Quan Updated Charlson Comorbidity Index Calculator
Quan Updated Charlson Comorbidity Index Calculator
Quan and colleagues re-derived the Charlson weights in 2004 Calgary discharge data and published twelve of them in 2011. Five of the original nineteen conditions now score nothing. This page carries no bands, and says why.
Updated Charlson (Quan 2011 weights)
12 scored conditions, 0 to 24The same patient as the 1987 page: a previous myocardial infarction, peripheral vascular disease, a transient ischaemic attack, dementia (2) and mild liver disease (2)
Scoring
Maximum 24 with the hierarchy applied
- the five that went to zero
- myocardial infarction, peripheral vascular disease, cerebrovascular disease, peptic ulcer disease and uncomplicated diabetes each carried a weight of 1 in 1987 and carry nothing here, with adjusted hazard ratios of 0.99, 1.10, 1.10, 1.08 and 1.12. They are not items on this page at all, which is why it has twelve selects and the 1987 page has sixteen
- the merge
- leukaemia, lymphoma and solid tumour are one item here and three in the original. That alone changes a haematology patient’s total by 2, independently of any weight change
- 0 to 24
- the twelve weights sum to 28 flat, but the diabetes, liver and malignancy pairs are mutually exclusive, so 24 is the ceiling a patient can reach. Quan’s own reproduction in the literature prints 24
- derived on
- Calgary, Alberta discharge data from 2004, against mortality within one year of discharge; validated against IN-HOSPITAL mortality in six countries. Note that the derivation outcome and the validation outcome are different outcomes
- discrimination
- c-statistics of 0.825, 0.828, 0.878, 0.727, 0.831 and 0.869 in Australia, Canada, France, Japan, New Zealand and Switzerland, against 0.808, 0.825, 0.882, 0.723, 0.836 and 0.876 for the original weights. The updated weights were better in three countries and WORSE in three, by margins of 0.003 to 0.017. Re-derived weights are not uniformly an improvement
- which weight set
- Two weight sets are in current use and they give different totals for the same patient. State which one a number came from, every time. A published Charlson score with no weight set named cannot be compared with anything
Worked example
The same patient as the 1987 page: a previous myocardial infarction, peripheral vascular disease, a transient ischaemic attack, dementia (2) and mild liver disease (2)
Myocardial infarction, peripheral vascular disease and cerebrovascular disease all carry a weight of 0 on these weights and are not scored items on this page
Dementia 2 + mild liver disease 2 = 4 points
The same five diagnoses score 5 on the 1987 weights, which is the top published stratum, and 4 here, which is not in any published stratum because this weight set does not have any
Three of the patient's five conditions contribute nothing here and two contribute double. The total fell by one point for reasons that cancel almost exactly — which is why a small difference between two published scores tells you nothing about how different the underlying weights are
Add a lymphoma. The 1987 total becomes 5 + 2 = 7; this total becomes 4 + 2 = 6. Add a localised solid tumour as well and the original goes to 9 while this stays at 6, because the three malignancy items are one item here
The five conditions Quan’s re-derivation sent to zero
| Condition | 1987 weight | 2011 weight | 2011 hazard ratio |
|---|---|---|---|
| Myocardial infarction | 1 | 0 | 0.99 |
| Peripheral vascular disease | 1 | 0 | 1.10 |
| Cerebrovascular disease | 1 | 0 | 1.10 |
| Peptic ulcer disease | 1 | 0 | 1.08 |
| Diabetes without chronic complications | 1 | 0 | 1.12 |
Discrimination for in-hospital mortality, six countries
| Country | Data year | Updated weights | 1987 weights |
|---|---|---|---|
| Australia | 2008 | 0.825 | 0.808 |
| Canada | 2008 | 0.828 | 0.825 |
| France | 2004 | 0.878 | 0.882 |
| Japan | 2008 | 0.727 | 0.723 |
| New Zealand | 2008 | 0.831 | 0.836 |
| Switzerland | 2008 | 0.869 | 0.876 |
Re-derived weights, and why newer is not the same as better
Quan and colleagues took the Charlson index’s nineteen conditions, mapped them to ICD-9 and ICD-10 codes, and re-fitted the weights against mortality within one year of discharge in Calgary, Alberta hospital data from 2004. The result, published in 2011, has twelve conditions with a non-zero weight instead of nineteen. Five conditions no longer predicted death at all: myocardial infarction, peripheral vascular disease, cerebrovascular disease, peptic ulcer disease and uncomplicated diabetes, with adjusted hazard ratios of 0.99, 1.10, 1.10, 1.08 and 1.12. Three weights fell, four rose, five were unchanged, and leukaemia, lymphoma and solid tumour were merged into a single malignancy item.
The obvious reading is that these weights are the modern ones and should replace the originals. The paper’s own validation does not support that. Tested against in-hospital mortality in six countries, the updated weights discriminated better in three and worse in three, with differences in the c-statistic of between 0.003 and 0.017. Both weight sets are in current use, and what matters is that a published score names which one produced it. The worked example here scores 4; the same patient scores 5 on the original weights, which is a stratum boundary, and the two totals run on scales of different lengths — 0 to 24 against 0 to 33 — so even the ratio of two scores is not comparable between them.
This page carries no coloured bands, and the omission is deliberate. The 1987 paper published four strata with one-year mortality rates attached, and those appear on the original weights’ page because that is where they belong. Quan’s paper published discrimination — c-statistics by country — and no strata. Drawing a bar on this page with the 1987 cohort’s 12, 26, 52 and 85 per cent against it would take one instrument’s outcomes and print them on another instrument’s scale, which is the error these two pages exist to separate. Where a weight set has not published strata, the honest page has no bands.
One further caution specific to this weight set. It was derived from administrative discharge abstracts, so every condition definition is an ICD code list rather than a chart finding. Scoring these weights from a chart review is a hybrid that neither paper validated, and the renamed items make it easier than it looks to do by accident: seven of the seventeen conditions have a different name here than in the original.
A score is not a diagnosis and a cohort figure is not this patient’s outcome: a stratum in which 52 per cent died within a year describes that stratum, not which 52 per cent. This page reports what a published instrument scored and what that score predicted in a named cohort. It recommends no action. Every weight, cut-off and outcome figure here comes from a named derivation cohort, and cohorts differ in case mix, era, coding and outcome definition; where your own institution’s protocol or analysis plan differs, it takes precedence.
Frequently asked questions
Which Charlson conditions score nothing on Quan’s weights?
Five: myocardial infarction, peripheral vascular disease, cerebrovascular disease, peptic ulcer disease and diabetes without chronic complications. Each carried a weight of 1 in 1987 and each had an adjusted hazard ratio within about 10 per cent of 1 in Quan’s 2004 Calgary data. They are not items on this page at all.
Are Quan’s 2011 weights better than the 1987 weights?
Not uniformly, on the paper’s own evidence. Validated against in-hospital mortality in six countries, the updated weights gave a higher c-statistic in Australia, Canada and Japan and a lower one in France, New Zealand and Switzerland. The differences run from 0.003 to 0.017. The choice between them is a choice to state, not a correction to apply.
Why does this page have no risk bands?
Because Quan et al. published c-statistics and did not publish strata with outcome rates. The four mortality rates usually quoted with a Charlson score belong to the 1987 derivation cohort and to its 0-to-33 scale, and this score runs 0 to 24. Printing them here would be attaching one instrument’s outcomes to another instrument’s numbers.
What is the maximum updated Charlson score?
24. The twelve weights sum to 28 if every item is scored, but the diabetes, liver and malignancy pairs are mutually exclusive and the severe member replaces the mild one, so 24 is the ceiling a patient can reach.
Can I compare a Quan score with a 1987 Charlson score?
Only as two different measurements. Two weight sets are in current use and they give different totals for the same patient. State which one a number came from, every time. They run on scales of different lengths, they count malignancy differently, and seven of the seventeen conditions are named differently between them. A difference of one point between two published scores may mean nothing at all about the patients.
Related calculators
References
- Quan H, Li B, Couris CM, Fushimi K, Graham P, Hider P, Januel JM, Sundararajan V. Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol. 2011;173(6):676–82.
- Health Data Research UK Phenotype Library. Charlson comorbidity index (Quan 2011), phenotype PH1657. phenotypes.healthdatagateway.org (accessed 9 October 2026).
- Kita S, Shirai Y, Yoshida T, Shiraishi K, Nakamura A, Kawano M, Kinoshita Y, Noguchi T, Ito S. Comparison of various risk scores for the prognosis of hemorrhagic upper gastrointestinal mucosal disorder. Int J Emerg Med. 2020;13:41, Table 1.
- Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–83.
Not medical advice. For healthcare professionals and education. Reference intervals vary by laboratory and assay — always use your own laboratory's. Never base a dose or a treatment decision on this page alone. Full disclaimer at calcengines.com/disclaimer/