Charlson Comorbidity Index (CCI)

Purpose

The Charlson Comorbidity Index (CCI) assesses comorbidity level by taking into account both the number and severity of 19 pre-defined comorbid conditions. It provides a weighted score of a client’s comorbidities which can be used to predict short term and long-term outcomes such as function, hospital length of stay and mortality rates. The CCI is the most widely used scoring system for comorbities used by researchers and clinicians (Charlson, Pompei, Ales, & Mackenzie, 1987; Elixhauser, Steiner, Harris, & Coffey, 1998).

In-Depth Review

Purpose of the measure

The Charlson Comorbidity Index (CCI) assesses comorbidity level by taking into account both the number and severity of 19 pre-defined comorbid conditions. It provides a weighted score of a client’s comorbidities which can be used to predict short term and long-term outcomes such as function, hospital length of stay and mortality rates. The CCI is the most widely used scoring system for comorbities used by researchers and clinicians (Charlson, Pompei, Ales, & Mackenzie, 1987; Elixhauser, Steiner, Harris, & Coffey, 1998).

Available versions

The CCI was published by Charlson, Pompei, Ales, and Mackenzie in 1987.

Features of the measure

Items:
The CCI is comprised of 19 comorbid conditions: myocardial infarct, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, connective tissue disease, ulcer disease, mild liver disease, diabetes, hemiplegiaComplete paralysis of the arm, leg, and trunk on one side of the body that results from damage to the parts of the brain that control muscle movements. Hemiplegia is not a progressive condition, nor is it a disease., moderate or several renal disease, diabetes with end organ damage, any tumor, leukemia, lymphoma, moderate or severe liver disease, metastatic solid tumor, AIDS. Each disease is given a different weight based on the strength of its association with 1-year mortality as follows (Charlson et al., 1987):

The CCI can be completed from medical records, administrative databases, or interview-based questionnaires (Bravo, Dubois, Hebert, De Wals, & Messier, 2002).

Scoring:
The total score in the CCI is derived by summing the assigned weights of all comorbid conditions presented by the client. Higher scores indicate a more severe condition and consequently, a worse prognosis (Charlson, Szatrowski, Peterson, & Gold, 1994).

Time:
Not reported

Subscales:
None

Equipment:
Not applicable.

Training:
No specific training is available.

Alternative forms of the CCI

The CCI has a weighted age version, two adaptations to be used with ICD-9 databases, and one version to be used with clients with amputations (Charlson et al., 1994; Deyo, Cherkin, & Ciol, 1992; Melchiore, Findley, & Boda, 1996; Romano, Roos, & Jollis, 1993).

Client suitability

Can be used with:

• Clients with strokeAlso called a “brain attack” and happens when brain cells die because of inadequate blood flow. 20% of cases are a hemorrhage in the brain caused by a rupture or leakage from a blood vessel. 80% of cases are also know as a “schemic stroke”, or the formation of a blood clot in a vessel supplying blood to the brain..
• The CCI is a general scoring system allowing for its use with a variety of populations (Groot, Beckerman, Lankhorst, & Bouter, 2003).

Should not be used in:

• To date, there is no information on restrictions for using the CCI.

In what languages is the measure available?

Not applicable

Summary

Psychometric Properties

Overview

We conducted a literature search to identify all relevant publications on the psychometric properties of the Charlson Comorbidity Index (CCI) in individuals with strokeAlso called a "brain attack" and happens when brain cells die because of inadequate blood flow. 20% of cases are a hemorrhage in the brain caused by a rupture or leakage from a blood vessel. 80% of cases are also know as a "schemic stroke", or the formation of a blood clot in a vessel supplying blood to the brain.. We identified 6 studies.

Reliability

Test-retest:
Katz, Chang, Sangha, Fossel, and Bates (1996) evaluated the test-retest reliabilityA way of estimating the reliability of a scale in which individuals are administered the same scale on two different occasions and then the two scores are assessed for consistency. This method of evaluating reliability is appropriate only if the phenomenon that the scale measures is known to be stable over the interval between assessments. If the phenomenon being measured fluctuates substantially over time, then the test-retest paradigm may significantly underestimate reliability. In using test-retest reliability, the investigator needs to take into account the possibility of practice effects, which can artificially inflate the estimate of reliability (National Multiple Sclerosis Society).
of the questionnaire format of the CCI in 25 inpatients with different diagnoses including strokeAlso called a “brain attack” and happens when brain cells die because of inadequate blood flow. 20% of cases are a hemorrhage in the brain caused by a rupture or leakage from a blood vessel. 80% of cases are also know as a “schemic stroke”, or the formation of a blood clot in a vessel supplying blood to the brain.. Participants were evaluated by the same rater twice within 24 hours. Test-retest reliabilityA way of estimating the reliability of a scale in which individuals are administered the same scale on two different occasions and then the two scores are assessed for consistency. This method of evaluating reliability is appropriate only if the phenomenon that the scale measures is known to be stable over the interval between assessments. If the phenomenon being measured fluctuates substantially over time, then the test-retest paradigm may significantly underestimate reliability. In using test-retest reliability, the investigator needs to take into account the possibility of practice effects, which can artificially inflate the estimate of reliability (National Multiple Sclerosis Society).
was excellent as calculated using Intraclass CorrelationThe extent to which two or more variables are associated with one another. A correlation can be positive (as one variable increases, the other also increases – for example height and weight typically represent a positive correlation) or negative (as one variable increases, the other decreases – for example as the cost of gasoline goes higher, the number of miles driven decreases. There are a wide variety of methods for measuring correlation including: intraclass correlation coefficients (ICC), the Pearson product-moment correlation coefficient, and the Spearman rank-order correlation.
Coefficient (ICC = 0.92) and Spearman’s Rank CorrelationThe extent to which two or more variables are associated with one another. A correlation can be positive (as one variable increases, the other also increases – for example height and weight typically represent a positive correlation) or negative (as one variable increases, the other decreases – for example as the cost of gasoline goes higher, the number of miles driven decreases. There are a wide variety of methods for measuring correlation including: intraclass correlation coefficients (ICC), the Pearson product-moment correlation coefficient, and the Spearman rank-order correlation.
(rho = 0.94).

Inter-rater:
Liu, Domen and Chino (1997) assessed the inter-rater reliability of the CCI in 10 clients with strokeAlso called a "brain attack" and happens when brain cells die because of inadequate blood flow. 20% of cases are a hemorrhage in the brain caused by a rupture or leakage from a blood vessel. 80% of cases are also know as a "schemic stroke", or the formation of a blood clot in a vessel supplying blood to the brain.. The CCI was administered by two examiners blinded to each other’s scores. Inter-rater reliabilityA method of measuring reliability . Inter-rater reliability determines the extent to which two or more raters obtain the same result when using the same instrument to measure a concept.
, as calculated using Intraclass CorrelationThe extent to which two or more variables are associated with one another. A correlation can be positive (as one variable increases, the other also increases - for example height and weight typically represent a positive correlation) or negative (as one variable increases, the other decreases - for example as the cost of gasoline goes higher, the number of miles driven decreases. There are a wide variety of methods for measuring correlation including: intraclass correlation coefficients (ICC), the Pearson product-moment correlation coefficient, and the Spearman rank-order correlation.
Coefficient, was adequate (ICC = 0.67).

Validity

Content:
Charlson et al. (1987) identified the comorbid conditions of 559 inpatients with breast cancer. They then calculated the relationship of potential prognostically important variables to survival using Cox’s regression analysis. Finally, the adjusted relative risk was estimated to each comorbid condition.

Criterion:
Concurrent:
No gold standardA measurement that is widely accepted as being the best available to measure a construct.
exists against which to compare the CCI.

Predictive:
Liu et al. (1997) estimated the ability of the CCI at hospital admission to predict length of stay and the Functional Independence Measure (FIM) score (Keith, Granger, Hamilton, & Sherwin, 1987) at discharge. Predictive validityA form of criterion validity that examines a measure’s ability to predict some subsequent event. Example: can the Berg Balance Scale predict falls over the following 6 weeks? The criterion standard in this example would be whether the patient fell over the next 6 weeks.
was calculated in 106 clients with Spearman’s Rank CorrelationThe extent to which two or more variables are associated with one another. A correlation can be positive (as one variable increases, the other also increases – for example height and weight typically represent a positive correlation) or negative (as one variable increases, the other decreases – for example as the cost of gasoline goes higher, the number of miles driven decreases. There are a wide variety of methods for measuring correlation including: intraclass correlation coefficients (ICC), the Pearson product-moment correlation coefficient, and the Spearman rank-order correlation.
. Correlation between the CCI and the FIM was poor(rho = -0.19) as was the correlationThe extent to which two or more variables are associated with one another. A correlation can be positive (as one variable increases, the other also increases – for example height and weight typically represent a positive correlation) or negative (as one variable increases, the other decreases – for example as the cost of gasoline goes higher, the number of miles driven decreases. There are a wide variety of methods for measuring correlation including: intraclass correlation coefficients (ICC), the Pearson product-moment correlation coefficient, and the Spearman rank-order correlation.
between the CCI and length of stay (rho = 0.16). These results suggest that the CCI measured at hospital admission may not be predictive of length of stay or the FIM at discharge.

Goldstein, Samsa, Matchar, and Horner (2004) examined in 960 clients with acute stroke whether the CCI measured at admission was able to predict the modified Rankin Scale (mRS) (Rankin, 1957) at hospital discharge, and, 1-year mortality rates. Predictive validityA form of criterion validity that examines a measure’s ability to predict some subsequent event. Example: can the Berg Balance Scale predict falls over the following 6 weeks? The criterion standard in this example would be whether the patient fell over the next 6 weeks.
was analyzed using logistic regression. The CCI was dichotomized into low comorbidity (scores <2) and high comorbidity (scores <2) and the mRS into good outcomes (scores <2) and poor outcomes (scores ≥2). Higher CCI scores were associated with a 36% increased odds of having poor outcomes on the modified Rankin Scale and 72% greater odds of death at 1 year post-stroke.

Fischer, Arnold, Nedeltchev, Schoenenberger, Kappeler, Hollinger et al. (2006) verified in 259 clients the ability of the CCI, as measured at admission to a stroke unitStroke units are designed to provide multidisciplinary specialized care for patients who have had a stroke. In the best units, the team consists of nurses, pharmacists, social workers, medical staff, and occupational, physical and speech therapists. Stroke units can be located in a special unit in a defined location, or can used as a roving stroke specialist team. (Hill, M. Stroke Units in Canada. CMAJ. 2002:167:649-50.), to predict poor outcomes on the modified Rankin Scale (mRS – Rankin, 1957) at 4 months after hospital discharge. The mRS was dichotomized into good outcomes (scores ≤ 2) and poor outcomes (scores >2). Logistic regression analyses revealed that the CCI was not able to predict poor outcomes on the mRS. In this study, the predictors of the mRS score at 4 months post-stroke were strokeAlso called a “brain attack” and happens when brain cells die because of inadequate blood flow. 20% of cases are a hemorrhage in the brain caused by a rupture or leakage from a blood vessel. 80% of cases are also know as a “schemic stroke”, or the formation of a blood clot in a vessel supplying blood to the brain. severity, atrial fibrilation, coronary artery disease and diabetes.

Tessier, Finch, Daskalopoulou, and Mayo (2008) examined, in 672 participants, the ability of the CCI, the Functional Comorbidity Index (Groll, Bombardier, & Wright, 2005), and a stroke-specific comorbidity index (developed by the same authors) to predict function 3 months post-stroke. Predictive validityA form of criterion validity that examines a measure’s ability to predict some subsequent event. Example: can the Berg Balance Scale predict falls over the following 6 weeks? The criterion standard in this example would be whether the patient fell over the next 6 weeks.
was calculated by use of c-statistics to calculate the area under the Receiver Operating Characteristic (ROC) curve. The ability of the CCI (AUC = 0.76), the Functional Comorbidity Index (AUC = 0.71) and the stroke-specific comorbidity index (AUC = 0.71) to predict function at 3 months post-stroke were all adequate. These results suggest that the percentage of patients correctly classified according to their function at 3 months post-stroke is slightly higher when using the CCI over these other comorbidity measures.

Construct:
Convergent/Discriminant:

Katz et al. (1996) tested the convergent validityA type of validity that is determined by hypothesizing and examining the overlap between two or more tests that presumably measure the same construct. In other words, convergent validity is used to evaluate the degree to which two or more measures that theoretically should be related to each other are, in fact, observed to be related to each other.
of the CCI by comparing it to self-reported number of prescription medications consumed, number of hospitalizations, length of stay and total financial costs in 170 hospital inpatients, including those with strokeAlso called a “brain attack” and happens when brain cells die because of inadequate blood flow. 20% of cases are a hemorrhage in the brain caused by a rupture or leakage from a blood vessel. 80% of cases are also know as a “schemic stroke”, or the formation of a blood clot in a vessel supplying blood to the brain.. Correlations, as calculated using Spearman’s Rank CorrelationThe extent to which two or more variables are associated with one another. A correlation can be positive (as one variable increases, the other also increases – for example height and weight typically represent a positive correlation) or negative (as one variable increases, the other decreases – for example as the cost of gasoline goes higher, the number of miles driven decreases. There are a wide variety of methods for measuring correlation including: intraclass correlation coefficients (ICC), the Pearson product-moment correlation coefficient, and the Spearman rank-order correlation.
, were all poor between the CCI and self-reported number of prescription medications (rho = 0.06), number of hospitalizations (rho = 0.22), length of stay (rho = 0.20) and total costs (rho = 0.26).

Liu et al. (1997) measured the convergent validityA type of validity that is determined by hypothesizing and examining the overlap between two or more tests that presumably measure the same construct. In other words, convergent validity is used to evaluate the degree to which two or more measures that theoretically should be related to each other are, in fact, observed to be related to each other.
of the CCI in 106 clients with strokeAlso called a “brain attack” and happens when brain cells die because of inadequate blood flow. 20% of cases are a hemorrhage in the brain caused by a rupture or leakage from a blood vessel. 80% of cases are also know as a “schemic stroke”, or the formation of a blood clot in a vessel supplying blood to the brain., by comparing it to the number of medication consumed, laboratory studies, therapeutic interventions, number of consultations during hospital’s stay, and days of interruption of participationAs defined by the International Classification of Functioning, Disability and Health, participation is an individual’s involvement in life situations in relation to health conditions, body functions or structures, activities, and contextual factors. Participation restrictions are problems an individual may have in the manner or extent of involvement in life situations. in rehabilitation due to complications. Adequate correlations were found between the CCI and the total number of medications consumed (rho = 0.48) and poor correlations were found between the CCI and laboratory studies (rho = 0.28), therapeutic interventions (rho = 0.19), consultations (rho = 0.25), and days of interruption of rehabilitation participationAs defined by the International Classification of Functioning, Disability and Health, participation is an individual’s involvement in life situations in relation to health conditions, body functions or structures, activities, and contextual factors. Participation restrictions are problems an individual may have in the manner or extent of involvement in life situations. (rho = 0.22).

Tessier et al. (2008) analyzed the convergent validityA type of validity that is determined by hypothesizing and examining the overlap between two or more tests that presumably measure the same construct. In other words, convergent validity is used to evaluate the degree to which two or more measures that theoretically should be related to each other are, in fact, observed to be related to each other.
of the CCI by comparing it to the Functional Comorbidity Index (Groll et al., 2005) in 437 clients with Correlations were found to be excellent (rho = 0.62).

Known groups:
No studies have examined known groups validityKnown groups validity is a form of construct validation in which the validity is determined by the degree to which an instrument can demonstate different scores for groups know to vary on the variables being measured.
of the CCI.

Responsiveness

No studies have examined the responsivenessThe ability of an instrument to detect clinically important change over time.
of the CCI.

References

• Bravo, G., Dubois, M.F., Hebert, R., De Wals, P., & Messier, L. (2002). A perspective evaluation of the Charlson Comorbidity Index for use in long-term care patients. JAGS, 50, 740-745.
• Charlson, M., Pompei, P., Ales, M.L., & Mackenzie C.R. (1987). A new method of classifying comorbidity in longitudinal studies: Development and validation. J Chronic Dis, 40, 373-393.
• Charlson, M., Szatrowski, T.P., Peterson, J., & Gold, J. (1994). Validationof a Combined Comorbidity Index. Journal of Clinical Epidemiology, 47(11), 1245-1251.
• De Groot, V., Beckerman, H., Lankhorst, G.J., & Bouter, L.M. (2003). How to measure comorbidity: A critical review of available methods. Journal of Clinical Epidemiology, 56, 221-229.
• Deyo, R.A., Cherkin, D.C., & Ciol, M.A. (1992). Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. Journal Clinical Epidemiology, 45, 613-619.
• Elixhauser, A., Steiner, C., Harris, D.R., & Coffey, R.M. (1998).Comorbidity measures for use with administrative data. Medical Care, 36(1), 8-27.
• Fischer, U., Nedeltchev, K., Schoenenberger, R.A., Kappeler, L., Hollinger, P.,Schroth, G. et al. (2006). Impact of comorbidity on ischemic stroke outcome. Acta Neurol Scand, 113, 108-113.
• Goldstein, L.B., Samsa, G.P., Matchar, D.B., & Horner, R.D. (2004). Charlson Index Comorbidity Adjustment for Ischemic Stroke Outcome Studies. Stroke, 35, 1941-1945.
• Groll, D., Bombardier, C., & Wright, J. (2005). The development of a comorbidity index with physical function as the outcome. Journal of Clinical Epidemiology, 58, 595-602.
• Hall, W. H., Ramachandran, R., Narayan, S., Jani, A. B., & Vijayakumar, S. (2004). An electronic application for rapidly calculating Charlson comorbidity score. BMC Cancer, 4, 94.
• Katz, J., Chang, L., Sangha, O., Fossel, A., & Bates, D. (1996). Can comorbidity be measured by questionnaire rather than medical record review? Medical Care, 34(1), 73-84.
• Keith, R.A., Granger, C.V., Hamilton, B.B., & Sherwin, F.S. (1987). The functional independence measure: A new tool for rehabilitation. Adv Clin Rehabil, 1, 6-18.
• Liu, M., Domen, K., & Chino, N. (1997). Comorbidity measures for stroke outcome research: A preliminary study. Arch Phys Rehabil, 78, 166-172.
• Melchiore, P.J., Findley, T., Boda, W. (1996). Functional outcome and comorbidity indexes in the rehabilitation of the traumatic versus the vascular unilateral lower limb amputee. Am J Phys Med Rehabil, 75, 9-14.
• Rankin, J. (1957). Cerebral vascular accidents in patients over the age of 60. Scott Med J, 2, 200-215.
• Romano, P.S., Roos, L.L., & Jollis, J.G. (1993). Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. Journal of clinical epidemiology, 46 (10) 1075-1079.
• Tessier, A., Finch. L., Daskalopoulou, S.S., Mayo, N.E. (2008). Validation of the Charlson Comorbidity Index for Predicting Functional Outcome of Stroke. Arch Phys Med Rehabil, 89, 1276-1283.

See the measure

How to obtain the CCI

An electronic application for rapidly calculating Charlson Comorbidity Index score

The following link will allow you to download an Excel Spread sheet calculator for Charlson Comorbidity Index: Excel calculator Charlson Index