Article Text
PDF
Abstract
Objective To compare pharmacy team input to medicines reconciliation (MR) in a variety of care areas in order to inform optimisation of service delivery.
Methods 30 acute hospital pharmacy departments evaluated their MR services in 10 care areas using a piloted data collection form. Omitted medicines and wrong dose discrepancies on the admission prescription were recorded and rated for clinical severity. Data were collected on whether the admission was planned, the number of coprescribed medicines and if the patient had brought their home medicines into hospital.
Results 3086 MRs were reviewed and 4041 unintended discrepancies (UDs) in prescribing were identified (mean 1.3/MR). 1616 UDs (0.52 per patient) were ranked as having the potential for moderate impact on patient care (Level 3). Level 3 UDs were identified in all care areas; however, Admissions, Care of the Elderly, General Surgery and Orthopaedic patients had more Level 3 UDs per patient than the total population (two-tailed Z test, 99% CI). More UDs was ranked Level 3 in Care of the Elderly and General Surgery patients (two-tailed Z test, 99% CI). Over 80% of recorded errors involved four prescribing categories (cardiovascular, central nervous system, endocrine, respiratory). Planned admissions and the presence of the patients’ own medicines had little impact on the accuracy of admission prescribing. The average time to carry out MR was 15 min.
Conclusions Prioritisation of pharmacy-led MR by care area or type of admission alone is not a safe option. Opportunities should instead be taken to explore multidisciplinary methods of implementing MR which optimise available information.
Statistics from Altmetric.com
Background
Incomplete or inaccurate communication at the time of admission has been shown to lead to prescribing errors in up to 67% of all admissions, with between 11% to 59% considered to be clinically important.1–6 Uncorrected errors on admission to hospital can impact on the inpatient stay, and may be continued after discharge, resulting in increased contact with healthcare providers, readmissions and even death.7 ,8 In a prospective study to measure the nature and prevalence of prescription errors by junior doctors in 20 hospitals in England, Dornan et al found that errors at admission were higher than at any other point in the patient stay.1 For these reasons, accurate reconciliation of medicines at admission has been core to health improvement strategies and a quality improvement goal for many health economies for almost 10 years. In England there is currently a drive to make medicines reconciliation (MR) a national measure of patient safety.
MR has been defined by the Institute of Healthcare Improvement as “being the process of identifying the most accurate list of a patient's current medicines—including the name, dosage, frequency and route—and comparing them to the current list in use, recognising any discrepancies and documenting any changes, thus resulting in a complete list of medications, accurately communicated”.9 In England national guidance stipulates that policies should ensure that “pharmacists are involved in medicine reconciliation as soon as possible after admission” and that “the responsibilities of pharmacists and other staff in the medicines reconciliation process are clearly defined”.10 This guidance was underpinned by a systematic review of the effectiveness and cost-effectiveness of interventions aimed at preventing medication errors at admission.11
The challenges associated with delivering accurate and timely MR are well recognised and pose a significant problem to organisations worldwide.12
To benchmark progress against national guidance, a collaborative baseline audit was facilitated across East and South-East England in January 2010. Fifty-six acute hospitals with a total of 33 120 beds participated. Nine thousand one hundred and one acute hospital admissions were audited. Eighty-seven per cent of the patients had received a pharmacy-led MR (pMR) but only 52% within 24 h of admission.13 An average of 1.32 unintentional discrepancies per pMR were documented, 73% of which were omitted medicines and 14% wrong doses.14
A service evaluation was next proposed as a way of identifying ways of optimising service delivery for maximum patient benefit. It aimed to:
-
Capture data on the process outcomes of pMR in different care areas
-
Capture data on factors that might impact on the quality of admission prescribing (defined as: total number of medicines prescribed for the patient; whether it was a planned or emergency admission; whether patient brought own medicines into hospital when admitted).
Methods
It was established that this was a service evaluation and research ethics approval was not required.
Data collection
A data collection form was designed and piloted. Pharmacy departments within East and South-East England were invited to participate and 30 acute hospitals prospectively collected data over a 3-week period in September 2010. Participants could choose whether they evaluated all or a selection of the beds they offered a pMR service to. pMR was carried out according to each organisation's standard operating policy for the service.
A nominated coordinator in each participating trust collated the data collected on the wards by pharmacy staff. Collated data were returned to the project coordinator (the author) and entered onto Excel spreadsheets.
For each pMR information was collected on: the care area; unintentional discrepancies (UDs) between the preadmission and admission prescription which were either omitted medicines (defined as medicines taken before admission but not accounted for on the medication chart or in the care notes) or medicines prescribed at the wrong dose; and the time taken to carry out the pMR. The service evaluation focused on omitted and wrong dose discrepancies as they comprised 87% of the discrepancies in the audit carried out a few months earlier and because they have a clear impact on patient care, so could thus provide a comparator across care areas.
For each UD of omission or wrong dose identified during the reconciliation the following data were recorded: British National Formulary (BNF) clinical category15; whether the admission was planned or unplanned; whether the pMR was completed within 24 h of the patient's admission; whether the patient had brought their medicines from home into hospital with them; number of medicines taken in total, as established by pMR (four or less vs five or more); the potential clinical impact of the prescribing omission or wrong dose error.
The potential clinical impact of each UD was rated by the practitioner completing the pMR or the nominated trust coordinator. Definitions of impact were based on those used by the National Reporting and Learning System for England and Wales incident reporting scheme16:
-
Level 1. None/Insignificant: No harm would have occurred to the patient.
-
Level 2. Low/Minor: Would have caused minimal harm. May have required extra observation or minor treatment.
-
Level 3. Moderate: Could have resulted in a moderate increase in treatment with significant but not permanent harm to the patient.
-
Level 4. Severe/Major: Could have resulted in permanent harm.
Medication-related examples were offered to help ensure participants were able to distinguish between the different levels of potential harm. For each UD ranked by participants as ‘moderate’ or ‘severe’ (Level 3 or 4) the following extra data were recorded: name of medicine; whether omitted or wrong dose; detail that may explain the ranking of the clinical impact. Clinical impact decisions were not peer reviewed.
Data analysis
The data were grouped into 10 care areas (figure 1). Specialist Medicine included data from HIV, oncology, renal, infectious diseases, sleep and endocrine wards. Specialist Surgery included data from gynaecology, cardiac and thoracic, GI surgery, transplant, urology, vascular, maxillofacial, burns and eye units. ‘Other’ comprised maternity, paediatric and private beds.
Number of omitted drugs and wrong doses identified per medicines reconciliation (MR) for each care area (number of reviewed MRs in brackets) compared with the total sample (N=3086).
Data for high risk drugs were reviewed separately. High risk drugs were defined as medicines which are the subject of a national safety alert or where short omissions could lead to destabilisation of the patient's clinical condition. The list included: warfarin; methotrexate; insulin; steroids and immunosuppressants; antiepileptics; antiparkinson agents.
As the patients were collected by participants from a wide variety of care areas, and the sample size was large, the data collected were assumed to be normally distributed. Analysis of variance using the two-tailed Z test was used to determine whether there were significant differences between the number of Level 3 UDs per patient in each of the care groups, as compared with the total population reviewed, and whether the probability of an identified UD having the potential to cause moderate harm varied between care groups.
Results
All hospitals chose to collect pMR data in a selection of care areas. The care area most frequently reviewed was the admissions unit (a bedded area from which patients may be discharged following a short stay or transferred to a more specialist ward for a longer stay).
Overall, 3086 pMRs were reviewed and 4041 omitted medicine and wrong dose UDs were identified. This represented a mean of 1.3 UDs per pMR reviewed. All the care areas included data from at least eight hospitals; however, the amount of data submitted by individual organisations for a particular care area was often small, resulting in four of the chosen care areas having less than 150 pMRs reviewed overall. Figure 1 displays the number of omitted dose and wrong dose UDs per MR by care area.
Table 1 reports the number of omitted medicines and wrong doses for 13 BNF categories for each of the care areas evaluated: No omitted medicines and wrong doses were identified for products in BNF category 14 (Immunological products and vaccines) or 15 (Anaesthesia).15
All identified unintentional omitted medicines and wrong doses by British National Formulary (BNF) category
No UDs in the population studied were deemed to be Level 4. Table 2 describes the UD distribution by care area. An average of 0.52 (range 0.21–1.18) Level 3 UDs and 0.005 (range 0.003–0.151) high risk drug UDs were recorded across the total population. Proportionally, Level 3 UDs represented 40% of overall discrepancies (range 31–52% across care areas). Analysis of variance (two-tailed Z test) indicated that there was >99% confidence that the differences in Level 3 UDs between the population average and the individual care group average were significant differences for all groups except respiratory medicine. Two-tailed Z tests also demonstrated that there was >99% confidence that the probability of a discrepancy being a Level 3 UD was higher in Care of the Elderly and General Surgery patients and lower in Cardiology and General Medical patients.
Analysis of unintentional discrepancies (UDs) by care area
The largest proportion of Level 3 UDs for most care areas related to cardiovascular medicines. The 260 omitted medicines and wrong dose errors associated with cardiovascular medicines recorded for patients on admission units were interrogated further: 23% related to medicines used to treat hypertension and heart failure; 18% comprised nitrates, calcium channel blockers and other antianginals; 15% lipid regulating agents; 11% β-adrenoceptor blocking agents; 11% diuretics; and 10% antiplatelets.
Overall 75% of recorded UDs were for patients taking five or more medicines (range 59–86% across care areas, information missing from 4.2% of returns); 40% of patients with UDs had brought some or all of their medicines into hospital with them (range 31–64% across care areas) and 80% of patients with a recorded UD had had unplanned admissions (range 27–93% across care areas, information missing from 2.8% of forms) (table 3).
Relationships between care area, Level 3 (moderate clinical impact) unintentional discrepancies (UDs), planned or unplanned admission, regular use of five or more medicines, and possession of own medicines in hospital (patient's own drugs)
The time frame within which pMR had been undertaken varied across the care areas. Overall an average of 65% of pMRs had been completed within 24 h of admission (range 34–83% across care areas, information missing from 2% of forms). Care areas with a higher proportion of pMRs completed within 24 h included: Admissions units (73%); Specialist surgery (73%), General surgery (70%) and Cardiac (68%). The average time to complete a pMR across the care areas reviewed was 15 min.
Discussion
A recent systematic review to identify the most successful method of delivery of MR in a hospital setting concluded that there are limited data and a lack of rigorously controlled studies comparing different approaches to MR.17 The authors noted that higher quality studies are needed but that available evidence supports interventions that heavily use pharmacy staff and that focus on patients at high risk for adverse events.17
This is the first collaborative service evaluation which compares measures related to pMR across a variety of care areas with the aim of using the information to identify patients at the highest risk for adverse events in order to inform improved service delivery.
The data collected from the 30 sites indicate that pharmacy teams in a naturalistic setting identify a mean of 1.3 omission and wrong dose UDs per pMR (range 0.6–2.1 across 10 care areas). The mean number of UDs per patient identified in this evaluation was similar to that observed in an earlier audit in the same geography14 and also to results obtained in published studies in defined patient groups.4–6 The data thus substantiate the added value of pharmacy involvement in MR.
Total numbers of UDs identified cannot alone provide a clear indication of the added value of a pMR in terms of averted adverse events, nor indicate whether UDs in some care areas are likely to have more impact on patient care. It is well recognised that not all UDs identified during MR will lead to an adverse drug event (ADE) and some authors have used conversion rate of discrepancies into prescription changes as a proxy marker for prevented ADEs.3 ,6 In this service evaluation a level of potential clinical impact was assigned to each discrepancy and an impact of Level 3 (moderate) used as a proxy marker to compare the potential for adverse events.
Although some differences were noted between care areas, unintentional discrepancies with the potential for moderate clinical impact and high risk drug errors were seen in all care areas reviewed. The data thus indicate that prioritising pMR service delivery by care area alone is not a safe option.
The potential limitations to this methodology are threefold. The pMRs reviewed were selected by the organisations themselves. It is thus possible that the clinical areas chosen had already been prioritised by the organisation to receive a pMR service. Despite this limitation, the pMRs evaluated overall covered a wide range of care areas so comparison of outcomes between care areas was possible.
Two further limitations are that the clinical impact category assigned to UDs was not peer reviewed and the conversion rate of UDs to prescription changes was not recorded. Nevertheless, the criteria for, and proportion of, overall UDs judged in this evaluation to have the impact for producing moderate harm were similar to those in studies where clinical significance was peer reviewed,3 ,4 ,6 and the rating levels are supported by the observation that 80% of the Level 3 UDs identified were for medicines in four BNF categories associated primarily with the management of long-term conditions (table 1): cardiovascular (39%), central nervous system (17%), endocrine (14%) and respiratory (9%). These four categories of medicines accounted for 64% of primary care prescribed items in England in 2010.18 Research studies have also noted high levels of UDs for cardiovascular and central nervous system medicines.4–6
Factors impacting on MR accuracy at the point of admission
If care area alone is not a safe way to prioritise patients for pMR, it may be feasible to use other issues to refine the prioritisation process. It has been proposed that increased age and increased number of prescription medications are risk factors for admission medication errors, and that the presentation of a medication list or actual medication is a protective factor for avoiding errors.6 ,17
In this service evaluation although patient age and the actual number of prescribed medicines were not recorded, patients were divided into one of two groups: on four or less, or five or more medicines. Over 75% (range 59–86%) of all Level 3 UDs were for patients on five or more medicines; however, it is important to note that over 20% of Level 3 UDs (range 14–41%) were in patients on five or less drugs. Cornish et al4 screened patients and only continued with pMR if the patient reported taking four or more medicines. It would appear from this service evaluation that prioritising MR based on number of regular medications taken also has limitations in terms of ensuring patient safety.
This evaluation of pMR is the first to include large numbers of patients with planned admissions, primarily for elective surgery. Planned admissions have the potential to support accurate MR at the point of admission as such patients often attend preadmission clinics where a medication history is recorded and they are usually asked to bring regular medicines into hospital with them, yet the data from this evaluation demonstrate that the groups of patients with the highest rates of planned admissions (specialist surgery and orthopaedics) had similar, or higher, rates of Level 3 UDs as care areas accepting predominantly unplanned admissions. Furthermore, over 40% of patients with a Level 3 UD had brought regular medicines into hospital with them, whether the admission was planned or unplanned, suggesting that admitting clinicians may not be making best use of available medicines-related information when prescribing at admission.
Improved multiprofessional working around admission processes for elective patients could be key to improving patient safety and to the optimisation of pMR services. Qualitative analysis into the causes of prescribing errors made by junior doctors in England noted that there was an absence of what was deemed a ‘safety culture’ among junior prescribers, with doctors relying heavily on pharmacists and nurses to identify and correct errors.1 A recent study examined the barriers to MR as perceived by healthcare professionals.19 Semistructured interviews highlighted that there was insufficient knowledge among the doctors, nurses and pharmacists interviewed about this healthcare problem and its solution, and that some clinicians were not convinced of the benefits of MR as evidence to support its implementation was minimal.19 Other barriers identified included unclear task allocation, with no clear agreement between professionals on tasks and responsibilities, coupled with no feedback on performance. This led to different approaches to delivering MR within an organisation, and replication of tasks such as taking the drug history. The authors concluded that a multidisciplinary implementation plan was required to optimise benefits to patient safety.19
Improved understanding of the importance of accurate MR at admission by admitting clinicians, plus training and support in how best to ensure reconciliation is accurate and when to refer to a pharmacy team may thus be key to improving MR at admission, particularly in areas where medicines or medicines lists are readily available. This may prove superior to a prioritisation process implemented solely by pharmacy teams based on certain preagreed criteria.
For certain patient groups, such as elective admissions, pharmacy resource may also be better used in supporting accurate admission prescribing through education and training rather than by actually carrying out pMR. Use of a structured form (paper or electronic) to gather relevant information has been proposed as a method to support reconciliation and may be a useful strategy in some settings, particularly preadmission clinics.5 ,20–23 In addition, organisations can consider implementing a policy which ensures priority pMR to patients taking nominated high risk drugs, such as those subject to safety alerts.16
Accurate MR is not just the responsibility of healthcare professionals admitting a patient: patients must also be encouraged to play their part. Many patients admitted to hospital in England now see a number of healthcare providers for their long-term and acute care. As in other countries worldwide, electronic or paper transfer of medicines information cannot always be relied upon to be accurate and patients admitted to hospital may not be well enough, or have sufficient health literacy to clarify any queries. There is thus a need to ensure patients take responsibility for keeping an up-to-date list of all their medications (including those purchased as well as those prescribed) at all times. This should be supported by health professionals22 and facilitated by media campaigns. Patients can also be advised always to take supplies of regular medicines into hospital with them.
Assigning benefit to pharmacy-led MR
The benefits of accurate MR currently appear to be insufficiently appreciated by clinicians and managers.18 Better sharing of local patient safety benefits and perhaps also the cost benefits to healthcare organisations, together with feedback on performance, could prove more successful in improving the quality of MR than national or organisational directives to deliver the service.
Research studies have attempted to quantify the clinical impact of accurate MR on patient care. Using a computerised MR tool and admission process, Boockvar et al found that MR reduced potential ADEs (pADEs) on general medical units by 43%.24 In 2007 Karnon et al carried out a systematic review of interventions aimed at preventing medication errors at admission in order to derive a model to underpin the UK guidance on MR.11 This model calculated a baseline figure of 2.8 (1.5–4.5) pADEs per 1000 admission prescription orders, 75% of which could be averted by pharmacist-led reconciliation and 50% of which could be averted by a systematic approach or pharmacy technician involvement. These calculations are based on data on pADEs identified in a number of studies.
Applying the model derived by Karnon et al to the data collected in this large service evaluation of 3086 patients, and assuming an average of five items prescribed at admission per patient, approximately 43 (range 23–69) pADEs could have been anticipated during the inpatient stay for this patient group, of which pMR would have averted between 22–33 (range 11–52), depending on whether pMR involved pharmacists or pharmacy technicians.11
The calculations described above apply only to pADEs experienced during an inpatient stay. If medicines for long-term conditions are omitted unintentionally during an inpatient stay and then at discharge, it may result in the next prescriber assuming it is an intentional omission. Long-term omission of a medicine for a chronic condition can thus contribute to a pADE after discharge. Bell et al carried out a population-based cohort study involving 396 380 patients aged 66 years and older with continuous use of at least one of five evidence based medicines (statins, antiplatelets and anticoagulants, levothyroxine, respiratory inhalers, gastric acid suppressing drugs).7 They concluded patients admitted to hospital were more likely to experience potentially unintentional discontinuation of medicines than controls and that at 1 year follow-up patients who had had statins or antiplatelets or anticoagulant drugs discontinued had an elevated adjusted OR for secondary composite outcome measures of death, emergency department visit or hospitalisation.7
Conclusions
This large service evaluation in a naturalistic setting has demonstrated that pMR identifies a mean of 1.3 unintended prescribing discrepancies per patient and thus potentially averts approximately one clinically important preventable adverse event per 100 reconciliations undertaken.
Although discrepancies with the potential to cause moderate harm were identified more frequently in some care groups than others, they comprised at minimum 30% of the overall discrepancies in all the care groups reviewed. In addition discrepancies associated with high risk medicines were identified across all care groups. These results indicate prioritisation of pMR services according to care area cannot be recommended if patient safety is to be assured.
In this service evaluation neither the availability of the patient's medication nor the fact that it was a planned admission appeared to improve the accuracy of admission prescribing, or reduce the proportion of unintended prescribing discrepancies likely to have a moderate adverse clinical impact on patient care.
The difficulties associated with implementing accurate MR at admission, despite worldwide directives making it imperative, demonstrate the need for organisations to rethink their current policies and procedures at admission and to encourage a multidisciplinary, cohesive approach to delivering accurate MR across the whole organisation. In particular, tasks and responsibilities should be clear to all healthcare professionals. While pharmacy staffing levels remain insufficient to deliver pMR to all patients in a timely fashion, the focus for pharmacy staff may need to move from establishing accurate drug histories for all patients, which facilitates MR, to training and supporting doctors and nurses to establish accurate drug histories, referring for pharmacy support only when needed. Moving to full 7 day working practices may also provide valuable opportunities to embed timely MR more effectively into patient care.
Key messages
-
What is already known on this subject
-
Medicines reconciliation (MR) at transfer of care is a patient safety imperative in many countries.
-
Evidence indicates that accurate MR at admission to hospital is most cost-effectively delivered by pharmacy teams.
-
Pharmacy services in many organisations lack the resources to deliver timely MR to all admitted patients. This has resulted in pharmacy managers prioritising delivery by care area based on a perception of possible patient benefit.
-
What this study adds
-
This collaborative service evaluation across 30 hospitals found similar levels of prescribing errors with the potential for moderate clinical impact were made in all the care areas reviewed, thus demonstrating that prioritisation by care area alone is not a safe option.
-
Alternative strategies are suggested. All strategies should make better use of information available at the time of admission.
Acknowledgments
Christine Masterson for her work in collating and manipulating the data. Graham Dodds MMath for help and advice on the use of statistical tests. Acute trusts in East and South-East England for piloting the data collection forms and contributing data to the service evaluation.
References
Supplementary materials
Bulgarian version
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.
-
Data sharing statement Available data may be shared on request to the author.