Assessing depression treatment and *outcomes using insurance claims data

Gregory Andrew Allard, University of Rhode Island


This retrospective, longitudinal, quasi-experimental study uses insurance data to compare long-term outcomes among an adult population by depression history and treatment level. Treatment levels were assigned based on claims history correspondence with Agency for Health Care Policy and Research (AHCPR) depression outpatient treatment guidelines. ^ Thirty-month outcomes included depression resurgence and total health care utilization (total utilization). Recurrent depressives (recurrents) were expected to experience resurgence at higher rates than single episode depressives (singles). Additionally, cases with AHCPR-recommended (preventative) treatments were expected to have lower resurgence rates compared to cases that received lower-level treatments. Also, total utilization in cases receiving preventative treatments was expected to be less than or negligibly different from that of other groups throughout follow up. ^ In a large Hawaiian insurance plan sample (N = 4014), only 12.1% received treatment corresponding to AHCPR guidelines. Only 4.7% received preventative treatments. Compared to singles, recurrents were less likely to have received preventative treatments, despite greater need. ^ The study provided partial support for hypotheses. With either of two methods used to detect resurgence, a higher proportion of recurrents experienced resurgence compared to singles. Compared to those without, resurgence rate findings for those with preventative treatments were mixed and were influenced strongly by the resurgence method used. ^ Compared to those without, cases with preventative treatments experienced higher total utilization during the first 18 months of treatment. Thereafter, total utilization differences were negligible. The occurrence of sentinel events differed negligibly among treatment groups throughout the follow up period, however. ^ Depression is a major public health problem and is among the top ten costliest in the nation. Major barriers to improving the problem can be overcome if existing longitudinal population-based data, such as insurance claims, can be exploited. ^ Although only partial support for hypotheses was reported, the findings, particularly the low prevalence of preventative treatments, warrant continued research using claims data. Validation studies are needed. Resurgence method was found to strongly influence outcomes and would benefit from additional testing. Selection factors apparently bias depression history and treatment group assignment methods. Validation of case assignment methods would benefit from concurrent validation with validated clinical measures. ^

Subject Area

Health Sciences, Mental Health|Health Sciences, Public Health|Health Sciences, Health Care Management

Recommended Citation

Gregory Andrew Allard, "Assessing depression treatment and *outcomes using insurance claims data" (2001). Dissertations and Master's Theses (Campus Access). Paper AAI3025527.