To determine the extent to which relying on only one source of data leads to incomplete assessment of pneumococcal polysaccharide vaccine (PPV) or mammography.DESIGN
An urban Midwestern academic medical center in 1998/99.PARTICIPANTS
Medicare beneficiaries aged 65 and older with at least one health encounter.MEASUREMENTS
Completion of PPV and mammography was assessed using local and Medicare records. The study compared sources of records and assessed association between services and demographics and comorbidity.RESULTS
Adding Medicare data to local data increased the computed 1-year PPV from 8.8% (264/3,002) to 15.0% and increased the 1998/99 mammography rate from 40% (343/847) to 67%. Local data sources missed 40% of PPV and 39% of mammography; Centers for Medicare and Medicaid Services sources missed 50% of PPV and 2% of mammography. The vaccinated were younger than the nonvaccinated (74 vs 76, P<.001). African Americans and those with more comorbidity were less likely to receive PPV over 8 years. Of 555 patients with a Medicare record of mammography, whites and those without Medicaid were significantly less likely to have a local record of mammography (P<.001).CONCLUSION
Neither administrative nor local clinical records provide a complete or accurate assessment of these quality indicators. Accurate assessment of quality indicators requires pooling data from multiple sources across a broad region.