Possible inaccuracy in Charlson comorbidity index calculation

[tchang1997]

Prerequisites

• Put an X between the brackets on this line if you have done all of the following:
  • Checked the online documentation: https://mimic.mit.edu/
  • Checked that your issue isn't already addressed: https://github.com/MIT-LCP/mimic-code/issues?utf8=%E2%9C%93&q=

Description

I noticed that the SQL scripts for Charlson comorbidity index (CCI) calculation appear to be calculating the age score incorrectly. The implementation in the repo is

SELECT
     ...age
  , CASE WHEN age <= 40 THEN 0
    WHEN age <= 50 THEN 1
    WHEN age <= 60 THEN 2
    WHEN age <= 70 THEN 3
    ELSE 4 END AS age_score

whereas, based on the original paper on age-adjusted CCI by Charlson, it appears that the correct mapping from age to score would be

SELECT 
     ...age
  , CASE WHEN age <= 50 THEN 0
    WHEN age <= 60 THEN 1
    WHEN age <= 70 THEN 2
    WHEN age <= 80 THEN 3
    ELSE 4 END AS age_score

as prescribed in the paper: "Each decade of age over 40 adds 1 point to risk (e.g. 50-59 years, 1 point; 60-69 years, 2 points; 70-79 years, 3 points) and these points for age are added to the score from the Charlson comorbidity index (e.g., 0, 1,2,3, etc.)" This is consistent with other implementations of CCI, e.g. MDcalc.

This issue affects the majority of admissions in MIMIC. I reran the updated query on my copy, and this update changed the CCI in 293409 of 454324 (64.6%) of admissions — the CCI in these admissions was previously overestimated by 1.

If needed, I'm happy to open a PR to patch this.

[alistairewj]

Good catch! Would love a PR. Slowly making my way through the backlog, things that make life easy are appreciated :)

[alistairewj]

Closed by #1463