Article Text

Download PDFPDF

Pharmacist- versus physician-acquired medication history: a prospective study at the emergency department
  1. Sabrina De Winter1,
  2. Isabel Spriet1,
  3. Christophe Indevuyst1,
  4. Peter Vanbrabant2,3,
  5. Didier Desruelles3,
  6. Marc Sabbe3,
  7. Jean Bernard Gillet3,
  8. Alexander Wilmer2,
  9. Ludo Willems1
  1. 1Pharmacy Department, University Hospitals Leuven, Leuven, Belgium
  2. 2Department of General Internal Medicine, University Hospitals Leuven, Leuven, Belgium
  3. 3Department of Emergency Medicine, University Hospitals Leuven, Leuven, Belgium
  1. Correspondence to Miss Sabrina De Winter, Herestraat 49, 3000 Leuven, Belgium; sabrina.dewinter{at}uzleuven.be

Abstract

Background Recent literature revealed that medication histories obtained by physicians and nurses are often incomplete. However, the number of patients included was often low.

Study objective In this study, the authors compare medication histories obtained in the Emergency Department (ED) by pharmacists versus physicians and identify characteristics contributing to discrepancies.

Methods Medication histories were acquired by the pharmacist from patients admitted to the ED, planned to be hospitalised. A structured form was used to guide the pharmacist or technician to ensure a standardised approach. Discrepancies, defined as any difference between the pharmacist-acquired medication history and that obtained by the physician, were analysed.

Results 3594 medication histories were acquired by pharmacy staff. 59% (95% CI 58.2% to 59.8%) of medication histories recorded by physicians were different from those obtained by the pharmacy staff. Within these inaccurate medication histories, 5963 discrepancies were identified. The most common type of error was omission of a drug (61%; 95% CI 60.4% to 61.6%), followed by omission of dose (18%; 95% CI 17.6% to 18.4%). Drugs belonging to the class of psycholeptics, acid suppressors and beta blocking agents were related to the highest discrepancy rate. Acetylsalicylic acid, omeprazole and zolpidem were most commonly forgotten.

Conclusion This large prospective study demonstrates that medication history acquisition is very often incomplete in the ED. A structured form and a standardised method is necessary. Pharmacists are especially suited to acquire and supervise accurate medication histories, as they are educated and familiar with commonly used drugs.

  • Quality assurance
  • medication reconciliation
  • hospital pharmacy services
  • psycholeptics
  • beta blocking agents
  • drug therapy
  • emergency department
  • medication
View Full Text

Statistics from Altmetric.com

Introduction

Recent literature revealed that medication histories obtained by physicians and nurses are often incomplete, particularly in the Emergency Department (ED).1 2 6–12 A complete medication history should include all prescription drugs, over-the-counter drugs, herbal medicines, vitamins and nutritional supplements.4 Reasons for incompleteness may include diagnostic and treatment priorities that prohibit gathering a detailed medication history, and additionally the patient may not be able to provide an accurate history upon admission.3 However, an inaccurate medication history can conceal a drug-related problem. Moreover, errors of incompleteness often lead to delayed medication administration due to the time required for the hospital pharmacist to clarify the order and dispense the right product.

During hospitalisation, medication errors most frequently result from incomplete knowledge of drugs used by the patient leading to the subsequent development of an inaccurate drug chart. These unintended discrepancies are particularly important, as they will not be detected by computerised physician order entry systems (CPOE). These systems only provide information about what has been prescribed.5

The goals of this prospective study were (1) to compare medication histories obtained in the ED by pharmacists versus physicians and (2) to identify the drug classes in which discrepancies were high and potentially clinically relevant.

Methods

This study was conducted at the ED of the University Hospitals Leuven, Belgium; a 1900-bed, tertiary care teaching hospital. The ED admits around 140 patients per day (51 000 patient visits per year), from which 40% are referred by a general practitioner (GP). Approximately 30% of patients are hospitalised. Depending on the suspected medical problem, patients will be admitted by a physician of internal medicine, surgery or emergency medicine. These separate disciplines use their own patient chart to obtain a medication history. The physician will use various sources of information, including patient and family interview, review of medication lists or boxes, referral notes and contact with the GP to complete the medication information. The ED also has a 24-bed observational unit, mostly used by all admitted patients in the hospital and thus providing an acceptable time window to obtain complete medication histories.

During a 19-month study period (February 2007 to August 2008), a clinical pharmacist and a well-trained pharmacy technician (one full-time-equivalent) obtained medication histories in the ED independent from the physician-acquired ones, from 08:30 to 17:00 during the week. Medication histories were acquired from patients planned to be hospitalised. An online computer-generated admission system was used to identify those patients. Patients were included, depending on the availability of the clinical pharmacist or pharmacy technician. Patients, planned to be transferred to a ward using CPOE, were preferably interviewed. At the time of the study, CPOE was only partially (±25%) implemented in the hospital. Patients younger than 16 years, intubated/mechanically ventilated, poisoned and psychiatric patients were excluded.

The team of pharmacy technicians consisted of five technicians assigned in separate shifts to the ED and trained in medication reconciliation. All medication histories obtained by these pharmacy technicians were supervised by the clinical pharmacist.

A structured form (a checklist, a table and a standardised list of questions) was used to guide the pharmacist or technician to obtain medication histories in detail, including dosage and frequency (see online Appendix 1). The patient was explicitly asked about prescription and non-prescription drugs easily forgotten, such as topical and ophthalmological drugs, inhalers and dietary supplements. Different sources were used to retrieve information: communication with the patient and family members, verification of medication boxes brought into the ED, previous medical records, referral notes from the GP, contact with other hospitals or nursing homes and contact with the GP or community pharmacist.

After completion, the medication history was entered into the CPOE. In case of transfer to a ward without CPOE, the completed form was added to the medical record.

A discrepancy was defined as any difference between the physician-acquired medication history and that obtained by the pharmacist. When no medication history was acquired by the admitting physician, and no GP record or referral note was found, it was registered as ‘no medication history found.’ In this case, every drug obtained by the pharmacy staff was registered as an omission. All discrepancies were reviewed (orally and/or written) with the admitting medical team.

Drugs were categorised according to the Anatomical Therapeutic Chemical (ATC) Classification System of the WHO Collaborating Centre for Drug Statistics Methodology.14

All data that were collected and analysed with a Microsoft Office Access database (version 2003 and version 2007; Microsoft, Seattle, Washington) are shown in box 1. SPSS version 17.0 (SPSS, Chicago, Illinois) for Windows software was used for statistical analysis. For the main results, a 95% CI was reported and presented as result ±95% CI range. p Values were calculated using one-way analysis of variance (ANOVA) for continuous variables. Pairwise comparisons were made for statistically significant differences with a Tukey post hoc test. Statistical significance was defined as a two-tailed p value <0.05.

Box 1 Collected data

  • Date of medication history obtained by pharmacist or pharmacy technician

  • Age of patient

  • Origin of patient

  • Admission time at the ED

  • Discipline of physician obtaining medication history

  • Ward on which the patient would be admitted

  • Average time required to complete each medication history by pharmacy staff

  • Number of drugs in computerised physician order entry systems

  • Discrepancies:

    • Drug omission

    • Drug commission

    • Dose omission

    • Incorrect dose

    • Incorrect dosage schedule

    • Discrepant route of administration

  • Name of drug

  • Number of information sources

  • Number of phone calls

    • To community pharmacist

    • To GP

    • To nursing home

  • Patient brought medication in original boxes?

Results

During the 19-month study period, 71 786 adult patients were admitted to the ED, and 16 523 were hospitalised. Of the ED admitted patients, 27 980 were treated by surgeons, 33 064 by internal medicine physicians and 3720 by emergency physicians. Respectively, 20%, 49% and 31% of these patients were hospitalised. The pharmacist or pharmacy technician acquired medication histories of 3594 hospitalised patients (22%). The demographics and clinical characteristics of the study population and all hospitalised patients are summarised in table 1.

Table 1

Patients' demographics and clinical characteristics

In the study population, the median number of drugs registered per patient was 5 (range 0–24, IQR 2–8). Of the 3594 study patients, 390 (11%) patients did not use any medication; 2108 (59%) medication histories were registered in the CPOE.

The pharmacist-acquired medication histories were compared with the physician-acquired histories: 2552 (71%) medication histories were recorded by internal medicine physicians, 989 (27.5%) by surgeons and 54 (1.5%) by emergency physicians.

A total of 2134 (59%±0.8%) medication histories recorded by physicians were different from those obtained by the pharmacy staff. Within these, 5963 discrepancies were identified. When comparing the discrepancy rate according to the specialty, we observed an average of 2.1 discrepancies per medication history for surgery, 1.5 for internal medicine and 1.7 for emergency physicians. The differences observed between internal medicine versus surgery (p<0.0001) and emergency medicine versus surgery (p<0.003) were statistically significant.

For 171 patients, no medication history was recorded by the physician. Forty-four (1.7%) histories were absent for internal medicine patients, 121 (12.2%) for surgical and two (3.7%) for emergency medicine patients.

The most common error (3640–61%±0.6%) was omission of a drug, taking into account that 171 medication histories were lacking, in which 468 drugs were registered as an omitted drug. The second most common error (1089–18%±0.4%) was omission of the dose. Other types of discrepancies were incorrect medication dose (413–7%±0.3%), commission of medication (388–6.5%±0.3%) and incorrect dosage schedule (427–7%±0.3%). The route of administration was incorrect in 0.1% (six) of cases.

Drugs belonging to the class of psycholeptics (N05), acid suppressors (A02) and beta-blocking agents (C07) were observed with the highest total discrepancy rate (figure 1). Acetylsalicylic acid, omeprazole and zolpidem were most frequently forgotten (table 2).

Figure 1

Top 10 drug classes (classified according to ATC level 2) involved in discrepancies.

Table 2

Number of drug omissions linked to Anatomical Therapeutic Chemical 5

The median time, measured by self reporting, required for a complete medication review by the pharmacy staff, was 15 min (range 5–90 min, IQR 10–20 min).

For 1049 (49%±1%) medication histories, a single source was considered sufficient. Eight hundred and sixty-three medication histories (40%±1%) required two sources to be completed, and for 194 (9%±0.6%) histories, three sources were needed. For the remaining 31 (2%±0.3%) medication histories, more than three sources were used, with a maximum of six. In total, 1156 phone calls were made. For 519 medication histories (24%±0.9%), the community pharmacist was contacted. The GP was contacted for 325 patients (15%±0.7%).

The median time between patient arrival and medication reconciliation by the clinical pharmacist or pharmacy technician was 4 h (range 1–17 h, IQR 3–6 h).

Discussion

To the best of our knowledge, this is the largest prospective analysis on medication reconciliation. In many published studies, the definition of a discrepancy and the methodology of medication history acquisition varies considerably. As a result, the amount of discrepant medication histories varies widely from 10% to 96%.2 11 In our study, 2134 (59%) medication histories were different, comparing those obtained by physicians versus pharmacy staff, resulting in a total of 5963 discrepancies. The comparator in different studies varied from physician-acquired medication histories (used in our study) versus admitting medication orders. When using the admitting medication order as comparator, it represents the patient's actual exposure to medication in the hospital. In this case, it is necessary to distinguish intentional from unintentional discrepancies.2 As a result, every discrepancy has to be discussed with the physician in order to register the discrepancies correctly. As this method is labour-intensive, we avoided the need for this discussion and decided to compare the physician-acquired versus the pharmacist-acquired medication histories.

Some studies underestimate errors by focussing only on omission errors, whereas others include frequency as well as dosing errors and errors on route of administration, similar to our approach.1 2 6–9 11 12

The most common type of discrepancy seen in our study was omission of a drug, followed by omission of dose. In some studies, lack of documentation of correct dose, frequency and route of administration were responsible for the greatest number of discrepancies, while other studies have similar results as our study.1 2 4 5 7 9 Moreover, when non-prescription drugs are included, as in our study, the frequency of discrepancies is even higher.1 2 4 6 7 10 When summarising prospective studies for incompleteness of medication histories obtained in the ED, literature results concur with our findings, as shown in table 3.1 7 8 11 12

Table 3

Summary of prospective studies of incompleteness of medication histories obtained at the ED

The systematic review by Tam et al stated that prescription drugs most often involved in medication history errors are cardiovascular agents, sedatives and analgesics.2 Cohen et al found mainly antihypertensives, supplements and analgesics.8 Stepherd et al demonstrated that cardiovascular agents, analgesics and pulmonary agents were most frequently unidentified at triage.12 In our study, the most important drug classes involved in discrepancies were psycholeptics, acid suppressors and beta-blocking agents. These results are not surprising, as these drugs are very often used in our community.

We observed differences in numbers of discrepancies along with the treating specialty. This was also discussed by Yusuff et al; they concluded that the depth and frequency of medication histories taken by physicians are significantly influenced by their specialities.9 Physicians appear to be more interested in documenting an accurate medication history, as it is likely to help them in establishing an exact diagnosis. In our study, in surgical patients with often clear traumatic diagnoses, a medication history was lacking in 12.1%.

A quantitative study of discrepancies is interesting and important, but a qualitative evaluation of potential harm caused by the discrepancies would provide more clinical relevance. A qualitative investigation, however, was not a pre-defined endpoint of our study. During the study, the more complete pharmacist-acquired histories were always documented in the medical record, avoiding further harm. Potential harm caused by incomplete medication reconciliation can only be evaluated in a pure observational study (without reporting of the pharmacist-acquired medication histories) or a randomised trial (pharmacist versus physician acquired medication histories).

The presented results finally show that obtaining an accurate medication history is very time-consuming. This finding corresponds with literature results.13 Tam et al reviewed that a comprehensive medication history takes 9–30 min to be completed.2 In the study by Reeder et al, the mean time (±SD) to complete a medication history was 16.4±7.3 min.9 The use of multiple sources, including external phone calls, explains the time needed to complete the drug history in our study.

This study has several limitations, however. First, we assumed that in the ED, the pharmacist-acquired medication history is the best assessment of a patient's actual medication use, and we used it as the gold standard for comparison. There was no attempt to further confirm these medication histories. However, it was previously confirmed that pharmacists and well-trained pharmacy technicians are especially suited to obtain medication histories because of their education, experience, medication knowledge and patient-counselling skills.1 2 6–12 Second, opposed to a long period of training for and continuous evaluation of pharmacy technicians, we did not determine inter-rater reliability, which should be done in future studies to validate our standardised approach of medication reconciliation. Third, patients were not included in a randomised manner. Patient enrolment was based on the presence of the pharmacy staff at the ED and patient availability for an interview. In addition, patients planned to be hospitalised on a ward using CPOE were preferably enrolled. When comparing the study population and the population of all hospitalised patients, the male–female distribution and mean age were similar. Our retrospective analysis of the comparability of five selected frequently occurring diagnoses (pneumonia, hip fracture, sepsis, syncope and cerebrovascular accident) demonstrates that all these results do correlate well (table 1). Only for hip fracture did groups differ, which can be explained by the fact that all surgical wards are working with CPOE. Fourth, there is a possibility for recall bias. When the patient is interviewed a second time by the pharmacy staff, he can possibly recall more of his medication regimen. However, as already mentioned, in 41% of cases, two sources were needed to complete the drug history. Fifth, a complete medication history is not a first priority for physicians in the ED, and they often rely for a more accurate record on the obligatory medication reconciliation by the admitting physicians on the ward. Finally, performing phone calls, we did not require a confirmation of medication information in print. Transcription errors or bias may occur with information given over the telephone.

Conclusion

This large prospective study demonstrates that medication histories obtained upon admission at the ED are very often incomplete. More than half of medication histories acquired by physicians were discrepant, as compared with those obtained by the pharmacy staff. A structured form and a standardised method are necessary to obtain a complete and reliable medication history. Pharmacists and well-trained pharmacy technicians are especially suitable for implementing and supervising the acquisition of a correct medication history at the time of hospital admission.

Acknowledgments

This project was approved by the Institutional Board in University Hospitals and was funded by the Health Department of the Belgian government. We thank all pharmacists and pharmacy technicians for their daily efforts in realising this study.

References

View Abstract

Footnotes

  • Linked articles 41020.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Commentary
    Edward Etchells
  • Quality lines
    David P Stevens