In this large population-based study of Iranian urban residents, we found four key findings. First, weight gain ≥ 5% was associated with a minimum 60% higher risk of incident T2DM throughout the 3-, 6-, and 9-year follow-up periods even after adjustment for the participants’ initial weights and well-known diabetes risk factors. Second, weight loss ≥ 5% from the baseline was associated with a risk reduction of over 40% for the incident diabetes during the same follow-up periods. Third, the association between weight change (gain or loss) and incident T2DM did not change after adjustment for attained weight in place of initial weight. Fourth, subgroup analysis showed that the association between weight change and incident diabetes was robust in three sensitivity analyses including age, gender, and BMI categories.

Comparing our results with other studies in this field is not simple and poses some challenges. There were some differences between these studies regarding the duration of follow-up (ranging from 2.5 to 10 years), weight change duration (ranging from 1 to 20 years), weight measurement (self-report or measured), and weight change categorization (absolute change or relative change). Furthermore, variations exist in the analysis approach and level of adjustments for confounders.

Numerous research studies have shown that weight gain increases the risk of developing diabetes [6,7,8,9,10,11]. In our study, we found that 4.4 kg weight gain during 3 years increases the risk of developing diabetes by 26–30% during short and long-term follow-up. These results are in line with the findings reported by Jung et al. [9], demonstrating that each 2.8 kg weight gain over 2 years was associated with a 28% increased risk of incident diabetes during a 5-year follow-up period in the Korean population [9]. In another study, Xu et al. [22] found that each 10-kg weight gain over 30 years increases the risk of incident T2DM among the Chinese population by about 40%. With a similar weight gain among the Japanese population, the risk of incident T2DM was more than threefold over a 20-year follow-up period [10]. In another study among a cohort of US adults, Ford et al. [7] demonstrated that an increase in weight gain > 11 kg over 9 years was associated with an elevated risk of developing diabetes. Based on the change ratio, our results suggest that a weight gain < than 5% over three years does not yield significant short- or long-term implications for the incident T2DM in the general population. Conversely, a weight gain ≥ 5% is linked to at least a 60% rise in the risk of developing T2DM. This aligns with Ohno et al. [6] research, which showed that an increase of ≥ 5% in BMI over 2.5 years, compared to a stable change group, was significantly associated with a 33% higher risk of developing T2DM and reducing by at least 5% from the initial BMI was associated with an 18% reduction in the risk of incident T2DM.

Although several studies have demonstrated the association between weight loss and the incidence of diabetes [6, 9, 10], results have been inconsistent. In our data analysis weight loss ≥ 5% showed a suggestive 52%, 43%, and 49% decreased risk of incident diabetes during 3-, 6-, and 9- year follow-up, respectively. In line with ours, results from the Japanese population show that BMI loss ≥ 5% over 2.5 years, decreased the risk of 5-year incident diabetes by 18% [6]; however, reviewing cohorts conducted among American [7] and Dutch populations [11], did not find a significant association between weight loss and incident T2DM. Moreover, post-trial monitoring of the Diabetes Prevention Program Outcomes Study [15] and the China Da Qing Diabetes Prevention Study [16] found that the beneficial impact of lifestyle intervention for the prevention of diabetes persists for the period as long as ten and twenty years, respectively. Similarly, in our observational study, we found that a 3-year weight change reduced the risk of incident T2DM during a ten-year follow-up.

Research indicates that lifestyle or medical interventions are playing a higher role in weight loss which causes a lower rate of incident diabetes [4]. Results from the Look AHEAD trial show that in the intensive lifestyle intervention group, about 5% weight loss during 8 years improves glycemic control in people with diabetes [23, 24]. Besides, the effect of weight change in general population cohorts without interventions was also discussed [7,8,9,10,11, 25], and generally, no significant benefits of weight loss on risks of developing T2DM were reported. This risk reduction might be related to the improved nutrition quality of Iranian adults including an increase in total dietary Dietary Approach to Stop Hypertension (DASH) score, consumption of whole grains, legumes, nuts, and seeds, and a decrease in consumption of refined grain, and solid fats over a decade follow-up [26]. In the current study, the impact of physical activity on weight change was not examined, although the level of physical activity among Iranians has not improved over the past decade [27]. Importantly, we excluded individuals with cancer or recent hospitalizations within the last three months at the beginning of the study to address the potential impact of comorbidities on weight reduction.

To explore whether, at a certain level of attained weight, weight change history is important we adjusted attained weight, an issue that was poorly addressed in previous studies. We found that weight gain or loss has a significant impact on the incidence of T2DM, independent of attained weight measurements. In line with our findings, Kaneto et al. [10] demonstrated that weight changes over 5 years remained significantly associated with the occurrence of diabetes even after adjusting for final weight among 13,700 Japanese population aged 35–55 years. However, in the Dutch population-based Doetinchem cohort study, the significant association between weight change and the incidence of diabetes reached null [11].

In our subgroup analysis, we did not find an effect modification of gender, age, and BMI categories on incident diabetes. Similarly, the pooled cohort study conducted in Germany did not show the effect of modification of sex for the incidence of T2DM. Moreover, in the Japanese cohort, Ohno et al. [6] reported that both BMI gain and loss of ≥ 5% increased and decreased the risk of developing incident T2DM over 2.5 years to the same level, across normal weight, overweight, and obese individuals, respectively.

Our study strengths include the utilization of objectively measured weight data in follow-up re-examinations, distinguishing us from several studies that rely on self-reported data questionnaires, susceptible to recall information bias. Furthermore, our evaluation encompassed short, medium, and long-term follow-ups for incident diabetes. However, it is imperative to acknowledge certain limitations. Firstly, due to the absence of HbA1c data, diabetes was defined solely based on FPG and 2h-PCG. Secondly, the measures of weight change do not differentiate between alterations in lean or fat mass. Thirdly, since nutrition information was not assessed at re-exam 1 and only among a limited sample of participants in re-exam 2 (≈ 20%), we were not able to examine the impact of change in dietary behavior and calorie intake on observed weight change. Lastly, our study population is confined to residents of Tehran, a metropolitan city, sharing a uniform ethnicity; hence, caution should be exercised in generalizing our findings to rural populations or other ethnicities.

In conclusion, our data suggest that each 4.4 kg of weight gain during 3 years increases the risk of developing T2DM by at least 26%, even after considering attained weight. Moreover, the beneficial effect of weight loss is sustained for a decade follow-up. Care should be considered regarding an individual’s weight change, and lifestyle interventions are highly recommended to achieve ≥ 5% weight loss to prevent T2DM.