Study design: A review of a cohort of 310 consecutive patients who underwent anterior spinal fusion was performed to evaluate the accuracy of hospital ICD-9-CM complication coding.
Objectives: To better understand the clinical significance of conclusions suggested by studies that rely on electronic administrative databases for their data source.
Summary of background data: Despite their availability, there have been no studies to date that have evaluated the accuracy of ICD-9-CM administrative databases as they relate to the actual clinical experience in spinal procedures.
Methods: A physician and a research technician independently reviewed the primary medical records for the occurrence of complications. This data was compared with the hospital-acquired ICD-9-CM coded complications.
Results: The physician reviewer identified 152 complications in 119 patients, with 32 different types of complications. The research abstracter identified 175 complications in 130 patients, with 34 different types of complications identified. Hospital ICD-9-CM coding identified 105 complications in 80 patients, including only 11 different ICD-9-CM codes. Overall, 27% of ICD-9-CM complication codes were listed as "unspecified or unclassified complications, reactions, or misadventures," and contained no meaningful clinical information. Cardiac and pulmonary complications were over-estimated and wound infections and genitourinary and gastrointestinal complications were underestimated by ICD-9-CM coding.
Conclusions: Studies of complications of spinal procedures using data derived from hospital ICD-9-CM complication codes may be intrinsically flawed because the data available to researchers from these electronic databases may be inaccurate.