Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Care homes’ use of medicines study: prevalence, causes and potential harm of medication errors in care homes for older people

Care homes’ use of medicines study: prevalence, causes and potential harm of medication errors in... Original research Care homes’ use of medicines study: prevalence, causes and potential harm of medication errors in care homes for older people 1 2 2 2 1 1 3 N D Barber, D P Alldred, D K Raynor, R Dickinson, S Garfield, B Jesson, R Lim, 1 2 3 4 1,5 5 I Savage, C Standage, P Buckle, J Carpenter, B Franklin, M Woloshynowych, A G Zermansky Department of Practice and ABSTRACT Policy, School of Pharmacy, Introduction: Care home residents are at particular risk Box 1 London, UK; School of from medication errors, and our objective was to Healthcare, University of Leeds, determine the prevalence and potential harm of Leeds, UK; Robens Centre for Context: Care homes and the English National prescribing, monitoring, dispensing and administration Public Health, University of Health Service Surrey, Guildford, Surrey, UK; errors in UK care homes, and to identify their causes. 4 Care homes may provide 24-hour nursing care London School of Hygiene and Methods: A prospective study of a random sample of (nursing homes), personal care only (residential Tropical Medicine, London, UK; residents within a purposive sample of homes in three Imperial College Healthcare homes) or a combination. They may be owned by NHS Trust, London, UK areas. Errors were identified by patient interview, note individuals or companies of various sizes, including review, observation of practice and examination of large private health providers, charities or by the Correspondence to: dispensed items. Causes were understood by observation local authorities. Care homes are reviewed against Professor N D Barber, and from theoretically framed interviews with home staff, standards by the Care Quality Commission, which is Department of Practice and Policy, School of Pharmacy, doctors and pharmacists. Potential harm from errors was the independent regulator of health and social care Tavistock House, Tavistock assessed by expert judgement. in England. Each resident is registered with a Square, London WC1H 9JP, UK; Results: The 256 residents recruited in 55 homes were general practitioner who provides their medical care [email protected] taking a mean of 8.0 medicines. One hundred and and keeps their medical record. When a patient seventy-eight (69.5%) of residents had one or more transfers from their own home to a care home, they Accepted 27 July 2009 errors. The mean number per resident was 1.9 errors. The can elect to keep their general practitioner if he or mean potential harm from prescribing, monitoring, she is local. administration and dispensing errors was 2.6, 3.7, 2.1 and In England, the National Health Service is delivered 2.0 (0 = no harm, 10 = death), respectively. Contributing through 152 primary care trusts, which are respon- factors from the 89 interviews included doctors who were sible for a geographical area of the country, for which not accessible, did not know the residents and lacked they commission healthcare from general practi- information in homes when prescribing; home staff’s high tioners (primary care physicians) who usually work workload, lack of medicines training and drug round as part of a group practice, pharmacies and others. interruptions; lack of team work among home, practice Pharmacies may be owned and run by a single and pharmacy; inefficient ordering systems; inaccurate pharmacist or may be part of a chain; the largest medicine records and prevalence of verbal communica- chains are run by international companies. Homes tion; and difficult to fill (and check) medication adminis- usually obtain their regular supply of medicines for all tration systems. their residents from one pharmacy. Conclusions: That two thirds of residents were exposed Repeat medicines are ordered from the GP practice to one or more medication errors is of concern. The will to (usually monthly) by the care home staff using the improve exists, but there is a lack of overall responsibility. previous 28-day medicine administration record or Action is required from all concerned. the repeat medicines slip provided by the GP practice. Generally the GP practice authorises and prints the repeat prescriptions and sends them to the care home for checking. They are then forwarded to the community pharmacy where they are dispensed and INTRODUCTION delivered to the care home with a new 28-day Older people living in care homes are potentially at medicine administration record. greater risk of medication error than most other groups. They are prescribed multiple medicines and this, coupled with age-related changes in pharma- sites; and medicines may be dispensed in, and cokinetics and pharmacodynamics, makes them administered from, one of several different types of particularly susceptible to adverse drug events. packaging systems known as ‘‘monitored dosage Many have some degree of cognitive impairment systems’’ (MDS). that prevents them from being actors in the In 2000, a report documenting the extent of detection of errors. In addition, the medicines medical error was published and the UK govern- management system in care homes (described in This paper is freely available ment committed to reducing errors ; medication box 1) is complex—for example, many residents online under the BMJ Journals receive clinical interventions from multiple errors were a particular concern. Prescribing has unlocked scheme, see http:// qshc.bmj.com/info/unlocked.dtl sources; medicines may be dispensed from multiple been found to be suboptimal in UK care homes ; Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 341 Original research however, we do not know the prevalence of medication errors in not, and if so, whether it was in a cassette MDS or blister- this setting in the UK. Such studies in the USA have been pack MDS. Two drug rounds per resident were observed to undertaken; however, they have often relied on spontaneous identify administration errors. The definitions and denomi- reporting, which significantly underestimates the prevalence nators of the four types of error are described in appendix A. 10 11 of error. A US study of medication administration errors To ensure inter- and intrarater reliability, we produced a using direct observation found an error rate of 22% in nursing handbook of definitions and procedures, and research staff 12 2 homes. Gurwitz et al found 4.1 preventable adverse drug were trained in homes together and met regularly to resolve events per 100 resident-months in two US long-term care difficult cases, which were then added to the handbook. If facilities. any errors were thought to be likely to cause patient harm, Although previous studies have measured prevalence of then the pharmacist intervened. specific types of medication errors, they have not been designed The potential harm of each error was assessed using a valid to look at all types of error and simultaneously understand their and reliable method. Each error was individually assessed by a 13–15 causes. The theory of causation of human error is widely GP, a consultant old age psychiatrist, a clinical pharmacologist used in healthcare and has been used to explain causes of and two clinical pharmacists, using a validated 10-point scale 16 17 medication errors previously. In this study, we wished to (0 = no harm; 10 = death); their mean score was taken as the determine the prevalence of all forms of medication errors in harm score for that error. care homes, to assess the potential of these errors for harm and to establish the underlying causes. Statistical analysis Sample size was calculated to give a precision of +/22% SD (95% confidence interval (CI)) if the prevalence of prescribing METHODS error was 10%. The estimation was derived from a combina- Ethical approval was obtained from the Central Office for tion of simulation and formulas for clustered data, and the Research Ethics Committees. output was a matrix of numbers of homes and number of patients per home (eg, 100 homes with three patients each Selection criteria/participants gave a precision of 1.9%; 50 homes with six patients each gave We chose to sample a large number of homes and small number a precision of 2.04%), allowing us to adapt the sample of residents per home. It is generally acknowledged that errors dependent on recruitment rates. Statistical analysis was are a consequence of the systems being used, so our sampling performed using the software R 2.3 (R project for statistical strategy was designed to increase the number of systems of care computing, www.r-project.org). Statistical ‘‘significance’’ was observed (homes, general practitioners (GPs) and pharmacies). predefined at the 5% level. Exact binomial CIs were calculated Care homes for older people (nursing, residential and mixed) for proportions. x tests were used to assess differences in error were sampled in three geographically spread and demographi- rates between areas. Generalised estimating equations (library cally diverse areas in the UK (West Yorkshire, Cambridgeshire geepack, V.1.0-10) were used to model patient level odds of and central London) to obtain a varied sample with respect to errors, allowing for clustering in homes and using an ownership, size and type of care provided. For each home we independence correlation structure. Multilevel models were approached the manager/head office to obtain cooperation and also used to model patient level odds of errors, using the obtained written consent from care home staff who were MLwiN 2.03 software (multilevel models project, University observed. Care home residents on one or more medicines were of Bristol, http://www.cmm.bristol.ac.uk), fitting variance randomly selected and included in the study if they provided components at the various levels. written, informed consent. For those lacking capacity, assent was obtained from the next of kin. Written, informed consent RESULTS was obtained from GPs providing a service to the homes, which Of those approached, 72% (79/108) of homes, 67% (269/399) of included access to their care home records and GP records. residents and 61% (54/89) of general practices agreed to take part. Two hundred and fifty-six residents were recruited from 55 care homes. The majority (38, 69%) of the homes provided Data collection both residential and nursing care (corresponding figures for We used a mixture of methods. The qualitative work was residential care only and nursing care only: 12 (22%) and 5 ethnographically informed and involved field notes, observation (9%)). Table 1 shows demographic data. The majority of the and semistructured interviews based on Reason and Vincent’s 14 15 residents were women (69%, 177/256) and very old (mean age frameworks. To help understand the causes of specific errors, 85 years). There were slightly more residential care residents we conducted interviews with the home staff, GP or pharma- (54%, 139/256) than nursing care residents. Two hundred and cist. Interviews were taped where possible and transcribed; twenty (86%) of the residents were dispensed some of their otherwise, notes were made that were expanded on after the medicines in a monitored dosage system. There was a mean of interview. Analysis of the causes of errors used Reason’s 14 18 framework. 3.8 (range 1–14) GP practices per care home. There was The quantitative work to identify prescribing and monitor- considerable variation between areas: in London, the median ing errors was undertaken by clinical pharmacists (one in was 1 (range 1–3) practice per home; in West Yorkshire, the each of the three areas) conducting clinical medication median was 5 (range 1–14). reviews for each randomly selected resident. This process We intended to interview those involved in specific errors, included review of GP and care home notes and consultation soon after they had happened. However, many errors (such as with the resident and/or staff. Dispensing errors were some prescribing errors) had been made some time in the past, identified by comparing the prescriptions and medication so most interviews with doctors and pharmacists focused more administration record sheets with the dispensed medicines. on problems in general in medicines prescribing, dispensing and For each drug, it was noted if it was contained in an MDS or use in care homes. We undertook 59 interviews relating to 66 342 Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 Original research Table 1 Demographic data No of residents approached 399 No of residents consented/assented (%) 269 (67.4) No of residents excluded 13 Consent given post cut-off 6 Died 4 In hospital 3 No of residents entered into study 256 Cambridgeshire 31 West Yorkshire 121 London 104 Women, no (%) 177 (69.1) Age (y), mean (range) 85.2 (60–102) Nursing residents, no (%) 117 (45.7) Residential residents, no (%) 139 (54.3) Mean no of medicines per resident (95% CI) 8.0 (7.5 to 8.5) Median no of medicines per resident (range) 7.5 (1–25) No of residents using monitored dosing systems (%) 220 (85.9) Cambridgeshire (%) 23 (74.2) West Yorkshire (%) 117 (96.7) London (%) 80 (76.9) administration errors, 34 dispensing errors, 18 prescribing and 8 monitoring errors. Further general interviews were carried out with 19 pharmacists and 11 GPs, and observations were carried out in five pharmacies. Table 2 details the prevalence and potential harm of errors. Overall, 178 of 256 residents (69.5%, 95% CI 63.5 to 75.1) had at least one medication error (prescribing, monitoring, adminis- tration or dispensing). There was a mean of 1.9 (95% CI 1.64 to 2.17) errors per resident. Appendix B gives examples of errors. Prescribing and monitoring errors One hundred residents (39.1%, 95% CI 33.0 to 45.3) had one or more prescribing errors, totalling 153 prescribing errors, and the prescribing error rate by opportunity for error was 8.3% (95% CI 7.1 to 9.7) (opportunity for error is an act that can be erroneous; in this case it would be the prescribing of one medicine or the monitoring of one). The most common types of prescribing error, accounting for 87.6% of the total, were categorised as ‘‘incomplete information’’ (37.9% meaning, eg, that no strength or route was specified when there was more than one option), ‘‘unnecessary drug’’ (23.5%), ‘‘dose/strength error’’ (14.4%) and ‘‘omission’’ (11.8%). In total, there were 147 residents who were prescribed a medicine that required monitoring, and 18.4% (27) of these had an error. There were 32 monitoring errors in the 218 prescribed items that required monitoring (14.7%). The mean number of monitoring errors per resident was 0.13 (95% CI 0.08 to 0.17). There was significant variation between areas, with 75% of monitoring errors occurring in just one geographical area (p,0.01). Nearly one third (30.8%) of medicines deemed to require monitoring in the problem area were not being monitored. The great majority of monitoring errors (90.6%) resulted from a failure to request monitoring. The drugs most commonly associated with monitoring errors were diuretics (53.1%), ACE inhibitors (15.6%), amiodarone (12.5%) and levothyroxine (9.4%). The mean harm scores for prescribing and monitoring errors were 2.6 (range 0.2 to 5.8) and 3.7 (range 2.8 to 5.2), respectively. The interviews categorised factors thought to contribute to the errors as the patient, the task, the team and the work environment. Patient factors related to prescribing errors Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 343 Table 2 Error prevalence and harm assessment Type of error Prescribing Monitoring MAE Dispensing Overall Number of residents with an error (n = 256) (%; 95% CI) 100 (39.1%; 33.0 to 45.3) 27 (10.5%; 7.1 to 15.0) 57 (22.3%; 17.3 to 27.9) 94 (36.7%; 30.8 to 42.9) 178 (69.5%; 63.5 to 75.1) Mean number of errors per resident (95% CI) 0.60 (0.48 to 0.71) 0.13 (0.08 to 0.17) 0.45 (0.32 to 0.58) 0.73 (0.56 to 0.90) 1.91 (1.64 to 2.17) Median number of errors per resident (range) 0 (0–6) 0 (0–3) 0 (0–7) 0 (0–9) 1 (0–13) Number of errors/opportunity for error (%; 95% CI) 153/1837 (8.3%; 7.1 to 9.7) 32/218 (14.7%; 10.3 to 20.1) 116/1380 (8.4%; 7.0 to 10.0) 187/1915 (9.8%; 8.5 to 11.2) 488/5350 (9.1%; 8.4 to 9.9) Mean harm score* (95% CI) 2.6 (2.4 to 2.8) 3.7 (3.4 to 4.0) 2.1 (1.9 to 2.3) 2.0 (1.8 to 2.2) 2.6 (2.5 to 2.7) Median harm score* (range) 2 (0.2 to 5.8) 3 (2.8 to 5.2) 2 (0.1 to 5.8) 1 (0.2 to 6.6) 2 (0.1 to 6.6) MAE, medication administration error. *Assessed by a 10-point scale (0 = no harm; 10 = death). Original research included that patients were generally home bound and could dedicated time to order medicines. In some homes, 12-hour not access services external to the home; some patients also shifts were usual. disliked blood tests or taking some medicines. Some patients were confused and unable to give histories. Task factors related Dispensing errors to lack of usual prescribing technical support, including Ninety-four residents (36.7%, 95% CI 30.8 to 42.9) had a total of computer-based aids and accessing the medical record. There 187 dispensing errors with a mean of 0.73 (95% CI 0.56 to 0.90) were problems with practice computers’ failure to prompt dispensing errors per resident. The dispensing error rate by monitoring tests clearly. It could be hard to get blood tests done. opportunity for error was 9.8% (95% CI 8.5 to 11.2). Labelling There were many ‘‘team factors’’ as there was little sense of errors were found in 7.3% of dispensed items, content errors in being a whole team (home, pharmacy and practice staff). The 2.3% and clinical errors in 0.21%. There was a borderline service offered by GP practices was very variable, from a statistically significant difference in the odds of a dispensing dedicated GP making weekly visits, to GPs with no knowledge error according to delivery system (p = 0.056), largely due to the of the patient making a home visit when requested. Hospital higher odds with the cassette type monitored dosage systems out-patient and discharge letters were sometimes unclear, (cassette vs blister: adjusted odds ratio, 2.88, 95% CI 1.5 to 5.55, delayed, missed or not adequately incorporated in the patient p = 0.0012), which was associated with more labelling errors. record. When changing between GP practices, patient notes The mean harm score was 2.0 (range 0.2–6.6). took up to 4 months to arrive. GPs often expressed concern Task factors included the computer systems used, with about the care home staff, including turnover and staff some identifying all interactions rather than the clinically shortages. The skills of care home staff were sometimes seen significant ones. Some monitored dosage systems did not have as low, which several linked to pay. the space to fit in all the required warning labels, and they were tedious and difficult to fill. The similar appearance of Medication administration errors many tablets when removed from their original container was Fifty-seven (22.3%, 95% CI 17.3 to 27.9) residents had a total of a potential source of error. The prescription and medication 116 administration errors. The mean number of administration administration record were often different, so it was unclear errors per resident was 0.45 (95% CI 0.32 to 0.58) and the which was correct. Individual factors included staff feeling prevalence of administration errors by opportunities for error hungry, tired or unwell while dispensing monitored dosage was 8.4% (95% CI 7.0 to 10.0). Nearly half (49.1%) of all systems. There was lack of knowledge by pharmacy staff of administration errors were categorised as ‘‘omissions’’ and just the care homes’ systems and their need for support. Team more than one fifth (21.6%) were ‘‘wrong dose’’. The odds of a factors overlapped with this—some pharmacists had no medication administration error occurring were higher in knowledge of the care home and its requirements, relation- residential care than in nursing care residents; however, this ships were sometimes tense and poor language skills of home just failed to reach statistical significance at the 5% level staff were sometimes cited. Some pharmacies had a poor (adjusted OR 1.77, 95% CI 0.96 to 3.25, p = 0.063; adjusted for checking process, and use of locums could be a problem. Work age, sex and medication delivery system). The higher apparent factors included the pharmacies being seen as being busy and risk of administration errors in residential compared with pressured, with interruptions and distractions (including nursing residents was largely attributable to more ‘‘omissions’’ noise) and some staff shortages. (38 vs 19) and ‘‘wrong doses’’ (18 vs 7). There was no statistically significant difference in the administration errors by the medicine delivery system. The mean harm score was 2.1 Organisational culture (range 0.1–5.8). When considering all forms of medication errors, factors relating Patient factors included many patients’ lack of awareness of to organisational culture (called ‘‘latent failures’’ by Reason ) their medicines. In addition, their physical condition could make were significant. It was clear from the interviews that no one it hard to administer medicines properly. Some patients had took responsibility for the whole system. We often saw well- fears about medicines, such as feeling they were being poisoned, intentioned people doing their best but in an uncoordinated and some were consequently aggressive. Finding mobile patients way. Communication, written and verbal, was another proble- during the drug round could be a problem. Changing or adding matic factor, within and among the home, GP practice and medicines, such as acute treatment, in the middle of the 4-week pharmacy. Consequently, it is difficult to know which supply cycle could be a problem. Task factors included inability medicines any patient should be having. Management within to find the medicine, failure to order the right quantity of ‘‘as each organisation was a factor, particularly when challenged to required medicines’’, the special requirements that some deliver a safe service within a tight budget. medicines had (eg, ‘‘take on an empty stomach’’), the difficulty many staff had in correctly administering inhalers and a lack of DISCUSSION adequate protocols. Individual factors related to the staff People in care homes are a frail and vulnerable population at included lack of knowledge about inhalers and the timing of particular risk from medication errors, and it is a cause for medicines with respect to food. Team factors included the concern that two thirds of care home residents in this study medication administration record chart, which should be the were exposed to one or more errors. For each event involving documentary line of communication among GP, home and prescribing, dispensing or administration of a medicine, there pharmacist; these records were often inaccurate. Communication within the home tended to be verbal. Work was an 8%–10% chance of an error happening and a 14% chance environment factors included homes being hot, airless, having of a monitoring error. Safety is a systems issue, and we believe unpleasant smells, being poorly lit, noisy and short of space. this is the first study to consider the whole system of There were often staffing problems in the morning round medication use in care homes; our simultaneous collection of (when most medicines were given and when staff also had most qualitative data has allowed us to understand the causes of error other tasks). Staff were frequently interrupted and did not have and suggest solutions. 344 Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 Original research The prevalence of prescribing error is similar to that found in unclear and some commissioners discourage its use. The use of primary care ; administration error prevalence was a little MDS drives efficiencies of scale, such as large centralised higher than that in hospital (and likely to be better than the repackaging units, which in turn leads to the dispensary patients’ adherence if in their own home). The prevalence of becoming remote from the customers (home and patient). dispensing errors was three times higher than the rate found in Research into the effectiveness of MDS is urgently required. primary care in the UK, although that study excluded MDS. Within homes the use and accuracy of the medication Our higher rate predominantly reflected one type of MDS that administration record requires constant review. The lack of was difficult to label fully. protocols and adequate staff training remains an issue. Drug rounds are very busy, and often interrupted in the morning, and Although our study was not primarily designed to identify some medicines should be prescribed for different times to ease the prevalence of harm, we saw several errors, particularly this. The commonest administration errors were omissions monitoring errors, which had caused harm or were likely to. In because the drug was not available, so omissions need to be addition, many errors would reduce the quality of life and ability to function of residents, such as inadequate treatment of monitored and ordering, particularly of ‘‘as required’’ medicines, pain, of bowels and of breathing. needs to be improved. We were very impressed by the proportion of homes Limitations to our study include that our sample only participating in a study, which was potentially very threatening contained those willing to be studied (although the acceptance to them. Several care home managers have told us that patient rate of homes was high) and that our home sampling was not harm from medication error is their greatest fear and that up to random. Judgement of the cause of error was sometimes half of staff time can be spent on medication related activities. difficult as there could be conflicting sources of evidence or a Given this motivation and resource, we are hopeful of change. lack of evidence; hence, judgements sometimes retained an element of subjectivity. Observation may theoretically have Acknowledgements: We thank the care home staff, residents, relatives, community affected the prevalence of administration error, although pharmacists and general practitioners. routine observation has been found to have no effect. Staff Funding: The study was funded by the Patient Safety Research Programme of the interviewed will have given accounts affected by hindsight bias; Department of Health. The authors are independent of the funders. The sponsor hence, imputations of causality are speculative. approved the study design. All authors had full access to all the data and can take What can be done, and who should do it? As our study responsibility for the integrity of the data and the accuracy of the data analysis. shows, there are currently many and varied subsystems that are Competing interests: None. not being seen in an integrated way. There is now the opportunity for a systems approach to the whole. Since 2008, REFERENCES chief pharmacists of provider organisations and commissioners 1. Zermansky AG, Alldred DP, Petty DR, et al. Clinical medication review by a in England should have the lead role in ensuring safe medication pharmacist of elderly people living in care homes—randomised controlled trial. Age practices are embedded in patient care ; a significant and Ageing 2006;35:586–91. 2. Gurwitz JH, Field TS, Judge J, et al. The incidence of adverse drug events in two pressing agenda for them. An additional system-based solution large academic long-term care facilities. Am J Med 2005;118:251–8. relates to several of the communication and records problems 3. Alldred DP, Petty DR, Bowie P, et al. Antipsychotic prescribing patterns in care observed—we would hope these would be ameliorated by homes and relationship with dementia. Psychiatr Bull 2007;31:329–32. programmes in the National Health Service’s information 4. Expert Group on Learning from Adverse Events in the NHS. An organisation with a memory. London: The Stationary Office, 2000. technology programme (NPfIT), such as the Summary Care 5. Department of Health. Building a safer NHS for patients. London: Department of Record (a brief GP record which can be accessed by others), Health, 2001. GP2GP (electronic transfer of patients’ notes between GPs) and 6. Fahey T, Montgomery AA, Barnes J, et al. Quality of care for elderly residents in nursing homes and elderly people living at home: controlled observational study. BMJ the Electronic Prescription Service (electronic transfer of 2003;326:580–3. primary care prescriptions). The final system issue is that most 7. Alldred DP, Zermansky AG, Petty DP, et al. Clinical medication review by a primary care is based on patients going to centres of care rather pharmacist of elderly people living in care homes: pharmacist interventions. than the other way around. Primary care services that are based Int J Pharm Pract 2007;15:93–9. 8. Pierson S, Hansen R, Greene S, et al. Preventing medication errors in long-term on care going to patients need to be commissioned, in order not care: results and evaluation of a large scale web-based error reporting system. Qual to disadvantage the home bound. Saf Health Care 2007;16:297–302. We suggest the idea of a lead (not sole) GP for each home 9. Greene S, Williams C, Hansen R, et al. Medication errors in nursing homes. J Pat Saf 2005;1:181–9. should be explored. This role would need protected time and 10. Barker KN, McConnell WE. The problems of detecting medication errors in hospitals. associated funding. In addition to caring for patients, they Am J Hosp Pharm 1962;19:360–9. should liaise with other GPs and have responsibility to ensure, 11. Dean B, Barber N. Validity and reliability of observational methods for studying possibly by commissioning services, that patients on riskier medication administration errors. Am J Health System Pharm 2001;58:54–9. 12. Barker KN, Flynn EA, Pepper GA, et al. Medication errors observed in 36 healthcare medicines are appropriately monitored and that all patients’ facilities. Arch Int Med 2002;162:1897–903. medication is regularly reviewed by a pharmacist. 13. Rasmussen J, Jensen A. Mental procedures in real life tasks: a case study of Consideration should be given to having one person with electronic trouble-shooting. Ergonomics 1974;17:293–307. 14. Reason J. Human error. Cambridge: University of Cambridge, 1990. overall responsibility for medicines use in one or more care 15. Vincent C, Taylor-Adams SE, Stanhope N. Framework for analysing risk and safety in homes. Many pharmacists have the skills and knowledge to clinical practice. BMJ 1998;316:1154–7. undertake this role, and such developments are described in the 16. Dean BS, Schachter M, Vincent C, et al. Prescribing errors in hospital inpatients— UK government’s recent proposals for making best use of incidence and clinical significance. Qual Saf Health Care 2002;11:340–4. 17. Taxis K, Barber N. Causes of intravenous medication errors: an ethnographic study. pharmacists’ expertise. Qual Saf Health Care 2003;12:343–7. Pharmacists supplying homes should ideally know the home, 18. Dean B, Schachter M, Vincent C, et al. Causes of prescribing errors in hospital its ways and needs, so that ordering and supply match the inpatients: a prospective study. Lancet 2002;16:271–8. 19. Dean B, Barber N. A validated, reliable method of scoring the severity of medication home’s (and patients’) needs. The widespread use of MDS unit errors. Am J Health System Pharm 1999;56:57–62. dose systems, requiring millions of tablets to be repackaged each 20. Reason J. Managing the risks of organisational accidents. Aldershot: Ashgate, 1997. week, is a vast, unfunded undertaking. It imposes demands on 21. Shah SDH, Aslam M, Avery AJ. A survey of prescription errors in general practice. home and pharmacy alike, yet its contribution to safety is Pharm J 2001;267:860–2. Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 345 Original research 22. Dean BS, Schachter M, Vincent C, et al. Prescribing errors in hospital inpatients— and Dean and Barber as ‘‘any deviation between the medication prescribed and that incidence and clinical significance. Qual Saf Health Care 2002;11:340–4. administered’’. The number of opportunities for error (denominator) was the number of 23. Horne R, Weinman J, Barber N, et al. Concordance, adherence and compliance in doses given, plus any doses that should have been given but were omitted. medicine taking. London: NCCSDO, 2005. 24. Franklin BD, O’Grady K. Dispensing errors in community pharmacy: frequency, clinical significance and potential impact of authentication at the point of dispensing. APPENDIX B: EXAMPLES OF ERRORS Int J Pharm Pract 2007;15:273–81. 25. Department of Health. Pharmacy in England. Building on strengths—delivering the future. London: DoH, 2008. Case 1: Prescribing error 26. Dean B, Barber N, Schachter M. What is a prescribing error? Qual Saf Health Care Tramadol capsules 50 mg prescribed ‘‘one to be taken up to four times a day’’ for 2000;9:232–7. chronic foot pain, resident also taking warfarin for long-term DVT prophylaxis 27. Alldred DP, Standage C, Zermansky AG, et al. Development and validation of criteria (Tramadol can enhance the effect of warfarin). International normalised ratio checked to identify medication-monitoring errors in care home residents. Int J Pharm Pract regularly with erratic results ranging from 0.9 to 4.5 (mean harm score 5.8). 2008;16:317–23. 28. Beso A, Franklin BD, Barber N. The frequency and potential causes of dispensing errors in a hospital pharmacy. Pharm World Sci 2005;27:182–90. Case 2: Prescribing error 29. Allan EL, Barker KN. Fundamentals of medication error research. Am J Hosp Pharm Donepezil 10-mg tablets prescribed ‘‘one daily’’ for dementia. Following a review from 1990;47:555–71. hospital specialist, a letter to the GP indicated Aricept (brand name for donepezil) should be stopped. The donepezil had not been discontinued by the GP and there was APPENDIX A: ERROR DEFINITIONS no indication in the medical notes that the continuation was deliberate (mean harm Prescribing errors were identified and classified according to the definition developed score 1.8). by Dean et al as ‘‘A prescribing decision or prescription-writing process that resultsin 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 Case 3: Monitoring error practice’’. The number of opportunities for error (denominator) was the number of Lisinopril 5-mg tablets prescribed ‘‘one daily’’ for hypertension for a resident with an prescription items written, plus any omissions. The three pharmacists worked to a estimated creatinine clearance of 19 ml/min. A potassium level had been checked common detailed protocol when reviewing the residents and their medicines, were 1 year ago and had revealed a high potassium level 5.8 mmol/l (range 3.5–5 mmol/l): trained together at the start of the study and had regular review meetings to ensure no action had been taken (mean harm score 5.8). consistency. Monitoring errors were identified according to the definition developed and validated by Alldred et al as ‘‘A monitoring error occurs when a prescribed medicine is not Case 4: Monitoring error monitored in the way which would be considered acceptable in routine general Amiodarone 200-mg tablets prescribed ‘‘one daily’’ for atrial fibrillation. Thyroid practice. It includes the absence of tests being carried out at the frequency listed, with function tests were last checked 9 months ago, when thyroid stimulating hormone a tolerance of +50%. This means—for example, that if a medicine requires liver was 12.9 mlU/l (range 0.3–5.5 mlU/l) and thyroxine was 18 nmol/l (range 50– function tests at three monthly intervals, an error would occur if a test was not 151 nmol/l), and no action had been taken (mean harm score 5.8). conducted within 18 weeks’’. The number of opportunities for error (denominator) was the number of prescribed items that required monitoring, according to the validated criteria. Case 5: Medication administration error The definition of a dispensing error developed by Beso et al was adopted. A Bendroflumethiazide 2.5-mg tablets prescribed ‘‘one each morning’’ for hypertension. dispensing error was defined as ‘‘One or more deviations from an interpretable written This was discontinued due to low serum sodium of 127 mmol/l (range 135– prescription or medication order, including written modifications to the prescription 145 mmol/l); however, it remained on the current medication administration record made by a pharmacist following contact with the prescriber’’. Dispensing errors were when the next monthly drug supply was made and hence continued to be identified and classified by the clinical pharmacists by comparing the prescriptions and administered (mean harm score 4.6). medicine administration record sheets with the dispensed medicines. The number of opportunities for error (denominator) was the number of prescription items dispensed or omitted. Case 6: Dispensing error The clinical pharmacists observed two drug rounds per resident to identify and classify Aspirin enteric-coated 75 mg tablets dispensed instead of zopiclone 7.5 mg tablets as medication administration errors as defined using previous work by Allan and Barker a 7-day supply in a cassette dispensing system (mean harm score 5.0). 346 Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Quality and Safety in Health Care Pubmed Central

Care homes’ use of medicines study: prevalence, causes and potential harm of medication errors in care homes for older people

Download PDF
6 pages

Loading next page...
 
/lp/pubmed-central/care-homes-use-of-medicines-study-prevalence-causes-and-potential-harm-jHxxIjaC70

References (41)

Publisher
Pubmed Central
Copyright
© Barber et al 2009
ISSN
1475-3898
eISSN
1475-3901
DOI
10.1136/qshc.2009.034231
Publisher site
See Article on Publisher Site

Abstract

Original research Care homes’ use of medicines study: prevalence, causes and potential harm of medication errors in care homes for older people 1 2 2 2 1 1 3 N D Barber, D P Alldred, D K Raynor, R Dickinson, S Garfield, B Jesson, R Lim, 1 2 3 4 1,5 5 I Savage, C Standage, P Buckle, J Carpenter, B Franklin, M Woloshynowych, A G Zermansky Department of Practice and ABSTRACT Policy, School of Pharmacy, Introduction: Care home residents are at particular risk Box 1 London, UK; School of from medication errors, and our objective was to Healthcare, University of Leeds, determine the prevalence and potential harm of Leeds, UK; Robens Centre for Context: Care homes and the English National prescribing, monitoring, dispensing and administration Public Health, University of Health Service Surrey, Guildford, Surrey, UK; errors in UK care homes, and to identify their causes. 4 Care homes may provide 24-hour nursing care London School of Hygiene and Methods: A prospective study of a random sample of (nursing homes), personal care only (residential Tropical Medicine, London, UK; residents within a purposive sample of homes in three Imperial College Healthcare homes) or a combination. They may be owned by NHS Trust, London, UK areas. Errors were identified by patient interview, note individuals or companies of various sizes, including review, observation of practice and examination of large private health providers, charities or by the Correspondence to: dispensed items. Causes were understood by observation local authorities. Care homes are reviewed against Professor N D Barber, and from theoretically framed interviews with home staff, standards by the Care Quality Commission, which is Department of Practice and Policy, School of Pharmacy, doctors and pharmacists. Potential harm from errors was the independent regulator of health and social care Tavistock House, Tavistock assessed by expert judgement. in England. Each resident is registered with a Square, London WC1H 9JP, UK; Results: The 256 residents recruited in 55 homes were general practitioner who provides their medical care [email protected] taking a mean of 8.0 medicines. One hundred and and keeps their medical record. When a patient seventy-eight (69.5%) of residents had one or more transfers from their own home to a care home, they Accepted 27 July 2009 errors. The mean number per resident was 1.9 errors. The can elect to keep their general practitioner if he or mean potential harm from prescribing, monitoring, she is local. administration and dispensing errors was 2.6, 3.7, 2.1 and In England, the National Health Service is delivered 2.0 (0 = no harm, 10 = death), respectively. Contributing through 152 primary care trusts, which are respon- factors from the 89 interviews included doctors who were sible for a geographical area of the country, for which not accessible, did not know the residents and lacked they commission healthcare from general practi- information in homes when prescribing; home staff’s high tioners (primary care physicians) who usually work workload, lack of medicines training and drug round as part of a group practice, pharmacies and others. interruptions; lack of team work among home, practice Pharmacies may be owned and run by a single and pharmacy; inefficient ordering systems; inaccurate pharmacist or may be part of a chain; the largest medicine records and prevalence of verbal communica- chains are run by international companies. Homes tion; and difficult to fill (and check) medication adminis- usually obtain their regular supply of medicines for all tration systems. their residents from one pharmacy. Conclusions: That two thirds of residents were exposed Repeat medicines are ordered from the GP practice to one or more medication errors is of concern. The will to (usually monthly) by the care home staff using the improve exists, but there is a lack of overall responsibility. previous 28-day medicine administration record or Action is required from all concerned. the repeat medicines slip provided by the GP practice. Generally the GP practice authorises and prints the repeat prescriptions and sends them to the care home for checking. They are then forwarded to the community pharmacy where they are dispensed and INTRODUCTION delivered to the care home with a new 28-day Older people living in care homes are potentially at medicine administration record. greater risk of medication error than most other groups. They are prescribed multiple medicines and this, coupled with age-related changes in pharma- sites; and medicines may be dispensed in, and cokinetics and pharmacodynamics, makes them administered from, one of several different types of particularly susceptible to adverse drug events. packaging systems known as ‘‘monitored dosage Many have some degree of cognitive impairment systems’’ (MDS). that prevents them from being actors in the In 2000, a report documenting the extent of detection of errors. In addition, the medicines medical error was published and the UK govern- management system in care homes (described in This paper is freely available ment committed to reducing errors ; medication box 1) is complex—for example, many residents online under the BMJ Journals receive clinical interventions from multiple errors were a particular concern. Prescribing has unlocked scheme, see http:// qshc.bmj.com/info/unlocked.dtl sources; medicines may be dispensed from multiple been found to be suboptimal in UK care homes ; Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 341 Original research however, we do not know the prevalence of medication errors in not, and if so, whether it was in a cassette MDS or blister- this setting in the UK. Such studies in the USA have been pack MDS. Two drug rounds per resident were observed to undertaken; however, they have often relied on spontaneous identify administration errors. The definitions and denomi- reporting, which significantly underestimates the prevalence nators of the four types of error are described in appendix A. 10 11 of error. A US study of medication administration errors To ensure inter- and intrarater reliability, we produced a using direct observation found an error rate of 22% in nursing handbook of definitions and procedures, and research staff 12 2 homes. Gurwitz et al found 4.1 preventable adverse drug were trained in homes together and met regularly to resolve events per 100 resident-months in two US long-term care difficult cases, which were then added to the handbook. If facilities. any errors were thought to be likely to cause patient harm, Although previous studies have measured prevalence of then the pharmacist intervened. specific types of medication errors, they have not been designed The potential harm of each error was assessed using a valid to look at all types of error and simultaneously understand their and reliable method. Each error was individually assessed by a 13–15 causes. The theory of causation of human error is widely GP, a consultant old age psychiatrist, a clinical pharmacologist used in healthcare and has been used to explain causes of and two clinical pharmacists, using a validated 10-point scale 16 17 medication errors previously. In this study, we wished to (0 = no harm; 10 = death); their mean score was taken as the determine the prevalence of all forms of medication errors in harm score for that error. care homes, to assess the potential of these errors for harm and to establish the underlying causes. Statistical analysis Sample size was calculated to give a precision of +/22% SD (95% confidence interval (CI)) if the prevalence of prescribing METHODS error was 10%. The estimation was derived from a combina- Ethical approval was obtained from the Central Office for tion of simulation and formulas for clustered data, and the Research Ethics Committees. output was a matrix of numbers of homes and number of patients per home (eg, 100 homes with three patients each Selection criteria/participants gave a precision of 1.9%; 50 homes with six patients each gave We chose to sample a large number of homes and small number a precision of 2.04%), allowing us to adapt the sample of residents per home. It is generally acknowledged that errors dependent on recruitment rates. Statistical analysis was are a consequence of the systems being used, so our sampling performed using the software R 2.3 (R project for statistical strategy was designed to increase the number of systems of care computing, www.r-project.org). Statistical ‘‘significance’’ was observed (homes, general practitioners (GPs) and pharmacies). predefined at the 5% level. Exact binomial CIs were calculated Care homes for older people (nursing, residential and mixed) for proportions. x tests were used to assess differences in error were sampled in three geographically spread and demographi- rates between areas. Generalised estimating equations (library cally diverse areas in the UK (West Yorkshire, Cambridgeshire geepack, V.1.0-10) were used to model patient level odds of and central London) to obtain a varied sample with respect to errors, allowing for clustering in homes and using an ownership, size and type of care provided. For each home we independence correlation structure. Multilevel models were approached the manager/head office to obtain cooperation and also used to model patient level odds of errors, using the obtained written consent from care home staff who were MLwiN 2.03 software (multilevel models project, University observed. Care home residents on one or more medicines were of Bristol, http://www.cmm.bristol.ac.uk), fitting variance randomly selected and included in the study if they provided components at the various levels. written, informed consent. For those lacking capacity, assent was obtained from the next of kin. Written, informed consent RESULTS was obtained from GPs providing a service to the homes, which Of those approached, 72% (79/108) of homes, 67% (269/399) of included access to their care home records and GP records. residents and 61% (54/89) of general practices agreed to take part. Two hundred and fifty-six residents were recruited from 55 care homes. The majority (38, 69%) of the homes provided Data collection both residential and nursing care (corresponding figures for We used a mixture of methods. The qualitative work was residential care only and nursing care only: 12 (22%) and 5 ethnographically informed and involved field notes, observation (9%)). Table 1 shows demographic data. The majority of the and semistructured interviews based on Reason and Vincent’s 14 15 residents were women (69%, 177/256) and very old (mean age frameworks. To help understand the causes of specific errors, 85 years). There were slightly more residential care residents we conducted interviews with the home staff, GP or pharma- (54%, 139/256) than nursing care residents. Two hundred and cist. Interviews were taped where possible and transcribed; twenty (86%) of the residents were dispensed some of their otherwise, notes were made that were expanded on after the medicines in a monitored dosage system. There was a mean of interview. Analysis of the causes of errors used Reason’s 14 18 framework. 3.8 (range 1–14) GP practices per care home. There was The quantitative work to identify prescribing and monitor- considerable variation between areas: in London, the median ing errors was undertaken by clinical pharmacists (one in was 1 (range 1–3) practice per home; in West Yorkshire, the each of the three areas) conducting clinical medication median was 5 (range 1–14). reviews for each randomly selected resident. This process We intended to interview those involved in specific errors, included review of GP and care home notes and consultation soon after they had happened. However, many errors (such as with the resident and/or staff. Dispensing errors were some prescribing errors) had been made some time in the past, identified by comparing the prescriptions and medication so most interviews with doctors and pharmacists focused more administration record sheets with the dispensed medicines. on problems in general in medicines prescribing, dispensing and For each drug, it was noted if it was contained in an MDS or use in care homes. We undertook 59 interviews relating to 66 342 Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 Original research Table 1 Demographic data No of residents approached 399 No of residents consented/assented (%) 269 (67.4) No of residents excluded 13 Consent given post cut-off 6 Died 4 In hospital 3 No of residents entered into study 256 Cambridgeshire 31 West Yorkshire 121 London 104 Women, no (%) 177 (69.1) Age (y), mean (range) 85.2 (60–102) Nursing residents, no (%) 117 (45.7) Residential residents, no (%) 139 (54.3) Mean no of medicines per resident (95% CI) 8.0 (7.5 to 8.5) Median no of medicines per resident (range) 7.5 (1–25) No of residents using monitored dosing systems (%) 220 (85.9) Cambridgeshire (%) 23 (74.2) West Yorkshire (%) 117 (96.7) London (%) 80 (76.9) administration errors, 34 dispensing errors, 18 prescribing and 8 monitoring errors. Further general interviews were carried out with 19 pharmacists and 11 GPs, and observations were carried out in five pharmacies. Table 2 details the prevalence and potential harm of errors. Overall, 178 of 256 residents (69.5%, 95% CI 63.5 to 75.1) had at least one medication error (prescribing, monitoring, adminis- tration or dispensing). There was a mean of 1.9 (95% CI 1.64 to 2.17) errors per resident. Appendix B gives examples of errors. Prescribing and monitoring errors One hundred residents (39.1%, 95% CI 33.0 to 45.3) had one or more prescribing errors, totalling 153 prescribing errors, and the prescribing error rate by opportunity for error was 8.3% (95% CI 7.1 to 9.7) (opportunity for error is an act that can be erroneous; in this case it would be the prescribing of one medicine or the monitoring of one). The most common types of prescribing error, accounting for 87.6% of the total, were categorised as ‘‘incomplete information’’ (37.9% meaning, eg, that no strength or route was specified when there was more than one option), ‘‘unnecessary drug’’ (23.5%), ‘‘dose/strength error’’ (14.4%) and ‘‘omission’’ (11.8%). In total, there were 147 residents who were prescribed a medicine that required monitoring, and 18.4% (27) of these had an error. There were 32 monitoring errors in the 218 prescribed items that required monitoring (14.7%). The mean number of monitoring errors per resident was 0.13 (95% CI 0.08 to 0.17). There was significant variation between areas, with 75% of monitoring errors occurring in just one geographical area (p,0.01). Nearly one third (30.8%) of medicines deemed to require monitoring in the problem area were not being monitored. The great majority of monitoring errors (90.6%) resulted from a failure to request monitoring. The drugs most commonly associated with monitoring errors were diuretics (53.1%), ACE inhibitors (15.6%), amiodarone (12.5%) and levothyroxine (9.4%). The mean harm scores for prescribing and monitoring errors were 2.6 (range 0.2 to 5.8) and 3.7 (range 2.8 to 5.2), respectively. The interviews categorised factors thought to contribute to the errors as the patient, the task, the team and the work environment. Patient factors related to prescribing errors Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 343 Table 2 Error prevalence and harm assessment Type of error Prescribing Monitoring MAE Dispensing Overall Number of residents with an error (n = 256) (%; 95% CI) 100 (39.1%; 33.0 to 45.3) 27 (10.5%; 7.1 to 15.0) 57 (22.3%; 17.3 to 27.9) 94 (36.7%; 30.8 to 42.9) 178 (69.5%; 63.5 to 75.1) Mean number of errors per resident (95% CI) 0.60 (0.48 to 0.71) 0.13 (0.08 to 0.17) 0.45 (0.32 to 0.58) 0.73 (0.56 to 0.90) 1.91 (1.64 to 2.17) Median number of errors per resident (range) 0 (0–6) 0 (0–3) 0 (0–7) 0 (0–9) 1 (0–13) Number of errors/opportunity for error (%; 95% CI) 153/1837 (8.3%; 7.1 to 9.7) 32/218 (14.7%; 10.3 to 20.1) 116/1380 (8.4%; 7.0 to 10.0) 187/1915 (9.8%; 8.5 to 11.2) 488/5350 (9.1%; 8.4 to 9.9) Mean harm score* (95% CI) 2.6 (2.4 to 2.8) 3.7 (3.4 to 4.0) 2.1 (1.9 to 2.3) 2.0 (1.8 to 2.2) 2.6 (2.5 to 2.7) Median harm score* (range) 2 (0.2 to 5.8) 3 (2.8 to 5.2) 2 (0.1 to 5.8) 1 (0.2 to 6.6) 2 (0.1 to 6.6) MAE, medication administration error. *Assessed by a 10-point scale (0 = no harm; 10 = death). Original research included that patients were generally home bound and could dedicated time to order medicines. In some homes, 12-hour not access services external to the home; some patients also shifts were usual. disliked blood tests or taking some medicines. Some patients were confused and unable to give histories. Task factors related Dispensing errors to lack of usual prescribing technical support, including Ninety-four residents (36.7%, 95% CI 30.8 to 42.9) had a total of computer-based aids and accessing the medical record. There 187 dispensing errors with a mean of 0.73 (95% CI 0.56 to 0.90) were problems with practice computers’ failure to prompt dispensing errors per resident. The dispensing error rate by monitoring tests clearly. It could be hard to get blood tests done. opportunity for error was 9.8% (95% CI 8.5 to 11.2). Labelling There were many ‘‘team factors’’ as there was little sense of errors were found in 7.3% of dispensed items, content errors in being a whole team (home, pharmacy and practice staff). The 2.3% and clinical errors in 0.21%. There was a borderline service offered by GP practices was very variable, from a statistically significant difference in the odds of a dispensing dedicated GP making weekly visits, to GPs with no knowledge error according to delivery system (p = 0.056), largely due to the of the patient making a home visit when requested. Hospital higher odds with the cassette type monitored dosage systems out-patient and discharge letters were sometimes unclear, (cassette vs blister: adjusted odds ratio, 2.88, 95% CI 1.5 to 5.55, delayed, missed or not adequately incorporated in the patient p = 0.0012), which was associated with more labelling errors. record. When changing between GP practices, patient notes The mean harm score was 2.0 (range 0.2–6.6). took up to 4 months to arrive. GPs often expressed concern Task factors included the computer systems used, with about the care home staff, including turnover and staff some identifying all interactions rather than the clinically shortages. The skills of care home staff were sometimes seen significant ones. Some monitored dosage systems did not have as low, which several linked to pay. the space to fit in all the required warning labels, and they were tedious and difficult to fill. The similar appearance of Medication administration errors many tablets when removed from their original container was Fifty-seven (22.3%, 95% CI 17.3 to 27.9) residents had a total of a potential source of error. The prescription and medication 116 administration errors. The mean number of administration administration record were often different, so it was unclear errors per resident was 0.45 (95% CI 0.32 to 0.58) and the which was correct. Individual factors included staff feeling prevalence of administration errors by opportunities for error hungry, tired or unwell while dispensing monitored dosage was 8.4% (95% CI 7.0 to 10.0). Nearly half (49.1%) of all systems. There was lack of knowledge by pharmacy staff of administration errors were categorised as ‘‘omissions’’ and just the care homes’ systems and their need for support. Team more than one fifth (21.6%) were ‘‘wrong dose’’. The odds of a factors overlapped with this—some pharmacists had no medication administration error occurring were higher in knowledge of the care home and its requirements, relation- residential care than in nursing care residents; however, this ships were sometimes tense and poor language skills of home just failed to reach statistical significance at the 5% level staff were sometimes cited. Some pharmacies had a poor (adjusted OR 1.77, 95% CI 0.96 to 3.25, p = 0.063; adjusted for checking process, and use of locums could be a problem. Work age, sex and medication delivery system). The higher apparent factors included the pharmacies being seen as being busy and risk of administration errors in residential compared with pressured, with interruptions and distractions (including nursing residents was largely attributable to more ‘‘omissions’’ noise) and some staff shortages. (38 vs 19) and ‘‘wrong doses’’ (18 vs 7). There was no statistically significant difference in the administration errors by the medicine delivery system. The mean harm score was 2.1 Organisational culture (range 0.1–5.8). When considering all forms of medication errors, factors relating Patient factors included many patients’ lack of awareness of to organisational culture (called ‘‘latent failures’’ by Reason ) their medicines. In addition, their physical condition could make were significant. It was clear from the interviews that no one it hard to administer medicines properly. Some patients had took responsibility for the whole system. We often saw well- fears about medicines, such as feeling they were being poisoned, intentioned people doing their best but in an uncoordinated and some were consequently aggressive. Finding mobile patients way. Communication, written and verbal, was another proble- during the drug round could be a problem. Changing or adding matic factor, within and among the home, GP practice and medicines, such as acute treatment, in the middle of the 4-week pharmacy. Consequently, it is difficult to know which supply cycle could be a problem. Task factors included inability medicines any patient should be having. Management within to find the medicine, failure to order the right quantity of ‘‘as each organisation was a factor, particularly when challenged to required medicines’’, the special requirements that some deliver a safe service within a tight budget. medicines had (eg, ‘‘take on an empty stomach’’), the difficulty many staff had in correctly administering inhalers and a lack of DISCUSSION adequate protocols. Individual factors related to the staff People in care homes are a frail and vulnerable population at included lack of knowledge about inhalers and the timing of particular risk from medication errors, and it is a cause for medicines with respect to food. Team factors included the concern that two thirds of care home residents in this study medication administration record chart, which should be the were exposed to one or more errors. For each event involving documentary line of communication among GP, home and prescribing, dispensing or administration of a medicine, there pharmacist; these records were often inaccurate. Communication within the home tended to be verbal. Work was an 8%–10% chance of an error happening and a 14% chance environment factors included homes being hot, airless, having of a monitoring error. Safety is a systems issue, and we believe unpleasant smells, being poorly lit, noisy and short of space. this is the first study to consider the whole system of There were often staffing problems in the morning round medication use in care homes; our simultaneous collection of (when most medicines were given and when staff also had most qualitative data has allowed us to understand the causes of error other tasks). Staff were frequently interrupted and did not have and suggest solutions. 344 Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 Original research The prevalence of prescribing error is similar to that found in unclear and some commissioners discourage its use. The use of primary care ; administration error prevalence was a little MDS drives efficiencies of scale, such as large centralised higher than that in hospital (and likely to be better than the repackaging units, which in turn leads to the dispensary patients’ adherence if in their own home). The prevalence of becoming remote from the customers (home and patient). dispensing errors was three times higher than the rate found in Research into the effectiveness of MDS is urgently required. primary care in the UK, although that study excluded MDS. Within homes the use and accuracy of the medication Our higher rate predominantly reflected one type of MDS that administration record requires constant review. The lack of was difficult to label fully. protocols and adequate staff training remains an issue. Drug rounds are very busy, and often interrupted in the morning, and Although our study was not primarily designed to identify some medicines should be prescribed for different times to ease the prevalence of harm, we saw several errors, particularly this. The commonest administration errors were omissions monitoring errors, which had caused harm or were likely to. In because the drug was not available, so omissions need to be addition, many errors would reduce the quality of life and ability to function of residents, such as inadequate treatment of monitored and ordering, particularly of ‘‘as required’’ medicines, pain, of bowels and of breathing. needs to be improved. We were very impressed by the proportion of homes Limitations to our study include that our sample only participating in a study, which was potentially very threatening contained those willing to be studied (although the acceptance to them. Several care home managers have told us that patient rate of homes was high) and that our home sampling was not harm from medication error is their greatest fear and that up to random. Judgement of the cause of error was sometimes half of staff time can be spent on medication related activities. difficult as there could be conflicting sources of evidence or a Given this motivation and resource, we are hopeful of change. lack of evidence; hence, judgements sometimes retained an element of subjectivity. Observation may theoretically have Acknowledgements: We thank the care home staff, residents, relatives, community affected the prevalence of administration error, although pharmacists and general practitioners. routine observation has been found to have no effect. Staff Funding: The study was funded by the Patient Safety Research Programme of the interviewed will have given accounts affected by hindsight bias; Department of Health. The authors are independent of the funders. The sponsor hence, imputations of causality are speculative. approved the study design. All authors had full access to all the data and can take What can be done, and who should do it? As our study responsibility for the integrity of the data and the accuracy of the data analysis. shows, there are currently many and varied subsystems that are Competing interests: None. not being seen in an integrated way. There is now the opportunity for a systems approach to the whole. Since 2008, REFERENCES chief pharmacists of provider organisations and commissioners 1. Zermansky AG, Alldred DP, Petty DR, et al. Clinical medication review by a in England should have the lead role in ensuring safe medication pharmacist of elderly people living in care homes—randomised controlled trial. Age practices are embedded in patient care ; a significant and Ageing 2006;35:586–91. 2. Gurwitz JH, Field TS, Judge J, et al. The incidence of adverse drug events in two pressing agenda for them. An additional system-based solution large academic long-term care facilities. Am J Med 2005;118:251–8. relates to several of the communication and records problems 3. Alldred DP, Petty DR, Bowie P, et al. Antipsychotic prescribing patterns in care observed—we would hope these would be ameliorated by homes and relationship with dementia. Psychiatr Bull 2007;31:329–32. programmes in the National Health Service’s information 4. Expert Group on Learning from Adverse Events in the NHS. An organisation with a memory. London: The Stationary Office, 2000. technology programme (NPfIT), such as the Summary Care 5. Department of Health. Building a safer NHS for patients. London: Department of Record (a brief GP record which can be accessed by others), Health, 2001. GP2GP (electronic transfer of patients’ notes between GPs) and 6. Fahey T, Montgomery AA, Barnes J, et al. Quality of care for elderly residents in nursing homes and elderly people living at home: controlled observational study. BMJ the Electronic Prescription Service (electronic transfer of 2003;326:580–3. primary care prescriptions). The final system issue is that most 7. Alldred DP, Zermansky AG, Petty DP, et al. Clinical medication review by a primary care is based on patients going to centres of care rather pharmacist of elderly people living in care homes: pharmacist interventions. than the other way around. Primary care services that are based Int J Pharm Pract 2007;15:93–9. 8. Pierson S, Hansen R, Greene S, et al. Preventing medication errors in long-term on care going to patients need to be commissioned, in order not care: results and evaluation of a large scale web-based error reporting system. Qual to disadvantage the home bound. Saf Health Care 2007;16:297–302. We suggest the idea of a lead (not sole) GP for each home 9. Greene S, Williams C, Hansen R, et al. Medication errors in nursing homes. J Pat Saf 2005;1:181–9. should be explored. This role would need protected time and 10. Barker KN, McConnell WE. The problems of detecting medication errors in hospitals. associated funding. In addition to caring for patients, they Am J Hosp Pharm 1962;19:360–9. should liaise with other GPs and have responsibility to ensure, 11. Dean B, Barber N. Validity and reliability of observational methods for studying possibly by commissioning services, that patients on riskier medication administration errors. Am J Health System Pharm 2001;58:54–9. 12. Barker KN, Flynn EA, Pepper GA, et al. Medication errors observed in 36 healthcare medicines are appropriately monitored and that all patients’ facilities. Arch Int Med 2002;162:1897–903. medication is regularly reviewed by a pharmacist. 13. Rasmussen J, Jensen A. Mental procedures in real life tasks: a case study of Consideration should be given to having one person with electronic trouble-shooting. Ergonomics 1974;17:293–307. 14. Reason J. Human error. Cambridge: University of Cambridge, 1990. overall responsibility for medicines use in one or more care 15. Vincent C, Taylor-Adams SE, Stanhope N. Framework for analysing risk and safety in homes. Many pharmacists have the skills and knowledge to clinical practice. BMJ 1998;316:1154–7. undertake this role, and such developments are described in the 16. Dean BS, Schachter M, Vincent C, et al. Prescribing errors in hospital inpatients— UK government’s recent proposals for making best use of incidence and clinical significance. Qual Saf Health Care 2002;11:340–4. 17. Taxis K, Barber N. Causes of intravenous medication errors: an ethnographic study. pharmacists’ expertise. Qual Saf Health Care 2003;12:343–7. Pharmacists supplying homes should ideally know the home, 18. Dean B, Schachter M, Vincent C, et al. Causes of prescribing errors in hospital its ways and needs, so that ordering and supply match the inpatients: a prospective study. Lancet 2002;16:271–8. 19. Dean B, Barber N. A validated, reliable method of scoring the severity of medication home’s (and patients’) needs. The widespread use of MDS unit errors. Am J Health System Pharm 1999;56:57–62. dose systems, requiring millions of tablets to be repackaged each 20. Reason J. Managing the risks of organisational accidents. Aldershot: Ashgate, 1997. week, is a vast, unfunded undertaking. It imposes demands on 21. Shah SDH, Aslam M, Avery AJ. A survey of prescription errors in general practice. home and pharmacy alike, yet its contribution to safety is Pharm J 2001;267:860–2. Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231 345 Original research 22. Dean BS, Schachter M, Vincent C, et al. Prescribing errors in hospital inpatients— and Dean and Barber as ‘‘any deviation between the medication prescribed and that incidence and clinical significance. Qual Saf Health Care 2002;11:340–4. administered’’. The number of opportunities for error (denominator) was the number of 23. Horne R, Weinman J, Barber N, et al. Concordance, adherence and compliance in doses given, plus any doses that should have been given but were omitted. medicine taking. London: NCCSDO, 2005. 24. Franklin BD, O’Grady K. Dispensing errors in community pharmacy: frequency, clinical significance and potential impact of authentication at the point of dispensing. APPENDIX B: EXAMPLES OF ERRORS Int J Pharm Pract 2007;15:273–81. 25. Department of Health. Pharmacy in England. Building on strengths—delivering the future. London: DoH, 2008. Case 1: Prescribing error 26. Dean B, Barber N, Schachter M. What is a prescribing error? Qual Saf Health Care Tramadol capsules 50 mg prescribed ‘‘one to be taken up to four times a day’’ for 2000;9:232–7. chronic foot pain, resident also taking warfarin for long-term DVT prophylaxis 27. Alldred DP, Standage C, Zermansky AG, et al. Development and validation of criteria (Tramadol can enhance the effect of warfarin). International normalised ratio checked to identify medication-monitoring errors in care home residents. Int J Pharm Pract regularly with erratic results ranging from 0.9 to 4.5 (mean harm score 5.8). 2008;16:317–23. 28. Beso A, Franklin BD, Barber N. The frequency and potential causes of dispensing errors in a hospital pharmacy. Pharm World Sci 2005;27:182–90. Case 2: Prescribing error 29. Allan EL, Barker KN. Fundamentals of medication error research. Am J Hosp Pharm Donepezil 10-mg tablets prescribed ‘‘one daily’’ for dementia. Following a review from 1990;47:555–71. hospital specialist, a letter to the GP indicated Aricept (brand name for donepezil) should be stopped. The donepezil had not been discontinued by the GP and there was APPENDIX A: ERROR DEFINITIONS no indication in the medical notes that the continuation was deliberate (mean harm Prescribing errors were identified and classified according to the definition developed score 1.8). by Dean et al as ‘‘A prescribing decision or prescription-writing process that resultsin 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 Case 3: Monitoring error practice’’. The number of opportunities for error (denominator) was the number of Lisinopril 5-mg tablets prescribed ‘‘one daily’’ for hypertension for a resident with an prescription items written, plus any omissions. The three pharmacists worked to a estimated creatinine clearance of 19 ml/min. A potassium level had been checked common detailed protocol when reviewing the residents and their medicines, were 1 year ago and had revealed a high potassium level 5.8 mmol/l (range 3.5–5 mmol/l): trained together at the start of the study and had regular review meetings to ensure no action had been taken (mean harm score 5.8). consistency. Monitoring errors were identified according to the definition developed and validated by Alldred et al as ‘‘A monitoring error occurs when a prescribed medicine is not Case 4: Monitoring error monitored in the way which would be considered acceptable in routine general Amiodarone 200-mg tablets prescribed ‘‘one daily’’ for atrial fibrillation. Thyroid practice. It includes the absence of tests being carried out at the frequency listed, with function tests were last checked 9 months ago, when thyroid stimulating hormone a tolerance of +50%. This means—for example, that if a medicine requires liver was 12.9 mlU/l (range 0.3–5.5 mlU/l) and thyroxine was 18 nmol/l (range 50– function tests at three monthly intervals, an error would occur if a test was not 151 nmol/l), and no action had been taken (mean harm score 5.8). conducted within 18 weeks’’. The number of opportunities for error (denominator) was the number of prescribed items that required monitoring, according to the validated criteria. Case 5: Medication administration error The definition of a dispensing error developed by Beso et al was adopted. A Bendroflumethiazide 2.5-mg tablets prescribed ‘‘one each morning’’ for hypertension. dispensing error was defined as ‘‘One or more deviations from an interpretable written This was discontinued due to low serum sodium of 127 mmol/l (range 135– prescription or medication order, including written modifications to the prescription 145 mmol/l); however, it remained on the current medication administration record made by a pharmacist following contact with the prescriber’’. Dispensing errors were when the next monthly drug supply was made and hence continued to be identified and classified by the clinical pharmacists by comparing the prescriptions and administered (mean harm score 4.6). medicine administration record sheets with the dispensed medicines. The number of opportunities for error (denominator) was the number of prescription items dispensed or omitted. Case 6: Dispensing error The clinical pharmacists observed two drug rounds per resident to identify and classify Aspirin enteric-coated 75 mg tablets dispensed instead of zopiclone 7.5 mg tablets as medication administration errors as defined using previous work by Allan and Barker a 7-day supply in a cassette dispensing system (mean harm score 5.0). 346 Qual Saf Health Care 2009;18:341–346. doi:10.1136/qshc.2009.034231

Journal

Quality and Safety in Health Care – Pubmed Central

Published: Sep 25, 2009

There are no references for this article.