This research presents results from an application of “capture-recapture” biostatistical methodology to the estimation of the size of a relatively “closed” population of public health interest: the number of adults with developmental disabilities in Montana. To calculate an ascertainment-corrected prevalence rate of the number of adult Montanans with developmental disabilities in 1998, data was linked using encrypted Social Security Numbers from four administrative data sources provided by Montana State's Disability Services Division's Vocational Rehabilitation and Developmental Disabilities Programs, Senior & Long-Term Care Division, and Medicaid Management Information System. The Montana statutory definition of developmental disabilities was used for determining case inclusion in each data set. Therefore, we selected ICD-9 codes for diagnoses of mental retardation, cerebral palsy, autism, and epilepsy among Montanans 18 years of age or older on the mid-year date of July 1, 1998. RESULTS: 2,849 cases were ascertained by at least one source, yielding an observed prevalence of 435.98 adults with developmental disabilities per 100,000 population. We selected a model under conditions of "partial three-way independence" that generated an estimate of 1,837.46 unobserved cases. When added to the observed cases, this yielded an estimated count of 4,686.46 total cases, and an ascertainment-corrected prevalence estimate of 717.17 per 100,000 population (95% C.I.: 4,679.08 to 4,694.11), representing a 64.5% increase above the observed prevalence. In conclusion, administratively observed counts might be insufficient indicators of the true number of adults with developmental disabilities in a rural state's population. Using a replicable, low-cost capture-recapture log-linear modeling procedure may generate more accurate estimates.
Learning Objectives: At the conclusion of the session, the participant will be able to: 1. Articulate the procedure for choosing and pursuing administrative data sets to link for the purpose of log-linear modeling. 2. Discuss the benefits and consequences of log-linear modeling used for estimating the population size of adults with developmental disabilities in a rural state.
Keywords: Public Health, Statistics
Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: None
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.