Association of Dialysis with the Risks of Cancers

Abstract

Background

To increase the survival span after dialysis in patients with end-stage renal disease (ESRD), identifying specific cancer risks is crucial in the cancer screening of these patients. The aim of this study was to investigate the risks of various cancers in an incident dialysis group in comparison with a non-dialysis group.

Method

We conducted a nationwide cohort study by using data from the Taiwan National Health Insurance Research Database. Patients who initially received long-term dialysis between January 1997 and December 2004, were selected and defined as the dialysis group and were matched with the non-dialysis patients (control group) according to age, sex, and index year. Competing risk analysis was used to estimate cumulative incidence and subdistribution hazard ratios (SHRs) of the first cancer occurrence.

Results

After consideration for the competing risk of mortality, the dialysis group showed a significantly higher 7-year cancer incidence rate than did the control group (6.4%; 95% confidence interval [CI], 6.0%-6.7% vs 1.7%; 95% CI, 1.4%-2.1%; P <0.001).The modified Cox proportional hazard model revealed that the dialysis group had significantly association with increased risks for all cancers (SHR, 3.43; 95% CI, 3.02-3.88). The risk of cancers was dominated in younger and female patients. Specific cancer risks were significantly higher in the dialysis group particularly in the development of oral, colorectal, liver, blood, breast, renal, upper urinary tract, and bladder cancer than in the control group. Multivariable stratified analyses confirmed the association between long-term dialysis and cancer in all subgroups of patients.

Conclusions

Dialysis is associated with a higher risk of cancer in patients with ESRD. However, cancer screening in ESRD population should be a selective approach, based on individual patient health condition and life expectancy.

About the Authors:

Ming Yen Lin

Affiliations Division of Nephrology, Department of Internal Medicine, Kaohsiung Medical University Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan, Faculty of Renal Care, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan, Instrument Technology Research Center, National Applied Research Laboratories, Hsinchu, Taiwan

Mei Chuan Kuo

Affiliations Division of Nephrology, Department of Internal Medicine, Kaohsiung Medical University Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan, Faculty of Renal Care, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan

Chi Chih Hung

Affiliations Division of Nephrology, Department of Internal Medicine, Kaohsiung Medical University Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan, Faculty of Renal Care, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan

Wen Jeng Wu

Affiliations Graduate Institute of Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan, Faculty of Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan, Department of Urology, Kaohsiung Medical University Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan, Kaohsiung Municipal Hsiao-Kang Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan

Li Tzong Chen

Affiliations Division of Gastroenterology, Department of Internal Medicine, Kaohsiung Medical University Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan, National Institute of Cancer Research, National Health Research Institutes, Tainan, Taiwan

Ming Lung Yu

Affiliations Faculty of Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan, Division of Hepatobiliary, Department of Internal Medicine, Kaohsiung Medical University Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan

Chih-Cheng Hsu

Affiliation: Institute of Population Health Sciences, National Health Research Institutes, Miaoli, Taiwan

Chien-Hung Lee

Affiliation: Department of Public Health, College of Health Science, Kaohsiung Medical University, Kaohsiung, Taiwan

Hung-Chun Chen

Affiliations Division of Nephrology, Department of Internal Medicine, Kaohsiung Medical University Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan, Faculty of Renal Care, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan

Shang-Jyh Hwang

* E-mail: [email protected]

Affiliations Division of Nephrology, Department of Internal Medicine, Kaohsiung Medical University Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan, Faculty of Renal Care, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan, Institute of Population Health Sciences, National Health Research Institutes, Miaoli, Taiwan

Introduction

The number of patients with end-stage renal disease (ESRD) worldwide has increased rapidly over the past few decades. By 2010, the estimated ESRD population was over 2 million, and the survival of these patients depends on expensive dialysis treatments and transplantation [1]. Advanced dialysis techniques increase patients’ life span; however, these techniques also increase the risk of complications. In recent years, researchers have increasingly examined specific complications of long-term dialysis, for example, the development of cancer rather than cardiovascular diseases in dialysis patients [2–10]. Studies on the risks of various cancers in dialysis patients are crucial in the development of cancer screening strategies, particularly in countries where the prevalence and incidence of ESRD is high.

Previous studies have showed the higher risks of various cancers in dialysis patients than general population, such as cancers of the liver [7, 9, 10], kidneys [2–4, 7–10], urinary tract [4, 8], bladder [2, 3, 7, 9], thyroid [2, 7, 8], and other organs [2, 8–10]. However, most the studies that reported these associations were either limited by the small scale of the study [8], inclusion of prevalent dialysis patients [10], or estimation of the risk ratio by using the excess standardized incidence ratio [2, 3, 7]. The shortcomings of these approaches were an inability to control the confounding effects on the outcome of interest and failure to obtain accurate comparison of risks for generalization to entire populations.

The purpose of this nationwide cohort study was to evaluate the association of dialysis and the risks of cancers. We used an incident dialysis population matched with the non-dialysis patients according to age, sex, and index year to explore this association.

Materials and Methods

Study population

We conducted a population-based cohort study through the Taiwan National Health Insurance Research Database (NHIRD), which contains claims information and health care data on more than 99% of the population of Taiwan. The accuracy of major disease diagnoses recorded in the NHIRD has been validated [11–13].This study was approved by the ethical review board of Kaohsiung Medical University Hospital (KMUH-IRB-EXEMP-20130014). All the studying processes were obeying the Declaration of Helsinki without violation any ethical principle. Because the patient identifiers in this national dataset were scrambled for research use, the review board waived the requirement for written informed consent.

We retrieved records on all patients who have received long-term dialysis since 1997 and comparable controls, who were matched with the dialysis patients according to age, sex, and index year, The patients who initially received dialysis according to the International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes for chronic renal failure under regular dialysis in the Registry for Catastrophic Illness Patient Database (RCIPD), a subset of the NHIRD (S1 Table). We obtained the initiation date of dialysis treatment for each outpatient by tracking the specific payment codes (S2 Table) from the NHIRD for the treatment and care received during hemodialysis and peritoneal dialysis.

The comparable controls were selected from the Longitudinal Health Insurance Data 2000 (LHID2000), which contains all claims data on 1 000 000 people. The subjects in LHID2000 were randomly sampled from the entire insured population from 1996 to 2000, and no differences in the age and sex distributions between the LHID2000 and the general population were observed. Patients (n = 996 346) who never received dialysis during 1997–2004 and had at least 1 outpatient visit (a total of 6 656 719 records) were considered potential matching controls.

A matching procedure was used to enhance the comparison between the dialysis and control groups. The index year was defined as the year of dialysis for the dialysis group, whereas the index year was the year of an outpatient visit for the control group. Age was calculated from the date of birth to the initiation date of dialysis for the dialysis group and from the date of birth to the date of the outpatient visit for the control group. Because the patients in the dialysis group were significantly older than those in the control group, we matched each patient with the control group according to age (within ± 2 y), sex, and index year by using the existing programs [14]. In each pair, the index date was the initiation date of dialysis in the dialysis group and the corresponding first outpatient visit date in the control group.

Main outcome measurements

Cancer occurrence was determined according to the ICD-9-CM codes from the RCIPD, and the first registry date was considered the date of cancer occurrence. All cancer registries in the database contained either surgical pathological information or a typical image presentation. Details on cancers at various sites, including oral, esophageal, gastric, colorectal, liver, pancreatic, lung cancer, hematological, breast, cervical, prostate, upper urinary tract, and bladder cancers, are shown in S1 Table. Because of the lack of accuracy in determining kidney cancer on the basis of the diagnostic code (ICD-9-CM code 189.0) [15], we used the surgical procedure codes (S2 Table), including those for radical nephrectomy, laparoscopic nephrectomy, and partial nephrectomy, to confirm renal cancer. Death was defined as death during hospitalization or self-discharges from hospitalization, without any outpatient follow up thereafter.

Covariate assessment

In covariate assessment, factors that possibly confound the risk of cancer occurrence were included. We obtained data on patient characteristics such as age, sex, index year (1997–1998, 1999–2000, 2001–2002, 2003–2004), insurance region (north, center, south, east), and urbanization (urban, suburban, rural) [16]. The Charlson comorbidity index (CCI) score was calculated for all diseases listed in a previous study [17]. Potential confounding comorbidities, which are listed using ICD-9-CM codes in S1 Table, included diabetes, hypertension, cerebrovascular disease, ischemic heart disease, and congestive heart disease. These were identified on the basis of ≥2 outpatient visits or ≥1 inpatient visit within 1 year before the index date.

Statistical Analysis

The mortality rate was considerably higher in dialysis patients than in non-dialysis patients. Therefore, the competing risk approach was used to calculate an accurate cancer incidence rate. The death-adjusted cumulative incidence for the marginal probability of cancer was obtained, and the cumulative incidences in the competing risk data were compared using the modified log-rank test [18].

To compare the subdistribution hazard ratios (SHRs) for various sites of cancer occurrence between the dialysis and control group, multivariable analyses were conducted using modified Cox regression hazard models [19]. To represent the competing risk hazard for each event, we adjusted for age, sex, index year, insurance region, urbanization, comorbidities, and the CCI score.

To determine the association between the duration of dialysis and cancer occurrence, we estimated SHR of cancer stratifying the follow-up duration. In addition, we used multivariable stratified analyses of different subgroups to validating these findings. Age-specific and sex-specific SHRs for various sites of cancer were estimated and are reported in S1–S4 Figs.

Data management and analyses were performed using SAS 9.3 (SAS Institute Inc., Cary, North Carolina, USA) software. The cumulative incidence in competing risk analyses was calculated using the cmprsk package of R [20]. A 2-sided P value lower than. 05 was considered statistically significant.

Results

Demographic characteristics of the observed cohort

We first selected 69 917 patients who newly received dialysis during 1997–2004. After patients who received dialysis for fewer than 3 months were excluded, 52 105 dialysis patients were included, and 51 957 (99.7%) age (±2 y)-, sex-, and index-year-matched patients were selected from the non-dialysis population as the control group. We excluded 3 119 matched pairs in which one of the patients developed cancer before the index date. Finally, 47 037 patients matched according to age, sex, and index year were included in the final analysis (Fig 1).

[Figure omitted. See PDF.]

Fig 1. Flow diagram illustrating the selection of patients and controls.

https://doi.org/10.1371/journal.pone.0122856.g001

The demographic characteristics, comorbidities, CCI scores, and follow-up durations of the study cohort are shown in Table 1. The mean (SD) follow-up duration for the dialysis group was 3.0 (2.1) years, and the median (interquartile range [IQR]) was 2.5 (1.2–4.4) years; these durations were significantly lower than those for the control group, for which the mean (SD) was 4.2 (2.2) years and the median (IQR) was 4.0 (2.0–6.0) years. The proportions of patients who lived in the southern insurance region (34.6% vs 27.7%) and urban areas (75.6% vs 56.7%) were higher in the dialysis group than in the control group.

[Figure omitted. See PDF.]

Table 1. Characteristics and prescription drugs in dialysis group and the age-, sex-, and index year-matched control group.

https://doi.org/10.1371/journal.pone.0122856.t001

The presence of comorbidities, namely diabetes, hypertension, cerebrovascular disease, ischemic heart disease, and congestive heart failure, was significantly higher in the dialysis group than in the control group. The mean (SD) CCI score was 3.9 (2.1) for the dialysis group and 3.3 (9.0) for the control group, and the median (IQR) was 3 (2–5) for the dialysis group and 0 (0–0) for the control group.

Seven-year cumulative incidence and risk of cancer

During the observation period, 1863 (4.0%) patients in the dialysis group and 644 (1.4%) patients in the control group developed cancer. The incidence of bladder cancer was the highest in the dialysis group, accounting for approximately 21%, whereas the proportion of this cancer was only 3% in the control group during the follow-up years. The mean age for cancer diagnosis in the dialysis group was 5 years lower than that in the control group, and the number of female patients diagnosed with cancer was higher in the dialysis group than in the control group (S3 Table and S4 Table). Death before cancer occurrence was considered a competing risk. The cumulative incidence of cancer after adjustment for competing mortality was significantly higher in the dialysis group (cumulative incidence, 6.4%; 95% confidence interval [CI], 6.0%–6.7%) than in the control group (cumulative incidence, 1.7%; 95% CI, 1.4%–2.1%; P < 0.001) (Fig 2). The dialysis group had a significantly higher risk of all cancers (SHR, 3.43; 95% CI, 3.02–3.88; P < 0.001) than the control group did. Furthermore, these cancers particularly occurred in the oral cavity (SHR, 2.50; 95% CI, 1.43–4.39; P = 00.001), colon (SHR, 2.34; 95% CI, 1.62–3.39; P <0.001), liver (SHR, 1.88; 95% CI, 1.35–2.61; P <0.001), blood (SHR, 2.30; 95% CI, 1.20–4.42; P = .01), breast (SHR, 4.21; 95% CI, 2.51–7.06; P < 0.001), kidneys (SHR, 40.31; 95% CI, 13.35–121.75; P < 00.001), upper urinary tract (SHR, 61.33; 95% CI, 18.58–202.45; P <00.001), and bladder (SHR, 41.95; 95% CI, 25.10–70.11; P <00.001) (Table 2).The risk of developing cancer in the dialysis group was consistently higher in all follow-up periods than that in the control group. The proportion of specific cancer in patient with cancer showed a great difference between groups during the first year follow up. Cancers of urinary system (kidney, upper urinary tract, and bladder cancers) accounted for considerable proportion of cancers in dialysis group (34.1%) compared to control group (5.3%) (data not shown). The highest risk of cancer in the dialysis group was observed ≥4 years after dialysis (SHR, 15.2; 95% CI, 10.8–21.5; P < 0.001) (Table 3).

[Figure omitted. See PDF.]

Fig 2. Cumulative incidence for cancer and mortality.

Data were compiled after consideration for competing mortality. The cumulative rate between dialysis and control group were calculated using Modified Kaplan-Meier and Gray methods [30] and test their difference between groups by modified log-rank test. The Modified log-rank P value for comparing the cumulative incidence for cancer in dialysis and control group is less than 0.001.

https://doi.org/10.1371/journal.pone.0122856.g002

[Figure omitted. See PDF.]

Table 2. Incident density of various sites of cancers.

https://doi.org/10.1371/journal.pone.0122856.t002

[Figure omitted. See PDF.]

Table 3. Cancer Occurrence in Relation to Dialysis, by Follow-up Duration.

https://doi.org/10.1371/journal.pone.0122856.t003

Stratified analysis

We confirmed the increased risks of all cancers in all stratified analyses (Fig 3). Higher risks were observed in patients aged <50 years (SHR, 6.91; 95% CI, 4.95–9.64; P < 0.001) and 50–59 (SHR, 5.06; 95% CI, 3.88–6.59; P < 0.001) than in those aged ≥60 years (SHR, 2.46; 95% CI, 2.09–2.89; P <00.001). The risks were higher in female patients (SHR, 4.44; 95% CI, 3.70–5.32; P <00.001) than in male patients (SHR, 2.7; 95% CI, 2.26–3.22; P < 00.001). The risk of overall cancer remained significantly higher in the subgroups of the dialysis group stratified according to various comorbidities, including diabetes (SHR, 4.58; 95% CI, 4.02–5.23; P < 0.001), hypertension (SHR, 3.05; 95% CI, 2.35–3.97; P < 0.001), cerebrovascular disease (SHR, 2.66; 95% CI, 1.51–4.71; P < 0.001), ischemic heart disease (SHR, 3.23; 95% CI, 2.02–5.17; P < 0.001), and congestive heart disease (SHR, 2.88; 95% CI, 1.36–6.11; P < 0.001).

[Figure omitted. See PDF.]

Fig 3. Multivariable stratified analyses for the association between dialysis and cancer.

Abbreviation: SHR, subdistribution hazard ratio; CI, confident interval.

https://doi.org/10.1371/journal.pone.0122856.g003

The risks of cancers at various sites stratified according age (using 60 y as the cutoff value) and sex showed marked differences in oral and gastric cancers in patients aged <60 years than in those aged >60 years, and a significantly higher incidence of upper urinary tract and bladder cancers was observed in female patients than in male patients (S1–S4 Figs).

Discussion

Through our large-scale nationwide cohort study and the competing risk analysis, we clearly showed that the risk of cancer in the Taiwanese incident dialysis population is increased. We also identified site-specific cancers, such as oral, colorectal, liver, breast, upper urinary tract, renal, bladder, and blood cancers, which were frequently observed in this population. Younger and female patients were at a higher risk than older and male patients, respectively.

The relationship between dialysis and the high risk of cancer has been widely investigated in several countries [2, 3]. However, an increasing risk of specific cancers in dialysis patients remains substantially inconclusive. An international collaborative study from the United States, Europe, Australia, and New Zealand reported an increased risk of cancers of the bladder, urinary tract, thyroid, and other endocrine organs in dialysis patients [3]. In Japanese dialysis patients, the excessive risk was observed on renal cancer, but not on upper urinary tract cancer [4]. Even within single country, studies that reported significant differences in specific cancer incidences in dialysis patients compared with the general population were not in complete agreement [7, 8, 10]. The discrepancy may be caused by various factors involved in the method for estimating the risk. The standard incidence ratio was often used in these studies, but this could be a problem if the ratio is not effectively controlled for confounding effects on the outcome of interest. Few studies have used an appropriate control cohort to identify the relative risk of cancer in incident dialysis patients. The traditional Cox regression model can be biased because the competing event, such as death, was significantly higher in the dialysis group than the general control group. Our study had the advantages of matched controls for age, sex, and index year; effective control of the confounding effects; and the use of the competing risk model for SHR estimation, potentially strengthening the reliability of the results.

The causes for the increased risks of specific cancers in the dialysis population have not been elucidated adequately. This could be attributed to differences in carcinogens to various cancers among countries. Aristolochia species of plants were either frequently mixed or misused in Southeastern Europe and Asia [21]. Plants containing aristolochic acid were considered a causative agent of renal damage and urinary tract cancer. This is a possible reason for the increased incidence of urinary tract cancer in Asian dialysis patients. The specific cancers including kidney, upper urinary tract, and bladder cancers showed high proportions in dialysis group during the first year follow up. These cancers occurred and accompanied with kidney disease even before dialysis, which implied certain causal factors linking the renal damage and tumorogenesis. The mechanisms may have a long-term period of development covering the whole process of kidney disease progression. From the similar results in current researches, young age and female are risk factors for aristolochic acid induced renal disease and cancers [22–24]. It suggests that cancer screening in patient with renal disease should be selective. The higher risk of oral, liver, colorectal, breast, and blood cancers shows in dialysis group compared to control group in the study, but the reason has not been well clarified. One large study considers viruses may play a key role on human papillomavirus-related- cancer such as vulvovaginal cancer in patient with ESRD [25]. The higher prevalence of Hepatitis C viral infections in patient with advance chronic kidney disease in Taiwan had been reported in our previous study [26]. It is likely that Hepatitis C virus may be involved in the progression and development of liver cancer in dialysis patients. Furthermore, chronic infection and inflammation, impaired functioning of the immune system, nutrition deficiency, and shortage of DNA repair mechanisms, which frequently appeared in dialysis patient [27, 28], may also contribute the tumorigenesis of these cancers.

Our results are consistent with those of related studies [2, 7, 9], which reported that the risk of cancer was greater in younger dialysis patients. One possible explanation for this is a stronger inflammatory response in young patients upon external stimulation. This response is likely to trigger tumorigenesis; however, the underlying mechanism requires further elucidation. Female dialysis patients have higher cancer risk than male patients; this observation is consistent with the early findings of other studies [8, 9]. Previous studies have reported that females are at higher risk for kidney disease and urinary tract cancer caused by Chinese herbs than are males [29–32]. These results suggest that sex, at least partly, influences the initiation of renal disease and urinary tract cancer. However, further research should be conducted to clarify the association of sex with cancerogenesis in dialysis patients. This study had several strengths. First, the competing risk for death approach substantially increased the ability to identify the risk of specific cancers. Second, the use of age-, sex-, and index year-matched control groups increased the comparability and reduced time-related bias of the observations. Third, patients in the study group had incident dialysis and were followed up for the development of incident cancer after dialysis. The time sequence between these two diseases obviously improves through this study design. Fourth, we stratified our population in subgroup analyses, and the results were similar, supporting the hypothesis that the risk of cancer in the dialysis population is increased.

Nevertheless, our study had certain limitations. First, it was impossible to confirm the cancer diagnoses according to histopathology reports in the NHIRD. However, the accuracy of diagnoses in the RCIPD can be considered valid because pathological and/or cytological evidence is required for approval. Second, we were unable to obtain major information on the patients’ history regarding smoking and other lifestyle choices. This limitation has been noted in previous observational studies that analyzed claim data, and the studies revealed a compromise between sample sizes and comprehensive collection of confounding factors. These results were obtained despite sex being assumed, in part, as a surrogate variable for smoking for adjustment in the models, as mentioned in a previous study [33]. Third, although an increase in the risk of all and site-specific cancers was observed in the dialysis population, these results must be explained in greater detail. Dialysis patients may be more aware of cancer because they are more frequently monitored for cancer development by medical staff, frequently visit the hospital for dialysis, and have easy access to cancer screening when abnormal symptoms arise. Finally, although the comparability between dialysis group and control group was substantially improved through matching their age and sex, we still cannot rule out the possibility that residual confounders could influence on our risk estimation.

Conclusion

In summary, we proved that the risks of all cancers and site-specific cancers (i.e., oral, colorectal, liver, breast, upper urinary tract, kidneys, bladder, and blood cancers) are increased in the incident dialysis population. Younger age and female sex increase susceptible to cancers. These findings suggest that cancer screening in ESRD population should be a selective approach, based on individual patient health condition and life expectancy.

Supporting Information

[Figure omitted. See PDF.]

S1 Fig. The association between dialysis and various sites of cancers in aged younger than 60 years.

https://doi.org/10.1371/journal.pone.0122856.s001

(TIF)

S2 Fig. The association between dialysis and various sites of cancers in aged equal and older than 60 years.

https://doi.org/10.1371/journal.pone.0122856.s002

(TIF)

S3 Fig. The association between dialysis and various sites of cancers in male.

https://doi.org/10.1371/journal.pone.0122856.s003

(TIF)

S4 Fig. The association between dialysis and various sites of cancers in female.

https://doi.org/10.1371/journal.pone.0122856.s004

(TIF)

S1 Table. The corresponding ICD-9-CM codes for the diagnosis of disease in the study.

https://doi.org/10.1371/journal.pone.0122856.s005

(DOCX)

S2 Table. The payment codes for the clinical treatment providing by Taiwan National Health Insurance.

https://doi.org/10.1371/journal.pone.0122856.s006

(DOCX)

S3 Table. Diagnosed age and proportion of female in dialysis patient with occurring cancer.

https://doi.org/10.1371/journal.pone.0122856.s007

(DOCX)

S4 Table. Diagnosed age by sex in dialysis patient with occurring cancer.

https://doi.org/10.1371/journal.pone.0122856.s008

(DOCX)

Acknowledgments

Parts of our results were presented in a poster at ASN Kidney Week 2007 in San Francisco, California.

Author Contributions

Conceived and designed the experiments: MYL WJW LTC MLY SJH. Analyzed the data: MYL C.C. Hung C.C. Hsu MCK CHL. Wrote the paper: MYL HCC SJH.

Citation: Lin MY, Kuo MC, Hung CC, Wu WJ, Chen LT, Yu ML, et al. (2015) Association of Dialysis with the Risks of Cancers. PLoS ONE 10(4): e0122856. https://doi.org/10.1371/journal.pone.0122856

References

1. Eggers PW. Has the incidence of end-stage renal disease in the USA and other countries stabilized? Curr Opin Nephrol Hypertens. 2011 May;20(3):241–5. pmid:21422925

2. Maisonneuve P, Agodoa L, Gellert R, Stewart JH, Buccianti G, Lowenfels AB, et al. Cancer in patients on dialysis for end-stage renal disease: an international collaborative study. Lancet. 1999 Jul 10;354(9173):93–9. pmid:10408483

3. Stewart JH, Buccianti G, Agodoa L, Gellert R, McCredie MR, Lowenfels AB, et al. Cancers of the kidney and urinary tract in patients on dialysis for end-stage renal disease: analysis of data from the United States, Europe, and Australia and New Zealand. J Am Soc Nephrol. 2003 Jan;14(1):197–207. pmid:12506152

4. Satoh S TN , Habuchi T, Ishiyama T, Seimo K, Kato T. Renal cell and transitional cell carcinoma in a Japanese population undergoing maintenance dialysis. J Urol. 2005;174(5):1749–53. pmid:16217276

5. Chang CH, Yang CM, Yang AH. Renal diagnosis of chronic hemodialysis patients with urinary tract transitional cell carcinoma in Taiwan. Cancer. 2007 Apr 15;109(8):1487–92. pmid:17330839

6. Stewart JH, Vajdic CM, van Leeuwen MT, Amin J, Webster AC, Chapman JR, et al. The pattern of excess cancer in dialysis and transplantation. Nephrol Dial Transplant. 2009 Oct;24(10):3225–31. pmid:19589786

7. Lin HF, Li YH, Wang CH, Chou CL, Kuo DJ, Fang TC. Increased risk of cancer in chronic dialysis patients: a population-based cohort study in Taiwan. Nephrol Dial Transplant. 2012 Apr;27(4):1585–90. pmid:21862456

8. Chung CJ, Huang CY, Tsai HB, Muo CH, Chung MC, Chang CH, et al. Sex Differences in the Development of Malignancies among End-Stage Renal Disease Patients: A Nationwide Population-Based Follow-Up Study in Taiwan. PLoS One. 2012;7(9):e44675. pmid:22957098

9. Liang JA, Sun LM, Yeh JJ, Sung FC, Chang SN, Kao CH. The association between malignancy and end-stage renal disease in Taiwan. Japanese journal of clinical oncology. 2011 Jun;41(6):752–7. pmid:21515610

10. Shebl FM, Warren JL, Eggers PW, Engels EA. Cancer risk among elderly persons with end-stage renal disease: a population-based case-control study. BMC nephrology. 2012;13:65. pmid:22834953

11. Lin CC, Lai MS, Syu CY, Chang SC, Tseng FY. Accuracy of diabetes diagnosis in health insurance claims data in Taiwan. Journal of the Formosan Medical Association = Taiwan yi zhi. 2005 Mar;104(3):157–63. pmid:15818428

12. Cheng CL, Kao YH, Lin SJ, Lee CH, Lai ML. Validation of the National Health Insurance Research Database with ischemic stroke cases in Taiwan. Pharmacoepidemiology and drug safety. 2011 Mar;20(3):236–42. pmid:21351304

13. Hsieh CY, Chen CH, Li CY, Lai ML. Validating the diagnosis of acute ischemic stroke in a National Health Insurance claims database. Journal of the Formosan Medical Association = Taiwan yi zhi. 2013 Oct 18.

14. Grandits G, Neuhaus J. Using SAS to Perform Individual Matching in Design of Case-Control Studies. SAS Global Forum 2010; Seattle, Washington SAS Institute Inc.; 2010.

15. Lin MY, Huang CN, Wu WJ, Chen LT, Chen HC, Hwang SJ. Overestimation of the incident rate of renal cancer in Registry Catastrophic Illness Patient Data-A population base study. Scientific Congress & Annual Meeting, 2012 Taiwan Society of Nephrology; December 9; Taipei, Taiwan2013. p. 308.

16. Chien IC, Hsu JH, Lin CH, Bih SH, Chou YJ, Chou P. Prevalence of diabetes in patients with schizophrenia in Taiwan: a population-based National Health Insurance study. Schizophr Res. 2009 Jun;111(1–3):17–22. pmid:19409756

17. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992 Jun;45(6):613–9. pmid:1607900

18. Gray R. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat. 1988;16:1141–54.

19. Fine JP GR. A proportional hazard model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496–509.

20. Gray B. cmprsk: Subdistribution Analysis of Competing Risks. Available from: http://cran.r-project.org/web/packages/cmprsk/index.html. Accessed 2014 July 20.

21. Debelle FD, Vanherweghem JL, Nortier JL. Aristolochic acid nephropathy: a worldwide problem. Kidney Int. 2008 Jul;74(2):158–69. pmid:18418355

22. Wang SM, Lai MN, Chen PC, Pu YS, Lai MK, Hwang JS, et al. Increased upper and lower tract urothelial carcinoma in patients with end-stage renal disease: a nationwide cohort study in Taiwan during 1997–2008. BioMed research international. 2014;2014:149750. pmid:25025033

23. Chen CH, Dickman KG, Huang CY, Moriya M, Shun CT, Tai HC, et al. Aristolochic acid-induced upper tract urothelial carcinoma in Taiwan: clinical characteristics and outcomes. International journal of cancer Journal international du cancer. 2013 Jul;133(1):14–20. pmid:23292929

24. Chen CH, Dickman KG, Moriya M, Zavadil J, Sidorenko VS, Edwards KL, et al. Aristolochic acid-associated urothelial cancer in Taiwan. Proceedings of the National Academy of Sciences of the United States of America. 2012 May 22;109(21):8241–6. pmid:22493262

25. Skov Dalgaard L, Fassel U, Ostergaard LJ, Jespersen B, Schmeltz Sogaard O, Jensen-Fangel S. Risk of human papillomavirus-related cancers among kidney transplant recipients and patients receiving chronic dialysis—an observational cohort study. BMC nephrology. 2013;14:137. pmid:23834996

26. Lee JJ, Lin MY, Yang YH, Lu SN, Chen HC, Hwang SJ. Association of hepatitis C and B virus infection with CKD in an endemic area in Taiwan: a cross-sectional study. American journal of kidney diseases: the official journal of the National Kidney Foundation. 2010 Jul;56(1):23–31.

27. Kato S, Chmielewski M, Honda H, Pecoits-Filho R, Matsuo S, Yuzawa Y, et al. Aspects of immune dysfunction in end-stage renal disease. Clin J Am Soc Nephrol. 2008 Sep;3(5):1526–33. pmid:18701615

28. Vamvakas S, Bahner U, Heidland A. Cancer in end-stage renal disease: potential factors involved-editorial. Am J Nephrol. 1998;18(2):89–95. pmid:9569948

29. Yang CS, Lin CH, Chang SH, Hsu HC. Rapidly progressive fibrosing interstitial nephritis associated with Chinese herbal drugs. American journal of kidney diseases: the official journal of the National Kidney Foundation. 2000 Feb;35(2):313–8.

30. Chang CH, Wang YM, Yang AH, Chiang SS. Rapidly progressive interstitial renal fibrosis associated with Chinese herbal medications. Am J Nephrol. 2001 Nov-Dec;21(6):441–8. pmid:11799260

31. Yang HY, Wang JD, Lo TC, Chen PC. Increased mortality risk for cancers of the kidney and other urinary organs among Chinese herbalists. Journal of epidemiology / Japan Epidemiological Association. 2009;19(1):17–23. pmid:19164871

32. Stefanovic V, Radovanovic Z. Balkan endemic nephropathy and associated urothelial cancer. Nature clinical practice Urology. 2008 Feb;5(2):105–12. pmid:18259188

33. Lai MN, Wang SM, Chen PC, Chen YY, Wang JD. Population-based case-control study of Chinese herbal products containing aristolochic acid and urinary tract cancer risk. Journal of the National Cancer Institute. 2010 Feb 3;102(3):179–86. pmid:20026811