Claims-based Measures of Disability

How does one measure disability?  This is a difficult question.  Many health economists face an even more difficult question: how does one measured disability in claims?

A paper by Ben-Shalom and Stapleton attempts to answer this question using six definitions:

• Chronic Illness and Disability Payment System (CDPS). Developed by Rick Kronick  et al. 2000 (one of my dissertation committee members) and coauthors as system to support risk-adjusted Medicaid payments to HMOs.  It is similar to the Medicare Advantage HCC system but applied to Medicaid.
• Access Risk Classification System (ARCS) algorithm,. This algorithm classifies people with disabilities according to their ability to access routine care. ARCS relies on ICD-9 diagnosis codes, procedure codes (HCPCS and CPT codes) and the number of prescriptions.  It was developed by Palsbo et al. (2008).
• SSDI eligibility. Another method of identifying disability is based on eligibility decisions by the Social Security Administration (SSA) for Social Security Disability Insurance (SSDI).
• Psychiatric disorders: The patient was considered to have a psychiatric disorder if they have schizophrenia, affective disorder, or psychotic, neurotic, personality, and other nonpsychotic mental disorders. The following ICD-9 codes were used: 295–301, 306–7, 309–10, and 314–315.
• Cognitive disorders. A person was classified as having cognitive disorders if they have denmentia or Alzheimers, which are based on the following ICD-9 codes: 290.x, 290.3, 290.4, 290.8 and 331.0.
• Intellectual disorders. A person was classified as having an intellectual disorder if they have a mild (ICD-9 code: 317), moderate (318.0), severe (318.1), profound (318.2) or unspecified (319) intellectual disability.

They test these algorithms against patients self-reports of disability in the Medicare Current Beneficiary Survey data.  The authors conclude that:

The predictive performance of the regression-based models is better than that of the individual claims-based indicators. At a predicted probability threshold chosen to maximize the sum of sensitivity and specificity, sensitivity is 0.72 for beneficiaries age 65 or older and specificity is 0.65. For those under 65, sensitivity is 0.54 and specificity is 0.67. The findings also suggest ways to improve predictive performance for specific disability populations of interest to researchers.

Source:

• Ben-Shalom, Y. and Stapleton, D. C. (2015), Predicting Disability among Community-Dwelling Medicare Beneficiaries Using Claims-Based Indicators. Health Services Research. doi: 10.1111/1475-6773.12316