Journal of Diabetes and Its Complications 29 (2015) 488–496
Contents lists available at ScienceDirect
Journal of Diabetes and Its Complications journal homepage: WWW.JDCJOURNAL.COM
Increasing body mass index identifies Chinese patients with type 2 diabetes mellitus at risk of poor outcomes Linong Ji a, Dajin Zou b, Li Liu c, Lei Qian c,⁎, Zbigniew Kadziola d, Steven Babineaux e, He Na Zhang c, Robert Wood f a
Department of Endocrinology and Metabolism, Peking University People's Hospital, Beijing, China Endocrinology Department, Affiliated Changhai Hospital to the 2nd Military Medical University, Shanghai, China c Medical Division, Lilly Suzhou Pharmaceutical Co., Ltd, Shanghai, China d Eli Lilly and Co., Real World Analytics, Vienna, Austria e Eli Lilly and Co., Indianapolis, IN, USA f Adelphi Real World, Bollington, Cheshire, UK b
a r t i c l e
i n f o
Article history: Received 24 November 2014 Received in revised form 14 February 2015 Accepted 26 February 2015 Available online 6 March 2015 Keywords: Body mass index China Diabetes mellitus Obesity Quality of life
a b s t r a c t Aims: Association between body mass index (BMI) and glycemic control, comorbidities/complications, and health-related quality of life (HRQoL) was assessed in Chinese patients with type 2 diabetes mellitus (T2DM) enrolled in the Diabetes Disease Specific Programme. Methods: Surveys of 200 physicians and 2052 patients with T2DM captured demographic, clinical, and HRQoL information. Adjusted and unadjusted analyses were conducted across 3 BMI groups; normal (18.5–b24.0, n = 998), overweight (24.0–b28.0, n = 822), and obese (≥28.0, n = 212). Results: There were no between group differences in the achievement of glycated hemoglobin (HbA1c) b 7.0% (48 mmol/mol); however, compared with the normal BMI group, more obese patients had an HbA1c N 9.0% (75 mmol/mol; 4.3% vs 10.2%, P = 0.002). More obese patients compared with normal BMI patients had hypertension (48.6% vs 35.3%, P b 0.001), dyslipidemia (35.4% vs 18.8%, P b 0.001), or both hypertension and dyslipidemia (24.1% vs 13.9%, P b 0.001). Patients in the obese group reported worse HRQoL and greater effects of diabetes on their daily living. Conclusions: Obesity in Chinese patients with T2DM results in poor glycemic control, more comorbidities, and worse HRQoL. Management of these patients should include efforts to reduce weight. Selection of weight-neutral or weight-reducing anti-diabetic medications maybe useful in these patients. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Diabetes is an increasingly prevalent health issue in China that carries a significant health and economic burden (International Diabetes Federation, 2013b; Li et al., 2012; Smith, Vlachopioti, & Colclough, 2010). The number of adults in China with diabetes was
Conflicts of Interest: L.L., L.Q., and H.Z. are employees of Lilly Suzhou Pharmaceutical Company, Ltd. Z.K. and S.B. are employees of Eli Lilly and Co. R.W. is an employee of Adelphi Real World. S.B. and Z.K. own shares in Eli Lilly and Co. L.J. has received honoraria for speaking engagements, participation in advisory boards, and clinical trials from Elli Lilly and Co, Novo Nordisk, Sanofi-Aventis, Merck-Serono, Astra Zeneca, Roche, and Lan & Lee. D.Z. has received honoraria for speaking engagements, participation in advisory boards, and clinical trials from Eli Lilly and Co, Novo Nordisk, Sanofi-Aventis, and Astra Zeneca. ⁎ Corresponding author at: Medical Division, Lilly Suzhou Pharmaceutical Co., Ltd, 21F, 1st Corporate Avenue, 222 Hu Bin Road 200021, Shanghai, China. Tel.: +86 21 23020845. E-mail address: [email protected] (L. Qian). http://dx.doi.org/10.1016/j.jdiacomp.2015.02.014 1056-8727/© 2015 Elsevier Inc. All rights reserved.
estimated to be 98 million in 2013 and is estimated to increase to 143 million adults in 2035 (International Diabetes Federation, 2013b). Diabetes is the third leading cause of death in China after cardiovascular disease and cancer (Li et al., 2012). Recent analyses estimate that the mean healthcare expenditure in China per person with diabetes is USD $656 per year (Yang et al., 2012). Identification of patients with diabetes who are at risk of poor outcomes may contribute to targeted health care strategies that can alleviate the impact of diabetes in China. Obesity complicates the management of diabetes by increasing insulin resistance and blood glucose concentrations (Maggio & Pi-Sunyer, 1997) and by making pharmacological treatment more difficult (Galloway, 1990; Genuth, 1990; Yki-Järvinen et al., 1997; Yurgin, Secnik, & Lage, 2008). In addition, the risk of developing dyslipidemia, hypertension, and cardiovascular disease is increased in people who are obese (Hubert, Feinleib, McNamara, & Castelli, 1983; Stamler, Stamler, Riedlinger, Algera, & Roberts, 1978). This risk is magnified further in patients with type 2 diabetes mellitus (T2DM)
L. Ji et al. / Journal of Diabetes and Its Complications 29 (2015) 488–496
(Albu, Konnarides, & Pi-Sunyer, 1995; Henry & Gumbiner, 1991; Ohlson et al., 1985). Numerous studies conducted in the United States and Europe have demonstrated that obese patients with T2DM have poorer glycemic control (El-Kebbi et al., 2003; Hillier & Pedula, 2001; Nguyen, Nguyen, Lane, & Wang, 2011), more commonly have complications such as neuropathy (Straub, Thum, Hollerbach, Palitzsch, & Scholmerich, 1994), and have a higher mortality rate (Maggio & Pi-Sunyer, 1997) compared with normal weight individuals with diabetes. However, few studies have included patients from China and it is not known whether body mass index (BMI) plays a role in glycemic control or clinical outcomes in Chinese patients with T2DM. A large proportion of Chinese patients with diabetes are classified as having normal weight (Boffetta et al., 2011; Chan et al., 2004; Xu et al., 2010); however, the prevalence of obesity is increasing in this population (Hou et al., 2013a). Therefore, it is important to assess the role that BMI has on clinical, health-related quality of life (HRQoL) outcomes in a contemporary population of patients with T2DM in China. Traditional large scale epidemiological surveys used to inform public health decision-making have limitations, notably the cost, administration time, and the timeliness of the data (Babineaux et al., 2014). The Diabetes Disease Specific Programme (DSP) utilizes up-to-date representative sampling of treated adult patients for national disease burden quantification and treatment pattern and outcomes assessment (Babineaux et al., 2014). Using the Diabetes DSP, the objectives of this exploratory study were to assess and summarize the association between BMI and glycemic control, comorbidities and complications, and HRQoL in Chinese patients with T2DM.
489
2.3. Survey design Both the PRF and PSC were voluntary, self-administered, used lay terminology, and were anonymous and confidential. The survey questionnaires were written in English and translated into Simplified Chinese language by Adelphi Consultech (Anderson et al., 2008). The PRF collected information on patient demographics and clinical characteristics, clinical management and outcomes (chart review was necessary), medication use and history, and rationale for treatment choices. Physicians also provided their opinions on the patients' life and involvement/engagement with their disease, weight management, and compliance/adherence to medication and lifestyle changes. Questions could be excluded if they did not apply to the particular patient. The PSC collected information about the patient's life, work, support- and knowledge-systems, and how diabetes affects their everyday life, as well as their opinions and understanding of their medications and glycemic control. The PSC also included the European Quality of Life instrument (EuroQoL EQ-5D 3L) (Dolan, 1997; The EuroQol Group, 1990), which comprises a questionnaire, validated in China, covering five dimensions (mobility, personal hygiene, performance of usual activities, pain/discomfort, and anxiety/depression) (Luo et al., 2003). United Kingdom adult population-based preferences for the EQ-5D health states were used to calculate the EQ-5D Total Health Index (where b0 = worse than death, 0 = death, and 1 = perfect health) (Dolan, 1997). Patients could choose to not complete any or all questions. 2.4. Survey outcomes
2. Materials and methods 2.1. Survey administration The Diabetes DSP was designed by Adelphi Real World and conducted in China (October 2011–March 2012) in accordance with European Pharmaceutical Marketing Research Association guidelines. Physicians were reimbursed for their participation in the study by local fieldwork partners at fair-market rates. DSP fieldwork teams adhered to national data collection regulations (Anderson, Benford, Harris, Karavali, & Piercy, 2008). Informed consent was obtained after physicians explained the study and patients reviewed the patient information sheet. A flow diagram illustrating the execution of the DSP program is shown in Appendix A.
2.2. Distribution of the Survey to the Population Internal medicine or diabetologists/endocrinologists treating patients with diabetes at tier 2 or 3 general hospitals participated in the survey. Physicians were identified by local DSP fieldwork teams from public lists using predefined selection criteria. Physicians were asked to complete the Patient Record Form (PRF) and distribute the Patient Self-completion Form (PSC) to consecutive patients as follows: internal medicine—T2DM (n = 9), T2DM on insulin only (n = 2), and type 1 diabetes on insulin (n = 1); diabetologist/endocrinologist physicians—T2DM (n = 8), T2DM on insulin only (n = 2), and type 1 diabetes on insulin (n = 1). The diagnosis of diabetes type was made by the physician and no additional testing was required by the survey. The patient completed the PSC in the waiting room and immediately returned the survey to the physician. Physicians completed the PRF and all surveys were returned to Adelphi Real World. To ensure patient privacy was maintained, the forms did not collect any identifying information.
Response rate was not calculated and missing item data were not imputed. The primary a priori outcome measures were: glycemic control (glycated hemoglobin [HbA1c]; fasting blood glucose [FBG]), prevalence of comorbidities and complications, and patient-reported HRQoL. HbA1c, FBG, and serum lipid levels were measured by hospital laboratories. In China, tier 3 hospitals use internationally recognized laboratory testing procedures. The prevalence of comorbidities was collected using a checkbox requesting the presence of medically diagnosed conditions. Adelphi Real World did not provide definitions for comorbidities; however, Appendix B lists the comorbidities captured and the Chinese criteria used to define these comorbidities. For hypertension, dyslipidemia, visual or renal impairment, and neuropathy, the physician was asked to provide the duration of the comorbidity/complication. Data regarding who or how the diagnoses were made were not captured by the survey. Systolic/diastolic blood pressure measurements, as well as serum lipid levels, captured in the patients' medical files, were also collected by the PRF; however, these values were not used to define the prevalence of hypertension or dyslipidemia, as patients may have been receiving antihypertensive or anticholesterolemic medications. 2.5. Statistical analysis The current study compared outcomes between BMI subgroups. BMI was calculated by the statistician using the most recent height (cm) and weight (kg) values collected by the PRF. Physicians were to measure height and weight using the methods customarily used in their practice; ie, specific guidance on how to measure height and weight were not provided by Adelphi Real world. The analysis was stratified by BMI b18.5 kg/m2 (underweight), BMI 18.5 to b 24.0 kg/m2 (normal weight), BMI 24.0 to b28.0 kg/m 2 (overweight), and BMI ≥28.0 kg/m2 (obese). The BMI cut-offs used are endorsed by the National Health & Family Planning Commission of the People's Republic of China (2013). The
490
L. Ji et al. / Journal of Diabetes and Its Complications 29 (2015) 488–496
Table 1 Patient demographics and baseline clinical characteristics. Characteristic
Body mass index group
Mean age, years Gender Male, n (%) Mean BMI, kg/m2 Time from diagnosis of type 2 diabetes, weeks Current insulin user, n (%) Smoking status Non-smoker Smoker/ex-smoker, n (%) Highest level of education, n (%) Less than high school/high school Some university/undergraduate/graduate degree Employment status, n (%) Full-time/part-time/self-employed/student Unemployed/retired/homemaker/other Occupation, n (%) Manual—skilled/unskilled Executive/managerial/clerical/other Marital status, n (%) Married/partnership Single/divorced/widowed Home circumstances, n (%) Lives with partner/family/friends or nursing/residential home Lives alone Caregiver status, n (%) Yes Relationship of caregiver, n (%) Professional carer Spouse/partner/family/friend/neighbour
P value
18.5 to b24, normal (n = 998)
24 to b28, overweight (n = 822)
≥28, obese (n = 212)
56.6
56.5
53.5a
495 (49.6) 22.0 192.2 300 (30.1)
405 (49.3) 25.6c 160.1e 213 (25.9)
70 (33.0)b 30.2d 175.0 49 (23.1)f
611 (63.1) 358 (36.9)
480 (60.0) 320 (40.0)
145 (71.8)g 57 (28.2)
588 (63.6) 336 (36.4)
483 (64.1) 271 (35.9)
119 (61.3) 75 (38.7)
403 (40.4) 593 (59.4)
319 (38.8) 503 (61.2)
103 (48.6)h 109 (51.4)
214 (21.4) 731 (73.2)
188 (22.9) 579 (70.4)
52 (24.5) 146 (68.9)
902 (90.8) 91 (9.2)
704 (86.6)i 109 (13.4)
184 (87.6) 26 (12.4)
951 (96.0) 40 (4.0)
785 (96.3) 30 (3.7)
198 (94.7) 11 (5.3)
210 (21.1)
186 (22.9)
36 (17.1)
4 (0.4) 204 (20.4)
3 (0.4) 182 (22.1)
3 (1.4) 32 (15.1)
b0.001, F b0.001, f b0.001, 0.003, 0.043, 0.007,
F F f f
0.781, f
0.036, f
0.444, f
0.014, f
0.550, f
0.192, f 0.076
Abbreviations: BMI = body mass index, f = Fisher’s exact test, F = analysis of variance F-test. a P b 0.001—normal vs. obese, overweight vs. obese; Student t-test. b P b 0.001—normal vs. obese, overweight vs. obese; Fisher’s exact test. c P b 0.001—normal vs. overweight; Student t-test. d P b 0.001 —normal vs. obese, overweight vs. obese; Student t-test. e P b 0.001—normal vs. overweight; Student t-test. f P = 0.045—normal vs. obese; Fisher’s exact test. g P = 0.02—normal vs. obese, P = 0.002—overweight vs. obese; Fisher’s exact test. h P = 0.03—normal vs. obese, P = 0.01—overweight vs. obese; Fisher’s exact test. i P = 0.005—normal vs. overweight; Fisher’s exact test.
number of patients in the underweight (b 18.5 kg/m2) group was 20. Therefore, these patients were excluded from the statistical analyses and associated data are not reported. For the three remaining BMI groups, categorical variables were compared, overall and pairwise, by Fisher's exact test or Monte Carlo simulation of Fisher's exact test if the former was not feasible.
Percentage of patients (%)
100
4.3
For continuous variables, the F-test and t-tests were used for overall and pairwise comparisons, respectively. Correction for multiple-testing was not performed as the analyses in this study were exploratory. Sensitivity analyses were performed to adjust for baseline variables demonstrating significant differences between the BMI
5.5
10.2
53.7
62.6
80 9.0%
60
62.4
7.0% to
Contents lists available at ScienceDirect
Journal of Diabetes and Its Complications journal homepage: WWW.JDCJOURNAL.COM
Increasing body mass index identifies Chinese patients with type 2 diabetes mellitus at risk of poor outcomes Linong Ji a, Dajin Zou b, Li Liu c, Lei Qian c,⁎, Zbigniew Kadziola d, Steven Babineaux e, He Na Zhang c, Robert Wood f a
Department of Endocrinology and Metabolism, Peking University People's Hospital, Beijing, China Endocrinology Department, Affiliated Changhai Hospital to the 2nd Military Medical University, Shanghai, China c Medical Division, Lilly Suzhou Pharmaceutical Co., Ltd, Shanghai, China d Eli Lilly and Co., Real World Analytics, Vienna, Austria e Eli Lilly and Co., Indianapolis, IN, USA f Adelphi Real World, Bollington, Cheshire, UK b
a r t i c l e
i n f o
Article history: Received 24 November 2014 Received in revised form 14 February 2015 Accepted 26 February 2015 Available online 6 March 2015 Keywords: Body mass index China Diabetes mellitus Obesity Quality of life
a b s t r a c t Aims: Association between body mass index (BMI) and glycemic control, comorbidities/complications, and health-related quality of life (HRQoL) was assessed in Chinese patients with type 2 diabetes mellitus (T2DM) enrolled in the Diabetes Disease Specific Programme. Methods: Surveys of 200 physicians and 2052 patients with T2DM captured demographic, clinical, and HRQoL information. Adjusted and unadjusted analyses were conducted across 3 BMI groups; normal (18.5–b24.0, n = 998), overweight (24.0–b28.0, n = 822), and obese (≥28.0, n = 212). Results: There were no between group differences in the achievement of glycated hemoglobin (HbA1c) b 7.0% (48 mmol/mol); however, compared with the normal BMI group, more obese patients had an HbA1c N 9.0% (75 mmol/mol; 4.3% vs 10.2%, P = 0.002). More obese patients compared with normal BMI patients had hypertension (48.6% vs 35.3%, P b 0.001), dyslipidemia (35.4% vs 18.8%, P b 0.001), or both hypertension and dyslipidemia (24.1% vs 13.9%, P b 0.001). Patients in the obese group reported worse HRQoL and greater effects of diabetes on their daily living. Conclusions: Obesity in Chinese patients with T2DM results in poor glycemic control, more comorbidities, and worse HRQoL. Management of these patients should include efforts to reduce weight. Selection of weight-neutral or weight-reducing anti-diabetic medications maybe useful in these patients. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Diabetes is an increasingly prevalent health issue in China that carries a significant health and economic burden (International Diabetes Federation, 2013b; Li et al., 2012; Smith, Vlachopioti, & Colclough, 2010). The number of adults in China with diabetes was
Conflicts of Interest: L.L., L.Q., and H.Z. are employees of Lilly Suzhou Pharmaceutical Company, Ltd. Z.K. and S.B. are employees of Eli Lilly and Co. R.W. is an employee of Adelphi Real World. S.B. and Z.K. own shares in Eli Lilly and Co. L.J. has received honoraria for speaking engagements, participation in advisory boards, and clinical trials from Elli Lilly and Co, Novo Nordisk, Sanofi-Aventis, Merck-Serono, Astra Zeneca, Roche, and Lan & Lee. D.Z. has received honoraria for speaking engagements, participation in advisory boards, and clinical trials from Eli Lilly and Co, Novo Nordisk, Sanofi-Aventis, and Astra Zeneca. ⁎ Corresponding author at: Medical Division, Lilly Suzhou Pharmaceutical Co., Ltd, 21F, 1st Corporate Avenue, 222 Hu Bin Road 200021, Shanghai, China. Tel.: +86 21 23020845. E-mail address: [email protected] (L. Qian). http://dx.doi.org/10.1016/j.jdiacomp.2015.02.014 1056-8727/© 2015 Elsevier Inc. All rights reserved.
estimated to be 98 million in 2013 and is estimated to increase to 143 million adults in 2035 (International Diabetes Federation, 2013b). Diabetes is the third leading cause of death in China after cardiovascular disease and cancer (Li et al., 2012). Recent analyses estimate that the mean healthcare expenditure in China per person with diabetes is USD $656 per year (Yang et al., 2012). Identification of patients with diabetes who are at risk of poor outcomes may contribute to targeted health care strategies that can alleviate the impact of diabetes in China. Obesity complicates the management of diabetes by increasing insulin resistance and blood glucose concentrations (Maggio & Pi-Sunyer, 1997) and by making pharmacological treatment more difficult (Galloway, 1990; Genuth, 1990; Yki-Järvinen et al., 1997; Yurgin, Secnik, & Lage, 2008). In addition, the risk of developing dyslipidemia, hypertension, and cardiovascular disease is increased in people who are obese (Hubert, Feinleib, McNamara, & Castelli, 1983; Stamler, Stamler, Riedlinger, Algera, & Roberts, 1978). This risk is magnified further in patients with type 2 diabetes mellitus (T2DM)
L. Ji et al. / Journal of Diabetes and Its Complications 29 (2015) 488–496
(Albu, Konnarides, & Pi-Sunyer, 1995; Henry & Gumbiner, 1991; Ohlson et al., 1985). Numerous studies conducted in the United States and Europe have demonstrated that obese patients with T2DM have poorer glycemic control (El-Kebbi et al., 2003; Hillier & Pedula, 2001; Nguyen, Nguyen, Lane, & Wang, 2011), more commonly have complications such as neuropathy (Straub, Thum, Hollerbach, Palitzsch, & Scholmerich, 1994), and have a higher mortality rate (Maggio & Pi-Sunyer, 1997) compared with normal weight individuals with diabetes. However, few studies have included patients from China and it is not known whether body mass index (BMI) plays a role in glycemic control or clinical outcomes in Chinese patients with T2DM. A large proportion of Chinese patients with diabetes are classified as having normal weight (Boffetta et al., 2011; Chan et al., 2004; Xu et al., 2010); however, the prevalence of obesity is increasing in this population (Hou et al., 2013a). Therefore, it is important to assess the role that BMI has on clinical, health-related quality of life (HRQoL) outcomes in a contemporary population of patients with T2DM in China. Traditional large scale epidemiological surveys used to inform public health decision-making have limitations, notably the cost, administration time, and the timeliness of the data (Babineaux et al., 2014). The Diabetes Disease Specific Programme (DSP) utilizes up-to-date representative sampling of treated adult patients for national disease burden quantification and treatment pattern and outcomes assessment (Babineaux et al., 2014). Using the Diabetes DSP, the objectives of this exploratory study were to assess and summarize the association between BMI and glycemic control, comorbidities and complications, and HRQoL in Chinese patients with T2DM.
489
2.3. Survey design Both the PRF and PSC were voluntary, self-administered, used lay terminology, and were anonymous and confidential. The survey questionnaires were written in English and translated into Simplified Chinese language by Adelphi Consultech (Anderson et al., 2008). The PRF collected information on patient demographics and clinical characteristics, clinical management and outcomes (chart review was necessary), medication use and history, and rationale for treatment choices. Physicians also provided their opinions on the patients' life and involvement/engagement with their disease, weight management, and compliance/adherence to medication and lifestyle changes. Questions could be excluded if they did not apply to the particular patient. The PSC collected information about the patient's life, work, support- and knowledge-systems, and how diabetes affects their everyday life, as well as their opinions and understanding of their medications and glycemic control. The PSC also included the European Quality of Life instrument (EuroQoL EQ-5D 3L) (Dolan, 1997; The EuroQol Group, 1990), which comprises a questionnaire, validated in China, covering five dimensions (mobility, personal hygiene, performance of usual activities, pain/discomfort, and anxiety/depression) (Luo et al., 2003). United Kingdom adult population-based preferences for the EQ-5D health states were used to calculate the EQ-5D Total Health Index (where b0 = worse than death, 0 = death, and 1 = perfect health) (Dolan, 1997). Patients could choose to not complete any or all questions. 2.4. Survey outcomes
2. Materials and methods 2.1. Survey administration The Diabetes DSP was designed by Adelphi Real World and conducted in China (October 2011–March 2012) in accordance with European Pharmaceutical Marketing Research Association guidelines. Physicians were reimbursed for their participation in the study by local fieldwork partners at fair-market rates. DSP fieldwork teams adhered to national data collection regulations (Anderson, Benford, Harris, Karavali, & Piercy, 2008). Informed consent was obtained after physicians explained the study and patients reviewed the patient information sheet. A flow diagram illustrating the execution of the DSP program is shown in Appendix A.
2.2. Distribution of the Survey to the Population Internal medicine or diabetologists/endocrinologists treating patients with diabetes at tier 2 or 3 general hospitals participated in the survey. Physicians were identified by local DSP fieldwork teams from public lists using predefined selection criteria. Physicians were asked to complete the Patient Record Form (PRF) and distribute the Patient Self-completion Form (PSC) to consecutive patients as follows: internal medicine—T2DM (n = 9), T2DM on insulin only (n = 2), and type 1 diabetes on insulin (n = 1); diabetologist/endocrinologist physicians—T2DM (n = 8), T2DM on insulin only (n = 2), and type 1 diabetes on insulin (n = 1). The diagnosis of diabetes type was made by the physician and no additional testing was required by the survey. The patient completed the PSC in the waiting room and immediately returned the survey to the physician. Physicians completed the PRF and all surveys were returned to Adelphi Real World. To ensure patient privacy was maintained, the forms did not collect any identifying information.
Response rate was not calculated and missing item data were not imputed. The primary a priori outcome measures were: glycemic control (glycated hemoglobin [HbA1c]; fasting blood glucose [FBG]), prevalence of comorbidities and complications, and patient-reported HRQoL. HbA1c, FBG, and serum lipid levels were measured by hospital laboratories. In China, tier 3 hospitals use internationally recognized laboratory testing procedures. The prevalence of comorbidities was collected using a checkbox requesting the presence of medically diagnosed conditions. Adelphi Real World did not provide definitions for comorbidities; however, Appendix B lists the comorbidities captured and the Chinese criteria used to define these comorbidities. For hypertension, dyslipidemia, visual or renal impairment, and neuropathy, the physician was asked to provide the duration of the comorbidity/complication. Data regarding who or how the diagnoses were made were not captured by the survey. Systolic/diastolic blood pressure measurements, as well as serum lipid levels, captured in the patients' medical files, were also collected by the PRF; however, these values were not used to define the prevalence of hypertension or dyslipidemia, as patients may have been receiving antihypertensive or anticholesterolemic medications. 2.5. Statistical analysis The current study compared outcomes between BMI subgroups. BMI was calculated by the statistician using the most recent height (cm) and weight (kg) values collected by the PRF. Physicians were to measure height and weight using the methods customarily used in their practice; ie, specific guidance on how to measure height and weight were not provided by Adelphi Real world. The analysis was stratified by BMI b18.5 kg/m2 (underweight), BMI 18.5 to b 24.0 kg/m2 (normal weight), BMI 24.0 to b28.0 kg/m 2 (overweight), and BMI ≥28.0 kg/m2 (obese). The BMI cut-offs used are endorsed by the National Health & Family Planning Commission of the People's Republic of China (2013). The
490
L. Ji et al. / Journal of Diabetes and Its Complications 29 (2015) 488–496
Table 1 Patient demographics and baseline clinical characteristics. Characteristic
Body mass index group
Mean age, years Gender Male, n (%) Mean BMI, kg/m2 Time from diagnosis of type 2 diabetes, weeks Current insulin user, n (%) Smoking status Non-smoker Smoker/ex-smoker, n (%) Highest level of education, n (%) Less than high school/high school Some university/undergraduate/graduate degree Employment status, n (%) Full-time/part-time/self-employed/student Unemployed/retired/homemaker/other Occupation, n (%) Manual—skilled/unskilled Executive/managerial/clerical/other Marital status, n (%) Married/partnership Single/divorced/widowed Home circumstances, n (%) Lives with partner/family/friends or nursing/residential home Lives alone Caregiver status, n (%) Yes Relationship of caregiver, n (%) Professional carer Spouse/partner/family/friend/neighbour
P value
18.5 to b24, normal (n = 998)
24 to b28, overweight (n = 822)
≥28, obese (n = 212)
56.6
56.5
53.5a
495 (49.6) 22.0 192.2 300 (30.1)
405 (49.3) 25.6c 160.1e 213 (25.9)
70 (33.0)b 30.2d 175.0 49 (23.1)f
611 (63.1) 358 (36.9)
480 (60.0) 320 (40.0)
145 (71.8)g 57 (28.2)
588 (63.6) 336 (36.4)
483 (64.1) 271 (35.9)
119 (61.3) 75 (38.7)
403 (40.4) 593 (59.4)
319 (38.8) 503 (61.2)
103 (48.6)h 109 (51.4)
214 (21.4) 731 (73.2)
188 (22.9) 579 (70.4)
52 (24.5) 146 (68.9)
902 (90.8) 91 (9.2)
704 (86.6)i 109 (13.4)
184 (87.6) 26 (12.4)
951 (96.0) 40 (4.0)
785 (96.3) 30 (3.7)
198 (94.7) 11 (5.3)
210 (21.1)
186 (22.9)
36 (17.1)
4 (0.4) 204 (20.4)
3 (0.4) 182 (22.1)
3 (1.4) 32 (15.1)
b0.001, F b0.001, f b0.001, 0.003, 0.043, 0.007,
F F f f
0.781, f
0.036, f
0.444, f
0.014, f
0.550, f
0.192, f 0.076
Abbreviations: BMI = body mass index, f = Fisher’s exact test, F = analysis of variance F-test. a P b 0.001—normal vs. obese, overweight vs. obese; Student t-test. b P b 0.001—normal vs. obese, overweight vs. obese; Fisher’s exact test. c P b 0.001—normal vs. overweight; Student t-test. d P b 0.001 —normal vs. obese, overweight vs. obese; Student t-test. e P b 0.001—normal vs. overweight; Student t-test. f P = 0.045—normal vs. obese; Fisher’s exact test. g P = 0.02—normal vs. obese, P = 0.002—overweight vs. obese; Fisher’s exact test. h P = 0.03—normal vs. obese, P = 0.01—overweight vs. obese; Fisher’s exact test. i P = 0.005—normal vs. overweight; Fisher’s exact test.
number of patients in the underweight (b 18.5 kg/m2) group was 20. Therefore, these patients were excluded from the statistical analyses and associated data are not reported. For the three remaining BMI groups, categorical variables were compared, overall and pairwise, by Fisher's exact test or Monte Carlo simulation of Fisher's exact test if the former was not feasible.
Percentage of patients (%)
100
4.3
For continuous variables, the F-test and t-tests were used for overall and pairwise comparisons, respectively. Correction for multiple-testing was not performed as the analyses in this study were exploratory. Sensitivity analyses were performed to adjust for baseline variables demonstrating significant differences between the BMI
5.5
10.2
53.7
62.6
80 9.0%
60
62.4
7.0% to
