Likelihood of ordering physical restraints: influence of physician characteristics

To determine physician knowledge regarding restraint regulations and effectiveness and effect of physician characteristics on likelihood of ordering restraints.

Cross-sectional, factorial research survey.

Academic medical center.

Interns in all specialties; residents in internal medicine, family practice, emergency medicine, psychiatry, and surgery; and attending faculty at an academic medical center.

Survey of demographic, professional, and restraint knowledge items and for each of five distinct vignettes; physician ratings of probability of patient harm and likelihood of ordering restraints. For each, physicians rated probability of patient harm and likelihood of ordering restraint.

One hundred eighty-nine of 246 (77%) surveys were returned. More than half (58%) were men; the median age was 30 (range 25-63), median years experience was 2 (range 0-33), and 60% were U.S. medical school graduates. Mean knowledge score was 68.4% (range 27-100%). Mean likelihood of ordering restraints ranged from 0.6 (not likely) to 9 (absolutely) (overall mean 3.9 +/- 2.2). Exploratory hierarchical regression on mean likelihood of ordering restraint (outcome) with independent variables of physician age and sex (Step 1), years experience and physician level (Step 2), specialization (Step 3), restraint knowledge (Step 4), and judgment of harm (Step 5) explained 31.9% of the variance (F=7.19, degrees of freedom 13,159, P<.001). Higher appraisal of harm (P<.001), less knowledge regarding restraint (P=.03), and male sex (P=.005) were unique indicators for the likelihood of ordering restraints. Psychiatry (P=.03) or internal medicine physicians (P=.05) were less likely to order restraints.

Physician characteristics and lack of restraint knowledge are associated with likelihood of ordering restraints. Results will guide medical education initiatives to reduce restraint rates.