Skip to main content
NCBI home page
Journal of Diabetes Investigation logoLink to Journal of Diabetes Investigation
. 2026 Apr 27;17(7):1218–1230. doi: 10.1111/jdi.70322

Causal effects of lifestyle and socioeconomic factors on diabetes and its complications and comorbidities: A Mendelian randomization study

Qian Yu 1, Zhaowei Kong 1,✉, Qingde Shi 2, Chak Kwong Cheng 3, Jia Zhang 4,5, Jinlei Nie 2
PMCID: PMC13327334  PMID: 42046344

ABSTRACT

Aims

Diabetes poses a growing global health burden. This study investigated the causal effects of lifestyle and socioeconomic factors on diabetes risk and related complications and comorbidities.

Materials and Methods

We applied two‐sample univariable and multivariable Mendelian randomization to assess the causal impact of 26 lifestyle and socioeconomic factors on diabetes, 7 complications, and 13 comorbidities.

Results

Genetically predicted protective factors included vigorous physical activity (odds ratio [OR], 0.98 [0.98–0.99]), computer use time (OR, 0.13 [0.05, 0.30]), carbohydrate intake (OR, 0.22 [0.15, 0.34]), short (OR, 0.04 [0.01, 0.16]) and long sleep duration (OR, 0.62 [0.47, 0.82]), moderate alcohol (OR, 0.13 [0.04, 0.50]) and caffeine (OR, 0.72 [0.64, 0.81]) consumption, education (OR, 0.25–0.67), and household income (OR, 0.52–0.65), which were associated with reduced risks of type 2 and gestational diabetes, stroke, coronary heart disease, heart failure, myocardial infarction, sleep apnea, and anxiety disorder (adjusted P‐values <0.05). Conversely, genetically predicted factors, such as television watching (OR, 1.39 [1.23, 1.57]) and driving time (OR, 3.28 [1.27, 8.48]), insomnia (OR, 1.21–1.82), smoking behaviors (OR, 1.17–1.77), alcohol dependence (OR, 1.17–1.28), coffee consumption (OR, 1.01 [1.00, 1.02]), and the Townsend deprivation index (OR, 1.51–1.57), are associated with increased risks of diabetes‐related outcomes (i.e., all diabetic types, neovascular glaucoma, heart failure, nonalcoholic fatty liver disease, sleep apnea, and eating disorder) (adjusted P‐values <0.05).

Conclusions

Our findings support causal roles of lifestyle and socioeconomic factors and diabetes‐related outcomes, emphasizing the need for targeted public health strategies to promote healthier living and socioeconomic equity.

Keywords: Diabetes, Lifestyle factors, Socioeconomic status

INTRODUCTION

Diabetes is a chronic metabolic disorder arising from complex interactions between genetic predispositions and environmental factors 1 . It encompasses type 1 diabetes, type 2 diabetes, and gestational diabetes, along with other less prevalent variants 2 . In 2021, the global prevalence of diabetes was estimated to be 537 million, with related expenditures reaching $966 billion 3 , 4 . This condition contributes significantly to morbidity and mortality through complications such as diabetic retinopathy, nephropathy, neuropathy, maculopathy, and cardiovascular diseases 5 . Furthermore, individuals with diabetes often experience comorbidities, including hypertension, stroke, coronary heart disease, heart failure, myocardial infarction, obesity, hyperlipidemia, sleep apnea, nonalcoholic fatty liver disease, osteoporosis, and mental health disorders 5 . Despite advances in pharmacological treatments, many individuals do not achieve optimal disease management, underscoring the need for preventive strategies.

Epidemiological studies suggest a strong link between lifestyle factors and diabetes risk. Meta‐analyses of randomized controlled trials (RCTs) indicate that regular physical activity and dietary modifications (e.g., carbohydrate‐restricted and calorie‐restricted diets) reduce the risk of type 2 6 , 7 and gestational diabetes 8 . Moreover, observational studies have linked short and long sleep durations to an increased risk of type 2 diabetes 9 , 10 . Moderate alcohol consumption 11 , 12 and nonsmoking 13 , 14 have also been associated with lower diabetes incidence, although recent smoking cessation initially increases risk before subsequent reductions over time 14 . Coffee and caffeine intake appear to lower the risk of type 2 diabetes 15 , though coffee intake is also associated with a higher occurrence of metabolic syndrome in type 1 diabetes 16 .

Beyond diabetes incidence, lifestyle factors may influence the progression of its complications and comorbidities. Weight loss through dietary interventions improves cardiovascular risk factors, such as hemoglobin A1C (HbA1C), blood pressure, and cholesterol levels, while exercise interventions showed significant effects on HbA1C and diastolic blood pressure 17 . However, meta‐analyses provide mixed evidence regarding the impact of lifestyle interventions on cardiovascular and microvascular complications in type 2 diabetes 18 , 19 . Observational studies suggest that a healthy lifestyle—including regular physical activity, reduced sedentary time, balanced diet, adequate sleep, moderate alcohol intake, and nonsmoking—lowers the risk of diabetic complications and comorbidities 20 , 21 . Additionally, higher socioeconomic status (e.g., educational attainment, income level) has been associated with reduced risks of diabetes and its complications, such as cardiovascular diseases and neuropathy 22 , 23 .

Although RCTs remain the gold standard for establishing causal relationships, they have inherent limitations, including high costs, ethical constraints, and difficulties in generalization 24 . Observational studies, despite their ability to capture real‐world effects, are subject to confounding by genetic, social, and behavioral factors 25 . Consequently, the causal role of many lifestyle and socioeconomic factors in diabetes and its complications remains unclear due to the lack of well‐designed prospective studies or large‐scale RCTs.

Mendelian randomization (MR) has emerged as a powerful approach to address these limitations 26 . This design minimizes confounding by assuming that genetic variants are randomly assigned at conception and remain stable throughout the lifetime 27 . Previous MR studies have linked insomnia, dietary protein intake, and smoking initiation to an increased risk of type 2 diabetes, while higher educational attainment and household income have been associated with a lower risk 28 , 29 . Additionally, among individuals with diabetes, higher levels of smoking initiation, lifetime smoking index, and insomnia have been linked to increased cardiovascular risk, whereas higher educational attainment and income demonstrate protective effects 29 . However, inconsistencies in MR findings—particularly regarding physical activity, alcohol, and coffee intake—highlight the need for refined methodologies and further investigation.

The growing availability of large‐scale genome‐wide association study (GWAS) data has enhanced the ability to conduct robust MR analyses. This study aims to utilize high‐quality GWAS datasets and apply two‐sample univariable MR (UVMR) to investigate the causal relationships between 26 lifestyle and socioeconomic factors—including physical activity, sedentary behavior, nutrition, sleep, smoking, alcohol and coffee intake, education, income, and material deprivation—and diabetes risk (type 1, type 2, and gestational diabetes). Additionally, this study explores the causal associations between these factors and seven common diabetes complications (diabetic retinopathy, nephropathy, neuropathy, maculopathy, neovascular glaucoma, Charcot foot, and acanthosis nigricans) as well as 13 comorbidities, including hypertension, stroke, coronary heart disease, heart failure, myocardial infarction, obesity, hyperlipidemia, sleep apnea, nonalcoholic fatty liver disease, osteoporosis, depressive disorder, anxiety disorder, and eating disorder. Finally, this study extends the analysis by employing multivariable MR (MVMR) to assess the interdependent effects of significant lifestyle and socioeconomic factors identified in UVMR. By incorporating MVMR, this study aims to disentangle direct and total effects of these factors on diabetes and its associated health outcomes, providing a comprehensive understanding of how lifestyle and socioeconomic variables contribute to the development and progression of diabetes.

METHODS

Study design

In a two‐sample MR approach, genetic variants served as instrumental variables to evaluate the causal effects of lifestyle and socioeconomic factors on the risk of diabetes, including its associated complications and comorbidities. While UVMR enables the examination of causality by using genetic variants explicitly associated with a single exposure 30 , MVMR extends this by allowing the investigation of causal relationships for SNPs associated with multiple exposures 31 . MVMR allows for distinguishing between total and direct causal effects 31 .

The overview of the study design and the core MR assumptions for selection of valid instrumental variables is provided in Figure 1. All summary statistics utilized were derived from GWAS conducted in populations of European ancestry. All studies received ethical approval from their institutional review boards, with participants providing written informed consent. This study followed the Strengthening the Reporting of Genetic Association Studies reporting guideline and Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization reporting guidelines.

Figure 1.

Figure 1

Open in a new tab

Schematic overview of the study design. IVW, inverse–variance weighted; MR, Mendelian randomization; MR‐PRESSO, Mendelian Randomization Pleiotropy RESidual Sum and Outlier; MVMR, multivariable Mendelian randomization; SNP, single‐nucleotide polymorphism; UVMR, univariable Mendelian randomization.

Summary data resources

For exposures, this study utilized GWAS summary statistics across multiple lifestyle (i.e., physical activity, sedentary behavior, nutrition, sleep, smoking, alcohol, and coffee intake) and socioeconomic domains (Table 1). The used datasets were from the MRC Integrative Epidemiology Unit database and published work as mentioned (Table 1).

Table 1.

Summary of exposure data sources

Unit Sample size Ancestry Consortium or cohort study Year of publication or release Identified SNPs PubMed ID
Lifestyle factors
Physical activity
Average acceleration (accelerometer‐based physical activity) 1‐SD increase in Milli‐g/week 91084 European UK Biobank 2018 2 29899525
Moderate physical activity 1‐SD increase in MET‐minutes/week 377234 European UK Biobank 2018 9 29899525
Vigorous physical activity 1‐SD increase in MET‐minutes/week 261055 European UK Biobank 2018 5 29899525
Sedentary behaviors
Television watching time 1‐SD increase in sedentary time 503325 European UK Biobank 2020 152 32317632
Computer use time 1‐SD increase in sedentary time 503325 European UK Biobank 2020 37 32317632
Driving time 1‐SD increase in sedentary time 503325 European UK Biobank 2020 4 32317632
Nutrition
Fat intake 1‐SD increase in nutrition intake 268922 European Meta 2021 6 32393786
Protein intake 1‐SD increase in nutrition intake 268922 European Meta 2021 7 32393786
Carbohydrate intake 1‐SD increase in nutrition intake 268922 European Meta 2021 13 32393786
Sugar intake 1‐SD increase in nutrition intake 235391 European Meta 2021 10 32393786
Sleep
Insomnia Odds of insomnia 1331010 European Meta 2019 248 30804565
Sleep duration Hours per day 446118 European UK Biobank 2019 78 30846698
Short sleep duration <7 h compared with 7–8 h per day 411934 European UK Biobank 2019 27 30846698
Long sleep duration ≥9 h compared with 7–8 h per day 339926 European UK Biobank 2019 8 30846698
Smoking
Smoking initiation SD in prevalence of smoking initiation 1232091 European Meta 2019 378 30643251
Smoking heaviness (cigarettes per day) 1‐SD increase in number of cigarettes smoked per day 337334 European Meta 2019 55 30643251
Smoking cessation SD change of smoking cessation 547219 European Meta 2019 24 30643251
Lifetime smoking index SD change of lifetime smoking index 462690 European UK Biobank 2019 126 31689377
Alcohol
Alcohol consumption 1‐SD increase of log‐transformed alcoholic drinks/week 941280 European Meta 2019 99 30643251
Alcohol dependence Odds of alcohol dependence 46568 European Meta 2018 10 30482948
Coffee
Coffee consumption 50% change 375833 European UK Biobank 2019 15 31046077
Caffeine consumption 80 mg increase (equivalent to dose from 1 cup of coffee) 9876 European Meta 2016 2 27702941
Socioeconomic factors
Educational attainment 1‐SD increase in years of educational attainment 766345 European Meta 2018 311 30038396
Age completed full‐time education 1‐SD increase in ages completed full‐time education 307897 European MRC‐IEU 2018 41 /
Annual household income 1‐SD increase in annual household income 397751 European MRC‐IEU 2018 47 /
Townsend deprivation index 1‐SD increase in Townsend deprivation index 462464 European MRC‐IEU 2018 17 /
Open in a new tab

MET, metabolic equivalent of task; MRC‐IEU, the MRC Integrative Epidemiology Unit at the University of Bristol; SD, standard deviation; SNP, single‐nucleotide polymorphism.

For outcomes, summary statistics for diabetes diagnoses—including type 2 diabetes, type 1 diabetes, and gestational diabetes—along with diabetes complications and comorbidities, were obtained from the FinnGen database (Table 2). All the cases were defined using diagnostic codes from the International Classification of Diseases, Eighth, Ninth, and Tenth Revisions (ICD‐8, ICD‐9, and ICD‐10). Further information related to the included GWAS datasets can be found in Data S1.

Table 2.

Summary of outcome data sources

Number of cases Number of controls Ancestry Year of release Lambda Diagnostic criteria
Diabetes
Type 2 diabetes 71,728 369,007 European 2024 1.35 ICD‐8, ICD‐9, ICD‐10
Type1 diabetes 4,526 369,007 European 2024 1.03 ICD‐8, ICD‐9, ICD‐10
Gestational diabetes 16,802 237,816 European 2024 1.10 ICD‐8, ICD‐9, ICD‐10
Complications
Diabetic retinopathy 12,681 71,596 European 2024 1.04 ICD‐8, ICD‐9, ICD‐10
Diabetic nephropathy 5,042 79,344 European 2024 1.04 ICD‐8, ICD‐9, ICD‐10
Diabetic neuropathy 3,503 51,410 European 2024 1.06 ICD‐8, ICD‐9, ICD‐10
Diabetic maculopathy 4,248 71,596 European 2024 1.05 ICD‐8, ICD‐9, ICD‐10
Neovascular glaucoma 1,353 440,239 European 2024 1.00 ICD‐8, ICD‐9, ICD‐10
Charcot foot 537 328,079 European 2024 0.91 ICD‐8, ICD‐9, ICD‐10
Acanthosis nigricans 108 423,041 European 2024 0.51 ICD‐8, ICD‐9, ICD‐10
Comorbidities
Hypertension 137,312 316,345 European 2024 1.35 ICD‐8, ICD‐9, ICD‐10
Stroke 47,841 328,079 European 2024 1.11 ICD‐8, ICD‐9, ICD‐10
Coronary heart disease 51,098 402,635 European 2024 1.18 ICD‐8, ICD‐9, ICD‐10
Heart failure 15,579 402,635 European 2024 1.12 ICD‐8, ICD‐9, ICD‐10
Myocardial infarction 28,546 378,019 European 2024 1.16 ICD‐8, ICD‐9, ICD‐10
Obesity 27,711 425,881 European 2024 1.21 ICD‐8, ICD‐9, ICD‐10
Hyperlipidemia 1,340 384,093 European 2024 1.01 ICD‐8, ICD‐9, ICD‐10
Sleep apnea 50,200 401,484 European 2024 1.19 ICD‐8, ICD‐9, ICD‐10
Nonalcoholic fatty liver disease 3,006 450,727 European 2024 1.03 ICD‐8, ICD‐9, ICD‐10
Osteoporosis 9,046 429,826 European 2024 1.06 ICD‐8, ICD‐9, ICD‐10
Depressive disorder 53,313 394,756 European 2024 1.18 ICD‐8, ICD‐9, ICD‐10
Anxiety disorder 50,486 330,460 European 2024 1.17 ICD‐8, ICD‐9, ICD‐10
Eating disorder 3,525 330,460 European 2024 1.05 ICD‐8, ICD‐9, ICD‐10
Open in a new tab

ICD, the International Classification of Diseases.

Instruments selection

To identify SNP‐exposure associations, we selected a set of instrumental SNPs for each exposure phenotype GWAS that achieved genome‐wide significance (P < 5 × 10−8) 32 . To ensure SNP independence, we applied the PLINK algorithm with an r 2 threshold of 0.001 and a window size of 10,000 kb for clustering 33 . Ambiguous and palindromic SNPs were excluded during harmonization. Associations between SNPs and potential confounders were assessed using PhenoScanner, and SNPs with F‐statistics less than 10 were excluded to minimize bias from weak genetic instruments 34 . SNPs exhibiting potential horizontal pleiotropy were excluded through the use of the MR‐Pleiotropy Residual Sum and Outlier (MR‐PRESSO) method 35 .

Statistical analysis

Univariable Mendelian randomization

We employed multiple two‐sample UVMR approaches to evaluate the influence of lifestyle and socioeconomic factors on diabetes, along with its related complications and comorbidities. The inverse–variance weighted (IVW) method was utilized as the principal analysis to assess causal relationships 36 . Cochran's Q statistic was computed to assess heterogeneity among individual causal effects 37 , 38 . To assess the potential pleiotropy of the genetic instruments, the UVMR analyses were conducted using supplementary methods, including MR Egger, weighted median, weighted mode, and simple mode. The MR‐PRESSO was additionally employed to detect and correct potential outliers 39 . An F‐statistic threshold of 10 was applied to determine whether an instrument demonstrated sufficient strength 40 . All UVMR analyses were conducted using the ‘TwoSampleMR’ package in R software (version 4.4.1).

Multivariable Mendelian randomization

Exposures were selected for inclusion in the MVMR model based on significant associations observed in UVMR and biological plausibility, ensuring that retained variables had minimal overlap in their effects to reduce collinearity. The joint instrument strength for exposures in the MVMR framework was assessed using the Sanderson–Windmeijer conditional F‐statistic. Instruments were classified as weak if the F‐statistic fell below the threshold of 10 41 . MVMR analysis was primarily performed using the IVW approach, providing adjusted causal estimates for each exposure. Additional sensitivity analyses included MVMR‐Egger and MVMR‐Median. Cochran's Q statistic was used to assess heterogeneity among SNP‐specific causal estimates, with a nonsignificant result indicating consistency across SNPs 38 . Additionally, a pleiotropy Q statistic was calculated to detect potential horizontal pleiotropy. The MVMR analyses were employed using the ‘MVMR’ R package.

Odds ratios (OR) and associated 95% confidence intervals (CI) are reported for the main results. Given the large number of exposure–outcome tests, correction for multiple testing was performed using the Benjamini–Hochberg false discovery rate (FDR) approach. Associations with an FDR‐adjusted P‐value <0.05 were considered statistically significant. All analyses were conducted using R version 4.4.1. Data for this study were analyzed between June 2024 and October 2024.

RESULTS

The results of the primary UVMR analysis using the IVW method are presented in Figures 2 and 3. Sensitivity analyses for UVMR, including MR‐Egger, weighted median, weighted mode, and simple mode methods, are provided in the Tables S1–S23. Additional UVMR results, including tests for directional pleiotropy and heterogeneity, are presented in Table S24. Results of the MVMR analyses are provided in Tables S25 and S26. Unless otherwise specified, all significant associations described below remained statistically significant after Benjamini–Hochberg FDR correction.

Figure 2.

Figure 2

Open in a new tab

Causal associations between lifestyle and socioeconomic factors and diabetes and complications. Forest plot of Mendelian randomization analysis results showing odds ratios [ORs] and 95% confidence intervals for each exposure. Dot colors represent statistical significance and direction of association: black dots indicate no statistically significant association [adjusted P > 0.05]; green dots indicate a statistically significant protective effect [adjusted P < 0.05, OR <1]; and red dots indicate a statistically significant risk‐enhancing effect [adjusted P < 0.05, OR >1]). (a), causal associations between lifestyle and socioeconomic factors and type 2 diabetes; (b) causal associations between lifestyle and socioeconomic factors and type 1 diabetes; (c) causal associations between lifestyle and socioeconomic factors and gestational diabetes; (d) causal associations between lifestyle and socioeconomic factors and diabetic retinopathy; (e) causal associations between lifestyle and socioeconomic factors and diabetic nephropathy; (f) causal associations between lifestyle and socioeconomic factors and diabetic neuropathy; (g) causal associations between lifestyle and socioeconomic factors and diabetic maculopathy; (h) causal associations between lifestyle and socioeconomic factors and diabetic neovascular glaucoma; (i) causal associations between lifestyle and socioeconomic factors and Charcot foot; (j) causal associations between lifestyle and socioeconomic factors and acanthosis nigricans). Exposure 1, average acceleration [accelerometer‐based physical activity]; Exposure 2, moderate physical activity; Exposure 3, vigorous physical activity; Exposure 4, television watching time; Exposure 5, computer use time; Exposure 6, driving time; Exposure 7, fat intake; Exposure 8, protein intake; Exposure 9, carbohydrate intake; Exposure 10, sugar intake; Exposure 11, insomnia; Exposure 12, sleep duration; Exposure 13, short sleep duration; Exposure 14, long sleep duration; Exposure 15, smoking initiation; Exposure 16, smoking heaviness [cigarettes per day]; Exposure 17, smoking cessation; Exposure 18, lifetime smoking index; Exposure 19, alcohol consumption; Exposure 20, alcohol dependence; Exposure 21, coffee consumption; Exposure 22, caffeine consumption; Exposure 23, educational attainment; Exposure 24, age completed full‐time education; Exposure 25, annual household income; Exposure 26, Townsend deprivation index.

Figure 3.

Figure 3

Open in a new tab

Causal associations between lifestyle and socioeconomic factors and diabetic comorbidities. Forest plot of Mendelian randomization analysis results showing odds ratios [ORs] and 95% confidence intervals for each exposure. Dot colors represent statistical significance and direction of association: black dots indicate no statistically significant association [adjusted P > 0.05]; green dots indicate a statistically significant protective effect [adjusted P < 0.05, OR <1]; and red dots indicate a statistically significant risk‐enhancing effect [adjusted P < 0.05, OR >1]). (a), causal associations between lifestyle and socioeconomic factors and hypertension; (b) causal associations between lifestyle and socioeconomic factors and stroke; (c) causal associations between lifestyle and socioeconomic factors and coronary heart disease; (d) causal associations between lifestyle and socioeconomic factors and heart failure; (e) causal associations between lifestyle and socioeconomic factors and myocardial infarction; (f) causal associations between lifestyle and socioeconomic factors and obesity; (g) causal associations between lifestyle and socioeconomic factors and hyperlipidemia; (h) causal associations between lifestyle and socioeconomic factors and sleep apnea; (i) causal associations between lifestyle and socioeconomic factors and nonalcoholic fatty liver disease; (j) causal associations between lifestyle and socioeconomic factors and osteoporosis; (k) causal associations between lifestyle and socioeconomic factors and depressive disorder; (l) causal associations between lifestyle and socioeconomic factors and anxiety disorder; (m) causal associations between lifestyle and socioeconomic factors and eating disorder). Exposure 1, average acceleration [accelerometer‐based physical activity]; Exposure 2, moderate physical activity; Exposure 3, vigorous physical activity; Exposure 4, television watching time; Exposure 5, computer use time; Exposure 6, driving time; Exposure 7, fat intake; Exposure 8, protein intake; Exposure 9, carbohydrate intake; Exposure 10, sugar intake; Exposure 11, insomnia; Exposure 12, sleep duration; Exposure 13, short sleep duration; Exposure 14, long sleep duration; Exposure 15, smoking initiation; Exposure 16, smoking heaviness [cigarettes per day]; Exposure 17, smoking cessation; Exposure 18, lifetime smoking index; Exposure 19, alcohol consumption; Exposure 20, alcohol dependence; Exposure 21, coffee consumption; Exposure 22, caffeine consumption; Exposure 23, educational attainment; Exposure 24, age completed full‐time education; Exposure 25, annual household income; Exposure 26, Townsend deprivation index.

Genetic associations between lifestyle and socioeconomic factors and diabetes

Genetically predicted carbohydrate intake (IVW OR, 0.22; 95% CI: 0.15–0.34; adjusted P‐value <0.001), smoking cessation (IVW OR, 0.86; 95% CI: 0.78–0.96; adjusted P‐value = 0.02), caffeine consumption (IVW OR, 0.72; 95% CI: 0.64–0.81; adjusted P‐value < 0.001), educational attainment (IVW OR, 0.64; 95% CI: 0.59–0.69; adjusted P‐value < 0.001), age completed full‐time education (IVW OR, 0.58; 95% CI: 0.47–0.70; adjusted P‐value < 0.001), and annual household income (IVW OR, 0.65; 95% CI: 0.56–0.76; adjusted P‐value < 0.001) were associated with protective effects against type 2 diabetes. Conversely, genetically instrumented television watching time (IVW OR, 1.39; 95% CI: 1.23–1.57; adjusted P‐value < 0.001), driving time (IVW OR, 3.28; 95% CI: 1.27–8.48; adjusted P‐value = 0.049), insomnia (IVW OR, 1.32; 95% CI: 1.22–1.42; adjusted P‐value < 0.001), smoking initiation (IVW OR, 1.17; 95% CI: 1.10–1.24; adjusted P‐value < 0.001), lifetime smoking index (IVW OR, 1.37; 95% CI: 1.22–1.53; adjusted P‐value < 0.001), alcohol dependence (IVW OR, 1.17; 95% CI: 1.06–1.29; adjusted P‐value = 0.01), coffee consumption (IVW OR, 1.01; 95% CI: 1.00–1.02; adjusted P‐value <0.001), and Townsend deprivation index (IVW OR, 1.51; 95% CI: 1.11–2.07; adjusted P‐value = 0.04) were associated with increased odds of type 2 diabetes. Additionally, genetically predicted smoking heaviness was associated with higher odds of type 1 diabetes (IVW OR, 1.77; 95% CI: 1.35–2.34; adjusted P‐value = 0.01).

Genetically predicted educational attainment (IVW OR, 0.64; 95% CI: 0.56–0.73; adjusted P‐value <0.001), age completed full‐time education (IVW OR, 0.44; 95% CI: 0.32–0.59; adjusted P‐value <0.001), and annual household income (IVW OR, 0.52; 95% CI: 0.41–0.66; adjusted P‐value <0.001) were associated with protective effects against gestational diabetes. In contrast, genetically instrumented insomnia (IVW OR, 1.40; 95% CI: 1.26–1.56; adjusted P‐value < 0.001) and smoking initiation (IVW OR, 1.16; 95% CI: 1.06–1.26; adjusted P‐value = 0.01) were associated with increased odds of gestational diabetes. In the MVMR analysis, the genetically predicted age at completion of full‐time education retained a strong inverse association with the risk of gestational diabetes (estimate, −1.12; 95% CI: −2.07 to −0.16).

Genetic associations between lifestyle and socioeconomic factors and diabetes complications

The genetically predicted lifetime smoking index was positively associated with an increased risk of neovascular glaucoma (IVW beta, 7.04; 95% CI: 4.60–9.48; adjusted P‐value < 0.001). MR‐Egger intercept analysis showed no evidence of horizontal pleiotropy (P‐value = 0.29). The Q‐statistic showed no evidence of significant heterogeneity (P‐value = 0.49). The F‐statistics for the genetic instruments confirmed the absence of weak instrument bias. Additionally, the MR analysis showed no evidence of associations between lifestyle and socioeconomic factors and complications such as diabetic retinopathy, nephropathy, neuropathy, maculopathy, Charcot foot, and acanthosis nigricans.

Genetic associations between lifestyle and socioeconomic factors and diabetes comorbidities

Genetically predicted alcohol consumption (IVW OR, 0.13; 95% CI: 0.04–0.50; adjusted P‐value = 0.03) and age completed full‐time education (IVW OR, 0.67; 95% CI: 0.53–0.84; adjusted P‐value = 0.01) were associated with protective effects against stroke. In the MVMR analysis, genetically predicted alcohol consumption retained a strong inverse association with stroke risk (estimate, −1.78; 95% CI: −2.96 to −0.61). Moreover, genetically predicted educational attainment was associated with a protective effect against coronary heart disease (IVW OR, 0.21; 95% CI: 0.08–0.59; adjusted P‐value = 0.03). Furthermore, genetically predicted short sleep duration (IVW OR, 0.04; 95% CI: 0.01–0.16; adjusted P‐value < 0.001), age completed full‐time education (IVW OR, 0.52; 95% CI: 0.37–0.72; adjusted P‐value < 0.001), and annual household income (IVW OR, 0.62; 95% CI: 0.49–0.79; adjusted P‐value < 0.001) were associated with protective effects against heart failure. In contrast, genetically instrumented insomnia (IVW OR, 1.21; 95% CI: 1.08–1.35; adjusted P‐value = 0.01), lifetime smoking index (IVW OR, 1.53; 95% CI: 1.25–1.88; adjusted P‐value <0.001), alcohol dependence (IVW OR, 1.28; 95% CI: 1.08–1.52; adjusted P‐value = 0.04), and coffee consumption (IVW OR, 1.01; 95% CI: 1.00–1.02; adjusted P‐value = 0.04) were associated with increased odds of heart failure. Additionally, genetically instrumented computer use time (IVW OR, 0.13; 95% CI, 0.05–0.30; adjusted P‐value < 0.001) was associated with a reduced risk of myocardial infarction.

Genetically instrumented vigorous physical activity (VPA) was associated with a reduced risk of sleep apnea (IVW OR, 0.98; 95% CI: 0.98–0.99; adjusted P‐value < 0.001). In contrast, the genetically instrumented Townsend deprivation index was associated with an increased risk of sleep apnea (IVW OR, 1.57; 95% CI; 1.18–2.09; adjusted P‐value = 0.01). In the MVMR analysis, genetically predicted VPA maintained a strong inverse association with the risk of sleep apnea (estimate, 0.02; 95% CI: 0.01–0.03). Moreover, genetically instrumented smoking heaviness was associated with a reduced risk of nonalcoholic fatty liver disease (IVW OR, 0.01; 95% CI: 0.00–0.09; adjusted P‐value = 0.01). In contrast, genetically predicted insomnia (IVW OR, 1.59; 95% CI: 1.29–1.98; adjusted P‐value < 0.001) and smoking initiation (IVW OR, 1.32; 95% CI: 1.11–1.57; adjusted P‐value = 0.04) were positively associated with an increased risk of nonalcoholic fatty liver disease. In addition, genetically instrumented insomnia was associated with a reduced risk of osteoporosis (IVW OR, 0.05; 95% CI: 0.02–0.12; adjusted P‐value < 0.001). Similarly, genetically instrumented smoking heaviness was associated with a reduced risk of osteoporosis (IVW OR, 0.08; 95% CI: 0.02–0.28; adjusted P‐value < 0.001).

Genetically predicted long sleep duration (IVW OR, 0.62; 95% CI: 0.47–0.82; adjusted P‐value = 0.01) and educational attainment (IVW OR, 0.25; 95% CI: 0.10–0.58; adjusted P‐value = 0.01) were associated with protective effects against anxiety disorders. Genetically predicted insomnia (IVW OR, 1.82; 95% CI, 1.47–2.25; adjusted P‐value < 0.001) and smoking initiation (IVW OR, 1.36; 95% CI, 1.14–1.63; adjusted P‐value = 0.04) were positively associated with an increased risk of eating disorders. Furthermore, no associations were found between genetically predicted lifestyle and socioeconomic factors and comorbidities such as hypertension, obesity, hyperlipidemia, and depressive disorders.

DISCUSSION

This MR study provides evidence supporting potential causal associations between lifestyle and socioeconomic factors and diabetes risk, complications, and comorbidities. Genetically instrumented factors—such as VPA, computer use time, carbohydrate intake, short and long sleep duration, alcohol and caffeine consumption, educational level, and annual household income—are associated with reduced risks of diabetes outcomes. Conversely, some genetically predicted factors, such as certain sedentary behaviors (i.e., television watching and driving time), insomnia, smoking behavior, alcohol dependence, coffee consumption, and the Townsend deprivation index, are associated with increased odds. These associations suggest that both lifestyle modifications and socioeconomic improvements could play an important role in the prevention and management of diabetes.

Causal pathways between lifestyle and socioeconomic factors and diabetes

In our MR analysis, several lifestyle and socioeconomic factors exhibited protective associations against diabetes. Specifically, genetically predicted carbohydrate intake, smoking cessation, caffeine consumption, educational level (i.e., educational attainment, age at completion of full‐time education), and annual household income were linked to a decreased risk of type 2 diabetes, which are consistent with previous observational 42 , 43 and meta‐analytic studies 6 , 7 , 15 , 23 , 44 . Prior MR studies 28 , 29 have similarly found that higher educational attainment and household income have protective effects against diabetes. Conversely, factors such as sedentary behaviors (i.e., television watching and driving), insomnia, smoking behaviors (i.e., smoking initiation and lifetime smoking exposure), alcohol dependence, coffee consumption, and the Townsend deprivation index were associated with an elevated risk of type 2 diabetes, indicating critical targets for intervention to prevent disease onset. Previous MR studies have also identified causal links between insomnia 45 , 46 , smoking initiation 47 , and increased diabetes risk.

However, unlike some systematic review and meta‐analyses of clinical trials 11 , 12 that suggest benefits of moderate alcohol consumption, our findings indicate that alcohol dependence significantly raises diabetes risk, underscoring the need for caution in viewing alcohol as a protective factor. In contrast to type 2 diabetes mellitus, which is more strongly linked to insulin resistance, beta‐cell dysfunction, and lifestyle‐related metabolic disturbances, type 1 diabetes mellitus is primarily an autoimmune disease with a distinct genetic architecture in which immune‐related loci, particularly the human leukocyte antigen (HLA) region, play a major role. Therefore, the observed association between genetically predicted smoking heaviness and type 1 diabetes mellitus should be interpreted cautiously. Although we applied several sensitivity analyses to assess horizontal pleiotropy, residual pleiotropy cannot be fully excluded in summary‐level MR analyses, especially for autoimmune outcomes such as type 1 diabetes mellitus. Accordingly, this finding should be viewed as exploratory rather than definitive evidence of a direct lifestyle‐mediated causal effect. For gestational diabetes, which represents a pregnancy‐specific metabolic condition characterized by gestational insulin resistance and altered maternal metabolic adaptation, the observed associations with educational level, annual household income, insomnia, and smoking initiation may be more plausibly interpreted within a lifestyle and metabolic framework.

Causal pathways between lifestyle and socioeconomic factors and diabetic complications

Our analysis identified a significant association between genetically predicted lifetime smoking index and an increased risk of neovascular glaucoma. This result aligns with prior observational studies that have reported smoking as a risk factor for vascular complications in diabetic patients, particularly those affecting ocular health 20 , 21 . However, unlike previous studies that primarily focused on associations, our study provides robust causal evidence linking smoking behaviors to an elevated risk of neovascular glaucoma in diabetes patients, emphasizing the mechanistic role of vascular stress induced by smoking. The findings suggest that integrating smoking reduction strategies into diabetes care may help mitigate the risk of such complications.

Interestingly, no associations were identified between lifestyle and socioeconomic factors and other common diabetes‐related complications, such as diabetic retinopathy, nephropathy, neuropathy, maculopathy, Charcot foot, and acanthosis nigricans. This contrasts with findings from observational studies 20 , 21 suggesting that a healthy lifestyle—characterized by regular physical activity, reduced sedentary time, healthy diet, 6–8 h of sleep per day, and moderate alcohol intake—is associated with lower risks of microvascular complications. Prior meta‐analyses also report mixed results on the effectiveness of lifestyle interventions for improving microvascular outcomes in diabetes, with some showing improvements in blood pressure and cholesterol 17 , while others indicate limited benefits 18 , 19 . Additionally, previous observational data associate higher socioeconomic status with lower risks of diabetes‐related complications, including cardiovascular disease, retinopathy, and neuropathy 22 . This discrepancy suggests that these complications may be less influenced by modifiable lifestyle and socioeconomic factors and could depend more on genetic or environmental determinants.

Causal pathways between lifestyle and socioeconomic factors and diabetic comorbidities

Our analysis revealed several protective and risk‐enhancing associations between lifestyle and socioeconomic factors and common diabetic comorbidities. Genetically predicted alcohol consumption and age at completion of full‐time education were associated with a reduced risk of stroke, which aligns with prior evidence suggesting potential benefits of moderate alcohol intake on vascular function 48 . In contrast, alcohol dependence was linked to an increased risk of heart failure, consistent with findings on its adverse cardiovascular effects, including oxidative stress 49 . Additionally, educational attainment was associated with a protective effect against coronary heart disease, which may reflect improved health literacy, access to healthcare, and healthier lifestyle choices. Short sleep duration, age at completion of full‐time education, and annual household income were associated with lower odds of heart failure, whereas insomnia, lifetime smoking index, alcohol dependence, and coffee consumption were positively associated with heart failure risk. These findings underscore the impact of sleep and substance use behaviors on cardiovascular comorbidities.

For noncardiovascular comorbidities, genetically predicted VPA was associated with a decreased risk of sleep apnea, whereas genetically predicted Townsend deprivation index was linked to an increased risk, indicating socioeconomic disparities in sleep‐related conditions. Furthermore, insomnia and smoking initiation showed positive associations with increased risks of nonalcoholic fatty liver disease. These findings suggest a complex interaction between sleep, socioeconomic factors, and substance use behaviors in the development of diabetic comorbidities.

Our results also identified protective associations of long sleep duration and educational attainment against anxiety disorders, while insomnia and smoking initiation were linked to increased risk of eating disorders. Unlike previous observational evidence that primarily reported associative relationships, our study provides novel causal evidence, emphasizing the direct impact of these factors on mental health outcomes. Notably, no associations were observed between the evaluated lifestyle and socioeconomic factors and certain comorbidities, such as hypertension, obesity, hyperlipidemia, and depressive disorder, highlighting the selective influence of these modifiable factors. These findings suggest that targeted interventions promoting healthy sleep behaviors, smoking prevention, and addressing socioeconomic disparities could effectively reduce the burden of specific mental health comorbidities in diabetes care.

Strengths and limitations

Our study has several strengths. First, we employed a rigorous MR framework with robust instruments, minimizing weak instrument bias and enhancing causal inference. Second, this study provides a comprehensive scope—spanning type 1, type 2, and gestational diabetes, as well as associated complications/comorbidities, offering a unified perspective on modifiable risk factors, surpassing prior MR studies. Third, the combined use of UVMR and MVMR enabled evaluation of both total and direct effects. Fourth, we applied extensive sensitivity analyses to validate key MR assumptions. Fifth, we applied extensive sensitivity analyses to validate key MR assumptions. Sixth, by focusing on policy‐relevant exposures (e.g., sleep duration, socioeconomic deprivation), the findings offer practical insights. Finally, large‐scale GWAS data from well‐characterized European cohorts enhanced statistical power and internal validity.

However, limitations should be noted. First, MR assumes that genetic variants affect outcomes only through the exposure of interest, and undetected pleiotropy may bias results 35 . This concern may be particularly relevant for type 1 diabetes mellitus, given its strong autoimmune and HLA‐related genetic architecture, and therefore type 1 diabetes mellitus‐specific findings should be interpreted with additional caution. Second, MR estimates reflect lifelong exposure, not short‐term effects, and are better suited for testing causality than quantifying precise effect sizes. Third, some exposures had few significant SNPs, possibly reducing power. Fourth, the restriction to European ancestry limits generalizability, and participation bias may influence findings.

Clinical implications and future perspectives

This study offers several clinical insights for diabetes prevention and management. First, while lifestyle interventions are already integral to care, our study provides causal evidence linking factors—such as VPA, healthy sleep, and reduced sedentary time—to lower diabetes risk. Second, protective associations with education and income emphasize the need to address social determinants of health, which are often overlooked in clinical settings. Integrating socioeconomic support into diabetes care could enhance lifestyle adherence and access to preventive resources. Third, these results support a more holistic model that combines medical treatment with behavioral and structural interventions to reduce diabetes burden.

CONCLUSION

This MR study provides evidence supporting potential causal roles of lifestyle and socioeconomic factors in diabetes risk, complications, and comorbidities. The protective associations of factors such as VPA, computer use time, carbohydrate intake, short and long sleep duration, alcohol and caffeine consumption, educational level, and annual household income highlight the potential impact of modifiable lifestyle behaviors and socioeconomic improvements on diabetes prevention. Conversely, risk‐enhancing associations observed with certain sedentary behaviors (i.e., television watching time and driving time), insomnia, smoking behaviors, alcohol dependence, coffee consumption, and the Townsend deprivation index suggest critical targets for interventions aimed at reducing diabetes risk. These findings support the development of public health strategies that address both individual behavior and social determinants of health. Future work should extend these insights to more diverse populations and evaluate their impact in real‐world interventions.

DISCLOSURE

The authors declare no conflict of interest.

Informed consent: This study used publicly available summary‐level genome‐wide association study data from the MRC IEU OpenGWAS database, the FinnGen consortium, and previously published studies. No individual‐level participant data were used. Ethical approval and written informed consent had been obtained in the original studies contributing the GWAS summary statistics.

Registry and registration no. of the study: N/A. This was a Mendelian randomization study based on publicly available summary‐level GWAS data and was not a clinical trial or prospectively registered interventional study.

Animal Studies: N/A. No animal experiments were performed in this study.

Supporting information

Data S1 (i) Description of GWAS summary statistics for exposures. (ii) Univariable Mendelian randomization (UVMR) results.

JDI-17-1218-s001.docx (784.5KB, docx)

ACKNOWLEDGMENTS

The authors wish to acknowledge Drs. Jonathan Little and Kaja Falkenhain (University of British Columbia‐Okanagan Campus) for constructive feedback on the manuscript. This work was supported by the Science and Technology Development Fund (FDCT) of Macau (Grant No.: 0011/2021/ITP) and the University of Macau (Grant No.: MYRG‐GRG2023‐00086‐FED. The funder had no role in the design, implementation, analysis, or interpretation of the data.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available in public repositories. These data were derived from the following resources available in the public domain: the MRC IEU OpenGWAS database (https://gwas.mrcieu.ac.uk/), the FinnGen consortium (https://www.finngen.fi/en), and previously published studies referenced in the manuscript.

REFERENCES

  • 1. Kharroubi AT, Darwish HM. Diabetes mellitus: The epidemic of the century. World J Diabetes 2015; 6: 850–867. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Association AD . Diagnosis and classification of diabetes mellitus. Diabetes Care 2010; 33(Supplement_1): S62–S69. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Magliano DJ, Boyko EJ, Atlas ID. COVID‐19 and diabetes. IDF Diabetes Atlas [Internet], 10th edn. Brussels, Belgium: International Diabetes Federation, 2021. [Google Scholar]
  • 4. Sun H, Saeedi P, Karuranga S, et al. IDF diabetes Atlas: Global, regional and country‐level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract 2022; 183: 109119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Deshpande AD, Harris‐Hayes M, Schootman M. Epidemiology of diabetes and diabetes‐related complications. Phys Ther 2008; 88: 1254–1264. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Gillies CL, Abrams KR, Lambert PC, et al. Pharmacological and lifestyle interventions to prevent or delay type 2 diabetes in people with impaired glucose tolerance: Systematic review and meta‐analysis. BMJ 2007; 334: 299. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Uusitupa M, Khan TA, Viguiliouk E, et al. Prevention of type 2 diabetes by lifestyle changes: A systematic review and meta‐analysis. Nutrients 2019; 11: 2611. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Guo XY, Shu J, Fu XH, et al. Improving the effectiveness of lifestyle interventions for gestational diabetes prevention: A meta‐analysis and meta‐regression. BJOG 2019; 126: 311–320. [DOI] [PubMed] [Google Scholar]
  • 9. Yaggi HK, Araujo AB, McKinlay JB. Sleep duration as a risk factor for the development of type 2 diabetes. Diabetes Care 2006; 29: 657–661. [DOI] [PubMed] [Google Scholar]
  • 10. Khalil M, Power N, Graham E, et al. The association between sleep and diabetes outcomes – A systematic review. Diabetes Res Clin Pract 2020; 161: 108035. [DOI] [PubMed] [Google Scholar]
  • 11. Howard AA, Arnsten JH, Gourevitch MN. Effect of alcohol consumption on diabetes mellitus: A systematic review. Ann Intern Med 2004; 140: 211–219. [DOI] [PubMed] [Google Scholar]
  • 12. Baliunas DO, Taylor BJ, Irving H, et al. Alcohol as a risk factor for type 2 diabetes: A systematic review and meta‐analysis. Diabetes Care 2009; 32: 2123–2132. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Willi C, Bodenmann P, Ghali WA, et al. Active smoking and the risk of type 2 diabetes: A systematic review and meta‐analysis. JAMA 2007; 298: 2654–2664. [DOI] [PubMed] [Google Scholar]
  • 14. Pan A, Wang Y, Talaei M, et al. Relation of active, passive, and quitting smoking with incident type 2 diabetes: A systematic review and meta‐analysis. Lancet Diabetes Endocrinol 2015; 3: 958–967. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Jiang X, Zhang D, Jiang W. Coffee and caffeine intake and incidence of type 2 diabetes mellitus: A meta‐analysis of prospective studies. Eur J Nutr 2014; 53: 25–38. [DOI] [PubMed] [Google Scholar]
  • 16. Stutz B, Ahola A, Harjutsalo V, et al. Association between habitual coffee consumption and metabolic syndrome in type 1 diabetes. Nutr Metab Cardiovasc Dis 2018; 28: 470–476. [DOI] [PubMed] [Google Scholar]
  • 17. Huang X‐L, Pan J‐H, Chen D, et al. Efficacy of lifestyle interventions in patients with type 2 diabetes: A systematic review and meta‐analysis. Eur J Intern Med 2016; 27: 37–47. [DOI] [PubMed] [Google Scholar]
  • 18. Schellenberg ES, Dryden DM, Vandermeer B, et al. Lifestyle interventions for patients with and at risk for type 2 diabetes: A systematic review and meta‐analysis. Ann Intern Med 2013; 159: 543–551. [DOI] [PubMed] [Google Scholar]
  • 19. Chen L, Pei J‐H, Kuang J, et al. Effect of lifestyle intervention in patients with type 2 diabetes: A meta‐analysis. Metabolism 2015; 64: 338–347. [DOI] [PubMed] [Google Scholar]
  • 20. Geng T, Zhu K, Lu Q, et al. Healthy lifestyle behaviors, mediating biomarkers, and risk of microvascular complications among individuals with type 2 diabetes: A cohort study. PLoS Med 2023; 20: e1004135. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Liu G, Li Y, Pan A, et al. Adherence to a healthy lifestyle in association with microvascular complications among adults with type 2 diabetes. JAMA Netw Open 2023; 6: e2252239. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Secrest AM, Costacou T, Gutelius B, et al. Associations between socioeconomic status and major complications in type 1 diabetes: The Pittsburgh epidemiology of diabetes complication (EDC) study. Ann Epidemiol 2011; 21: 374–381. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Scott A, Chambers D, Goyder E, et al. Socioeconomic inequalities in mortality, morbidity and diabetes management for adults with type 1 diabetes: A systematic review. PLoS One 2017; 12: e0177210. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Cartwright N, Munro E. The limitations of randomized controlled trials in predicting effectiveness. J Eval Clin Pract 2010; 16: 260–266. [DOI] [PubMed] [Google Scholar]
  • 25. Benson K, Hartz AJ. A comparison of observational studies and randomized, controlled trials. Research Ethics. London: Routledge, 2017; 213–221. [Google Scholar]
  • 26. Sanderson E, Glymour MM, Holmes MV, et al. Mendelian randomization. Nat Rev Methods Primers 2022; 2: 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Richmond RC, Smith GD. Mendelian randomization: Concepts and scope. Cold Spring Harb Perspect Med 2022; 12: a040501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Adams CD, Boutwell BB. Can increasing years of schooling reduce type 2 diabetes (T2D)? Evidence from a Mendelian randomization of T2D and 10 of its risk factors. Sci Rep 2020; 10: 12908. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Mai Z, Wang S, Chen H, et al. Genetically predicted lifestyle factors, socioeconomic status and risk of coronary artery disease in individuals with diabetes: A Mendelian randomization study. Front Public Health 2023; 11: 1284958. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. de Leeuw C, Savage J, Bucur IG, et al. Understanding the assumptions underlying Mendelian randomization. Eur J Hum Genet 2022; 30: 653–660. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Sanderson E. Multivariable Mendelian randomization and mediation. Cold Spring Harb Perspect Med 2021; 11: a038984. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Jannot A‐S, Ehret G, Perneger T. P < 5 × 10−8 has emerged as a standard of statistical significance for genome‐wide association studies. J Clin Epidemiol 2015; 68: 460–465. [DOI] [PubMed] [Google Scholar]
  • 33. Purcell S, Neale B, Todd‐Brown K, et al. PLINK: A tool set for whole‐genome association and population‐based linkage analyses. Am J Hum Genet 2007; 81: 559–575. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Burgess S, Thompson SG, Collaboration CCG . Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol 2011; 40: 755–764. [DOI] [PubMed] [Google Scholar]
  • 35. Hemani G, Bowden J, Davey SG. Evaluating the potential role of pleiotropy in Mendelian randomization studies. Hum Mol Genet 2018; 27(R2): R195–R208. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36. Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol 2013; 37: 658–665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Greco MFD, Minelli C, Sheehan NA, et al. Detecting pleiotropy in Mendelian randomisation studies with summary data and a continuous outcome. Stat Med 2015; 34: 2926–2940. [DOI] [PubMed] [Google Scholar]
  • 38. Bowden J, Del Greco MF, Minelli C, et al. Improving the accuracy of two‐sample summary‐data Mendelian randomization: Moving beyond the NOME assumption. Int J Epidemiol 2019; 48: 728–742. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Verbanck M, Chen C‐Y, Neale B, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018; 50: 1196. [DOI] [PubMed] [Google Scholar]
  • 40. Davies NM, Holmes MV, Smith GD. Reading Mendelian randomisation studies: A guide, glossary, and checklist for clinicians. BMJ 2018: 362: k601. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Sanderson E, Windmeijer F. A weak instrument F‐test in linear IV models with multiple endogenous variables. J Econom 2016; 190: 212–221. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42. Connolly V, Unwin N, Sherriff P, et al. Diabetes prevalence and socioeconomic status: A population based study showing increased prevalence of type 2 diabetes mellitus in deprived areas. J Epidemiol Community Health 2000; 54: 173–177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Bo S, Menato G, Bardelli C, et al. Low socioeconomic status as a risk factor for gestational diabetes. Diabetes Metab 2002; 28: 139–140. [PubMed] [Google Scholar]
  • 44. Agardh E, Allebeck P, Hallqvist J, et al. Type 2 diabetes incidence and socio‐economic position: A systematic review and meta‐analysis. Int J Epidemiol 2011; 40: 804–818. [DOI] [PubMed] [Google Scholar]
  • 45. Gao X, Sun H, Zhang Y, et al. Investigating causal relations between sleep‐related traits and risk of type 2 diabetes mellitus: A Mendelian randomization study. Front Genet 2020; 11: 607865. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Liu M, Yu X, Shi J, et al. Establishing causal relationships between insomnia and gestational diabetes mellitus using Mendelian randomization. Heliyon 2024; 10: e33638. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47. Yuan S, Larsson SC. A causal relationship between cigarette smoking and type 2 diabetes mellitus: A Mendelian randomization study. Sci Rep 2019; 9: 19342. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. Fernández‐Solà J. Cardiovascular risks and benefits of moderate and heavy alcohol consumption. Nat Rev Cardiol 2015; 12: 576–587. [DOI] [PubMed] [Google Scholar]
  • 49. Tsermpini EE, Plemenitaš Ilješ A, Dolžan V. Alcohol‐induced oxidative stress and the role of antioxidants in alcohol use disorder: A systematic review. Antioxidants 2022; 11: 1374. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data S1 (i) Description of GWAS summary statistics for exposures. (ii) Univariable Mendelian randomization (UVMR) results.

JDI-17-1218-s001.docx (784.5KB, docx)

Data Availability Statement

The data that support the findings of this study are available in public repositories. These data were derived from the following resources available in the public domain: the MRC IEU OpenGWAS database (https://gwas.mrcieu.ac.uk/), the FinnGen consortium (https://www.finngen.fi/en), and previously published studies referenced in the manuscript.

Articles from Journal of Diabetes Investigation are provided here courtesy of Wiley

RESOURCES