Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data

Department of Community Health Sciences, University of Calgary, Calgary, Alberta, Canada.
Medical Care (Impact Factor: 3.23). 12/2005; 43(11):1130-9.
Source: PubMed


Implementation of the International Statistical Classification of Disease and Related Health Problems, 10th Revision (ICD-10) coding system presents challenges for using administrative data. Recognizing this, we conducted a multistep process to develop ICD-10 coding algorithms to define Charlson and Elixhauser comorbidities in administrative data and assess the performance of the resulting algorithms.
ICD-10 coding algorithms were developed by "translation" of the ICD-9-CM codes constituting Deyo's (for Charlson comorbidities) and Elixhauser's coding algorithms and by physicians' assessment of the face-validity of selected ICD-10 codes. The process of carefully developing ICD-10 algorithms also produced modified and enhanced ICD-9-CM coding algorithms for the Charlson and Elixhauser comorbidities. We then used data on in-patients aged 18 years and older in ICD-9-CM and ICD-10 administrative hospital discharge data from a Canadian health region to assess the comorbidity frequencies and mortality prediction achieved by the original ICD-9-CM algorithms, the enhanced ICD-9-CM algorithms, and the new ICD-10 coding algorithms.
Among 56,585 patients in the ICD-9-CM data and 58,805 patients in the ICD-10 data, frequencies of the 17 Charlson comorbidities and the 30 Elixhauser comorbidities remained generally similar across algorithms. The new ICD-10 and enhanced ICD-9-CM coding algorithms either matched or outperformed the original Deyo and Elixhauser ICD-9-CM coding algorithms in predicting in-hospital mortality. The C-statistic was 0.842 for Deyo's ICD-9-CM coding algorithm, 0.860 for the ICD-10 coding algorithm, and 0.859 for the enhanced ICD-9-CM coding algorithm, 0.868 for the original Elixhauser ICD-9-CM coding algorithm, 0.870 for the ICD-10 coding algorithm and 0.878 for the enhanced ICD-9-CM coding algorithm.
These newly developed ICD-10 and ICD-9-CM comorbidity coding algorithms produce similar estimates of comorbidity prevalence in administrative data, and may outperform existing ICD-9-CM coding algorithms.

1 Follower
104 Reads
    • "Diabetes mellitus and hypertension were identified using validated algorithms from hospital discharge records and physician claims [19] [20]. Other comorbid conditions based on the Deyo classification of Charlson comorbidities were identified from the physician claims and hospitalization records using validated ICD-9-CM and ICD-10 coding algorithms [21]. Proteinuria was estimated by urine albumin: creatinine ratio (ACR) or urine dipstick based on outpatient random spot urine measurements . "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background: Using a community-based cohort we sought to investigate the association between change in estimated glomerular filtration rate (eGFR) and risk of incident cardiovascular disease including congestive heart failure (CHF), acute myocardial infarction (AMI), and stroke. Methods: We identified 479,126 adults without a history of cardiovascular disease who had at least 3 outpatient eGFR measurements over a 4year period in Alberta, Canada. Change in eGFR was estimated as the absolute annual rate of change (categorized as ≤-5, -4, -3, -2, -1, 0, 1, 2, 3, 4, and ≥5mL/min/1.73m(2)/year). In a sensitivity analysis we also estimated change as the annual percentage change (categorized as ≤-7, -6 to -5, -4 to -3, -2 to -1, 0, 1 to 2, 3 to 4, 5 to 6, and ≥7%/year). The adjusted risk of incident CHF, AMI, and stroke associated with each category of change in eGFR was estimated, using no change in eGFR as the reference, RESULTS: There were 2622 (0.6%) CHF, 3463 (0.7%) AMI, and 2768 (0.6%) stroke events over a median follow-up of 2.5years. Compared to participants with stable eGFR, those with the greatest decline (≤-5mL/min/1.73m(2)/year) had more than a two-fold increased risk of CHF (HR 2.57; 95% CI: 2.28 to 2.89). Risk for AMI and stroke was increased by 31% and 29%, respectively. After adjusting for the last eGFR at the end of the accrual period, the observed association remained significantly higher for CHF but diminished for AMI and stroke. A similar pattern was observed when change in eGFR was quantified as annual percentage change. Conclusion: In this large community-based cohort, we observed that a declining eGFR was associated with an increased risk of CHF, AMI, and stroke. However, when the risk of CVD events was adjusted for the last eGFR measurement, decline in eGFR per se was no longer associated with increased risk of AMI or stroke, and the association with CHF remained significant but was attenuated. These results demonstrate the importance of monitoring change in eGFR over time to improve cardiovascular risk prognostication.
    International Journal of Cardiology 09/2015; 202. DOI:10.1016/j.ijcard.2015.09.090 · 4.04 Impact Factor
  • Source
    • "Information on comorbidities was generated from ICD-9-CM codes in the clinical database with coding algorithms as described by Quan et al. [17]. The CHA 2 DS 2 -VASc score and Charlson comorbidity index were also calculated for each patient for risk stratification [18] [19] [20]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Introduction Although there are many different antiarrhythmic drugs (AADs) approved for rhythm management of atrial fibrillation (AF), little comparative effectiveness data exist to guide drug selection. Methods We followed 5952 consecutive AF patients who were prescribed amiodarone (N = 2266), dronedarone (N = 488), dofetilide (N = 539), sotalol (N = 1718), or class 1C agents (N = 941) to the primary end point of AF recurrence. Results Median follow-up time was 18.2 months (range 0.1–101.6 months). Patients who were prescribed amiodarone had the highest, while patients on class 1C agents had the lowest baseline CHA2DS2-VASc score, Charlson comorbidity index, and burden of comorbid illnesses including coronary artery disease, congestive heart failure, diabetes mellitus, hyperlipidemia, chronic obstructive lung disease, chronic kidney disease, or cancer (p < 0.05 for all comparisons). After adjusting for baseline characteristics, using dronedarone as benchmark, amiodarone [hazard ratio (HR) 0.58, p < 0.001], class 1C agents (HR 0.70, p < 0.001), and sotalol (HR 0.79, p = 0.008), but not dofetilide (HR 0.87, p = 0.178) were associated with less AF recurrence. In addition, compared to dronedarone, amiodarone and class 1C agents were associated with lower rates of admissions for AF (HR 0.55, p < 0.001 for amiodarone; HR 0.71, p = 0.021 for class 1C agents) and all-cause mortality was lowest in patients treated with class 1C agents (HR 0.42, p = 0.018). The risk of stroke was similar among all groups. Conclusion Compared with dronedarone, amiodarone, class 1C agents, and sotalol are more effective for rhythm control, while dofetilide had similar efficacy. These findings have important implications for clinical practice.
    Journal of Cardiology 07/2015; DOI:10.1016/j.jjcc.2015.07.001 · 2.78 Impact Factor
  • Source
    • "From this group, cases surviving the admission and with any communication disorder or a relevant procedure coded in any admission during the time period will be identified and analysed by Aboriginal status, type of communication disorder, region, age-group, sex and Stroke/TBI. Co-morbidities will be described for patients with and without communication disorders in terms of the Charlson co-morbidity index (Quan et al., 2005) and specific conditions like diabetes, chronic renal disease, alcohol-related admissions and other injuries/trauma using a 5-year look-back period, gleaned from hospital records. A similar analysis will be completed for each of the selected hospitals. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Aboriginal and Torres Strait Islander Australians experience stroke and traumatic brain injury (TBI) with much greater frequency than non-Aboriginal Australians. Acquired communication disorders (ACD) can result from these conditions and can significantly impact everyday life. Yet few Aboriginal people access rehabilitation services and little is known about Aboriginal peoples’ experiences of ACD. This paper describes the protocol surrounding a study that aims to explore the extent and impact of ACD in Western Australian Aboriginal populations following stroke or TBI and develop a culturally appropriate screening tool for ACD and accessible and culturally appropriate service delivery models. Method/Design: The 3-year, mixed methods study is being conducted in metropolitan Perth and five regional centres in Western Australia. Situated within an Aboriginal research framework, methods include an analysis of linked routine hospital admission data and retrospective file audits, development of a screening tool for ACD, interviews with people with ACD, their families, and health professionals, and drafting of alternative service delivery models. Discussion: This study will address the extent of ACD in Aboriginal populations and document challenges for Aboriginal people in accessing speech pathology services. Documenting the burden and impact of ACD within a culturally secure framework is a forerunner to developing better ways to address the problems faced by Aboriginal people with ACD and their families. This will in turn increase the likelihood that Aboriginal people with ACD will be diagnosed and referred to professional support to improve their communication, quality of life and functioning within the family and community context.
    Brain Impairment 07/2015; -1:1-12. DOI:10.1017/BrImp.2015.15 · 0.81 Impact Factor
Show more