Charlson Comorbidity Index Adjustment in Intracerebral Hemorrhage
Background and Purpose—Previous studies of intracerebral hemorrhage (ICH) outcome prediction models have not systematically included adjustment for comorbid conditions. The purpose of this study was to assess whether the Charlson Comorbidity Index (CCI) was associated with early mortality and long-term functional outcome in patients with intracerebral hemorrhage.
Methods—We performed a retrospective analysis on a prospective observational cohort of patients with ICH admitted to 2 University of California San Francisco hospitals from June 1, 2001 to May 31, 2004. Components of the ICH score and use of early care limitations were recorded. Outcome was assessed using the modified Rankin Scale to 12 months. The CCI was derived using hospital discharge International Classification of Diseases, revision 9 codes and patient history obtained from standardized case report forms.
Results—In this cohort of 243 ICH patients, comorbid conditions were common, with CCI scores ranging from 0 to 12. Only 29% of patients with high CCI scores (≥3) achieved a 12-month modified Rankin Scale score of ≥3 compared with 48% of patients with CCI scores of 0 (P=0.02). CCI score was independently predictive of 12-month functional outcome, with higher CCI having a greater impact (CCI=2: odds ratio, 2.3; P=0.06; CCI=≥3: odds ratio, 3.5; P=0.001).
Conclusions—Comorbid medical conditions as measured by the CCI independently influence outcome after ICH. Future ICH outcome studies should account for the impact of comorbidities on patient outcome.
Medical comorbidities are common in patients with intracerebral hemorrhage (ICH).1 Although it seems intuitive that comorbid conditions influence outcome after ICH, studies assessing specific predictors of ICH outcome generally have not evaluated the independent impact of these comorbidities and none has used a systematic approach to comorbidity adjustment.2 The Charlson Comorbidity Index (CCI) accounts for multiple comorbidities by creating a sum score weighted according to the presence of various conditions. The CCI has been validated in various diseases in longitudinal studies including ischemic stroke.3 Although some studies have included patients with ICH,4 none has focused on the impact of the CCI on ICH outcome. The purpose of this study was to assess whether the CCI is associated with long-term functional outcome and 30-day mortality independent of factors known to influence ICH outcome.
Subjects and Methods
Study Design and Subjects
We performed a retrospective analysis on a prospective observational cohort of patients with nontraumatic ICH who presented to 2 University of California San Francisco hospital emergency departments from June 1, 2001 through May 31, 2004. The details of the study design have been published previously.5
The CCI was derived from discharge International Classification of Diseases, revision 9 (ICD-9) codes using a method adapted for this use.6 If patients were readmitted to the same hospital within the 12-month follow-up period after initial ICH, then discharge ICD-9 codes from these additional hospitalizations were also used in calculating the CCI. Additional information obtained from study case report forms was then used to supplement ICD-9–derived data.
For the purposes of analysis, CCI scores were categorized as 0, 1, 2, and 3 or greater. Outcome on the modified Rankin Scale (mRS) was considered in 2 ways: as an ordinal variable using all mRS categories and dichotomized (favorable outcome mRS score, 0–3; unfavorable outcome mRS score, 4–6). Univariate analysis was performed using the χ2 test for categorical variables. To assess whether the CCI independently influenced 12-month mRS score, multivariable ordinal logistic regression was performed using the entire mRS score and adjusting for components of the ICH score. Because the use of do not resuscitate orders within the first day of hospital admission (early do not resuscitate orders) is known to influence ICH mortality, this was included in the multivariable model.7 Multivariable logistic regression was also preformed to assess the independent impact of the CCI on risk of 30-day mortality. Statistical analysis was performed using Stata (version 11.1), and significance was considered as P<0.05.
The demographics of this cohort have been previously described.5 Table 1 shows the frequency of comorbidities represented in the CCI and the Figure shows the distribution of sum CCI scores throughout the ICH cohort. The CCI score was significantly associated with an unfavorable 12-month outcome (P=0.02). Overall, 29% of patients with high CCI scores (≥3) had a favorable 12-month functional outcome compared with 48% of patients with a CCI score of 0. CCI scores were not associated with the use of early do not resuscitate orders (P=0.66) or withdrawal of medical support (P=0.08), suggesting that these care limitations were likely not instituted principally because of the presence of comorbid medical conditions.
In multivariate analysis, higher CCI was independently associated with worsened 12-month functional outcome independent of the components of the ICH score and use of early do not resuscitate orders (Table 2). Patients with a high CCI score (≥3) had 3.5-times greater odds of having a 1-point increase in the mRS score at 12 months compared with those patients without CCI comorbidities (P=0.001). In Table 2, there is a trend toward a dose–response relationship between increasing CCI score and increasing odds of worsened 12-month functional outcome. CCI score was also independently predictive of 30-day mortality. In multivariate analysis adjusting for the components of the ICH score, early do not resuscitate orders, and withdrawal of support, high CCI score (≥3) was associated with an 8.8-times increased odds of death at 30 days (95% CI, 1.3–60.1; P=0.03). Sensitivity analyses including 3 patients with recurrent ICH during the study period or excluding patients with possible underlying lesions did not change the impact of CCI on outcome or study conclusions.
In this study, the CCI was independently associated with worsened 12-month functional outcome and higher 30-day mortality after adjusting for the components of the ICH score and the use of early care limitations. This supports the use of the CCI as a validated method of accounting for patient comorbidities in ICH studies. The findings of this study are consistent with previous research evaluating the validity of the CCI in assessing comorbidity and functional outcome in ischemic stroke.3
A possible disadvantage to the CCI relates to the method of its determination. The use of ICD-9 codes for determining comorbidities is limited by the accuracy of coding and may underestimate the true incidence of comorbidities.8,9 Even so, a previous systematic review of the CCI found no difference in the adjusted odds ratio for mortality regardless of whether the CCI was derived from administrative data or chart review.9 Two other limitations inherent in the CCI are that not all possible medical comorbidities are represented and that the contribution of individual comorbidities to patient outcome is unable to be assessed because of the sum score nature of the CCI. There are several limitations of this current study, including the relatively small number of patients (n=243) and the fact that the CCI was determined retrospectively. Strengths were the ability to adjust for known ICH outcome predictors and the availability of prospectively acquired long-term functional outcome data. Also, the prospective nature of this study allowed the contribution of previous stroke and hemiplegia to be included.
Sources of Funding
This work was funded in part by National Institutes of Health/National Institute of Neurological Disorders and Stroke (NIH/NINDS) grant K23 NS41240 (J.C.H.) and an unrestricted research grant from Novo Nordisk. This publication was supported by NIH/National Center for Research Resources/Office of the Director, University of California, San Francisco–Clinical and Translational Science Institute grant KL2 RR024130. Dr Hemphill is also funded by NIH/NINDS grant U10 NS058931.
J.C.H. has received more than $10 000 in research support from Novo Nordisk, with the last having been received 4 years ago. He has received less than $5000 for review of medical records of legal cases in the past 2 years related to intracerebral hemorrhage but has not provided testimony in deposition or trial.
The online-only Data Supplement is available at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.111.617639/-/DC1.
- Received February 23, 2011.
- Accepted April 11, 2011.
- © 2011 American Heart Association, Inc.
- Rost NS,
- Smith EE,
- Chang Y,
- Snider RW,
- Chanderraj R,
- Schwab K,
- et al
- Goldstein LB,
- Samsa GP,
- Matchar DB,
- Horner RD
- Hemphill JC III.,
- Farrant M,
- Neill TA Jr.
- Becker KJ,
- Baxter AB,
- Cohen WA,
- Bybee HM,
- Tirschwell DL,
- Newell DW,
- et al