatment of thromboembolic disease. However, the use of VKAs increases the risk of bleeding complications. The risk of bleeding is related to the international normalized ratio (INR). An INR ≥ 4.5 is a marker for bleeding risk because above 4.5 the risk of bleeding sharply increases. Many factors, such as dietary intake of vitamin K, interacting drugs and comorbidities affect the risk of bleeding. Existing models to predict bleeding complications are not applicable for electronic clinical decision support and do not concern the general hospitalized population. Therefore, we aim to develop a model predicting the risk of an INR ≥ 4.5 during hospital stay in a medical or surgical ward, for adult patients treated with VKAs, based on risk factors that are electronically collected during routine care.

Methods

We analyzed a cohort of patients admitted to a tertiary hospital. We included admissions of adult patients between 2006 and 2010 treated with VKAs. Bleeding risk was operationalized as an INR value ≥ 4.5 during VKA treatment. Multivariable logistic regression analysis was used to assess the association between potential predictors and an INR ≥ 4.5. Model validity was studied in an independent cohort of patients from the same hospital, admitted between 2011 and 2014.

Results

We identified 8,996 admissions of patients treated with VKAs. There were 1507 admissions (17%) with an INR ≥ 4.5. The final model included the following predictors for an INR ≥ 4.5: gender, age, an INR ≥ 4.5 during a previous admission, type of VKA, type of ward, concomitant medication and several biochemical parameters. Temporal validation showed a c statistic of 0.71.

Conclusion

We developed and validated a clinical prediction model for an INR ≥ 4.5 in patients admitted to medical or surgical wards who are treated with VKAs. The model includes factors that are collected during routine care and are extractable from electronic patient records, enabling easy use of this model to predict an increased bleeding risk in clinical practice.

 

Medication reconciliation at the preoperative screening by pharmacy technicians compared to anaesthesiologists

M.M. Ebbens ^a*, K.B. Gombert-Handoko ^a, E.J. Wesselink ^b and P.M.L.A. Van den Bemt ^c

^a Leiden University Medical Center.

^b Zaans Medical Centre, Zaandam.

^c Erasmus University Medical Center, Rotterdam.

* Correspondence: marieke.ebbens@gmail.com.

Background

Medication reconciliation is a safety intervention aimed at reducing the risk of medication discrepancies. Mekonnen et al showed in a systematic review that pharmacy-led medication reconciliation is more effective to reduce medication discrepancies. For elective surgery patients medication reconciliation takes place at the preoperative screening (POS). For efficiency reasons, patients with two or more medications are seen by a pharmacy technician in the Leiden University Medical Center (LUMC). For patients who use one or none medications the anaesthesiologist performs medication reconciliation. To evaluate if this method is safe, we repeated the medication reconciliation at admission to evaluate if the medication is correctly noted in the patients electronic record.

Method

An observational prospective study was performed. To evaluate the medication reconciliation process at the POS we performed medication reconciliation at admission on four surgical wards in the LUMC. All patients between October 17^th 2016 and August 29 2017 that had been to the POS, were admitted for at least 24 hours, ≥ 18 years and were able to participate in a second medication reconciliation interview were included. The medication reconciliation result at admission was compared to the actual medication in the patients’ medical record, any unintended medication discrepancy was defined as a medication error at admission (MEA). We compared the percentage of patients with at least one MEA that were seen by a pharmacy technician at the POS to the percentage of patients with at least one medication error at admission seen by the anaesthesiologist.

Results

Of all 1020 patients eligible for inclusion, 367 patients were included in the study. Of the 367 included patients 167 (45.4%) had at least one MEA. 201 patients were seen by the pharmacy technician at the POS, 166 were seen by the anaesthesiologist. The percentage of patients with at least one MEA was 44% respectively 47%. Most of the medication errors at admission occurred because of changes in medication between POS and the actual admission.

Conclusion

Medication errors at admission are still occurring often despite medication reconciliation at the preoperative screening. Medication errors occur as often in patients seen by the anaesthesiologist as seen by the pharmacy. Therefore the current method can be maintained, however an extra intervention prior to admission is necessary to prevent medication errors at admission.

 

Management of chronic medication in the preoperative period; what are we missing?

K. van Soest ^a*, J.K.M. Jansen ^a, K.P.G. Hurkens ^a, B.P.C. van Oijen ^a and P.M. van der Kuy ^b

^a Zuyderland Medisch Centrum, Sittard-Geleen.

^b Erasmus Medisch Centrum, Rotterdam.

* Correspondence: k.vansoest@zuyderland.nl.

Background

Advances in both surgical and anesthetic techniques result in more patients undergoing operative procedures. The number of medicines per patient increases with age. Patients who take medicines before admission are at an increased risk of developing a postoperative complication. During preoperative assessments clinicians must reconsider chronic medication and adjust when necessary. Preoperative discontinuation-requiring medication (DRM) is defined as medication that should be stopped during the preoperative period because of increased risk of surgical complications. Not all DRMs are identified in the preoperative period. The purpose of this study was to identify the missing DRMs in the preoperative assessment.

Methods

Medication data from the national electronic health system (Landelijk Schakelpunt (LSP)) are collected prospectively from patients who are checked routinely at the preoperative screening in Zuyderland Medical Centre during a 4 month period. Medication is considered to be possibly “active” when the end-date of the prescription is maximal 45 days before retrieving the information from the LSP. Based on literature a list of DRMs is defined. To identify missing DRMs during the preoperative assessment, information from the LSP was compared with the actual medication known in the hospital prescription system.

Results

A total of 3,421 patients was seen at the preoperative screening in the Zuyderland MC during the 4 month study period. Information regarding current medicines was available in the LSP for 2,650 patients (77%). The mean number of medicines taken by patients was 5.7 and 1,749 (66%) patients used at least 1 DRM (average 2.3). The majority of these DRMs were cardiovascular (27%) or anticoagulation medicines (25%). Thirty-one DRMs would not have been identified at the preoperative screening when the LSP was not used. These 31 medicines were taken by 24 patients and included anti-coagulation (35%), cardiovascular (20%), non-steroidal anti-inflammatory drugs (16%) and diabetes-regulating medicines (16%).

Conclusion

Delaying surgery or especially development of postoperative complications due to continuation of a DRM can be reduced by reviewing the actual medication in the preoperative assessment. Information from the LSP is useful for this review. ICT solutions (e.g. automated clinical rules) can be used to identify the DRMs. No conclusions can be drawn at this stage regarding delay of surgery or effects on surgery complications. The mean number of DRM used by individual patients (2.3 in patients taking DRM) seems relevant enough to support the use of standardized clinical rules identifying these patients.

 

Prospective validation of a risk prediction model for medication errors at admission after preoperative screening

M.M. Ebbens ^a*, S.A. Van Laar ^a, E.J. Wesselink ^b, K.B. Gombert-Handoko ^a and P.M.L.A. Van den Bemt ^c

^a Leiden University Medical Center.

^b Zaans Medical Centre, Zaandam.

^c Erasmus University Medical Center, Rotterdam.

* Correspondence: marieke.ebbens@gmail.com.

Background

Pharmacy led medication reconciliation is one of the most effective interventions to prevent clinically relevant medication errors at admission (MEA) [Mekonnen]. For elective surgery medication reconciliation is often performed at the preoperative screening (POS). However, since the time lag between POS and admission changes in medication may occur leading to MEA. In a previous study we developed a risk model to predict patients with MEA [Ebbens, 2017]. The number of medications, cardiovascular comorbidity and respiratory comorbidity were included in this risk model. This study was performed to validate this risk model.

Methods

A prospective observational study was carried out on four surgical wards of the Leiden University Medical Center (LUMC). Patients were included at admission if admitted for at least 24 hours, ≥ 18 years, visited the POS and able to participate in a second medication reconciliation interview. From October 17^th 2016 to August 29^th 2017 medication reconciliation was repeated within 48 hours of admission. The result of the medication reconciliation at admission was compared to the actual medication in the electronic patient record and discrepancies were discussed with the attending physician. If a discrepancy was unintended it was defined as an MEA. The sensitivity and specificity of the prediction model in this patient population was assessed. Also, the prediction tool was tested composing a ROC curve of the result of the prediction model against actual occurrence of an MEA. Finally risk factors for MEA were analysed using logistic regression.

Results

Of the 1020 patients screened, 368 patients were included. Of these patients 167 (45.4%) had at least one MEA. The specificity of the prediction model was 65% and the sensitivity was 40%, with a positive predictive value of 47% and a negative predictive value of 59%. The AUC of the ROC curve of the original data was 0.752, compared to 0.535 of this study. The two AUC’s were statistically different and the prediction model hardly predicted a medication error better than chance. With univariate logistic regression number of medications at the POS and having difficulties with climbing two stairs were identified as risk factors for medication errors.

Conclusion

The prediction model positively predicts 47% of the MEA. Therefore this model is not suitable to identify high risk patients. An intervention for all patients is still needed to prevent medication errors at admission after medication reconciliation at the POS, until a more precise model can be constructed.

 

Somatic screening in outpatients with bipolar disorders. A MOPHAR implementation study

M. Simoons ^abc*, H. Mulder ^ad, H.G. Ruhé ^bcef and E.N. Van Roon ^cg

^a Wilhelmina Ziekenhuis Assen.

^b UMC Groningen.

^c Rijksuniversiteit Groningen.

^d GGZ Drenthe, Assen.

^e Warneford Hospital Oxford.

^f Radboudumc Nijmegen.

^g Medisch Centrum Leeuwarden.

* Correspondence: mirjam.simoons@wza.nl.

Background

Somatic complications account for the majority (about 60%) of the 13-30 years shorter life expectancy in psychiatric patients compared to the general population. A structured monitoring program for somatic comorbidities may reduce the risk for these complications. We aimed to investigate the implementation of somatic screening at intake of the Monitoring Outcomes of Psychiatric Pharmacotherapy (MOPHAR) monitoring program at the outpatient department for bipolar disorders at Mental Health Services (MHS) Drenthe.

Methods

We investigated the influence of the implementation of the MOPHAR monitoring program on the frequencies of anthropometric measurements and laboratory tests in a cohort of patients with bipolar disorders. Included patients were over 18 years and already in treatment when they entered the study with a renewed MOPHAR-intake. We compared the patient records on somatic screening around the MOPHAR-inclusion (prospectively) to those around the intake at start of treatment (retrospectively). Primary outcome was the rate of the 25 standard anthropometric and laboratory measurements obtained as described in the MOPHAR monitoring protocol. Secondary outcome was the rate of patients with aberrant test results after implementation of MOPHAR.

Results

In total 155 patients were included. Of all protocolled measurements, on average 21.3 ± 6.8 (mean ± SD; 85.2%) standard anthropometric and laboratory measurements were obtained per patient. For 92 patients (59.4%), all 25 measurements were obtained. In comparison, a median of 3.0 (12.0%; range 0-19) measurements were performed around the original intake at the start of treatment (p < 0.0001). In 136/154 (88.3%) an increased waist circumference was determined; the highest rate of aberrant test results after the MOPHAR screening. Most aberrant values in the remaining parameters were also found in metabolic parameters: BMI (73.4%), blood pressure (34.4%), glucose (20.0%) and lipids (22.6-33.1%). The metabolic syndrome according to the National Cholesterol Education Program Adult Treatment Protocol III (NCEP ATP-III) criteria was present in 47.8% of the patients.

Conclusion

We conclude that implementation of the MOPHAR monitoring program substantially improved monitoring practices at the outpatient department for bipolar disorder. The rate of detected metabolic aberrances is high in this sample of bipolar outpatients. We could not determine whether these were a result of initiated treatment or were pre-existent. Whether the MOPHAR monitoring program results in better patient outcomes is part of ongoing research.

 

Insight into the use and indication of expensive drugs: linkage of in-patient pharmacy data to a quality registry

A.M.P. Eldering-Heldens ^a*, P.H.A.M. Kloeg ^a, D. Ten Oever ^b and D. Hilarius ^c

^a Department of Pharmacy, Noordwest Ziekenhuisgroep, Alkmaar and Den Helder.

^b Department of Oncology, Noordwest Ziekenhuisgroep, Alkmaar and Den Helder.

^c Department of Pharmacy, Rode Kruis Ziekenhuis, Beverwijk.

* Correspondence: a.eldering@nwz.nl.

Background

In the past years there has been increasing focus on the costs and outcomes of treating patients with expensive drugs. In The Netherlands there is a yearly increase of costs with 100 million euros, the estimated costs in 2016 are 1.9 billion euros.

Methods

Expensive drugs registration data and DICA registry data from three different hospitals over the period January up to and including July 2017 were combined in a dashboard, through a database. The Nabon Breast Cancer Audit (NBCA) registries of DICA were linked to the expensive drugs registration data.

Results

For one of the participating hospitals, Noordwest Ziekenhuisgroep, a teaching hospital, oncology drugs account for 47% of the total expensive drug budget. Expensive drugs for the treatment of breast cancer represent 10% of the total expensive drug budget and specifically trastuzumab and pertuzumab represent 8% of the budget. Focusing on the NBCA population in all three hospitals, 242 patients are found in the dashboard. On average 32% of these patients has a HER2Neu receptor positive tumor, 39% has a HER2Neu receptor negative tumor and 29% has a tumor which was not classified or results were not registered yet. Of all 76 patients with HER2Neu positive tumors 98% has been treated with trastuzumab, of which 5% was also treated with pertuzumab. The most common indication for treatment of trastuzumab was HER2Neu positive EBC in adults and for the combination of trastuzumab and pertuzumab it was neoadjuvant breast cancer. Average costs for trastuzumab treatment are €29.000 euro per patient per year and for trastuzumab and pertuzumab combination therapy costs are €72.000,- euro. 60% of all patients with HER2Neu positive tumors, who were treated with trastuzumab underwent breast conserving surgery and 40% had an ablatio.

Conclusion

Combining NBCA and expensive drugs data leads to more insight in the relation between drug treatment, costs, patient/disease specific characteristics and potential outcomes. The dashboard which is now available will be complemented with data from additional hospitals and for a longer period of time. Eventually it will be possible to compare treatment and outcomes per indication throughout The Netherlands.

 

Medication optimization in elderly patients with cancer and a limited life expectancy

E.J. Brokaar ^a*, A. Sobels ^a, L. Binkhorst ^a, F. van den Bos ^a, J.E.A. Portielje ^b and L.E. Visser ^a

^a HagaZiekenhuis, Den Haag.

^b LUMC, Leiden.

* Correspondence: e.brokaar@hagaziekenhuis.nl.

Background

With the increasing number of elderly people, the number of elderly patients with cancer and polypharmacy is increasing. Preventive medication does not contribute to the quality of life of patients’ cancer with a limited life expectancy, but is often not discontinued despite a poor prognosis. Our aim was to reduce drug related problems (DRPs) and possibly inappropriate medications (PIMs) in these patient to improve the quality of life in their last year, and to measure patient satisfaction with the medication review process.

Methods

A pharmacist performed a medication review and the resulting recommendations were discussed in a multidisciplinary team (MDT) with a clinical oncologist and geriatrician until consensus was reached. A month after implementation of the recommendations, the patient was asked to complete a short questionnaire using a five-point scale to measure satisfaction with the medication review process. Implementation of the MDT’s recommendations was evaluated using the pharmaceutical file of the community pharmacy.

Results

An interim analysis showed high to very high patient satisfaction. The mean number of recommendations was 4.7 per patient, of which 56% was to discontinue a drug. The MDT accepted 84% of the recommendations of the pharmacist directly and 5% was accepted after a modification of the original recommendation. Of the MDT’s recommendations, 67% was implemented in the patients’ pharmaceutical file. The number of DRPs decreased from 4.9 to 2.5 and the number of PIMs from 2.9 to 1.2 per patient. The mean decrease of drugs in use was 1.5 per patient. Drug classes with the most recommendations were antihypertensive drugs, antihyperlipidemic drugs, and vitamin D.

Conclusion

The pharmacist-led medication review led to a high or very high patient satisfaction in these elderly patients with cancer and to a decrease in the prescribed number of drugs, DRPs, and PIMs.