Introduction

Adverse events experienced by patients at discharge from hospital are commonly related to medication use [1–4]. Approximately one third of adverse drug events occurring around the discharge period are the result of error [3]. Effective and accurate prescription information should be transferred from the hospital to the patient’s GP and community pharmacist to ensure optimal patient care [2, 4].

The inaccuracy of medication documentation at discharge has been demonstrated in other jurisdictions [5–7]. Policy documents outlining strategies to promote medication safety for patients moving from one care environment to another have been produced in the United Kingdom (UK) [8] and the United States (US) [9]. At present, there exists no such policy or recommendations in the Irish setting, nor is there any published evidence to suggest the need. This study investigated the quality and consistency of medication documentation produced for transfer to the next care provider at patient discharge.

Participants and methods

Study setting

The study was conducted in a 600 bed academic teaching hospital in Ireland. Written consent for participation in the study was obtained, and the study was granted Hospital Ethics Committee approval.

The current prescribing process involves a junior doctor hand writing inpatient medication orders onto a drug prescription form. At discharge, the doctor writes the discharge prescription and transcribes medication details onto a discharge summary sheet.

The study took place over a 3 month period from 17 January to 17 April 2006. Included were all patients admitted to and discharged from inpatient cardiology care in four chosen medical wards. Excluded were patients who were transferred from the hospital to another care facility for further surgical intervention.

Measures and data collection

The assessment was conducted at discharge. All documents listing medication were reviewed and assessed for accuracy and congruency of details. This included:

  • inpatient drug prescription form

  • discharge summary sheet

  • discharge prescription

  • case notes.

Any anomalies were analysed for their potential to cause patient harm using a validated visual analogue scale between 1 (no harm) and 10 (death) [10]. A panel of five experienced healthcare professionals (a drug safety co-ordinator, two clinical pharmacists, a cardiology registrar and a cardiology clinical nurse manager) independently scored the clinical importance of every error. The mean score for each error was calculated and categorised as potential to cause none or minor (mean score <3), moderate (mean score 3–7), or severe (mean score >7) patient harm.

Definitions

Where medication details documented for any given patient were not harmonised across the case notes, drug prescription form, discharge summary sheet and the discharge prescription, it was recorded as an inconsistency. For the purpose of the study a prescribing error was defined as a prescribing decision or prescription writing process that results in an unintentional, significant reduction in the probability of treatment being timely and effective or increase in the risk of harm, when compared with generally accepted practice [11].

Omission of medication prescribed on an as required basis during inpatient stay was not recorded as an inconsistency, for example discontinuation of deep vein thrombosis prophylaxis once patient became mobile upon discharge, discontinuation of sleeping tablets or laxatives prescribed on an as required basis during patient stay.

Ethical considerations

Throughout the study, where potentially serious adverse patient outcomes were uncovered, steps were taken by the research pharmacist and a cardiology consultant to take remedial action.

Data analysis

Data was entered onto and analysed using Microsoft Excel® and EpiCalc 2000®.

Results

A total of 192 patients admitted to inpatient care under the cardiology consultants were eligible for inclusion in the study. Of these 139 patients (72%) consented to participation (28 patients were admitted and discharged within such a short timeframe as to prohibit the opportunity to recruit them to the study; 25 patients refused to take part). Patient age ranged from 18 to 89 years (median 63). The median number of regular medications prescribed per patient upon discharge from hospital was 7 (range 1–19). A total of 70.5% of patients were male, 56.8% were medical card holders and 85% were discharged on a weekday. The median length of stay was 6 days (range 1–46).

Identification and classification of inconsistencies

Of the 139 patients recruited, inconsistency in documentation of discharge medications was observed in 91 (65.5%, 95% confidence interval (CI) 57 to 73%).

Any patient may have suffered more than one inconsistency and these may have been present in one or more of the documents (Table 1). Drug omission was the commonest inconsistency type observed.

Table 1 Inconsistencies observed
Full size table

Clinical importance of inconsistencies

A sum of 113 errors were identified in 1,046 medication orders (10.8%, 95% CI 9-12%). Errors were judged to have the potential to cause none or minor (47%), or moderate (53%) patient harm. None were judged to have the potential to cause severe patient harm. Examples of minor and moderate errors are provided (Table 2). Remedial action was taken for three (2%) patients to mitigate patient harm.

Table 2 Examples of inconsistencies identified
Full size table

Inconsistencies related to day of discharge

When analysed according to day of discharge, there was no discernible difference in inconsistency rate between those discharged on a Saturday or Sunday and those discharged on a weekday (p = 0.11, χ2 = 2.56, DF = 1; relative risk 1.28).

Inconsistencies related to number of medications prescribed

When analysed according to increasing number of medications prescribed, the incidence of inconsistencies is significantly different between groups (p = 0.0004, χ2 = 16.17, DF = 2) (Table 3).

Table 3 Risk of inconsistency/number of medications per patient
Full size table

Discussion

Findings in this study add an Irish context to the evidence demonstrating inefficiencies and risk for patients moving between care environments [1, 2, 12, 13]. The number of patients who experienced a discrepancy is higher than reported elsewhere (range 11–53% of patients) [1–4]. The error rate per medication order is consistent with findings from UK but differs in that in the UK errors were detected in the routine course of clinical pharmacy service, providing an opportunity to take corrective action: Franklin et al. [14] reported an error rate of 9.2% across newly written, as required and discharge prescriptions; Tully and McElduff [15] reported an error rate of 10.7% identified by clinical pharmacists across all stages of patient stay from admission to discharge. Interestingly the current study found that patients discharged at the weekend, with on-call staff and limited support services, did not experience a greater risk of discrepancy than those discharged on a weekday. This illustrates the complexity of the process that involves the same level of input independent of day or level of support.

The authors acknowledge the limitations of this study, performed in a single service in one hospital involving a small number of patients. However the finding that almost two thirds of patients experienced an inconsistency, many of which had the potential to result in moderate patient harm, merits further investigation. It is worth noting that the discharge prescription is the only legal document on foot of which medication is dispensed, and discrepancies identified on the discharge summary may carry less potential for patient harm. Indeed this is borne out by the fact that the authors felt the need to contact only three patients to take corrective action. However, conscious of the importance of addressing systemic problems that underlie human errors, the authors seek to adopt a generative organisational safety culture [16] and focus on the discharge prescribing process, and improvements that can be made to it [17]. The prescribing process described is typical of that used in other specialty areas, and other hospitals in Ireland, and so similar findings would be expected elsewhere. The survey is currently being repeated in general medical and surgical services in the study hospital. Such investigation is a constructive and beneficial tool only when it leads to systematic improvements in patient outcomes. There are a number of evidence based strategies to improve standards of care at admission and discharge. The question is whether these strategies are adaptable to and feasible in the Irish setting.

The current discharge prescribing process involves reconciling four different medication lists, resulting in a high transcription burden, which is known to increase the risk of prescribing error [18]. It is perhaps not surprising therefore that drug omission was the commonest error type identified. This study demonstrates that the handwritten discharge summary is an unreliable source of medication information and in its current format a non-value added step in the process. This finding is supported by a recent systematic review which demonstrated that discharge summaries lack information on discharge medication in 2–40% of cases [7]. Electronic prescribing has been shown to decrease prescribing errors and increase prescribing quality in the US [19, 20] and the UK [21]. A pilot study in our hospital of electronic generation of the discharge prescription and summary using a surgical patient management system has demonstrated a significant reduction in prescribing errors. Results of this study will be reported elsewhere. The system is being adapted for use in a medical setting and will be piloted in cardiology later this year.

Intensive clinical pharmacy input from admission to discharge has been shown to improve patient outcomes in the UK [4, 14, 15], US [22] and Northern Ireland [23, 24]. Such services are the norm in the UK since the early 1990s [25, 26], but not in the Republic of Ireland. In the study hospital clinical pharmacy services are provided on admission and during patient stay but not at discharge. Resource constraints and a ceiling on the number of staff employed in each hospital represent barriers to service expansion [27]. Local research is warranted to investigate the feasibility and impact of extending clinical pharmacy services to the point of discharge.

The data presented in this study demonstrates the inefficiencies and medication safety risks inherent in the current discharge prescribing process. The authors hope to highlight the existence of this problem, assess the impact in other services and hospitals and demonstrate the need to design a new, sustainable and evidence based process which will improve patient outcomes in the Republic of Ireland, consistent with the Health Services Executive’s strategy for transformation [28].

Conclusions

Quality and accuracy of medication information imparted at discharge is poor and carries with it the potential to cause minor to moderate patient harm. This is largely attributable to the complexity of the process. Strategies for improvement include involvement of clinical pharmacists at the point of discharge and simplification of the process through the use of an electronic solution.