To The Editor:
"Treatment of Osteoporosis: Are Physicians Missing an
Opportunity?" (82-A: 1063-1070, Aug. 2000), by Freedman
et al., is an important study, and I agree with the authors’ conclusions.
However, the authors use claims data and state that some patients
who had a distal radial fracture might have been excluded from their
cohort. While this limitation is acknowledged, they do not mention what
might be a more important limitation regarding their design.
They did not mention the limitation regarding inclusion of nonradial
fractures. Nonfractures coded as "rule out radial fractures" are
still (often) coded as distal radial fractures on the X-Ray/ER/Office
claim. These patients are also (often) given splints at the time
of the claim-coding visit and therefore would be captured in the
authors’ cohort.
While patients coded as having "arm pain" or
a "broken arm" who indeed had a distal radial
fracture would be excluded from their cohort, there is little that
can be done to capture those patients short of sweeping all the
vague diagnostic codes (and there would be tens of thousands) and
chart-sampling. The patients included in the authors’ cohort,
however, represent a much smaller number (around 1000), and a representative
chart sample would have addressed the limitation mentioned above.
Don A. Saroff, MDGallatin Medical Foundation
10720 South Paramount
Downey, CA 90241
K.B. Freedman, F.S. Kaplan, W.B. Bilker, B.L. Strom,
and R.A. Lowe reply:
We would like to thank Dr. Saroff for his kind remarks and astute
observations regarding our study. He addresses the possibility that
patients with non-distal radial fractures were included in our cohort.
This may have occurred if patients were coded from the emergency
department as "rule out distal radial fractures" but
did not actually have a fracture. He also suggests that we confirm
the diagnoses in our database with the use of medical records.
Previous studies have shown a high degree of accuracy for fracture
diagnoses made with database files when one combines the use of
both ICD-9-CM and CPT codes1.
When comparing information from a Medicare database with actual
medical records, Ray et al.1 found
that the combination of ICD-9-CM diagnostic codes and CPT procedural
codes had a positive predictive value of 96% for the diagnosis
of an acute fracture of the radius and ulna. Although confirmation
with medical records in our study would have been ideal, medical
records are not available from the database provider that we used
(Protocare Sciences, Inc.) due to issues of patient confidentiality.
Of the 1162 fracture patients in our study, 966 were identified
by an ICD-9-CM code specific for distal radial fracture and one
of the confirming CPT codes specific for the treatment of distal radial
fractures (25600, 25605, 25611, 25620). There were an additional
196 patients who were identified by an ICD-9-CM code for distal
radial fracture and a CPT code for an upper extremity cast or splint.
For the sake of Dr. Saroff’s argument, assume that all
of these patients were miscoded and did not truly have a distal
radial fracture. These patients can then be excluded from the fracture
cohort. Even with this extreme assumption, only 29% (279)
of 966 women with a distal radial fracture were diagnosed or treated
for osteoporosis following the fracture. Therefore, despite this
potential limitation, the conclusions of our study remain the same.
—Kevin B. Freedman, MD, MSCE
Frederick S. Kaplan, MD
Warren B. Bilker, PhD
Brian L. Strom, MD, MPH
Robert A. Lowe, MD, MPHCorresponding author: Kevin B.
Freedman, MD, MSCE
Orthopaedic Associates of Allentown
Suite 2500
1243 South Cedar Crest Boulevard
Allentown, PA 18103