The predictive value of ICD-10 diagnostic coding used to assess Charlson comorbidity index conditions in the population-based Danish National Registry of Patients.
Thygesen SK, Christiansen CF, Christensen S, Lash TL, Sørensen HT. · 2011
Read The predictive value of ICD-10 diagnostic coding used to assess Charlson comorbidity index conditions in the population-based Danish National Registry of Patients. by Thygesen SK, Christiansen CF, Christensen S, Lash TL, Sørensen HT. (2011) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc3125388.txt.
Introduction
Sandra Thygesen and colleagues checked hospital coding accuracy in Denmark in a 2011 paper in BMC Medical Research Methodology. Researchers often use a standard index of nineteen long-term diseases to adjust hospital studies, trusting that registry codes match real diagnoses. The team audited that trust against patient records.
They reviewed discharge papers and charts for fifty hospital contacts per disease, drawn from a decade of northern Danish admissions. Nearly every coded diagnosis was confirmed on review. Twelve of the nineteen diseases scored perfectly, and the weakest, diabetes with complications, still scored well above four fifths correct. The authors conclude the registry codes are reliable enough for comorbidity adjustment in research, at least in this Danish setting.
Catalog record
| CREATOR | Thygesen SK, Christiansen CF, Christensen S, Lash TL, Sørensen HT. |
|---|---|
| YEAR | 2011 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The predictive value of ICD-10 diagnostic coding used to assess Charlson comorbidity index conditions in the population-based Danish National Registry of Patients By Sandra K Thygesen, Christian F Christiansen, Steffen Christensen, Timothy L Lash, Henrik T Sørensen Published in: BMC Medical Research Methodology DOI: 10.1186/1471-2288-11-83 ABSTRACT Abstract Background The Charlson comorbidity index is often used to control for confounding in research based on medical databases. There are few studies of the accuracy of the codes obtained from these databases. We examined the positive predictive value (PPV) of the ICD-10 diagnostic coding in the Danish National Registry of Patients (NRP) for the 19 Charlson conditions. Methods Among all hospitalizations in Northern Denmark between 1 January 1998 and 31 December 2007 with a first-listed diagnosis of a Charlson condition in the NRP, we selected 50 hospital contacts for each condition. We reviewed discharge summaries and medical records to verify the NRP diagnoses, and computed the PPV as the proportion of confirmed diagnoses. Results A total of 950 records were reviewed. The overall PPV for the 19 Charlson conditions was 98.0% (95% CI; 96.9, 98.8). The PPVs ranged from 82.0% (95% CI; 68.6%, 91.4%) for diabetes with diabetic complications to 100% (one-sided 97.5% CI; 92.9%, 100%) for congestive heart failure, peripheral vascular disease, chronic pulmonary disease, mild and severe liver disease, hemiplegia, renal disease, leukaemia, lymphoma, metastatic tumour, and AIDS. Conclusion The PPV of NRP coding of the Charlson conditions was consistently high. BACKGROUND Comorbidities are coexistent diseases to a disease of interest [1] or index disease [2]. Comorbidities may directly affect the prognosis of the disease of interest, or may indirectly affect the prognosis by affecting the choice of treatment [1-3]. An index of comorbidity level has the advantage of bringing several comorbidities into a single numeric score, thereby reducing the number of candidate variables into a manageable set of proxy variables, which is especially beneficial when using administrative databases or medical registries [4,5]. The Charlson comorbidity index is the most widely used comorbidity index [3,6]. It was developed to predict 1-year mortality among 604 medical patients based on comorbidity data obtained from hospital chart review in a single US hospital in 1984 [6]. The 19 Charlson conditions were selected and weighted according to their potential influence on mortality and validated for predicting 1-year mortality in a cohort of 685 breast cancer patients [6]. Since then, the Charlson comorbidity index has been adapted for use with data from administrative databases and medical registries that record medical conditions using the International Classification of Diseases, 9th revision (ICD-9), a Clinical Modification of ICD-9 (ICD-9-CM) [7,8], and recently also the International Classification of Diseases, 10th revision (ICD-10) [9-11]. Only two studies have examined the accuracy of diagnosis coding of Charlson comorbidity conditions in administrative hospital registries compared with diagnoses obtained through medical records. A study by Quan et al [12] was carried out in 1996-1997 in Calgary, Canada using ICD-9 diagnosis codes, but the results may not necessarily hold today for European registries using ICD-10 codes. Henderson et al. assessed the quality of coding in Victoria, Australia soon after the implementation of ICD-10 in 1998-2001 compared with ICD-9 coding in the earlier years and found high coding quality in both time periods [13]. Medical registries and administrative databases offer an important resource for studies of public health issues [14]. Scandinavian population-based medical registries can be linked using unique personal identifiers and are therefore used extensively for epidemiologic research [15,16]. Data collection and coding procedures may vary across countries [17], and no Nordic study has validated the coding procedure on all conditions included in the Charlson comorbidity index. Measuring comorbidity accurately is important, since controlling for confounding by comorbidity affects the validity of such epidemiologic studies [3,18]. Sufficient control for confounding requires high data quality [19]. We therefore conducted this study to assess the positive predictive value (PPV) of the coding of each of the 19 Charlson comorbidity conditions assessed by ICD-10 diagnoses from a population-based medical registry. METHODS Study population This study was conducted in the North Jutland Region, Denmark using patients with diagnoses registered in the Danish National Registry of Patients (NRP) between 1 January 1998 and 31 December 2007. The population of this region is approximately 500,000 people, corresponding to about 11% of the total Danish population. The Danish population receives tax-supported health care without additional charge. The National Registry of Patients The Danish NRP includes data on all non-psychiatric hospital admissions in Denmark since 1977 and outpatient clinic and emergency room visits since 1995. The NRP includes data on date of admissions and discharges, surgical procedures performed, and up to 20 diagnoses classified according to the International Classification of Diseases, 8th revision (ICD-8) until the end of 1993 and 10th revision (ICD-10) thereafter. The physician who discharges the patient reviews the medical record and makes a discharge summary including discharge diagnoses coded using ICD codes. ICD codes are then entered by the medical secretary into the hospital registry. From there, the data are electronically transmitted to the NRP at the National Board of Health (Figure 1) [16,20]. [Figure: Coding process during an admission to a hospital in Denmark.] We used the NRP to identify all hospital contacts (comprising both in- and outpatients) in the study population with one of the Charlson comorbidity conditions as a first-listed diagnosis [21]. We did not make any age restriction. Outpatients included emergency room contacts and patients followed in outpatient clinics. According to Danish practice, the first-listed diagnosis in the discharge record is the main reason for the hospital contact. For each of the 19 Charlson comorbidity conditions, we selected randomly five hospital contacts per year in the ten year study period, yielding a total of 950 hospital contacts. We used the patient's unique personal identification number and date of admission and discharge to identify the discharge summary for each of the 950 hospital contacts. The discharge summary describes the most important events during the hospital contact, including reason for admission, diagnostic work-up, treatment, prescribed medications, and plan after discharge [22]. All discharge summaries were reviewed by the same physician (SKT). Record review The review of the discharge summary and medical record began with a confirmation of the personal identification number and the date of admission and discharge for the hospital contact retrieved from the NRP. We then proceeded with confirmation of the diagnosis. We considered a diagnosis to be confirmed if the discharge summary described the exact diagnosis or a diagnosis within the same Charlson comorbidity condition. For example, if a discharge summary or medical record indicated non-insulin dependent diabetes mellitus, and this disease was coded as insulin dependent diabetes mellitus in the NRP, then the diagnosis was considered to be confirmed because the Charlson comorbidity index does not distinguish between non-insulin dependent and insulin dependent diabetes mellitus. If the diagnosis was not described in the discharge summary or if the discharge summary was not available, then the medical record was reviewed to determine whether the diagnosis code could be confirmed. Discharge summaries from outpatient clinic contacts may only include a description of treatment, and in these cases the medical record was retrieved. When the reviewing physician had any doubt about whether the discharge summary or