Research objective: We sought to identify initial nontraumatic lower-extremity amputations (ILEAs) and compare rates of ILEAs with different coding algorithms and varying lengths of observation period prior (look-back) to the first observed amputation.
Study design: We used a retrospective design on merged Medicare claims and patient treatment files of the Veteran Healthcare Administration for fiscal years 1998 to 2000 of veterans with diabetes ages 18 years and older. Three different algorithms using least-inclusive to most-inclusive procedure codes and "look-back" periods of 12, 18, and 24 months were used to identify ILEAs.
Findings: Overall, 2997 (8.2 per 1000) veterans experienced an amputation in fiscal year 2000. Using 24 months of previous data, the rates of ILEA under the 3 different algorithms I, II, and III were 6.4/1000, 6.2/1000, and 6.0/1000 respectively. Using the most stringent algorithm (III), ILEA rates were 6.6/1000 with 12 months of prior data, 6.2/1000 with 18 months of prior data. Compared with any amputations in fiscal year 2000, 12-, 18-, and 24-month look-back period decreased the ILEA rates by 19%, 24%, and 27%, respectively.
Conclusion: These findings highlight the usefulness of administrative data in identifying ILEAs. All 3 algorithms performed comparably well in the detection of ILEAs. The extent of identification was most complete with 24 months prior data, with marginal gains in extending the prior observation period from 18 to 24 months. Policy level analysis should consider reporting ILEAs, in addition to total amputation rates when examining trends and disparities in amputations.