2038. Predicting the Quality
of Preventive Care: A Prospective
Evaluation of Three Measurement Methods
TR Dresselhaus, HSR&D REAP-Center for Research in Patient-Center
Care, VA San Diego Healthcare System, J Luck, Center for the Study of
Healthcare Provider Behavior, School of Public Health, University of California,
Los Angeles, VA Greater Los Angeles Healthcare System, DS Bertenthal, San
Francisco VA Research Enhancement Award Program, JW Peabody, San
Francisco VA Research Enhancement Award Program, Institute for Global Health,
University of California, San Francisco and Los Angeles
Objectives: To
prospectively evaluate variation in quality of preventive care at multiple sites
using standardized patients chart abstraction and vignettes and to determine
whether clinical vignettes predicts these variations.
Methods: 72 physicians in
2 VAMC primary care clinics and 2 community clinics were randomly selected among
consenting physicians (95%). We
compared 3 measurement methods for 4 common conditions:
(1) standardized patients (SPs) presenting unannounced to physicians’
clinics; (2) abstraction of the SP medical records and; (3) clinical vignettes
that exactly corresponded to the SPs. Physicians
completed 480 visits. Scoring
criteria were based on national guidelines for 12 prevention measures and
categorized as Vaccines, Vascular, Cancer Screening, or Personal Habits.
We calculated the proportion of prevention items completed for the 3
methods. We also developed a
multiple regression model to predict performance using half the data;
subsequently, this model was applied to the remaining data to determine if site,
training level, or clinical condition predicted quality of preventive care.
Results: Measurements
of the quality of preventive care ranged from 57% (SPs) to 54% (vignette) to 46%
(chart abstraction). Vignettes
matched or exceeded SP scores for 3 of 4 categories (Vaccine, Vascular, Cancer
Screening); charts were lowest in all 4 categories.
We found significant variation in most sites (p <.05 for SPs and
charts at 4 sites, and p <.05 for Vignettes at 3 sites), conditions (p
<.05 for SPs and charts for 3 conditions, and p <.05 for vignettes for 4
conditions) but no difference by training
level. Quality of preventive care at VA sites was superior to community sites
(vignettes: p < 0.05).
Conclusions: These data
indicate overall poor quality of preventive care regardless of measurement
method. Chart abstraction, as a
method, appears to be subject to recording bias.
As hypothesized, quality varies significantly depending upon site and
condition, but not training level. Clinical
vignettes are comparable to SPs in measuring variation in quality and therefore
may be useful for evaluating preventive care performance.