2009 The Author Journal compilation  2009 Nordic Pharmacological Society. Basic & Clinical Pharmacology & Toxicology, 106, 100–105

Doi: 10.1111/j.1742-7843.2009.00481.x

Drug Dispensing Errors in a Ward Stock System Stig Ejdrup Andersen Department of Clinical Pharmacology, Bispebjerg Hospital, Copenhagen, Denmark (Received 17 April 2009; Accepted 4 August 2009)

Abstract: The aim of this study was to determine the frequency of drug dispensing errors in a traditional ward stock system operated by nurses and to investigate the effect of potential contributing factors. This was a descriptive study conducted in a teaching hospital from January 2005 to June 2007. In five wards, samples of dispensed solid drugs were collected prospectively and compared with the prescriptions. Data were evaluated using multivariable logistic regression. Overall, 2173 samples were collected, 95.5% of which were correctly dispensed (95% CI 94.5–96.2). In total, 124 errors in 6715 opportunities for error were identified; error rate of 1.85 errors per 100 opportunities for error (95% CI 1.54–2.20). Omission of a dose was the predominant type of error while vitamins and minerals, drugs for acid-related diseases and antipsychotic drugs were the drugs most frequently affected by errors. Multivariable analysis showed that surgical and psychiatric settings were more susceptible to involvement in dispensing errors and that polypharmacy was a risk factor. In this ward stock system, dispensing errors are relatively common, they depend on speciality and are associated with polypharmacy. These results indicate that strategies to reduce dispensing errors should address polypharmacy and focus on high-risk units. This should, however, be substantiated by a future trial.

Drug distribution systems vary, yet all are designed to ensure that each patient receives medication exactly as prescribed. One of the key processes in drug distribution is dispensing, which denotes apportioning and preparation of prescribed drugs. Because the output of drug dispensing is the input into the administration process, the frequency of dispensing errors is an important measure of the quality of the drug distribution system. Hitherto, research into dispensing errors has focused more on pharmacy-based systems, such as ward pharmacy systems or unit dose systems and less on the traditional ward stock system operated by nurses. Moreover, dispensing has been studied as part of studies on administration error rates. Thus, Taxis et al. reported a 5.1% error rate at a site using the traditional system [1]. At two other sites using a ward pharmacy system and a unit dose system, the error rates were 8.0% and 2.4% respectively. Combining both direct observation and unannounced control visits, Lisby et al. detected an equivalent dispensing error rate of 4% at a site using the traditional system [2]. Approximately one in four errors was potentially serious. None of these studies addressed the underlying causes of dispensing errors. Each task in drug distribution is associated with some intrinsic risk that can be modified by contributing factors [3]. Consequently, errors rarely have single causes but originate from a complex interaction between several factors [4]. Thus, an analysis of medication errors unveiled an adverse effect of temporary staff, patient load, shift work and sea-

Author for correspondence: Stig Ejdrup Andersen, Department of Clinical Pharmacology, Bispebjerg Hospital, Bispebjerg Bakke 23, DK-2400 Copenhagen NV, Denmark (fax +45 35313711, e-mail [email protected]).

sonal variation in daylight [5] while an ethnographic study showed that lack of training, technology design, poor communication and heavy workload contributed to intravenous medication errors [6]. As few studies have considered contributing factors in the context of dispensing errors, this study was set up. The aim was to determine the frequency of drug dispensing errors in a traditional ward stock system and to investigate the effect of potential contributing factors. Materials and Methods Study wards. The study was approved by the hospital managers and conducted in a 696-bed teaching hospital with approximately 30,000 discharges per year and 86% mean bed occupancy. All participating wards kept a floor stock of formulary drugs. Nursing staff or pharmacy technicians could order supplementary medication from the pharmacy on a daily basis. Day and night, the pharmacy could execute emergency orders at hours’ notice. Table 1 shows the characteristics of the wards. Three of the wards received pharmacy service with daily visits from the ward pharmacy technician on weekdays while a pharmacist controlled all medication rooms once a year. In January 2005, at the outset of the study, doctors wrote prescriptions on a drug-prescribing sheet that nursing staff used when dispensing and administering the medication. During the course of the study, a computerized physician order entry system was gradually implemented (table 1). In this system, the physicians enter prescriptions into the computer and verify each prescription on a daily basis. The sparse decision support linked to the system includes pointing out drugs from the wards’ standard drug assortment and producing allergy warnings. Besides computerized physician order entry, the system comprises digitalized medication administration records, bar code scanners and personal digital assistants, which allows verification of barcode-labelled medication containers, patient bracelets and caregiver badges. Fig. 1 illustrates the drug distribution system. According to the hospital policy, drug administration in acute ward sections should follow immediately after dispensing and needs no re-verification of drug or dose (uninterrupted procedure) (fig. 1).

WARD STOCK DISPENSING ERRORS

101

Table 1. Characteristics of the participating wards.

Ward

Speciality

Beds

Mean length of stay (days)

Geriatrics Neurology Psychiatry Abdominal surgery Orthopaedic surgery

Other Other Psychiatry Surgery Surgery

51 63 151 70 98

24 8 26 3 6

Mean bed occupancy (%)

No. of stationary ward sections

Yearly discharge rate

Start date of CPOE

Service by pharmacy technician

93 82 86 76 81

3 4 12 5 7

740 2442 1126 6306 4933

May 2006 March 2006 November 2005 November 2006 November 2006

Yes No No Yes Yes

CPOE, computerized physician order entry system.

Medication flow Information flow Measure of error rates

Interrupted process MD orders (CPOE or drug prescribing sheet)

Patient care units Dispensed into a multicompartment medication box

Delivered from the pharmacy

Stored in the medication room

Medication retrieved for dispensing

Dose and medication re-verified

Transferred to a drug cup

Administration

Dose retrieved and verified

CPOE = Computerised physician order entry

Dispensed into a drug cup

MD = Physician

Administration Uninterrupted process

Fig. 1. Overview of the decentralized ward stock dispensing process.

In non-acute ward sections, nursing staff were allowed to dispense medication for up to 3 days into a multi-compartment pill box (interrupted process). During the following drug rounds, nurses were to re-verify drugs and doses before pouring the medication from the appropriate compartment into a drug cup and administering the medication. Data collection. From January 2005 to June 2007, five trained nurses (sample collectors) visited the ward sections at the time of a drug round and collected samples of dispensed medication. Up to 10 samples per month from all shifts and all days of the week were to be collected. Each sample consisted of all solid doses dispensed for a patient at the time of the visit. The samples were convenience samples (fig. 1). Having had the opportunity to correct errors and leaving the medication room to administer the drugs, the nurse was stopped by a sample collector and requested to hand over the drug cup. With a drug formulary and a digital tablet atlas at hand, the sample collector subsequently recorded details of each dispensed dose and compared with the relevant prescriptions. After correction of errors, the drug sample was returned and administered as quickly as possible. All visits were unannounced. Blinded to the specific purpose of the study, staff were explained that samples would be collected to assess the safety of multi-compartment pill boxes. As the samples were collected at the time of dispensing, the nurses could not be blinded to the visits. No samples were collected during the periods when the computerized physician order entry system was being implemented. No patient data were recorded. Supplementary administrative data on the discharge rates were collected from the hospital management system.

Definitions. An error was defined as a dose of medication dispensed that deviated from the doctor’s prescription. Five types of error were included: unordered drugs, omission of a dose, wrong dose, wrong form and wrong time. Time deviations of less than 2 hr were accepted. Prescribing and administration errors were not recorded. The total number of opportunities for error was the denominator used to calculate the error rate. An opportunity for error was defined as any dose dispensed plus any dose prescribed but not dispensed. Data analysis. The distribution of dispensing errors was calculated with 95% confidence intervals (CI). Nominal data were compared using the chi-squared test and continuous data were compared using one-way analysis of variance. The effect of potentially contributing factors on dispensing errors was evaluated using multiple logistic regression. Ordinal and nominal independent variables were incorporated into the analysis by transforming them into multiple categorical variables. The dependent variable was samples of dispensed medication (correct ⁄ incorrect). Six independent variables were incorporated into the model at the same time: computerized physician order entry and barcode verification in the ward section (yes ⁄ no); uninterrupted process (yes ⁄ no); clinical speciality (surgical ⁄ psychiatric ⁄ other specialities); shift (day ⁄ afternoon ⁄ night); number of dispensed doses per sample (1, 2, 3, 4, 5, >5); higher than median discharge rate (yes ⁄ no). Malfunction of the computerized physician order entry system was not recorded and neither were incidents when electronic prescribing was used, but the bar code verification could not be completed. To take into account potential clustering, ward was nested within speciality. Because the length of the day in Copenhagen is approximately 10 hr and 30 min. shorter at winter than at summer solstice and variation in daylight has previously been

 2009 The Author Journal compilation  2009 Nordic Pharmacological Society. Basic & Clinical Pharmacology & Toxicology, 106, 100–105

STIG EJDRUP ANDERSEN

102

associated with medication errors [5], the analysis was adjusted for data collection period (in 1 year categories) and season (in four categories). Samples with missing data were excluded from the analysis. A two-tailed p-value of 5 Discharge rate Low High

2075 (95.5)

98 (4.5)

1407 (95.1) 668 (96.4)

Adjusted

OR

CI

OR

CI

73 (4.9) 25 (3.6)

0.72

0.45–1.15

0.69

0.44–1.09

562 (94.6) 1513 (95.8)

32 (5.4) 66 (4.2)

0.77

0.50–1.18

0.85

0.66–1.10

756 (99.0) 930 (96.9) 389 (86.6)

8 (1.0) 30 (3.1) 60 (13.4)

3.05 14.6

1.39–6.69 6.90–30.8

2.90 3.08

1.16–5.06 1.73–5.48

1506 (95.4) 328 (94.5) 241 (97.2)

72 (4.6) 19 (5.5) 7 (2.8)

1.21 0.61

0.72–2.04 0.28–1.34

1.30 0.96

0.94–1.79 0.62–1.47

604 (99.3) 487 (97.2) 363 (96.5) 227 (91.2) 141 (92.8) 253 (88.2)

4 (0.7) 14 (2.8) 13 (3.5) 22 (8.8) 11 (7.2) 34 (11.8)

4.34 5.41 14.6 11.8 20.3

1.42–13.3 1.75–16.7 4.99–42.9 3.70–37.5 7.13–57.8

1.95 2.09 3.38 3.36 4.37

1.11–3.45 1.18–3.72 1.94–5.89 1.84–6.14 2.52–7.57

1188 (95.0) 887 (96.2)

63 (5.0) 35 (3.8)

0.74

0.49–1.14

0.93

0.67–1.28

Each odds ratio (OR) was adjusted for all variables in the table and for time of data collection (in 1 year categories) and season (four categories). CPOE, computerized physician order entry system; CI, 95% confidence interval. 1 Neurology and geriatrics.

Rate of dispensing errors (%)

40 Other specialities1 Psychiatry Surgery

30

20

10

0

1

2

3 4 Number of drugs

5

>5

Fig. 2. Rate of dispensing errors by number of drugs in each sample. 1 Neurology and geriatrics.

association. Still, the positive impact of the technology is probably underestimated. First, the computerized systems were studied at a certain stage of development, yet software and user interface are continuously being updated in response to problems. Secondly, we did not record when electronic prescribing was used but the bar code verification

could not be completed. This happens regularly, as about 10–20% of the drug containers are still not barcode-labelled. Thirdly, the analysis did not take into account periods with malfunctioning computer systems. And finally, though no data were collected intentionally during the implementation periods, data might still have been collected from a changing system. Polypharmacy use is a known risk factor of noncompliance, adverse events, drug–drug interactions and inappropriate medication use [18,19]. The complexity of the problem is underlined by the fact that the probability of underprescribing also increases with the number of drugs used [20]. Requiring more operations, polypharmacy makes the task of drug dispensing more complex. As this study demonstrates a positive association between the numbers of dispensed drugs and dispensing errors, it also reveals a hitherto unrecognized hazard of polypharmacy that tends to be of particular importance in surgical units (fig. 2). It follows that measures against polypharmacy such as considered prescribing and review of medication lists might benefit dispensing as well. This should, however, be substantiated by a future trial. Compared with previous studies, the error rate in the present study seems rather low [1,2]. And though the severity and

 2009 The Author Journal compilation  2009 Nordic Pharmacological Society. Basic & Clinical Pharmacology & Toxicology, 106, 100–105

STIG EJDRUP ANDERSEN

104

potential clinical impact of the observed errors was not assessed, most of the observed dispensing errors appear minor, and the majority would have no serious adverse consequences should they reach the patients (table 2). Still, medication errors are closely linked to adverse drug events [21], and given the high volume of drug dispensed, even a low error rate may translate into a large number of adverse events. As only samples of solid doses were collected, this study represents a best-case scenario as regards the potential to harm. Although doses of anticoagulants, cardiovascular drugs and some antibiotics were sampled, other frequently used high alert drugs like concentrated solutions of electrolytes, insulin and antineoplastic drugs were not included [21–24]. Moreover, the study design did not allow observation of the preparation of intravenous drugs which is a known high-risk activity [6,25]. This study has additional methodological shortcomings that deserve attention. The most important is that drug administration was not observed. The most common type of error was omissions and though these errors appear to be a built-in problem in many drug distribution systems, our finding is in keeping with previous reports [26,27]. These errors might have been detected and corrected during the administration process. Moreover, pharmaceutical or administration errors such as wrong storage, use of expired drugs, lack of identity control or wrong route were not included. The method relied on nurses identifying dispensed drugs in a tablet atlas. Although this method leads to a high degree of recognition, as much as 8% of the drugs are misrecognized [2]. Direct observation of the drug dispensing process, which is considered the most reliable method of identifying medication errors, might have led to a higher yield [28]. Moreover, the data collectors were not masked to the study purpose, and blinding of nursing staff could not be ensured. Finally, as the study was carried out in a single hospital, the applicability to other systems is unknown. The strength of this prospective study is that it was performed under real practice settings. And despite limitations, this study shows that wrong dispensing is relatively common in this ward stock system operated by nurses. Surgical and psychiatric settings are more susceptible to involvement in errors, and polypharmacy seems an important risk factor. These results indicate that it might be appropriate to address polypharmacy and focus on high-risk clinical units when redesigning for safer dispensing in ward stock systems. This strategy should, however, be substantiated by a future trial. Acknowledgement The author is indebted to the five dedicated nurses who collected the drug samples.

References 1 Taxis K, Dean B, Barber N. Hospital drug distribution systems in the UK and Germany – a study of medication errors. Pharm World Sci 1999;21:25–31.

2 Lisby M, Nielsen LP, Mainz J. Errors in the medication process: frequency, type, and potential clinical consequences. Int J Qual Health Care 2005;17:15–22. 3 McDowell SE, Ferner HS, Ferner RE. The pathophysiology of medication errors: how and where they arise. Br J Clin Pharmacol 2009;67:605–13. 4 Karsh BT, Holden RJ, Alper SJ, Or CK. A human factors engineering paradigm for patient safety: designing to support the performance of the healthcare professional. Qual Saf Health Care 2006;15(Suppl 1):i59–65. 5 Roseman C, Booker JM. Workload and environmental factors in hospital medication errors. Nurs Res 1995;44:226–30. 6 Taxis K, Barber N. Causes of intravenous medication errors: an ethnographic study. Qual Saf Health Care 2003;12:343–7. 7 Dean B, Schachter M, Vincent C, Barber N. Causes of prescribing errors in hospital inpatients: a prospective study. Lancet 2002;359:1373–8. 8 Howard R, Avery A, Bissell P. Causes of preventable drug-related hospital admissions: a qualitative study. Qual Saf Health Care 2008;17:109–16. 9 Nichols P, Copeland TS, Craib IA, Hopkins P, Bruce DG. Learning from error: identifying contributory causes of medication errors in an Australian hospital. Med J Aust 2008;188:276–9. 10 Franklin BD, O’grady K, Donyai P, Jacklin A, Barber N. The impact of a closed-loop electronic prescribing and administration system on prescribing errors, administration errors and staff time: a before-and-after study. Qual Saf Health Care 2007;16:279–84. 11 Bates DW, Leape LL, Cullen DJ, Laird N, Petersen LA, Teich JM et al. Effect of computerized physician order entry and a team intervention on prevention of serious medication errors. JAMA 1998;280:1311–6. 12 Donyai P, O’grady K, Jacklin A, Barber N, Franklin BD. The effects of electronic prescribing on the quality of prescribing. Br J Clin Pharmacol 2008;65:230–7. 13 Ioannidis JP, Lau J. Evidence on interventions to reduce medical errors: an overview and recommendations for future research. J Gen Intern Med 2001;16:325–34. 14 Oren E, Shaffer ER, Guglielmo BJ. Impact of emerging technologies on medication errors and adverse drug events. Am J Health Syst Pharm 2003;60:1447–58. 15 Poon EG, Cina JL, Churchill WW, Mitton P, McCrea ML, Featherstone E et al. Effect of bar-code technology on the incidence of medication dispensing errors and potential adverse drug events in a hospital pharmacy. AMIA Annu Symp Proc 2005;1085. 16 van Gijssel-Wiersma DG, van den Bemt PM, Walenbergh-van Veen MC. Influence of computerised medication charts on medication errors in a hospital. Drug Saf 2005;28:1119–29. 17 Koppel R, Metlay JP, Cohen A, Abaluck B, Localio AR, Kimmel SE et al. Role of computerized physician order entry systems in facilitating medication errors. JAMA 2005;293:1197–203. 18 Sorensen L, Stokes JA, Purdie DM, Woodward M, Roberts MS. Medication management at home: medication-related risk factors associated with poor health outcomes. Age Ageing 2005;34:626– 32. 19 Steinman MA, Landefeld CS, Rosenthal GE, Berthenthal D, Sen S, Kaboli PJ. Polypharmacy and prescribing quality in older people. J Am Geriatr Soc 2006;54:1516–23. 20 Kuijpers MA, van Marum RJ, Egberts AC, Jansen PA. Relationship between polypharmacy and underprescribing. Br J Clin Pharmacol 2008;65:130–3. 21 Bates DW, Boyle DL, Vander Vliet MB, Schneider J, Leape L. Relationship between medication errors and adverse drug events. J Gen Intern Med 1995;10:199–205. 22 Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D et al. Incidence of adverse drug events and potential adverse drug

 2009 The Author Journal compilation  2009 Nordic Pharmacological Society. Basic & Clinical Pharmacology & Toxicology, 106, 100–105

WARD STOCK DISPENSING ERRORS events. Implications for prevention. ADE Prevention Study Group. JAMA 1995;274:29–34. 23 Bates DW, Spell N, Cullen DJ, Burdick E, Laird N, Petersen LA et al. The costs of adverse drug events in hospitalized patients. Adverse Drug Events Prevention Study Group. JAMA 1997;277:307–11. 24 Thomas EJ, Studdert DM, Burstin HR, Orav EJ, Zeena T, Williams EJ et al. Incidence and types of adverse events and negligent care in Utah and Colorado. Med Care 2000;38:261–71. 25 Cousins DH, Sabatier B, Begue D, Schmitt C, Hoppe-Tichy T. Medication errors in intravenous drug preparation and adminis-

105

tration: a multicentre audit in the UK, Germany and France. Qual Saf Health Care 2005;14:190–5. 26 Barker KN, Flynn EA, Pepper GA, Bates DW, Mikeal RL. Medication errors observed in 36 health care facilities. Arch Intern Med 2002;162:1897–903. 27 Ridge KW, Jenkins DB, Noyce PR, Barber ND. Medication errors during hospital drug rounds. Qual Health Care 1995;4:240–3. 28 Flynn EA, Barker KN, Pepper GA, Bates DW, Mikeal RL. Comparison of methods for detecting medication errors in 36 hospitals and skilled-nursing facilities. Am J Health Syst Pharm 2002;59:436–46.

 2009 The Author Journal compilation  2009 Nordic Pharmacological Society. Basic & Clinical Pharmacology & Toxicology, 106, 100–105