Age-related risk of mortality from bladder cancer in diabetic patients: A 12-year follow-up of a national cohort in Taiwan

Abstract

Background. To compare bladder cancer mortality between diabetic patients and the general population.

Materials and methods. Yearly sex-specific mortality rates for age 25–64, 65–74, and ≥75 years in Taiwanese general population for 1995–2006 were calculated; 113,347 diabetic men and 131,573 diabetic women aged ≥25 years recruited in 1995–1998 were followed prospectively.

Results. In the general population, 4,943 men and 2,291 women died of bladder cancer, and aging was associated with increased risk. Although the trend of crude mortality was increasing in either sex, the trend of age-standardized rates had been steady. The average crude and age-standardized mortality rates were 5.35 and 5.98 (per 100,000 population), respectively, for men; and were 2.63 and 3.02 for women. A total of 224 diabetic men and 126 diabetic women died of bladder cancer with crude mortality of 26.0 and 11.9 per 100,000 person-years, respectively. The relative risk of bladder cancer mortality (95% confidence interval) for diabetic patients was 2.18 (1.75–2.72), 2.50 (2.06–3.04), and 5.95 (4.57–7.74), in men, and 1.34 (0.96–1.89), 2.48 (1.92–3.19), and 7.44 (5.46–10.15), in women, for ages ≥75, 65–74, and 25–64 years, respectively.

Conclusions. Diabetic patients had a higher risk of bladder cancer mortality, which is more remarkable in the younger population.

• Bladder cancer
• diabetes mellitus
• mortality
• secular trend
• Taiwan

Introduction

A meta-analysis of 16 studies (7 case-control, 3 cohort, and 6 cohort of diabetic patients with external population comparisons) published in 2006 concluded that diabetes mellitus (probably type 2 diabetes mellitus because of the exclusion of diabetes diagnosed at an early age of ≤30 years) was associated with bladder cancer with a summary relative risk of 1.24 (95% confidence interval 1.08–1.43) [1]. While analyzing by study design in that meta-analysis, it was noted that diabetes was significantly associated with bladder cancer in case-control studies and cohort studies, but not in cohort studies of diabetic patients using population controls (relative risk = 1.01, 95% confidence interval 0.91–1.12) [1]. As pointed out by the authors, publication bias with missing of studies of small sample size and reporting low relative risk is possible, which may result in overestimation of the relationship between diabetes mellitus and bladder cancer [1].

Most of the studies included in the meta-analysis were conducted in Western countries, and only 1 among the 16 studies was conducted in the Asian population in Korea [1]. This Korean study did show a significantly higher risk of bladder cancer incidence, but not mortality, associated with diabetes mellitus with an age-adjusted odds ratio of 1.32 (95% confidence interval 1.10–1.57) in men. However, no data were given for women [2].

Key messages

• The association between diabetes and bladder cancer is rarely studied in the Asian people.

• This study evaluated the trend of bladder cancer mortality in the Taiwanese general population and the mortality rate ratios between diabetic patients and the general population.

• Our findings suggest that the secular trend of bladder cancer mortality in the Taiwanese general population is steady but a significantly higher risk of mortality from bladder cancer in diabetic patients is observed.

A later case-control study evaluating the association between diabetes mellitus and cancer risk in the Japanese did not confirm such an association in men, and they were not able to estimate the association in women due to the lack of bladder cancer cases [3]. A common limitation in both the Korean and the Japanese studies is the neglect of the subtypes of diabetes mellitus. Furthermore the association in different age groups is worthy of investigation but had not been looked at. The prevalence [4], [5] and incidence [6] of diabetes (especially type 2) have been increasing during the past decades in Taiwan. According to the population-based Taiwanese Survey on Hypertension, Hyperglycemia, and Hyperlipidemia conducted in 2002, the current crude prevalence of diabetes in the population aged ≥15 years is 7.5%, with a rate of 6.6% after standardization to the World Health Organization standard population for the years 2000–2025 [7]. A clarification of the association between diabetes and bladder cancer is important for disease prevention and management. Therefore, the purpose of this study was to evaluate, based on different strata of age and sex, 1) the secular trends of mortality from bladder cancer in the general population of Taiwan; and 2) the association between diabetes mellitus and mortality from bladder cancer in all diabetic patients and in patients with type 2 diabetes mellitus using the general population as controls.

Materials and methods

Mortality from bladder cancer in the general population

The study was approved and supported by the Department of Health, Executive Yuan, Taiwan. Because bladder cancer is actually very rare among the young generation, and peak incidence may be as old as ≥85 years in either the USA [8] or Taiwan [9], we chose to analyze the data for those aged 25 years or older, and the following age groups were applied: 25–64, 65–74, and ≥75 years old. The data of mortality from bladder cancer were abstracted from vital statistics reported annually by the government of Taiwan. The causes of death were coded according to the ninth revision of the International Classification of Diseases (ICD-9) since 1981. Bladder cancer was diagnosed based on the ICD-9 code of 188.

The crude and standardized mortality rates (per 100,000 persons) were calculated for both men and women aged ≥25 years. The standard population used for such standardization was the average of 1995–2006 mid-year population of Taiwan. The age- and sex-specific mortality rates were then calculated by dividing the total number of deaths ascribed to bladder cancer by the mid-year population in the specific age and sex groups from the years 1995–2006. Average mortality rates during the period of 1995–2006 for specific age and sex were also calculated by dividing the average numbers of deaths due to bladder cancer within these 12 years by the average mid-year population of the specific age and sex within these 12 years.

Linear regression was used to evaluate whether the trends of mortality from bladder cancer significantly changed with regard to calendar years, where the mortality rate was the dependent variable and the calendar year was the independent variable.

Mortality from bladder cancer in the diabetic patients

Since March 1995 a compulsory and universal system of health insurance which covered >96% of the total population, the so-called National Health Insurance (NHI), was implemented in Taiwan. From 1995 to 1998 a cohort of 256,036 diabetic patients (labeled as ‘the original cohort’ in this study) using the NHI was established as described in detail elsewhere [10]. The diagnosis of diabetes mellitus was defined by the ICD-9 code 250 or the A-code (abridged code) of A181.

All patients were followed until the end of the year 2006. The vital status and the date and cause of death were obtained by matching the computerized data file of the National Register of Deaths using the unique identification number. Mortality rates were computed using a person-years denominator. The person-years of follow-up for each patient were calculated as the duration from the date of recruitment until the end of 2006 for those who were alive or to the date of death for those who died. Age- and sex-specific mortality rates and the mortality rate ratios for the diabetic patients versus the general population of the same sex were calculated. The age- and sex-specific mortality rate ratios were calculated using the average mortality rates within the 12 years in the general population as referents. Because we noticed that the yearly mortality from bladder cancer in the general population slightly fluctuated during the 12-year period, the mortality rate ratios were also calculated using the highest age- and sex-specific mortality rates within the calendar years in the general population as comparison groups.

For sensitivity analyses, we also calculated the mortality rates and the mortality rate ratios in the subgroup of patients who had been interviewed with a base-line questionnaire, as described in detail elsewhere [6], [11]. The total number of diabetic patients interviewed was 93,484, and among them 91,665 patients (42,260 men and 49,405 women) were aged ≥25 years (labeled as ‘subcohort diabetic patients aged ≥25 years’). To evaluate whether the association was found in patients with type 2 diabetes mellitus, the mortality rates and mortality rate ratios were also calculated after excluding patients of type 1 diabetes mellitus based on the following criteria: diabetic ketoacidosis at the onset of diabetes mellitus, or the need for insulin injection within 1 year after diagnosis of diabetes. There were 40,820 diabetic men and 47,874 diabetic women after such exclusion, and they were labeled as ‘subcohort type 2 diabetic patients aged ≥25 years’. Because there were only 1,440 diabetic men and 1,531 diabetic women excluded with possible diagnosis of type 1 diabetes mellitus and among them only 2 men and 2 women died of bladder cancer, we did not analyze the association with these patients.

Because arsenic exposure is associated with the development of bladder cancer [12], and there are two counties (Ilan and Chiayi) afflicted with environmental exposure to high levels of arsenic from drinking well-water in Taiwan [12–14], the above sensitivity analyses were also performed after excluding 2,488 and 5,163 diabetic patients who reported to live in Ilan and Chiayi, respectively.

Results

A total of 4,943 men and 2,291 women aged 25 years or older in the general population died of bladder cancer from 1995 to 2006 according to data obtained from vital statistics. Table I shows the crude, standardized, and age-specific mortality rates (per 100,000) from bladder cancer during the period in the general population and the age-specific average mortality rates within this 12-year period in separate sexes. Although the trend of the crude rates seemed to increase in either sex, the trend of the standardized rates was not statistically significant. Basically the age-specific mortality rates increased dramatically with increasing age in any specific calendar year in either men or women. The highest age-specific mortality rates were observed in years 1997, 2005, and 1996, respectively, for men; and in the years 1995, 1998, and 2004, respectively, for women, for the age group of 25–64, 65–74, and ≥75 years. Except for the age group of 25–64 years in men showing a significant decrease in the trend of mortality from bladder cancer, none of the other age groups showed a significant change in the secular trend in either sex.

Table I.  The mortality rate (per 100,000) from bladder cancer by sex and age in the general population of Taiwan during the years 1995–2006 and the average mortality rate within these 12 years.

AGE (YR)		1995	1996	1997	1998	1999	2000	2001	2002	2003	2004	2005	2006	Average	P - value
Men
25–64	n	86	85	91	66	80	79	61	82	75	74	75	65	77	0.0028
	MYP	5,480,406	5,556,750	5,634,555	5,724,144	5,813,705	5,899,488	5,999,779	6,103,567	6,201,093	6,304,555	6,405,632	6,507,085	5,969,230
	MR	1.57	1.53	1.62	1.15	1.38	1.34	1.02	1.34	1.21	1.17	1.17	1.00	1.28
65–74	n	115	114	122	137	137	129	145	98	113	115	147	124	125	0.4677
	MYP	633,594	651,416	661,022	666,575	668,093	663,768	654,953	646,223	638,278	631,604	628,407	629,220	647,763
	MR	18.15	17.50	18.46	20.55	20.51	19.43	22.14	15.17	17.70	18.21	23.39	19.71	19.25
≥75	n	139	171	158	182	203	210	190	211	246	276	284	258	211	0.2881
	MYP	239,815	256,536	275,487	295,092	315,060	338,169	363,853	389,650	415,976	438,526	466,054	488,446	356,889
	MR	57.96	66.66	57.35	61.68	64.43	62.10	52.22	54.15	59.14	62.94	60.94	52.82	59.03
≥25	CMR	5.35	5.72	5.65	5.76	6.18	6.06	5.64	5.48	5.98	6.31	6.75	5.86	5.35	0.0450
	SMR	5.98	6.35	6.02	5.98	6.30	6.01	5.46	5.20	5.58	5.78	6.14	5.23	5.98	0.0508
Women
25–64	n	36	35	35	22	33	34	32	29	34	40	39	41	34	0.8938
	MYP	5,289,084	5,387,872	5,485,202	5,588,887	5,689,657	5,784,728	5,892,585	6,003,528	6,107,151	6,216,807	6,329,650	6,450,148	5,852,108
	MR	0.68	0.65	0.64	0.39	0.58	0.59	0.54	0.48	0.56	0.64	0.62	0.64	0.58
65–74	n	50	44	47	61	51	59	58	52	56	68	62	56	55	0.2871
	MYP	479,129	496,486	513,451	532,303	551,983	571,732	589,566	607,053	624,888	643,246	663,676	683,524	579,753
	MR	10.44	8.86	9.15	11.46	9.24	10.32	9.84	8.57	8.96	10.57	9.34	8.19	9.54
≥75	n	72	78	86	82	84	89	101	93	110	147	139	136	101	0.3731
	MYP	244,166	256,893	271,872	287,173	302,716	319,719	338,961	359,404	380,367	397,931	425,506	450,726	336,286
	MR	29.49	30.36	31.63	28.55	27.75	27.84	29.80	25.88	28.92	36.94	32.67	30.17	30.16
≥25	CMR	2.63	2.56	2.68	2.57	2.57	2.73	2.80	2.50	2.81	3.51	3.24	3.07	2.63	0.0083
	SMR	3.02	2.92	3.00	2.80	2.73	2.83	2.86	2.48	2.72	3.34	2.99	2.77	3.02	0.8640

P-value by linear regression testing the trend during 1995–2006.

CMR = crude mortality rate; n =number of deaths from bladder cancer; MR = mortality rate; MYP = mid-year population; SMR = standardized mortality rate using the average general population in Taiwan during the years 1995–2006 as standard population.

The age-specific mortality rates from bladder cancer in the diabetic patients and their mortality rate ratios compared to the general population in Taiwan are shown in Table II for men and in Table III for women. A total of 113,347 diabetic men and 131,573 diabetic women aged 25 years or older in the original cohort at recruitment were followed. Among them 224 diabetic men and 126 diabetic women died of bladder cancer with a calculated mortality rate of 26.0 and 11.9 per 100,000 person-years, respectively (Tables II and IIII). Except for the age group of ≥75 years in women, all the age-specific mortality rate ratios were significant in the original cohort in either sex while using either the average or the highest mortality rates of the general population as referents (Tables II and III).

Table II.  Age-specific mortality rates from bladder cancer in the diabetic men and their mortality rate ratios compared to men in the general population of Taiwan.

AGE (YEARS)	n	N	Person-years	Mortality rate (per 100,000 person-years)	Mortality rate ratio (compared to average mortality rate of the general population)	Mortality rate ratio (compared to the highest mortality rate of the general population)
I. Original cohort analysis
Diabetic men aged ≥25 years
25–64	45	70,974	589,654.3	7.63	5.95 (4.57–7.74)	4.73 (3.42–6.53)
65–74	102	30,704	211,811.1	48.16	2.50 (2.06–3.04)	2.06 (1.61–2.64)
≥75	77	11,669	59,761.6	128.85	2.18 (1.75–2.72)	1.93 (1.48–2.52)
II. Sensitivity analyses
Subcohort diabetic men aged ≥25 years
25–64	13	26,291	185,345	7.01	5.47 (3.36–8.89)	4.34 (2.55–7.39)
65–74	44	12,395	76,349.6	57.63	2.99 (2.25–3.98)	2.46 (1.78–3.41)
≥75	21	3,574	17,710.4	118.57	2.01 (1.32–3.06)	1.78 (1.14–2.78)
Subcohort diabetic men aged ≥25 years, excluding those living in arseniasis-endemic counties
25–64	13	24,251	170,442.3	7.63	5.94 (3.67–9.62)	4.72 (2.79–8.00)
65–74	40	11,474	70,485.0	56.75	2.95 (2.19–3.98)	2.43 (1.73–3.40)
≥75	21	3,317	16,302.9	128.81	2.18 (1.44–3.32)	1.93 (1.24–3.02)
Subcohort type 2 diabetic men aged ≥25 years
25–64	13	25,321	178,819.1	7.27	5.67 (3.49–9.20)	4.50 (2.65–7.65)
65–74	43	12,051	74,413.3	57.79	3.00 (2.25–4.01)	2.47 (1.78–3.43)
≥75	20	3,448	17,171.3	116.47	1.97 (1.28–3.04)	1.75 (1.11–2.76)
Subcohort type 2 diabetic men aged ≥25 years, excluding those living in arseniasis-endemic counties
25–64	13	23,352	164,397.7	7.91	6.16 (3.82–9.95)	4.90 (2.90–8.27)
65–74	39	11,153	68,701.0	56.77	2.95 (2.18–3.99)	2.43 (1.72–3.42)
≥75	20	3,193	15,775.8	126.78	2.15 (1.40–3.30)	1.90 (1.21–3.00)

n =number of deaths from bladder cancer.

N = Number in study.

Table III.  Age-specific mortality rates from bladder cancer in the diabetic women and their mortality rate ratios compared to women in the general population of Taiwan.

Age (years)	n	N	Person-years	Mortality rate (per 100,000 person-years)	Mortality rate ratio (compared to average mortality rate of the general population)	Mortality rate ratio (compared to the highest mortality rate of the general population)
I. Original cohort analysis
Diabetic women aged ≥25 years
25–64	31	81,066	713,590.5	4.34	7.44 (5.46–10.15)	6.38 (4.20–9.69)
65–74	61	35,194	258,155.6	23.63	2.48 (1.92–3.19)	2.06 (1.46–2.92)
≥75	34	15,313	83,905.5	40.52	1.34 (0.96–1.89)	1.10 (0.76–1.59)
II. Sensitivity analyses
Subcohort diabetic women aged ≥25 years
25–64	11	29,485	214,239.3	5.13	8.79 (5.36–14.43)	7.54 (4.25–13.38)
65–74	23	15,029	96,058.4	23.94	2.51 (1.68–3.75)	2.09 (1.31–3.34)
≥75	17	4,891	25,384.8	66.97	2.22 (1.39–3.54)	1.81 (1.11–2.97)
Subcohort diabetic women aged ≥25 years, excluding those living in arseniasis-endemic counties
25–64	11	27,053	196,094.8	5.61	9.61 (5.90–15.65)	8.24 (4.69–14.49)
65–74	22	13,534	86,165.2	25.53	2.68 (1.78–4.02)	2.23 (1.39–3.58)
≥75	16	4,457	22,816.3	70.13	2.33 (1.44–3.75)	1.90 (1.14–3.15)
Subcohort type 2 diabetic women aged ≥25 years
25–64	11	28,548	207,570.8	5.30	9.08 (5.55–14.86)	7.79 (4.40–13.77)
65–74	22	14,570	93,223.1	23.60	2.47 (1.64–3.73)	2.06 (1.28–3.32)
≥75	16	4,756	24,716.7	64.73	2.15 (1.33–3.47)	1.75 (1.05–2.92)
Subcohort type 2 diabetic women aged ≥25 years, excluding those living in arseniasis-endemic counties
25–64	11	26,175	189,808.0	5.80	9.93 (6.11–16.12)	8.51 (4.86–14.93)
65–74	21	13,115	83,602.3	25.12	2.63 (1.73–4.00)	2.19 (1.35–3.55)
≥75	15	4,334	22,223.9	67.49	2.24 (1.36–3.67)	1.83 (1.08–3.08)

n =number of deaths from bladder cancer.

N = Number in study.

The mortality rates and mortality rate ratios in the sensitivity analyses with or without excluding diabetic patients living in the arseniasis-endemic areas showed similar results as observed in the original cohort (Tables II and III). In all analyses the magnitude of mortality rate ratios was largest for the youngest age group of 25–64 years (Tables II and III).

Discussion

This study suggested that although the secular trend of crude mortality rate from bladder cancer was increasing during the 12-year period of 1995–2006 in both men and women in Taiwan, the trend of the age-standardized rates were steady (Table I). The trends of age-specific mortality from bladder cancer were also steady except in men aged 25–64 years where a decreasing trend was observed (Table I). This is in contrast to an increasing trend in cancers of the lung, liver, oral cavity, colon and rectum, breast, and prostate; and a decreasing trend in cancers of the stomach and uterine cervix, as observed in an analysis of the secular trends of cancer mortality in Taiwan [15]. This secular trend also seemed to be contradictory to some studies showing an increasing trend of bladder cancer in white people [16].

Furthermore, except for the age group of ≥75 years in women in the original cohort analysis, the risk of mortality from bladder cancer between the diabetic patients and the general population as indicated by the mortality rate ratios increased significantly in both sexes (Tables II and III). The mortality rate ratios were especially remarkable in the youngest age group of 25–64 years and supported an inverse association with increasing age (Tables II and III). These observations strongly supported a close link between diabetes mellitus (or more specifically type 2 diabetes mellitus) and bladder cancer. Such a consistent association in both sexes had not been previously observed by other investigators in either the USA [17], Korea [2], or Japan [3]. The age-related attenuation in the mortality rate ratios was also not previously reported. One of the explanations for the highest relative risk of bladder cancer mortality in young diabetic subjects aged <65 years in comparison to the population controls is a reflection of the fairly low rate of bladder cancer in the general population in this age group (Table I). While diabetes may increase the risk of bladder cancer in this age group, the mortality rate ratio would increase manifold more than those observed in the older age groups (Tables II and III). Because the incidence of type 2 diabetes mellitus is increasing in Taiwan, which is more remarkable in the young generation [6], the observation of a higher mortality rate ratio in the young age group has important clinical implications. It is expected that with the cumulative prevalent cases of diabetes there will be more and more cases of bladder cancer, and the trend of mortality from bladder cancer will surely increase and not remain steady some time in the future. However, the mortality rates of bladder cancer in the diabetic patients in this younger age group of 25–64 years, i.e. 7.01–7.91 per 100,000 person-years for diabetic men (Table II) and 4.34–5.80 per 100,000 person-years for diabetic women (Table III), were still much lower than the respective figures in subjects ≥65 years in the general population (Table I). This finding suggested that age was a much more important risk factor than the effect of diabetes.

Men have a higher risk of developing bladder cancer than do women, and smoking has been identified as the most important risk factor [12], [18]. In the present study we also observed a higher mortality from bladder cancer in men than in women in either the general population (Table I) or the diabetic patients (Tables II and III). The link between smoking and bladder cancer can also be supported by an observation that stopping smoking might exert a beneficial effect on reducing bladder cancer recurrence after its diagnosis [19]. However, the effect of smoking might be different between different sexes. In a study conducted in Taiwan smoking is not a major risk factor for bladder cancer for women [20]. Although without supporting data, we believed that the impact of smoking in our diabetic women might be small because of the very low smoking rate (4.5%) in comparison to that of their male counterparts (63.8%) [21]. A recent study in the USA suggested a significantly higher risk of bladder cancer in women exposed to environmental tobacco smoke in a dose-responsive pattern [22]. Therefore even though smoking rate is low in our diabetic women, the impact of environmental tobacco smoke on bladder cancer to our diabetic women and also to other non-smokers should be attended to.

End-stage renal disease (ESRD) [23] and renal transplantation [24] are associated with bladder cancer. One might attribute the increased mortality from bladder cancer in the diabetic patients to the higher risk of ESRD and higher chance of receiving renal transplantation in these patients. However, both of these could not explain the link between diabetes mellitus and bladder cancer due to the following reasons: 1) The percentage of renal transplantation in patients with ESRD is actually very small therefore the impact might not be significant; and 2) The magnitudes of mortality rate ratios were consistently so large in the age group of 25–64 years, and ESRD in the diabetic patients rarely occurred in such a young age. Furthermore, a recent study in Taiwan suggested that although patients with ESRD had a higher risk of developing transitional cell carcinoma involving mainly the bladder, such a higher risk was associated with age at the time of dialysis, female sex, analgesic use, and Chinese herb use, but not with diabetes [25].

Arsenic is a significant risk factor for the development of bladder cancer [12]. In Taiwan there are two endemic areas of environmental exposure to high levels of arsenic from drinking well-water, i.e. the south-western coast confined to the so-called blackfoot disease-endemic area [13], [14] and the north-eastern coast confined to the Ilan basin [12]. However, it is hard to simply ascribe the observed higher risk of mortality from bladder cancer in the diabetic patients to a higher arsenic exposure for the following reasons: 1) The diabetic patients were recruited from all over Taiwan and not restricted to the arseniasis-endemic areas; 2) The exposure to arsenic from drinking-water in the blackfoot disease-endemic area has been terminated for nearly three decades. If such exposure did exert an effect, it should be more obviously observed in the older age rather than in the younger age groups in the present study (Tables II and III); 3) If such an exposure did exert an effect, it would only attenuate the mortality rate ratios because the denominator of the mortality rates in the general population would be inflated; and 4) The sensitivity analyses after excluding diabetic patients living in the arseniasis-endemic counties did not affect our conclusion (Tables II and III).

Cadmium exposure also increases the risk of bladder cancer [26], and an early study in Taiwan suggested that tea consumption was associated with an increased risk [27]. On the other hand alpha-tocopherol and retinol [28], vitamin B6 [29], alcohol consumption [30], consumption of raw cruciferous vegetables [31], and total water intake [32] might be potentially protective. Family history [33], birth weight [34], genomic DNA hypomethylation [35], microRNA profile [36], telomere length [37] and genetic variants involved in the inflammatory pathway [38], the metabolism of tobacco carcinogens [39], and DNA repair [40], etc. might also contribute to the development of bladder cancer. Physical activity seems to be unrelated to bladder cancer [41], and the association with body mass index is controversial [1], [41]. We were not able to evaluate the effects of these potential confounders because of lack of data collection.

The mechanisms underlying the association between type 2 diabetes mellitus and various cancers have been postulated as the activation of insulin pathway caused by hyperinsulinemia as a result of insulin resistance and the activation of insulin-like growth factor pathway, which stimulate cell proliferation and inhibit apoptosis [42]. Animal studies support a role of insulin-like growth factor on bladder carcinogenesis [43], and a case-control study in the USA provided evidence for such a link [44]. Diabetic patients may also suffer from a higher risk of urinary tract infection, which potentially contributes to a higher risk of bladder cancer [45]. However, such infection-induced bladder cancer risk could not be confirmed in a recent case-control study [46].

The strengths of this study included a prospective follow-up of a large cohort of diabetic patients, the probable completeness of ascertainment of the vital status by matching the computerized data file of the National Register of Deaths, and the exclusion of possible cases of type 1 diabetes mellitus to demonstrate a close link with type 2 diabetes mellitus. However, there are some limitations. First, we evaluated cancer mortality and not incidence. If diabetes mellitus has an effect on the case fatality rate, then the estimation of the mortality rate ratio might not properly reflect the incidence rate ratio. Second, there might be an increased detection bias in the diabetic patients because they might have visited their physicians more frequently, rendering a higher probability of detecting the existence of cancers. However, this might only suggest a higher detection rate of early cancer resulting in a better prognosis in the diabetic patients, which would only attenuate the magnitude of the mortality rate ratios. Furthermore, because the mortality from bladder cancer in the general population actually included patients with diabetes, the mortality rates from bladder cancer in non-diabetic subjects should be lower than those derived from the general population. This would also attenuate the true relative risk. Third, since diabetes mellitus is always underdiagnosed, the misclassification of some diabetic patients as non-diabetic controls would further attenuate the relative risk of bladder cancer between diabetic patients and non-diabetic controls. Fourth, diabetic patients might have taken more medications than people without diabetes, which might have complicated the situation. For example, long-term use of aspirin is preventive for colorectal cancer [47], and metformin, an oral anti-diabetic agent, is preventive for breast cancer [48]. If these medications exert a similar protective effect on bladder cancer, it is reasonable that the estimated mortality rate ratios would attenuate the true relative risk because the patients with type 2 diabetes mellitus might have been using metformin for controlling their blood glucose, and a higher proportion of the diabetic patients might have been using aspirin for the prevention of cardiovascular diseases. However, we were not able to evaluate the effects of these medications because of the lack of collecting such information. Fifth, the diabetic cohort got old during the observational period, therefore the effect of aging could not be completely excluded. In secondary analyses we re-stratified the patients according to their ages at the end of follow-up and not at entry of the study and re-calculated the mortality rate ratios comparing with the highest mortality rates of the general population. By using such a stringent method, the mortality rate ratios between diabetic patients and the general population remained significant for the age groups of 25–64 and 65–74 years for men; and for the age group of 25–64 years for women (data not shown). Therefore, the association between diabetes and bladder cancer mortality was especially significant for the younger patients. Lastly, because we observed a significant decreasing trend of mortality from bladder cancer in the younger men aged 25–64 years (Table I), the effect of treatment improvement for bladder cancer over the time-course of the study could not be excluded.

In summary, we have demonstrated a general steady trend of mortality from bladder cancer in both sexes in the Taiwanese general population during a period of 12 years from 1995 to 2006. Furthermore, the risk of mortality from bladder cancer is increased in the diabetic patients of either sex, and the magnitude of such risk is most remarkable in patients below 65 years of age. Given that the population is ageing and the incidence of type 2 diabetes mellitus is increasing in Taiwan, which is more remarkable in the younger generation [6], the impact of the link between diabetes and bladder cancer on the mortality of the population should warrant public health attention.

Acknowledgements

The authors thank the following institutes for their continuous support on epidemiologic studies of diabetes mellitus and arsenic-related health hazards: the New Century Health Care Promotion Foundation; the Department of Health (DOH89-TD-1035; DOH97-TD-D-113-97009), the National Taiwan University Hospital Yun-Lin Branch (NTUHYL96.G001), and the National Science Council (NSC-86-2314-B-002-326, NSC-87-2314-B-002-245, NSC88-2621-B-002-030, NSC89-2320-B002-125, NSC-90-2320-B-002-197, NSC-92-2320-B-002-156, NSC-93-2320-B-002-071, NSC-94-2314-B-002-142, NSC-95-2314-B-002-311, and NSC-96-2314-B-002-061-MY2), Taiwan. Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.