Real-world use of semaglutide for obesity treatment: A cohort study in an integrated health care delivery system

Abstract

BACKGROUND:

Novel antiobesity medications have demonstrated significant weight loss in clinical trials. Data on real-world use of semaglutide are lacking. Such data are vital to inform health benefit design of this and other costly incretin mimetics, as well as inform patients of expected outcomes outside of clinical trials.

OBJECTIVE:

To evaluate real-world persistence and adherence to semaglutide for obesity treatment and impacts on weight loss.

METHODS:

The primary outcome was persistence to therapy. Secondary outcomes were adherence to therapy and its approved dosing schedule. Weight loss observations for all outcomes were also analyzed. This retrospective, cohort, observational study evaluated prescription and clinical data from a single health system between 2021 and 2024. The index date was defined as the first semaglutide claim filled for weight loss, and all refill data were obtained up to 12 months following. Electronic medical record data were collected at 6 and 12 months. Persistence was defined as the absence of any therapy gap exceeding 60 days within 12 months. Adherence was defined as proportion of days covered greater than or equal to 80%. Weight loss was calculated as percent change from baseline, defined as the collected weight closest to the index date. Multivariable logistic regression was used to identify factors independently associated with persistence and adherence at 12 months.

RESULTS:

A total of 393 patients were included. Throughout the 12-month follow-up period, 44% (172) of patients remained persistent and lost a mean of 13.8% of their baseline weight, compared with 6.4% for nonpersistent patients. At 12 months, 44% (175) patients were adherent and experienced 13% weight loss. For those closely following the approved titration schedule, a 12.5% average weight loss at month 12 was observed, almost double that of patients using other titration patterns (7.2% average weight loss at month 12). In multivariable logistic regression analyses, compared with male sex, female sex was associated with lower odds of persistence (odds ratio = 0.32; 95% CI = 0.14-0.75) and adherence (odds ratio = 0.33; 95% CI = 0.14-0.77), whereas race, ethnicity, and comorbidities were not statistically significantly associated with persistence or adherence.

CONCLUSIONS:

Less than 45% of patients were persistent or adherent to semaglutide for weight loss during the initial 12 months of use. The results of this study suggest that adding support measures and/or altering coverage criteria based on persistence or adherence patterns may be reasonable to consider. Efforts to promote persistence and adherence during the first year of treatment, as well as proper implementation of titration schedules, may lead to improved weight loss in patients using semaglutide.

Plain language summary

This study found that more than half (56%) of patients with obesity who were taking semaglutide for obesity were neither persistent nor adherent. These behaviors appear to compromise weight loss results. Further, some patients did not use semaglutide appropriately, yet those who do are observed to have impressive weight loss. Semaglutide, when used according to package labeling, appears to increase weight loss results during the first 6 months of use.

Implications for managed care pharmacy

This study evaluated semaglutide use in employees of a single health system. Results showed that less than half of those using semaglutide for obesity treatment were persistent or adherent. Semaglutide remains highly utilized for obesity and its related complications. Results from this study can aid payers in designing appropriate levels of coverage for this and other emerging incretin mimetics.

The chronic disease of obesity has reached epidemic proportions in the United States, with data to support an increasing rate in the coming years.^1,2 There has and continues to be significant growth in medication options that aid in weight loss. Newer antiobesity medications (AOMs) known as “incretin mimetics” target pathways in the gut, liver, and pancreas to produce weight loss results, exceeding efficacy of previously approved AOMs.³ One such incretin mimetic, semaglutide, gained popular use for the management of type 2 diabetes mellitus (T2DM) starting in 2017, prior to its approval as an AOM in 2021.^4–6 Although semaglutide and other incretin mimetics (eg, liraglutide, tirzepatide) like it are effective, they come with a considerable cost burden to both payers and patients.⁷

Semaglutide and similar agents are costly, ranging from roughly $300 to $1,300 per month when insurance coverage is not applied, depending on the medication dose selected and product source.^8–10 Uncovering adherence trends and understanding factors associated with nonadherence is vital, particularly for insurers and institutions offering this and similar agents for obesity treatment. Even as the cost of incretin mimetics continues to evolve, this burden remains considerable for patients, payers, and the US health care system as they attempt to address costs of the obesity epidemic.^11,12

The current behaviors of patients when using semaglutide is imperative to understand. This is due to the significant impact that patient adherence can have on semaglutide’s effectiveness. Persistence with use of incretin mimetics for obesity has been shown to be higher than other AOMs in a prior study evaluating liraglutide, yet overall low (42% at 6 months).¹³ Adherence (as measured by proportion of days covered [PDC]) to incretin mimetics for obesity has also been assessed, and a PDC of 80% or greater was identified in 27% of individuals prescribed any incretin mimetic for weight loss.¹⁴ In the same study, adherence with this semaglutide was found to be higher (63%).¹⁴ Although these results are useful, correlation of adherence to weight loss is needed to fully describe the current state of real-world semaglutide use in obesity. Only such data will allow stakeholders involved in the payment of semaglutide for obesity to grasp potential implications of nonpersistence and nonadherence.

The objective of this study was to evaluate persistence and adherence to semaglutide used for obesity treatment in a real-world setting. We also sought to determine whether weight loss was associated with adherence and persistence to semaglutide and whether dosing aligned with US Food and Drug Administration (FDA) labeling. This study used data from commercially covered members who received care at 1 health system. Such employees and their spouses (if applicable) in this single health system who had coverage for and were treated with semaglutide were studied.

Methods

STUDY DESIGN

This cohort study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline (Supplementary Figure 1 ^{(457.9KB, pdf)} , available in online article). It was a retrospective, observational study using prescription claims data and clinical data from electronic medical records (EMRs) of a single health system, Cone Health. This health system is a nonprofit entity, with employees across multiple hospitals, outpatient clinics, and additional care sites that provide care to 450,000 to 500,000 patients annually.¹⁵ This study was approved by the Cone Health Institutional Review Board (ID 2242034-1).

PATIENT SELECTION

Employee beneficiaries (and their spouses, if applicable) were eligible for coverage for semaglutide for obesity (brand name Wegovy) under Cone Health’s prescription coverage during June 31, 2021, to April 31, 2024. Employees using such coverage (n = 617 in June 2024) were identified from prescription claims through a text search for “semaglutide.” Patients exclusively receiving products other than the brand name Wegovy were excluded, as were claims for obesity or diabetes products other than Wegovy.

From this cohort, employees with prescriptions written by providers outside of Cone Health and those who received outpatient physician services outside of Cone Health during the collection period were excluded, verified through reviewing prescribers against a list provided by Cone Health administration. These exclusion criteria were established based on the inability to verify ongoing follow-up and management of obesity and semaglutide. Employees were also excluded if they were younger than 18 years.

OUTCOME VARIABLES

The primary outcome for this study was persistence in patients receiving semaglutide for weight loss. Persistence was defined as continuous semaglutide use without a gap of more than 60 days between prescription fills. Persistence was evaluated at 2 prespecified time points—6 months (180 days) and 12 months (365 days) after the index date—and using multiple time intervals (months 1-6, 7-12, and 1-12 from the index date). The index date was defined as the date of the first semaglutide prescription claim. Patients with incomplete or missing prescription claims data within either follow-up period were classified as nonpersistent for that time point.

Secondary outcomes were patient adherence to semaglutide, and implementation of its approved dosing strategy for obesity. Adherence was assessed using the PDC, calculated over the same 6- and 12-month intervals following the index date, and using multiple time intervals (months 1-6, 7-12, and 1-12 from the index date). A PDC of at least 80% was used to define adherence to semaglutide, consistent with prior literature.¹⁴ Patients with a PDC below this threshold, or with missing prescription data within the assessment window, were classified as nonadherent.

To determine whether the FDA-approved dose titration schedule was followed, we assessed dosing at the 6-month time point as the maximum amount of time required for dose titration per the FDA labeling. Persistence and adherence was also assessed at 6 months from initiation to serve as a logical midpoint evaluation for the 12-month time point. Both of these time points have been used in prior, relevant work.^13,14 A 1-month supply of semaglutide was counted as 28 days based on its labeled packaging.¹⁶ Semaglutide’s FDA-approved dose titration schedule follows a progression of monthly dose increases from a starting dose of 0.25 mg weekly. Patients reach a target dose of either 1.7 mg or 2.4 mg weekly with a possible delay in dose escalation due to tolerability for 1 month. Regression to prior doses is not advised, nor is maintaining doses other than 1.7 mg or 2.4 mg weekly long term for the indication of obesity.

When the FDA schedule is followed, titration to maintaining 1.7 mg takes 4 to 5 months, and 5 to 6 months for maintaining 2.4 mg. Implementation of this schedule was assessed in this study by examining doses received over time in relation to the approved dosing regimen. Regression to lower doses and maintaining doses other than 1.7 mg or 2.4 mg were considered to be nonapproved dosing. A single, 4-week delay while maintaining the previously prescribed dose was permitted. Patients who did not follow the FDA-approved schedule within the 6-month assessment period, including for reasons of nonpersistence and nonadherence, were considered to have used a nonapproved dosing schedule. Missing data were not imputed for this analysis, and a single, incorrect, or missed fill or missing data point 28 days or more after the prior fill was considered nonapproved dosing.

The bivariate association between persistence/adherence and weight loss was also determined, using the percent change from baseline weight. Baseline weight was the weight measured during routine care closest to the index date, within a ±60-day window. Follow-up weights were obtained approximately 6 and 12 months after the index date, again using values recorded within ±60 days of each measurement period. Percent weight change at multiple time intervals (months 1-6 and 1-12 from the index date) was calculated as (weight at time point − baseline weight)/baseline weight × 100. Months 7 to 12 represent additional, incremental weight loss achieved during this period and was calculated as (weight at month 12 – weight at month 6)/weight at month 6 × 100. Body mass index (BMI) was calculated at each time point using the corresponding weight and the baseline height measurement, reported in kilograms per square meter.

INDEPENDENT VARIABLES

Independent variables were selected a priori based on clinical relevance to better characterize factors associated with persistence and adherence. Demographic characteristics included age, sex, race, and ethnicity. Age, measured in years at the index date, was categorized into 5 age groups: 18 to 30 years, 31 to 40 years, 41 to 50 years, 51 to 60 years, and 60 years and older. Sex was extracted from patient records and coded as male or female. Race and ethnicity were obtained from the electronic medical record at the index date and were modeled as categorical variables based on the classifications recorded in routine clinical care, including identification as Hispanic or Latino vs not Hispanic or Latino.

Clinical characteristics consisted of comorbid conditions present at baseline, including hypertension, hyperlipidemia, T2DM, cardiovascular disease, and major depressive disorder. Each comorbidity was coded as a dichotomous variable indicating the presence or absence of that condition at the index date (see Supplementary Table 1 ^{(457.9KB, pdf)} for a complete list of International Classification of Diseases codes to identify comorbidities). Measures of baseline anthropometrics were also included: baseline weight, recorded in kilograms, and BMI. These values were obtained from the measurement closest to the index date within a ±60-day window. Calculated BMI values were categorized into 4 classes according to current classification guidelines (overweight, BMI 25-29.5; obesity class I, BMI 30-34.9; obesity class II, BMI 35-39.9; and obesity class III, BMI 40 and above).¹⁷

DATA ANALYSIS

Demographic and clinical data were extracted from the EMR and linked with claims data using a unique identifier prior to deidentification. All demographic and clinical data measured and collected in the EMR as part of routine patient care were extracted in November 2024.

Persistence and adherence were first evaluated using unadjusted bivariate analyses at 3 predefined assessment periods: months 1 to 6, months 7 to 12, and months 1 to 12 following the index date. Differences in demographic and clinical characteristics between persistent or adherent and nonpersistent or nonadherent patients at each time point were assessed using Pearson χ² or Fisher exact tests for categorical variables and Wilcoxon rank-sum tests for continuous variables, as appropriate.

Multivariable logistic regression models were used to examine factors associated with persistence and adherence. Separate models were fit for 12-month persistence and 12-month adherence, each coded as a binary dependent variable (1 = persistent/adherent, 0 = nonpersistent/nonadherent). No single independent variable was prespecified as the primary exposure of interest; rather, the objective of these models was to estimate the association between patient demographic and clinical independent variables, the odds of being persistent or adherent at 12 months. Clinical variables were modeled as a dichotomous variable, and reference categories were male sex, White or Caucasian race, not Hispanic or Latino ethnicity, absence of each comorbid condition, and age 18 to 30 years. BMI categories at baseline were not statistically significantly associated with persistence or adherence in bivariate analyses; therefore, this variable was not included in the multivariable models.

Results from the multivariable logistic regression models were reported as odds ratios (ORs) with corresponding 95% CIs. Results were statistically significant if the 95% CI did not include the null value. The analyses were performed in R using version 4.4.0 and RStudio version 2024.04.1 build 748 on Windows 11.^18–20

No modification for potential missing data was made for demographic and clinical variables. Missing data for weight and height were not imputed, but rather, patients without data for baseline, 6-month, or 12-month time points were analyzed for the data that were available within the collection time boundaries. Any potential missing prescription claims that were filled but not collected in data analysis contributed to nonpersistent and nonadherent classification.

Results

A total of 393 patients were included in the final analysis. Supplementary Figure 2 ^{(457.9KB, pdf)} outlines identification of patients and prescription claims data included. The average age of patients in this cohort was 47 years (SD = 11). The majority of patients were female (358 patients, 91%), White or Caucasian (241, 61%), and not Hispanic or Latino (373, 95%). The most common comorbidity among the cohort in this study was hyperlipidemia (229, 58%), followed by hypertension (196, 50%). A minority of patients had comorbid T2DM (49, 12%). See Supplementary Table 2 ^{(457.9KB, pdf)} for all missing data for this cohort. Patients for whom baseline data were available regarding weight (83%) and BMI (77%) averaged 227 lb and 37 kg/m², respectively.

Table 1 shows persistence and adherence results at month 12 and associations between demographic and clinical variables for each. During the entire analysis period (months 1-12), the cohort demonstrated similar persistence (172, 44%) and adherence (175, 45%) rates. Most patients were persistent during months 1 to 6 (275, 70%), which declined during months 7 to 12 (197, 50%, a 28.6% decrease). Similar findings were seen for adherence (231, 59% were adherent for months 1-6; 145, 37% were adherent for months 7-12). Figure 1 displays results for persistence and adherence during months 1 to 6, 7 to 12 and 1 to 12. Supplementary Table 3 ^{(457.9KB, pdf)} contains statistical analysis of demographic and clinical variables with persistence and adherence for months 1 to 6, whereas Supplementary Table 4 ^{(457.9KB, pdf)} contains this analysis for months 7 to 12.

TABLE 1.

Persistence and Adherence With Semaglutide at 12 Months

Characteristic	Not persistent at 12 months (n = 221, 56.2%)	Persistent at 12 months (n = 172, 43.8%)	P valuea	Nonadherent at 12 months (n = 218, 55.5%)	Adherent at 12 months (n = 175, 44.5%)	P valuea
Age, mean (SD), years	47 (11)	47 (11)	0.495	46 (12)	48 (11)	0.239
Female sex, n (%)	194 (54.2)	164 (45.8)	0.015	191 (53.4)	167 (46.6)	0.012
Male sex, n (%)	27 (77.12)	8 (22.9)		27 (77.1)	8 (22.9)
Race, n (%)
White or Caucasian	134 (55.6)	107 (44.4)	0.764	131 (54.4)	110 (45.6)	0.688
Black or African American	70 (56)	55 (44)		70 (56)	55 (44)
Otherb	17 (63)	10 (27)		17 (63)	10 (27)
Ethnicity, n (%)
Not Hispanic or Latino	206 (55.2)	167 (44.8)	0.132	203 (54.4)	170 (9,745.6)	0.112
Hispanic or Latino	15 (75)	5 (25)		15 (75)	5 (25)
Comorbidities, n (%)c
HTN	111 (56.6)	85 (43.4)	0.915	107 (54.6)	89 (45.4)	0.948
HLD	119 (54.3)	100 (45.7)		121 (55.3)	98 (44.7)
T2DM	25 (51)	24 (49)		25 (51)	24 (49)
CVD	29 (54.7)	24 (45.3)		30 (56.6)	23 (43.4)
MDD	78 (52)	72 (48)		78 (52)	72 (48)
BMI categories, n (%)d
Overweight	23 (74.2)	8 (25.8)	0.282	18 (58.1)	13 (41.9)	0.965
Obesity class I	55 (55.5)	44 (44.5)		58 (58.6)	41 (41.4)
Obesity class II	45 (56.3)	35 (43.7)		44 (55)	36 (45)
Obesity class III	51 (56)	40 (44)		51 (56)	40 (44)

^a Based on chi-square test or Mann-Whitney U-test.

^b Asian, American Indian or Alaskan Native, and all other races specified.

^c Values listed are only for those diagnosed with the listed comorbidity.

^d Missing values reported for BMI categories are 92 in total.

BMI = body mass index; CVD = cardiovascular disease; HLD: hyperlipidemia; HTN = hypertension; MDD = major depressive disorder; T2DM = type 2 diabetes.

FIGURE 1.

Persistence and Adherence With Semaglutide for Weight Loss at Various Time Points

Persistence was defined as continuous semaglutide use without a gap of 60 days or more between prescription fills. Adherence was defined as a proportion of days covered of at least 80%.

Table 2 shows results from the multivariable logistic regression evaluating factors associated with persistence and adherence at 12 months. This analysis showed that female sex was associated with lower odds of persistence compared with male sex (OR = 0.32; 95% CI = 0.14-0.75). Similarly, in the multivariable logistic regression assessing factors associated with adherence at 12 months, female sex was associated with lower odds of adherence compared with male sex (OR = 0.33; 95% CI = 0.14-0.77). Race, ethnicity, age group, and comorbidities were not significantly associated with persistence or adherence.

TABLE 2.

Adjusted Odds of Treatment Persistence and Adherence According to Demographic Characteristics and Comorbidities

Characteristic	Persistence from months 1 to 12		Adherence from months 1 to 12
	Odds ratio	95% CI	Odds ratio	95% CI
Age younger than 30 years	Reference
31-40 years	1.22	0.51-2.93	0.84	0.34-2.06
41-50 years	0.82	0.34-1.95	0.57	0.24-1.39
51-60 years	0.88	0.36-2.18	0.55	0.22-1.40
>60 years	1.24	0.46-3.35	0.76	0.28-2.10
Male sex	Reference
Female sex	0.32	0.14-0.75	0.33	0.14-0.77
Race
White or Caucasian	Reference
Black or African American	1.10	0.70-1.74	1.19	0.76-1.89
Othera	0.88	0.36-2.18	0.95	0.39-2.35
Ethnicity
Not Hispanic or Latino	Reference
Hispanic or Latino	2.88	0.95-8.71	2.76	0.92-8.29
Comorbidities
HTN (present vs absent)	1.23	0.77-1.95	1.06	0.67-1.67
HLD (present vs absent)	0.83	0.53-1.29	1.04	0.67-1.61
T2DM (present vs absent)	0.78	0.41-1.47	0.83	0.44-1.56
CVD (present vs absent)	0.79	0.41-1.53	0.98	0.50-1.89
MDD (present vs absent)	0.79	0.51-1.2	0.81	0.53-1.25

^a Asian, American Indian or Alaskan Native, and all other races specified.

CVD = cardiovascular disease; HLD = hyperlipidemia; HTN = hypertension; MDD = major depressive disorder; T2DM = type 2 diabetes.

Weight loss was compared by persistence and adherence status at each time point. In unadjusted analyses of observed weight changes, patients who maintained early persistence (months 1-6) and persistence overall (months 1-12) were observed to experience the greatest weight loss (8.9% at month 6 and 13.8% at month 12 mean weight loss from baseline, respectively), compared with patients who were nonpersistent (3.3% mean weight loss at month 6 and 6.4% mean weight loss at month 12). Early adherence (months 1-6) and adherence overall (months 1-12) also were associated with observed weight loss (9.5% at month 6 and 13% at month 12, respectively). Patients demonstrating early persistence lost an average of 11.7% at month 12, whereas patients demonstrating early adherence lost 12.6% at month 12. Both groups lost more weight than those who did not display early persistence or adherence behaviors, who lost 4.4% and 5.2% mean weight loss at month 12, respectively. Figure 2 displays weight loss results for persistence and adherence variations across measurement periods.

FIGURE 2.

Persistence, Adherence, and Weight Change Across Time

All comparative groups were found to be statistically significant from the nonpersistent/nonadherent groups (P < 0.001; t-test or Mann-Whitney U).

Months 1 to 6 and months 1 to 12 represent cumulative percent weight loss from baseline. Months 7 to 12 represent additional, incremental weight loss achieved during this period.

Nearly all patients (388, 99%) had semaglutide fill/dose patterns that were not aligned with the approved dosing schedule. Among the 5 (1%) patients with patterns indicative of the approved dosing schedule, mean weight loss at month 12 was 12.5%, compared with 7.2% for those with divergent patterns (irrespective of persistence). The most common reason for divergence from the schedule was receiving the wrong dose at month 1 (282, 73% of the 388 total divergent patients). Of these patients, almost half (134, 34% of the total cohort) initiated semaglutide at 0.5 mg, designated as the first therapeutic dose for most patients (ie, able to induce significant weight loss). The remaining patients were fairly evenly spread across the remaining 3 available doses (15% for 1 mg, 11% for 1.7 mg, 12% for 2.4 mg).

Discussion

Real-world use of semaglutide used for obesity treatment among this cohort demonstrated that less than half of all patients were consistently persistent or adherent, and these patients were observed to have decreased weight loss as compared with patients classified as persistent or adherent. Patients who were persistent across all 12 months were observed to have lost an average of 13.8% of their baseline weight, comparable to the 12.4% placebo-adjusted weight loss demonstrated for those receiving semaglutide for obesity in one of its clinical trials with similar follow-up time.²⁰ Persistence rates in our cohort appear to be similar to those of previous studies: 44% across all 12 months in this study and 47% to 50% in a prior study assessing incretins and another study assessing AOMs from a variety of classes.^13,14

In a prior study by Gleason et al evaluating semaglutide among other incretin mimetics, higher real-world adherence to semaglutide was observed (63%) than that seen in our cohort (45% across all 12 months in this study).¹⁴ Their study used commercial pharmacy and medical claims from a large pool of patients, in a population of patients with obesity but without T2DM, and defined PDC as less than or equal to a 60-day gap (while also making allowance for glucagon-like peptide-1 switching).¹⁴ Although these 2 studies are not directly comparable with our study given that semaglutide was the only incretin mimetic evaluated in our study and prior studies evaluated either multiple incretins or AOMs from a variety of classes, the improved adherence shown in other study populations is promising for potential improvement.^13,14

Persistence below 50% suggests that assistance with behavior change may be needed for semaglutide and similar weight loss agents.^13,14 Assistance needs will vary depending on the patient and can include prevention and management of adverse effects, reinforcement of and assistance with comprehensive lifestyle changes, and facilitation of access for these costly agents. Effective, repeated counseling efforts on ways to prevent and manage adverse effects from semaglutide and similar agents may improve tolerability, persistence, and adherence rates. Additionally, AOMs are one “tool in the toolbox” for patients with obesity to use to reach their weight loss goals. When used alone without adequate engagement in comprehensive lifestyle changes, their efficacy can be diminished both in the short and long term.

Early adherence (months 1-6) led to greater weight loss than early persistence when directly compared, while both outperformed “late” or nonearly adherence or persistence (months 7-12). For example, patients demonstrating early persistence behaviors were observed to have a difference of 7.3 percentage points average weight loss at month 12 than those without early persistence behaviors. For early adherence, the difference in weight loss at month 12 was observed to be 7.4 percentage points higher. If programs and benefits were designed to highlight and reinforce such early persistence and/or adherence behaviors, greater weight loss may be observed.

Patients with a gap greater than 60 days between fills and/or a PDC less than 80% were observed to have a markedly low mean weight loss (as low as 6.4% for persistence, 6.6% for adherence). Still, across all 12 months, patients who did not demonstrate persistence or adherence maintained an average weight loss of approximately 6.5%, which is defined as meaningful weight loss according to the FDA approval guidance for AOMs and other clinical guidelines for obesity management.^21,22 These results support the beneficial impact of these agents in providing meaningful weight loss, despite suboptimal use. It is critical for employers and payers to determine whether weight loss of 6.5% across 12 months is substantial enough to justify the financial investment semaglutide requires, or if a more inexpensive AOM would achieve a similar result.

Multivariable logistic regression analyses indicated that female individuals were associated with lower odds of persistence and adherence compared with male individuals; however, this finding should be interpreted cautiously given the majority female composition of the cohort (91%), which may limit generalizability. Having 1 or more comorbid conditions did not impact persistence or adherence to semaglutide, suggesting that individuals may view semaglutide for weight loss as “separate” from their other, chronic conditions. Overall, our finding that persistence and adherence were not associated with age, race, ethnicity, or comorbidities suggests that the underlying drivers are not well understood.

A unique aspect of this study, not addressed in prior work, is evaluation of semaglutide’s titration schedule, of which a minority of participants followed (5 patients, 1%). Given that semaglutide is associated with adverse effects partially mitigated by careful titration during initiation and dose escalation, evaluation of why titration was not followed is warranted to more fully grasp its impacts on adherence behaviors and weight loss results. The number of patients who started at 0.5 mg (134, 34%) or 1 mg (58, 15%) is concerning from a tolerability, persistence, and adherence standpoint. Potential reasons for this trend include a desire by patients for greater initial weight loss produced by higher starting doses of semaglutide and/or a lack of knowledge by prescribing clinicians of the correct starting dose.

For those closely following the approved titration schedule with no more than one 4-week delay, a 12.5% average weight loss at month 12 was observed, almost double that for patients using other titration patterns (7.2% average weight loss at month 12). Benefit plans who build in requirements for prescribing clinicians to adhere to the approved titration schedule may be appropriate to study. To appropriately inform benefit decisions, health entities must evaluate a variety of outcomes, including clinical outcomes examined in this study. Benefit decisions have proved significantly challenging for incretin mimetics, beginning with their use in T2DM, and now for obesity. What began as determinations for incretin mimetic use in roughly 10% of beneficiaries with T2DM has now evolved to determining use in roughly 40% of beneficiaries.^23,24 With such a costly agent class, this expansion in coverage determinations results in a massive financial burden for payers and patients.

A multifaceted approach to understanding the clinical, economical, and ethical factors at play in such determination is needed for optimal benefit design. This study supports understanding of some of the relevant clinical factors (ie, analyzing adherence behaviors and observing weight loss). Questions that remain include whether coverage should be limited to persistent or adherent patients and whether persistence or adherence can be predicted based on baseline clinical or other factors.

LIMITATIONS

Our study’s limitations begin with not adjusting for covariates in our weight loss analyses, such as due to varying demographic and clinical variables in our cohort. As a result, differences in weight loss between persistent and nonpersistent groups should be interpreted with caution. Our study also has potential bias due to time-varying covariates and unmeasured confounding variables. These variables include a patient’s disease severity (eg, class of obesity, number of weight-related complications) and social determinant of health factors (eg, access to recreation facilities, healthy food). Even for patients who had optimal persistence and adherence, confounding variables can affect these outcomes as well as those of weight loss, skewing results negatively or positively. This study took place in a single health system, limiting application of its findings to other institution types. Missing data for height and initial weight limited more complete analysis of BMI differences at each time point for this cohort.

Our study has a relatively short duration of follow-up and would be strengthened by extending follow-up to include a reasonable minimum time for dose titration (eg, 5 months). This study did not evaluate reasons for nonpersistence and nonadherence for any of its outcomes, which limits its ability to inform specific efforts to improve either measure of adherence to semaglutide. It also did not capture access issues related to shortages of semaglutide, which took place during the majority of the data collection period.^25,26 Given that shortage of semaglutide for both T2DM and obesity has recently ended, future studies on this agent should prove to be less impacted by access issues.

Another limitation of this study is its outcome measurements for 12 months after initiation of semaglutide, rather than once a maintenance dose has been reached. In clinical trials of semaglutide, 4 months were allowed to reach either a 1.7-mg or 2.4-mg weekly maintenance dose, after which patients were followed for 12 months.^20,27,28 The approach used in this study was beneficial because an analysis of dosing was allowed. Still, the small number of patients who followed the approved schedule (n = 5) limits full exploration of the real-world factors behind the use of this schedule. This study also did not evaluate overlap in persistence assessments (ie, overfills, possibly to prevent gaps in therapy due to shortages), which may have resulted in underestimation for persistence rates. Early refills were not accounted for, which were allowed by the health system up to 6 days before the scheduled refill date. Lastly, switching to another incretin or AOM was considered nonpersistence, which limits findings regarding this trend in patient care.

Conclusions

Less than half of patients receiving semaglutide for obesity were persistent or adherent, and the recommended titration schedule was not followed for the majority of patients. Patients displaying persistent or adherent behaviors appeared to experience more weight loss than those who did not. Future studies are needed to evaluate clinical outcomes in relation to persistence and adherence to further inform coverage policies.

Disclosures

The authors declare no conflicts of interest.