Validation of CKD and Related Conditions in Existing Datasets: A Systematic Review

doi:10.1053/j.ajkd.2010.05.013

Am J Kidney Dis. Author manuscript; available in PMC 2012 January 1.

Published in final edited form as:

Am J Kidney Dis. 2011 January; 57(1): 44–54.

Published online 2010 August 6. doi: 10.1053/j.ajkd.2010.05.013

PMCID: PMC2978782

NIHMSID: NIHMS219374

Validation of CKD and Related Conditions in Existing Datasets: A Systematic Review

Morgan E. Grams, MD,¹ Laura C. Plantinga, ScM,² Elizabeth Hedgeman, MS, MPH,³ Rajiv Saran, MD, MRCP, MS,³ Gary L. Myers, PhD,⁴ Desmond E. Williams, MD, PhD,⁴ and Neil R. Powe, MD, MPH, MBA², on behalf of the CDC CKD Surveillance Team

¹Department of Medicine, Johns Hopkins School of Medicine, Baltimore, MD

²Department of Medicine, San Francisco General Hospital and University of California, San Francisco, CA

³Department of Medicine, University of Michigan, Ann Arbor, MI

⁴Division of NHANES, National Center for Health Statistics, Centers for Disease Control and Prevention, Atlanta, GA

Correspondence and Reprint Requests: Neil R. Powe, MD, San Francisco General Hospital and University of California, San Francisco, 1001 Potrero Avenue, Building 5F38, San Francisco, CA 94110. Phone: 415-206-3465; Fax: 415-206-5586; Email: npowe/at/medsfgh.ucsf.edu

The publisher's final edited version of this article is available at Am J Kidney Dis

Publisher's Disclaimer

Abstract

Background

Accurate classification of individuals with kidney disease is vital to research and public health efforts aimed at improving health outcomes. Our objective was to identify and synthesize published literature evaluating the accuracy of existing data sources related to kidney disease.

Study design

A systematic review of studies seeking to validate the accuracy of the underlying data relevant to kidney disease.

Setting & Population

U.S.-based and international studies covering a wide range of both outpatient and inpatient study populations.

Selection Criteria for Studies

Any English-language study investigating the prevalence or etiology of kidney disease, the existence of co-morbid conditions, or the cause of death in those with CKD. All definitions and stages of CKD, including end-stage renal disease (ESRD), were accepted.

Index Tests

Presence of a kidney disease-related variable in existing datasets, including administrative datasets and disease registries.

Reference Tests

Presence of a kidney disease-related variable defined by laboratory criteria or medical record review.

Results

Thirty studies were identified. Most studies investigated the accuracy of kidney disease reporting, comparing coded renal disease to that defined by estimated glomerular filtration rate (eGFR). Sensitivity of coded renal disease varied widely (0.08–0.83). Specificity was higher, with all studies reporting values of ≥0.90. The studies evaluating the etiology of CKD, comorbidities, and the cause of death in CKD all used ESRD or transplant populations exclusively, and accuracy was highly variable when compared to ESRD registry data.

Limitations

Only English-language studies were evaluated.

Conclusions

Given the heterogeneous results of validation studies, a variety of attributes of existing data sources, including the accuracy of individual data items within these sources, should be carefully considered prior to use in research, quality improvement and public health efforts.

Data validation is fundamental to effective clinical research. Without accounting for underlying assumptions and algorithms used in measurement, systematic error can be introduced into generated data. With the increasing emphasis on research using secondary data sources – the U.S. government’s augmented support of comparative effectiveness research performed with existing data sources, for example – validating the accuracy of underlying information acquires great importance.¹^,² Data items particularly prone to error are those related to diseases that are under-reported or variable in definition.

Chronic kidney disease (CKD) is common, morbid, and a growing area of focus in clinical research.³^–⁵ Low awareness of CKD, in the public and among providers, is a well-recognized public health concern.⁶^,⁷ Only recently have commercial and hospital-based laboratories begun to report estimated glomerular filtration rate (eGFR), a measure currently integral to the classification of CKD.⁸^,⁹ Some medical professionals may not routinely screen for CKD, even in high-risk patients.¹⁰^–¹² In addition, translation of medical charts to administrative datasets may not be perfect.¹³ Traditional tools for identifying disease in large databases (such as International Classification of Disease (ICD) codes) may therefore poorly reflect the true extent of disease.

Misclassifying CKD can lead to multiple problems. For individual patients, failure to make a timely diagnosis delays delivery of effective treatment.¹⁴ In research efforts, misclassification can bias results. Clearly, the development of surveillance programs to track prevalence of CKD would be hindered by inaccurate diagnosis or coding of CKD. In outcomes research and quality improvement, such inaccuracies could affect the generalizability of results. Missed cases may share characteristics, such as older age or female sex, that influence the observed mix of patients, and thus case-associated characteristics and outcomes may not be representative of the group as a whole. Therefore, before we can have confidence in results, verification of data accuracy is essential.

Our objective was to summarize and interpret studies that sought to validate the accuracy of existing data drawn from different settings as it relates to kidney disease. Cognizant of recent calls for secondary data analysis and comparative effectiveness research, we were particularly interested in large datasets that contained information on the presence, etiology, and associated comorbid conditions of kidney disease, as well as the cause of death in patients with kidney disease. Here, we present a systematic review of studies that address the validity of these kidney disease-related data items in healthcare data.

Methods

Study Design

We performed a systematic review of the literature to identify, review, and summarize peer-reviewed studies that address the accuracy of reported or calculated kidney disease prevalence, cause of CKD, or information on co-morbidities and mortality. All stages of CKD were included, as well as relevant studies examining acute kidney injury and kidney transplant. All human, full-text studies published in English were considered. Studies that compared kidney disease identifiers to the same, re-abstracted identifier (e.g., a comparison of ICD-9-CM codes to re-abstracted ICD-9-CM codes) were excluded.

Identification of the Literature

With guidance from an information specialist, we developed an initial search strategy based on MEDLINE Medical Subject Headings (MeSH) commonly used in a few previously identified relevant studies. Because there is no single MeSH term for CKD, we linked the initial search strategy to a combination of free text key words and additional MeSH terms indicating kidney disease by Boolean operators. We tailored the search to the PubMed electronic database, using limits of Humans and English Language. The full search strategy can be found in Item S1 (provided as online supplementary material available with this article at www.ajkd.org). In addition, we reviewed the references of all included articles. Finally, using the Web of Science’s citation manager, we reviewed titles of studies citing any previously identified included article.

Extraction of the Data

Once a study was identified as meeting inclusion criteria, two review authors (MEG, LCP) independently abstracted data items and entered them into predesigned templates. Data extracted included characteristics of the study population and the origin of the datasets used. Information on kidney disease included type (stage 1–5 CKD, ESRD, acute kidney injury (AKI) or kidney transplant) and the method of determination, as well as its gold standard. Prevalence of each variable, if reported, was abstracted. We also noted any reported measure of accuracy, such as sensitivity and specificity, agreement between sources, or kappa statistics, and any subgroup analysis designed to determine which patients were misclassified. We manually calculated unreported sensitivity and specificity if the study supplied the necessary underlying information.

Synthesis of the Data

From the abstracted data, we reported the range of sensitivities and specificities for each variable of interest: presence of kidney disease, cause of kidney disease, comorbid conditions with kidney disease, and cause of death in kidney disease. We produced evidence tables for each variable of interest, including both U.S.-based and international studies.

Results

The flow chart of study selection is displayed in Figure 1. Our PubMed search resulted in 6705 titles, from which we identified 20 articles meeting inclusion criteria. Eight additional studies were identified through references of these identified studies, and two additional studies were identified through the Web of Science citation manager. Characteristics of the final 30 studies (18 of which were US-based) are shown in Table 1. Initial inter-reviewer agreement for abstracted data items was 86%; after clarification and resolution of errors, agreement was 100%.

Figure 1

Flow Chart of Study Selection

Table 1

Characteristics of studies validating the accuracy of existing data sources for examining kidney disease

Accuracy of Reported Presence of Kidney Disease

Ten U.S. studies evaluated the accuracy of the variable “kidney disease” within their data source (Table 2).¹⁰^,¹⁵^–²³ Six evaluated the accuracy of reported presence of CKD¹⁰^,¹⁵^,¹⁶^,¹⁹^,²¹^,²²; four evaluated AKI.¹⁷^,¹⁸^,²⁰^,²³ Data sources included Medicare claims data, the National Hospital Discharge Survey data, medical records, laboratory requisition data, a heart failure registry, and hospital administrative databases. Study populations varied widely: healthy outpatients, veterans with diabetes, elderly Medicare beneficiaries hospitalized for acute myocardial infarction, and inpatient hospitalizations, both in general and for specific causes, such as heart failure, COPD, or diabetes. As such, the prevalence of “kidney disease” varied, from 28% to 67% for CKD, and from 1% to 12% in the AKI studies. All but two²¹^,²² defined “kidney disease” by ICD codes in the data source (underlying coding algorithms are listed in Table S1). The gold standard was uniform in the CKD studies; CKD was defined as an estimated glomerular filtration rate (eGFR) of less than 60 ml/min/1.73 m² based on serum creatinine, although Ferris et al.¹⁵ included only Stages 3–4. The AKI studies used a gold standard of medical record mention, a 100% increase in inpatient serum creatinine from nadir to peak, or a graded definition depending on baseline serum creatinine proposed by Hou et al.²⁴ Using these definitions, sensitivity of ranged from 0.11 to 0.45. Specificity was higher, ranging between 0.93 and 1.00.

Table 2

Characteristics of Studies Addressing the Accuracy of a Presence of Kidney Disease Variable, by Country and Type of Kidney Disease

Eleven international studies also met inclusion criteria (Table 2).²⁵^,²⁶^,²⁸^–³¹^,³³^–³⁷ Seven were Canadian studies;²⁵^,²⁶^,²⁸^–³¹^,³³ four of the seven had common authorship.²⁹^–³¹^,³³ All but one²⁶ evaluated code-defined “kidney disease” in administrative databases against medical record mention of CKD. Study populations included patients hospitalized for acute myocardial infarction, patients undergoing percutaneous coronary intervention, admissions for heart failure, and selected hospital discharges. Prevalence of kidney disease was lower than the U.S. studies, ranging from 1.5 to 14.1%. Sensitivity ranged from 0.42 to 0.83. Specificity, when reported, exceeded 0.96 in all studies. The UK³⁵ and Italian³⁷ studies were similar, comparing general practice medical charts to a gold standard of eGFR. Prevalence of CKD ranged from 4.5% in the UK study (with the denominator as the entire study population, some of whom did not have measured serum creatinine) to 9.3% in the Italian study. Sensitivity was low, at 0.08 and 0.15 in the UK and Italian studies, respectively. The Israeli study³⁴ evaluated heart failure registry data, comparing the coded presence of “renal insufficiency” from a mandated admission checklist to an eGFR gold standard. Here, the prevalence was higher, at 56.5%, and the corresponding sensitivity was 0.59. Finally, the Australian study³⁶ estimated the accuracy of yearly incident and prevalent cases of ESRD, comparing ICD-coded administrative data to national registry data. Over the 8-year study period, the authors found an annual average variation between the two systems of 7% in both incident and prevalent cases.

Overall, sensitivity estimates varied by gold standard. Using the most inclusive definition of kidney disease (i.e., the reported algorithm with the highest estimate of sensitivity), we compared sensitivities among studies using the medical record as the gold standard to those using eGFR or, in the AKI studies, a creatinine-based criteria (Figure 2). Studies comparing coded renal disease to laboratory-defined kidney disease reported, on average, lower estimates of sensitivities than those using a medical record comparison (p<0.001). Of note, six of the seven studies using the medical record as a gold standard were international.²⁵^,²⁸^–³¹^,³³

Figure 2

Estimates of Sensitivity in Studies Evaluating the Presence of Kidney Disease, by Gold Standard. Abbreviation: eGFR, estimated glomerular filtration rate.

Five studies analyzed misclassified cases.¹⁰^,¹⁵^,¹⁸^,¹⁹^,³⁴ Ferris et al.¹⁵ found that women, whites, and those with less severe CKD were more likely to go unlabeled by ICD-codes; diabetes and hypertension were associated with correct classification. Amsalem et al.³⁴ found that women, older adults, and those with less severe kidney impairment were more likely to be misclassified. Stevens et al.¹⁰ estimated higher sensitivity in severe CKD, but no difference in sensitivity or specificity by presence of CKD risk factors. Waikar et al.¹⁸ observed higher rates of kidney disease identification in men, older adults, patients hospitalized on a medical service, and those requiring dialysis or dying as an inpatient. Finally, Winkelmayer et al.¹⁹ found increased CKD identification in patients with more physician visits, more hospitalized days, or a previous nursing home stay.

Accuracy of Reported Etiology of Kidney Disease

Three studies reported the accuracy of “etiology of kidney disease”³⁸^–⁴⁰; all evaluated a U.S. ESRD or transplant population (Table 3). The data sources included the Health Care Financing Administration Program Medical and Management Information System (HCFA PMMIS) database (precursor to Center for Medicare and Medicaid Studies or CMS), the United States Renal Data System (USRDS), the New York Statewide Planning and Research Cooperative System (SPARCS) database, the Michigan Kidney Registry (MKR), and data collected from individual dialysis units and transplant centers. The populations were similar: New York Medicare ESRD patients with a hospitalization over a 5-year period, Michigan Medicare ESRD patients, and a national random sample of Medicare ESRD patients. Agreement between data sources on disease etiology ranged between 59% and 89%, or from 79% and 89% if secondary codes were used.

Table 3

Characteristics of Studies Addressing the Accuracy of Reported Etiology of Kidney Disease

Two studies reviewed misclassified cases. Byrne et al.³⁸ observed a higher match rate in blacks, younger patients, and those with more hospital admissions, and a lower rate in patients with polycystic kidney disease. The USRDS study⁴⁰ showed lowest match rates for patients whose etiology of disease was classified as “other.”

Accuracy of Reported Comorbid Conditions in Patients with Kidney Disease

Three studies estimated accuracy of reported comorbid conditions in patients with kidney disease (Table 4).⁴¹^–⁴³ Longenecker et al.⁴¹ and Merkin et al.⁴² identified a range of conditions in an ESRD population, and Lentine et al.⁴³ evaluated cardiovascular disease in kidney transplant recipients. Data sources include the CMS Medical Evidence Form (a.k.a. Form 2728), data from a national prospective cohort of incident dialysis patients, Medicare billing claims, and a single center kidney transplant administrative database. Two study populations were examined in the three studies: kidney transplant recipients with Medicare as their primary insurance,⁴³ and a national prospective cohort of incident dialysis patients.⁴¹^,⁴² In the ESRD populations,⁴¹^,⁴² prevalence of comorbid conditions varied, from 10% for cancer to 96% for hypertension. Compared to the medical record, sensitivity estimates for Form 2728 were lower for pulmonary disorders (COPD, 33%) and higher for diabetes (75%) and hypertension (77%).⁴¹ Specificity was high (91% to 100%).

Table 4

Characteristics of Studies Addressing the Accuracy of Reported Comorbid Conditions with Kidney Disease

Accuracy of Reported Cause of Death in Patients with Kidney Disease

Two U.S. studies addressed the accuracy of the variable “cause of death” in ESRD populations (Table 5).⁴⁴^,⁴⁵ Perneger et al.⁴⁴ compared death certificate reports to the Maryland ESRD registry, finding an overall agreement of 31% between sources, with a kappa of 0.24. Rocco et al.⁴⁵ compared the adjudicated classification of death from a national prospective randomized controlled trial of prevalent hemodialysis patients with the HCFA death notification form. Notably, the HCFA death notification form helped to inform the cause of death classification by committee. Agreement varied considerably by cause of death; no overall agreement was calculated.

Table 5

Characteristics of Studies Addressing the Accuracy of Reported Cause of Death in Kidney Disease

An Australian study compared “cause of death” from registry data to the death codes to death certificate data (Table 5).⁴⁶ Kappa ranged from 0.01 to 0.73, depending on cause of death. As in the U.S.-based studies, agreement was highest in death from neoplasm. Indigenous patients were less likely to match. There was no clear temporal change in accuracy overall.

Discussion

This review is, to our knowledge, the first to systematically summarize studies that assess the validity of existing data related to kidney disease. Our results show large variability in accuracy of data items, depending on variable of interest, study population, and, importantly, the comparison gold standard. These results have important implications for clinicians, researchers, provider organizations and payers who seek to improve the health of kidney disease patients through analyses of secondary data sources.

Validation of the variable “kidney disease” is essential in CKD research, as many analyses are predicated on the correct diagnosis and coding of disease. In the 21 studies that address the presence of kidney disease, specificity is high, regardless of gold standard, indicating relatively few false positives. In a study population with a high prevalence of CKD, therefore, a researcher using one of the study datasets could be confident that few patients described as having kidney disease were misclassified. However, if one wanted to assemble a cohort of CKD patients, caution and scrutiny are advisable. Sensitivity of the studies varied widely. Studies using a gold standard of medical record documentation reported higher sensitivities than those using eGFR. No studies incorporated an assessment of kidney function using repeated measurements, proteinuria or measured GFR, the latter arguably the best gold standard.

Gold standard choice affects the case population. Defining CKD by medical record likely underestimates the number of cases due to lack of physician recognition or documentation. Using eGFR as a gold standard may overestimate CKD cases by misclassifying normal elderly or people with AKI; further, not all patients may have measured serum creatinine, and eGFR calculated from non-standardized creatinine measurements may be inaccurate.⁴⁷ As an example, despite similar study populations, 67% of Winkelmayer’s¹⁹ population had CKD (defined by eGFR) whereas So et al³³ found only 17% with either AKI or CKD (defined by medical record). This discrepancy may introduce bias in certain studies. For instance, if white patients with less severe CKD are systematically misclassified as without kidney disease, a comparative effectiveness study of therapeutic interventions may not be generalizable to such patients.

Studies with the same gold standard were also variable. For example, So et al.,³³ Ferris et al.,¹⁵ and Quan et al.²⁹ all evaluated ICD-coded kidney disease against the medical record, estimating sensitivities of 0.83, 0.50, and 0.42, respectively. Many reasons may underlie this variability. Geographic variation in coding practices is well described.¹⁹ Practices may differ between inpatient and outpatient settings. Coding may change over time to incorporate new reimbursement practices, alterations in the numbers of allowable billing codes, and updates in the disease codes themselves.¹⁷ For example, new CKD codes were recently introduced. Using data which differ by medical center, time, or coding practices may systematically bias results.

Validation of CKD-related variables such as etiology, associated comorbid conditions, and cause of death is inherently difficult. First, defining gold standards for variables can be subjective. Tissue diagnosis is considered the gold standard for most etiologies of CKD. However, some patients present too late in the course of disease for an informative biopsy, and biopsy is not often done for certain types of patients, such as those with longstanding diabetes or hypertension. Second, cause of death or kidney disease, if known, is often multifactorial and may not be reducible to one or two ICD codes. For instance, in a patient with multiple comorbidities, clinical differentiation between diabetes and hypertension as the cause of CKD is difficult. Third, many databases use the same sources of information, such as the Medical Evidence Report (Form 2728). Constructing a gold standard on a common source introduces incorporation bias, and overestimates the accuracy of the underlying data.

Despite these limitations, some interesting patterns emerge in the validation of CKD-related data. Among studies evaluating the cause of kidney failure, less agreement was found in polycystic kidney disease, perhaps because fewer PKD patients are hospitalized for complications of PKD, as compared to, for instance, diabetics with diabetic nephropathy. Using inpatient claims data for analysis of a CKD cohort, therefore, may not only under-capture CKD patients, but may also skew the cause of disease toward the etiologies more commonly hospitalized for non-renal reasons, such as diabetes and hypertension.

In conclusion, this review shows that existing data sources need careful scrutiny before use in any research effort, particularly surveillance studies, outcomes research, quality improvement research and comparative effectiveness research. There is a wide range in accuracy of kidney disease-related variables, and additional research is required to investigate the sources of this variation. In addition, initiatives to improve the quality of underlying kidney disease data should be undertaken, including standardized reporting for prevalence estimates. In particular, efforts must be made to increase providers’ recognition and documentation of disease, such as through more systematic testing and standardized ICD coding. Finally, an electronic medical record, integrated with laboratory and vital statistic data, would greatly facilitate clinical and epidemiologic research.

Supplementary Material

01: Item S1

PubMed Search Strategy

Click here to view.^{(27K, pdf)}

02: Table S1

International Classification of Disease Codes Used in Presence of Kidney Disease Studies

Click here to view.^{(123K, pdf)}

ACKNOWLEDGEMENTS

Support: This project was supported under a cooperative agreement from the CDC through the Association of American Medical Colleges (AAMC), grant number MM-1143-10/10. Report contents are solely the responsibility of the authors and do not necessarily represent the official views of the AAMC or CDC. Dr. Powe is partially supported by grant K24DK02643.

Footnotes

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

In addition to authors Powe, Plantinga, Saran, Hedgeman, Williams, and Myers, the CDC CKD Surveillance Team consists of Chi-yuan Hsu, Kirsten Bibbins-Domingo, Delphine Tuot (University of California, San Francisco); Josef Coresh, Edgar Miller III, Deidra Crews (Johns Hopkins University); Lesley Stevens (Tufts University); Brenda Gillespie, William Herman, Freidrich Port, Bruce Robinson, Vahakn Shahinian, Jerry Yee, Eric Young (University of Michigan); and Mark Eberhardt, Paul Eggers, Nicole Flowers, Linda Geiss, Susan Hailpern, Regina Jordan, Juanita Mondeshire, Bernice Moore, Meda Pavkov, Nilka Ríos-Burrows, Deborah Rolka, Sharon Saydah, Anton Schoolwerth, Rodolfo Valdez, Larry Waller (CDC).

None of the funding agencies listed below had any role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Financial Disclosure: The authors declare that they have no relevant financial interests.

REFERENCES

1. Conway PH, Clancy C. Comparative-effectiveness research--implications of the Federal Coordinating Council's report. N Engl J Med. 2009;361(4):328–330. [PubMed]

2. Iglehart JK. Prioritizing comparative-effectiveness research--IOM recommendations. N Engl J Med. 2009;361(4):325–328. [PubMed]

3. Coresh J, Selvin E, Stevens LA, et al. Prevalence of chronic kidney disease in the United States. JAMA. 2007;298(17):2038–2047. [PubMed]

4. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med. 2004;351(13):1296–1305. [PubMed]

5. Weiner DE, Tighiouart H, Stark PC, et al. Kidney disease as a risk factor for recurrent cardiovascular disease and mortality. Am J Kidney Dis. 2004;44(2):198–206. [PubMed]

6. Plantinga LC, Boulware LE, Coresh J, et al. Patient awareness of chronic kidney disease: Trends and predictors. Arch Intern Med. 2008;168(20):2268–2275. [PMC free article] [PubMed]

7. Nickolas TL, Frisch GD, Opotowsky AR, Arons R, Radhakrishnan J. Awareness of kidney disease in the US population: Findings from the National Health and Nutrition Examination Survey (NHANES) 1999 to 2000. Am J Kidney Dis. 2004;44(2):185–197. [PubMed]

8. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease study group. Ann Intern Med. 1999;130(6):461–470. [PubMed]

9. Levey AS, Eckardt KU, Tsukamoto Y, et al. Definition and classification of chronic kidney disease: A position statement from Kidney Disease: Improving Global Outcomes (KDIGO) Kidney Int. 2005;67(6):2089–2100. [PubMed]

10. Stevens LA, Fares G, Fleming J, et al. Low rates of testing and diagnostic codes usage in a commercial clinical laboratory: Evidence for lack of physician awareness of chronic kidney disease. J Am Soc Nephrol. 2005;16(8):2439–2448. [PubMed]

11. Kinchen KS, Sadler J, Fink N, et al. The timing of specialist evaluation in chronic kidney disease and mortality. Ann Intern Med. 2002;137(6):479–486. [PubMed]

12. Hsu CY, Vittinghoff E, Lin F, Shlipak MG. The incidence of end-stage renal disease is increasing faster than the prevalence of chronic renal insufficiency. Ann Intern Med. 2004;141(2):95–101. [PubMed]

13. Fisher ES, Whaley FS, Krushat WM, et al. The accuracy of Medicare's hospital claims data: Progress has been made, but problems remain. Am J Public Health. 1992;82(2):243–248. [PubMed]

14. Jungers P, Massy ZA, Nguyen-Khoa T, et al. Longer duration of predialysis nephrological care is associated with improved long-term survival of dialysis patients. Nephrol Dial Transplant. 2001;16(12):2357–2364. [PubMed]

15. Ferris M, Shoham DA, Pierre-Louis M, Mandhelker L, Detwiler RK, Kshirsagar AV. High prevalence of unlabeled chronic kidney disease among inpatients at a tertiary-care hospital. Am J Med Sci. 2009;337(2):93–97. [PubMed]

16. Kern EF, Maney M, Miller DR, et al. Failure of ICD-9-CM codes to identify patients with comorbid chronic kidney disease in diabetes. Health Serv Res. 2006;41(2):564–580. [PMC free article] [PubMed]

17. Waikar SS, Curhan GC, Wald R, McCarthy EP, Chertow GM. Declining mortality in patients with acute renal failure, 1988 to 2002. J Am Soc Nephrol. 2006;17(4):1143–1150. [PubMed]

18. Waikar SS, Wald R, Chertow GM, et al. Validity of International Classification of Diseases, Ninth Revision, Clinical Modification codes for acute renal failure. J Am Soc Nephrol. 2006;17(6):1688–1694. [PubMed]

19. Winkelmayer WC, Schneeweiss S, Mogun H, Patrick AR, Avorn J, Solomon DH. Identification of individuals with CKD from Medicare claims data: A validation study. Am J Kidney Dis. 2005;46(2):225–232. [PubMed]

20. Liangos O, Wald R, O'Bell JW, Price L, Pereira BJ, Jaber BL. Epidemiology and outcomes of acute renal failure in hospitalized patients: A national survey. Clin J Am Soc Nephrol. 2006;1(1):43–51. [PubMed]

21. Ryan TP, Sloand JA, Winters PC, Corsetti JP, Fisher SG. Chronic kidney disease prevalence and rate of diagnosis. Am J Med. 2007;120(11):981–986. [PubMed]

22. Heywood JT, Fonarow GC, Costanzo MR, et al. High prevalence of renal dysfunction and its impact on outcome in 118,465 patients hospitalized with acute decompensated heart failure: A report from the ADHERE database. J Card Fail. 2007;13(6):422–430. [PubMed]

23. Geraci JM, Ashton CM, Kuykendall DH, Johnson ML, Wu L. International Classification of Diseases, 9th Revision, Clinical Modification codes in discharge abstracts are poor measures of complication occurrence in medical inpatients. Med Care. 1997;35(6):589–602. [PubMed]

24. Hou SH, Bushinsky DA, Wish JB, Cohen JJ, Harrington JT. Hospital-acquired renal insufficiency: A prospective study. Am J Med. 1983;74(2):243–248. [PubMed]

25. Humphries KH, Rankin JM, Carere RG, Buller CE, Kiely FM, Spinelli JJ. Co-morbidity data in outcomes research: Are clinical data derived from administrative databases a reliable alternative to chart review? J Clin Epidemiol. 2000;53(4):343–349. [PubMed]

26. Lee DS, Donovan L, Austin PC, et al. Comparison of coding of heart failure and comorbidities in administrative and clinical data for use in outcomes research. Med Care. 2005;43(2):182–188. [PubMed]

27. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45(6):613–619. [PubMed]

28. Levy AR, Tamblyn RM, Fitchett D, McLeod PJ, Hanley JA. Coding accuracy of hospital discharge data for elderly survivors of myocardial infarction. Can J Cardiol. 1999;15(11):1277–1282. [PubMed]

29. Quan H, Parsons GA, Ghali WA. Validity of information on comorbidity derived from ICD-9-CCM administrative data. Med Care. 2002;40(8):675–685. [PubMed]

30. Quan H, Parsons GA, Ghali WA. Assessing accuracy of diagnosis-type indicators for flagging complications in administrative data. J Clin Epidemiol. 2004;57(4):366–372. [PubMed]

31. Quan H, Li B, Saunders LD, et al. Assessing validity of ICD-9-CM and ICD-10 administrative data in recording clinical conditions in a unique dually coded database. Health Serv Res. 2008;43(4):1424–1441. [PMC free article] [PubMed]

32. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130–1139. [PubMed]

33. So L, Evans D, Quan H. ICD-10 coding algorithms for defining comorbidities of acute myocardial infarction. BMC Health Serv Res. 2006;6:161. [PMC free article] [PubMed]

34. Amsalem Y, Garty M, Schwartz R, et al. Prevalence and significance of unrecognized renal insufficiency in patients with heart failure. Eur Heart J. 2008;29(8):1029–1036. [PubMed]

35. Anandarajah S, Tai T, de Lusignan S, et al. The validity of searching routinely collected general practice computer data to identify patients with chronic kidney disease (CKD): A manual review of 500 medical records. Nephrol Dial Transplant. 2005;20(10):2089–2096. [PubMed]

36. Brameld KJ, Thomas MA, Holman CD, Bass AJ, Rouse IL. Validation of linked administrative data on end-stage renal failure: Application of record linkage to a 'clinical base population'. Aust N Z J Public Health. 1999;23(5):464–467. [PubMed]

37. Minutolo R, De Nicola L, Mazzaglia G, et al. Detection and awareness of moderate to advanced CKD by primary care practitioners: A cross-sectional study from italy. Am J Kidney Dis. 2008;52(3):444–453. [PubMed]

38. Byrne C, Vernon P. Validation of the cause of renal failure of patients in the Medicare end-stage renal disease program. Am J Kidney Dis. 1991;18(3):375–378. [PubMed]

39. How good are the data? USRDS data validation special study. Am J Kidney Dis. 1992;20(5 Suppl 2):68–83. [PubMed]

40. Completeness and reliability of USRDS data: Comparisons with the Michigan Kidney Registry. Am J Kidney Dis. 1992;20(5 Suppl 2):84–88. [PubMed]

41. Longenecker JC, Coresh J, Klag MJ, et al. Validation of comorbid conditions on the End-Stage Renal Disease Medical Evidence Report: The CHOICE study. Choices for Healthy Outcomes in Caring for ESRD. J Am Soc Nephrol. 2000;11(3):520–529. [PubMed]

42. Merkin SS, Cavanaugh K, Longenecker JC, Fink NE, Levey AS, Powe NR. Agreement of self-reported comorbid conditions with medical and physician reports varied by disease among end-stage renal disease patients. J Clin Epidemiol. 2007;60(6):634–642. [PMC free article] [PubMed]

43. Lentine KL, Schnitzler MA, Abbott KC, Bramesfeld K, Buchanan PM, Brennan DC. Sensitivity of billing claims for cardiovascular disease events among kidney transplant recipients. Clin J Am Soc Nephrol. 2009;4(7):1213–1221. [PubMed]

44. Perneger TV, Klag MJ, Whelton PK. Cause of death in patients with end-stage renal disease: Death certificates vs registry reports. Am J Public Health. 1993;83(12):1735–1738. [PubMed]

45. Rocco MV, Yan G, Gassman J, et al. Comparison of causes of death using HEMO study and HCFA end-stage renal disease death notification classification systems. The National Institutes of Health-funded Hemodialysis. Health care Financing Administration. Am J Kidney Dis. 2002;39(1):146–153. [PubMed]

46. Li SQ, Cass A, Cunningham J. Cause of death in patients with end-stage renal disease: Assessing concordance of death certificates with registry reports. Aust N Z J Public Health. 2003;27(4):419–424. [PubMed]

47. Myers GL, Miller WG, Coresh J, et al. Recommendations for improving serum creatinine measurement: A report from the laboratory working group of the National Kidney Disease Education Program. Clin Chem. 2006;52(1):5–18. [PubMed]