Exploring ways to respond to rising obesity and diabetes in the Caribbean using a system dynamics model

**Background:** Diabetes and obesity present a high and increasing burden in the Caribbean that has failed to respond to prevention policies. These conditions result from a complex system of drivers, making intervention impact prediction difficult. This study developed a system dynamics simulation model to map the system driving diabetes and obesity in the Caribbean using Jamaica as a test case, aiming to assess changes needed to achieve WHO Global Action Plan targets and test intervention scenarios.

**Methods:** The model was developed in partnership with stakeholders through group model building sessions, causal loop diagrams, and in-depth interviews. Jamaica was selected due to its population-based health surveillance surveys and relevance to other English-speaking Caribbean countries. The model uses a stock-and-flow structure tracking adults (20+ years) through glycemic states: normoglycemia, prediabetes, and diabetes. Inputs included population data from UN World Population Prospects, baseline estimates from Jamaican Health and Lifestyle Surveys, incidence rates from systematic reviews, physical activity trends from Ng and Popkin, and dietary intake data from Caribbean and Latin American studies. The model was implemented using Silico App and R with deSolve package, projecting from 1990 to 2050. Three scenario sets were tested: achieving WHO global targets, stakeholder-determined priorities, and single interventions.

**Key Results:** Continuing current trends, the model predicts diabetes prevalence in Jamaican adults will rise from 12% in 2018 to 15.4% in 2030 and 20.9% by 2050. Obesity prevalence is projected to rise from 28.6% in 2018 to 32.1% by 2030 and 39.2% by 2050. Women have three times the obesity prevalence of men in 2020 (38.4% vs 12.5%), narrowing to 48.8% vs 22.2% by 2050. To achieve WHO global targets (no increase from 2010 to 2025), the model required a tripling of moderate-to-vigorous physical activity and a 30% reduction in caloric intake—deemed unachievable. For more plausible Caribbean-relevant scenarios, a combined intensive upstream and downstream intervention reduced diabetes prevalence by 1.9 percentage points (12.3%) by 2030 and 3.9 p.p. (18.7%) by 2050. The most effective single intervention was reducing sugar-sweetened beverage consumption by 25%, yielding a 2.3 p.p. (14.7%) reduction in diabetes prevalence by 2030 and 4.8 p.p. (23.1%) by 2050. Adding 30 minutes/day of MVPA reduced diabetes prevalence by 0.75 p.p. (4.8%) by 2030 and 1.42 p.p. (6.8%) by 2050. For obesity, reducing ultra-processed food consumption by 25% reduced prevalence by 0.3 p.p. (1.0%) by 2030 and 1.2 p.p. (3.1%) by 2050. Downstream interventions alone had minimal population-level impact (0.2 p.p. reduction in diabetes by 2050).

**Clinical Implications:** The WHO global targets for halting obesity and diabetes rise by 2025 are unrealistic given the magnitude of change required. No single intervention is as effective as a combination of upstream and downstream approaches. The model strongly supports adopting sustained, coordinated multi-sector policies including SSB taxation, front-of-package labelling, public information campaigns, improved infrastructure for active transport, and health system interventions targeting high-risk individuals. Even modest relative reductions compared to baseline trends are clinically valuable given the broader benefits of obesity reduction on quality of life, cancer risk, and cardiovascular disease. The findings reinforce that NCD prevention requires long-term commitment beyond single political cycles.