The Impact of Alternative Lookback Periods and Sources of Information on Charlson Comorbidity Adjustment in Medicare Claims

Department of Medicine, the University of Chicago, Illinois, USA.
Medical Care (Impact Factor: 3.23). 12/1999; 37(11):1128-39. DOI: 10.1097/00005650-199911000-00005
Source: PubMed


The Charlson Score is a particularly popular form of comorbidity adjustment in claims data analysis. However, the effects of certain implementation decisions have not been empirically examined.
To determine the effects of alternative data sources and lookback periods on the performance of Charlson scores in the prediction of mortality following hospitalization.
A representative sample of 1,387 elderly patients hospitalized in 1993, drawn from the Medicare Current Beneficiary Survey (MCBS). Three years of linked Medicare claims and survey instruments were available for all patients, as was 2-year mortality follow-up.
Nested Cox regression and comparisons of areas under the Receiver Operating Characteristic (ROC) curve were used to evaluate ability to predict mortality.
Compared with a 1-year lookback involving solely inpatient claims, statistically and empirically significant improvements in the prediction of mortality are obtained by incorporating alternative sources of data (particularly 2 years of inpatient data and 1 year of outpatient and auxiliary claims), but only if indices derived from distinct sources of data are entered into the regression distinctly. The area under the ROC curve for 1-year mortality predication increases from 0.702 to 0.741 (P = 0.002). Furthermore, these improvements in explanatory power obtained whether one also controls for Charlson scores based on self-reported health history and/or secondary diagnoses from the claim for the index hospitalization itself. Finally, claims-based comorbidity adjustment performs comparably to survey-derived adjustment, with areas under the ROC curve of 0.702 and 0.704, respectively.
The widespread practice of comorbidity adjustment in pre-existing administrative data sources can be improved by taking more complete advantage of existing administrative data sources.

Download full-text


Available from: James X Zhang,
  • Source
    • "Two additional studies have compared chart review vs. administrative data-derived CCIs [48] [49]. One study, evaluating the Quan CCI, found that kappa agreement ranged greatly (from 0.02 to 0.47) according to the comorbidity identified [44]. Another study [49], evaluating a study-specific administrative data CCI using ICD-10 codes, found CCI scores derived from the two sources to be well correlated (r 5 0.88, P ! "
    [Show abstract] [Hide abstract]
    ABSTRACT: Objectives: To conduct a systematic review of studies reporting on the development or validation of comorbidity indices using administrative health data and compare their ability to predict outcomes related to comorbidity (ie, construct validity). Study design and setting: We conducted a comprehensive literature search of MEDLINE and EMBASE, until September 2012. After title and abstract screen, relevant articles were selected for review by two independent investigators. Predictive validity and model fit were measured using c-statistic for dichotomous outcomes and R(2) for continuous outcomes. Results: Our review includes 76 articles. Two categories of comorbidity indices were identified: those identifying comorbidities based on diagnoses, using International Classification of Disease codes from hospitalization or outpatient data, and based on medications, using pharmacy data. The ability of indices studied to predict morbidity-related outcomes ranged from poor (C statistic ≤ 0.69) to excellent (C statistic >0.80) depending on the specific index, outcome measured, and study population. Diagnosis-based measures, particularly the Elixhauser Index and the Romano adaptation of the Charlson Index, resulted in higher ability to predict mortality outcomes. Medication-based indices, such as the Chronic Disease Score, demonstrated better performance for predicting health care utilization. Conclusion: A number of valid comorbidity indices derived from administrative data are available. Selection of an appropriate index should take into account the type of data available, study population, and specific outcome of interest.
    Journal of Clinical Epidemiology 10/2014; 68(1). DOI:10.1016/j.jclinepi.2014.09.010 · 3.42 Impact Factor
  • Source
    • "The other method involves linking multiple hospitalisation records belonging to the same patient and determining whether a smoking diagnosis is present on any of these records. This is referred to as using a lookback period [7], with lookback periods of 1 year [8] and 5 years [9]–[11] variously used to identify smoking. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Adjustment for the differing risk profiles of patients is essential to the use of administrative hospital data for epidemiological research. Smoking is an important factor to include in such adjustments, but the accuracy of the diagnostic codes denoting smoking in hospital records is unknown. The aims of this study were to measure the validity of current smoking and ever smoked status identified from diagnoses in hospital records using a range of algorithms, relative to self-reported smoking status; and to examine whether the misclassification of smoking identified through hospital data is differential or non-differential with respect to common exposures and outcomes. Data from the baseline questionnaire of the 45 and Up Study, completed by 267,153 residents of New South Wales (NSW), Australia, aged 45 years and older, were linked to the NSW Admitted Patient Data Collection. Patients who had been admitted to hospital for an overnight stay between 1 July 2005 and the date of completion of the questionnaire (1 January 2006 to 2 March 2009) were included. Smokers were identified by applying a range of algorithms to hospital admission histories, and compared against self-reported smoking in the questionnaire ('gold standard'). Sensitivities for current smoking ranged from 59% to 84%, while specificities were 94% to 98%. Sensitivities for ever smoked ranged from 45% to 74% and specificities were 93% to 97%. For the majority of algorithms, sensitivities and/or specificities differed significantly according to principal diagnosis, number of comorbidities, socioeconomic status, residential remoteness, Indigenous status, 28 day readmission and 365 day mortality. The identification of smoking through diagnoses in hospital data results in differential misclassification. Risk adjustment based on smoking identified from these data will yield potentially misleading results. Systematic capture of information about smoking in hospital records using a mandatory item would increase the utility of administrative data for epidemiological research.
    PLoS ONE 04/2014; 9(4):e95029. DOI:10.1371/journal.pone.0095029 · 3.23 Impact Factor
  • Source
    • "Specificity for these conditions is also very high and ranges from 99 percent to 100 percent (Romano and Mark 1994). The properties of claims in the more global assessment of overall morbidity burden have also been validated (Zhang, Iwashyna, and Christakis 1999). For certain results, there is one further proviso, namely, that neither member of the couple be a member of a staff-model HMO. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The health oftwo people connected by a social tie may be interdependent. The impact of the death of one spouse on the risk of death of the other, known as the widow/er effect, is a classic example (Parkes and Fitzgerald 1969; Lillard and Waite 1995; Martikainen and Valkonen 1996a; and Schaefer, Quesenberry, and Soora 1995). The impact of illness in one spouse on the risk of ill health or death in the other spouse (the proband under study), is another example. This latter phenomenon, often termed caregiver burden, is typically studied as ifit were unrelated to the widower effect (Clipp and George 1993; Dunkin and Anderson-Hanley 1998; Shaw et al. 1997; Schulz et aI. 2003)-as if ill health in a spouse affects the morbidity, but not necessarily the mortality, of caregiving probands. Indeed, most prior work on caregiver burden has focused on how spousal illness worsens the health of probands, but not on whether it increases their mortality, with the exception of one influential study that suggested that caregiving to dementia patients was a risk factor for death (Schulz and Beach 1999). Moreover, comparisons across different types of spousal diseases, in terms of how they may affect caregiver health, are lacking. Some studies have found that worse physical health in a spouse is
Show more