Prostate cancer mortality in Taiwanese men: Increasing age-standardized trend in general population and increased risk in diabetic men

Abstract

Background. To evaluate the trend of prostate cancer mortality in Taiwanese general population and the association between diabetes and prostate cancer mortality.

Materials and methods. In the general population during 1995–2006, the trends of prostate cancer mortality were evaluated, followed by calculation of age-specific mortality rates for age 40–64, 65–74, and ≥ 75 years. A cohort of 102,651 diabetic men aged ≥ 40 years recruited in 1995–1998 was followed prospectively.

Results. The trends of crude and age-standardized mortality from prostate cancer in the general population increased significantly (P < 0.0001). In the general population, 7,966 men aged ≥ 40 years died of prostate cancer, and aging was associated with increased risk. Age-specific prostate cancer mortality suggested significantly increasing trend for ages 65–74 and ≥ 75 years. A total of 321 diabetic men died of prostate cancer (crude mortality rate 41.9/100,000 person-years). Mortality rate ratios (95% confidence interval) showed higher risk of prostate cancer mortality in the diabetic patients, with magnitude increased with decreasing age: 1.55 (1.29–1.86), 2.68 (2.29–3.13), and 6.84 (5.34–8.75) for age ≥ 75, 65–74, and 40–64 years, respectively.

Conclusions. Prostate cancer mortality in the Taiwanese general population is increasing. Diabetic patients have a higher risk of prostate cancer mortality, which is more remarkable with decreasing age.

Key words::

• Diabetes
• mortality
• mortality rate ratio
• prostate cancer
• underlying cause of death

Key messages

• Diabetic patients show a lower risk of prostate cancer in some Caucasian studies. However, the association between diabetes and prostate cancer has rarely been evaluated in Asians.

• This study evaluated the trend of prostate 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 prostate cancer mortality in the Taiwanese general population is increasing and diabetic patients have a higher risk of prostate cancer mortality, which is more remarkable with decreasing age.

Introduction

Asian men have the lowest incidence of and mortality from prostate cancer (1). However, the incidence has risen rapidly in Asian men in recent decades (1), probably due to Westernization of life-style, increasing prevalence of obesity, and increased intake of fat (1). In Taiwan, a study analyzing the secular trend of prostate cancer mortality from 1964 to 1994 showed a 2-fold increase in age-adjusted mortality rate within this 30-year period (2). On the other hand, the secular trend of prostate cancer mortality in the USA was increasing in the late 1980s and declined from 1993 to 2003 (3). It is not known whether the secular trend in Taiwan keeps on rising since the publication of the previous study (2) or begins to decrease as shown in the USA (3).

Diabetic patients seem to carry a lower risk of prostate cancer in two recent meta-analysis studies, with a 9% (4) and 16% (5) reduction of risk, respectively. When we examined the data presented in the two meta-analysis studies, it was noticed that many of the component studies were case-control in design and only three of them focused on the follow-up of cohorts of diabetic patients (6–8). Among these three cohorts of diabetic patients, the cases of prostate cancer were 9 (6), 498 (7), and 2,455 (8), respectively, and only the last study (8) showed a significant risk reduction of 9% in prostate cancer associated with diabetes. Actually except for the first study, which was conducted in residents with diabetes in Rochester, Minnesota of the USA (6), the diabetic patients in the other two studies were recruited from hospitalized patients in Denmark (7) and Sweden (8), respectively. Although some recent studies still suggest that diabetes is protective for prostate cancer (9–11), this was not similarly observed by others (12). More recent studies suggest the existence of heterogeneity in the association between diabetes and prostate cancer (10,13–15). For example, the cohort study of the Kaiser Permanente Northern California showed an inverse association in subjects with glucose aberration and diabetes when the follow-up began 10 years after enrollment; but a positive (though not significant) association with glucose aberration or diabetes was observed during the first 10 years of follow-up (13). Another recently published prospective study from the USA also suggested divergent relations between diabetes and prostate cancer: diabetes had an inverse association with early-stage prostate cancer, showed no relation with aggressive prostate cancer, but had a positive association with aggressive prostate cancer in the subgroup of men with low body mass index (15).

It is also true that most of the studies did not differentiate between type 1 and type 2 diabetes. A recent population-based case-control study in the USA concluded that diabetes was not associated with prostate cancer (odds ratio 0.98; 95% confidence interval (CI) 0.76–1.27) (12). However, they did find a reduction of prostate cancer risk in the subgroup of early-onset diabetes diagnosed before age 30, which was considered as a proxy for type 1 diabetes (odds ratio 0.27; 95% CI 0.07–0.97) (12). The investigators suggested that the protective effect of diabetes on prostate cancer in earlier studies might be due to a confounding of the mixture of type 1 diabetes (12).

Most of the above-mentioned studies were not conducted in Asian populations. The early Japanese study included in the meta-analysis by Kasper and Giovannucci (5) was conducted in 1976 and used a case-control design (100 men with prostate cancer and 100 controls) with an estimated relative risk of 1.17 (0.90–1.39) (16). A more recent Japanese cohort study identifying 230 incident cases of prostate cancer among 22,458 men suggested that a history of diabetes was not predictive for total prostate cancer, but diabetic patients did show a significantly higher risk of advanced prostate cancer with an adjusted hazard ratio of 1.89 (1.02–3.50) (17). Although a multi-ethnic study showed a significant and inverse association between diabetes and prostate cancer, the association was not statistically significant in African Americans and Native Hawaiians (9). Insulin level seems to be neutral in the association with the risk of developing prostate cancer in white people (18), but a study conducted in Chinese people showed that those with the highest tertile of insulin level carried a significantly 2.6-fold higher risk of developing prostate cancer than those men in the lowest tertile (19). Therefore, the effect of diabetes on the development of prostate cancer might not be the same among different ethnicities.

The inverse association between diabetes and prostate cancer required further confirmation, especially in the Asian populations. The purpose of this study was to evaluate, based on different strata of age, 1) the secular trends of prostate cancer mortality in the general population of Taiwan; and 2) the association between diabetes and prostate cancer mortality in all diabetic patients or patients with type 2 diabetes.

Materials and methods

Prostate cancer mortality in men of the general population

The study was approved and supported by the Department of Health, Executive Yuan, Taiwan. In Taiwan, every resident has a unique identification number, and events like birth, death, marriage, or migration should be registered in the household registration offices. If a person dies, a death certificate should be issued by a physician, and this certificate should be reported to the household registration offices within 30 days as required by law. Therefore the database is accurate and complete and is centrally controlled by the government. The data of the death certificates including the unique identification number, date of birth, sex, and date and cause of death have been computerized since 1971 and can be used for academic research. It is easy to link the death certificate data of those who died by using the unique identification number. The causes of death were coded according to the ninth revision of the International Classification of Diseases (ICD-9) in Taiwan since 1981. Death due to prostate cancer was defined according to the ICD-9 code 185. Because the contributing causes of death were not available, only the underlying cause of death was used for analyses in this study.

The age-and-sex-specific population numbers are reported annually by the government of Taiwan, and the data are also available for use in electronic files in recent decades. The secular trends of both crude and age-standardized (to the 2000 World Health Organization population) mortality rates from prostate cancer from 1995 to 2006 in the Taiwanese male general population were first calculated for all ages based on the death certificate database and the age-specific population numbers in men in each year. Linear regression was used to evaluate whether the trends of mortality significantly changed with regard to calendar years, where the mortality rate was the dependent variable and the calendar year was the independent variable.

Because prostate cancer is very rare among young men, we chose to analyze the data for men aged ≥ 40 years, and the following age groups were applied: 40–64, 65–74, and ≥ 75 years old. The age-specific mortality rates were calculated by dividing the total number of deaths ascribed to prostate cancer by the mid-year male population in the specific age groups from the years 1995 to 2006. Average mortality rates during the period of 1995–2006 for specific age groups were also calculated by dividing the average numbers of deaths due to prostate cancer within these 12 years by the average mid-year male population of the specific age groups within the period. Furthermore, the yearly age-specific overall mortality rates from all-causes in the general population were calculated. The trends for these age-specific mortality rates during the period were also examined by linear regression.

Prostate cancer mortality in the diabetic patients

Figure 1 shows a flow chart for the follow-up of a cohort of diabetic patients for prostate cancer mortality. 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 national cohort of 256,036 diabetic patients using the NHI was established for epidemiologic studies related to diabetes, which was described in detail elsewhere (20). The diagnosis of diabetes was defined by the ICD-9 code 250 or the A-code (abridged code) of A181. Among them 102,651 were men aged ≥ 40 years (labeled as ‘the original cohort’ in this study, Figure 1).

Figure 1. Flow chart showing the procedures in the calculation of mortality from prostate cancer in the diabetic cohorts.

All patients were followed until the end of 2006. The vital status and the date and cause of death in those who died were obtained by matching the computerized data file of the death certificates by 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 enrollment until the end of 2006 for those who were alive or to the date of death for those who died. The patients were categorized into age subgroups according to their age at enrollment. Age-specific mortality rates and the mortality rate ratios for the diabetic men versus the general population of the male sex were calculated. The age-specific mortality rate ratios were calculated using the average mortality (from prostate cancer) rates within the 12 years in the general population as referents. It is recognized that age is a strong risk factor for prostate cancer and the subjects would age during the long follow-up period after enrollment. In order to reduce the possible aging effect on age subgroup categorization of the subjects, the analyses for the original cohort were also performed by splitting the follow-up duration into two periods: 1) from enrollment to end of 2000: age was categorized at enrollment, and mortality was followed from enrollment to end of 2000; and 2) from 2001 to end of 2006: those who died before the end of 2000 were excluded, age was calculated at 2001, and mortality was followed from 2001 to end of 2006.

For subcohort 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 (Figure 1), as described in detail elsewhere (21,22). The total number of diabetic patients interviewed was 93,484, and among them 39,927 were men aged ≥ 40 years (labeled as ‘subcohort diabetic men aged ≥ 40 years’, Figure 1). To evaluate whether the association was found in patients with type 2 diabetes, the mortality rates and mortality rate ratios were also calculated after excluding patients with type 1 diabetes based on the following criteria: diabetic ketoacidosis at the onset of diabetes, or the need for insulin injection within 1 year after diagnosis of diabetes. There were 38,696 diabetic men after such exclusion, and they were labeled as ‘subcohort type 2 diabetic men aged ≥ 40 years’ (Figure 1). Because there were only 1,231 patients excluded with possible diagnosis of type 1 diabetes, and among them only 3 died of prostate cancer, we did not analyze the association in patients with type 1 diabetes. In order to exclude the possibility that diabetes might be caused by incipient prostate cancer, the analyses were also done by dividing the diabetic patients into subgroups with a duration of diabetes at enrollment of < 10 years and ≥ 10 years.

Results

Figure 2 shows the unstandardized and age-standardized secular trends of prostate cancer mortality in the male general population from 1995 to 2006. Both trends are increasing with statistical significance (P < 0.0001). A total of 7,966 men aged ≥ 40 years in the general population died of prostate cancer during the period. Table I shows the age-specific mortality rates (per 100,000), the age-specific average mortality rates, and the age-specific overall mortality rates from all-causes in the male general population during the period. Basically the mortality rates from prostate cancer and from all-causes increased dramatically with increasing age in any specific calendar year. During this 12-year period a significantly increasing trend of prostate cancer mortality was observed for the age groups of 65–74 and ≥ 75 years (P < 0.0001), but not for the age group of 40–64 years. However, the overall mortality from all-causes decreased significantly in all age groups.

Figure 2. Secular trends of prostate cancer mortality in the Taiwanese male general population from 1995 to 2006.

Table I. The mortality rate (per 100,000) from prostate cancer and overall mortality rate (per 100,000) from all-causes by age in the Taiwanese male general population in 1995–2006 and the average mortality rate within this 12-year period.

Age (years)	Prostate and overall MR in general population	1995	1996	1997	1998	1999	2000	2001	2002	2003	2004	2005	2006	Average	P^a
40–64	n of PC	35	34	47	46	29	61	60	44	48	40	56	60	47
	MYP	2576753	2654115	2737115	2831340	2933998	3037988	3142535	3246734	3350715	3452074	3546682	3640548	3095883
	MR from PC	1.4	1.3	1.7	1.6	1.0	2.0	1.9	1.4	1.4	1.2	1.6	1.6	1.5	0.7948
	Overall MR	810.7	804.7	769.9	751.7	737.2	719.8	708.4	683.1	678.3	686.5	701.9	684.3	723.5	<0.0001
65–74	n of PC	119	163	166	173	167	174	184	201	174	198	214	204	178
	MYP	633594	651416	661022	666575	668093	663768	654953	646223	638278	631604	628407	629220	647763
	MR from PC	18.8	25.0	25.1	26.0	25.0	26.2	28.1	31.1	27.3	31.3	34.1	32.4	27.5	<0.0001
	Overall MR	3156.1	3105.7	3041.4	3002.2	3004.8	2959.2	2999.3	2896.5	2867.6	2844.7	2811.4	2653.4	2946.8	<0.0001
≥ 75	n of PC	216	266	317	321	365	399	449	505	520	581	639	691	439
	MYP	239815	256536	275487	295092	315060	338169	363853	389650	415976	438526	466054	488446	356889
	MR from PC	90.1	103.7	115.1	108.8	115.9	118.0	123.4	129.6	125.0	132.5	137.1	141.5	123.0	<0.0001
	Overall MR	8869.8	8824.9	8258.5	8331.6	8199.7	7902.9	7755.6	7555.8	7479.8	7600.0	7681.5	7162.1	7861.7	<0.0001

MR = mortality rate; MYP = mid-year population; PC = prostate cancer.

^aThe trend tested by linear regression.

Among the original diabetic cohort, 321 patients died of prostate cancer (Figure 1) with a calculated mortality rate of 41.9 per 100,000 person-years. The age-specific mortality rates from prostate cancer in the diabetic patients and their mortality rate ratios compared to the general population are shown in Table II. The age-specific mortality rate ratios were all significant in the original cohort analyses and in the subcohort analyses (except for the oldest age group of ≥ 75 years with duration of diabetes at enrollment < 10 years). An increase in the magnitude of mortality rate ratio was observed with decreasing age in all the analyses, suggesting that the youngest age group of diabetic patients might have the highest risk of prostate cancer mortality compared to the general population (Table II). It is unlikely that diabetes was caused by incipient prostate cancer, because diabetes diagnosed at least 10 years before prostate cancer mortality can hardly be a consequence of the carcinogenic process. The possible effect of aging during the long follow-up period on age subgroup categorization was also minimal because the results were similar when the long duration of follow-up was split into two shorter parts in the original cohort analyses.

Table II. Age-specific mortality rates (per 100,000 person-years) from prostate cancer in the diabetic male patients and the mortality rate ratios comparing diabetic male patients to the average mortality rates in the Taiwanese male general population.

Age (years)	n	N	Person-year	Mortality rate	Mortality rate ratio
I. Original cohort analysis
Followed from enrollment to end of 2006
40–64	51	60278	494704.5	10.31	6.84 (5.34–8.75)
65–74	156	30704	211811.1	73.65	2.68 (2.29–3.13)
≥ 75	114	11669	59761.6	190.76	1.55 (1.29–1.86)
Followed from enrollment to end of 2000
40–64	19	60278	203137.2	9.35	6.21 (4.16–9.25)
65–74	53	30704	94488.8	56.09	2.04 (1.56–2.66)
≥ 75	58	11669	30353.6	191.08	1.55 (1.20–2.01)
Followed from 2001 to end of 2006
40–64	22	48431	267973.6	8.21	5.45 (3.73–7.96)
65–74	75	26548	133463.8	56.19	2.04 (1.63–2.56)
≥ 75	94	13653	59608.3	157.70	1.28 (1.05–1.57)
II. Subcohort diabetic men analyses
Diabetes of any duration at enrollment
40–64	17	23958	167823.3	10.13	6.72 (4.43–10.19)
65–74	58	12395	76349.6	75.97	2.76 (2.15–3.55)
≥ 75	33	3574	17710.4	186.33	1.51 (1.08–2.13)
Diabetes duration < 10 years at enrollment
40–64	14	18376	131755.2	10.63	7.05 (4.47–11.11)
65–74	38	7887	49825.0	76.27	2.77 (2.04–3.77)
≥ 75	18	2138	10946.7	164.43	1.34 (0.84–2.12)
Diabetes duration ≥ 10 years at enrollment
40–64	3	5582	36070.3	8.32	5.52 (2.01–15.11)
65–74	20	4508	26541.7	75.35	2.74 (1.80–4.18)
≥ 75	15	1436	6785.4	221.06	1.80 (1.09–2.96)
III. Subcohort type 2 diabetic men analyses
Diabetes of any duration at enrollment
40–64	16	23197	162814.6	9.83	6.52 (4.24–10.03)
65–74	56	12051	74413.3	75.26	2.74 (2.12–3.53)
≥ 75	33	3448	17171.3	192.18	1.56 (1.11–2.19)
Diabetes duration < 10 years at enrollment
40–64	13	17991	129175.8	10.06	6.68 (4.15–10.74)
65–74	37	7742	48999.2	75.51	2.75 (2.01–3.75)
≥ 75	18	2088	10708.6	168.09	1.37 (0.86–2.17)
Diabetes duration ≥ 10 years at enrollment
40–64	3	5206	33638.1	8.92	5.92 (2.18–16.05)
65–74	19	4309	25428.1	74.72	2.72 (1.76–4.19)
≥ 75	15	1360	6484.3	231.33	1.88 (1.14–3.10)

n = number of prostate cancer.

Discussion

This study suggested that prostate cancer mortality in the general population of Taiwan is increasing during the 12-year period from 1995 to 2006 (Figure 2), which was more remarkable in the older population (Table I). The secular trend of prostate cancer mortality has been increasing over the past four decades in Taiwan if the observation of Chang et al. showing a 2-fold increase in age-adjusted mortality from 1964 to 1994 (2) is taken into account simultaneously. This secular trend in Taiwan is contradictory to that observed in the USA, which showed an increasing trend in late 1980s and declined from 1993 to 2003 (3). Although the etiology for such an increasing trend of prostate cancer in the Taiwanese remains largely unknown, some suggested that this could probably be associated with the Westernization of life-style, increased intake of fat, and high prevalence of obesity (1).

Furthermore, we found that diabetic patients might have a higher risk of prostate cancer mortality with increasing magnitude of mortality rate ratios with decreasing age, which remained consistent in all of the subgroup analyses (Table II). This would have public health importance if verified by future studies because diabetes is increasing dramatically worldwide, especially in the young generation. As well demonstrated in our previous study, the incidence of type 2 diabetes in Taiwan had increased approximately 2-fold and 3.5-fold, respectively, among the male population aged 35–64 and < 35 years during the period from 1992 to 1996 (22). With the increasingly accumulating number of diabetic patients in these age groups, it is expected that the incidence of and mortality from prostate cancer will be increasing in the years to come.

The increasing risk of prostate cancer mortality in the diabetic patients was somewhat inconsistent with the conclusions of an inverse association in the two meta-analysis studies (4,5) and in some recent studies (9–11). It should be pointed out that overall incidence of prostate cancer and mortality from the disease are two different entities and probably linked to different factors. In the present study we evaluated cancer mortality and not incidence. Diabetic patients may have poorer prognosis with higher mortality rate after they are diagnosed to have cancers (23). If diabetes has an effect on the case fatality rate in prostate cancer in the present study, the estimated mortality rate ratio might not properly reflect the incidence rate ratio.

Diabetic patients may develop more advanced prostate cancer than non-diabetic men (17). It is possible that diabetic men might have at the same time a lower risk of low-grade prostate cancer but a higher risk of high-grade cancer. The lower testosterone levels associated with obesity and hyperglycemia (4) or the reduced bioavailability of testosterone in the diabetic patients (24) may place these patients at a lower risk of developing low-grade prostate cancer. However, the high levels of insulin, insulin-like growth factors, and cytokines associated with diabetic patients might, on the other hand, place them at a higher risk of developing high-grade prostate cancer (25). Therefore, a higher risk of prostate cancer mortality might be seen in the diabetic patients even in the presence of an inverse association with the overall incidence of prostate cancer.

The introduction of prostate-specific antigen (PSA) testing for screening of prostate cancer in 1990s may also lead to discrepant observations between mortality and incidence of prostate cancer. The more advanced prostate cancer in the diabetic patients (17) is less dependent on PSA testing for its detection. Furthermore PSA levels are lower in patients with diabetes (9,26,27). Therefore PSA testing may increase the detection and incidence of low-grade cancer in non-diabetic men but probably would not affect the detection of high-grade cancer in diabetic men. Indeed, the meta-analysis by Kasper and Giovannucci observed an inverse association between prostate cancer incidence and diabetes mainly in studies performed during post-PSA era (5).

Glycemic control and duration of diabetes might also complicate the link between diabetes and prostate cancer. For example recent studies demonstrated an increased risk during the early phase of hyperglycemia or diabetes and a lower risk in those with a prolonged duration (13,14). Insulin resistance with elevated levels of insulin, insulin-like growth factors, and pro-inflammatory factors might be inducible for prostate cancer during the early stage of diabetes (25). However, insulin levels in patients with type 2 diabetes vary during the long natural course of the disease. Because we did not have data on the hormonal or biochemical profiles of our diabetic patients and our analyses did not support a variation in prostate cancer mortality risk dependent on different duration of diabetes (Table II), we were not able to evaluate the effect of such hormonal or biochemical changes in our diabetic patients and did not know whether an analysis of the prostate cancer incidence (rather than mortality) in the diabetic patients would show a different result.

If the mortality rate ratio in the present study did reflect the incidence rate ratio in the Taiwanese, some clinical facts may help to explain the contradictory findings of a positive association in our population and an inverse association in Caucasians. Statin use is associated with a lower risk of prostate cancer, which could possibly be responsible for a declining secular trend of prostate cancer in the USA (3). The prescription rate of statins in diabetic patients was 21.1% in 2004 in the USA (28), 24.9% in 2005 in Austria (29), 40% in 2002 in Ireland (30), but only 7.8% in 1997 and increased to 10.1% in 2003 in Taiwan (31). The much higher prescription rates of statins in diabetic patients in European countries and in the USA could partly be responsible for the lower risk of prostate cancer in the Caucasian diabetic patients. On the other hand, serum creatinine is significantly predictive for increased prostate cancer risk in a dose-dependent pattern in a recent Finnish nested case-control study (32). It is well recognized that chronic kidney disease is much more common in the Asian populations with diabetes, including the Taiwanese (33). Therefore the higher prevalence and incidence of chronic kidney disease characterized by increased creatinine in our diabetic patients might also confer a higher risk of prostate cancer.

The strengths of this study included a prospective follow-up of a large cohort of diabetic patients, the completeness of ascertainment of the vital status by matching the computerized data file of death certificates on a nation-wide basis, and the exclusion of possible cases of type 1 diabetes to demonstrate a close link with type 2 diabetes. Patients with the use of insulin within 1 year of diabetes diagnosis were excluded for the classification of type 2 diabetes in order to avoid the possible misclassification of type 1 as type 2 diabetes. However, a recent study showed that the inverse association between diabetes and prostate cancer could be seen among users of oral antidiabetic drugs and among insulin users (11). In the present study, similar findings were seen with or without the exclusion of possible diagnosis of type 1 diabetes (including those who used insulin within 1 year of diabetes diagnosis) (Table II). Because the number of cases of type 1 diabetes was small and there were only three cases of prostate cancer mortality among these patients, we were not able to evaluate the association in patients with type 1 diabetes and believed that the positive association with prostate cancer mortality in the diabetic patients mainly reflected an association in patients with type 2 diabetes with or without the use of insulin.

There are some limitations. First, diabetic patients might have taken more medications than people without diabetes, which might have complicated the situation. For example, the use of statins (3), metformin (34), or aspirin (35) might be protective for some forms of cancer. On the other hand, increased cancer risk was observed in diabetic patients using sulfonylurea or insulin (36). We were not able to evaluate the effects of these medications because of the lack of collecting such information. Second, we did not have sufficient information of some important confounding factors, such as obesity, smoking, alcohol drinking, family history, life-style, dietary factors, hormonal profiles, and genetic parameters, etc., for adjustment in our analyses. However, because age is the strongest confounding factor for prostate cancer, we have tried to reduce its confounding effect by analyzing the data in different age groups and by using the age-standardized mortality rates in the secular trend analysis. Third, the low number of prostate cancer deaths in the current study can also be a limitation. However, a lack of sufficient power with small sample size would most likely lead to a lack of statistical significance and not a significant association as shown in the present study.

In conclusion, we have demonstrated an increasing trend of prostate cancer mortality in the Taiwanese general population from 1995 to 2006, which is more remarkable in the older ages. Furthermore, the risk of prostate cancer mortality is increased in the diabetic patients compared to the general population, and the magnitude of such risk increment becomes larger with decreasing age. This study highly suggests that the generally accepted concept that diabetes confers a lower risk of prostate cancer might not be universal for different countries or ethnicities. We recognize that mortality and incidence are different entities, and the association between diabetes and incidence of prostate cancer in our population requires further investigation. Given that the population is aging and the incidence of type 2 diabetes is increasing in Taiwan (22), the impact of prostate cancer on the mortality of the population should warrant public health attention.

Acknowledgements

The author thanks the following institutes in Taiwan for their continuous support on the epidemiologic studies of diabetes and arsenic-related health hazards: the New Century Health Care Promotion Foundation; the National Genotyping Center of National Research Program for Genomic Medicine, National Science Council; 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).

Declaration of interest: The author states no conflict of interest and has received no payment in preparation of this manuscript.