Mortality of patients with type 2 diabetes in Taiwan: A 10-year nationwide follow-up study

diabetes research and clinical practice 107 (2015) 178–186

Contents available at ScienceDirect

Diabetes Research and Clinical Practice jou rnal hom ep ag e: w ww.e l s e v i er . c om/ loca te / d i ab r es

Mortality of patients with type 2 diabetes in Taiwan: A 10-year nationwide follow-up study Wei-Hung Lin a,b, Chih-Hui Hsu c, Hua-Fen Chen d, Chi-Chu Liu e,*,1, Chung-Yi Li c,f,1 a

Institute of Clinical Medicine, College of Medicine, National Cheng Kung University, Tainan, Taiwan Division of Nephrology, Department of Internal Medicine, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University, Tainan, Taiwan c Department and Graduate Institute of Public Health, College of Medicine, National Cheng Kung University, Tainan, Taiwan d Department of Endocrinology, Far Eastern Memorial Hospital, Panchiao, New Taipei City, Taiwan e Department of Anesthesia, Sin-Lau Hospital, Tainan City, Taiwan f Department of Public Health, College of Public Health, China Medical University, Taichung, Taiwan b

article info

abstract

Article history:

Aims: This study aims to investigate the distribution of underlying-causes-of-death (UCOD)

Received 6 March 2014

among deceased patients with type 2 diabetes mellitus (DM) in Taiwan and assess the

Received in revised form

influence of socio-demographic characteristics on mortality in type 2 DM patients.

6 June 2014

Methods: A cohort study on patients who sought medical care for type 2 DM from 2000 to

Accepted 14 September 2014

2008 was conducted on 65,599 type 2 DM patients retrieved from the 1-million beneficiaries

Available online 5 October 2014

randomly selected from Taiwan’s National Health Insurance Database. The study cohort

Keywords:

2000 and 2009. We examined the distribution of UCOD in the deceased subjects. The hazard

was then linked to Taiwan’s Mortality Registry to ascertain the patients who died between Type 2 diabetes mellitus

ratios of mortality in relation to socio-demographic characteristics were estimated from Cox

Urbanization

proportional hazard model.

Geographic variation

Results: The leading causes of death in type 2 DM included neoplasm (22.68%), cardiovas-

Mortality rate

cular diseases (21.46%), and endocrine diseases (20.78%). Male gender and older ages were

Cohort studies

associated with significantly increased risk of mortality. In addition, lower urbanization and greater co-morbidity score were also significantly associated with an increased risk of mortality with a dose-gradient pattern. Conclusions: Neoplasm accounts for the largest portion (22.68%) of deaths in type 2 DM patients closely followed by with cardiovascular diseases (21.46%). An increased risk of mortality in type 2 DM patients in lower urbanized areas may reflect poor diabetes care in these areas. # 2014 Published by Elsevier Ireland Ltd.

* Corresponding author at: Department of Anesthesia, Sin-Lau Hospital, #1, No. 57, Sec. 1, Dongmen Rd, Tainan City, Taiwan. E-mail address: [email protected] (C.-C. Liu). 1 Both authors contributed equally to this article. http://dx.doi.org/10.1016/j.diabres.2014.09.021 0168-8227/# 2014 Published by Elsevier Ireland Ltd.

diabetes research and clinical practice 107 (2015) 178–186

1.

Introduction

Diabetes mellitus (DM) is an epidemic disease in the world. Marked changes in human health behaviors and lifestyle have resulted in higher incidence and prevalence of DM [1]. It has been recently estimated that the global prevalence of diabetes is 8.3% [2]. The number of people with diabetes is also rising because of population growth, aging, urbanization, and increasing prevalence of obesity and physical inactivity. The potential for increase in patients with diabetes is greatest in Asia [3]. Type 2 DM has become an important public health threat for the ethnic Chinese population living in mainland China, Hong Kong, Taiwan, and Singapore, with a prevalence of onefifth of the adult population [4]. Given the genetic susceptibility and rapid westernization of food and lifestyle, a striking increase in incidence and prevalence of type 2 DM is anticipated [5]. The rapid increase in incidence and prevalence of type 2 DM is a health or medical issue and an economic and social problem for most governments. In developed countries, the largest increase in the number of type 2 DM is recorded in the elderly population aged more than 65 years, but the larger part of new onset type 2 DM occurred in the 45–64 year old population,2 who are vulnerable to premature death from various complications related to DM. Increased public awareness on the adverse health consequences of type 2 DM has resulted in intensive monitoring and aggressive clinical management of DM worldwide. However, type 2 DM still accounts for a considerable number of deaths from discrete complications each year globally. Hence, cause-specific mortality statistics based on the underlying-cause-of-death (UCOD) recorded on the death certificate are important to compare the cause-of-death statistics between countries and across time. Most countries follow the guidelines determined by the World Health Organization (WHO) and would register mortality data according to the UCOD [6]. The UCOD is characterized as the disease or injury that triggers the sequence of morbid events leading directly to death. Information on the UCOD of diabetic patients may help estimate the disease burden of DM. Although several previous studies have suggested an association between urbanization and higher type 2 DM incidence [7,8], only few studies have examined the association of urbanization with mortality in DM. Taiwan introduced a universal health insurance to cover all citizens in 1995. The national health insurance (NHI) program was intended to assure the accessibility of health care at acceptable cost [9] and eliminate the financial barrier that prevents the poor from receiving health care services. Recognizing the association between urbanization and mortality in patients with type 2 DM under the above medical care system is important. This study aimed to investigate the distribution of UCOD in a nationally representative sample of type 2 DM patients. In addition, this study also sought to assess the influence of urbanization on the risk of mortality in patients with type 2 DM.

2.

Methods

2.1.

Source of data

179

Data investigated in this study were retrospectively retrieved from the medical claims of the National Health Insurance Research Database (NHIRD) provided by the Bureau of National Health Insurance (BNHI). NHIRD provides all inpatient and ambulatory medical claims for about 99% of Taiwanese [10]. To confirm the accuracy of claim files, the BNHI performs periodical expert reviews on a random sample for every 50–100 ambulatory and inpatient claims [9]. Therefore, information attained from NHIRD is complete and accurate [11,12]. We used the medical claims data from 1997 to 2009 to obtain a representative sample of one million people randomly selected from all beneficiaries registered in 2000. Using the scrambled personal identification number; all NHI datasets can be interlinked and linked externally to Taiwan’s Death Registry (TDR). Access to research data has been reviewed and approved by the Review Committee of the National Health Research Institute. In Taiwan, the law dictates that all live births and deaths must be registered within 10 days. The TDR is considered exact and complete because registering deaths in Taiwan is necessary for physicians to complete death certificates [13]. We retrieved the information on the date of death and UCOD for each deceased individual.

2.2.

Study cohorts and covariates

The study cohort consisted of all patients who sought inpatient or outpatient care for type 2 DM. Patients with type 2 DM were determined by the diagnostic codes of diabetes (International Classification of Disease, 9th Version Clinical Modification (ICD-9-CM): 250  0 or 250  2). If a patient had the code for type 2 DM at the time of discharge or had more than three ambulatory care claims for type 2 DM within a 1-year period, he or she was considered suitable for inclusion in the study. We limited our patients to those who had at least three ambulatory claims to avoid unexpected inclusion of miscoded patients from outpatient settings [14]. We obtained a total of 66,108 patients with an ambulatory claim for type 2 DM from 2000 to 2008. We also identified 38,854 patients with type 2 DM from inpatient claims. We excluded patients with ages less than 30 years (n = 2950) to avoid potential contamination by type 1 DM. Patients diagnosed with type 1 DM from 1997 to 2008 were also excluded (n = 4918). Finally, 65,599 patients with prevalent type 2 diabetes were included in the analysis. The flow chart for the cohort setup is shown in Fig. 1. The index date for each study patient was the date when he or she received inpatient or outpatient care for type 2 DM. Covariates analyzed in this study included gender, age, urbanization, and co-morbidity indicated by Charlson’s score. Age was determined on the index date. Information on a patient’s underlying illnesses was retrieved from inpatient and outpatient claims from the first day of 1997 up to the index date. Underlying illnesses included all diseases used to calculate the Charlson’s score, which is a weighted summary measure of clinically important concomitant diseases that has

180

diabetes research and clinical practice 107 (2015) 178–186

Fig. 1 – Flow chart of identification and follow-up of study subjects.

been adapted for use with ICD-9-CM coded administrative database [15]. We adopted the Charlson’s score developed by Deyo et al. [15]. In brief, the Charlson comorbidity index was successfully translated into a set with ICD-9-CM administrative database. The index has added explanatory power to patient age in accounting for variations in outcomes following lumbar spine surgery. The index has been widely used in studying the outcomes of other surgical procedures or medical services using the administrative database. We categorized each patients’ residential or employment area at the time of the index date into various levels of urbanization according to the classification scheme proposed by Liu et al. [16]. This scheme classified all 316 cities and townships of Taiwan into seven ordered levels of urbanization based on various indicators including population density, proportion of residents with college or higher education, percentage of elderly (>65 years) people, proportion of agriculture workforce, and

number of physicians per 105 people [16]. A greater number for urbanization level indicates higher level of urbanization.

2.3.

Follow-up and study end-points

The patients included in the study were linked to the TDR from 2000 to 2009 for possible episode of mortality. A minimum of one year was given to collect information on the mortality for each patient. After the episode of mortality was found, we retrieved the UCOD information of the deceased subject. The total number of deceased subjects was 14,712. However, 340 out of the 14,712 subjects had missing information on UCOD. Person-years (PYs) of follow-up were calculated for each diabetic patient from his/her index date to the date of mortality or censoring (i.e., December 31, 2009). We also chose patients without a code of type 2 DM within the study period as control group. The control group was matched to type 2 DM on

181

diabetes research and clinical practice 107 (2015) 178–186

age, sex, and year of first-time ambulatory care visit for DM from 2000 to 2008 with a ratio of 1:1. A total of 66,108 patients were identified for the control group from 2000 to 2008. Similar to the type 2 DM cohort, PYs of the follow-up were also calculated for each control patient from his or her index date to the date of mortality or censoring (i.e., December 31, 2009), and the underlying illnesses was retrieved from the first day of 1997 to the index date to calculate the Charlson’s score.

2.4.

Statistical analysis

We first described the characteristics of the study cohort and listed the UCOD for the deceased patients. We then estimated the mortality rate according to various covariates. Bivariate and multivariable Cox proportion hazard regression analyses were used to assess the adjusted effects of various covariates on the mortality rate. We also employed the life-table method to calculate the n-year survival rate of type 2 DM [17]. We investigated whether or not the association of predictors for all causes of mortality in patients with T2DM differed in males and females by fitting appropriate interaction terms. Genderstratified analyses were subsequently conducted after finding evidence of gender differences, such as gender-age (P < 0.01), gender-urbanization (P = 0.17), and gender-Charlson’s score interactions (P = 0.03). All analyses were performed by SAS statistical software (version 9.3 for Windows; SAS Institute, Inc., Cary, NC), and the results were considered to be statistically significant when two-tailed P values were less than 0.05.

3.

Results

Approximately 52.1% of the patients in our study cohort were males and patients aged 50–69 years accounted for more than half of the patients. Approximately 30.6% of the patients first appeared in the medical claims for type 2 DM in 2000, whereas 7–10% were first seen in the subsequent years between 2001 and 2008. Most of the patients lived in areas with higher urbanization. With respect to the co-morbidity noted three years prior to date of recruitment, the most prevalent comorbidity was respiratory disease (90.0%), followed by digestive disorders (89.3%), musculoskeletal disorders (75.6%), and diseases of the nerve systems (71.5%). The mean Charlson’s score was 4.4, with 39.8% of the patients having scores greater than five (Table 1). In the 358,806 person-years (5.5 person-years per person) of follow-up, 14,712 patients with type 2 DM encountered mortality of all causes, representing a mortality rate of 41.00 per 1000 person-years. In addition, the 1-year, 5-year, and 10year survival rate of the patients with type 2 DM was 93.3, 79.0, and 61.1%, respectively (data not shown). After the exclusion of cases with missing UCOD, malignant neoplasm was found to be the most common cause of death, accounting for 22.68% of the deaths in patients with type 2 DM. For the sit-specific cancer, neoplasms of liver and intrahepatic bile ducts (5.15%), trachea, bronchus and lung (4.31%), and colon and rectum (2.45%) were more common. Other leading causes of death included diseases of the circulation system (21.46%), endocrine, nutritional, and metabolic diseases, and immunity

Table 1 – Characteristics of study cohort (n = 65,599). Characteristics

No. of patients

Sex Men 34,199 31,400 Women Age (years) 30–39 3588 11,393 40–49 17,182 50–59 16,545 60–69 70–79 12,598 3921 80–89 372 90 60.5 (12.9) Mean (SD) Year of recruitment 20,097 2000 6175 2001 2002 5746 5382 2003 6531 2004 5799 2005 2006 5387 5814 2007 4668 2008 Urbanization levela 7 18,315 18,978 6 10,904 5 4 9629 1632 3 2919 2 2567 1 Missing 655 History of co-morbidity in 3 years prior to recruitmentb Infection (001–139) 25,813 12,389 Neoplasm (140–239) 42,239 Endocrine (240–279) 5046 Blood (280–289) 20,889 Mental (290–319) 46,909 Nerve (320–389) Circulation (390–459) 43,626 59,015 Respiratory (460–519) 58,571 Digestive (520–579) Genitourinary (580–629) 34,887 1008 Pregnancy and reproductive (630–676) 37,913 Skin (680–709) 49,617 Musculoskeletal (710–739) 40,107 Injury (800–999) Charlson’s scorec 0–2 21,179 18,289 3–4 26,131 5 4.4 (3.1) Mean (SD)

Percent

52.1 47.9 5.5 17.4 26.2 25.2 19.2 6.0 0.6

30.6 9.4 8.8 8.2 10.0 8.8 8.2 8.9 7.1 27.9 28.9 16.6 14.7 2.5 4.5 3.9 1.0 39.3 18.9 64.4 7.7 31.8 71.5 66.5 90.0 89.3 53.2 1.5 57.8 75.6 61.1 32.3 27.9 39.8

a

A higher number is indicative of a higher level of urbanization. Numbers in parenthesis are ICD-9-CM codes. c Calculation was based on the information of selected comorbidity in 1-year period prior to recruitment. b

disorders (20.78%). Cerebrovascular and ischemic heart disease accounted for 9.70 and 5.82% of total deaths, respectively (Table 2). Table 3 shows the overall and sex-specific predictors for all causes of mortality in patients with type 2 DM. Compared with females, male patients had a 1.43-fold hazard (95%

182

diabetes research and clinical practice 107 (2015) 178–186

Table 2 – The underlying causes of death among the deceased study subjects. Underlying causes of death

n

Diseases of the circulation system (ICD-9: 390–459, ICD-10: I00–I99) Ischemic heart disease (ICD-9: 410–414, ICD–10: I20–I25) Cerebrovascular disease (ICD-9: 430–438, ICD-10: I60–I69) Other disease of the circulation system Endocrine, nutritional and metabolic diseases, and immunity disorders (ICD-9: 240–279, ICD-10: E00–E90) Diabetes mellitus (ICD-9: 250, ICD-10: E10–E14) Malignant neoplasm (ICD-9: 140–208, ICD-10: C00–C97) Liver and intrahepatic bile ducts (ICD-9: 155, ICD-10: C22) Trachea, bronchus, and lung (ICD-9: 162, ICD-10: C349) Colon and rectum (ICD-9: 153–154, ICD-10: C18–C20) Pancreas (ICD-9: 157, ICD-10: C259) Stomach (ICD-9: 151, ICD-10: C169) Breast (ICD-9: 174–175, ICD-10: C50) Prostate (ICD-9: 185, ICD-10: C61) Other malignant neoplasm Diseases of the respiratory system (ICD-9: 460–519, ICD-10: J00–J99) Diseases of the digestive system (ICD-9: 520–579, ICD-10: K00–K93) Nephritis, nephrotic syndrome, and nephrosis (ICD-9: 580–589, ICD-10: N00–N07, N17–N19, N25–N27) Diseases of the nervous system and sense organs (ICD-9: 320–389, ICD-10: G00–G99) Infectious and parasitic diseases (ICD-9: 001–139, ICD-10: A00–B99) Injury and poisoning (ICD-9: 800–999, ICD-10: V0l-Y98) Other causes All Missing information All mortality

%

3084 836 1394 854 2987

21.46 5.82 9.70 5.94 20.78

2948 3260 740 620 352 190 177 93 88 1000 1331 1200 647

20.51 22.68 5.15 4.31 2.45 1.32 1.23 0.65 0.61 6.96 9.26 8.35 4.50

136

0.95

363 536 828 14,372 340 14,712

2.53 3.73 5.76 100.00

ICD: international classification for disease.

CI = 1.39–1.48) of mortality in 10 years. The adjusted hazard ratio (HR) of mortality significantly increased with increasing age and Charlson’s score, with a dose-gradient pattern ( p for trend <0.001). The adjusted HR ranged from 1.28 (95% CI = 1.10–1.50) (40–49 years) to 31.27 (95% CI = 26.06–37.51) (90 years). We also noted an inverse dose-gradient relationship ( p for trend < 0.01) between the level of urbanization and risk of mortality. Compared with patients residing in the most urbanized areas (i.e., urbanization level 7), patients from less urbanized regions experienced significantly increased hazards of mortality, with an adjusted HR ranging from 1.11 (95% CI = 1.06–1.16) (urbanization level 6) to 1.38 (95% CI = 1.28–1.49) (urbanization 1). Gender differences were associated with sociodemographic factors and all-cause mortality (Pinteraction values ranged between 0.01 and 0.17). Stratified analyses by gender showed that age 90 years was associated with greater increased risk of mortality in women [adjusted HR (95% CI) = 53.53 (38.27–74.89)] than in men [adjusted HR (95% CI) = 23.66 (18.45–30.35)]. A Charlson’s score  5 was associated with increased risk of mortality in women [adjusted OR (95% CI) = 3.02 (2.86–3.19)] than in men [adjusted OR (95% CI) = 2.97 (2.83–3.11)]. Urbanization was not related to mortality when other risk factors were controlled. Regardless of gender, the adjusted HR of mortality significantly increased with increasing age and Charlson’s score in a dosegradient pattern ( p for trend <0.001). An inverse dosegradient relationship ( p for trend <0.01) was found between level of urbanization and risk of mortality in both genders (Table 3).

We further analyzed age-specific relative hazards of allcause mortality in patients with type 2 diabetes in comparison with non-diabetic controls, with adjustments for sex, urbanization, and co-morbidity (Supplement Table S1). Compared with the non-diabetic controls with the same age, patients with type 2 DM showed a ‘‘J curve’’ for age-specific adjusted HRs. The highest adjusted HR was noted in type 2 DM patients aged 30–39 years at 2.23 (95% CI = 1.50–3.32), followed by patients aged 90 years (HR = 1.60, 95% CI = 1.33–1.94) and 80–89 years (HR = 1.39, 95% CI = 1.29–1.49). Supplementary Table S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.diabres. 2014.09.021.

4.

Discussion

We conducted a 10-year population-based follow-up study and found that the leading causes of death in Taiwanese patients with type 2 DM included neoplasms and cardiovascular diseases (CVD). In addition, compared with female type 2 DM patients, male patients had a higher risk of mortality within 10 years. The risk of mortality was also significantly associated with older ages, lesser urbanization, and higher comorbidity scores. One US study reported that CVD are involved in the preponderance of mortality (69.5%) in patients with DM, particularly ischemic heart disease [18]. CVD is also the leading cause of death for patients with DM, with the proportion of death ranging from 47 to 74%, among European

Table 3 – Overall and sex-specific predictors for all causes mortality in patients with type 2 diabetes. Total

Sex Men Women Age (years) 30–39 40–49 50–59 60–69 70–79 80–89 90

Person-years observed

No. of death

Mortality rate (per 10,000 person-years)

Adjusted hazard ratio (95% CI)

181,088 177,720

8313 6399

459 360

19,920 66,459 98,146 96,771 63,442 13,319 750

197 917 1967 3768 5106 2458 299

99 138 200 389 805 1845 3987

1.00 1.28 (1.10–1.50) 1.76 (1.52–2.05) 3.10 (2.68–3.59) 5.93 (5.13–6.84) 13.70 (11.82–15.87) 31.27 (26.06–37.51) p for trend test <0.001

Urbanizationa 7 6 5 4 3 2 1

100,330 105,452 59,250 52,424 8457 15,433 13,630

3435 4017 2481 2582 518 878 769

342 381 419 493 613 569 564

1.00 1.11 (1.06–1.16) 1.19 (1.13–1.25) 1.26 (1.19–1.32) 1.34 (1.22–1.47) 1.37 (1.27–1.48) 1.38 (1.28–1.49) p for trend test <0.01

Charlson’s scoreb 0–2 3–4 5

93,383 102,506 162,917

1562 2877 10,273

167 281 631

1.00 1.40 (1.32–1.50) 2.51 (2.37–2.65) p for trend test <0.01

a b

Interaction by gender

Women

Adjusted hazard ratio (95% CI)

Adjusted hazard ratio (95% CI)

1.00 1.33 (1.12–1.59) 1.79 (1.51–2.11) 2.82 (2.39–3.32) 5.06 (4.30–5.96) 10.72 (9.04–12.70) 23.66 (18.45–30.35) p for trend test <0.01

1.00 1.22 (0.88–1.70) 1.95 (1.43–2.67) 3.95 (2.91–5.38) 8.14 (5.99–11.06) 20.78 (15.26–28.30) 53.53 (38.27–74.89) p for trend test <0.01

1.00 1.08 (1.02–1.15) 1.19 (1.11–1.27) 1.26 (1.18–1.35) 1.34 (1.17–1.52) 1.38 (1.24–1.52) 1.44 (1.29–1.60) p for trend test <0.01

1.00 1.15 (1.08–1.24) 1.19 (1.10–1.29) 1.22 (1.13–1.31) 1.35 (1.18–1.54) 1.33 (1.19–1.48) 1.25 (1.11–1.41) p for trend test <0.01

1.00 1.60 (1.51–1.69) 2.97 (2.83–3.11) p for trend test <0.01

1.00 1.50 (1.41–1.61) 3.02 (2.86–3.19) p for trend test <0.01

1.43 (1.39–1.48) 1.00 p < 0.01

p = 0.17

p = 0.03

diabetes research and clinical practice 107 (2015) 178–186

Predictors

Men

A higher number is indicative of a higher level of urbanization. Calculation was based on the information of selected co-morbidity in 1-year period prior to recruitment.

183

184

diabetes research and clinical practice 107 (2015) 178–186

and Asian cities including London, Switzerland, Warsaw, Berlin, Zagreb, and Hong Kong according to the multinational study of WHO [19]. A previous study from Japan mentioned that malignant neoplasms and CVD accounted for 25 and 19.5% of mortality in DM patients, respectively. The decreased prevalence of CVD as a cause of mortality in type 2 DM can be explained by the difference in the clinical manifestations of DM in Japan and in Western countries [20]. Similar to the Japanese study, we also reported a lesser proportion of CVD as the leading cause of mortality in Taiwanese patients with type 2 DM. The disease characteristics of DM tend to vary with geographical regions and ethnic groups. Contrary to our findings, an earlier Taiwanese study pointed out that death related to CVD were ranked as the most common cause of mortality in type 2 DM [21]. Over the last decade, enormous advances in the prevention and treatment of heart disease, as well as reduction in smoking and application of improved lifesaving technology in people suffering from CVD and other diabetes complications, have been reported [22]. The decline in hospitalization rates for CVD, ischemic heart disease, and stroke among people with DM was also noticed [23], and the CVD incidence and death risk in Taiwan has declined annually after 2000. However, the incidence of cancer and the number of cancer mortality gradually increased after 2000 in the general population [24]. The increase in cancer deaths may be because of more patients growing older and the development of technology used for detecting early cancer development over the past 10 years. All the above statements may explain, at least to some extent, the transition in the leading cause of mortality among type 2 DM in Taiwan. The dissimilarity in the distributions of leading causes of death in diabetic populations from different countries may also reflect the distribution of the causes of death in the general populations of different countries. For example, death from CVD has been the leading cause of death in the US, whereas malignancy has become the leading cause of death in Taiwan since 1982 [24]. Male patients had a 1.43-fold higher hazard of mortality over the study period, which was compatible with the findings from a previous study in Taiwan [21]. Risk of mortality significantly increased with increasing age, with a dose-gradient pattern from 1.28 (40–49 years) to 31.27 (90 years). Previous studies also found that adults with DM have 2- to 4-fold higher risk of CVD events relative to those without DM and have about 60% higher risk of early mortality [25,26]. The power of primary and secondary prevention strategies seems to be less adequate in older type 2 DM patients [27,28], despite evidence from clinical trials suggesting that cardioprotective agents confer a compatible degree of benefit in high-risk patients of all ages [29]. Therefore, our study suggests that aggressive risk-reduction strategies are needed in type 2 DM of middle to old ages to further reduce the mortality. A greater prevalence of DM among urban residents compared with rural residents was noticed in many areas of the world [7,8]. A higher mortality found in type 2 DM from less urbanized areas may be attributed to poor care and treatment of type 2 DM. An earlier Taiwanese study noted that a significant urban–rural difference in the relative risk of hip

fracture incidence in DM, notably in male patients, suggested poor diabetes care for patients with type 2 DM in rural areas [30]. Certain physical environments may account for a higher mortality rate in rural type 2 DM. For example, residential segregation of people with lower socio-economic positions in rural areas may influence the inequality of resources in rural areas. In addition, given that the universal health care system of Taiwan largely removes the barrier of receiving medical care services, minimal difference in affordability of health care services for people living in different areas exists. However, the accessibility of health care services in the rural parts of Taiwan is still a concern for residents in rural areas, mainly because of inadequate local medical care resources and inappropriate utilization pattern of some residents [31]. Inadequate accessibility to medical care services may be responsible for the observed urban–rural difference in mortality of type 2 DM patients in Taiwan. Given that patients with type 2 DM usually suffer from various life threatening complications, such as macrovascular disease and cancer, the urban–rural difference in mortality of type 2 DM patients observed in this study may result from geographic variations in the delivery of appropriate clinical management for diabetic complications experience by patients with type 2 DM. Another possible reason is that in the rural region, people who had severe DM status seek medical assistance had higher risk of death. Our study had several methodological strengths. First, the diabetic cohort and control group were collected from the NHI database, which is population based and is highly representative, allowing little room for recall and selection bias; the likelihood of nonresponse and loss to follow-up of cohort members is minimal. Second, the advantage of using insurance claim datasets in clinical research gave us easy access to the longitudinal records of a large sample of geographically disperse patients. Third, a large number of study subjects also made it possible for us to make sex-stratified analyses of mortality in relation to age, urbanization, and Charlson’s score. Despite the strength of our study, several limitations exist. First, the exclusive reliance on the claim data may have resulted in potential misclassification bias in our study. We used at least three diabetes-related diagnoses with the first and the last visits >30 days apart, which largely reduced the likelihood of disease misclassification. Second, to determine type 2 DM, patients who had been admitted because of diabetic ketoacidosis, discharged with the diagnosis of insulin-dependent diabetes mellitus, and those who received a certificate for potential type 1 DM were excluded. We cannot overlook the possibility that some type 2 DM patients were also be excluded by our algorithm. However, we believe that our algorithm only had a small impact on patients being excluded because of potential type 1 DM accounted for only 3% of the total number of identified cases. Finally, the national diabetes surveillance system was similar to that implemented in other countries. Information on the population, such as body mass index or waist circumference, blood pressure, laboratory data, smoking and diabetes, and family history were not available in the claims database, limiting the specific interpretations that can be made by our study.

diabetes research and clinical practice 107 (2015) 178–186

5.

Conclusion [8]

We found that neoplasm accounts for the largest proportion of deaths in Taiwanese patients with type 2 DM, implying that cancer screening programs is important in patients with type 2 DM. Patients with type 2 DM should be carefully assessed for liver and biliary, lung, and colorectal neoplasms. Our study also indicated an increased risk of mortality in type 2 DM patients residing in areas with lesser urbanization, which may suggest inadequacy of health care and cancer screening for type 2 DM patient in less urbanized areas. Future studies should look into the underlying causes of the observed increased risk of mortality in rural diabetic patients and implement multifaceted intervention programs to ensure the effective prevention and treatment of type 2 DM in high-risk populations and reduce the social disparities in the mortality of type 2 DM patients in Taiwan.

[9]

[10]

[11]

[12]

[13]

[14]

Conflict of interest The authors declare that they have no conflict of interest.

Acknowledgments This study was partially supported by a grant with National Scientific Council (NSC101-2314-B-006-076-MY3), who however has no role in this study. The interpretation and conclusions contained herein do not represent those of BNHI, Department of Health or NHRI.

[15]

[16]

[17] [18]

[19]

references [20] [1] Zimmet P, Alberti KG, Shaw J. Global and societal implications of the diabetes epidemic. Nature 2001;414: 782–7. [2] Guariguata L, Whiting DR, Hambleton I, Beagley J, Linnenkamp U, Shaw JE. Global estimates of diabetes prevalence in adults for 2013 and projections for 2035 for the IDF Diabetes Atlas. Diabetes Res Clin Pract 2014;103(2):137–49. [3] Amos AF, McCarty DJ, Zimmet P. The rising global burden of diabetes and its complications: estimates and projections to the year 2010. Diabet Med 1997;14: S1–85. [4] Gu D, Reynolds K, Duan X, Xin X, Chen J, Wu X, et al. Prevalence of diabetes and impaired fasting glucose in the Chinese adult population: International Collaborative Study of Cardiovascular Disease in Asia (InterASIA). Diabetologia 2003;46:1190–8. [5] Wong KC, Wang Z. Prevalence of type 2 diabetes mellitus of Chinese populations in Mainland China, Hong Kong, and Taiwan. Diabetes Res Clin Pract 2006;73:126–34. [6] World Health Organization. 2nd ed. International statistical classification of diseases and related health problems, vol. 2, 2nd ed. Geneva, Switzerland: World Health Organization; 2004. Tenth Revision. [7] Attard SM, Herring AH, Mayer-Davis EJ, Popkin BM, Meigs JB, Gordon-Larsen P. Multilevel examination of diabetes in modernising China: what elements of urbanisation are

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

185

most associated with diabetes? Diabetologia 2012;55: 3182–92. King H, Aubert RE, Herman WH. Global burden of diabetes, 1995–2025: prevalence, numerical estimates, and projections. Diabetes Care 1998;21:1414–31. Cheng TM. Taiwan’s new national health insurance program: genesis and experience so far. Health Aff (Millwood) 2003;22:61–76. Lu JF, Hsiao WC. Does universal health insurance make health care unaffordable? Lessons from Taiwan. Health Aff (Millwood) 2003;22:77–88. Sun Y, Chang YH, Chen HF, Su YH, Su HF, Li CY. Risk of Parkinson disease onset in patients with diabetes: a 9-year population-based cohort study with age and sex stratifications. Diabetes Care 2012;35:1047–9. Chen HF, Chen P, Li CY. Risk of malignant neoplasms of liver and biliary tract in diabetic patients with different age and sex stratifications. Hepatology 2010;52:155–63. Lu TH, Lee MC, Chou MC. Accuracy of cause-of-death coding in Taiwan: types of miscoding and effects on mortality statistics. Int J Epidemiol 2000;29:336–43. Chen PC, Chan YT, Chen HF, Ko MC, Li CY. Populationbased cohort analyses of the bidirectional relationship between type 2 diabetes and depression. Diabetes Care 2013;36:376–82. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–9. Liu CY, Hung YT, Chuang YL, Chen YJ, Weng WS, Liu JS, et al. Incorporating development stratification of Taiwan townships into sampling design of large scale health interview survey. J Health Manage 2006;4:1–22. Cutler SJ, Ederer F. Maximum utilization of the life table method in analyzing survival. J Chronic Dis 1958;8:699–712. Gu K, Cowie CC, Harris MI. Mortality in adults with and without diabetes in a national cohort of the U.S. population, 1971–1993. Diabetes Care 1998;21:1138–45. Morrish NJ, Wang SL, Stevens LK, Fuller JH, Keen H. Mortality and causes of death in the WHO Multinational Study of Vascular Disease in Diabetes. Diabetologia 2001;44:S14–21. Sasaki A, Horiuchi N, Hasegawa K, Uehara M. Mortality and causes of death in type 2 diabetic patients. A long-term follow-up study in Osaka District, Japan. Diabetes Res Clin Pract 1989;7:33–40. Tseng CH. Mortality and causes of death in a national sample of diabetic patients in Taiwan. Diabetes Care 2004;27:1605–9. Ford ES, Ajani UA, Croft JB, Critchley JA, Labarthe DR, Kottke TE, et al. Explaining the decrease in U.S. deaths from coronary disease, 1980–2000. N Engl J Med 2007;356:2388–98. Geiss L, Engelgau M, Pogach L, Acton K, Fleming B, Roman S, et al. A national progress report on diabetes: successes and challenges. Diabetes Technol Ther 2005;7:198–203. Department of Health, Executive Yuan, Taiwan, ROC. Health and vital statistics: 1982–2013. Taipei, Taiwan, ROC: Department of Health, Executive Yuan; 2014. Hu FB, Stampfer MJ, Solomon CG, Liu S, Willett WC, Speizer FE, et al. The impact of diabetes mellitus on mortality from all causes and coronary heart disease in women: 20 years of follow-up. Arch Intern Med 2001;161:1717–23. Fox CS, Coady S, Sorlie PD, Levy D, Meigs JB, D’Agostino Sr RB, et al. Trends in cardiovascular complications of diabetes. JAMA 2004;292:2495–9. Ko DT, Mamdani M, Alter DA. Lipid-lowering therapy with statins in high-risk elderly patients: the treatment-risk paradox. JAMA 2004;291:1864–70. McLaughlin TJ, Soumerai SB, Willison DJ, Gurwitz JH, Borbas C, Guadagnoli E, et al. Adherence to national

186

diabetes research and clinical practice 107 (2015) 178–186

guidelines for drug treatment of suspected acute myocardial infarction: evidence for undertreatment in women and the elderly. Arch Intern Med 1996;156:799–805. [29] MRC/BHF. Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet 2002;360:7–22.

[30] Chen HF, Ho CA, Li CY. Increased risks of hip fracture in diabetic patients of Taiwan: a population-based study. Diabetes Care 2008;31:75–80. [31] Tan HF, Tseng HF, Chang CK, Lin W, Hsiao SH. Accessibility assessment of the Health Care Improvement Program in rural Taiwan. J Rural Health 2005;21:372–7.