Demographics and Clinical Characteristics

A total of 1,493,964 people with T2D met the study selection criteria, of which 1,061,354 (71%) were in the 1-year follow-up cohort and 308,320 (20.6%) in the 5-year follow-up cohort (Fig. S1). Table 1 presents a summary of demographic and baseline characteristics for both cohorts. The mean age was similar across all weight-change categories in each of the cohorts. The lowest mean age (58.7 years) was observed among participants with ≥ 15% weight loss in the 1-year follow-up period. Conversely, in the 5-year follow-up period, the lowest mean age (59.2 years) was observed among participants with weight increase > 3%. For the cohorts, large decreases in body weight (i.e., 10% or higher) occurred in < 5% of participants during 1-year follow-up and < 15% of participants during 5-year follow-up. There were no substantial differences among the weight-change categories in both follow-up periods for the following variables: race, geographic region, weight loss medications, CCI, or DCSI scores.

Table 1 Patient demographic and clinical characteristics at 1-year and 5-year follow-up

Obesity and sleep apnea were the most commonly diagnosed comorbidities of interest across all weight-change categories. Moreover, these comorbidities were most prevalent among those who experienced greater weight loss percentages. Furthermore, a numerically higher prevalence of bariatric surgery was also observed among people who had a weight loss of ≥ 15% (1-year follow-up cohort: 6.8% vs. 0.2–1.0% for the remaining weight-change categories; 5-year follow-up cohort: 4.0% vs. 0.3–0.7% for the remaining weight-change categories). Overall, the percentage of participants who underwent bariatric surgery tended to increase among patients with greater weight loss in both follow-up cohorts.

Changes in Longitudinal Weight

In the 1-year follow-up cohort, participants with the greatest weight loss, i.e., ≥ 15%, had a 23% (25.9 kg) reduction in weight from the index weight date (Fig. 1a). A similar pattern was observed in the 5-year follow-up cohort, that is, participants with ≥ 15% weight loss had a 22% (22.9 kg) reduction in weight from the index weight date (Fig. 1b). Overall, participants with numerically greater weight loss in follow-up tended to have a higher index weight in both follow-up cohorts.

Fig. 1

a, b Changes in longitudinal weight in 1-year and 5-year follow-up cohorts

Association Between Weight Loss and Glycemic Parameters

During both 1-year and 5-year follow-up periods, the largest decrease in HbA1c was observed among participants who lost ≥ 15% of their index weight: 1.2% and 0.5%, respectively (Fig. 2a, b). The percentage of participants with HbA1c < 7% at weight index date was relatively similar across weight-change categories (Fig. 3). The percentage of participants who attained HbA1c < 7% in the 1-year follow-up cohort was numerically highest among participants with ≥ 10% to < 15% and ≥ 15% weight loss (74% and 77%, respectively, vs. 48–65% for the remaining weight-change categories). A similar pattern was observed for the 5-year follow-up cohort, that is, 56% and 64%, respectively, vs. 40–50% for the remaining weight-change categories. Overall, increasing weight loss percentages were associated with numerically greater reductions in HbA1c levels and a higher proportion of participants achieving HbA1c targets.

Fig. 2

a, b Changes in longitudinal HbA1c values by weight-change categories in 1-year and 5-year follow-up cohorts. HbA1c glycated hemoglobin

Fig. 3

a, b Percentage of participants with HbA1c < 7.0% by weight-change categories in 1-year and 5-year follow-up cohorts. HbA1c glycated hemoglobin

In the 1-year follow-up cohort, those who lost ≥ 15% of their index weight had the largest mean reduction in FBG levels (26.3 mg/dL; Fig. 4a). However, in the 5-year follow-up cohort, a mean increase of 3.6 mg/dL was observed within this category (Fig. 4b).

Fig. 4

a, b Changes in longitudinal fasting blood glucose by weight-change categories in 1-year and 5-year follow-up cohorts

Association Between Weight Change and Metabolic Parameters

In the 1-year follow-up cohort, reductions in LDL (5.4–9.6 mg/dL) and total cholesterol (6.4–13.8 mg/dL) plateaued beyond ≥ 10% weight loss from the index weight date (Fig. 5a, b). In addition, the weight-change categories in the 5-year follow-up cohort (Fig. 6a, b) revealed a similar plateauing effect beyond ≥ 10% weight loss for LDL (9.9–11.9 mg/dL) and total cholesterol (11.6–15.0 mg/dL). Numerically greater improvements in HDL, triglycerides, ALT, and AST levels were associated with increasing weight loss percentages, among both 1-year and 5-year follow-up cohorts.

Fig. 5

a–f Changes in metabolic parameters by weight-change categories in 1-year follow-up cohort. LDL low-density lipoprotein cholesterol, HDL high-density lipoprotein cholesterol, ALT alanine aminotransferase, AST aspartate aminotransferase

Fig. 6

a–f Changes in metabolic parameters by weight-change categories in 5-year follow-up cohort. LDL low-density lipoprotein cholesterol, HDL high-density lipoprotein cholesterol, ALT alanine aminotransferase, AST aspartate aminotransferase

The increase from the index date in HDL ranged from 0.4 to 5.0 mg/dL in the 1-year cohort and 0.6–4.5 mg/dL in the 5-year follow-up cohort (Figs. 5c, 6c). The reduction from the index date in triglycerides ranged from 5.3 to 53.0 mg/dL in the 1-year cohort and 0.3–40.4 mg/dL in the 5-year follow-up cohorts (Figs. 5d, 6d). The level of HDL cholesterol was lower than baseline (an undesirable change) among participants who had a weight increase of > 3% at both 1-year and 5-year follow-up. Total reduction in liver enzymes from index date to follow-up ranged from 0.4 to 9.7 IU/L in both follow-up cohorts (Figs. 5e, f, 6e, f).

Predictors of Weight Loss

A total of 1,049,618 and 307,455 people were categorized into the five index BMI categories for the 1-year and 5-year follow-up cohorts, respectively (Fig. 7a, b). Index BMI data for the remaining participants were unknown or < 18.5 kg/m2 and are not included in this analysis. The percentage of participants who experienced ≥ 15% weight loss in the 1-year follow-up cohort increased from 0.6% to 4.2% as the index BMI increased. Similarly, this percentage increased from 2.3% to 11.0% in the 5-year follow-up cohort.

Fig. 7

a, b BMI at index date among weight-change categories in 1-year and 5-year follow-up cohorts. BMI body mass index, n total number of patients in each weight-change category

The multivariable logistic regression analysis indicated that among all BMI categories, BMI ≥ 35 kg/m2 was the strongest predictor of ≥ 15% weight loss in the 1-year and 5-year follow-up cohorts (1-year follow-up cohort: OR 4.8, 95% CI 4.4–5.3; 5-year follow-up cohort: OR 5.3, 95% CI 4.9–5.8; p < 0.001 for both; Table 2). All other covariates held constant, the strongest predictor of ≥ 15% and ≥ 10% weight loss in both follow-up cohorts was the presence of bariatric surgery (1-year follow-up cohort: OR 17.0, 95% CI 15.8–18.3 and OR 9.6, 95% CI 8.9–10.2, respectively; 5-year follow-up cohort: OR 8.5, 95% CI 7.7–9.4 and OR 4.9, 95% CI 4.5–5.4, respectively; p < 0.001 for all).

Table 2 Predictors of ≥ 15% and ≥ 10% weight loss at 1-year and 5-year follow-up

Additionally, female sex and the use of antidepressants were factors that were positively associated with weight loss in both follow-up cohorts. In the 1-year follow-up cohort, the use of weight loss medications, sulfonylureas, and dipeptidyl peptidase 4 (DPP4) inhibitors were significantly associated with ≥ 10% and ≥ 15% weight loss. However, this association was not significant in the 5-year follow-up cohort, except for DPP4 inhibitors in the ≥ 15% weight loss category (Table 2). The use of GLP-1 RA was not associated with either ≥ 10% or ≥ 15% weight loss. The combined effect of sex, BMI, and age (as interactive predictors) on weight loss was also evaluated, and it was observed that women aged < 65 years with BMI ≥ 35 kg/m2 were significantly more likely to achieve ≥ 10% and ≥ 15% weight loss in the 1-year follow-up cohort. In the 5-year follow-up cohort, women aged > 75 years with BMI ≥ 35 kg/m2 had the highest likelihood of weight loss among all evaluated interactive predictors. Data for all interactive predictors of ≥ 10% and ≥ 15% weight loss are provided in Supplementary Table S2.