Identifying physician-recognized depression from administrative data: consequences for quality measurement

Area of Science:

• Health Services Research
• Medical Informatics
• Psychiatry

Background:

• Administrative data is crucial for healthcare quality measurement, including systems like the Health Plan Employer Data and Information Set (HEDIS).
• Accurate identification of patients with depression from administrative data is limited by several factors.

Purpose of the Study:

• To investigate and compare two algorithms for identifying physician-recognized depression using administrative data.
• To assess the performance of these algorithms in terms of sensitivity, specificity, and positive predictive value.

Main Methods:

• Two algorithms were developed to identify depression: Algorithm 1 required two criteria (depression diagnosis or antidepressant claim), while Algorithm 2 added a mandatory depression diagnosis.
• The study utilized data from a large managed care organization, analyzing member panels enrolled throughout 1997.
• Medical records were reviewed for a subset of patients, and large patient cohorts were identified based on each algorithm.

Main Results:

• Algorithm 1 demonstrated higher sensitivity (95%) and positive predictive value (49%), while Algorithm 2 showed higher specificity (88%) and a better positive predictive value (60%).
• Algorithm 2 identified patients with higher rates of follow-up visits and appropriate antidepressant dosing compared to Algorithm 1.
• Both algorithms exhibited significant false positive rates, impacting the reliability of depression identification.

Conclusions:

• The construction of the denominator significantly influences quality metrics derived from administrative data.
• High false positive rates in both algorithms raise concerns regarding the interpretation of depression quality reports based on administrative data.
• Further research is needed to refine methods for accurate depression identification in administrative datasets.