Written and maintained by CASRAI Editorial Board
Last updated
The Charlson Comorbidity Index (CCI) is the most widely used tool for quantifying a patient’s overall comorbidity burden as a single number for risk-adjustment purposes. Developed by Mary Charlson and colleagues (Charlson ME, Pompei P, Ales KL, MacKenzie CR, “A new method of classifying prognostic comorbidity in longitudinal studies: development and validation,” Journal of Chronic Diseases, 1987), it assigns weighted points to 19 categories of comorbid disease and, in its age-adjusted form, to a patient’s age — producing a single score that is inversely related to estimated survival. Unlike SOFA or qSOFA (which measure acute, in-the-moment organ dysfunction), CCI measures a patient’s chronic disease burden going into an episode of care, which is why it shows up constantly in risk-adjusted outcome reporting and secondary-data research rather than at the acute bedside.
The 19 weighted comorbidity categories
Each condition is scored once, based on whether it is present in the patient’s history, and the weights are summed into an unadjusted CCI:
| Points | Conditions |
|---|---|
| 1 | Myocardial infarction; congestive heart failure; peripheral vascular disease; cerebrovascular disease (CVA or TIA); dementia; chronic pulmonary disease; connective tissue disease; peptic ulcer disease; mild liver disease; diabetes without end-organ damage |
| 2 | Diabetes with end-organ damage; hemiplegia; moderate-to-severe chronic kidney disease; solid tumor (localized, non-metastatic); leukemia; lymphoma |
| 3 | Moderate-to-severe liver disease |
| 6 | Metastatic solid tumor; AIDS |
The weighting itself is the point of the instrument: a metastatic malignancy or AIDS diagnosis (6 points) is treated as roughly six times the mortality-relevant burden of a single 1-point condition like peripheral vascular disease, and the weights were derived empirically from one-year mortality data in the original 1987 validation cohort, not assigned by consensus opinion.
Age adjustment (the ACCI)
The same 1987 paper introduced an age-adjusted version, commonly called the ACCI, that adds points for age on top of the comorbidity total — treating age itself as an independent contributor to mortality risk, separate from diagnosed comorbid disease:
| Age | Points added |
|---|---|
| <50 | 0 |
| 50–59 | +1 |
| 60–69 | +2 |
| 70–79 | +3 |
| ≥80 | +4 |
Whether to report the unadjusted CCI or the ACCI depends on the study or program: age-adjusted scores are common where age itself is not otherwise controlled for in the analysis, while some risk-adjustment models that already include age as a separate covariate use the unadjusted comorbidity count instead to avoid double-counting age’s effect. Either way, the total score is inversely proportional to estimated survival — a higher score means a lower estimated survival probability — and published literature commonly groups scores into rough severity bands (0 = no comorbidity, low single digits = mild-to-moderate burden, higher scores = severe burden), though the exact cut points used vary by study and should be reported explicitly rather than assumed.
How CCI is actually calculated in practice
In a research or quality-reporting setting, CCI is rarely scored from a clinician reading a chart by hand — it’s almost always derived from coded administrative or claims data, which means the coding algorithm used to map ICD codes to the 19 Charlson categories matters as much as the weights themselves. Two coding adaptations dominate the literature: Deyo, Cherkin, and Ciol’s 1992 ICD-9-CM adaptation (Journal of Clinical Epidemiology) and Quan et al.’s 2005 update covering both ICD-9-CM and ICD-10 administrative data (Medical Care). These algorithms don’t always agree on which codes map to which category or how many diagnosis fields to search, and studies using different versions on the same underlying data can produce measurably different comorbidity scores for the same patients — a real source of cross-study and cross-site variability that a reader has to check for, not assume away, when comparing risk-adjusted results across papers or registries.
Its role in risk-adjusted outcome reporting and research
CCI’s core use case is as a covariate: comparing raw mortality, readmission, or complication rates across hospitals, surgeons, or treatment arms is misleading if one group’s patients simply carry more chronic disease going in, and CCI is the standard way to adjust for that before comparing outcomes. It appears throughout the same risk-adjustment methods covered in our Observed-to-Expected (O/E) Ratio and Risk Adjustment guide and in outcome models feeding public reporting programs like the ones covered in Hospital Readmissions Reduction Program, CMS Overall Hospital Star Rating Methodology, and AHRQ Patient Safety Indicators Explained. For research administrators and quality-registry teams building or validating a risk-adjustment model, that dependence on the coding algorithm above is the practical takeaway: which version of CCI, calculated from which coding adaptation, over what lookback window, needs to be stated explicitly in methods sections and registry documentation — not just “Charlson score” as a black-box covariate — if the resulting comparison is going to be reproducible or defensible against a different site’s numbers.
Frequently asked questions
What is a “high” Charlson Comorbidity Index score?
There’s no single universal cutoff — published studies use different severity bands, and a score’s clinical meaning depends on the population and outcome being studied. What’s consistent across the literature is the direction: higher scores are consistently associated with lower estimated survival and worse outcomes, and any specific threshold used should be stated explicitly by whoever is reporting it.
Is the Charlson Comorbidity Index the same as the Elixhauser Comorbidity Index?
No, though they serve a similar purpose. Elixhauser is a separate, later comorbidity-measurement system with a different (and larger) condition list and no single summary weighting scheme in its original form; the two are not interchangeable, and a study or registry should specify which one it uses.
Do I need lab values to calculate CCI?
No. Unlike SOFA, CCI is based on diagnosed comorbid conditions (from a chart review or, more commonly, coded administrative/claims data) rather than acute physiological measurements, which is exactly why it captures chronic disease burden rather than in-the-moment organ dysfunction.
Why do two studies report different Charlson scores for similar patients?
Most commonly because they used different ICD-to-Charlson-category coding algorithms (Deyo’s 1992 adaptation and Quan’s 2005 update are the two most common, and they don’t always agree), different lookback windows for identifying comorbidities, or different choices about age-adjustment — all of which should be reported explicitly in methods.
