OSTEOPOROSIS RECOGNITION: CORRECTING GEHLBACH ET AL.

Gehlbach et al.1 analyzed 1993–1997 data from the National Ambulatory Medical Care Survey (NAMCS) to determine the degree to which primary care physicians recognized osteoporosis. The authors’ analysis suggested that primary care physicians were underdiagnosing and, in consequence, undertreating this condition. Specifically, Gehlbach et al. reported that fewer than 2% of elderly White women were correctly diagnosed, while the estimated prevalence in this age group was 29%. (The report cited for this prevalence estimate2 excluded women who had ever received hormone therapy, and thus does not provide a population-based estimate.)

Gehlbach et al. based their conclusions on the diagnoses associated with the patient’s current office visit. They did not examine a condition-specific checklist included on the NAMCS survey form. In 1993 and 1994, physicians were asked whether patients had any of 5 conditions in addition to the presenting complaint: asthma, diabetes, HIV, obesity, or osteoporosis. We reassessed physicians’ recognition of osteoporosis using this check-off item.

Like Gehlbach et al., we obtained NAMCS data sets and limited the analysis to White women aged 60 years and older visiting primary care physicians: family physicians, general practitioners, internists, obstetricians, and gynecologists. We also included geriatricians, which were not explicitly mentioned by Gehlbach et al. but, based on raw visit counts, were included in their analysis. In addition, we corrected a methodological flaw in the analysis of Gehlbach et al. by using weighting procedures appropriate to the NAMCS’s stratified design.

In 1993, 13.4% of visits involved women with diagnosed osteoporosis. Age-specific prevalence was 7.3% among women aged 60 through 69 years, 13.8% among women aged 70 through 79 years, and 25.4% among women aged 80 years and older. In 1994, 9.5% of visits involved women with osteoporosis, with a prevalence of 5.8% among women aged 60 through 69 years, 9.05% among women aged 70 through 79 years, and 18.1% among women aged 80 years and older.

Gehlbach and colleagues’ characterization of “low rates of recognition and treatment”1 by primary care physicians is not supported by a correct analysis of NAMCS data. Rather, physicians’ recognition of osteoporosis paralleled what one might expect to find in a visit-based group of women.3,4 Although there was room for improvement, the data do not suggest that physicians were failing to recognize osteoporosis in 1993 and 1994. There were no clinical guidelines in place at that time recommending routine or universal screening for osteoporosis.

The presence of errors in a report by established investigators reminds us all to be cognizant of definitions and restrictions when analyzing secondary data. Review of data fields, appropriate weighting to reflect the sampling design, and acknowledgement of statistically unreliable estimates are essential. Such care is particularly important for sponsored research.