Authors:
J.W. WILLIAMS; Durham VAMC, Durham, NC
W. GOLDEN; Arkansas Foundation for Medical Care, Inc., Ft. Smith, AR
P.E. KECK; University of Cincinnati, Cincinnati, OH
M. JEWELL; EPI-Q, Inc, Oak Brook, IL
C. BREWSTER; EPI-Q, Inc, Oak Brook, IL

Abstract:

BACKGROUND:
In people age 15-44, bipolar disorder (BD) is the 6th leading cause of loss in disability adjusted life years. For individuals with BD, the gap between usual and guideline concordant is substantial. As a tool to facilitate quality improvement, we developed and field tested performance indicators (PI) for BD.

METHODS:
A multidisciplinary expert panel used a modified Delphi process and the RAND appropriateness criteria to develop 16 candidate PI. A convenience sample of 80 practices (48 psychiatry, 32 primary care) in 28 states used chart audit, guided by a detailed data collection form and data dictionary, to field test the PI. Each practice identified patients via automated data (e.g., ICD-9 codes) with depression or bipolar disorder for chart audit. A subset of patient records were re-abstracted for inter-rater reliability. Feasibility was evaluated by: the number of patients eligible for the PI, the ability to apply the PI based on information available in medical records, and the variability in performance on the PI. Inter-rater reliability was described using simple agreement and the kappa statistic.

RESULTS:
The 16 candidate indicators included measures to screen for mania and for a family history of mental health disorders in unipolar depression, and measures to assess suicide risk and alcohol or substance use in patients with unipolar or BD. Twelve measures specific to bipolar disorder addressed treatment (5), monitoring for treatment adverse effects (4), response to treatment (2), and education (1). 802 cases (419 BD; 383 Depression) were audited. Six BD-specific PI (3 treatment, 2 adverse effect monitoring, 1 treatment response) had low feasibility based on few patients qualifying for the indicator or little variability in measured performance. For the depression specific PI, practice performance was as follows: assessment for mania (47.6%) and obtained a family history of mental health disorders (40.8%). Performance on measures applicable to unipolar or BD were: suicide risk assessment (71.5%) and assessment for substance abuse (60.5%). For the BD specific PI, practice performance was: serum level when treated with lithium (44.4%), level of functioning assessed at baseline and with 12 weeks (41.4%), appropriate psychosocial interventions recommended (39.9%), provided education on BD (33.4%), assessment for hyperglycemia after initiating an atypical antipsychotic (19.7%) and weight measurement (15.5%). Psychiatric practices performed better than primary care practices on 7 of the 10 feasible PI (p<0.05). Primary Care Practices were more likely to measure weight, and did not differ significantly for hyperglycemia monitoring or obtaining serum lithium levels. Inter-rater reliability was substantial in the 68 charts that were double abstracted; simple agreement ranged from 89.5% to 100% and kappa ranged from 0.75 to 1.0.

CONCLUSION:
There is a need to improve the quality of care for individuals with bipolar disorder. These ten reliable PI can be implemented using chart audits and may be a useful component of quality improvement efforts.