Skip to main content Back to Top


Computerized Systems Can Change Prescribing

Nancy Tarleton Landis

Electronic order entry by physicians can improve medication prescribing in several ways, research at a large academic medical center confirms.1 Physicians at Brigham and Women's Hospital in Boston have used computer work stations to write all orders for adult inpatients since 1993. Orders for the recommended drug within a class have increased, while variability in dosage and administration has decreased.

System features. For each order entered, the drug name, route, dose, and frequency of administration must be entered on a structured form. Space is also provided on the form for duration, instructions, and whether the drug is to be given routinely or as needed. When the physician enters the drug name, the system displays suggested doses (the physician can also choose "other" to enter a different dose) and highlights the recommended dose and frequency. The system's clinical decision-support features also check for allergies and drug–drug interactions, duplicate medications, and potential alternatives for the particular patient's condition. The physician is alerted if one order prompts the need for another order; for example when bed rest is ordered, the system suggests subcutaneous heparin to prevent thrombosis.

Study findings. Data for periods before and after implementation of the computerized order entry system showed the following:

  • Orders for nizatidine, the preferred oral H2-blocking agent, accounted for 12% and 16% of all orders for drugs in the class in two four-week periods before implementation, compared with 81% and 95% afterward. One and two years later, 97% of all orders for oral H2-blockers were for nizatidine. 
  • Mean dosages and standard deviations for each medication ordered during a six-month period before implementation were compared with those for a one-month period after implementation. The means were similar, but the standard deviation decreased 11% after implementation and this trend continued through the third year of system use. In a similar analysis, the standard deviation for frequency of administration decreased 30%.
  • Before implementation, 2.1% of orders were for doses exceeding the recommended maximum; this decreased to 0.56% in the first month after system implementation and continued to decrease thereafter.

Additionally, the pharmacy and therapeutics committee decided that i.v. ondansetron could effectively be administered three instead of four times per day, and the recommended frequency on the ordering screen was changed. The percentages of orders for ondansetron three times per day rose from 6% before the screen change to 75% and 94% in the two four-week periods after the change, and the pattern persisted in three annual follow-up samples.

The system prompt for subcutaneous heparin when bed rest is ordered was implemented in 1994. In the five-month periods before and after implementation, heparin orders accompanied 24% and 47%, respectively, of bed rest orders. One and two years later, the percentages were 48% and 54%.

Conclusions and cautions. Computerized recommendations for changing dose, frequency, or drug choice within a class "were readily accepted and had a very large impact," wrote the researchers. It is more difficult, said the authors, to effectively implement recommendations to stop an intended action that do not offer an alternative action; such recommendations are seen as changing the plan of care.

Physician order entry with clinical decision support can increase compliance with recommendations and guidelines "by presenting them at the exact moments when they are most relevant," the authors concluded. They urged caution in using order entry databases to identify prescribers who fail to follow guidelines and recommendations. Acceptance of physician order entry could be reduced, they said, if prescribers see the computer database primarily as a way for others to monitor their performance.

An accompanying editorial2 notes that electronic ordering systems are needed in ambulatory care environments as well as hospitals. In addition, the editorialists call for improvements in electronic systems for pharmacy-based drug-use evaluation and review. Alerts should be sent only for drug therapy problems for which there is high-quality evidence, and the emphasis should be on problems associated with severe adverse outcomes. The authors suggest that standards be established, by a nonprofit, nongovernmental organization, for computerized prescribing and drug-use monitoring.

1. Teich JM, Merchia PR, Schmiz JL et al. Effects of computerized physician order entry on prescribing practices. Arch Intern Med. 2000; 160:2741-7.
2. Armstrong EP, Chrischilles EA. Electronic prescribing and monitoring are needed to improve drug use. Arch Intern Med. 2000; 160:2713-4. Editorial.