Over-reliance on pharmacy computer systems may place patients at great risk

From the February 10, 1999 issue

Our recent computer field test and survey results resemble conclusions drawn from a Georgetown University study on the state of pharmacy computer systems. Published August 26, 1996 in US News and World Report, the cover story, "Danger at the Drugstore," concludes that pharmacy computer systems in this country are of limited reliability when used to detect and correct prescription errors, most notably serious drug interactions. Our findings, from over 300 respondents, show that pharmacy computer systems need serious improvement. Many respondents' systems performed poorly when tested with specific questions to assess the computer's ability to detect serious or fatal errors reported to ISMP in the previous year. Chillingly, only 4 out of 307 systems tested detected all unsafe orders presented in the field test.

It's frightening that the majority of our nation's pharmacy computer systems may be incapable of detecting orders that exceed a set maximum dose. Yet, only about 38% of respondents' systems detected lethal overdoses of both cisplatin (204 mg for a 26 kg child) and vincristine (3 mg for a 2 year old). Only about 34% detected lethal colchicine doses (both single and cumulative). 87% did not detect an excessive antibiotic dose for a patient with renal impairment (tobramycin 120 mg IV q8h for a patient whose creatinine clearance was 10 mL/min). Only 42% of the respondents' systems are even linked between the pharmacy and laboratory to obtain this important information! Moreover, less than half (41%) of the linked systems automatically screen orders against current laboratory values to alert staff to drug therapy concerns. Respondents' systems fared somewhat better detecting a ketorolac and aspirin cross allergy (88%), drug interaction between ketoconazole and cisapride (85%) and therapeutic class duplication with famotidine and nizatidine (82%). However, only 35% detected drug ingredient duplication with acetaminophen and Percocet. Also, just 39% alerted staff or blocked entry for an oral suspension ordered IV or would not allow the order to be entered at all.

On the plus side, 75% of respondents considered the interaction warnings appearing during order entry to be clinically significant or easily eliminated if considered insignificant. However, with the exception of a warning or blocked entry for an oral suspension by IV route, most respondents' systems require no staff action (such as entering a password) to ensure that the warning is acknowledged. Thus, each warning can be bypassed simply by pressing a function key and may be overlooked by those working under pressure.

The most common computer vendors used by respondents include HBOC (27%), Meditec (16%), Cerner (10%), SMS (6%), Pharmakon (6%), and Digimedics (6%). The most frequent drug information providers include First Data Bank (61%), Medispan (25%) and Medicom (7%). When results for each vendor's computer system and drug information provider were analyzed separately, no computer system or drug information provider was better than another at detecting unsafe orders. Each had significant limitations similar to those listed above. Also, respondents at different sites using the same vendor's computer system and/or drug information provider reported considerable variation in the ability to detect unsafe orders. This is likely due to differing system applications, variations in maximizing system capabilities, and varying frequency of receiving/integrating drug information updates. While most respondents reported that they receive and integrate drug information updates either monthly (36%;40%) or quarterly (48%;54%), 3% of respondents reported that they never receive drug information updates and 8% reported that they never integrate the information! Additionally, almost one third of respondents indicated that it would take six months (22%) to one year or longer (9%) to integrate new drug interactions reported by the FDA.

Although no system detected unsafe orders significantly better than others, Meditec users consistently reported a higher level of additional system capabilities when compared to other systems. These included the ability to build alerts for error prone situations (84%); produce warning override reports (94%); provide clinically significant drug interaction alerts or eliminate insignificant alerts (88%); produce computer generated MARs (92%) to their satisfaction (81%); directly link with the laboratory system (78%) and automatically screen orders (50%); integrate inpatient and outpatient records (79%); direct physician order entry capability (27%); and satisfaction with vendor support (71%). However, users also reported frustration with the complexity of the system. Overall, including all vendors' systems, respondents reported the lowest capability for integrating inpatient and outpatient records (37%) and direct physician order entry (13%).

While pharmacy computer systems have become a common and essential professional tool to increase staff efficiency and support effective drug therapy monitoring, our study shows that pharmacists must not rely on this tool alone. Vendors may market systems by promoting the detection of unsafe orders. Yet, it is clear that, in practice, complex self-programming and the unrealistic time commitment necessary to achieve desired results may prohibit full use of the systems' capabilities. Thus, such vendor claims are meaningless unless the system applications are user friendly, allow maximum capabilities to easily be achieved, and they are not cost prohibitive. Nonetheless, with 69% of respondents reporting that their systems allow them to build alerts, it is crucial to maximize a systems' capabilities whenever possible by responding to serious error prone situations reported in the ISMP Medication Safety Alert! Meanwhile, we plan to use the survey results to promote improved pharmacy computer technology for more effective recognition of subtherapeutic and excessive doses, allergic and cross allergic reactions, drug class and ingredient duplication, drug interactions, and contraindicated drugs or those needing dose adjustment for specific diseases or other clinical situations. We thank all who responded to the survey to make this goal a reality.

Need a new computer system? Check out ISMP's top ten signs