Practice guidelines to reduce testing in the hospital.

OBJECTIVE: To reduce testing among hospitalized patients using practice guidelines for any of 14 medical problems.
DESIGN: Comparison of test use before and after implementation of guidelines. The guidelines were developed by consensus panels of self-selected participating physicians. Non-participating physicians were monitored during the same periods. In addition, the two groups of physicians were evaluated similarly for their management of three medical problems for which guidelines were not developed.
SETTING: Acute care hospital. PATIENTS/PARTICIPANTS: 1,638 hospitalized patients and their 79 physicians. INTERVENTION: Implementation of practice guidelines for the care of hospitalized patients. MEASUREMENT AND MAIN
RESULTS: Geometric mean charges expressed in inflation-adjusted dollars were used as measures of test use. For the intervention group, laboratory tests decreased by 20.6%, x-rays by 42.3%, and EKGs by 34.2%. All the decreases were significant (p = 0.001). The non-participating physicians who were higher test users during both years of the study also achieved significant (p less than 0.05) but smaller reductions during the intervention year: 13.9% for laboratory tests, 30.3% for x-rays, and 21.8% for EKGs, perhaps because the same residents were involved in the care of both groups of patients. For the non-guideline diagnoses, the participating physicians achieved reductions of 11.1% for laboratory tests and 19.2% for x-rays, and a 3.5% increase in EKGs. Two-way analyses of variance that took into account the reductions in testing achieved by non-participants, or by participants with non-guideline diagnoses, revealed no significant reduction in testing attributable directly to the guidelines except for EKGs. Follow up of the participating physicians during the six months after the end of the intervention revealed that testing remained at the lower level achieved while the guidelines were in use. Outcome of care, as measured by deaths in the hospital, deaths within 90 days of discharge, and readmissions within 90 days of discharge, was not affected by the use of the guidelines.
CONCLUSIONS: 1) A large group of physicians could be recruited in a hospital to establish practice guidelines by group consensus. 2) These self-selected physicians were willing to use the guidelines (or allow the housestaff to use them) while caring for their patients. 3) Participating physicians were able to achieve substantial and significant reductions in testing without any demonstrable adverse effect on quality of care as measured by deaths and readmissions, and without any demonstrable shifting of resources from the inpatient to the outpatient setting of care. 4) The reductions in testing, whether caused by the guidelines or not, persisted for at least six months beyond the end of the period of implementation.