Needs-based planning of mental health and substance use services in British Columbia
For the purposes of the administrative data analysis component of this project, we have three research questions. The first two are condition specific, the third will combine results from across disorder groupings.
1) How to define a prevalent case of [a specific MHSU disorder] within the administrative data, so as to estimate the treated prevalence of [a specific MHSU disorder] within a defined period of time (e.g. annual prevalence)?
Hypothesis: For each disorder, some combination of MSP and DAD claims of a disorder-specific diagnosis will define a case for that specific MHSU disorder. Applying this definition across the treated population will produce a treated prevalence rate at or below the expected population prevalence estimate produced in Phase 1.
Based on previous exploratory analyses of administrative data, we believe that the success of this question will vary by disorder, as some disorders are better represented in the administrative data than others. Identification of an MHSU disorder within administrative data can be challenging due to issues around the reporting of specific MHSU diagnoses. Some of these issues exist at the level of care. For example, an individual may interact several times with the health system before a definitive diagnosis is achieved. Or individuals may end treatment midway through a recommended treatment plan. Stigma often plays a role in both these patient-related decisions, as well as in the diagnosis process of the healthcare provider. Other issues exist at the level of data reporting. For example, some diagnostic codes may not be consistently reported at the the level required for identification of specific disorders (e.g. substance-related diagnoses). In some cases, gaps in the reporting of health service data may lead to incomplete pictures of treatment (e.g. an absence of community-level treatment data at the provincial-level). Solutions to these challenges have, historically, been developed on an ad-hoc basis according to the immediate needs of a given project. This can generate inconsistencies in how cases are reported in the literature. We will address these issues by applying a systematic approach, which will standardize the way in which cases are defined both within and across MHSU disorders. These efforts will also be informed by the evidence obtained in Phase 1, regarding expected population-level prevalence and distribution by severity.
2) For a given disorder, is it possible to subdivide the prevalent cases according to severity (reflective of the severity levels established in Phase 1) using combinations of the MSP, DAD, and PharmaNet administrative data?
Hypothesis: Within a disorder cohort, severity levels can be determined by some combination of physician visits (MSP data) as well as hospitalizations (DAD, if relevant) and medication regimes (PharmaNet data, if relevant).
We anticipate that the degree of success in establishing these definitions will once again vary by disorder type. But in general, more severe cases should be reflected in higher service utilization, which should be reflected in the MSP, DAD, and PharmaNet data.
3) For disorders where question 1 and 2 are possible, what is the commodity pattern across these disorders, according to severity?
Hypothesis: Comorbidity among mental illness and substance abuse is predicted to be high. For some disorders, it is expected to vary by severity level.
We anticipate that our ability to address this question will vary be disorder, in accordance with the success of research questions 1 and 2. The ability to produce comorbidity estimates by severity is expected to better inform the service planning integration and system sizing components of Phase 2.