Incidence of bleeding from gastroduodenal ulcers in patients with end-stage renal disease receiving hemodialysis

Data sources

National Health Insurance is a mandatory health insurance program in Taiwan that provides comprehensive coverage for medical care. Up to 96% of the population was enrolled by 2002.12 Claims data are collected in the National Health Insurance Research Database at the National Health Research Institutes. For research purposes, the institute released to the public a cohort dataset for one million randomly selected individuals and a dataset for 46 576 patients with some catastrophic illnesses (e.g., end-stage renal disease, cancer) who were alive in 2000. The institute collected all records for these individuals from 1995 to 2006. There were no statistically significant differences in age, sex and health care costs between the one million randomly sampled individuals and those enrolled in our study. Comprehensive health care data include the enrolment files, claims data, catastrophic illness files and a registry for drug prescriptions.

Each patient’s original identification number was encrypted for privacy in the cohort dataset. The encryption procedure was consistent between datasets so that all claims belonging to each patient could be linked. Because the dataset consisted of de-identified secondary data released to the public for research purposes, this study was exempt from full review by the institutional review board.

Study groups

Our study included almost all patients in Taiwan receiving long-term hemodialysis, as well as age- and sex-matched controls. We did not include patients who had started hemodialysis in the three months before enrolment and those with acute renal failure because of a high rate of ulcer bleeding in patients with critical medical conditions and acute renal failure.15^,16

In Taiwan, patients with end-stage renal disease who are in need of long-term (more than three months) renal replacement therapy can apply for a catastrophic illness registration card from the National Health Research Institutes. We identified 36 474 patients with end-stage renal disease who had received a diagnosis of chronic renal failure (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code 585) and who had received hemodialysis for more than three months from Jan. 1, 2000 through 2006. We excluded patients receiving peritoneal dialysis and those who had received a kidney transplant before or after enrolment. Figure 1 shows the flow of patients during the course of this study.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1:

Flow of patients with end-stage renal disease undergoing hemodialysis during the seven-year study period. In total, 26 145 patients were alive at the end of follow-up in 2007. *All patients who were excluded from follow-up had died during the study period (28.3% of patients undergoing hemodialysis).

We included controls without kidney disease (absence of ICD-9-CM codes: 580.x–588.x, 250.4x, 274.1x, 283.11, 403.x1, 404.x2, 404.x3, 440.1, 442.1, 447.3, 572.4, 642.1x and 646.2x) from the database in a 1:1 ratio. We matched the controls with patients receiving hemodialysis in terms of age, sex and use of some ulcerogenic medications (nonsteroidal anti-inflammatory drugs [NSAIDs], acetylsalicylic acid [ASA], steroids, ticlopidine and warfarin) and year of enrolment. Medications were identified and classified by use of the National Drug Classification System and the Anatomic Therapeutic Chemical Code, which is an internationally accepted classification system of drugs coordinated by the World Health Organization Collaborating Centre for Drug Statistics Methodology.13

We included a comparison group of 6320 patients with chronic kidney disease (ICD-9-CM codes 582.0, 582.4, 582.8x, 586, 250.4x, 274.1, 403.x1, 404.x2, 404.x3, corresponding to stage 4 and 5 chronic kidney disease) who were identified from the same cohort from the National Health Research Institutes.

We also investigated the covariables age, sex, pre-existing hypertension (ICD-9-CM 401.0–405.99), diabetes mellitus (ICD-9-CM 250.0–250.93), congestive heart failure (ICD-9-CM 428–428.9), coronary artery disease (ICD-9-CM 410–414.9) and cirrhosis of the liver (ICD-9-CM 571.2, 571.5 and 571.6).

Statistical analysis

All data were expressed as frequency (percentage) or mean and standard deviation. We compared parametric continuous data between groups using the Student t test, and we compared categorical data using the χ² test and Yates correction or the Fisher exact test. We performed survival analysis using the Kaplan–Meier method, with significance based on the log-rank test. We calculated survival time as being the time from the date of enrolment in the cohort to the date of admission to hospital because of ulcer bleeding. We performed multiple regression analysis using Cox proportional hazard regression analysis. We defined statistical significance as a two-sided p value of less than 0.05.