Risk of progression to type 2 diabetes or regression to normoglycaemia among Ghanaians with prediabetes living in Ghana and the Netherlands: The RODAM prospective study Authors: Emmanuel Bannerman‐Williams, Yacoba Atiase, Bert‐Jan H. van den Born, Felix Patience Chilunga, Eva van der Linden, Karlijn A. C. Meeks, Erik Beune, Charles Agyemang, Charles Frederick Hayfron‐Benjamin Published: 2026 ABSTRACT AbstractIntroductionEvidence shows that ethnicity affects the risk of progression of prediabetes to type 2 diabetes(T2D) or regression to normoglycaemia However, little is known about progression rates among sub‐Saharan African populations or whether geographical context affects risk variations. We compared the risk of progression to T2D or regression to normoglycemia in West Africans with prediabetes living in Africa and their migrant compatriots living in Europe.MethodsWe analysed data from 615 Ghanaians with prediabetes at baseline (245 in Ghana/370 in Amsterdam) in the prospective RODAM cohort with an average follow‐up of 6.7 years. Prediabetes was defined according to the WHO criteria. Incidence risk ratios for progression to T2D and regression to normoglycaemia were calculated with adjustment for age, sex, follow‐up duration, socioeconomic status and baseline BMI and fasting plasma glucose.ResultsOverall, the cumulative incidences of T2D and normoglycaemia from prediabetes were 12.6% and 28.9%, respectively. While the proportions of individuals who progressed to T2D did not differ between non‐migrants and migrants (16.0% vs. 10.9%, p‐value = 0.108), the proportion who regressed to normoglycaemia was higher in non‐migrants (51.5% vs. 17.1%, p < 0.001) with an incident risk ratio of 3.48 (95% CI 2.39–5.05, p < 0.001) in the fully adjusted model. However, the risk of progression of prediabetes to T2D remained similar in the two groups after full adjustment [1.20 (0.71–2.05), 0.498].ConclusionsAmong Ghanaians with prediabetes, regression to normoglycaemia was more frequent than progression to T2D, and more likely in non‐migrants than in migrants. Future studies could identify factors underlying the higher rates of regression to normoglycaemia in non‐migrants to improve outcomes in individuals with prediabetes. What's new Africans with prediabetes are more likely to regress to normoglycemia than to progress to type 2 diabetes.African migrants and non‐migrants with prediabetes have a similar risk of progressing to type 2 diabetes.However, African non‐migrants are more likely to regress to normoglycemia than migrants. What's new Africans with prediabetes are more likely to regress to normoglycemia than to progress to type 2 diabetes.African migrants and non‐migrants with prediabetes have a similar risk of progressing to type 2 diabetes.However, African non‐migrants are more likely to regress to normoglycemia than migrants. Africans with prediabetes are more likely to regress to normoglycemia than to progress to type 2 diabetes. Africans with prediabetes are more likely to regress to normoglycemia than to progress to type 2 diabetes. African migrants and non‐migrants with prediabetes have a similar risk of progressing to type 2 diabetes. African migrants and non‐migrants with prediabetes have a similar risk of progressing to type 2 diabetes. However, African non‐migrants are more likely to regress to normoglycemia than migrants. However, African non‐migrants are more likely to regress to normoglycemia than migrants. INTRODUCTION Prediabetes or intermediate hyperglycaemia is associated with an increased risk of type 2 diabetes (T2D), 1 cardiovascular disease, chronic kidney disease and a higher risk of death. 2 , 3 Prediabetes is believed to capture early‐stage glucose dysregulation and is determined using measures of glycaemic control like glycosylated haemoglobin (HbA1c) and fasting plasma glucose (FPG) at thresholds below those used in defining diabetes. 4 According to the WHO, prediabetes affects 7.3% of the world's adult population; this is projected to increase to 8.3% by 2045. 5 Prediabetes can be used to identify individuals at risk for developing T2D. However, prediabetes can also regress to normoglycaemia, which is associated with a better prognosis, including a reduction in the future risk of cardiovascular disease and all‐cause mortality. 6 , 7 While most studies have reported identical rates of progression to diabetes and conversion to normoglycaemia in individuals with prediabetes, 8 there is evidence that conversion to normoglycaemia may be more frequent. 9 The rates of progression and regression have been found to differ across populations. For example, South Asian and mixed‐race individuals are known to have an increased risk of progression from prediabetes to T2D, compared with those of European ethnic origin living in the same geographical setting. 10 , 11 However, data on changes in glycaemic status in sub‐Saharan Africans with prediabetes are limited, even though the burden of prediabetes in sub‐Saharan Africans may be high 12 and sub‐Saharan Africans have the highest projected incidence of T2D. 13 In addition to ethnic differences in progression rates, evidence also suggests that environmental factors may influence these rates. 14 This suggests that the progression of prediabetes to T2D may vary depending on context. Including geographical and lifestyle‐related factors, such as diet and physical activity. 15 , 16 , 17 Leveraging data from the prospective RODAM study, we assessed the risk of progression to T2D or regression to normoglycaemia in West Africans with prediabetes living in West Africa and their migrant compatriots living in Europe. Based on a higher exposure to obesogenic environments, it would be expected that the risk of progression to T2D is higher and the risk of reversal to normoglycaemia is lower among migrants compared to non‐migrants. METHODS Study design The Research on Obesity and Diabetes among African Migrants Prospective (RODAM‐Pros) cohort study design, recruitment and data collection are described in a separate publication. 18 All the participants in the RODAM study that completed the baseline assessment (2012–2015) were eligible for the follow‐up, of which 2165 participants (n = 1246 living in Ghana and n = 919 Ghanaian migrants in Amsterdam) were included in the RODAM‐Pros cohort study (2019–2021). For the current analyses, 615 adults with prediabetes based on either the WHO or ADA criteria at the baseline evaluation and with data on follow‐up visits were included (Figure 1). This included data on 245 Ghanaians living in Ghana and 370 Ghanaians living in Amsterdam. Flow chart of study design and inclusion in analysis. FPG, Fasting Plasma Glucose; HbA1c, Glycated Haemoglobin. Assessments The assessment of sociodemographic, health‐related behaviours, clinical characteristics and physical measurements, including body mass index (BMI) and blood pressure at baseline and follow‐up visits has been described in detail elsewhere. 18 , 19 Educational status was adapted to the local context at each study site and classified into four categories: (1) no formal education or elementary school; (2) lower vocational or lower secondary education; (3) intermediate vocational or intermediate/higher secondary education; and (4) higher vocational education or university. Smoking status was defined by a positive response to the question, ‘Do you smoke at all?’ Physical activity was assessed using the International Physical Activity Questionnaire, and participants were categorized into three levels of total physical activity: low, moderate or high. 20 Physical examinations were conducted at all study sites using validated devices and standardized operating procedures. Weight was measured to the nearest 0.1 kg using a SECA 877 scale, with participants wearing light clothing and no shoes. Height was measured to the nearest 0.1 cm using a portable stadiometer (SECA 217), without shoes. BMI was calculated as weight in kilograms divided by height in meters squared (kg/m2). Waist circumference was measured in centimetres at the midpoint between the lower margin of the least palpable rib and the top of the iliac crest. All anthropometric measurements were taken twice by the same assessor, and the average of the two readings was used for analysis. Blood pressure was measured three times after participants had rested for at least 5 min in a seated position, using a validated semi‐automated device (Microlife WatchBP Home, Widnau, Switzerland) with appropriately sized cuffs. The mean of the last two measurements was used for analysis. Hypertension was defined as systolic blood pressure ≥140 mmHg, and/or diastolic blood pressure ≥90 mmHg, and/or current use of antihypertensive medication. Sample collection and biochemical analyses Standardized methods were implemented to ensure consistency in data collection across all sites and over time. At every site, skilled research assistants were employed to collect venous blood samples from subjects who had fasted for at least 10 h. After collection (within 1–3 h of the venipuncture), all samples were separated into aliquots and processed. They were then temporarily stored at 4°C–7°C at the local research location before being stored at −80°C. Following their transportation to the laboratories of the nearby research centres, the separated samples were examined, registered and kept at −80°C until being sent to their ultimate analysis location. Serum, EDTA whole blood and heparin plasma were among these samples. Fasting glucose and cholesterol concentrations were determined using the ABX Pentra 400 Chemistry Analyzer (HORIBA ABX, Montpellier, France). Definition of prediabetes and T2D Prediabetes was defined according to criteria from the World Health Organization (WHO) and the American Diabetes Association (ADA). The WHO definition is based on FPG concentration of 6.1–6.9 mmol/L and/or HbA1c of 42–47 mmol/mol (6.0%–6.4%) in individuals without diabetes by either HbA1c or FPG and not receiving glucose‐lowering therapy. 21 The ADA uses lower thresholds, defining prediabetes as an FPG concentration of 5.6–6.9 mmol/L and/or HbA1c of 39–47 mmol/mol (5.7–6.4%) in individuals without diabetes and not on glucose‐lowering therapy. 22 Individuals meeting either the WHO or ADA criteria were eligible for inclusion. For the primary analyses, prediabetes was defined using the WHO criteria; supplementary analyses applied ADA criteria. Participants were selected on both HbA1c and/or FPG criteria. T2D at follow‐up was defined as the use of glucose‐lowering medication, HbA1c level ≥47 mmol/mol (≥6.5%), or FPG concentration ≥7 mmol/L at follow‐up. Patient and public involvement Community leaders played a central role in the design, recruitment, and dissemination phases of the study. The leaders were identified through local organizations such as churches and supported by key community figures. Statistical analysis Baseline characteristics were summarised as means with standard deviations (SD) for normally distributed continuous variables and frequencies with percentages for categorical variables. The distribution of continuous variables was assessed using graphical methods (histograms and Q–Q plots) in addition to formal normality testing. Given the moderate sample size, visual inspection was performed in conjunction with the Kolmogorov–Smirnov test to evaluate normality. Changes in HbA1c and FPG between baseline and follow‐up were evaluated using paired methods to account for within‐person correlation. Because glycaemic measures were available at only two time points, more complex longitudinal modelling approaches were not considered necessary. Where relevant, regression models adjusting for baseline values were applied to reduce bias and improve precision. Incidence rates of progression to T2D and regression to normoglycaemia were calculated using person‐time at risk. Incidence risk ratios (IRRs) and corresponding 95% confidence intervals (CIs) were estimated using Poisson regression. Follow‐up duration was included as an offset term to account for variation in person‐time across participants. Four models were specified: Model 1 was unadjusted. Model 2 was adjusted for age, sex, and follow‐up duration; Model 3 was adjusted for age, sex, follow‐up duration, educational level and baseline BMI; and Model 4 was adjusted for age, sex, follow‐up duration, educational level, baseline BMI and baseline FPG. A statistical test of significance was set at a p‐value < 0.05. Data were analysed using the IBM SPSS version 26 for Windows. RESULTS Baseline characteristics Table 1 summarizes the baseline characteristics of the study population. Migrants were similar in age but more frequently males, smokers and highly educated compared with the non‐migrant group. The average BMI and waist circumference of migrants were higher than those of non‐migrants. The mean HbA1c, systolic BP, diastolic BP, and the proportion of individuals with hypertension at baseline were higher in migrants than in non‐migrants. Conversely, mean FPG and mean LDL‐cholesterol were higher in non‐migrants than in migrants. At follow‐up, glucose‐lowering therapy use was more frequent in non‐migrants than in migrants. Participants lost to follow‐up were more often male, older, highly educated and smokers, with higher baseline fasting plasma glucose and cholesterol, but lower blood pressure and BMI, compared with those retained. Both groups were otherwise similar in physical activity, waist circumference and HbA1c levels (Table S1). TABLE 1Baseline characteristics of study participants.All participantsNon‐migrantsMigrants p value N 615245370Follow‐up time, in months80.19 (±8.50)79.37 (±4.62)80.74 (±10.25)0.051Sex0.001Male251 (40.8%)80 (32.7%)171 (46.2%)Female364 (59.2%)165 (67.3%)199 (53.8%)Age (years)48.36 (± 8.842)48.56 (±9.54)48.24 (±8.36)0.657Highest education0.002Secondary education and higher143 (24.5%)42 (17.8%)101 (29.1%)Below secondary education440 (75.5%)194 (82.2%)246 (70.9%)Physical activity0.112Low level85 (21.7%)59 (25.2%)26 (16.6%)Moderate level66 (16.9%)36 (15.4%)30 (19.1%)High level240 (61.4%)139 (59.4%)101 (64.3%)Smoking status (%)0.009Never smoked557 (96.9%)234 (99.2%)323 (95.3%)Current or previous smoker18 (3.1%)2 (0.8%)16 (4.7%)BMI, kg/m2 28.05 (± 4.71)27.02 (±5.44)28.73 (±4.03)<0.001Waist circumference, cm93.16 (±11.03)90.70 (±11.51)94.78 (±10.39)<0.001Diagnosis of hypertension321 (52.2%)96 (39.2%)225 (60.8%)<0.001Systolic BP, mmHg132.70 (±18.45)127.40 (±18.93)136.23 (±17.27)<0.001Diastolic BP, mmHg83.05 (±11.11)80.14 (±11.25)84.99 (±10.59)<0.001Statin therapy (%)19 (3.1%)0 (0.0%)19 (5.1%)<0.001FPG, mmol/L5.34 (±1.01)5.52 (± 0.57)5.23 (±1.21)0.001HbA1c, mmol/mol37.95 (± 9.53)34.91 (±12.34)39.96 (±6.36)<0.001Total cholesterol, mmol/L5.09 (±1.11)5.16 (±1.22)5.04 (± 1.02)0.199HDL‐cholesterol, mmol/L1.33 (± 0.50)1.22 (± 0.32)1.40 (± 0.57)<0.001Triglyceride, mmol/L0.97 (±0.60)1.22 (± 0.59)0.80 (± 0.54)<0.001LDL‐cholesterol, mmol/L3.19 (±1.11)3.38 (±1.05)3.07 (±1.13)0.001Glucose‐lowering medication32 (74.4%)17 (6.9%)15 (4.1%)0.044 Note: Values for categorical variables are given as a number (percentage); for continuous variables, like mean (± standard deviation). Differences in categorical variables were assessed by χ 2 test or Fisher's exact test for categorical variables, t‐test for continuous variables, or Mann–Whitney U test for variables not normally distributed.Abbreviations: BMI, body mass index; CVD, cardiovascular disease; FPG, fasting plasma glucose; HbA1c, glycated haemoglobin; HDL, high‐density lipoprotein; LDL, low‐density lipoprotein; Migrants, Ghanaians resident in Amsterdam; Non‐migrants, Ghanaians resident in Ghana. Baseline characteristics of study participants. Note: Values for categorical variables are given as a number (percentage); for continuous variables, like mean (± standard deviation). Differences in categorical variables were assessed by χ 2 test or Fisher's exact test for categorical variables, t‐test for continuous variables, or Mann–Whitney U test for variables not normally distributed. Abbreviations: BMI, body mass index; CVD, cardiovascular disease; FPG, fasting plasma glucose; HbA1c, glycated haemoglobin; HDL, high‐density lipoprotein; LDL, low‐density lipoprotein; Migrants, Ghanaians resident in Amsterdam; Non‐migrants, Ghanaians resident in Ghana. Glycaemic status at follow‐up In the overall study population, the incidence of T2D from prediabetes ranged from 8.1% to 12.6%, while the incidence of normoglycaemia from individuals with prediabetes ranged from 21.0% to 77.2%, depending on the diagnostic criteria used (Figure 2a–f). This varied from non‐migrants to migrants: the incidence of T2D at 6.7 years of follow‐up in non‐migrants and migrants ranged from 9.0% to 16.0% and 7.1% to 11.1%, respectively. The incidence of normoglycaemia at 6.7 years of follow‐up in non‐migrants and migrants ranged from 30.3% to 82.2% and 14.6% to 74.5%, respectively. Glycaemic outcomes of patients with prediabetes after 6.7 years of follow‐up in migrants and non‐migrants. Glycaemic states in Figure 1a–c were based on the WHO criteria (N = 491 for all participants, N = 322 for migrants, and N = 168 for non‐migrants). Glycaemic states in Figure 1d–f were based on ADA criteria (N = 615 for all participants, N = 322 for migrants, and N = 168 for non‐migrants). WHO Criteria: (1) Definitions based on HbA1c only: Remained prediabetes if follow‐up HbA1c concentration was 42–47 mmol/mol (6.0%–6.4%) with FPG 6.9 mmol/L and not on glucose‐lowering therapy. Progressed to Diabetes if follow‐up HbA1c concentration >47 mmol/mol (>6.4%) regardless of FPG concentrations. Regressed to normoglycemia if follow‐up HbA1c concentration <42 mmol/mol (<6.0%) and FPG concentration <6.1 mmol/L and not on glucose‐lowering therapy. (2) Definitions based on FPG only: Remained prediabetes if follow‐up FPG concentration was 6.1–6.9 mmol/L with HbA1c <47 mmol/mol (<6.4%) and not on glucose‐lowering therapy. Progressed to Diabetes if follow‐up FPG concentration was ≥7 mmol/L regardless of HbA1c concentrations. Regressed to normoglycemia if follow‐up FPG concentration was <6.1 mmol/L and HbA1c concentration <42 mmol/mol (<6.0%) and not on glucose‐lowering therapy. (3) Definitions based on both FPG and HbA1c: Remained prediabetes if follow‐up FPG concentration was 6.1 to 6.9 mmol/L with HbA1c of 42–47 mmol/mol (6.0%–6.4%) and not on glucose‐lowering therapy. Progressed to diabetes if follow‐up FPG concentration was ≥7 mmol/L and/or HbA1c ≥6.5% or use of glucose‐lowering medication at follow‐up. Regressed to normoglycemia if follow‐up FPG concentration was <6.1 mmol/L and HbA1c concentration <42 mmol/mol (<6.0%) and not on glucose‐lowering therapy. ADA Criteria: (1) Definitions based on HbA1c only: Remained prediabetes if follow‐up HbA1c concentration was 39–47 mmol/mol (5.7%–6.4%) with FPG <6.9 mmol/L and not on glucose‐lowering therapy. Progressed to Diabetes if follow‐up HbA1c concentration >47 mmol/mol (>6.4%) regardless of FPG concentrations. Regressed to normoglycemia if follow‐up HbA1c concentration <39 mmol/mol (<5.7%) and FPG concentration <5.6 mmol/L and not on glucose‐lowering therapy. (2) Definitions based on FPG only: Remained prediabetes if follow‐up FPG concentration was 5.6 to 6.9 mmol/L with HbA1c <47 mmol/mol (<6.4%) and not on glucose‐lowering therapy. Progressed to Diabetes if follow‐up FPG concentration was ≥7 mmol/L regardless of HbA1c concentrations. Regressed to normoglycemia if follow‐up FPG concentration was <5.6 mmol/L and HbA1c concentration <39 mmol/mol (<5.7%) and not on glucose‐lowering therapy. (3) Definitions based on both FPG and HbA1c: Remained prediabetes if follow‐up FPG concentration was 5.6 to 6.9 mmol/L with HbA1c of 39–47 mmol/mol (5.7%–6.4%) and not on glucose‐lowering therapy. Progressed to diabetes if follow‐up FPG concentration was ≥7 mmol/L and/or HbA1c >47 mmol/mol (>6.4%) or use of glucose‐lowering medication at follow‐up. Regressed to normoglycemia if follow‐up FPG concentration was <5.6 mmol/L and HbA1c concentration <39 mmol/mol (<5.7%) and not on glucose‐lowering therapy. In Figure 2a–c, glycaemic outcomes between non‐migrants and migrants based on HbA1c, FPG, and both HbA1c and FPG cut‐off values are compared. Based on HbA1c cut‐off values only (Figure 2a), the proportion of individuals who progressed from prediabetes to T2D was similar in non‐migrants and migrants (13.5% vs. 9.0%, p‐value = 0.298). However, non‐migrants had greater percentages of those who regressed from prediabetes to normoglycaemia compared with migrants (30.3% vs. 18.4%, p‐value = 0.018). Based on FPG cut‐off values only (Figure 2b), the proportion of individuals who progressed from prediabetes to T2D was similar in non‐migrants and migrants (10.1% vs. 7.1%, p‐value = 0.298). While the proportion of individuals who regressed from prediabetes to normoglycaemia tended to be higher in non‐migrants than migrants, the difference was not statistically significant (82.2% vs. 74.5%, p‐value = 0.060). In analyses based on both HbA1c and FPG cut‐off values (Figure 2c), the proportion of individuals who progressed to T2D did not differ between non‐migrants and migrants (16.0% vs. 10.9%, p‐value = 0.108). However, the proportion of individuals who regressed to normoglycaemia was thrice as high in non‐migrants as in migrants (51.5% vs. 17.1%, p < 0.001). Similar observations were made using cut‐off values based on the ADA criteria (Figure 2d–f) except for analyses based on FPG alone, where the proportion who regressed to normoglycaemia was significantly higher in non‐migrants than in migrants (69.8% vs. 48.4%, p < 0.001). Risk of progression to T2D and regression to normoglycaemia Table 2 shows the IRR for progression to T2D and regression to normoglycaemia in non‐migrants and migrants. Based on the WHO criteria, the risk of progression from prediabetes to T2D was 47% higher in non‐migrants than in their migrant compatriots (Incidence Risk Ratio 1.47, 95% CI 0.89–2.43, p‐value = 0.133) in the unadjusted model, although the strength of association was not statistically significant. In a model adjusted for age, sex and follow‐up duration, the strengths of association were also not statistically significant [1.37 (0.82–2.29), 0.228]; similar observations were made in models that further adjusted for BMI [1.28 (0.76–2.16), 0.362] and for BMI and baseline FPG [1.20 (0.71–2.05), 0.498]. Based on the ADA criteria, the risk of progression from prediabetes to T2D did not differ between migrants and migrants. TABLE 2Incident rate ratio for progression from prediabetes to diabetes or regression from prediabetes to normoglycemia in non‐migrant West Africans relative to migrant West Africans (reference group is migrants).Incidence risk ratio (95% confidence interval), p‐valueModel 1Model 2Model 3Model 4WHO Criteria (N = 491)Progression to T2DMigrants1.00 (ref)1.00 (ref)1.00 (ref)1.00 (ref)Non‐migrants1.47 (0.89–2.43), 0.1331.37 (0.82–2.29), 0.2281.28 (0.76–2.16), 0.3621.20 (0.71–2.05), 0.498Regression to normoglycemiaMigrants1.00 (ref)1.00 (ref)1.00 (ref)1.00 (ref)Non‐migrants3.01 (2.15–4.22), <0.0013.03 (2.16–4.27), <0.0013.39 (2.35–4.89), <0.0013.48 (2.39–5.05), <0.001ADA Criteria (N = 615)Progression to T2DMigrants1.00 (ref)1.00 (ref)1.00 (ref)1.00 (ref)Non‐migrants1.18 (0.74–1.87), 0.4861.07 (0.67–1.72), 0.7751.18 (0.71–1.91), 0.5391.06 (0.65–1.75), 0.809Regression to normoglycemiaMigrants1.00 (ref)1.00 (ref)1.00 (ref)1.00 (ref)Non‐migrants3.41 (2.48–4.70), <0.0013.43 (2.48–4.73), <0.0013.35 (2.36–4.78), <0.0013.48 (2.42–5.01), <0.001 Note: Model 1 = unadjusted; Model 2 = adjusted for age, sex and follow‐up duration; Model 3 = adjusted for age, sex, follow‐up duration, education and BMI; Model 4 = adjusted for age, sex, follow‐up duration, education, BMI and baseline FPG. Incident rate ratio for progression from prediabetes to diabetes or regression from prediabetes to normoglycemia in non‐migrant West Africans relative to migrant West Africans (reference group is migrants). Note: Model 1 = unadjusted; Model 2 = adjusted for age, sex and follow‐up duration; Model 3 = adjusted for age, sex, follow‐up duration, education and BMI; Model 4 = adjusted for age, sex, follow‐up duration, education, BMI and baseline FPG. Based on the WHO criteria, the likelihood of regression from prediabetes to normoglycaemia was higher in non‐migrants in all models, including the fully adjusted model [3.48 (2.39–5.05), <0.001]. Similar observations were made when the glycaemic state was assessed based on the ADA criteria, where in the fully adjusted model, the likelihood of regression to normoglycemia was nearly 250% higher in non‐migrants [3.48 (2.42–5.01), <0.001]. DISCUSSION Key findings Using a representative sample of Ghanaian individuals, we noted that among West Africans with prediabetes, regression to normoglycaemia is more frequent than progression to T2D, despite an overall increase in HbA1c and FPG levels. Non‐migrant West Africans with prediabetes were more likely to regress to normoglycaemia compared to migrants, which could not be fully explained by differences in BMI and FPG at baseline and had a tendency towards a lower risk of progression to T2D. The combination of an elevated FPG and HbA1C was a stronger predictor for the development of T2D than FPG or HbA1C alone. Discussion of key findings During the 6.7‐year follow‐up period, only about 13% of the entire study population with prediabetes progressed to T2D regardless of the definition of prediabetes at baseline, and despite a significant overall increase in FPG and HbA1c and adequate follow‐up duration given the evidence that prediabetes typically progresses to diabetes within 5 years. 23 A systematic review and meta‐analysis by Richter et al. that included 103 prospective cohort studies in primarily middle‐aged adults reported that the 6‐year cumulative incidence of diabetes was 17% (95% CI, 14%–20%) in individuals with prediabetes based on HbA1c levels and 22% (95% CI, 15%–31%) for FPG using similar definitions as in our study. 24 The review excluded sub‐Saharan African populations, with the only African study involving South African Indians, and lacked data on high‐risk transitioning migrants. 25 This current study suggests that the risk of progression of prediabetes to T2D may be lower in West Africans compared with their counterparts of different ethnic origins of similar age. Indeed, findings from this study were rather better compared to those in older adults 26 , 27 in other populations. For example, in a community‐based cohort in the United States that included older adults with a mean age of about 76 years and 17% of African Americans, the authors reported that fewer than 12% of the study participants developed diabetes after a follow‐up period of 6.5 years. 26 The biological basis of a relatively lower proportion of West Africans with prediabetes progressing to T2D compared with studies in other ethnic groups of similar ages remains uncertain but may reflect an interaction of environmental factors such as better physical activity 28 with metabolic peculiarities such as a higher disposition index (an estimate of beta‐cell function for a given level of insulin sensitivity), which has been reported to be higher in people of African descent compared to European descent. 29 , 30 Some aspects of the pathophysiology of T2D vary from one ethnic group to another. For example, insulin clearance plays a greater role in the pathophysiology of T2D in individuals with African ancestry than in other ethnic groups. 31 Also, correlations between measures of glycaemic state, such as HbA1c, vary substantially across ethnic groups, partly contributed to by conditions affecting red cell lifespan. 32 Like the pooled results in the meta‐analysis by Richter et al., our study also found that regression to normoglycaemia at follow‐up was more frequent compared to progression to T2D. Similar to the progression from prediabetes to T2D, regression rates are time‐dependent. 24 Based on results from the 47 studies that assessed regression of prediabetes to normoglycaemia in the review by Richter et al., regression ranged from 33% to 59% within 1–5 years of follow‐up, and from 17% to 42% for 6–11 years of follow‐up. 24 Our study agrees with the findings of this review as well as other studies conducted after the review. 24 , 26 , 27 This study also shows that regression to normoglycaemia was more pronounced when FPG instead of HbA1c cut‐off values were used. This is consistent with previous population studies showing that regression from prediabetes to normoglycaemia was three times as frequent at the follow‐up visit among those with HbA1c levels of 5.7%–6.4% compared with those with FPG levels between 5.6 and 6.9 mmol/L. 26 The higher likelihood of regression to normoglycemia using FPG measures compared with HbA1c measures most likely reflects the higher within‐person variability for FPG compared with HbA1c. 33 However, since the same measures were used consistently in comparing the risk of progression to T2D and regression to normoglycemia, the observation that regression to normoglycaemia is more frequent than progression to T2D in West Africans may be valid. The different predictive values of FPG and HbA1c cut‐off values for assessing outcomes of prediabetes may also reflect the influence of residual beta‐cell function, stability of FPG and HbA1c in individuals of African descent, and the influence of hemoglobinopathies. This study reports that the risk of progression from prediabetes to T2D did not significantly differ between non‐migrants and migrants in both crude and adjusted models. We found no prior longitudinal study that compared the risk of progression of prediabetes to T2D in migrants and non‐migrants of the same ethnic origin. However, a cross‐sectional study by Gujral et al., which compared the prevalence of T2D in Indians living in India and the USA, found that Indians who lived in India had a higher prevalence of T2D than those who lived in the USA. 34 This unexpected finding challenges the prevailing notion of migration as a universal risk factor for diabetes and emphasizes the complex relationship between environmental and lifestyle factors in diabetes pathogenesis. In addition to pre‐migration and post‐migration‐related factors, which will drive the progression of prediabetes to T2D, genetic predisposition also contributes to the progression to T2D. 35 The assumption is that the barriers that prevent migrants from leading healthy lives could increase their risk of T2D. 35 A unique finding of this study is that the risk of regression to normoglycaemia in a homogeneous population varies by geographical site of residence. In particular, study participants with prediabetes living in Ghana were more likely to regress to normoglycemia compared to those living in Amsterdam, even after adjustment for a wide range of cardiometabolic risk factors. To prevent the progression of prediabetes to diabetes, the ADA guidelines recommend that adults with prediabetes aim for an intensive lifestyle behaviour change program to achieve and maintain a weight reduction of at least 7% of initial body weight via a healthy reduced‐calorie diet and at least 150 min/week of moderate‐intensity physical activity. 36 In the current study, this may be relevant given that at baseline, migrants had a higher mean BMI than non‐migrants. However, adjusting for baseline BMI did not influence the likelihood of progression to T2D or regression to normoglycemia. These findings suggest that other factors outside the included conventional cardiometabolic risk factors could drive these differences. It may well be that Ghanaian environmental variables have some protective effects against persistent glucose dysregulation in individuals with prediabetes. The lower likelihood of regression to normoglycaemia among migrants may also reflect factors such as lifestyle changes associated with migration, reduced access to preventive health services, and challenges related to social and health system integration. Further studies are needed to clarify how these factors contribute to the observed differences in risk. Studies exploring the underlying mechanisms of the differential risks of regression of prediabetes to normoglycemia are needed to help enhance migrant health. Strengths and limitations To our knowledge, our study is the first to simultaneously report the risk of progression to T2D and regression to normoglycaemia in sub‐Saharan Africans with prediabetes living in different geographical settings. Another strength of this study is that data collection was highly standardized, ensuring comparability over time and between different study sites. This study has several limitations. Analyses were not stratified by sex due to limited sample size and a substantial proportion of participants with prediabetes were lost to follow‐up, largely due to COVID‐19, which may have biased results. Those lost to follow‐up had higher FPG but similar HbA1c, suggesting their exclusion may have underestimated both progression to T2D and regression to normoglycaemia, particularly when defined by FPG. The average follow‐up period of 6.7 years further limits our ability to capture longer‐term transitions from prediabetes to T2D, which are known to vary over time. 24 We relied on only two time points to assess the progression to T2D and regression to normoglycemia, limiting the assessment of intermediate transitions between glycaemic states over time and leading to potential misclassification because of known within‐person variability in FPG and HbA1c. Furthermore, we could not account for changes in diet and physical activity, which are known to influence the progression of prediabetes to T2D as well as regression to normoglycaemia Finally, 2‐h glucose testing was not conducted in the current study, although its additional value is limited by a high within‐person coefficient of variation compared with FPG concentration and HbA1c concentration. 33 CONCLUSION In this longitudinal study across different geographic sites, we found that among West Africans with prediabetes, regression to normoglycaemia was more frequent than progression to T2D, and regression to normoglycaemia was more common in non‐migrants than in migrants. The higher risk of regression to normoglycaemia compared with progression to T2D makes prediabetes a valuable point for interventions targeting regression to normoglycaemia. The lower likelihood of regression to normoglycaemia among migrants compared with non‐migrants suggests a relative disadvantage in achieving normal glycaemic status after the development of prediabetes. Future studies should identify the biological mechanisms underlying the higher rates of regression to normoglycaemia in non‐migrants to improve outcomes in individuals with prediabetes. Potential factors to be explored include measures of insulin secretion, insulin sensitivity, haemoglobin variants, lifestyle changes associated with migration, reduced access to preventive health services, and challenges related to social and health system integration. Future studies are also needed to assess how the differential regression of prediabetes to normoglycaemia affects the health of individuals with prediabetes. AUTHOR CONTRIBUTIONS Emmanuel Bannerman‐Williams: Conceptualization; formal analysis; methodology; writing – original draft; writing – review and editing. Yacoba Atiase: Conceptualization; methodology; supervision; writing – original draft; writing – review and editing. Bert‐Jan H. van den Born: Conceptualization; methodology; supervision; writing – original draft; writing – review and editing. Felix Patience Chilunga: Investigation; methodology; writing – original draft; writing – review and editing. Eva van der Linden: Investigation; methodology; project administration; writing – original draft; writing – review and editing. Karlijn A. C. Meeks: Investigation; methodology; writing – original draft; writing – review and editing. Erik Beune: Investigation; methodology; project administration; resources; writing – original draft; writing – review and editing. Charles Agyemang: Conceptualization; funding acquisition; investigation; methodology; project administration; resources; supervision; writing – original draft; writing – review and editing. Charles Frederick Hayfron‐Benjamin: Conceptualization; formal analysis; methodology; supervision; writing – original draft; writing – review and editing. FUNDING INFORMATION The study was funded by the European Research Council (Grant number 772244). KACM is supported by the NIH Pathway to Independence Award (DK131018). CONFLICT OF INTEREST STATEMENT None declared. Supporting information Table S1. Baseline characteristics of participants included in the study and those lost to follow‐up.