Abstract
Purpose: :
With the expanding use of health care claims data to study patients with ocular diseases, it is important to know whether the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic billing codes that providers submit for reimbursement during a patient encounters accurately captures the ocular conditions documented in patient’s medical records.
Methods: :
We searched the computerized billing records at the University of Michigan Kellogg Eye Center (KEC) and the Duke University Eye Center (DUEC) for 150 consecutive patients at each site who received billing codes for each of the following ocular conditions: cataracts (ICD-9-CM code 366.1x), non-exudative macular degeneration (362.51), glaucoma suspect (365.0), primary open-angle glaucoma (365.11), diabetes with no retinopathy (250.01), and proliferative diabetic retinopathy (362.02). Next, we reviewed the actual medical records to determine whether there was evidence documenting that each patient had the condition corresponding to the billing code submitted. For each condition, we determined the proportion of patients who were coded properly and improperly. Analyses were conducted to determine whether coding accuracy varied among the two institutions and whether it varied based on the type of eye care provider evaluating the patient (optometrist, general ophthalmologist, subspecialist).
Results: :
Overall, among the 1,888 billing records reviewed, 1689 (89.46%) accurately captured the ocular condition listed in the medical records. 199 (10.54%) discrepancies were noted between the billing and medical records. The proportion of accurately coded records was not significantly different amongst the two institutions, 91.02% versus 86.18%. (p=0.0028) The proportion of accurately coded records was highest for cataract (100%) and was ≥90% for macular degeneration, suspected glaucoma, open-angle glaucoma, and proliferative diabetic retinopathy. The only ocular condition that demonstrated poor correspondence between medical and billing records was for diabetes mellitus without retinopathy.
Conclusions: :
For the majority of common ocular diseases, billing records accurately capture the types of ocular conditions listed in the actual medical records. Given that health care claims data accurately identifies patients with common ocular diseases, researchers can feel confident using this data source to study the epidemiology, utilization patterns, and outcomes of patients with common ocular diseases.
Keywords: clinical (human) or epidemiologic studies: health care delivery/economics/manpower • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • clinical research methodology