Competing demands can limit the ability of primary care clinicians to provide high quality depression treatment. In the primary care medical setting competing demands can be conceptualized and measured at the level of the visit, patient, provider, and practice ecosystem. Provider burden, a source of competing demands arising from the collective effect of biomedical complexity and psychosocial needs of patients, may be particularly salient for depression care. If high provider burden is shown to be a significant predictor of depression quality and/or outcomes in usual care, it will be critical to determine whether current depression interventions are effective in this provider group. While provider burden is a promising construct, it has been under-conceptualized and under-investigated. Rigorous investigations of the effects of provider burden on process of care and clinical outcomes will require multilevel modeling, which has been increasingly incorporated into mental health services investigation over the past decade, but with fewer applications to process of care outcomes that often consist of binary or count-type data. Composition modeling, developed in organizational and educational research, addresses measurement issues in multilevel research but is relatively new in health research. Three approaches to measuring provider burden include (1) direct surveys, (2) construction of summary measures using data from providers/practices, and (3) construction of summary measures using data from patients, aggregated to the provider level using composition models. Because direct surveys are expensive, the current application focuses on strategies (2) and (3) as a preliminary step in exploring alternative ways of measuring provider burden. We propose to develop measures and conduct analyses using data from 1117 depressed patients of 156 primary care providers enrolled in the Quality Improvement in Depression (QID) studies. We plan to pilot test existing survey measures and alternative measures in multilevel models, examining whether provider burden impacts depression quality of care during the 6 months post-intervention, after adjusting for visit and patient-level competing demands, the intervention effect, and other relevant covariates. We also propose to pilot test existing and alternative measures in multilevel models to explore whether provider burden moderates depression intervention effectiveness, after adjusting for visit and patient level competing demands and other relevant covariates.
| Dickinson, L Miriam; Dickinson, W Perry; Rost, Kathryn et al. (2008) Clinician burden and depression treatment: disentangling patient- and clinician-level effects of medical comorbidity. J Gen Intern Med 23:1763-9 |
