Current labeling supports only short-term (≤3 month (mo)) use of phentermine due to limited research & safety concerns. We studied whether longer-term, off-label users had more weight loss or greater cardiovascular disease (CVD) risk than on-label users


Using electronic health record data from 8 U.S. health systems, we identified 13,972 adults 18-65 years old with a first phentermine fill between 2010-2015. We created the following time-varying exposure categories according to duration and pattern of phentermine use: On-Label (single episode ≤3mo; referent); Intermittent on-label (multiple episodes, none >3mo); Intermittent off-label (multiple episodes, ≥3mo); Medium-Continuous (single episode >3mo to1yr) and Long-term-continuous (single episode >1yr). We used multivariable linear models to compare % weight loss (%WL) across groups at 6,12&24 mo, and Cox proportional hazards models with time-varying exposure to compare risk of composite incident CVD or death, up to 3yrs after starting phentermine. Covariates included race, gender, age, baseline BMI, diabetes, hypertension, and smoking


Our cohort was 84% female, 45% white, with mean(sd) baseline age 43.5(11)yrs and BMI 37.8(7)kg/m2. Weight change differed significantly between groups at all points in follow-up (p<0.001), with modeled % WL relative to on-label users as follows: Intermittent On-Label: -2.2% at 6mo, -2.0% at 12mo, and -0.5% at 24mo ; Intermittent off-label: -4.7% at 6mo, -6.0% at 12mo, and -3.4% at 24mo ; Medium-Cont: -5.5% at 6mo, -4.7% at 12mo, and -1.5% at 24mo; and Long-term-Cont: -7.5% at 24mo. Overall incidence of the composite CVD/Death outcome was low at 0.6% (n= 84 events), with no significant difference in hazard ratios between exposure groups


We observed greater weight loss without increased risk of incident CVD or death in patients using phentermine for longer than the indicated 3mo. A clinical trial with longer follow-up could inform whether long-term phentermine use is safe and effective