Are there more drug errors in intensive care units?

Synopsis: Potential or actual adverse drug events were more common in ICUs than on general wards, and they happened more often in medical than in surgical ICUs, but when the data were corrected for the number of drugs administered, these differences disappeared.

Source: Cullen DJ, et al. Crit Care Med 1997; 25:1,289-1,297.

Adverse drug events account for the largest percentage of preventable patient injuries and have been estimated to cost around $6,000 per incident. They are a frequent occurrence in ICUs, often attributed to caregiver stress, fatigue, and poor communication. Certainly, the consequences of these occurrences are expected to be greater in more fragile, critically ill patients than in general ward patients. Cullen and colleagues attempted to determine the frequency of potential and actual adverse drug events (ADEs) in ICUs and general wards, their severity, underlying systemic causes, and economic effect. They prospectively studied all patients admitted to one of five ICUs (two SICUs, three MICUs) or six general wards (two surgical, four medical) during a six-month period. Investigators visited each unit twice daily, reviewed records, and interviewed pharmacists, nurses, and physicians to identify drug administration problems.

Once an actual or potential adverse drug event was identified, it was analyzed through nonjudgmental interviews with all involved, analysis of the circumstances, quantification of the severity of the event, and patient factors. Caregiver circumstances included workload, perceived stress, fatigue, amount of sleep in the preceding day, staffing patterns, staff vacancies, and quality of communication. Patient factors included age, location (unit), severity of illness (TISS points), and number of drugs prescribed in the preceding 24 hours. Length of stay was determined for all patients studied. The number of ADEs was reported per 1000 patient days (PD).

Included in the study were 4,031 patients in 11 units (both ICUs and wards). There were 19 actual and potential ADEs/1000 PD in the ICUs, compared with 10/1000 PD in the general wards (P is less than 0.01), 25/1000 PD in MICUs, compared with 9.7/1000 PD on medical wards (P is less than 0.01); and 14.4/1000 PD in SICUs, compared with 10.8/1000 pd on surgical wards (P is less than 0.01). There was no difference between medical and surgical wards. When the number of ADEs was divided by the number of drugs given, the rates in all areas were identical — the range being 0.61-0.65 ADE/1000PD/drug ordered. No differences in patient demographics or degree of illness were identified to account for the findings. ADEs had more serious effects on ICU patients than on those in wards, with no differences between medical and surgical patients. No patients had permanent injury, and no statistical effect on length of stay was found.

In analyzing the individual events, no pattern of staffing or caregiver stress was apparent.

Staffing was perceived as adequate, and communications appeared reasonable. ADEs often occurred when patient demands were easily met and in patients representing the spectrum of sickness. Caregivers appeared to appreciate the opportunity to candidly unburden themselves of associated guilt during the interview, making dishonesty less of a concern for the study validity. From this study, Cullen and colleagues were unable to identify systemic factors contributing to ADEs and further evaluation will be necessary.

Since the only consistent predictor of rate of ADEs was the number of drugs the patients were receiving, reduction in total drug therapy would be a strategy for decreasing ADEs.

COMMENT BY CHARLES G. DURBIN, Jr., MD, FCCM, medical director of respiratory care, University of Virginia, Charlottesville.

Accidents and errors contribute to patient morbidity, mortality, and cost of care. There is also a mental cost to caregivers involved in adverse events in terms of guilt and burnout. Prevention of adverse events improves the life of patients and caregivers. Adverse drug events are among the easiest types of events to identify and study in the ICU environment, and they can serve as a surrogate for other types of events.

The surprising finding of this study is that the most important factor contributing to an ADE was the total number of different drugs the patient received during the day before the event. Although event severity was related to patient factors, the likelihood of an adverse event in the first place was a function of the total number of drugs given. This suggests that, with every drug, there is a finite risk of an adverse event and the risk then becomes cumulative. This being true, analysis of the process of selection, ordering, preparing, and administering medications should identify weaknesses in the system. In this study, no predominant single aspect of the medication chain was found, although errors in ordering and in administration were frequent.

The incidence reported in this paper is undoubtedly lower than the actual rate of ADEs due to voluntary reporting. Creating a supportive, nonjudgmental environment rather than searching for the guilty person encourages self-reporting by caregivers and is important in improving patient care. The relief reported by ICU staff members in this study after being interviewed by the investigators lends support for this practice.

This report has several shortcomings. It included only two hospitals and a relatively small number of units. A larger variety of hospitals and units will be necessary to confirm these findings. As mentioned, self-reporting contributes significant concerns about under-reporting. Although Cullen et al made independent chart reviews, this is unlikely to identify potential or minor ADEs.

This study failed to confirm the conventional wisdom that stress is a major contributor to human errors. That the units studied were well-staffed with little staff turnover may be a historical anomaly. Stress from shortages and change from redesign seems to be accelerating lately. It would be interesting to repeat this study now to evaluate these factors and their contribution to ADEs. Applying the authors’ methods to daily care could contribute to quality improvement.