- Email: [email protected]
Predictors of mortality in hospitalized diabetic patients: A 7-year prospective study
diabetes research and clinical practice 80 (2008) 449–454
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/diabres
Predictors of mortality in hospitalized diabetic patients: A 7-year prospective study Mei-Chun Wu a, Wen-Jane Lee b, Scheng-Ming Tschen c, Shih-Yi Lin d,e,f, I-Te Lee d,e,f, Chii-Yuan Jeng d,e,f, Li-Nien Tseng d,e,f, Ying-Mei Liu a, Been-Yuan Chen a, Wayne H-H Sheu c,d,e,f,* a
Department of Pharmacy, Taichung Veterans General Hospital, Taichung, Taiwan Department of Medical Education and Research, Taichung Veterans General Hospital, Taichung, Taiwan c Department of Life Sciences, National Chung-Hsing University, Taichung, Taiwan d Division of Endocrinology and Metabolism, Department of Internal Medicine, Taichung Veterans, General Hospital, Taichung, Taiwan e Institute of Medical Technology, National Chung-Hsing University, Taichung, Taiwan f College of Medicine, National Yang-Ming University, Taipei, Taiwan b
article info
abstract
Article history:
This study was conducted to investigate the mortality rate, causes of death, and standar-
Received 7 January 2007
dized mortality ratio (SMR), and to identify the significant predictive factors of mortality in
Accepted 29 November 2007
diabetic patients at a medical center in Taiwan. Clinical data were obtained from 1792
Published on line 4 March 2008
diabetic inpatients discharged from the metabolism department of a medical center during the years 1996–2002. Underlying causes of death were determined from death certificates.
Keywords:
Predictors of mortality were assessed by uni- and multivariate Cox survival analyses. Of
Diabetes mellitus
1792 patients studied, 410 (22.9%) patients died. The crude mortality rate was 93.2/1000
Crude mortality rate
person-years, and the overall SMR was 2.98 (2.71–3.28). The percentages of causes of death
Standardized mortality ratios (SMRs)
ascribed to diabetes, cancer, cardiopulmonary disease, infection, stroke, digestive diseases,
Predictive factors
nephropathy, accidents, suicide, and disease of arteries, arterioles, and capillaries were 38.0, 13.2, 9.5, 7.8, 7.6, 6.8, 5.1, 2.0, 0.5, and 0.2%, respectively. The independent predictors of mortality were age greater than 65, duration of hypertension more than 5 years, 24 h proteinuria greater than 0.3 g, and estimated creatinine clearance less than 60 mL/min. Conclusion: The mortality of diabetic inpatients was about threefold that of the general population. The predictors of mortality included older age, longer duration of hypertension, increased 24 h proteinuria, and decreased creatinine clearance. # 2008 Elsevier Ireland Ltd. All rights reserved.
1.
Introduction
Diabetes has become endemic worldwide [1]. There are up to one million diabetic patients in the Taiwan area alone [2] and the average expense incurred for every
diabetic patient is 4.3 times that for the non-diabetic patient [3]. It is also estimated that the total medical expenditure would be reduced by 20% if the rate of poor glycemic control could be reduced to one-third of the current level [3].
* Corresponding author at: No. 160, Section 3, Chung-Kang Road, Division of Endocrinology and Metabolism, Taichung Veterans General Hospital, Taichung 407, Taiwan. Tel.: +886 4 2374 1300; fax: +886 4 2374 1318. E-mail address: [email protected] (W.-H. Sheu). 0168-8227/$ – see front matter # 2008 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.diabres.2007.11.020
450
diabetes research and clinical practice 80 (2008) 449–454
A long-term study of the mortality and the predictive factors of mortality of diabetic patients found that the mortality rate of diabetic patients in Taiwan was 1.63 times that of the general population [4]. Another long-term study found that the mortality rate of Brazilian type 2 diabetic outpatients was 3.36 times that of the general population, and the predictive factors for mortality in these diabetic individuals included old age, increased 24 h proteinuria, preexisting vascular disease, and decreased HDL cholesterol [5]. To the best of our knowledge, there have not been any followup studies on the mortality rate and mortality factors of hospitalized diabetic patients. Therefore, we collected the clinical characteristics of a group of diabetic patients after their discharge from hospital in order to investigate their longterm mortality rate and the predictive factors of mortality.
2.
Materials and methods
This study protocol was approved and supported by the Institutional Review Board of Taichung Veterans General Hospital (TCVGH), Taiwan. Data of study subjects were collected from 1841 consecutive hospitalized diabetic inpatients from 1 January 1996 to 31 December 2002, excluding the 49 patients who died either in hospital or within 1 month after their discharge from hospital, with the remaining 1792 as the subjects of the study (999 men and 793 women). The follow-up period lasted from the date when the patient entered the study till the date of death or until 31 December 2002, with median follow-up time of 27 months. The list of the 1792 subjects’ names was linked to the national mortality database to ascertain the death records before 31 December 2002. Underlying causes of death were determined from death certificates coded according to the 9th version of the International Classification of Diseases (ICD-9). The causes of death in the government computer files have been coded according to the ICD-9 since 1981. The diagnosis of diabetes was according to the definition of the ICD-9 code 250 or the A-code of A181. The subjects were divided into two groups, survivors (n = 1382) and non-survivors (n = 410), for compiling and comparing the medical data, and the data collected for analysis were based on the items of the laboratory examinations.
2.1.
Statistical analysis
The individual description and examination data in this study were implemented in the SPSS version 10.0 software for processing and analysis, in which continuous variables were described as mean S.D. For continuous variables in the examination, the Mann–Whitney U-test was used as comparison method. As for categorical variables, Yate’s correction of contingency or Fisher’s exact test was used. To indicate a significant difference in the statistics, p was set at <0.05. The mortality rate applied in this study and that of the general population in Taiwan (average of 1996–2002 mid-year population) were indicated as standardized mortality ratios, SMRs, in which Byar’s approximation was used to calculate the 95% CIs for SMRs [6]. Uni- and multivariate proportional hazards Cox regression models were used for survival analysis to observe the impact that each risk factor had on the
surviving patients. Specifically, after treating variables with a P < 0.05 in Cox univariate analysis and deleting multicollinearity, we computed a forward stepwise regression on the variables that remained. The p was examined by the Wald test and the results were indicated by relative risk, and 95% CIs.
3.
Results
Mortality rates were computed using a person-year denominator. The total follow-up person-years were 4399, of which 2405 were for men and 1994 for women. The number of the follow-up deaths was 410, representing 22.9% of total enrollment. The overall crude mortality rate was 93.2/1000 personyears, with 250 men at a crude mortality rate of 104.0/1000 person-years, and 160 women at a crude mortality rate of 80.2/ 1000 person-years. Table 1 depicts the total clinical characteristics and comparisons between males and females, with mean age of the patients of 62.4 15.0 years (mean S.D.). Compared with diabetic women, diabetic men were older at hospitalization, older when they died, with higher concentrations of serum creatinine, and more likely to smoke, with family diabetic history, family hypertension, and retinopathy, but with lower systolic blood pressure, as well as lower concentrations of serum cholesterol, triglyceride, cholesterol/HDL-cholesterol, LDL-cholesterol. No difference was found between men and women as far as other variables. Percentages of the cause of death and SMRs are shown in Table 2. Diabetes was the main cause of death with 156 patients (38.0%), other causes were in descending order as follows: cancer in 54 patients (13.2%), cardiopulmonary disease in 39 patients (9.5%), infection in 32 patients (7.8%), stroke in 31 patients (7.6%), digestive diseases in 28 patients (6.8%), nephropathy in 21 patients (5.1%), accidents in 8 patients (2%), suicide in 2 patients (0.5%), and disease of artery, arterioles, and capillaries in 1 patient (0.2%). Among them, the mortality of 124 patients (30.2%) could be categorized as due to diabetes-related causes, including infection in 32 patients, nephropathy in 21 patients, and cardiovascular disease in 71 patients (cardiopulmonary in 39 patients, stroke in 31 patients, artery, arterioles, and capillaries in 1 patient). Thus, causes of death ascribed to diabetes directly (38.0%) and indirectly (30.2%) totaled 68.2%. The most common non-diabetes-related cause of death was cancer (54 patients, 13.2%). Diabetes was the leading direct cause of death with SMR of 14.92 (95% CIs: 12.76–17.46), with 17.38 (95% CIs: 14.20–21.26) in men and 13.84 (95% CIs: 10.79–17.74) in women. The SMR of cardiovascular diseases was 1.97 (95% CIs: 1.57–2.49), with 1.79 (95% CIs: 1.33–2.41) in men and 2.43 (95% CIs: 1.68–3.52) in women. The SMR in total cases was 2.98 (95% CIs: 2.71–3.28), with 2.58 in men and 3.80 in women (Table 2). Table 3 shows the results of univariate Cox survival analysis in these diabetic patients with the significant factors being: hospitalized age 65 years, BMI < 25 kg/m2, duration of hypertension > 5 years, white blood count > 8000 count/mL, cholesterol < 5.17 mmol/L, triglyceride < 1.69 mmol/L, LDLcholesterol < 2.59 mmol/L, creatinine clearance < 60 mL/ min, daily protein loss > 0.3 g/day, presence of retinopathy, and family history of diabetes.
diabetes research and clinical practice 80 (2008) 449–454
451
Table 1 – Clinical characteristic analysis of hospitalized diabetic individuals Clinical characteristics Inpatient age (years) Age at death (years) Follow-up time (m) Height (cm) Weight (kg) BMI (kg/m2) SBP (mmHg) DBP (mmHg) DM-Hx (years) HTN-Hx (years) HbA1c (%) WBC (count/mL) Cholesterol (mmol/L) Triglyceride (mmol/L) HDL-C (mmol/L) Cholesterol/HDL-C LDL-C (mmol/L) Scr (mmol/L) Ccr (mL/min) DPL (g/day) Smoker N Y
Total (N = 1792)
Males (N = 999)
Females (N = 793)
P value
62.4 15.0 70.7 11.2 29.5 21.0 159.7 8.6 60.9 12.2 23.9 4.2 130.2 16.0 74.7 10.1 10.3 7.6 5.4 7.3 10.2 3.8 8551 4571 4.9 1.7 2.2 2.6 1.0 0.4 5.7 2.9 2.8 1.5 133.1 131.1 65.5 55.0 1.0 2.2
64.3 15.0 71.9 11.5 28.9 20.8 165.0 6.4 64.6 12.2 23.8 4.1 129.3 15.3 74.8 10.2 10.5 7.9 5.7 7.5 10.2 3.1 8676 4335 4.8 1.6 2.1 2.6 0.9 0.3 5.5 2.9 2.7 1.5 143.9 110.2 65.6 38.0 1.2 2.6
60.1 14.8 68.7 10.3 30.2 21.3 153.0 6.0 56.2 10.6 24.0 4.3 131.2 16.8 74.5 10.0 10.1 7.2 5.1 6.9 10.2 4.6 8384 4869 5.1 1.8 2.3 2.6 1.0 0.4 5.8 2.9 2.9 1.6 119.6 152.2 65.4 71.0 0.9 1.5
<0.001 <0.001 0.220 <0.001 <0.001 0.409 0.014 0.409 0.425 0.337 0.398 0.088 <0.001 <0.001 0.060 0.017 0.002 <0.001 0.257 0.957
m*
<0.001
y*
<0.001
y*
0.023
y*
0.012
y*
1061 (71.3%) 428 (28.7%)
357 (47.7%) 391 (52.3%)
704 (95.0%) 37 (5.0%)
Family history of DM N Y
683 (56.2%) 533 (43.8%)
399 (62.4%) 240 (37.6%)
284 (49.2%) 293 (50.8%)
Family history of HTN N Y
953 (81.7%) 213 (18.3%)
519 (84.3%) 97 (15.7%)
434 (78.9%) 116 (21.1%)
Retinopathy N Y
566 (50.4%) 557 (49.6%)
329 (53.9%) 281 (46.1%)
m* m m* m* m m* m m m m m m* m* m m* m* m* m m
237 (46.2%) 276 (53.8%)
Data: mean standard deviation. SBP, systolic blood pressure; DBP, diastolic blood pressure; DM-Hx, history of diabetes; HTN-Hx, history of hypertension; WBC, white blood count; Scr, serum creatinine; Ccr, creatinine clearance; DPL, daily protein loss; HTN, hypertension; m, Mann– Whitney U-test. Y, Yate’s correction of contingency; F, Fisher’s exact test. * P < 0.05.
Table 4 shows the predictive multivariate Cox regression models for hospitalized diabetic patient mortality. The independent mortality predictors were: hospitalized age 65 65 years old, duration of hypertension > 5 years, creatinine clearance < 60 mL/min, and daily protein loss > 0.3 g, with values of relative risk ranging between 1.44 and 8.37, respectively.
4.
Discussion
The mortality rate of diabetic patients has declined gradually in recent years, although the rate is still higher than that of the general population [7]. The subjects of this study were inpatients who had been hospitalized because of poor glucose control or related complications. Thus, the survival expectation was found to be poorer [8]. According to a previous study, the average life expectancy of diabetic patients is 5–10 years shorter than that of the general population [9]. Our data indicated that the crude mortality rate was 93.2/1000 personyears, and overall SMR was 2.98 (95% CIs: 2.71–3.28), which was three times higher than the general mortality rate. Our
findings are thus in line with the rate indicated in other studies [5]. Tseng, who collected records of diabetes patients in Taiwan [4], between 1995 and 1998, and followed them up until 2001, found that the overall SMR of the diabetes patients was 1.63 (95% CIs: 1.62–1.65). The overall SMR of the present study was higher than that of Tseng’s, probably because our study subjects were diabetes patients who had been hospitalized and, presumably, had more severe disease entity. For about 62% of diabetes patients in this study, diabetes was not registered as an underlying cause of death (Table 2). One earlier study showed that only 12.5% of all death certificates of diabetic patients listed diabetes as an underlying cause [10], and that diabetes was not even mentioned on 51% of death certificates for diabetic patients. Another study indicated that diabetes was listed as the underlying cause of death for only 7.7% of diabetic men and 13.4% of diabetic women [11]. One possibility is that at the time of the patient’s death, diabetes had not been diagnosed, and therefore diabetes was not recorded on the death certificate. Another reason may have been that the diabetes patients died from unrelated causes such as cancer. In addition, diabetes patients might also die from other chronic diseases related to diabetes
452
diabetes research and clinical practice 80 (2008) 449–454
Table 2 – Observed number (percentage) of deaths, SMRs (95% CI) and cause of death in hospitalized diabetic individuals classified according to the ICD-9 Cause of death Diabetes Cancer Cardiopulmonary disease Stroke Disease of arteries, arterioles, and capillaries Nephropathy Infection
Digestive diseases Accidents Suicide Other causes
CVD All causes
ICD-9 code
Men
Women
Total
n (%)
SMR (95% CI)
n (%)
SMR (95% CI)
n (%)
SMR (95% CI)
250 140–208 401–429
94 (37.6) 34 (13.6) 23 (9.2)
17.38 (14.20–21.26) 1.47 (1.05–2.06) 1.91 (1.27–2.87)
62 (38.8) 20 (12.5) 16 (10.0)
13.84 (10.79–17.74) 2.34 (1.51–3.61) 2.77 (1.71–4.50)
156 (38.0) 54 (13.2) 39 (9.5)
14.92 (12.76–17.46) 1.78 (1.37–2.32) 2.15 (1.58–2.95)
430–438 440–448
19 (7.6) 1 (0.4)
1.65 (1.06–2.58) 1.87 (0.33–10.59)
12 (7.5) 0 (0.0)
2.16 (1.24–3.78) –
31 (7.6) 1 (0.2)
1.80 (1.27–2.56) 1.46 (0.26–8.27)
580–589 001–139, 320, 321, 326, 421, 460–466, 480–487, 510, 513, 551, 567, 590, 599, 680–686, 711, 730 520–579 (excludes 551) 800–949 950–959 ICD-9, excluding above 401–438, 440–448 –
14 (5.6) 22 (8.8)
4.27 (2.54–7.16) 2.24 (1.48–3.39)
7 (4.4) 10 (6.3)
4.01 (1.94–8.27) 3.07 (1.67–5.66)
21 (5.1) 32 (7.8)
4.16 (2.72–6.36) 2.56 (1.81–3.61)
12 (4.8)
1.85 (1.06–3.24)
16 (10.0)
5.92 (3.65–9.62)
28 (6.8)
3.08 (2.13–4.45)
5 (2.0) 1 (0.4) 25 (10.0)
1.29 (0.55–3.03) 1.12 (0.20–6.34) 1.25 (0.85–1.85)
3 (1.9) 1 (0.6) 13 (8.1)
2.23 (0.76–6.57) 2.99 (0.53–16.96) 1.58 (0.93–2.71)
8 (2.0) 2 (0.5) 38 (9.3)
1.57 (0.80–3.10) 1.69 (0.46–6.15) 1.36 (0.99–1.87)
43 (17.2)
1.79 (1.33–2.41)
28 (17.5)
2.43 (1.68–3.52)
71 (17.3)
1.97 (1.57–2.49)
250 (100.0)
2.58 (2.28–2.92)
160 (100.0)
3.80 (3.25–4.43)
410 (100.0)
2.98 (2.71–3.28)
SMR, standardized mortality ratio; ICD-9, International Classification of Diseases, Ninth Revision; CVD, cardiovascular disease. Age ranges of SMRs: <45, 45–54, 55–64, 65–74, 75 years.
Table 3 – Results of univariate Cox survival analysis Clinical characteristics
Hazard ratio (95% CI) Total
Inpatient age (65 years) BMI (<25 kg/m2) DM-Hx (>10 years) HTN-Hx (>5 years) HbA1c (<7%) WBC (>8000 count/mL) Cholesterol (<5.17 mmol/L) Triglyceride (<1.69 mmol/L) HDL-C (<1.03 mmol/L) HDL-C (M) (1.03 mmol/L) HDL-C (F) (<1.29 mmol/L) LDL-C (<2.59 mmol/L) Scr (61.88–123.76 mmol/L) Ccr (<60 mL/min) DPL (>0.3 g/day) Smoker (Y) Family history of DM (N) Retinopathy (Y)
2.83 1.50 1.09 1.44 1.53 1.38 1.45 1.34 1.21
Males **
(2.27–3.52) (1.10–2.05)** (0.88–1.35) (1.13–1.85)** (1.16–2.01)** (1.06–1.80)* (1.15–1.83)** (1.07–1.68)** (0.93–1.57)
2.68 1.49 1.13 1.38 1.54 1.05 1.32 1.51
(1.96–3.67) (1.01–2.20)* (0.86–1.48) (1.00–1.90) (1.09–2.18)* (0.75–1.48) (0.97–1.80) (1.11–2.04)**
2.98 1.50 0.99 1.52 1.46 2.04 1.56 1.11
(2.17–4.10)** (0.90–2.48) (0.69–1.41) (1.04–2.23)* (0.92–2.32) (1.34–3.10)** (1.09–2.23)* (0.79–1.56)
1.40 1.58 1.55 3.22 2.70 1.77 1.40 1.85
(0.77–2.55) (1.09–2.29)* (1.12–2.15)** (2.01–5.14)** (1.58–4.61)** (0.93–3.37) (0.93–2.11) (1.14–3.01)*
1.12 (0.79–1.57) 1.32 1.66 3.53 2.00 1.06 1.42 1.92
(1.05–1.67)* (1.35–2.04)** (2.62–4.76)** (1.45–2.76)** (0.84–1.33) (1.09–1.86)** (1.43–2.57)**
1.13 1.76 3.79 1.68 0.81 1.33 2.09
(0.84–1.53) (1.35–2.28)** (2.57–5.58)** (1.12–2.52)* (0.62–1.08) (0.93–1.91) (1.45–3.02)**
Y, yes; N, no. HTN-Hx, history of hypertension; Ccr, creatinine clearance; DPL, daily protein loss. P < 0.05. ** P < 0.01. *
Females **
diabetes research and clinical practice 80 (2008) 449–454
Table 4 – Cox proportional hazards model analysis of survival-related predictors Risk factor
Relative risk
Patient age during hospitalization (years) <65 65
1 1.44
HTN-Hx (years) 25 >5
1 2.68
Ccr (mL/min) 60 <60
1 8.37
DPL (g/day) 20.3 >0.3
1 1.70
95% CI of RR
P valuea 0.0019
1.15–1.82 <0.0001 2.14–3.36 <0.0001 6.28–11.15 0.0002 1.29–2.23
HTN-Hx, history of hypertension; Ccr, creatinine clearance; DPL, daily protein loss; RR, relative risk. a P value by Wald statistic.
such as ischemic heart disease, stroke and kidney diseases, which may explain in part why diabetes has been underreported as the major cause of death. Therefore, it is suggested that the importance of diabetes as cause of death in public health records might be underestimated if the analysis is based on the data collected from death certificates [12]. Conversely, it is also possible that doctors in Taiwan were more inclined to list diabetes as the cause of death than those in other countries, or the coders in Taiwan were inclined to record diabetes as the main cause of death [13]. A previous study from Taiwan [4] indicates that the main causes of death of diabetes patients were diabetes (28.8%), cardiovascular diseases (19.8%), cancer (18.5%), digestive diseases (7.9%), and infection (6.4%). The three main causes of death in our study were in line with those in Tseng’s report. Our study also showed the percentage of infection to be slightly higher (7.8%), which might be because our study dealt mostly with older patients whose immune functions had deteriorated, and whose chance of infection thereby increased. A recent multinational study [14] that compared the effect of diabetes on people residing in the Asia-Pacific area (including the patients in Taiwan) found that more cases of stroke were reported as the cause of death for Asians than for Caucasian residents in New Zealand and Australia. In Japan, stroke was also more common than heart disease as the cause of death [14,15]. Our study showed that fewer diabetes patients died from stroke than from cardiopulmonary disease (including coronary heart disease). When cardiopulmonary disease, stroke, and artery, arterioles, and capillaries were combined to be presented as cardiovascular cause of death, the rate was 17.3% in our study population (a rate higher than that of cancer at 13.2%). In total, the percentage of all the diabetic patients whose death was ascribed to diabetes plus cardiovascular disease was 55.3% (17.3% + 38.0%). This figure was higher than the 49.4% in the abovementioned U.S. report [11] and 49.1% in a British report [16]. Nephropathy is another major cause of death for diabetes patients. Nephropathy represented 1.6–2.9% of the cause of
453
death among diabetes patients, a relative risk about two times that of the general population [17]. Based on the national data of the causes of death in the U.S. in 1979, Geiss et al. [18] found that the death rate of the diabetes patients who died from nephropathy was higher than that of non-diabetes patients. In our study, nephropathy reported as the cause of death was 5.1%, about 4.16 times that of the general population. Proteinuria is considered an indicator of renal disease, which is a major cause of death in diabetes patients. Nelson et al. [19] found that the death rate of type 2 diabetic patients with proteinuria was 3.5 times that of those without it. Mattock et al.’s prospective study [20] investigated the prevalence rate and risk factors of coronary heart disease of 141 type 2 diabetes patients and found that proteinuria was a major risk factor for mortality of type 2 diabetic patients. Our study indicated that proteinuria and creatinine clearance were significant predictors of mortality in diabetes inpatients. Some studies have indicated that hypertension is a predictor of rate of mortality for diabetes patients [21]. However, other reports show that hypertension is not considered to have an impact on the mortality rate of diabetes patients [22]. A follow-up cohort study certified that the unverified cause of atherosclerosis was the real cause of the increased mortality of type 2 diabetic patients [7]. In our multivariate analysis, duration of hypertension greater than 5 years was found to be an independent predictor of mortality. There are some limitations to our study. To begin with, type 1 and type 2 diabetes are shown in the fifth code in the International Classification of Diseases. However, the fifth code is rarely marked in the current disease coding, as observed in the prevailing data. In our study, a distinction between type 1 and type 2 diabetes was not made. Since only a small number of diabetes patients suffered from type 1 diabetes in this area [23], we therefore believe that the results of the study reflect the mortality rate of type 2 diabetes. In addition, the diabetes type may be wrongly classified because the cause of death is not traceable on the death certificate. Finally, the number of deaths may also be underestimated due to underreporting of death records in those who emigrated from Taiwan. In conclusion, the mortality rate of diabetes inpatients was three times higher than that of the general population in these hospitalized diabetes subjects. Among those deceased, only 38% of direct causes of death were ascribed to diabetes. The independent predictors of mortality included: age greater than 65, duration of hypertension more than 5 years, 24 h proteinuria greater than 0.3 g, and estimated creatinine clearance less than 60 mL/min. The mortality rate of diabetes patients may be reduced if the risk factors can be noted in the early period of the disease, and then strictly controlled.
Acknowledgments We wish to thank the Biostatistics Task Force of Taichung Veterans General Hospital, Taichung, Taiwan, ROC, for their help with statistical analysis. This study was supported in part by grants from the National Science Council (NSC 95-2314B075A-020-MY3), Taiwan, ROC, and Taichung Veterans General Hospital (TCVGH-953505C), Taichung, Taiwan, ROC.
454
diabetes research and clinical practice 80 (2008) 449–454
Conflict of interest All authors state that they have no conflict of interest.
references
[1] H. King, R.E. Aubert, W.H. Herman, Global burden of diabetes, 1995–2025: prevalence, numerical estimates, and projections, Diabetes Care 21 (1998) 1414–1431. [2] T. Tai, Current status of diabetes in Taiwan, Diabetes Res. Clin. Pract. 50 (2000) S1–S2. [3] T. Lin, P. Chou, M.S. Lai, S.T. Tsai, T.Y. Tai, Direct costs-ofillness of patients with diabetes mellitus in Taiwan, Diabetes Res. Clin. Pract. 54 (2001) S43–S46. [4] C.H. Tseng, Mortality and causes of death in a national sample of diabetic patients in Taiwan, Diabetes Care 27 (2004) 1605–1609. [5] G.F. Salles, K.V. Bloch, C.R. Cardoso, Mortality and predictors of mortality in a cohort of Brazilian type 2 diabetic patients, Diabetes Care 27 (2004) 1299–1305. [6] N.E. Breslow, N.E. Day, Statistical Method in Cancer Research: The Design and Analysis of Cohort Studies, IARC Scientific, Lyon, France, 1987. [7] S.E. Moss, R. Klein, B.E. Klein, Cause-specific mortality in a population-based study of diabetes, Am. J. Public Health 81 (1991) 1158–1162. [8] I.I. Kessler, Mortality experience of diabetic patients. A twenty-six-year follow-up study, Am. J. Med. 51 (1971) 715– 724. [9] G.S. Bale, P.S. Entmacher, Estimated life expectancy of diabetics, Diabetes 26 (1977) 434–438. [10] A.P. Bender, J.M. Sprafka, H.G. Jagger, K.H. Muckala, C.P. Martin, T.R. Edwards, Incidence, prevalence, and mortality of diabetes mellitus in Wadena, Marshall, and Grand Rapids, Minnesota: the Three-City Study, Diabetes Care 9 (1986) 343–350. [11] K. Gu, C.C. Cowie, M.I. Harris, Mortality in adults with and without diabetes in a national cohort of the U.S. population, 1971–1993, Diabetes Care 21 (1998) 1138–1145.
[12] G. Panzram, Mortality and survival in type 2 (non-insulindependent) diabetes mellitus, Diabetologia 30 (1987) 123– 131. [13] T.H. Lu, P.Y. Hsu, R.N. Anderson, C.N. Huang, Mortality and causes of death in a national sample of diabetic patients in Taiwan: response to Tseng, Diabetes Care 28 (2005) 1266– 1267. [14] M. Woodward, X. Zhang, F. Barzi, W. Pan, H. Ueshima, A. Rodgers, et al., The effects of diabetes on the risks of major cardiovascular diseases and death in the Asia-Pacific region, Diabetes Care 26 (2003) 360–366. [15] M. Katakura, M. Naka, T. Kondo, N. Nishii, M. Komatsu, Y. Sato, et al., Prospective analysis of mortality, morbidity, and risk factors in elderly diabetic subjects: Nagano study, Diabetes Care 26 (2003) 638–644. [16] C.L. Morgan, C.J. Currie, J.R. Peters, Relationship between diabetes and mortality: a population study using record linkage, Diabetes Care 23 (2000) 1103–1107. [17] M.I. Uusitupa, L.K. Niskanen, O. Siitonen, E. Voutilainen, K. Pyorala, Ten-year cardiovascular mortality in relation to risk factors and abnormalities in lipoprotein composition in type 2 (non-insulin-dependent) diabetic and nondiabetic subjects, Diabetologia 36 (1993) 1175–1184. [18] L.S. Geiss, W.H. Herman, S.M. Teutsch, Diabetes and renal mortality in the United States, Am. J. Public Health 75 (1985) 1325–1326. [19] R.G. Nelson, D.J. Pettitt, M.J. Carraher, H.R. Baird, W.C. Knowler, Effect of proteinuria on mortality in NIDDM, Diabetes 37 (1988) 1499–1504. [20] M.B. Mattock, N.J. Morrish, G. Viberti, H. Keen, A.P. Fitzgerald, G. Jackson, Prospective study of microalbuminuria as predictor of mortality in NIDDM, Diabetes 41 (1992) 736–741. [21] E. Barret Connor, M.H. Criqui, M.R. Klauber, M. Holdbrook, Diabetes and hypertension in a community of older adults, Am. J. Epidemiol. 113 (1981) 276–284. [22] W.B. Kannel, D.L. McGee, Diabetes and cardiovascular risk factors: the Framingham study, Circulation 59 (1979) 8–13. [23] A.F. Amos, D.J. McCarty, P. Zimmet, The rising global burden of diabetes and its complications: estimates and projections to the year 2010, Diabet. Med. 14 (1997) S1–S85.
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/diabres
Predictors of mortality in hospitalized diabetic patients: A 7-year prospective study Mei-Chun Wu a, Wen-Jane Lee b, Scheng-Ming Tschen c, Shih-Yi Lin d,e,f, I-Te Lee d,e,f, Chii-Yuan Jeng d,e,f, Li-Nien Tseng d,e,f, Ying-Mei Liu a, Been-Yuan Chen a, Wayne H-H Sheu c,d,e,f,* a
Department of Pharmacy, Taichung Veterans General Hospital, Taichung, Taiwan Department of Medical Education and Research, Taichung Veterans General Hospital, Taichung, Taiwan c Department of Life Sciences, National Chung-Hsing University, Taichung, Taiwan d Division of Endocrinology and Metabolism, Department of Internal Medicine, Taichung Veterans, General Hospital, Taichung, Taiwan e Institute of Medical Technology, National Chung-Hsing University, Taichung, Taiwan f College of Medicine, National Yang-Ming University, Taipei, Taiwan b
article info
abstract
Article history:
This study was conducted to investigate the mortality rate, causes of death, and standar-
Received 7 January 2007
dized mortality ratio (SMR), and to identify the significant predictive factors of mortality in
Accepted 29 November 2007
diabetic patients at a medical center in Taiwan. Clinical data were obtained from 1792
Published on line 4 March 2008
diabetic inpatients discharged from the metabolism department of a medical center during the years 1996–2002. Underlying causes of death were determined from death certificates.
Keywords:
Predictors of mortality were assessed by uni- and multivariate Cox survival analyses. Of
Diabetes mellitus
1792 patients studied, 410 (22.9%) patients died. The crude mortality rate was 93.2/1000
Crude mortality rate
person-years, and the overall SMR was 2.98 (2.71–3.28). The percentages of causes of death
Standardized mortality ratios (SMRs)
ascribed to diabetes, cancer, cardiopulmonary disease, infection, stroke, digestive diseases,
Predictive factors
nephropathy, accidents, suicide, and disease of arteries, arterioles, and capillaries were 38.0, 13.2, 9.5, 7.8, 7.6, 6.8, 5.1, 2.0, 0.5, and 0.2%, respectively. The independent predictors of mortality were age greater than 65, duration of hypertension more than 5 years, 24 h proteinuria greater than 0.3 g, and estimated creatinine clearance less than 60 mL/min. Conclusion: The mortality of diabetic inpatients was about threefold that of the general population. The predictors of mortality included older age, longer duration of hypertension, increased 24 h proteinuria, and decreased creatinine clearance. # 2008 Elsevier Ireland Ltd. All rights reserved.
1.
Introduction
Diabetes has become endemic worldwide [1]. There are up to one million diabetic patients in the Taiwan area alone [2] and the average expense incurred for every
diabetic patient is 4.3 times that for the non-diabetic patient [3]. It is also estimated that the total medical expenditure would be reduced by 20% if the rate of poor glycemic control could be reduced to one-third of the current level [3].
* Corresponding author at: No. 160, Section 3, Chung-Kang Road, Division of Endocrinology and Metabolism, Taichung Veterans General Hospital, Taichung 407, Taiwan. Tel.: +886 4 2374 1300; fax: +886 4 2374 1318. E-mail address: [email protected] (W.-H. Sheu). 0168-8227/$ – see front matter # 2008 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.diabres.2007.11.020
450
diabetes research and clinical practice 80 (2008) 449–454
A long-term study of the mortality and the predictive factors of mortality of diabetic patients found that the mortality rate of diabetic patients in Taiwan was 1.63 times that of the general population [4]. Another long-term study found that the mortality rate of Brazilian type 2 diabetic outpatients was 3.36 times that of the general population, and the predictive factors for mortality in these diabetic individuals included old age, increased 24 h proteinuria, preexisting vascular disease, and decreased HDL cholesterol [5]. To the best of our knowledge, there have not been any followup studies on the mortality rate and mortality factors of hospitalized diabetic patients. Therefore, we collected the clinical characteristics of a group of diabetic patients after their discharge from hospital in order to investigate their longterm mortality rate and the predictive factors of mortality.
2.
Materials and methods
This study protocol was approved and supported by the Institutional Review Board of Taichung Veterans General Hospital (TCVGH), Taiwan. Data of study subjects were collected from 1841 consecutive hospitalized diabetic inpatients from 1 January 1996 to 31 December 2002, excluding the 49 patients who died either in hospital or within 1 month after their discharge from hospital, with the remaining 1792 as the subjects of the study (999 men and 793 women). The follow-up period lasted from the date when the patient entered the study till the date of death or until 31 December 2002, with median follow-up time of 27 months. The list of the 1792 subjects’ names was linked to the national mortality database to ascertain the death records before 31 December 2002. Underlying causes of death were determined from death certificates coded according to the 9th version of the International Classification of Diseases (ICD-9). The causes of death in the government computer files have been coded according to the ICD-9 since 1981. The diagnosis of diabetes was according to the definition of the ICD-9 code 250 or the A-code of A181. The subjects were divided into two groups, survivors (n = 1382) and non-survivors (n = 410), for compiling and comparing the medical data, and the data collected for analysis were based on the items of the laboratory examinations.
2.1.
Statistical analysis
The individual description and examination data in this study were implemented in the SPSS version 10.0 software for processing and analysis, in which continuous variables were described as mean S.D. For continuous variables in the examination, the Mann–Whitney U-test was used as comparison method. As for categorical variables, Yate’s correction of contingency or Fisher’s exact test was used. To indicate a significant difference in the statistics, p was set at <0.05. The mortality rate applied in this study and that of the general population in Taiwan (average of 1996–2002 mid-year population) were indicated as standardized mortality ratios, SMRs, in which Byar’s approximation was used to calculate the 95% CIs for SMRs [6]. Uni- and multivariate proportional hazards Cox regression models were used for survival analysis to observe the impact that each risk factor had on the
surviving patients. Specifically, after treating variables with a P < 0.05 in Cox univariate analysis and deleting multicollinearity, we computed a forward stepwise regression on the variables that remained. The p was examined by the Wald test and the results were indicated by relative risk, and 95% CIs.
3.
Results
Mortality rates were computed using a person-year denominator. The total follow-up person-years were 4399, of which 2405 were for men and 1994 for women. The number of the follow-up deaths was 410, representing 22.9% of total enrollment. The overall crude mortality rate was 93.2/1000 personyears, with 250 men at a crude mortality rate of 104.0/1000 person-years, and 160 women at a crude mortality rate of 80.2/ 1000 person-years. Table 1 depicts the total clinical characteristics and comparisons between males and females, with mean age of the patients of 62.4 15.0 years (mean S.D.). Compared with diabetic women, diabetic men were older at hospitalization, older when they died, with higher concentrations of serum creatinine, and more likely to smoke, with family diabetic history, family hypertension, and retinopathy, but with lower systolic blood pressure, as well as lower concentrations of serum cholesterol, triglyceride, cholesterol/HDL-cholesterol, LDL-cholesterol. No difference was found between men and women as far as other variables. Percentages of the cause of death and SMRs are shown in Table 2. Diabetes was the main cause of death with 156 patients (38.0%), other causes were in descending order as follows: cancer in 54 patients (13.2%), cardiopulmonary disease in 39 patients (9.5%), infection in 32 patients (7.8%), stroke in 31 patients (7.6%), digestive diseases in 28 patients (6.8%), nephropathy in 21 patients (5.1%), accidents in 8 patients (2%), suicide in 2 patients (0.5%), and disease of artery, arterioles, and capillaries in 1 patient (0.2%). Among them, the mortality of 124 patients (30.2%) could be categorized as due to diabetes-related causes, including infection in 32 patients, nephropathy in 21 patients, and cardiovascular disease in 71 patients (cardiopulmonary in 39 patients, stroke in 31 patients, artery, arterioles, and capillaries in 1 patient). Thus, causes of death ascribed to diabetes directly (38.0%) and indirectly (30.2%) totaled 68.2%. The most common non-diabetes-related cause of death was cancer (54 patients, 13.2%). Diabetes was the leading direct cause of death with SMR of 14.92 (95% CIs: 12.76–17.46), with 17.38 (95% CIs: 14.20–21.26) in men and 13.84 (95% CIs: 10.79–17.74) in women. The SMR of cardiovascular diseases was 1.97 (95% CIs: 1.57–2.49), with 1.79 (95% CIs: 1.33–2.41) in men and 2.43 (95% CIs: 1.68–3.52) in women. The SMR in total cases was 2.98 (95% CIs: 2.71–3.28), with 2.58 in men and 3.80 in women (Table 2). Table 3 shows the results of univariate Cox survival analysis in these diabetic patients with the significant factors being: hospitalized age 65 years, BMI < 25 kg/m2, duration of hypertension > 5 years, white blood count > 8000 count/mL, cholesterol < 5.17 mmol/L, triglyceride < 1.69 mmol/L, LDLcholesterol < 2.59 mmol/L, creatinine clearance < 60 mL/ min, daily protein loss > 0.3 g/day, presence of retinopathy, and family history of diabetes.
diabetes research and clinical practice 80 (2008) 449–454
451
Table 1 – Clinical characteristic analysis of hospitalized diabetic individuals Clinical characteristics Inpatient age (years) Age at death (years) Follow-up time (m) Height (cm) Weight (kg) BMI (kg/m2) SBP (mmHg) DBP (mmHg) DM-Hx (years) HTN-Hx (years) HbA1c (%) WBC (count/mL) Cholesterol (mmol/L) Triglyceride (mmol/L) HDL-C (mmol/L) Cholesterol/HDL-C LDL-C (mmol/L) Scr (mmol/L) Ccr (mL/min) DPL (g/day) Smoker N Y
Total (N = 1792)
Males (N = 999)
Females (N = 793)
P value
62.4 15.0 70.7 11.2 29.5 21.0 159.7 8.6 60.9 12.2 23.9 4.2 130.2 16.0 74.7 10.1 10.3 7.6 5.4 7.3 10.2 3.8 8551 4571 4.9 1.7 2.2 2.6 1.0 0.4 5.7 2.9 2.8 1.5 133.1 131.1 65.5 55.0 1.0 2.2
64.3 15.0 71.9 11.5 28.9 20.8 165.0 6.4 64.6 12.2 23.8 4.1 129.3 15.3 74.8 10.2 10.5 7.9 5.7 7.5 10.2 3.1 8676 4335 4.8 1.6 2.1 2.6 0.9 0.3 5.5 2.9 2.7 1.5 143.9 110.2 65.6 38.0 1.2 2.6
60.1 14.8 68.7 10.3 30.2 21.3 153.0 6.0 56.2 10.6 24.0 4.3 131.2 16.8 74.5 10.0 10.1 7.2 5.1 6.9 10.2 4.6 8384 4869 5.1 1.8 2.3 2.6 1.0 0.4 5.8 2.9 2.9 1.6 119.6 152.2 65.4 71.0 0.9 1.5
<0.001 <0.001 0.220 <0.001 <0.001 0.409 0.014 0.409 0.425 0.337 0.398 0.088 <0.001 <0.001 0.060 0.017 0.002 <0.001 0.257 0.957
m*
<0.001
y*
<0.001
y*
0.023
y*
0.012
y*
1061 (71.3%) 428 (28.7%)
357 (47.7%) 391 (52.3%)
704 (95.0%) 37 (5.0%)
Family history of DM N Y
683 (56.2%) 533 (43.8%)
399 (62.4%) 240 (37.6%)
284 (49.2%) 293 (50.8%)
Family history of HTN N Y
953 (81.7%) 213 (18.3%)
519 (84.3%) 97 (15.7%)
434 (78.9%) 116 (21.1%)
Retinopathy N Y
566 (50.4%) 557 (49.6%)
329 (53.9%) 281 (46.1%)
m* m m* m* m m* m m m m m m* m* m m* m* m* m m
237 (46.2%) 276 (53.8%)
Data: mean standard deviation. SBP, systolic blood pressure; DBP, diastolic blood pressure; DM-Hx, history of diabetes; HTN-Hx, history of hypertension; WBC, white blood count; Scr, serum creatinine; Ccr, creatinine clearance; DPL, daily protein loss; HTN, hypertension; m, Mann– Whitney U-test. Y, Yate’s correction of contingency; F, Fisher’s exact test. * P < 0.05.
Table 4 shows the predictive multivariate Cox regression models for hospitalized diabetic patient mortality. The independent mortality predictors were: hospitalized age 65 65 years old, duration of hypertension > 5 years, creatinine clearance < 60 mL/min, and daily protein loss > 0.3 g, with values of relative risk ranging between 1.44 and 8.37, respectively.
4.
Discussion
The mortality rate of diabetic patients has declined gradually in recent years, although the rate is still higher than that of the general population [7]. The subjects of this study were inpatients who had been hospitalized because of poor glucose control or related complications. Thus, the survival expectation was found to be poorer [8]. According to a previous study, the average life expectancy of diabetic patients is 5–10 years shorter than that of the general population [9]. Our data indicated that the crude mortality rate was 93.2/1000 personyears, and overall SMR was 2.98 (95% CIs: 2.71–3.28), which was three times higher than the general mortality rate. Our
findings are thus in line with the rate indicated in other studies [5]. Tseng, who collected records of diabetes patients in Taiwan [4], between 1995 and 1998, and followed them up until 2001, found that the overall SMR of the diabetes patients was 1.63 (95% CIs: 1.62–1.65). The overall SMR of the present study was higher than that of Tseng’s, probably because our study subjects were diabetes patients who had been hospitalized and, presumably, had more severe disease entity. For about 62% of diabetes patients in this study, diabetes was not registered as an underlying cause of death (Table 2). One earlier study showed that only 12.5% of all death certificates of diabetic patients listed diabetes as an underlying cause [10], and that diabetes was not even mentioned on 51% of death certificates for diabetic patients. Another study indicated that diabetes was listed as the underlying cause of death for only 7.7% of diabetic men and 13.4% of diabetic women [11]. One possibility is that at the time of the patient’s death, diabetes had not been diagnosed, and therefore diabetes was not recorded on the death certificate. Another reason may have been that the diabetes patients died from unrelated causes such as cancer. In addition, diabetes patients might also die from other chronic diseases related to diabetes
452
diabetes research and clinical practice 80 (2008) 449–454
Table 2 – Observed number (percentage) of deaths, SMRs (95% CI) and cause of death in hospitalized diabetic individuals classified according to the ICD-9 Cause of death Diabetes Cancer Cardiopulmonary disease Stroke Disease of arteries, arterioles, and capillaries Nephropathy Infection
Digestive diseases Accidents Suicide Other causes
CVD All causes
ICD-9 code
Men
Women
Total
n (%)
SMR (95% CI)
n (%)
SMR (95% CI)
n (%)
SMR (95% CI)
250 140–208 401–429
94 (37.6) 34 (13.6) 23 (9.2)
17.38 (14.20–21.26) 1.47 (1.05–2.06) 1.91 (1.27–2.87)
62 (38.8) 20 (12.5) 16 (10.0)
13.84 (10.79–17.74) 2.34 (1.51–3.61) 2.77 (1.71–4.50)
156 (38.0) 54 (13.2) 39 (9.5)
14.92 (12.76–17.46) 1.78 (1.37–2.32) 2.15 (1.58–2.95)
430–438 440–448
19 (7.6) 1 (0.4)
1.65 (1.06–2.58) 1.87 (0.33–10.59)
12 (7.5) 0 (0.0)
2.16 (1.24–3.78) –
31 (7.6) 1 (0.2)
1.80 (1.27–2.56) 1.46 (0.26–8.27)
580–589 001–139, 320, 321, 326, 421, 460–466, 480–487, 510, 513, 551, 567, 590, 599, 680–686, 711, 730 520–579 (excludes 551) 800–949 950–959 ICD-9, excluding above 401–438, 440–448 –
14 (5.6) 22 (8.8)
4.27 (2.54–7.16) 2.24 (1.48–3.39)
7 (4.4) 10 (6.3)
4.01 (1.94–8.27) 3.07 (1.67–5.66)
21 (5.1) 32 (7.8)
4.16 (2.72–6.36) 2.56 (1.81–3.61)
12 (4.8)
1.85 (1.06–3.24)
16 (10.0)
5.92 (3.65–9.62)
28 (6.8)
3.08 (2.13–4.45)
5 (2.0) 1 (0.4) 25 (10.0)
1.29 (0.55–3.03) 1.12 (0.20–6.34) 1.25 (0.85–1.85)
3 (1.9) 1 (0.6) 13 (8.1)
2.23 (0.76–6.57) 2.99 (0.53–16.96) 1.58 (0.93–2.71)
8 (2.0) 2 (0.5) 38 (9.3)
1.57 (0.80–3.10) 1.69 (0.46–6.15) 1.36 (0.99–1.87)
43 (17.2)
1.79 (1.33–2.41)
28 (17.5)
2.43 (1.68–3.52)
71 (17.3)
1.97 (1.57–2.49)
250 (100.0)
2.58 (2.28–2.92)
160 (100.0)
3.80 (3.25–4.43)
410 (100.0)
2.98 (2.71–3.28)
SMR, standardized mortality ratio; ICD-9, International Classification of Diseases, Ninth Revision; CVD, cardiovascular disease. Age ranges of SMRs: <45, 45–54, 55–64, 65–74, 75 years.
Table 3 – Results of univariate Cox survival analysis Clinical characteristics
Hazard ratio (95% CI) Total
Inpatient age (65 years) BMI (<25 kg/m2) DM-Hx (>10 years) HTN-Hx (>5 years) HbA1c (<7%) WBC (>8000 count/mL) Cholesterol (<5.17 mmol/L) Triglyceride (<1.69 mmol/L) HDL-C (<1.03 mmol/L) HDL-C (M) (1.03 mmol/L) HDL-C (F) (<1.29 mmol/L) LDL-C (<2.59 mmol/L) Scr (61.88–123.76 mmol/L) Ccr (<60 mL/min) DPL (>0.3 g/day) Smoker (Y) Family history of DM (N) Retinopathy (Y)
2.83 1.50 1.09 1.44 1.53 1.38 1.45 1.34 1.21
Males **
(2.27–3.52) (1.10–2.05)** (0.88–1.35) (1.13–1.85)** (1.16–2.01)** (1.06–1.80)* (1.15–1.83)** (1.07–1.68)** (0.93–1.57)
2.68 1.49 1.13 1.38 1.54 1.05 1.32 1.51
(1.96–3.67) (1.01–2.20)* (0.86–1.48) (1.00–1.90) (1.09–2.18)* (0.75–1.48) (0.97–1.80) (1.11–2.04)**
2.98 1.50 0.99 1.52 1.46 2.04 1.56 1.11
(2.17–4.10)** (0.90–2.48) (0.69–1.41) (1.04–2.23)* (0.92–2.32) (1.34–3.10)** (1.09–2.23)* (0.79–1.56)
1.40 1.58 1.55 3.22 2.70 1.77 1.40 1.85
(0.77–2.55) (1.09–2.29)* (1.12–2.15)** (2.01–5.14)** (1.58–4.61)** (0.93–3.37) (0.93–2.11) (1.14–3.01)*
1.12 (0.79–1.57) 1.32 1.66 3.53 2.00 1.06 1.42 1.92
(1.05–1.67)* (1.35–2.04)** (2.62–4.76)** (1.45–2.76)** (0.84–1.33) (1.09–1.86)** (1.43–2.57)**
1.13 1.76 3.79 1.68 0.81 1.33 2.09
(0.84–1.53) (1.35–2.28)** (2.57–5.58)** (1.12–2.52)* (0.62–1.08) (0.93–1.91) (1.45–3.02)**
Y, yes; N, no. HTN-Hx, history of hypertension; Ccr, creatinine clearance; DPL, daily protein loss. P < 0.05. ** P < 0.01. *
Females **
diabetes research and clinical practice 80 (2008) 449–454
Table 4 – Cox proportional hazards model analysis of survival-related predictors Risk factor
Relative risk
Patient age during hospitalization (years) <65 65
1 1.44
HTN-Hx (years) 25 >5
1 2.68
Ccr (mL/min) 60 <60
1 8.37
DPL (g/day) 20.3 >0.3
1 1.70
95% CI of RR
P valuea 0.0019
1.15–1.82 <0.0001 2.14–3.36 <0.0001 6.28–11.15 0.0002 1.29–2.23
HTN-Hx, history of hypertension; Ccr, creatinine clearance; DPL, daily protein loss; RR, relative risk. a P value by Wald statistic.
such as ischemic heart disease, stroke and kidney diseases, which may explain in part why diabetes has been underreported as the major cause of death. Therefore, it is suggested that the importance of diabetes as cause of death in public health records might be underestimated if the analysis is based on the data collected from death certificates [12]. Conversely, it is also possible that doctors in Taiwan were more inclined to list diabetes as the cause of death than those in other countries, or the coders in Taiwan were inclined to record diabetes as the main cause of death [13]. A previous study from Taiwan [4] indicates that the main causes of death of diabetes patients were diabetes (28.8%), cardiovascular diseases (19.8%), cancer (18.5%), digestive diseases (7.9%), and infection (6.4%). The three main causes of death in our study were in line with those in Tseng’s report. Our study also showed the percentage of infection to be slightly higher (7.8%), which might be because our study dealt mostly with older patients whose immune functions had deteriorated, and whose chance of infection thereby increased. A recent multinational study [14] that compared the effect of diabetes on people residing in the Asia-Pacific area (including the patients in Taiwan) found that more cases of stroke were reported as the cause of death for Asians than for Caucasian residents in New Zealand and Australia. In Japan, stroke was also more common than heart disease as the cause of death [14,15]. Our study showed that fewer diabetes patients died from stroke than from cardiopulmonary disease (including coronary heart disease). When cardiopulmonary disease, stroke, and artery, arterioles, and capillaries were combined to be presented as cardiovascular cause of death, the rate was 17.3% in our study population (a rate higher than that of cancer at 13.2%). In total, the percentage of all the diabetic patients whose death was ascribed to diabetes plus cardiovascular disease was 55.3% (17.3% + 38.0%). This figure was higher than the 49.4% in the abovementioned U.S. report [11] and 49.1% in a British report [16]. Nephropathy is another major cause of death for diabetes patients. Nephropathy represented 1.6–2.9% of the cause of
453
death among diabetes patients, a relative risk about two times that of the general population [17]. Based on the national data of the causes of death in the U.S. in 1979, Geiss et al. [18] found that the death rate of the diabetes patients who died from nephropathy was higher than that of non-diabetes patients. In our study, nephropathy reported as the cause of death was 5.1%, about 4.16 times that of the general population. Proteinuria is considered an indicator of renal disease, which is a major cause of death in diabetes patients. Nelson et al. [19] found that the death rate of type 2 diabetic patients with proteinuria was 3.5 times that of those without it. Mattock et al.’s prospective study [20] investigated the prevalence rate and risk factors of coronary heart disease of 141 type 2 diabetes patients and found that proteinuria was a major risk factor for mortality of type 2 diabetic patients. Our study indicated that proteinuria and creatinine clearance were significant predictors of mortality in diabetes inpatients. Some studies have indicated that hypertension is a predictor of rate of mortality for diabetes patients [21]. However, other reports show that hypertension is not considered to have an impact on the mortality rate of diabetes patients [22]. A follow-up cohort study certified that the unverified cause of atherosclerosis was the real cause of the increased mortality of type 2 diabetic patients [7]. In our multivariate analysis, duration of hypertension greater than 5 years was found to be an independent predictor of mortality. There are some limitations to our study. To begin with, type 1 and type 2 diabetes are shown in the fifth code in the International Classification of Diseases. However, the fifth code is rarely marked in the current disease coding, as observed in the prevailing data. In our study, a distinction between type 1 and type 2 diabetes was not made. Since only a small number of diabetes patients suffered from type 1 diabetes in this area [23], we therefore believe that the results of the study reflect the mortality rate of type 2 diabetes. In addition, the diabetes type may be wrongly classified because the cause of death is not traceable on the death certificate. Finally, the number of deaths may also be underestimated due to underreporting of death records in those who emigrated from Taiwan. In conclusion, the mortality rate of diabetes inpatients was three times higher than that of the general population in these hospitalized diabetes subjects. Among those deceased, only 38% of direct causes of death were ascribed to diabetes. The independent predictors of mortality included: age greater than 65, duration of hypertension more than 5 years, 24 h proteinuria greater than 0.3 g, and estimated creatinine clearance less than 60 mL/min. The mortality rate of diabetes patients may be reduced if the risk factors can be noted in the early period of the disease, and then strictly controlled.
Acknowledgments We wish to thank the Biostatistics Task Force of Taichung Veterans General Hospital, Taichung, Taiwan, ROC, for their help with statistical analysis. This study was supported in part by grants from the National Science Council (NSC 95-2314B075A-020-MY3), Taiwan, ROC, and Taichung Veterans General Hospital (TCVGH-953505C), Taichung, Taiwan, ROC.
454
diabetes research and clinical practice 80 (2008) 449–454
Conflict of interest All authors state that they have no conflict of interest.
references
[1] H. King, R.E. Aubert, W.H. Herman, Global burden of diabetes, 1995–2025: prevalence, numerical estimates, and projections, Diabetes Care 21 (1998) 1414–1431. [2] T. Tai, Current status of diabetes in Taiwan, Diabetes Res. Clin. Pract. 50 (2000) S1–S2. [3] T. Lin, P. Chou, M.S. Lai, S.T. Tsai, T.Y. Tai, Direct costs-ofillness of patients with diabetes mellitus in Taiwan, Diabetes Res. Clin. Pract. 54 (2001) S43–S46. [4] C.H. Tseng, Mortality and causes of death in a national sample of diabetic patients in Taiwan, Diabetes Care 27 (2004) 1605–1609. [5] G.F. Salles, K.V. Bloch, C.R. Cardoso, Mortality and predictors of mortality in a cohort of Brazilian type 2 diabetic patients, Diabetes Care 27 (2004) 1299–1305. [6] N.E. Breslow, N.E. Day, Statistical Method in Cancer Research: The Design and Analysis of Cohort Studies, IARC Scientific, Lyon, France, 1987. [7] S.E. Moss, R. Klein, B.E. Klein, Cause-specific mortality in a population-based study of diabetes, Am. J. Public Health 81 (1991) 1158–1162. [8] I.I. Kessler, Mortality experience of diabetic patients. A twenty-six-year follow-up study, Am. J. Med. 51 (1971) 715– 724. [9] G.S. Bale, P.S. Entmacher, Estimated life expectancy of diabetics, Diabetes 26 (1977) 434–438. [10] A.P. Bender, J.M. Sprafka, H.G. Jagger, K.H. Muckala, C.P. Martin, T.R. Edwards, Incidence, prevalence, and mortality of diabetes mellitus in Wadena, Marshall, and Grand Rapids, Minnesota: the Three-City Study, Diabetes Care 9 (1986) 343–350. [11] K. Gu, C.C. Cowie, M.I. Harris, Mortality in adults with and without diabetes in a national cohort of the U.S. population, 1971–1993, Diabetes Care 21 (1998) 1138–1145.
[12] G. Panzram, Mortality and survival in type 2 (non-insulindependent) diabetes mellitus, Diabetologia 30 (1987) 123– 131. [13] T.H. Lu, P.Y. Hsu, R.N. Anderson, C.N. Huang, Mortality and causes of death in a national sample of diabetic patients in Taiwan: response to Tseng, Diabetes Care 28 (2005) 1266– 1267. [14] M. Woodward, X. Zhang, F. Barzi, W. Pan, H. Ueshima, A. Rodgers, et al., The effects of diabetes on the risks of major cardiovascular diseases and death in the Asia-Pacific region, Diabetes Care 26 (2003) 360–366. [15] M. Katakura, M. Naka, T. Kondo, N. Nishii, M. Komatsu, Y. Sato, et al., Prospective analysis of mortality, morbidity, and risk factors in elderly diabetic subjects: Nagano study, Diabetes Care 26 (2003) 638–644. [16] C.L. Morgan, C.J. Currie, J.R. Peters, Relationship between diabetes and mortality: a population study using record linkage, Diabetes Care 23 (2000) 1103–1107. [17] M.I. Uusitupa, L.K. Niskanen, O. Siitonen, E. Voutilainen, K. Pyorala, Ten-year cardiovascular mortality in relation to risk factors and abnormalities in lipoprotein composition in type 2 (non-insulin-dependent) diabetic and nondiabetic subjects, Diabetologia 36 (1993) 1175–1184. [18] L.S. Geiss, W.H. Herman, S.M. Teutsch, Diabetes and renal mortality in the United States, Am. J. Public Health 75 (1985) 1325–1326. [19] R.G. Nelson, D.J. Pettitt, M.J. Carraher, H.R. Baird, W.C. Knowler, Effect of proteinuria on mortality in NIDDM, Diabetes 37 (1988) 1499–1504. [20] M.B. Mattock, N.J. Morrish, G. Viberti, H. Keen, A.P. Fitzgerald, G. Jackson, Prospective study of microalbuminuria as predictor of mortality in NIDDM, Diabetes 41 (1992) 736–741. [21] E. Barret Connor, M.H. Criqui, M.R. Klauber, M. Holdbrook, Diabetes and hypertension in a community of older adults, Am. J. Epidemiol. 113 (1981) 276–284. [22] W.B. Kannel, D.L. McGee, Diabetes and cardiovascular risk factors: the Framingham study, Circulation 59 (1979) 8–13. [23] A.F. Amos, D.J. McCarty, P. Zimmet, The rising global burden of diabetes and its complications: estimates and projections to the year 2010, Diabet. Med. 14 (1997) S1–S85.