Arthritis & Rheumatism, Volume 62,
November 2010 Abstract Supplement
Abstracts of the American College of
Rheumatology/Association of Rheumatology Health Professionals
Annual Scientific Meeting
Atlanta, Georgia November 6-11, 2010.
Validation of Diagnostic Codes for Subtrochanteric, Diaphyseal, and Typical Hip Fractures Using Administrative Claims Data.
Narongroeknawin, Pongthorn, M. Patkar, Nivedita, Shakoory, Bita, Jain, Archana, R. Curtis, Jeffrey, Delzell, Elizabeth, H. Lander, Philip
As part of developing administrative claims-based algorithms for identifying subtrochanteric and diaphyseal fractures of the femur (SFF and DFF), we assessed the accuracy of hospital and physician diagnosis codes for these fractures. Algorithms that combined hospital discharge with surgeon's diagnosis codes to identify cases of these fractures yielded relatively high positive predictive values (PPV) and will be useful in future population-based observational studies.
Candidates for study inclusion were all patients with an ICD-9 diagnosis code for femoral fracture, at a University hospital from 01/2004 to 12/2008. Individuals with hospital discharge diagnosis for open femoral fracture and distal end femoral fractures were excluded because these fractures are usually caused by major trauma and not associated with longer term bisphosphonates use.
Based on the clinical history, radiology report, operative note, and discharge summary accounting for the fracture admission, patients were assigned to one of the following classifications: (1) SFF, defined as fracture at the region between the lesser trochanter and a point 5 centimeters distal; (2) DFF, defined as fracture at the region between subtrochanteric and supracondylar areas; or (3) "typical" hip fracture, other than SFF and DFF. We classified suspected SFF and DFF cases according to several alternative algorithms based on ICD-9 diagnosis codes, with the goal of identifying the algorithm that maximized PPV.
We identified 137 persons with a suspected SFF and randomly selected 50 persons each with a suspected DFF or typical hip fracture. The PPV of case algorithms, which were varied based on the position and source of the diagnosis codes on medical claims, ranged from 6989% for SFF, from 8998% for DFF, and from 8598% for typical hip fracture.
In a subgroup analysis, the PPVs of the various algorithms for SFF and typical hip fracture among patients age > 55 years and among patients without major trauma ICD-9 codes were the same as those in the primary analysis. In contrast, the PPVs of algorithms for DFF were lower for these subgroups than those in primary analysis. However, the number of confirmed DFF cases in these subgroups was small.
Table 1. Characteristic of Validation Sample
Table 2. Positive Predictive Value (PPV) of Various Case Identification Algorithms for Subtrochanteric, Diaphyseal, and Typical Hip Fractures
Administrative claims data-based algorithms that combined hospital discharge with surgeon's diagnosis codes to identify cases of SFF, DFF, and typical hip fracture yielded high PPVs. These claims algorithms will be useful in future population-based observational studies to evaluate the association between osteoporosis medications and subtrochanteric and diaphyseal fractures.
To cite this abstract, please use the following information:
Narongroeknawin, Pongthorn, M. Patkar, Nivedita, Shakoory, Bita, Jain, Archana, R. Curtis, Jeffrey, Delzell, Elizabeth, et al; Validation of Diagnostic Codes for Subtrochanteric, Diaphyseal, and Typical Hip Fractures Using Administrative Claims Data. [abstract]. Arthritis Rheum 2010;62 Suppl 10 :1573