Medication Use Among U.S. Adults After Bariatric Surgery: A Population-Based Analysis of NHANES 2015–2018

Bariatric surgery for patients with morbid obesity results in greater improvements in weight loss and type 2 diabetes outcomes compared with nonsurgical interventions (1). However, individuals with prior morbid obesity who had bariatric surgery can have altered drug absorption, distribution, and metabolism (2). The cumulative effect of bariatric surgery on medication use patterns has not been systematically investigated to improve postsurgery patient care. Using National Health and Nutrition Examination Survey (NHANES) data from 2015 to 2018, we identified and compared medication use in adults who had bariatric surgery with their former peers who were morbidly obese but did not have bariatric surgery.

This study included adults aged ≥18 years who responded to bariatric surgery questions during the NHANES 2015–2018 cycles and excluded participants with missing values for self-reported height and weight. The BMI of each subject was calculated by weight in kilograms divided by the square of height in meters. The subpopulations of our interest were 1) participants who had BMI ≥35 and did not have bariatric surgery (MO w/o BS) and 2) participants who had bariatric surgery, regardless of their BMI (BS). Demographic characteristics for both groups were analyzed. Subjects who had not taken prescription medications and who had missing values for generic drug names in both groups were further excluded. Estimates were weighted to account for the unequal probabilities of selection resulting from the complex sample design, nonresponse, and planned oversampling of selected populations. The weighted percentages of subjects on single drugs and drug categories were calculated and ranked. All analyses were conducted with R (version 4.0.2) and RStudio (version 1.3.959). This study, using a nationally representative sample, was exempt from institutional review board examination because we used deidentified publicly available data.

The BS group had 131 sampled participants, representing an estimated 1.5% of the U.S. population, or over 3 million adults. The MO w/o BS group had 1,637 sampled participants, representing an estimated 14.4% of the U.S. population, or over 33 million adults. Compared with the MO w/o BS group, subjects in the BS group were more likely to be older, female, of non-Hispanic White ethnicity, and to have a college degree or above and a higher annual household income. A Student t test showed the BMIs of BS subjects were significantly lower than those of MO w/o BS subjects (t₁₇₆₆ = 11.35, P < 0.05).

The percentages of subjects who had not taken prescription medications in the past 30 days at the time of the survey were comparable between the two groups (29.8% of BS subjects vs. 32.9% of MO w/o BS subjects; χ² = 0.5, P = 0.46). After excluding subjects with missing values for medications, medication use was analyzed for 89 subjects in the BS group (corresponding to 2.4 million U.S. adults) and 1,082 subjects in the MO w/o BS group (corresponding to 22.2 million U.S. adults). For those who had taken medications, the average number of prescription drugs taken per subject in the BS group is 5.2, compared with 4.6 in the MO w/o BS group (t₁₁₆₉ = 1.66, P = 0.10).

As shown in Fig. 1A, most of the top used drugs in the BS group were used at lower rates than those in the MO w/o BS group, including metformin and lisinopril. Compared with MO w/o BS subjects, BS subjects took approximately half the drugs prescribed for essential hypertension and type 2 diabetes, while they took more drugs to treat anxiety, neuralgia, neuritis, depression, and lower back pain (Fig. 1B). This is in contrast to some current evidence that suggests that bariatric surgery can reduce anxiety, depression, and lower back pain (3,4). Our finding on neurologic complications postbariatric surgery is in agreement with recent reports (5).

A small sample study and inability to evaluate medication use before and after BS limit our findings. There were no data for the specific type of bariatric surgery performed or amount of weight lost. Furthermore, individuals qualified for the surgery may not have it performed due to lack of health insurance coverage, untreated comorbidities, or psychiatric issues, which might introduce confounders into this analysis.

In summary, the numbers of medications used by both groups were comparable, while the indications differed. This suggests that the medication needs of patients post-BS shift over time, and different medications are needed to address potential consequences of bariatric surgery (e.g., neurologic/psychiatric). Future work is needed to analyze electronic health record data of larger numbers of samples of individuals who had bariatric surgery with a prepost study design to confirm findings in this study.