(223) Modeling Medical, Prescription, Disability, and Absence Costs for Employees with Major Depressive Disorder (MDD): Using Shorter Pre-Index Time Periods for Charlson Comorbidity Indexes (CCI) and Smaller Sample Sizes

Background: Models of disease outcomes often control for comorbidities using measures such as the Charlson Comorbidity Index scores (CCI). Predicting future outcomes based on prior period CCIs are limited by sample sizes with continuous eligibility for the entire pre- and post-periods.

Objectives: Assess precision of Rx, medical, absence and disability cost models using various durations of pre-index CCIs to predict annual post-index metrics for employees (EMPs) with major depressive disorder (MDD).

Methods: Retrospective analysis of US EMPs in Workpartners’ Research and Reference Database (RRDb). EMPs with MDD ICD-9/ICD-10 codes were identified from medical claims (2010—2022). EMPs had ≥1 year continuous eligibility post-index date (1st claim for MDD cohort, average of 1st claim date for control group). Ten random cohorts with 3 controls were randomly selected per eligible EMP. For each cohort cost models were developed using 2-stage (logistic followed by generalized linear) model stepwise regression. All models controlled for differences in demographics (age, gender, self-reported race, marital status), job-related factors (salary, full-/part-time status, exempt/non-exempt status), US region, and comorbidities (using 6, 9, and 12 month [M] pre-index baseline CCIs) . Variables selected ≥8 of 10 times were used in the final models. Costs were inflation adjusted by CPI to December 2022 US dollars and mean modeled differences (MMD) presented for medical and Rx costs, short-term disability (STD) costs and absence costs. EMP STD and absence days were also modelled and reported.

Results: All cost models included CCI at p< 0.05 for ≥1of the 2-part models, regardless of CCI time frame. For every metric tested, all 10 models run had statistically significant differences (MDD vs. controls), except the 12M Rx cost baseline CCIs [bCCIs] with 9/10 models (P < 0.05). Medical costs MMDs were $6,370, $6,395 and $6,470 for 12M, 9M and 6M baseline CCIs [bCCIs], respectively. Rx costs were $1,043, $1,140 and $1,055 for 12M, 9M and 6M bCCIs, respectively. STD costs MMDs were $641, $656 and $621 for 12M, 9M and 6M bCCIs, respectively. MMD in 12M post-index absence costs between cohorts were $945, $935 and $949 for 12M, 9M and 6M bCCIs, respectively. STD day MMDs were 3.80, 3.86 and 3.70 and absence day MMDs were 3.98, 3.97 and 3.91 for 12M, 9M and 6M bCCIs, respectively.

Conclusions: A six month CCI value is an effective control covariate for predicting annual post-index outcomes for direct costs (prescription and medical), and indirect costs and times associated with absences and disability leaves. This strongly suggests the potential for pre-index eligibility periods less than one year, which allows increased sample sizes for populations with significant turnover.