Is the TMPM-ICD9 revolution in trauma risk-adjustment compatible with imperfect administrative coding?

Area of Science:

• Trauma registry data analysis
• Injury severity scoring systems
• Health informatics

Background:

• TMPM-ICD9 is a new injury severity measure using ICD-9-CM codes, proposed as a replacement for AIS measures.
• Administrative ICD coding is often less complete than dedicated AIS coding, potentially impacting TMPM-ICD9 performance.
• This study investigates the effect of ICD coding completeness on TMPM-ICD9 performance in Italy.

Purpose of the Study:

• To compare the predictive performance of TMPM-ICD9 and the Injury Severity Score (ISS) using trauma registry data.
• To evaluate the impact of ICD-9-CM coding completeness on the performance of these injury severity measures.
• To ascertain if administrative coding quality affects TMPM-ICD9's suitability for clinical use.

Main Methods:

• Hierarchical logistic regression models were used to assess discrimination (c-statistics) and calibration (Akaike's criterion, calibration curves).
• Data from 3570 trauma patients (2007-2009) were analyzed.
• Coding completeness was assessed by the ratio of ICD-9-CM to AIS codes per patient, with models stratified by this ratio.

Main Results:

• TMPM-ICD9 models demonstrated worse performance compared to ISS, particularly in calibration and discrimination (p<0.05).
• ICD-9-CM coding was less complete than AIS coding in 68% of patients.
• TMPM-ICD9 model discrimination improved when ICD coding completeness was higher (ratio >1).

Conclusions:

• Sub-optimal ICD-9-CM coding quality significantly reduced the predictive performance of TMPM-ICD9 compared to previous studies.
• Imperfect administrative coding may hinder the widespread adoption of TMPM-ICD9.
• Despite performance limitations, the ready availability of administrative data may still justify the use of TMPM-ICD9 in settings with incomplete coding.