The object of this article was to systematically review available methods to measure comorbidity and to assess their validity and reliability. A search was made in Medline and Embase, with the keywords comorbidity and multi-morbidity, to identify articles in which a method to measure comorbidity was described. The references of these articles were also checked, and using a standardized checklist the relevant data were extracted from these articles. An assessment was made of the content, concurrent, predictive and construct validity, and the reliability. Thirteen different methods to measure comorbidity were identified: one disease count and 12 indexes. Data on content and predictive validity were available for all measures, while data on construct validity were available for nine methods, data on concurrent validity, and interrater reliability for eight methods, and data on intrarater reliability for three methods. The Charlson Index is the most extensively studied comorbidity index for predicting mortality. The Cumulative Illness Rating Scale (CIRS) addresses all relevant body systems without using specific diagnoses. The Index of Coexisting Disease (ICED) has a two-dimensional structure, measuring disease severity and disability, which can be useful when mortality and disability are the outcomes of interest. The Kaplan Index was specifically developed for use in diabetes research. The Charlson Index, the CIRS, the ICED and the Kaplan Index are valid and reliable methods to measure comorbidity that can be used in clinical research. For the other indexes, insufficient data on the clinimetric properties are available.

No abstract

The object of this article was to systematically review available methods to measure comorbidity and to assess their validity and reliability. A search was made in Medline and Embase, with the keywords comorbidity and multi-morbidity, to identify articles in which a method to measure comorbidity was described. The references of these articles were also checked, and using a standardized checklist the relevant data were extracted from these articles. An assessment was made of the content, concurrent, predictive and construct validity, and the reliability. Thirteen different methods to measure comorbidity were identified: one disease count and 12 indexes. Data on content and predictive validity were available for all measures, while data on construct validity were available for nine methods, data on concurrent validity, and interrater reliability for eight methods, and data on intrarater reliability for three methods. The Charlson Index is the most extensively studied comorbidity index for predicting mortality. The Cumulative Illness Rating Scale (CIRS) addresses all relevant body systems without using specific diagnoses. The Index of Coexisting Disease (ICED) has a two-dimensional structure, measuring disease severity and disability, which can be useful when mortality and disability are the outcomes of interest. The Kaplan Index was specifically developed for use in diabetes research. The Charlson Index, the CIRS, the ICED and the Kaplan Index are valid and reliable methods to measure comorbidity that can be used in clinical research. For the other indexes, insufficient data on the clinimetric properties are available.

This is the first study to investigate the recovery of arm function after stroke over a period of 4 years. It is encouraging to note that even after 16 weeks improvement still occurred in some patients. However, considerable long-term loss of arm function, associated disability and perceived problems were found. There is an obvious need to develop effective treatment methods for hemiplegic arm function.

SummaryResearch articles on the prognosis of stroke patients were analysed to identify studies that met sound methodological principles of prognostic research as well as to identify variables capable of predicting functional outcome (ADL) after stroke. Data sources comprised a computer-aided search of published prognostic studies and references to literature used in prognostic studies. Seventy-eight studies were tested for adherence to the following key methodological criteria: reliability and validity of measurement instruments used to assess dependent and independent variables; inclusion of an inception cohort; adequate and uniform end-point of observation; control for drop-outs during period of observation; statistical testing of presumed relationship between dependent and independent variables; sufficient sample size in relation to number of determinants; control for multicollinearity; specification of patient characteristics (i.e. age, type, recurrent stroke and localization of stroke); description of interfering treatment effects during the period of observation; and crossvalidation of the prediction model in a second independent group of patients.Only three studies satisfied nine out of 11 criteria and ten studies eight criteria for the determination of valid prognostic research. The results of these studies indicate that the following variables are valid predictors for functional recovery after stroke: age; previous stroke; urinary continence; consciousness at onset; disorientation in time and place; severity of paralysis; sitting balance; admission ADL score; level of social support and metabolic rate of glucose outside the infarct area in hypertensive patients. This study supports the general opinion that not only are differences in objectives and heterogeneity in stroke patients responsible for the lack of accuracy in predicting functional outcome, but also the methodological flaws in published prognostic research.

A small but statistically significant intensity-effect relationship in the rehabilitation of stroke patients was found. Insufficient contrast in the amount of rehabilitation between experimental and control conditions, organizational setting of rehabilitation management, lack of blinding procedures, and heterogeneity of patient characteristics were major confounding factors.

ABSTRACT. van der Lee JH, de Groot V, Beckerman H, Wagenaar RC, Lankhorst GJ, Bouter LM. The intra-and interrater reliability of the Action Research Arm test: a practical test of upper extremity function in patients with stroke. Arch Phys Med Rehabil 2001;82:14-9.Objectives: To determine the intra-and interrater reliability of the Action Research Arm (ARA) test, to assess its ability to detect a minimal clinically important difference (MCID) of 5.7 points, and to identify less reliable test items.Design: Intrarater reliability of the sum scores and of individual items was assessed by comparing (1) the ratings of the laboratory measurements of 20 patients with the ratings of the same measurements recorded on videotape by the original rater, and (2) the repeated ratings of videotaped measurements by the same rater. Interrater reliability was assessed by comparing the ratings of the videotaped measurements of 2 raters. The resulting limits of agreement were compared with the MCID.Patients: Stratified sample, based on the intake ARA score, of 20 chronic stroke patients (median age, 62yr; median time since stroke onset, 3.6yr; mean intake ARA score, 29.2).Main Outcome Measures: Spearman's rank-order correlation coefficient (Spearman's rho); intraclass correlation coefficient (ICC); mean difference and limits of agreement, based on ARA sum scores; and weighted kappa, based on individual items.Results: All intra-and interrater Spearman's rho and ICC values were higher than .98. The mean difference between ratings was highest for the interrater pair (.75; 95% confidence interval, .02-1.48), suggesting a small systematic difference between raters. Intrarater limits of agreement were Ϫ1.66 to 2.26; interrater limits of agreement were Ϫ2.35 to 3.85. Median weighted kappas exceeded .92. Conclusion:The high intra-and interrater reliability of the ARA test was confirmed, as was its ability to detect a clinically relevant difference of 5.7 points.

The aim of this study is to show the relationship between test-retest reproducibility and responsiveness and to introduce the smallest real difference (SRD) approach, using the sickness impact profile (SIP) in chronic stroke patients as an example. Forty chronic stroke patients were interviewed twice by the same examiner, with a 1-week interval. All patients were interviewed during the qualification period preceding a randomized clinical trial. Test-retest reproducibility has been quantified by the intraclass correlation coefficient (ICC). the standard error of measurement (SEM) and the related smallest real difference (SRD). Responsiveness was defined as the ratio of the clinically relevant change to the SD of the within-stable-subject test-retest differences. The ICC for the total SIP was 0.92, whereas the ICCs for the specified SIP categories varied from 0.63 for the category 'recreation and pastime' to 0.88 for the category 'work'. However, both the SEM and the SRD far more capture the essence of the reproducibility of a measurement instrument. For instance, a total SIP score of an individual patient of 28.3% (which is taken as an example, being the mean score in the study population) should decrease by at least 9.26% or approximately 13 items, before any improvement beyond reproducibility noise can be detected. The responsiveness to change of a health status measurement instrument is closely related to its test-retest reproducibility. This relationship becomes more evident when the SEM and the SRD are used to quantify reproducibility, than when ICC or other correlation coefficients are used.