The intention-to-treat population
Data handling is shown in Fig. 1. A total of 574,004 unique adult patients were seen in 1 of 56 clinics between 17 March 2020 and 16 March 2024. Of these, 274,182 had a body mass index (BMI) ≥ 25 kg m2 with at least two weight measurements recorded in the EHR and are included in the primary outcome analysis assessing change in weight. Of these patients, 189,227 were first weighed during usual care and 84,955 were first weighed in the intervention (147,455 weighed in usual care were also weighed in the intervention). The demographics and health metrics of patients included in this analysis are shown in Table 1 and are highly representative of the demographics of adults residing in Colorado. Supplementary materials provide additional details, including participating clinics (Extended Data Table 1), an operational definition of the Edmonton Obesity Staging System (EOSS; Extended Data Table 2), captured weight-related comorbidities (Extended Data Table 3) and modeling outputs for Fig. 2a,b (Extended Data Tables 4 and 5).
Participant flow diagram. Hatched bars indicate the usual care phase. Solid bars indicate the intervention phase. UC, usual care; Int, intervention.
a,b, Predicted weight trajectories for eligible patients with a measured weight in the usual care (blue line) and intervention (red line) phases from 0 month to 6 months and 6 months to 18 months from their first weight regardless of whether they received discernable weight-related care in the ITT sample (a), and who received weight-related care in the usual care (blue line) or intervention phase (pink line), and those who never received weight-related care in the usual care (purple line) or intervention phase (green line) in the prespecified secondary analysis (b). The two-piece (0–6 months and 6–18 months) solid lines are predicted weights for a hypothetical average patient during follow-up with 95% prediction intervals in gray.
Coprimary outcomes comparing usual care and the intervention in eligible patients
The intervention (PATHWEIGH) included 3 components: (1) health system primary care leadership endorsement, (2) an EHR-driven care process designed to prioritize, facilitate and expedite weight management and (3) implementation strategies to support use of the care process and educate clinicians on obesity treatment. The coprimary outcomes were average patient weight loss at 6 months and weight loss maintenance from 6 months to 18 months. From the intention-to-treat (ITT) population, the average time these patients spent in usual care was 9.1 months and the average time these patients spent in the intervention phase was 13.7 months. Model-adjusted predicted average weight increased by 0.29 kg (95% confidence interval (CI): 0.27 kg, 0.32 kg) from the first weight to 6 months later (P < 0.001) and by 0.18 kg (95% CI: 0.15 kg, 0.21 kg) from 6 months to 18 months later (P < 0.001) for an average total weight gain of 0.47 kg (95% CI: 0.45 kg, 0.50 kg) in the usual care phase (P < 0.001). Model-adjusted predicted weight decreased by 0.00 kg (95% CI: −0.03 kg, 0.03 kg) from the first weight to 6 months later (P = 0.98) and by 0.10 kg (95% CI: 0.07 kg, 0.12 kg) from 6 months to 18 months later (P < 0.001) for an average total weight loss of 0.10 kg (95% CI: 0.07 kg, 0.13 kg) in the intervention phase (P < 0.001). A counterfactual analysis comparing differences in weight between the intervention and usual care suggests that the intervention decreased average weight by 0.29 kg (95% CI: 0.27 kg, 0.32 kg) from the first weight to 6 months later (P < 0.001) and 0.28 kg (95% CI: 0.26 kg, 0.31 kg) from 6 months to 18 months (P < 0.001) for a total difference of 0.58 kg (95% CI: 0.54 kg, 0.61 kg) (P < 0.001), showing the intervention’s ability to eliminate the population weight gain observed in usual care (Fig. 2a and Extended Data Table 4).
Approximately 25% of eligible patients received discernable care for their weight at least once during the trial (in one phase or both; Table 1). Of these patients, 50,482 were first weighed during usual care and 18,900 were first weighed in the intervention (44,215 weighed in usual care were also weighed in the intervention). Notably, a higher proportion of patients who received discernable care for their weight (versus those who did not) were commercially insured (67% versus 62%) women (61% versus 51%) with a higher average BMI (36 kg m−2 versus 30 kg m−2) at their initial weight measurement (Table 1). Results from the generalized estimating equations (GEE) logistic model indicated that the intervention increased the likelihood of a patient receiving discernable care for their weight by 23% (odds ratio = 1.23 versus usual care; 95% CI: 1.16, 1.31; P < 0.001).
Prespecified secondary analysis among patients identified as having received discernable care for their weight
Among patients receiving discernable care for their weight during the study period, the average time these patients spent in usual care was 11.3 months and the average time these patients spent in the intervention phase was 32 months. Counterfactual analysis model-adjusted predicted average weight decreased by 0.06 kg (95% CI: 0.00 kg, 0.12 kg) from the first weight to 6 months later (P = 0.037) and by an additional 0.39 kg (95% CI: 0.35 kg, 0.43 kg) from 6 months to 18 months later (P < 0.001) for a total weight loss of 0.45 kg (95% CI: 0.40 kg, 0.49 kg) (P < 0.001) for those receiving weight-related care during usual care. Model-adjusted predicted average weight decreased by 0.88 kg (95% CI: 0.81 kg, 0.95 kg) from the first weight to 6 months later (P < 0.001) and by an additional 1.30 kg (95% CI: 1.25 kg, 1.35 kg) from 6 months to 18 months later (P < 0.001) for a total weight loss of 2.18 kg (95% CI: 2.12 kg, 2.24 kg) (P < 0.001) for those receiving weight-related care during the intervention phase. The adjusted difference between usual care and the intervention was 1.73 kg more weight loss over 18 months in the intervention for those receiving weight-related care (95% CI: 1.68 kg, 1.78 kg, P < 0.001; Fig. 2b and Extended Data Table 5).
Prespecified secondary analysis among patients identified as having never received discernable care for their weight
Among patients never receiving discernable care for their weight during the study period, 138,745 were first weighed during usual care and 66,055 were first weighed in the intervention phase (103,240 weighed in usual care were also weighed in the intervention; Table 1). The average time these patients spent in usual care was 9.56 months, and the average time these patients spent in the intervention phase was 26.9 months. Counterfactual analysis model-adjusted predicted average weight increased by 0.29 kg (95% CI: 0.26 kg, 0.32 kg) from the first weight to 6 months later (P < 0.001) and by an additional 0.26 kg (95% CI: 0.23 kg, 0.29 kg) from 6 months to 18 months later (P < 0.001) for a total weight gain of 0.55 kg (95% CI: 0.52 kg, 0.58 kg) (P < 0.001) during usual care. Model-adjusted predicted average weight increased by 0.08 kg (95% CI: 0.045 kg, 0.11 kg) from the first weight to 6 months later (P < 0.001) and by an additional 0.10 kg (95% CI: 0.07 kg, 0.14 kg) from 6 months to 18 months later (P < 0.001) for a total weight gain of 0.18 kg (95% CI: 0.15 kg, 0.22 kg) (P < 0.001) during the intervention phase. The adjusted difference of 0.32 kg over 18 months in usual care versus intervention (95% CI: 0.30 kg, 0.35 kg; P < 0.001) is the amount of intervention-mitigated weight gain even when patients did not receive weight-related care (Fig. 2b and Extended Data Table 5).
Weight trajectories in those who did versus did not receive weight-related care
An associative counterfactual analysis comparing the weight trajectories of patients who received discernable care for their weight and those who did not indicated that these two subpopulations have different weight trajectories. The model-adjusted difference in weight for those with an initial visit in usual care who received weight-related care weighed was 0.35 kg (95% CI: 0.30 kg, 0.40 kg) lower at 6 months (P < 0.001) and an additional 0.65 kg (95% CI: 0.63 kg, 0.68 kg) lower from 6 months to 18 months later (P < 0.001) than would be expected without weight-related care. The adjusted difference of 1.00 kg over 18 months (95% CI: 0.96 kg, 1.04 kg; P < 0.001) represents the difference in weight for those who did versus did not receive weight-related care during usual care (Fig. 2b). The model-adjusted difference in weight for those with an initial visit in the intervention phase who received weight-related care weighed was 0.96 kg (95% CI: 0.89 kg, 1.03 kg) less at 6 months (P < 0.001) and an additional 1.41 kg (95% CI: 1.36 kg, 1.45 kg) from 6 months to 18 months later (P < 0.001) than would be expected without weight-related care. The adjusted difference of 2.37 kg over 18 months (95% CI: 2.33 kg, 2.40 kg; P < 0.001) represents the difference in weight for those who did versus did not receive weight-related care during the intervention phase (Fig. 2b).
Delivery of weight-related care
Trackable weight-related care included referrals, performance of a bariatric procedure and patient acknowledgement that an anti-obesity medication was actively being used. Clinician counseling on lifestyle modification was not trackable, but rather presumed when the clinician used a weight-related International Classification of Disease-10 code for billing without ordering of the treatments above. Chi-square tests (Table 2) indicated that the proportion of patients receiving referrals to the Health and Wellness Center (a weight loss clinic) was lower in the intervention phase compared with usual care, χ2(1) = 8.38, P = 0.004, 95% CI (5.9%, 4.9%) as was the proportion of patients who received bariatric surgery, χ2(1) = 7.22, P = .007, 95% CI (0.6%, 0.08%). In contrast, the proportion of patients reporting use of anti-obesity medications, χ2(1) = 107.77, P < 0.001, 95% CI (6.5%, 4.4%), was higher in the intervention phase than in usual care. No other significant differences in the proportions of patients receiving the remaining referrals were observed between the intervention and usual care phases.
Engagement of the clinics with the implementation strategies
A crude estimate of clinic engagement was quantified using an engagement score (0–8; low to high engagement) and was based on the clinics’ and/or at least one clinician per clinic documented participation in up to 8 implementation activities (participation in each activity by the 56 clinics is shown in parentheses): (1) virtual introductory meeting (55/56), (2) in-person all-clinic training (49/56), (3) individual consultation (29/56), (4) obesity e-learning module (35/56), (5) World Obesity Federation SCOPE training (17/56), (6) posted signage informing patients that weight-prioritized visits were available (18/56), (7) attending a learning community meeting (14/56) and/or (8) identifying a champion for PATHWEIGH (18/56). A total of 36 clinics (64%) showed moderate engagement (score 3–5), 12 clinics (21%) engaged to a greater degree (score 6–8) and 8 clinics (14%) engaged to a lesser degree (score 0–2; Fig. 3).
Percentage of the 56 clinics that participated in each implementation activity. WOF training, World Obesity Federation SCOPE training.
Safety
No health metric (Table 1) changed ≥1% in an unfavorable direction in the intervention. Death rates were very low in our patient population, 0.6% during usual care and 1.7% during the intervention (each over 3 years). The higher death rate during the intervention is probably due to the enrichment of patients seen in both phases who were older in the intervention versus usual care (>50% of our population; Table 1). Due to the timing of the trial (data capture began in March 2020; the first intervention group started in March 2021), COVID-related deaths became much less common as more patients were being exposed to the intervention.
link