International Journal for Quality in Health Care 15:207-212 (2003)
© 2003 International Society for Quality in Health Care
Paper |
Common types of medication errors on long-term psychiatric care units
1Healthcare Evaluation Section, Department of Management Science, National Institute of Public Health, Wako, Saitama
2Hanazono Hospital, Yamanashi, Japan
Objective. This multi-center study identified the most frequent types of medication errors in long-term psychiatric care hospitals.
Setting. Japan.
Design and study participants. We asked 132 units in 44 Japanese psychiatric hospitals to introduce an in-patient incident reporting system on potential adverse drug events (PADEs) for the period 1 October to 30 November 2000. We analyzed types of PADE, outcomes, and characteristics of patients, staff, and units.
Results. We received 221 PADE incident reports from 85 units of 44 hospitals. One-quarter (24.9%) of the incidents were intercepted before reaching patients. The frequency of monitoring of the patients by clinical staff in response to medication errors increased by 8.1%. Wrong drug administration, i.e. giving a drug to a patient that was not the drug prescribed for that patient, was the most common type of incident (35.7%). Logistic regression analysis revealed that wrong drug administration occurred more frequently on units with either fewer registered nurses, or two or more patients with the same (or similar) name staying on the same unit. Incident reporters evaluated wrong drug administration as being potentially more serious than the other types of medication error. Wrong drug administration was seen more frequently in units with no patient name printed on medication drug pouches.
Conclusions. Wrong drug administration was the most common type of PADE, and may result in more serious consequences than others. Even a simple organizational quality improvement effort, in which printed patients’ names are placed on the drug pouch (not only with each prescription but with each drug administration), could reduce risk to patients from adverse outcomes due to medication errors.
Keywords: adverse drug event, drug administration, medication error, mental health, psychiatry, quality of care, safety
Medication errors, especially administration of the wrong drug, are among the most common types of error in health care services [1,2]. A system to verify that the proper drug is delivered to the right patient is both basic and essential for ensuring patient safety and improving quality of care.
Although drug identity checking errors (wrong drug error) have been classified in the same category as wrong dose error [1], this categorization may need to be reconsidered, especially when considering quality improvement actions in clinical practice. The implications for corrective action are different for wrong drug errors than they may be for wrong dose errors. Wrong drug errors, defined as pharmacists’ and nurses’ checking errors that result in patients receiving (or nearly receiving, so-called ‘near misses’) the wrong medication [1], are different from wrong dosage errors where there is a failure to ensure that a proper dose is dispensed or administered. The potential adverse effects of wrong drug errors may be very serious when patients receive an inappropriate drug for their condition, since then the disorder is not properly treated and the patient is exposed to an unnecessary medication with attendant side- and adverse effects [3].
Wrong drug errors are important in long-term care facilities such as nursing homes and long-term psychiatric institutions. A study by Kapborg and Svensson showed that the administration of wrong drugs was one of the common errors made in nursing homes [2]. In these chronic care settings, drugs are rarely changed due to the relatively stable conditions of nursing home residents; as a consequence, less attention may be paid to administration of medications. Staff to patient ratios are also lower in long-term care facilities than in acute care facilities. In Japan, the nurse:patient ratio is 1:3 for acute care beds in general hospitals, whereas it is 1:5 for long-term beds in psychiatric hospitals.
Smidt and McQueen studied drug adverse events in a group of hospitals covering all types of in-patient services, and found that these events occur most frequently on medical and psychiatric units [4]. Drug therapy is a principal treatment in internal medicine and psychiatry, and thus the more medicines are used, the greater the risk of medication errors. Moreover, a special feature of psychiatric treatment may also compound the problem of medication errors. Psychiatric patients, especially in long-term care facilities, are often trained in self-administration of medications to increase their self-care and improve their compliance. Compliance, defined as a complex phenomenon that includes a personal contribution to the management of one’s illness [5], is strongly related to factors such as unpleasant medication side-effects and attitudes towards medication [6,7]. Self-administration of medications can improve awareness of illness and enhance medication compliance. However, the patient’s participation in drug administration may also increase the risks of taking the wrong drug. Psychiatry was alerted to the high risk of adverse drug events in 1972 [4]. To our knowledge, scientific studies specific to medication errors in psychiatric care have still not been reported.
Leape et al. emphasize that a first step in error prevention is to develop a better understanding of the types of incidents and their causes [8]. We conducted a multi-center study of potential adverse drug events (PADEs) on long-term psychiatric care units in order to identify the most common types of events and their contributing factors. Our aim was to understand this phenomenon as a basis for constructing prevention strategies in day-to-day practice.
Methods
Subjects
Subjects (facility subjects) were recruited from private psychiatric hospital members of the Kaiseikai Association of Psychiatric Hospitals in Japan. Of 162 members of the association, 44 hospitals participated in the study. We asked these hospitals to select three representative units in each hospital (a total of 132 units) and to introduce a multi-center incident reporting system on PADEs. The reporting system is described below. The study took place between 1 October and 30 November 2000. We received at least one reported incident of a PADE from each of 85 units (64.4%), while there were no incidents in the remaining 47 units. Since we could not assure that no incident reports meant no PADEs and suspected under reporting [9], we excluded units where no PADEs were reported during the study period. The mean number of beds in each unit was 58.2 [standard deviation (SD) 13.9]. More than half the patients in 91.1% of the units stayed for >3 months. Since some psychiatric hospitals in Japan have many long-stay aging patients, these facilities have become de facto nursing homes [10]; in this study, we call these units ‘long-term psychiatric care units’. These units in Japan may be seen as similar to a variety of residential and community facilities in the USA, UK, and Europe, in which the most severely ill patients live and receive their care.
This study was approved by the review board of the Kaiseikai Association. In addition, each participating hospital underwent its own approval through its institutional review board or board of directors. The proposed incident report was also submitted for review and approval to the risk management committee in each facility. If patterns of care potentially harmful to patients were discovered, each hospital’s risk management committee would undertake appropriate quality improvement measures. All incident reports were provided to the Association, with the identity of the patients and hospitals protected.
Incident reporting
The definition of a PADE incident used in this study was ‘any event that could be harmful to patients on participating units, whether the incident was intercepted or not’. These events are preventable if preceded by a medication error. Some hospitals in Japan have already introduced incident reporting systems in which clinical staff are asked to note any medication errors and anything occurring that they believe to be an adverse event. These reports are collected by a head nurse, and are reviewed at risk management or manager-level meetings. Each hospital uses a different format, and information obtained by these reports varies. In this study, we developed an incident report for the multi-center study that was based on previous reporting systems. We also reviewed a monitoring system developed by the Ohio State University Medical Center; its format and definitions of terms were clear and thought to be applicable in Japan [11]. Our study is the first attempt to collect and analyze incidents in psychiatric hospitals in Japan, using the same reporting system.
Incident reporters were clinical staff who were implicated in or responded to harmful events for patients during the study period. Before the study, all clinical staff members in charge of participating units were briefed on the system by head nurses.
The incident report used consisted of three parts: patient characteristics, type of incident, and staff characteristics. Patient characteristics included gender, age, ICD-10 diagnosis, compliance, frequency of medication administration per day, and the number of drugs. With respect to compliance, reporters and the patient’s team members rated the degree of compliance as either ‘consistent’ or ‘inconsistent’. We asked that the type of incident (i.e. missed dose, wrong dose, and/or wrong drug administration), outcomes (e.g. change in frequency of patient monitoring or changes in vital signs), and potential severity be noted on the incident report. We defined wrong drug administration as giving a drug to a patient that was not the drug prescribed for that patient in this study. Potential severity was self-rated by incident reporters using severity definitions of ‘potentially serious’, ‘potentially significant’, and ‘clinically insignificant’, originally developed by the Ohio State University Medical Center [11]. The cutoff point of potentially serious/potentially significant was determined by several definitions such as ‘the drug ordered could exacerbate the patient’s condition, such as a drug–drug interaction or drug–disease interaction’.
We also carried out a survey of unit characteristics on those units involved in the incident reporting system. The survey included: number of beds; staff:patient ratios, including number of psychiatrists, pharmacists, and nurses; the process by which medications are dispensed; and characteristics of the patients, such as age and diagnosis. We also examined the registered nurse ratio among nursing staff members. In Japan, there is a category of ‘licensed assistant nurse’; while nursing practice does not differ greatly between registered nurses and licensed assistant nurses, registered nurses have more education and a more advanced practice degree.
Analysis
We used the Mann–Whitney and Kruskal–Wallis tests for comparison of rank data. Logistic regression was conducted to test the relationships among the types of incidents and related factors. Variables were entered in a hierarchical fashion, beginning with characteristics of incidents followed by variables associated with patient, staff, and hospital characteristics. We used SPSS version 10.0 (SPSS Inc, Chicago, IL) for the analysis.
Results
We received 221 reports of PADEs during the study period, from 85 units (64.4%) in 44 hospitals. Since there were 278 520 total patient-days on these units, the rate of errors reported was 0.79 per 1000 patient-days. The mean age of the patients was 56.9 years (SD 16.1 years). The numbers of men and women affected were 117 (52.9%) and 104 (47.1%), respectively. Regarding psychiatric diagnosis, 122 (55.2%) had schizophrenia, 33 (14.9%) had organic disorders, 16 (7.2%) had mood disorders, 15 had (6.8%) mental retardation, and 35 (15.8%) were diagnosed as ‘other disorder’. One hundred and twenty-two (55.2%) patients were rated as showing good compliance. The mean frequency of drug administration was 3.7 (SD 1.2) times a day. The mean number of tablets was 10.6 (SD 9.7).
Of the 221 reported incidents, 79 (35.7%) were PADEs of wrong drug administered (0.28 per 1000 patient-days) and 142 (64.3%) were other PADEs (0.51 per 1000 patient-days). Of the 79 incidents of wrong drug administration, 12 (5.4%) were caused by patients who self-administered the wrong drug and 11 (5.0%) occurred when patients with the same name were on the same unit. Other reported PADEs were 44 (19.9%) wrong time, 35 (15.8%) missed dose, 30 (13.6%) wrong dose, and 33 (14.9%) ‘other’ incident types. All incidents occurred at the drug administration stage of care in each unit. Twenty-two incidents occurred in seven hospitals that did not have a system to print patient names on each drug pouch. Of these, five hospitals (15 incidents) had only hand-written systems and two hospitals (seven incidents) delivered drugs without patient names printed on the drug pouch, although the name was printed on the prescription package, which is a collection of drug pouches.
Table 1 shows the types of PADEs and outcomes. Of the 221 reported incidents, 55 (24.9%) were intercepted before reaching the patients. Of the 166 (75.1%) not intercepted, 148 patients (67.0%) did not show any clinical changes, though frequent monitoring was instituted; while this is an important safety intervention, it is costly and preferably prevented if possible. The frequency of patient monitoring was increased in 18 patients (8.1%) in response to the PADEs. The rate of wrong drug administration that reached patients was significantly higher than for other PADEs [odds ratio (OR) 1.27; 95% confidence interval (CI) 1.10–1.47]. The registered nurse ratio among nursing staff members was significantly more related to wrong drug administration than to other PADEs (OR 4.12; 95% CI 2.34–7.25). More of this type of PADE was found on units with fewer registered nurses per patient. There were no significant differences in other clinical variables (such as gender, age, ICD-10 diagnosis, compliance, frequency of medications per day, and total number of drugs) or system variables (such as number of beds or staff:patient ratio) between PADEs of the wrong drug administration type when compared with other types.
|
We also investigated potential severity, as it might be related to error type (Table 2). Although 125 (56.6%) incidents were clinically insignificant, 33 (14.9%) were potentially significant and 63 (28.5%) were potentially serious. Potential severity of wrong drug administration was significantly more serious than for other PADE types (OR 1.19; 95% CI 1.03–1.37).
|
Logistic regression was performed for all univariate predictor variables that were significant at P < 0.01. Table 3 shows the final multivariate model. Variables predictive of incidents of wrong drug administration were: (1) the presence of patients with the same name on the same unit; and (2) fewer registered nurses on the unit.
|
Discussion
Our results indicate that more than one-third of reported incidents in long-term psychiatric care hospitals were PADEs of the wrong drug administration type. Although previous studies have generally not focused on this type of PADE, our results are consistent with a previous study in which drug identity checking errors and wrong dose errors were reported as among the most frequent type of errors [1]. The wrong drug administration PADE differs from what might be termed quantitative types of PADEs, such as wrong dose or wrong time. Wrong drug administration is a qualitative incident since it is not a matter of degree. Also, it is more difficult to intercept this potentially serious type of PADE because it occurs at the drug administration stage of care, not at the ordering or dispensing stage of care [1]. Identifying the causes of PADEs is essential if quality improvements in psychiatric care are to be achieved.
Our study points to several causes for PADEs. Firstly, there were identification incidents caused by confusing patients with the same or a similar name. Our results indicate that this type of PADE occurred more often when more than two patients with the same name were resident in the same unit. In addition, similarity of name also increased the risk (e.g. Yama-da and Yama-ji, where the first four characters are the same, much like when there is similarity of spelling or sound-alike names in English). It has been reported that similarity of drug names is also a risk factor for errors [12]. Similarity of patient names should therefore be regarded as another risk factor for wrong drug administration PADEs.
Secondly, we discovered that the registered nurse ratio was a significant predictor of PADEs, consistent with a previous study which reported that staffing levels influenced medication errors [13]. Our results showed a significant relationship between wrong drug administration incidents and the registered nurse ratio, whereas there was no significant relationship between the other types of PADE and staff:patient numbers. Nurses were responsible for 85% of all interceptions. The leading cause of adverse drug events reported is lack of knowledge of medication [1]. Thus, having the appropriate number or ratio of registered nurses on a unit is essential. Increasing the educational experience of all staff members involved in drug administration may also be a very important clinical intervention in reducing medication errors.
Thirdly, the nature of the patient’s disease and psychiatric care were also related to PADEs. A significant proportion of patients with psychotic illness resist taking medication. While most in-patients took the drugs that were offered to them, many patients stated that they were not ill and did not need to take medication [14]. These patients, who lack insight into their illness, tend to be re-hospitalized more frequently and are more non-compliant with treatment [15]. Unit staff can engage patients in drug self-administration to improve their insight and compliance. When successful, more patients will take their medication by themselves. One concern is that those patients who self-administer medications may have a tendancy to take other patients’ drugs, which we found in this study. Although this was not the focus of this study, a more detailed investigation into self-administered medications and safety is required.
Patient safety can be promoted by averting preventable PADEs before they occur. Our results suggest several systematic approaches to help prevent wrong drug administration incidents. Firstly, wrong drug administration was less frequent when patient names were printed on their drug administration pouch. In the hospitals studied, drugs are delivered to each unit in a package, with each patient’s name printed on it. The package has multiple prescription pouches within it. However, some hospitals do not print patient names on each prescription pouch; and even when this is done, it may be hand-written which can often be difficult to read and may cause confusion. Printing patient names on each individual drug pouch enables the patient to check the drug to determine whether it is correct for them or not, thereby adding a patient checking element into the error reduction effort. If this simple and relatively inexpensive intervention were introduced in Japan, or other countries where this is not currently done, staff and patients could check names (i.e. check to determine the correct identity of the recipient) and thereby reduce the risk of this preventable PADE. It may also be advisable to introduce a special checking system for patients with the same or similar name on the same unit. Secondly, a greater number or increased ratio of registered nurses among nursing staff and education for all clinical staff members are other potential error reduction strategies. Increased training and greater professionalism can be achieved in these ways. Finally, it may be very useful to re-examine the participation of patients in drug administration. Engaging the patient in self-care is an important aspect of all rehabilitation strategies and is likely to aid in the prevention of errors [16].
This study has a number of limitations. Firstly, we relied on voluntary reporting. Many researchers have discussed the problem of voluntary participation [17], in particular underreporting [9]. Since PADEs in this study may not be representative of the general long-term psychiatric patient population, important patient predictors may be uncovered should data on the full psychiatric population be known. Secondly, because this was a cross-sectional study, data on the effect of preventative measures (i.e. quality interventions) on PADE reduction could not be obtained. Based on the results of this study, we are now planning to conduct such an interventional approach. Thirdly, our results on compliance are limited and open to interpretation because we did not assess the degree of the therapeutic alliance. We could not find any significant relationship between compliance and the wrong drug administration incident. Further study is needed to assess therapeutic alliance in psychiatric care and to understand its role in patient safety.
Errors are an inevitable outcome in all processes of care. By studying PADEs we can better understand the distinction between preventable and non-preventable incidents, the factors associated with preventable PADEs, and may devise organizational efforts to enhance patient safety. From our study, we suggest that long-term psychiatric care hospitals should focus on reducing incidents of the wrong drug administration type. Implementation of advanced informational systems, including electronic medical records and bar coding of medication, require a large amount of money. This paper offers various strategies to reduce this type of error without such costly investment.
This work was supported by a Health Sciences Research Grant from the Ministry of Health, Labor and Welfare, Japan (13162401). We thank Lloyd I. Sederer, M.D., Director, Division of Clinical Services, American Psychiatric Association, for his advice on this research.
Address reprint requests to Dr Hiroto Ito, Healthcare Evaluation Section, Department of Management Science, National Institute of Public Health, 2-3-6 Minami, Wako, Saitama 351-0104, Japan. E-mail: ito{at}niph.go.jp 
Accepted for publication February 13, 2003.
References
- Leape LL, Bates DW, Cullen DJ et al. System analysis of adverse drug events. J Am Med Assoc 1995; 274: 35–43.
[Abstract/Free Full Text] - Kapborg I, Svensson H. The nurse’s role in drug handling within municipal health and medical care. J Adv Nurs 1999; 30: 950–957.[CrossRef][Web of Science][Medline]
- Bates DW. A 40-year-old woman who noticed a medication error. J Am Med Assoc 2001; 285: 3134–3140.
[Free Full Text] - Smidt NA, McQueen EG. Adverse reactions to drugs: a comprehensive hospital inpatient survey. NZ Med J 1972; 76: 397–401.[Medline]
- Sackett DL, Haynes BR. Compliance with Therapeutic Regimens. Baltimore, MD: The Johns Hopkins University Press, 1976.
- David AS. Insight and psychosis. Br J Psychiatry 1990; 156: 798–808.
[Abstract/Free Full Text] - Kampman O, Lehtinen K. Compliance and psychosis. Acta Psychiatr Scand 1999; 100: 167–175.[Web of Science][Medline]
- Leape LL, Brennan TA, Laird N et al. The nature of adverse events in hospitalized patients: results of Harvard Medical Practice Study II. N Engl J Med 1991; 324: 377–384.[Abstract]
- Gandhi TK, Seger DL, Bates DW. Identifying drug safety issues: from research to practice. Int J Qual Health Care 2000; 12: 69–76.
[Abstract/Free Full Text] - Ito H, Sederer LI. Mental health services reform in Japan. Harvard Rev Psychiatry 1999; 7: 208–215.[CrossRef][Web of Science][Medline]
- Joint Commission on Accreditation of Healthcare Organizations. Medication Use: a Systems Approach to Medication Use. Oakbrook Terrace: Joint Commission on Accreditation of Healthcare Organizations, 1998.
- Lambert BL, Lin SJ, Chang KY, Gandhi SK. Similarity as a risk factor in drug-name confusion errors: the look-alike (orthographic) and sound-alike (phonetic) model. Med Care 1999; 37: 1214–1225.[CrossRef][Web of Science][Medline]
- O’Shea E. Factors contributing to medication errors: a literature review. J Clin Nurs 1999; 8: 496–504.[CrossRef][Web of Science][Medline]
- McEvoy JP, Apperson LJ, Appelbaum PS et al. Insight in schizophrenia: Its relationship to acute psychopathology. J Nerv Ment Dis 1989; 177: 43–47.[Web of Science][Medline]
- McEvoy JP, Freter S, Everett G et al. Insight and clinical outcome of schizophrenic patients. J Nerv Ment Dis 1989; 77: 48–51.
- Cohen MR, Anderson RW, Attilio RM, Green L, Muller RJ, Pruemer JM. Preventing medication errors in cancer chemotherapy. Am J Health Syst Pharm 1996; 53: 737–746.
- Cullen DJ, Bates DW, Small SD, Cooper JB, Nemeskal AR, Leape LL. Incident reporting system does not detect adverse drug events: a problem for quality improvement. Joint Comm J Qual Improv 1995; 21: 541–548.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Haw, J. Stubbs, and G. Dickens An observational study of medication administration errors in old-age psychiatric inpatients Int. J. Qual. Health Care, August 1, 2007; 19(4): 210 - 216. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. D Maidment, P. Lelliott, and C. Paton Medication errors in mental healthcare: a systematic review Qual. Saf. Health Care, December 1, 2006; 15(6): 409 - 413. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. D. Maidment and A. Thorn A medication error reporting scheme: analysis of the first 12 months Psychiatr. Bull., August 1, 2005; 29(8): 298 - 301. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||