Abstract
Antiretroviral therapy (ART) is complex and has high propensity for medication error and drug-drug interactions (DDIs). We evaluated the clinical utility of pharmacist prescreening for DDIs, adherence to
Introduction
Current treatment for HIV-positive patients is complex due to lifelong antiretroviral therapy (ART) comprising regimens with significant propensity for drug-drug interactions (DDIs). In the developed world, the ageing population of HIV-positive patients increases risk of polypharmacy, medication error and DDIs.1–4 However, during complex outpatient consultations which, for example, may involve lipid monitoring, cardiovascular risk assessment, monitoring of renal and hepatic function, bone mineral density or therapeutic drug level monitoring, updating medication lists and checking for DDIs may frequently be overlooked. Prescribing errors have been described in up to 20% of patients taking ART, the majority of which included co-administration of contraindicated combinations. 1 Measured prevalence of DDIs is high, affecting 27% of 159 outpatients in the UK, 2 and 40% of 1497 patients in Switzerland; 4 however, physician awareness of ART DDIs is low, with only 36% of clinically significant DDIs correctly identified by physicians in 159 patients in the UK. 2
Identification and management of DDIs requires that a current, accurate medication history is maintained. Patients may take medication prescribed from other hospital specialties, primary care or purchased over the counter (OTC), including traditional and complimentary medicines, 4 recreational drugs and, increasingly, medicines purchased from other sources including via the Internet. Medication-recording discrepancies for HIV-positive patients have been reported as common between hospital departments, 5 as well as between hospital and community settings.
Medicines reconciliation is a fundamental role of clinical pharmacists and, in secondary care, has been found to reduce incidence of prescribing errors in some settings.6–8
In many centres, pharmacists have active roles in HIV management,1,9 although one study showed that while availability of complete medication lists reduced number of DDIs per patient, pharmacist advice on DDI management added no further benefit. 10
We sought to evaluate whether detailed medication history-taking, prescreening for DDIs and ART adherence assessment by a pharmacist prior to HIV outpatient consultations added benefit to patient care.
Methods
This was an evaluation of a new service conducted in two HIV outpatient clinics (Genitourinary Medicine and Infectious Diseases) at a single hospital site. The evaluation took place between 12/2009 and 10/2011 at a large teaching hospital. Consecutive patients who were aged 18 years or over, taking current ART and willing to speak to a pharmacist about their medication were seen prior to their routine outpatient appointment for HIV care. Patients with acute opportunistic infection or AIDS-defining illness, or those unable to speak English were not included. A structured consultation took place, in which patients were asked for a detailed medication history including: ART, hospital-prescribed medication, medication prescribed in primary care, OTC medication, herbal medicines, vitamins, supplements and recreational drugs. Where patients were unsure of their regular medication, their general practitioner was contacted for confirmation. Any medicines taken by the patient that were not recorded in the clinical notes entry or clinic summary letter for the previous appointment were deemed ‘not previously recorded’. The pharmacist then screened the full medication list for DDIs involving ART using the University of Liverpool online resource www.hiv-druginteractions.org, 11 and the summaries of product characteristics of each medication. DDIs included were those rated on the HIV drug interactions website as ‘red’ (Drugs should not be co-administered) or ‘amber’ (Potential interaction which may require close monitoring, alteration of drug dosage or timing of administration). 11 Combinations were included as potential DDIs if there were clinical data suggesting an interaction, or if a theoretical interaction could be expected based on elimination, metabolism or toxicity profiles of the co-administered drugs. DDIs between ART and co-medications and between components of ART regimens were included, with the exception of the ‘amber’ warning between ritonavir and protease inhibitors, as all patients taking protease inhibitors in the cohort were taking ritonavir at standard doses for pharmacokinetic boosting.
Adherence to ART was assessed using a simplified Medication Adherence Self-Report Inventory (M-MASRI). 12 Patients were shown a visual analogue scale from 0 to 100% and were asked to estimate how much medication they had taken in the preceding month, and how many doses had been missed.
A template detailing current medication, any DDIs and adherence assessment was filed in the clinical notes, including a personalized drug interaction report. Treating physicians were asked to give feedback via a structured questionnaire, whether the information provided told them something they did not previously know, and if they changed their management of the patient as a result. The questionnaire was divided into three sections for both questions, allowing physicians to state whether each component of the pharmacist consultation told them something they did not know, i.e. provided new information. For example, whether the medication history made them aware of a new drug the patient was taking from their general practitioner, or a dose change; if the drug interaction check highlighted a previously unknown interaction or confirmed a lack of interaction between a new combination; and whether the adherence check highlighted issues of which the physician was unaware. Change in management constituted any change in the treatment of the patient made by the treating physician, which resulted from the information provided by the pharmacist. This included a change in prescribing of any medication, and any monitoring which was undertaken, for example in order to manage a DDI.
The utility of the service was audited following its introduction. The same pharmacist saw all of the patients included. Fisher's exact tests (StatsDirect version 2.6.8, Statsdirect Ltd, Cheshire, UK) and Spearman's Rank (PASW Statistics version 17.0, SPSS Ltd, Quarry Bay, Hong Kong) were utilized for statistical analyses.
Results
A total of 200 patients aged from 19 to 71 years (median 43), including 127 (63.5%) male, were included in the analysis. No patients refused the pharmacist consultation, and all patients approached were included in the analysis. Of the total 192 (96%) patients were prescribed an ART regimen containing a nucleoside reverse transcriptase inhibitor (NRTI), 92 (46%) were taking a non-nucleoside reverse transcriptase inhibitor (NNRTI), 89 (44.5%) were taking a protease inhibitor (PI), and 10 (5%) patients were taking both an NNRTI and a PI.
From 200 patients who had pharmacist consultations, 103 physician responses were available (Table la). Using Fisher's exact test, there was no significant difference in the rate of physician questionnaire return for patients who had a medication change (39/103 versus 32/97, P = 0.28), a DDI (64/103 versus 52/97, P = 0.14) or adherence <90% (6/103 versus 13/97, P = 0.11), suggesting that the patients for whom physician questionnaires were returned were likely to be representative of the cohort. Demographic features of the patients for whom physician questionnaires were returned were comparable with the evaluated cohort (data not shown).
Summary of findings from pharmacist consultations in 200 patients taking antiretrovirals and responses to physician questionnaire
ARV= antiretroviral
AII values are n (%)
Including new medication/regular medication not recorded, discontinued medication and dose changes to regular medication
In 63 (61.2%) cases, physicians reported that one or more component of the pharmacist consultation told them something they did not know. In 14 (13.6%) cases, physicians reported changing management of a patient based on information from the pharmacist consultation. Outcomes of physician responses are shown in Table 1b.
Summary of responses to physician questionnaire (n = 103) from 200 pharmacist consultations
All values are n (%)
All components includes medication history, drug interaction check, adherence assessment
Physicians reported that a medication history told them something they did not know in 20 (19.4%) patients. Seventy-one (35.5%) patients had a change in medication from that previously recorded in their case-notes, including one or more of the following: a new medication or medication not recorded since the previous clinic appointment (50, 25%); a medication discontinued (23,11.5%) or a dose change (14, 7%).
A total of 143 (71.5%) patients were taking one or more prescribed co-medications alongside their ART. These are shown in Table 2. Nineteen (9.5%) patients were taking one or more recreational drug and 7 (3.5%) took a herbal or homeopathic remedy. Thirty-five (17.5%) purchased vitamins or supplements and 20 (10%) were taking medicines purchased from a pharmacy. Medicines purchased OTC were loperamide, pseudoephedrine, ibuprofen, antacids, codeine phosphate/paracetamol, antihistamine eye drops, paracetamol, nicotine replacement products, diphenhydramine and aspirin.
Co-medication use and potential drug-drug interactions observed between antiretrovirals and different therapeutic classes of co-medications in 200 HIV-positive outpatients
Co-med = co-medication; DDI = drug-drug interaction; ARV = antiretroviral; OTC = over the counter
AII values are n (%)
Physicians reported that DDI checks told them something they did not know in 39 (37.9%) consultations (Table 1a), and in three cases, for patients where no potential DDIs were observed. One or more potential DDI was observed in 115 (58%) patients, with a total of 246 DDIs (mean 1.2). Of these, 22 (9%) potential DDIs occurred with medication which was not previously recorded in the patients’ notes. The DDIs observed with the different classes of co-medications taken are shown in Table 2. Only 2 (1%) DDIs involved contraindicated ‘red’ combinations. These were atazanavir/ritonavir with nevirapine, and atazanavir/ritonavir with simvastatin.
The most commonly prescribed medicines in this cohort of patients taking ART were cardiovascular agents (including antihypertensives, antiplatelets, cholesterol reducing agents) and central nervous system (CNS) drugs (including antidepressants, hypnotics, antipsychotics). DDIs occurred most frequently with CNS drugs, affecting 20% of the patient cohort and accounting for almost 25% of the DDIs observed. (Table 2).
Patients taking a Pi-containing regimen were more likely to experience a DDI (80/99 versus 37/101, P < 0.0001) although co-medication use was comparable. Consequently, in these patients, the pharmacist consultation was more likely to tell the doctor something they did not know than in patients taking non-PI-containing regimens (42/55 versus 21/48, P = 0.0011).
There was a correlation between increasing age and the number of co-medications taken by patients, and also between increasing age and increasing number of potential DDIs (P = 0.011, P < 0.0001 respectively; Spearmans Rank) This analysis was unadjusted for factors such as time on ARVs.
Patient-reported ART adherence ranged from 0 to 100% (mean 95.4%). Sixty-two (31%) patients reported ART adherence less than 100%.
Physicians were more likely to report that a pharmacist consultation told them something they did not know, in patients who were taking two or more concomitant medications in addition to ART (41/54 versus 22/49, P = 0.0012). When components of the consultation were analysed separately, medication history and DDI checks were more likely to add benefit in patients taking two or more co-medications (15/54 versus 5/49, P = 0.0213, 29/54 versus 10/49, P = 0.0005, respectively), whereas the likelihood of an ART adherence assessment adding benefit did not differ with number of co-medications.
Discussion
We aimed to determine whether pharmacist prescreening prior to HIV outpatient appointments including medication history taking, screening for DDIs and adherence assessment can add benefit. The pharmacist consultation provided new information in 63 (61.2%) patients, and physicians reported changing management of patients in 14 (13.6%) cases. Potential DDIs were observed in 115 (58%) of all 200 patients, and were more common in patients taking regimens containing Pis. Seventy-one (35.5%) patients had a change in medication (a new medication or medication not recorded, a medication discontinued, or a dose change) since their previous clinic appointment.
The DDI check was the component of the pharmacist consultation that physicians most frequently reported told them something they did not know (37.9%), which is in keeping with our previous findings. 2 In three cases, physicians reported that the DDI check told them something they did not know in patients for whom no DDIs were observed, which illustrates the benefit of confirming safe combinations.
Although only 6 (5.8%) of 103 patients estimated their adherence to be <90%, physicians reported that the ART adherence assessment told them something they did not know in 32 (31.1%) patients. In four cases, physicians additionally annotated the questionnaire that the ART adherence assessment saved time. Although physician time saved by pharmacist prescreening was not formally assessed, it is an example of an outcome measure which could be investigated in future evaluations.
There was a change in prescribed medication history (including newly-prescribed medication or medication not previously recorded, discontinued medication and dose changes) in 35.5% of 200 patients, with physicians reporting that a medication history told them something they did not know in 19.4% of 103 responses (Tables 1a and b). It is possible in some cases that the treating physician was aware of medication changes, although they were not recorded. It is also possible that in some cases, although there had been a change to non-ART medication, physicians did not believe that this information added benefit.
Pharmacist consultations were more likely to add benefit in patients taking two or more co-medications alongside their ART, and in patients taking a PI-containing regimen. There is therefore potential to prioritize patients that may be most likely to benefit from a pharmacist consultation. However, a complete current medication history is likely to benefit all patients, as these data show that HIV-positive patients may access medication from various settings, and have regular changes to non-ART medication.
As previously described,3,13 the number of co-medications taken, and DDIs observed increased with increasing age.
The prevalence of DDIs reported here (58%) is relatively high compared with other studies.2,4,10 All potential interactions were reported to physicians and included in the analysis if there was potential for interaction via metabolic pathways, renal elimination pathways or overlapping toxicity profiles, including theoretical interactions which may require clinical monitoring in some patients. DDIs were more common in regimens containing PIs, which has been reported elsewhere. 2 Other studies have reported only those interactions considered to be clinically significant. We did not evaluate the clinical significance of the DDIs reported, or assess clinical harm. The potential DDIs reported therefore represent potential for risk, however an unknown proportion may translate to actual clinical harm.
Limitations include lack of evaluation of clinical implications of potential DDIs, and to what extent actual harm was caused. Only 19 (9.5%) patients reported adherence <90%. As adherence was assessed by patient self report, there is potential for over-estimation. Reporting of recreational drug use and medicines purchased without prescription may be underestimated, depending on patient willingness to disclose. The patients for which physician responses were available were found to have comparable rates of medication changes, DDIs and adherence to other patients for whom responses were not available. However, it is unknown whether doctors who changed management of patients were more likely to return questionnaires. We did not evaluate the grades of prescriber who returned forms, and whether more junior grades were more likely to return a form, or respond that the pharmacist consultation told them something they did not know or changed management of their patient. We did not differentiate in the analysis of medication recording between patients with a newly-initiated medicine since the last clinic visit, and patients who were taking a medicine which was not recorded in their notes.
Pharmacists are equipped with the skills and experience to perform accurate medicines reconciliation and to screen for DDIs in patients taking ART regimens. Although pharmacists in some centres may already be involved in outpatient clinics for HIV-positive patients, these data show that medicines reconciliation, DDI screening and ART adherence assessment, which are core skills for pharmacists, can add benefit in the HIV outpatient setting.
Footnotes
Acknowledgements
We thank the National Institute of Health Research (NIHR – Department of Health) and the Northwest Development Agency (NWDA) for infrastructural and project support. No funding has been received specifically for this evaluation. SK has research grants, travel support and honoraria for lecturing from Abbott, Merck, Janssen, BMS, ViiV and Gilead. The Liverpool Drug Interactions website (SK, KS) receives sponsorship from ViiV, Merck, Bristol Myers Squibb, Boehringer Ingelheim, Janssen, Gilead. The website also receives funding from the UK National Institute for Health Research, the British HIV Association, The European AIDS Clinical Society and proceeds from the International Conference on Drug Therapies in HIV Infection. MB, AROM has accepted consultancy fees, speaker fees, travel grants and/or research funding from Janssen, MSD, Viiv/GSK, Boehringer Ingelheim, BMS, Abbott and Gilead, MB: None to declare.