Summary
Background
The ongoing war in Ukraine have exposed the population to unprecedented stressors the long-term metabolic consequences of which remain insufficiently characterised. This study evaluated changes in glycaemic control and related biomarkers in the Ukrainian population during ongoing war between 2021 and 2024.
Methods
This large-scale, single centre repeated cross-sectional study analysed routine laboratory data across Ukraine from January 2021 to December 2024. We included a convenience sample of adults aged 18–90 years with complete demographic data who underwent on fasting plasma glucose (FPG) (108,346 individuals), of whom 86,077 also had glycated haemoglobin (HbA1c) results. Additional data was available for 2502 insulin and 9000 cortisol measurements. Biomarker trends were assessed annually and quarterly, stratified by sex, age, and geographic region, and categorised as corresponding to normoglycemia, prediabetes, and diabetes.
Findings
In 2021, median HbA1c was 5.71% (interquartile range; IQR 5.34–6.34), with 28.7% of individuals in the prediabetic and 22.3% in the diabetic range. By 2024, median HbA1c rose to 5.91%, with prediabetes increasing from 28.7% to 37.7%, and diabetic levels rate growing from 22.3% to 26.4%. As a result, 64% of all individuals tested in 2024 had HbA1c above the normoglycaemic threshold. Women and working-age adults showed the greatest glycaemic deterioration, despite men having higher absolute levels.
Interpretation
The Ukrainian population experienced significant metabolic deterioration during the war. Rising HbA1c and the growing prevalence of prediabetes and diabetes, particularly among women and younger adults, indicate escalating cardiometabolic risks. Future prospective longitudinal studies are needed to evaluate the long-term effectiveness of integrated metabolic and psychosocial interventions in conflict-affected populations.
Funding
National Research Foundation of Ukraine.
Keywords: Ukraine, Armed conflicts, Diabetes mellitus, Glucose metabolism disorders, Prediabetic state, Glycated haemoglobin
Research in context.
Evidence before this study
We searched PubMed, Embase, and Google Scholar for articles published in English from database inception up to June 30, 2024, using combinations of the terms “war”, “conflict”, “stress”, “glycaemia”, “HbA1c”, “insulin”, and “cortisol”, which yielded 215 studies. Previous studies demonstrated that acute and chronic stress related to pandemics, displacement, and different military conflict zones is associated with impaired glycaemic control, elevated rates of diabetes, and increased risk of cardiovascular disease linked to stress physiology and disrupted healthcare access. However, there is a lack of large-scale, population-wide longitudinal analyses quantifying real-world metabolic biomarker trends during an active, full-scale armed conflict.
Added value of this study
To our knowledge, this is the first large-scale study examining the long-term metabolic consequences of war on a civilian population. Using routine laboratory data covering the 2021–2024 period, we quantified changes in metabolic biomarkers (glycated haemoglobin (HbA1c), fasting plasma glucose, insulin, cortisol) across time, geography, sex, and age groups in Ukraine.
The study revealed that the war was associated with a measurable deterioration in glycaemic control at the population level, with the proportion of individuals in the prediabetic or diabetic HbA1c ranges rising from 51% in 2021 to 64% in 2024. Notably, women and working-age adults were the most vulnerable groups, exhibiting the most significant rise in HbA1c levels, despite lower baseline values. We also revealed regional disparities linked to the intensity of military activity, supporting a stress-mediated mechanism for metabolic shift.
By linking objective laboratory data with major societal stressors, this study provides unique data of the metabolic costs of prolonged military conflict, highlighting the need for integrated metabolic and psychosocial interventions in conflict-affected settings.
Implications of all the available evidence
Our findings contribute to the growing evidence that humanitarian disasters and societal stressors, including pandemics and armed conflict, accelerate metabolic deterioration in populations. This study revealed that the ongoing war in Ukraine is accompanied by a marked rise in dysglycemia, particularly among women and younger adults, signalling a high future burden of diabetes and cardiovascular complications in the absence of targeted interventions. Taken together with existing global evidence, these results underscore the need to integrate metabolic monitoring, diabetes prevention, and mental health support into humanitarian health responses.
Future prospective cohort studies tracking individual biomarker trajectories alongside specific psychological and environmental stressors are needed to optimize tailored metabolic interventions in conflict settings.
Introduction
Over the past four decades, the global number of adults living with diabetes has quadrupled, rising from 108 million in 1980 to 463 million in 2019, with 90–95% of cases being type 2 diabetes (T2D).1 In 2024, 589 million adults aged 20–79 were living with diabetes. According to projections by the International Diabetes Federation (IDF), this number is expected to reach 853 million by 2050.2
However, the real-world challenges and data witness a much faster progression of diabetes spread. Various global health threats escalate T2D development and progression. According to the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2021 Diabetes Collaborators, around 529 (95% uncertainty interval [UI] 500–564) million people lived with diabetes (96.0%–with T2D), corresponding to 6.1% (5.8–6.5) of the world population.3 Moreover, in 2021 1.70 million died due to diabetes defining about 37.8 million years of life lost, and 79.2 million disability-adjusted life-years (DALYs) from diabetes.4 For the last four years, Ukraine has been experiencing Russia's full-scale invasion (February 24, 2022), triggering the largest humanitarian crisis in Europe since World War II affecting various aspects of healthcare.5, 6, 7
Recent studies have identified a cumulative negative impact of war-related factors on the health of people with T2D in Ukraine, manifesting in worsened glycaemic control and elevated glycated haemoglobin (HbA1c) levels.8 The military aggression, which began on February 24, 2022, has significantly disrupted diabetes care9 Living in occupied territories under ongoing invasion, destruction of housing and infrastructure, and forced displacement are among the most significant war-related stressors that irreversibly affect T2D progression.8 As a result, a substantial part of the Ukrainian population has been left without adequate glycaemic control and essential treatment during the war, undermining the course and management of diabetes.8
Continuous exposure to psychological stress causes repeated or prolonged dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis and sympathetic nervous system, leading to hypercoagulability, oxidative stress, dyslipidemia, immune dysfunction, chronic inflammation, and cardiovascular threats—all of which increase the risk of metabolic disturbance and insulin resistance over time10 and T2D.11 Although the number of studies reporting the influence of war on Ukrainian civilians, internally displaced persons, and refugees is growing, there remains limited data about the population changes in glycaemic control at different regions of Ukraine. Since the start of the full-scale invasion in February 2022, exposure to war-related threats have varied geographically across Ukraine. Eastern regions located near the frontline have experienced the most intense and sustained military activity, while central regions including Kyiv, as well as southern Ukraine, have been frequently affected by missile and drone attacks, resulting in repeated disruption of civilian infrastructure and healthcare services.
We therefore analysed routine laboratory data from 2021 to 2024 to quantify changes in fasting plasma glucose (FPG), HbA1c, insulin, and cortisol in Ukraine, and to examine differences by sex, age, and region. Our goal was to assess whether war as a humanitarian crisis is associated with deterioration in glycaemic control in people living in Ukraine.
Methods
Ethics
All data were de-identified prior to analysis. The study was approved by the Ethics Committee of the Medical laboratory CSD (№25/2 from 03.02.2025) and adhered to the ethical principles of the Declaration of Helsinki and did not require separate informed consent due to its retrospective design and the anonymised nature of the data sources.
Study design
We conducted a repeated cross-sectional study of routine laboratory data from Ukraine covering the pre-war period and the ongoing war (January 2021–December 2024). That was a single centre study, based on the data retrieved from laboratory data base of Medical Laboratory CSD (Kyiv, Ukraine), which operates all over Ukraine. Analyses were conducted using aggregated measures of laboratory biomarkers data across several years with additional stratification by sex, age group, and geographic region. Relative distributions and stratified comparisons within each time period were assessed to reduce the impact of unequal sample sizes across years. Regional stratification (northern, central, western, eastern, and southern Ukraine) was performed to explore potential geographic differences in glycaemic biomarkers, considering that the intensity of military actions and war-related threads, including population displacement and disruption of healthcare services, varied substantially across different geographic parts of Ukraine during the study period.
Data source and laboratory procedures
The anonymised data on the levels of FPG (mmol/L), HbA1c (%), insulin (μU/mL), and cortisol (nmol/L) were retrieved from the CSD laboratory database, and included mixed patient's categories tested for check-up, diagnostic or treatment monitoring reasons. The reported numbers correspond to test observations rather than a longitudinal cohort of uniquely followed individuals. Because the data were retrieved in anonymised form, it was not possible to link repeated measurements belonging to the same person across different time periods. Therefore, analyses were conducted using independent observations within each time period, and temporal comparisons reflect changes in aggregated laboratory measurements over time rather than within-person longitudinal trajectories.
For laboratory testing, fasting venous blood samples were collected in outpatient laboratory settings via venipuncture between 8:00 and 9:00 a.m. following an overnight fast. Samples were immediately processed and sent for testing to an ISO15189 (an international standard that specifies the requirements for quality and competence in medical laboratories) certified laboratory (Medical Laboratory CSD, Kyiv), that is one of the several large laboratories in Ukraine. The laboratory network of CSD has blood collecting offices over all regions of Ukraine, excluding the occupied territories. FPG concentration was measured by enzymatic test (hexokinase method) (Beckman Coulter DxC700, USA). HbA1c was evaluated via high-performance liquid chromatography (Toson G11, Japan). Cortisol levels were measured by electrochemiluminescent method (Cobas e801; Roche, Switzerland), insulin concentration in serum was assessed by immunochemiluminescent methods (DxI800, Beckman Coulter, USA). All assays were performed according to manufacturer instructions and internal quality-control procedures.
Study population and eligibility criteria
We included adults aged 18–90 years with complete demographic data (sex, age, and region). We excluded records missing any of these variables and records for individuals <18 or >90 years. The sampling approach can be characterised as a large-scale convenience sample of individuals who underwent laboratory testing for clinical reasons (check-up, diagnostic evaluation, or treatment monitoring), rather than a probabilistic population-based sample. The dataset included all eligible adult individuals (18–90 years old) with complete demographic data (sex, age, region) who had FPG and HbA1 testing between January 2021 and December 2024 (n = 108,346). No additional sampling or selection procedures were applied beyond the articulated eligibility criteria.
Records with missing key demographic variables (sex, age, or geographic region) were excluded from the analysis to ensure consistency in stratified comparisons.
Measures and definitions
The dataset comprised 108,346 FPG measurements, of which 86,077 were accompanied with HbA1c data. Additionally, 2502 records had insulin and 9000 had cortisol measurements. Insulin and cortisol measurements were available only in subsets of individuals undergoing additional laboratory testing and were analysed as exploratory variables. HbA1c categories were defined as: <5.7% (normoglycemia), 5.7–6.4% (prediabetes), and ≥6.5% (diabetes).
Temporal trends and subgroup analyses of HbA1c
At the first step, the general dynamics of the biomarkers reflecting carbohydrates metabolism were analysed annually and quarterly. The overall period of observation was subdivided into 16 quartiles (four per year).
The next step included analysis of the HbA1c levels at the sample of 86,077 observations, with following sex and age analysis for defining the groups of people who were the most vulnerable to war-related factors. In addition to HbA1c value, the percentage of observations representing HbA1c levels below 5.7%, 5.7–6.4% and ≥6.5% were calculated for each period. The laboratory dataset did not contain clinical information on prior diabetes diagnosis, medication use, treatment adherence, or other clinical characteristics. Consequently, it was not possible to distinguish individuals with previously diagnosed diabetes from those without known disease or to determine treatment status at the time of testing.
To assess the dynamics in prevalence of cases with increased HbA1c levels the segmented (year-by-year) annual percentage change (APC) was calculated using the formula (R{t+1}/R{t}–1) × 100%, where represents the prevalence in year t. In addition, the average APC (AAPC) was counted for the 2021–2024 period, reflecting the general dynamics.
Stratification variables
For comparative analysis, the data were stratified by year (2021—pre-war period; 2022–2024—wartime), sex (man, woman), age (according to the World Health Organization: 18–45 years—young adults, 46–65 years—middle-aged adults, 66–75 years—older, 75+ years—oldest age), and region of residence. Considering that the composition of the tested population may have been influenced by population displacement, demographic shifts related to the mobilization of military-aged men, and changes in healthcare-seeking behaviour, the annual distribution of samples by sex and age group was analysed across the study period. To account for the geographic differences in the effects of war on adults, the following regions of Ukraine were considered: northern, central, western, eastern, and southern. Since the onset of the full-scale invasion in 2022, hostilities have been distributed heterogeneously across Ukraine, with eastern and southern regions experiencing the most sustained frontline combat and extensive infrastructure damage, while western regions were relatively less exposed. In addition, central regions, including Kyiv, have been repeatedly affected by large-scale missile and drone attacks. Across multiple regions, targeted strikes on critical energy infrastructure resulted in recurrent, large-scale blackouts, further exacerbating regional differences in living conditions, healthcare provision, and environmental stressors. Such spatial heterogeneity in military exposure, combined with differences in displacement patterns and exposure duration to stressors, provides an important context for examining whether chronological trends in glycaemic biomarkers differ across regions.
Statistical analysis
Data processing was conducted using Statistical software EZR v. 1.67 (graphical user interface for R statistical software version 4.3.1). Descriptive statistics (median and interquartile range) were used to characterise the sample. Although HbA1c values had a normal distribution when assessed at the annual aggregate level, our analysis was performed quarterly and included additional stratification by sex, age, and geographic region. In several of these smaller subgroups, particularly during early 2022, when the number of laboratory tests temporarily declined due to the russian invasion, the distribution of HbA1c deviated from normality. Given the heterogeneity in subgroup sizes and instability of distributional assumptions, non-parametric statistical methods were applied consistently across all analyses. To compare intergroup differences, the following statistical methods were applied: Mann–Whitney test for two groups; Kruskal–Wallis test for more than two groups and Dunn test (for post-hoc pairwise comparisons). Differences in the distribution of HbA1c categories across years were assessed using the chi-square test. Statistical significance was set at p < 0.05.
To partially account for differences in sample composition over time, analyses were stratified by sex, age group, and geographic region. In addition, the annual distribution of samples by sex and age group was examined to assess potential shifts related to population displacement, demographic changes, and altered healthcare-seeking behaviour during the war. No multivariable adjustment for confounding was performed, as the anonymised laboratory dataset did not include individual-level information on prior diabetes diagnosis, medication use, treatment adherence, body mass index, lifestyle factors, or socioeconomic status. Therefore, the findings should be interpreted as descriptive associations and temporal trends within the tested sample rather than adjusted causal estimates.
Several analytical approaches were applied to assess the robustness of the findings. Temporal trends were evaluated using both annual and quarterly aggregation to ensure consistency across different time resolutions. In addition, analyses were stratified by sex, age group, and geographic region to examine whether observed patterns were consistent across subgroups. Given substantial variation in sample sizes across years, relative distributions and within-period comparisons were emphasised to reduce the influence of unequal sample composition. Consistent patterns observed across these complementary analyses support the stability of the reported findings.
Role of the funding source
The funder of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.
Results
The sample of HbA1 data included both men (n = 30,728; 35.7%) and women (n = 55,349; 64.3%) of various age groups. Assessment of the metabolic profile of the Ukrainian people based on routine laboratory data of 2021 (Table S1) revealed a median FPG of 5.17 (4.72–5.79), while median HbA1c was 5.71 (5.34–6.34) within the prediabetic range.
Analysis of data revealed significant annual and quarterly differences in metabolic parameters (Table S1). The highest values of FPG were observed in 2021 and 2022, with medians of 5.17 (4.72–5.79) and 5.2 (4.78–5.74; p < 0.001), respectively. These levels decreased to 4.92 (4.53–5.45) in 2023 and 4.94 (4.57–5.44) in 2024. Notably, the highest FPG values of FPG were detected during periods 4–6, corresponding to Q4/2021–Q2/2022 (Table S1), which aligned with the prewar and early wartime crises.
HbA1c levels differed significantly across the observed periods, rising from a median of 5.71 (5.34–6.34) in 2021 to 5.91 (5.53–6.56) in 2024 (p < 0.001). The gradual increase indicated worsening glycaemic control in Ukraine during the wartime (Table S1). The highest HbA1c values were recorded in mid-2021 followed by a slight decline and subsequent upward trend from mid-2022 through the end of 2024 (Table S1). As the dataset was derived from anonymised laboratory records lacking detailed clinical information on prior diabetes diagnosis, treatment status, and medication access, it was not possible to determine whether the observed increase in HbA1c reflects progression from prediabetes to diabetes among previously unaffected individuals, worsening glycaemic control in those with established diabetes, or a combination of both.
Insulin and cortisol analyses were available only in smaller subsets of individuals undergoing additional laboratory testing. Insulin levels showed a gradual increase throughout the study, peaking in early 2023 (Period 9, Q1-2023) at 12.07 (7.47–16.53), coinciding with the long-term blackout period (Table S1), with a decline in late 2023, elevation at the beginning of 2024 to 11.13 (7.01–14.98), and following reduction. Cortisol showed the highest values in 2022–2023, fluctuating over the analysed period. Considering the limited availability and potential indication bias, insulin and cortisol findings should be considered exploratory.
Assessing the number of people with different ranges of HbA1c, we found a high proportion of people with prediabetic and diabetic levels. In 2021, an HbA1c level of 5.7–6.4 was found in 28.7% (1052 of 3670 tested) and an HbA1c level of 6.5 and higher was recorded in 22.3% (820 of 3670 tested) of the retrieved cases. Thus, about half of the observed sample demonstrated elevated levels of HbA1c. Over the war period (2022–2024), the prevalence of both prediabetic and diabetic levels of HbA1c were significantly elevated in comparison to 2021 (2021 vs 2022, 2021 vs 2023, and 2021 vs 2024 chi-square test with Bonferroni correction p < 0.001; Fig. 1). In 2024, the percentage of people with HbA1c of 5.7–6.4% rose to 37.7% (22,106 of 58,645 cases). Similarly, the proportion of people with HbA1c ≥ 6.5% increased from 22.3% in 2021 to 26.4% (15,459 of 58,645 cases) in 2024. Overall, 64.1% (37,565 of 58,645 cases) of tested individuals in 2024 had HbA1c above the normoglycaemic threshold (Fig. 1).
Fig. 1.
The changes in HbA1c levels (a) and the proportions of the people with different HbA1c ranges over the period of observation (b). A–demonstrates the dynamics of median and 95% CI. B–reflects the proportion of people with different levels of HbA1c. Definitions: HbA1c, glycated haemoglobin; CI, confidence interval.
Importantly, classification into normoglycaemic, prediabetic, and diabetic ranges was based solely on HbA1c thresholds and reflects biochemical categories rather than confirmed clinical diagnoses. Therefore, individuals with HbA1c values in the diabetic range may include both persons with previously diagnosed diabetes and individuals without known disease. Notably, in 2022, the percentage of people with prediabetic HbA1c levels increased in Q2 and Q3 (periods 6 and 7), corresponding to the spring and summer of 2022 following the Russian full-scale invasion. Moreover, from the second half of 2023 (periods 11 and 12), we recorded another sustained increase in the proportion of people with elevated levels of HbA1c (Table S1) (p < 0.001).
Assessing the geographic difference in glycaemic profile of people underwent laboratory testing in 2021, higher HbA1c levels were observed in eastern and southern regions of Ukraine (p = 0.019, Table 1). After the full-scale invasion in 2022, the highest values of HbA1c were recorded in eastern and central regions of Ukraine, while the lowest HbA1c was found in western Ukraine. A further rise in HbA1c across all the geographic zones led to significantly higher levels of HbA1c by 2024, when the median in every region's reached 5.9–6.0%, compared with 5.5–5.9% in 2021–2023 (Table 1).
Table 1.
The annual differences in HbA1c levels at different regions of Ukraine.
| Year | Меdian (IQR) |
p | |||||
|---|---|---|---|---|---|---|---|
| Total | Northern | Central | Western | Eastern | Southern | ||
| 2021 | 5.71 (5.34–6.34) | 5.71 (5.34–6.30) | 5.74 (5.31–6.46) | 5.62 (5.32–6.21) | 5.86 (5.45–7.01) | 5.83 (5.44–6.97) | 0.019 |
| 2022 | 5.66 (5.28–6.33) | 5.66 (5.30–6.37)c | 5.72 (5.31–6.44)c,e | 5.53 (5.21–6.06)a,b | 5.76 (5.33–6.24) | 5.48 (5.18–6.10)b | <0.001 |
| 2023 | 5.73 (5.34–6.38) | 5.70 (5.33–6.31)b,d,e | 5.83 (5.40–6.58)a,c,e | 5.68 (5.29–6.31)b,d,e | 5.82 (5.44–6.56)a,c,e | 5.55 (5.22–6.0)a,b,c,d | <0.001 |
| 2024 | 5.91 (5.54–6.50) | 5.91 (5.54–6.50)b,c,d | 5.96 (5.58–6.6)a,c,d,e | 5.87 (5.45–6.57)a,b,d,e | 5.99 (5.58–6.9)a,b,c,e | 5.92 (5.56–6.5)b,c,d | <0.001 |
Notes: Data are presented as Median (IQR).
p—by the Kruskal–Wallis test, Dunn's test was used for posterior pairwise comparisons.
Definitions: HbA1c–glycated haemoglobin.
Statistical significance difference from the Northern region, p < 0.05.
Statistical significance difference from the Central region, p < 0.05.
Statistical significance difference from the Western region, p < 0.05.
Statistical significance difference from the Easter region, p < 0.05.
Statistical significance difference from the Southern region, p < 0.05.
Thus, from 2022 to 2024, HbA1c levels rose steadily across all regions, with the eastern and central regions showing the highest values.
Considering the potential mechanisms of altered glycaemic control across geographic zones we assessed the differences in insulin and cortisol levels. The highest insulin level during the pre-war period and in 2022 was recorded in the eastern region: 14.18 (8.20–20.38) in 2021 and 14.42 (9.24–21.07) in 2022, followed by a decline in 2023 and 2024 (Table S2). In contrast, a gradual increase in median insulin was observed in the southern region, peaking in 2023 at 14. 88 (8.79–19.39) (p < 0.001). By 2024, insulin levels converged (≈9–11) and dropped as compared to 2023, although interregional differences remained significant (p < 0.001).
In contrast to insulin, cortisol levels were relatively uniform across regions in pre-war period (p = 0.656). However, in 2022 a prominent interregional divergence appeared (p < 0.001) with median cortisol increasing mostly in eastern and southern regions of Ukraine, while the western region of Ukraine demonstrated the lowest cortisol levels (6.71; IQR 4.8–12.4). In 2023, cortisol levels remained heterogeneous (p < 0.001) but began to drift back toward baseline values (11–13).
Thus, in the pre-war period, HbA1c levels already predisposed approximately 50% (1872 of 3670 cases in 2021) of individuals to prediabetes or diabetes. War-related factors further promoted shift, with 64% (37,565 of 58,645 cases in 2024) of tested individuals demonstrating prediabetic or diabetic HbA1c values.
The sex distribution of the tested cases remained relatively stable throughout the study period despite a substantial increase in the number of samples analysed. Women consistently represented approximately two-thirds of the tested individuals, accounting for 64–66% each year (Table S4).
Although the highest HbA1c levels throughout the entire period were consistently observed among men, women demonstrated more dramatic growth of HbA1c in 2023–2024 resulting in similar values compared to men. By 2024, HbA1c levels reached the highest values in men 6.02 (5.62–6.93) and women 5.94 (5.56–6.50). Sex-specific trends over the years are shown in Table S3.
The analysis of HbA1c categories (normal, prediabetic, and diabetic) showed an earlier and more pronounced rise in the proportion of women with prediabetic (5.7–6.4%) and diabetic (≥6.5%) HbA1c levels (Fig. 2). From 2021 to 2024, the prevalence of women in the prediabetic range increased by 35.2% (from 698 of 2429 cases (28.7%) in 2021–14,620 of 37,720 cases (38.8%) in 2024), while the proportion in the diabetic range rose by 32.4% (from 442 of 2429 cases (18.2%) in 2021–9109 of 37,720 (24.1%) in 2024).
Fig. 2.
Sex differences in the proportions of different HbA1c levels (A, C, E) and annual percentage changes of HbA1c (B, D, F) during 2021–2024. Women demonstrated an increase in prediabetic and diabetic levels of HbA1c in 2023 and 2024, resulting in an extensive increase. Definitions: HbA1c, glycated haemoglobin.
In contrast, men demonstrated initial improvement in 2022 as compared to 2021; the portion of men with the normal HbA1c range increased markedly (+13.7%, from 509 of 1241 (41.0%) in 2021 to 1093 of 2346 (46.6%) in 2022), while both pre-diabetic and diabetic proportions decreased. However, in 2023, annual percentage changes of prediabetic HbA1c levels was 10.8% with additional growth in 2024 by 28.8%, ending at 35.8% (7486 of 20,925 cases) prediabetic prevalence being 25.6% higher than in 2021 (354 of 1241 cases. 28.5%). Diabetic HbA1c prevalence declined during 2022–2023, but climbed again in 2024 (+12.6%; 6350 of 20,925; 30.3%), returning almost to its 2021 level (378 of 1241; 30.5%). Overall, from 2021 to 2024, men demonstrated a 25.6% rise in prediabetic HbA1c, accompanied by a 17.3% reduction in normal HbA1c values, with no significant net changes in the diabetic category (Table 2).
Table 2.
Sex specific differences in prevalence and annual percentage changes of HbA1c of various ranges.
| Years | Normal HbA1c range (<5.7%) |
Pre-diabetic HbA1c range (5.7–6.4%) |
Diabetic HbA1c range (>6.4%) |
|||
|---|---|---|---|---|---|---|
| Proportion (%) | APC (%) | Proportion (%) | APC (%) | Proportion (%) | APC (%) | |
| The whole population | ||||||
| 2021 | 49 | 28.7 | 22.3 | |||
| 2022 | 52.3 | +6.73 | 25.3 | −11.85 | 22.4 | +0.45 |
| 2023 | 47.9 | −8.41 | 29.4 | +16.21 | 22.7 | +1.34 |
| 2024 | 35.9 | −25.05 | 37.7 | +28.23 | 26.4 | +16.30 |
| AAPC (2021–2024) | −26.7 | +31.4 | +18.4 | |||
| Women | ||||||
| 2021 | 53.1 | 28.7 | 18.2 | |||
| 2022 | 55.5 | +4.5 | 25.3 | −11.8 | 19.2 | +5.5 |
| 2023 | 49.4 | −11.0 | 30.2 | +19.4 | 20.4 | +6.3 |
| 2024 | 37.1 | −24.9 | 38.8 | +28.5 | 24.1 | +18.1 |
| AAPC (2021–2024) | −30.1 | +35.2 | +32.4 | |||
| Men | ||||||
| 2021 | 41 | 28.5 | 30.5 | |||
| 2022 | 46.6 | +13.7 | 25.1 | −11.9 | 28.3 | −7.2 |
| 2023 | 45.3 | −2.8 | 27.8 | +10.8 | 26.9 | −4.9 |
| 2024 | 33.9 | −25.2 | 35.8 | +28.8 | 30.3 | +12.6 |
| AAPC (2021–2024) | −17.3 | +25.6 | −0.7 | |||
Notes: APC, annual percentage change; AAPC, average annual percentage change.
Definitions: HbA1c–glycated haemoglobin.
The age distribution of the tested sample remained broadly consistent throughout the study period (Table S5). The largest proportion of the cases was represented by individuals aged 66–75 years (Age group 3)–their number was about ≈40–42% each year. Middle-aged adults (46–65 years; Age group 2) accounted for about one-third of observations, although their proportion gradually declined from 34.3% (1257 of 3670 cases) in 2021 to 27.5% (16,142 of 58,645) in 2024. The oldest age group (≥75 years; Age group 4) increased slightly over time, reaching 26.8% (15,709 of 58,645 cases) in 2024, while young adults (18–45 years; Age group 1) consistently represented a small and stable share of the tested cases (nearly 4%, including 160 of 3670 cases (4.4%) in 2021 and 2435 of 58,645 cases (4.2%) in 2024).
All age groups shifted from normal toward prediabetes by 2024, with larger relative increases in young (18–45) and middle-aged (46–65) adults than in older groups (Table 3). Assessing age differences in HbA1c dynamics, we found that in 2021, 90% (144 of 160) of young adults had normal HbA1c levels, while 4.4% (7 of 160) and 5.6% (9 of 160) had prediabetic and diabetic values, respectively. Among middle-aged adults, 77.7% (977 of 1257) were within the normal range, 14% (176 of 1257) were prediabetic, and 8.3% (104 of 1257) had diabetic HbA1c levels. In contrast, older and oldest groups showed significantly higher prevalence of prediabetes (37% (547 of 1479) and 41.6% (322 of 774) respectively) and diabetes (28.6% (423 of 1479) and 36.7% (284 of 774), respectively).
Table 3.
Age differences in prevalence and annual percentage changes of HbA1c of various ranges.
| Years | Normal HbA1c range (<5.7%) |
Pre-diabetic HbA1c range (5.7–6.4%) |
Diabetic HbA1c range (>6.4%) |
|||
|---|---|---|---|---|---|---|
| Proportion (%) | APC (%) | Proportion (%) | APC (%) | Proportion (%) | APC (%) | |
| Age group 1 young adults (18–45 years) | ||||||
| 2021 | 90 | 4.4 | 5.6 | |||
| 2022 | 90.3 | +0.33 | 4.1 | −6.8 | 5.6 | 0.0 |
| 2023 | 86.5 | −4.21 | 6.3 | +53.7 | 7.1 | +26.8 |
| 2024 | 80.5 | −6.94 | 12 | +90.5 | 7.5 | +5.6 |
| AAPC (2021–2024) | −10.56 | +172.7 | +33.9 | |||
| Age group 2 middle-aged adults (46–65 years) | ||||||
| 2021 | 77.7 | 14 | 8.3 | |||
| 2022 | 78.4 | +0.9 | 11.8 | −15.7 | 9.8 | +18,1 |
| 2023 | 74.1 | −5.5 | 16.6 | +40.7 | 9.3 | −5.1 |
| 2024 | 64.4 | −13.1 | 25.6 | +54.2 | 10 | 7.5 |
| AAPC (2021–2024) | −17.1 | +82.9 | +20.5 | |||
| Age group 3 older age (66–75 years) | ||||||
| 2021 | 34.4 | 37 | 28.6 | |||
| 2022 | 40.6 | +18.0 | 31.3 | −15.4 | 28.1 | −1.7 |
| 2023 | 36.9 | −9.1 | 35.8 | +14.4 | 27.3 | −2.8 |
| 2024 | 26.5 | −28.2 | 43.6 | +21.8 | 29.9 | +9.5 |
| AAPC (2021–2024) | −23.0 | +17.8 | +4.5 | |||
| Age group 4 oldest age (75+ years) | ||||||
| 2021 | 21.7 | 41.6 | 36.7 | |||
| 2022 | 29.2 | +34.6 | 37.6 | −9.6 | 33.1 | −9.8 |
| 2023 | 23.9 | −18.2 | 39.6 | +5.3 | 36.5 | +10.3 |
| 2024 | 14.5 | −39.3 | 44.9 | +13.4 | 40.6 | +11.2 |
| AAPC (2021–2024) | −33.2 | +7.9 | +10.6 | |||
Notes: APC, annual percentage change; AAPC, average annual percentage change.
Definitions: HbA1c–glycated haemoglobin.
Every age group experienced a decline in normal HbA1c and an increase in prediabetes prevalence by 2024. Although younger people started from a healthier baseline, they showed faster relative deterioration, indicating greater sensitivity to war-related stress and a higher risk of diabetes (Table 3).
People of different ages responded differently to war-related factors, with younger individuals showing the greatest vulnerability. While the older and oldest groups exhibited only a modest increase in the rate of prediabetic HbA1c (by 18% and 12.5% respectively), the prevalence among young and middle-aged adults rose sharply by more than 2.5-fold (from 4.4% (7 of 160) in 2021 to 12% (291 of 2435) in 2024 in Age group 1) and about 2-fold (from 176 of 1257 (14.0%) in 2021 to 4129 of 16,142 (25.6%) in 2024 in Age group2), respectively. Among adults aged 66–75 years, glycaemic control briefly improved in 2022 (number of cases of prediabetic levels dropped from 547 of 1479 (37%) in 2022 to 833 of 2665 (31.3%) in 2022), but by 2023 and 2024 prediabetes prevalence increased by 14.4% (to 2565 of 7158; 35.8%) and 21.8% (10,626 of 24,359), reaching 43.6% in this group. while diabetes prevalence remained largely unchanged. In contrast, individuals over 75 years experienced a continuous decline in the proportion with normal HbA1c, leaving only 14.5% (2272 of 15,709) in the normal range by 2024, alongside increases of 9.7% (from 322 of 774 (41.6%) in 2021–7060 of 15,709 (44.9%) in 2024) in prediabetes and 10.6% (from 284 of 774 (36.7%) in 2021 to 6377 of 15,709 (40.6%) in 2024) in diabetes. Overall, young and middle-aged adults were the most prone to war-related metabolic shift, though the impact of war deepened the glycaemic disturbances in older people.
Discussion
The findings of this study contribute to the growing body of evidence linking prolonged societal stress and healthcare disruption with worsening metabolic health.
Our analysis of routine laboratory data from Ukraine (2021–2024) showed that glycaemia shifted upward during humanitarian shocks. Median HbA1c increased due to the increased proportion of people with HbA1c in the 5.7–6.4% and ≥6.5% ranges. The observed increase in HbA1c values over time may reflect several mechanisms. These include progression from normoglycemia or prediabetes to diabetes, deterioration of glycaemic control among individuals with previously diagnosed diabetes, or disruptions in access to diabetes care and medications during the wartime period. Because treatment and diagnosis data were not available, the relative contribution of these mechanisms cannot be determined. Younger and middle-aged adults exhibited greater relative shifts from normal to prediabetes than older groups. Regional medians increased in all macro-regions. Higher HbA1c levels observed in the eastern and central regions may reflect the greater intensity of military activity and associated stressors in these areas during the study period.
The discovered metabolic costs of the ongoing war seem to be substantial for public-health system. Our findings are consistent with evidence from other conflict- and displacement-affected populations, demonstrating that disruptions in continuity of care and limited access to essential medications are associated with higher risks of diabetes, hypertension and other non-communicable diseases.12,13 In addition, chronic exposure to psychological stress in conflict environments has been shown to adversely affect metabolic control through both physiological and behavioural pathways.14 These patterns occur within a broader systemic context in which non-communicable diseases remain under-recognised in emergency preparedness and response efforts. Recent analysis highlighted the need to integrate non-communicable disease care, including diabetes management, into emergency health systems through accessible primary and community-based services, with respect to crisis-adaptable care models.15
Continuous increases in HbA1c levels and an increased proportion of people with prediabetic (5.7–6.4) and diabetic (<6.5%) HbA1c levels suggests an elevated proportion of people at risk cardiovascular disease.16 Previous studies showed that people with HbA1c level of 6.0% or higher were at elevated risk for the development of diabetes and cardiovascular complications.17 Increases in Hb1Ac levels are associated with increased microvascular risk.14,15 In contrast, a 1% decline in Hb1Ac has been shown to significantly reduce the risk of macro- and microvascular complications, as well as deaths related to diabetes.18
In the second part of study, we focused on HbA1c levels as an indicator of glycaemic control. In contrast to highly variable glucose levels, HbA1c values reflect the long-term changes in carbohydrates metabolism with a relatively low intraindividual variability.19 Sex-stratified analyses showed higher absolute HbA1c in men each year, but larger relative deterioration in women. Our data revealed a dual pattern: men demonstrated higher baseline HbA1c in 2021, but women deteriorated more significantly once the war began in 2022. During 2021 the HbA1c median in men reached the prediabetic threshold (reaching 5.8%), whereas women demonstrated lower values of HbA1c around 5.6%.20 From 2021 to 2024, the glycaemic trajectory diverged sharply—while women showed a significant increase of prediabetic and diabetic HbA1c levels prevalence, men exhibited a modest rise in pre-diabetes with no significant changes in diabetic HbA1c values prevalence. This suggests greater vulnerability of women to war-related metabolic stressors. Such divergence is consistent with reports that women experience higher subjective stress reactivity as compared to men. Besides, women possess lower mass of skeletal muscles but more adipose tissue as compared to men of the same age, so they typically have higher levels of free fatty acids and intracellular content of lipids which can affect insulin resistance.21 In addition, hormonal factors, particularly peri-menopausal oestrogen decline could be important mechanism of stress-induced insulin resistance diabetogenic effect, accelerating glycaemic drift in women aged 45 years and older.22
The social and cultural context of the wartime may also contribute to the observed sex differences. Women are more involved in primary caregiving responsibilities and household, while managing displacement, family separation and economic uncertainty. These cumulative stressors together with the disruptions in healthcare access and daily routines could contribute to deterioration in metabolic control, causing more prominent shift in cortisol-mediated signalling and insulin resistance. Next, pre-existing sex gaps in healthcare utilization, typical for Eastern Europe, might be amplified under wartime resource scarcity, delaying detection or optimization of glycaemic control until more advanced stages. Finally mental health disorders provoked by grief of loss, displacement and uncertainty due to the war could facilitate diabetes development especially in women. It was indicated that depression and anxiety increases T2D risk and correlates positively with fasting glucose, insulin, and HbA1c.23
Using the GBD 2021 benchmark of 915 DALYs per 100,000 for diabetes, the detected 4.1% increase of diabetes prevalence would raise Ukraine's diabetes DALY rate by about 156 DALYs per 100,000. Applied to approximately 30 million adults in Ukraine, this could count nearly 47,000 additional DALYs annually.3 Under such conditions, years lived with disability (YLD) could dominate, given that microvascular sequelae are reported to account for ∼52% of total diabetes DALYs.3 Considering the cumulative effects of stress and diabetes complications, years of life lost (YLL) could rise in Ukraine due to increased cardiovascular mortality.3
These preliminary estimates underscore the urgency of developing and deploying scalable diabetes-prevention and control programs in Ukraine, especially in conflict-affected regions.
Strengths include a large sample size, multi-year observation period, and use of consistent standardised laboratory methods within an ISO15189-certified laboratory network. These features define high analytical reliability and allow for robust assessment of temporal trends.
Several limitations should be considered. First, the study is based on a large-scale convenience sample of individuals who underwent laboratory testing for clinical reasons (check-up, diagnostic evaluation, or treatment monitoring), rather than a probabilistic population-based sample. The data from only one of several large private laboratories were analysed in the study. As such, the dataset reflects real-world testing patterns within a nationwide laboratory network rather than a randomly selected representative cohort of the Ukrainian population. Consequently, the findings may be affected by selection bias, as individuals seeking testing may differ from the general population, particularly during wartime when healthcare access, displacement, and health-seeking behaviour are substantially altered.
Second, the number of observations differed markedly across study years (HbA1c: 3670 in 2021, 6587 in 2022, 17,175 in 2023, and 58,645 in 2024), which may introduce reporting bias. Temporal changes may therefore partially reflect shifts in the composition of the tested population rather than true changes in underlying glycaemic status.
Third, the use of anonymised laboratory data excluded identification of individuals across time. As a result, duplicate observations could not be identified, and some individuals may have contributed multiple test results. Consequently, repeated measurements could not be distinguished from independent observations, and observed temporal trends may partly reflect individual changes rather than population-level shifts. Besides, the dataset lacked clinical and behavioural information, including prior diabetes diagnosis, medication use, treatment adherence, body mass index, lifestyle factors, and socioeconomic status. Therefore, it was not possible to differentiate between incident dysglycemia and worsening control among individuals with established diabetes, and to adjust for potential confounding factors. The findings should thus be interpreted as descriptive associations rather than causal relationships.
In addition, geographic analyses might be influenced by uneven regional coverage and the inability to account for internal displacement during the war. Migration and shifting healthcare access may have altered the regional composition of the tested population, potentially biasing comparisons across regions. Finally, insulin and cortisol measurements were available only in smaller, clinically selected subgroups, introducing potential indication bias. In particular, cortisol levels may reflect diagnostic evaluation rather than baseline physiological stress, limiting their interpretation as markers of population-level stress responses.
These constraints likely bias estimates toward conservative or context-specific effects and justify cautious interpretation. However, several considerations mitigate these concerns. The analysis included more than 100,000 observations over four consecutive years, providing substantial statistical power and chronological stability. All measurements were performed within the same ISO15189 certified laboratory network using standardised analytical platforms throughout the study period. Therefore, observed dynamics of the data are unlikely to be explained by assay variability. Moreover, the same eligibility criteria were applied across years of the observation. This supports the validity of temporal comparisons within the dataset, even if the sample is not fully population representative. Finally, subgroup analyses were performed to account for geographic heterogeneity and demographic structure, which reduces but does not fully remove the impact of differential regional access or war-related displacement.
In this large, laboratory-based analysis covering all regions of Ukraine from 2021 to 2024, we observed consistent upward shifts in glycaemic biomarkers within a broad testing sample during the war. The proportion of individuals with HbA1c in the prediabetic and diabetic ranges increased over time with the highest rate at eastern part of Ukraine exposing the highest military burden. Women and younger adults, despite lower baseline levels of HbA1c, showed the sharpest deterioration, highlighting vulnerable groups at long-term risk of diabetes and cardiovascular complications. The observed patterns underscore the potential metabolic impact of a prolonged humanitarian crisis.
Continued monitoring of glycaemic biomarkers, implementation of targeted screening and prevention strategies might be helpful in mitigating the emerging metabolic burden of humanitarian emergencies in Ukraine.
Contributors
OS and NK conceived the study, developed the protocol, led the data approvals, and wrote the first draft of the manuscript. VY, YI and VG performed the statistical analysis. VY, YI, AS, PBG, DS, LP, MT, OL, VP and VPop helped with data interpretation and provided critical feedback throughout the project to shape the analysis and the manuscript. OS and NK provided study leadership. OS, NK and VG assessed and verified data. All authors were permitted full access to the study data, approved the final submission, and accepted responsibility for the decision to submit for publication.
Data sharing statement
Deidentified individual participant data supporting the findings of this study are available upon request from the corresponding author. The data is not publicly available due to privacy or ethical restrictions.
Editor note
The Lancet Group takes a neutral position with respect to territorial claims in published maps and institutional affiliations.
Declaration of interests
All authors declare no competing interests.
Acknowledgements
We would like to thank to Medical Laboratory CSD (Kyiv, Ukraine) for supporting the study. This research was funded by National Research Foundation of Ukraine (project ‘Study of the course and consequences of COVID-19 in patients with diabetes and the impact of SARS-COV-2 infection on the rate of biological aging"‘, number 2021.01/0213). This organization was not involved in the idea, writing, editing of, or decision to submit this manuscript. The authors thank the National Research Foundation of Ukraine (NRFU) for their continued support. Preparation of this article was supported by the Giving Voice initiative.
During the preparation of this work the author(s) used ChatGPT in order to support a comprehensive literature search and to improve the English language during the editing stage. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the published article.
Footnotes
Supplementary data related to this article can be found at https://doi.org/10.1016/j.eclinm.2026.104077.
Appendix A. Supplementary data
References
- 1.Boyle J.P., Thompson T.J., Gregg E.W., Barker L.E., Williamson D.F. Projection of the year 2050 burden of diabetes in the US adult population: dynamic modeling of incidence, mortality, and prediabetes prevalence. Popul Health Metr. 2010;8:29. doi: 10.1186/1478-7954-8-29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Genitsaridi I., Salpea P., Salim A., et al. 11th edition of the IDF Diabetes Atlas: global, regional, and national diabetes prevalence estimates for 2024 and projections for 2050. Lancet Diabetes Endocrinol. 2026;14:149–156. doi: 10.1016/S2213-8587(25)00299-2. [DOI] [PubMed] [Google Scholar]
- 3.Ong K.L., Stafford L.K., McLaughlin S.A., et al. Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050: a systematic analysis for the Global Burden of Disease Study 2021. Lancet. 2023;402:203–234. doi: 10.1016/S0140-6736(23)01301-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Watkins D.A., Ali M.K. Measuring the global burden of diabetes: implications for health policy, practice, and research. Lancet. 2023;402:163–165. doi: 10.1016/S0140-6736(23)01287-4. [DOI] [PubMed] [Google Scholar]
- 5.Sulaieva O., Dudin O., Botsun P., et al. The detrimental consequences of two consecutive disasters impacting cytopathology in Ukraine: COVID followed by the war. Cytopathology. 2023;34:442–449. doi: 10.1111/cyt.13236. [DOI] [PubMed] [Google Scholar]
- 6.Martsenkovskyi D., Shevlin M., Ben-Ezra M., et al. Mental health in Ukraine in 2023. Eur Psychiatry. 2024;67:e27. doi: 10.1192/j.eurpsy.2024.12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Poberezhets V. Healthcare crisis in Ukraine - worrying consequences of the Russian-Ukrainian war. Croat Med J. 2022;63:315–316. doi: 10.3325/cmj.2022.63.315. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sulaieva O., Yerokhovych V., Zemskov S., et al. The impact of war on people with type 2 diabetes in Ukraine: a survey study. eClinicalMedicine. 2025;79 doi: 10.1016/j.eclinm.2024.103008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Alessi J., Yankiv M. War in Ukraine and barriers to diabetes care. Lancet. 2022;399:1465–1466. doi: 10.1016/S0140-6736(22)00480-9. [DOI] [PubMed] [Google Scholar]
- 10.McEwen B.S., Wingfield J.C. The concept of allostasis in biology and biomedicine. Horm Behav. 2003;43:2–15. doi: 10.1016/s0018-506x(02)00024-7. [DOI] [PubMed] [Google Scholar]
- 11.Wentworth B.A., Stein M.B., Redwine L.S., et al. Post-traumatic stress disorder: a fast track to premature cardiovascular disease? Cardiol Rev. 2013;21:16–22. doi: 10.1097/CRD.0b013e318265343b. [DOI] [PubMed] [Google Scholar]
- 12.Ansbro É., Issa R., Willis R., Blanchet K., Perel P., Roberts B. Chronic NCD care in crises: a qualitative study of global experts' perspectives on models of care for hypertension and diabetes in humanitarian settings. J Migr Health. 2022;5 doi: 10.1016/j.jmh.2022.100094. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kehlenbrink S., Smith J., Ansbro É., et al. The burden of diabetes and use of diabetes care in humanitarian crises in low-income and middle-income countries. Lancet Diabetes Endocrinol. 2019;7:638–647. doi: 10.1016/S2213-8587(19)30082-8. [DOI] [PubMed] [Google Scholar]
- 14.Aebischer Perone S., Martinez E., Du Mortier S., et al. Non-communicable diseases in humanitarian settings: ten essential questions. Confl Health. 2017;11:17. doi: 10.1186/s13031-017-0119-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ansbro É., Cooper K., Tharwat M., et al. Non-communicable diseases in emergencies: time for action on converging crises. Lancet. 2025;406:1197–1200. doi: 10.1016/S0140-6736(25)01852-5. [DOI] [PubMed] [Google Scholar]
- 16.Wild S.H., Dunn C.J., McKeigue P.M., Comte S. Glycaemic control and cardiovascular disease in type 2 diabetes: a review. Diabetes Metab Res Rev. 1999;15:197–204. doi: 10.1002/(sici)1520-7560(199905/06)15:3<197::aid-dmrr28>3.0.co;2-f. [DOI] [PubMed] [Google Scholar]
- 17.Mitsios J.P., Ekinci E.I., Mitsios G.P., Churilov L., Thijs V. Relationship between glycated hemoglobin and stroke risk: a systematic review and meta-analysis. J Am Heart Assoc. 2018;7 doi: 10.1161/JAHA.117.007858. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Stratton I.M., Adler A.I., Neil H.A.W., et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ. 2000;321:405–412. doi: 10.1136/bmj.321.7258.405. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Rohlfing C., Wiedmeyer H.M., Little R., et al. Biological variation of glycohemoglobin. Clin Chem. 2002;48:1116–1118. [PubMed] [Google Scholar]
- 20.Kautzky-Willer A. Does diabetes mellitus mitigate the gender gap in COVID-19 mortality? Eur J Endocrinol. 2021;185:C13–C17. doi: 10.1530/EJE-21-0721. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Mauvais-Jarvis F. Gender differences in glucose homeostasis and diabetes. Physiol Behav. 2018;187:20–23. doi: 10.1016/j.physbeh.2017.08.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Mauvais-Jarvis F., Manson J.E., Stevenson J.C., Fonseca V.A. Menopausal hormone therapy and type 2 diabetes prevention: evidence, mechanisms, and clinical implications. Endocr Rev. 2017;38:173–188. doi: 10.1210/er.2016-1146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Rotella F., Mannucci E. Depression as a risk factor for diabetes: a meta-analysis of longitudinal studies. J Clin Psychiatry. 2013;74:31–37. doi: 10.4088/JCP.12r07922. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.