Prescription caps, sometimes called drug limits, restrict the number of prescriptions a Medicaid beneficiary can fill in a given month. States adopted them widely beginning in the early 2000s as a blunt instrument for controlling pharmacy spending, and they remain in force in a number of states today. The logic behind them is straightforward: limit the volume of reimbursed prescriptions and the state saves money. But decades of research, dating back to landmark studies in the late 1980s on Medicaid payment restrictions, have shown that these caps can disrupt therapy for chronic conditions, sometimes shifting costs elsewhere in the health system through increased hospitalizations and emergency care. For people with HIV, the stakes are particularly high, because consistent antiretroviral therapy is what keeps the virus suppressed, protects long-term health, and prevents onward transmission.

What made the South Carolina case scientifically valuable was the timing and the specificity of the policy change. On July 1, 2017, the state removed its Medicaid prescription cap outright, creating a natural experiment. The research team compared Medicaid beneficiaries with diagnosed HIV in South Carolina against similar beneficiaries in nine control states that maintained their caps throughout the 2016 to 2019 study window. To make the comparison as clean as possible, the investigators used propensity score matching to pair individuals in the policy-change state with statistically comparable people in the control states, then applied a difference-in-differences framework that nets out baseline differences between the groups and shared temporal trends. An event-study design allowed them to trace how the effect evolved month by month after the cap came off.

The study population was carefully defined. The sample included adults aged 21 to 64 years with diagnosed HIV and continuous Medicaid claims spanning both the pre-policy and post-policy periods. Beneficiaries who moved between states, lived in long-term care facilities, or were pregnant were excluded, since each of these circumstances could confound the relationship between policy and medication use. After matching, the analytic sample comprised 1,135 individuals in both South Carolina and the control states, drawn from the Transformed Medicaid Statistical Information System analytic files, the federal data infrastructure that underpins much of contemporary Medicaid research.

The headline result concerns adherence, measured as the proportion of days covered, or PDC, a standard metric that captures the fraction of a period during which a patient has medication on hand. In the post-policy period, the change in PDC was 4.6 percentage points greater in South Carolina than in the control states, with a 95 percent confidence interval of 2.7 to 6.6 percentage points. In practical terms, beneficiaries whose therapy had been constrained by the cap filled their antiretroviral prescriptions more reliably once the limit disappeared, closing a meaningful fraction of the adherence gap that caps are believed to create. Adherence of this kind is not an abstract quality measure; prior work has linked higher PDC to achieving viral suppression, the clinical goal that underpins both individual health and the public health benefit of HIV treatment.

The event-study analysis added temporal texture to the finding. The adherence improvement did not appear uniformly across the follow-up period. Instead, it peaked at 8.4 percentage points roughly nine months after the cap was removed, suggesting that beneficiaries needed time to re-establish their refill patterns, work through accumulated prescription needs, and coordinate care with providers before the full effect materialized. Notably, the effect was driven by individuals whose baseline prescription fills had been below the cap, the group for whom the policy constraint was most plausibly binding. That pattern is exactly what one would expect if the cap itself, rather than some unrelated change in South Carolina, was responsible for suppressed medication use.

Alongside adherence, the study tracked prescription fills and days’ supply, both overall and for HIV-related medications specifically. Cap removal was accompanied by increases in HIV-related prescription fills and days’ supply, consistent with beneficiaries obtaining more of the antiretroviral regimens and supporting medications they needed. These utilization shifts represent the direct mechanical consequence of lifting a quantity restriction: when the ceiling rises, fills that would have been deferred, split, or foregone become possible. For a population that often manages multiple medications, including treatments for comorbid conditions alongside antiretroviral therapy, the removal of a global prescription limit can unblock access across an entire regimen, not just a single drug.

Perhaps the most sobering finding is what did not change. The researchers examined inpatient admissions, outpatient evaluation and management visits, and emergency department visits, and found no consistent changes in health care utilization following cap removal. This null result matters for the policy debate. A common argument for lifting drug caps is that improved medication access will pay for itself by reducing downstream acute care. In this study, at least over the observed window, that offset did not materialize in measurable utilization shifts. The authors are careful in their interpretation: removing the cap improved HIV-related prescription fills, days’ supply, and ART adherence, but achieving optimal adherence and improvements in health outcomes and health care utilization may require additional strategies beyond the policy change alone.

That interpretation aligns with a broader clinical reality. Adherence to antiretroviral therapy is shaped by far more than prescription availability. Housing instability, substance use, mental health conditions, transportation barriers, stigma, and the complexity of modern polypharmacy all influence whether people take their medications as prescribed. Real-world studies of adherence among people with HIV in the United States have repeatedly documented that a substantial share of patients fall short of the levels associated with reliable viral suppression, even in systems without fill limits. A cap removal lowers one structural barrier, and the 4.6 percentage point adherence gain shows that the barrier was real, but it does not address the others. The study’s own framing suggests that policy makers who want viral suppression and reduced acute care use should pair pharmacy benefit reform with adherence support services, case management, and interventions targeting the social determinants that drive missed refills.

The findings also carry weight for the many states that still maintain prescription caps. Prior research has associated state drug caps with fewer Medicaid-covered prescriptions for opioid use disorder treatment, reduced access to medications among young adults with disabilities, and disrupted therapy for schizophrenia and depression, building a consistent cross-disease picture of caps as policies with clinical costs that are easy to overlook in budget negotiations. Because Medicaid is the largest source of coverage for people with HIV in the United States, the population studied here sits at the intersection of two policy domains, disability and infectious disease, where fill limits can have outsized consequences. The South Carolina experience, analyzed with a rigorous quasi-experimental design and funded by the National Institutes of Health, provides a template for evaluating similar reforms elsewhere. As states continue to weigh pharmacy cost containment against access, the evidence from this natural experiment suggests that lifting caps can meaningfully improve medication-taking for people living with HIV, even if the full promise of that improvement, in viral suppression and avoided hospital care, depends on supports that go beyond the pharmacy counter.

Subject of Research: Effects of Medicaid prescription cap removal on antiretroviral adherence and health care use among people with HIV

Article Title: Medication Adherence and Health Care Use After Medicaid Prescription Cap Removal Among People with HIV: a Difference-in-Differences Study

Article References: Geng, F., Lu, Y., Santostefano, C. M., Lu, X., Lee, Y., Dow, P. M., Shireman, T. I., & Galarraga, O. (2026). Medication Adherence and Health Care Use After Medicaid Prescription Cap Removal Among People with HIV: a Difference-in-Differences Study. Journal of General Internal Medicine. https://doi.org/10.1007/s11606-026-10794-9

Image Credits: AI Generated

DOI: 10.1007/s11606-026-10794-9

Keywords: Medicaid, prescription caps, HIV, antiretroviral therapy, medication adherence, difference-in-differences, health policy, South Carolina, health care utilization, proportion of days covered, pharmacy benefit, viral suppression

The study, conducted by researchers at the University of Mississippi School of Pharmacy in partnership with the Mississippi Division of Medicaid and published in Advances in Therapy, analyzed deidentified administrative claims from January 2023 through December 2025. The team identified 8,495 GLP-1 receptor agonist users overall and narrowed the cohort to 3,968 adults who initiated therapy between July 2023 and June 2025, each with six months of continuous enrollment before and after starting the medication. The population was strikingly homogeneous: 92.52 percent were female, 50.28 percent were Black, and nearly 62 percent were younger adults aged 18 to 39. Hypertension and depression were the most common coexisting conditions, and most members were enrolled in managed care organizations rather than fee-for-service plans.

The headline numbers tell a story of steady erosion. Mean adherence, measured by the proportion of days covered, a standard metric that divides the total days a patient has medication on hand by the total days in the observation window, stood at 0.76 at three months but fell to 0.65 by six months. The proportion of members meeting the conventional adherence threshold of at least 80 percent of days covered dropped from 56.85 percent at three months to 38.16 percent at six months. Persistence, defined as continuous therapy without a gap of 60 days or more, followed the same downward slope, falling from 72.98 percent to 51.03 percent over the same interval. In the subgroup followed for a full year, mean adherence sank to 0.54, with only 25.26 percent adherent and 36.67 percent persistent.

What makes this study particularly valuable is its timing. Most prior real-world analyses of GLP-1 adherence were conducted during the nationwide drug shortage that ran from 2022 through February 2025, a period when supply constraints made it nearly impossible to distinguish voluntary discontinuation from involuntary interruption. Because this analysis extends through December 2025, it captures one of the first pictures of adherence under conditions of stable drug availability. The researchers found that adherence improved progressively across initiation cohorts: members who started therapy in the first half of 2025 consistently outperformed those who began in 2024, who in turn fared better than the 2023 pioneers. Among 2023 initiators, only 47.77 percent were adherent at three months, compared with 61.01 percent of 2025 initiators.

To move beyond aggregate statistics, which can mask enormous individual variation, the team employed group-based trajectory modeling, a statistical technique that clusters individuals who follow similar longitudinal patterns of behavior. The model input was the continuous proportion of days covered calculated for each 30-day interval, and the optimal five-group quadratic solution was selected using the Bayesian Information Criterion, the logged Bayes factor, a minimum group membership threshold of five percent, and clinical interpretability. The result was a taxonomy of medication-taking behavior that is far more nuanced than a single adherence rate could ever convey.

The largest group, comprising 52.27 percent of members, was labeled consistent adherent, maintaining high medication coverage throughout follow-up. The remaining members scattered across four suboptimal pathways. The delayed decliner group, 13.53 percent of the cohort, held steady for roughly three months before sliding. The consistent decliner group, 12.85 percent, showed a steady reduction from the outset. The fluctuating adherent group, 11.14 percent, cycled through declines and partial recoveries, a pattern that may reflect intermittent access barriers or wavering motivation. Most concerning was the rapid discontinuer group, 10.21 percent, who abandoned therapy by the second month and never returned, a trajectory often associated with early adverse effects such as nausea and gastrointestinal distress, or with immediate obstacles to refilling prescriptions.

Multinomial logistic regression, using the consistent adherent group as the reference, illuminated who was most at risk of falling off course. Members who initiated therapy in July through December 2023 had more than twice the odds of becoming rapid discontinuers compared with 2025 initiators, an odds ratio of 2.10, likely reflecting the chaos of the shortage era. Age mattered profoundly: members aged 40 to 64 had significantly lower odds of suboptimal trajectory membership than those aged 18 to 39, perhaps because younger patients perceive less disease severity, have shallower roots in the healthcare system, and face fewer obesity-related complications that motivate sustained treatment. Black members had higher odds of belonging to the rapid discontinuer and fluctuating adherent groups than white members, a disparity the authors attribute to structural and socioeconomic barriers rather than race itself. Enrollment in a managed care organization was protective, cutting the odds of rapid discontinuation by roughly a third, presumably because care coordination, medication management programs, and proactive outreach help keep patients engaged in ways that fragmented fee-for-service coverage cannot.

Clinical characteristics painted a more complicated picture. Type 2 diabetes and cardiovascular comorbidities were associated with higher odds of suboptimal trajectories, possibly reflecting greater clinical complexity and competing therapeutic priorities that crowd out obesity treatment. Hypertension, paradoxically, was protective, reducing the odds of delayed decliner membership. Sex showed no significant association, a departure from prior studies, though the authors note their cohort was so overwhelmingly female, reaching 93 percent of initiators in early 2025, that detecting sex differences may have been statistically impractical. Sleep apnea, fatty liver disease, hyperlipidemia, kidney conditions, and depression were not significantly linked to trajectory membership.

The policy implications are considerable. As of January 2026, only 13 state Medicaid programs covered at least one GLP-1 receptor agonist approved for weight management, and several have since withdrawn or restricted coverage. Coverage alone, the Mississippi experience suggests, is insufficient to guarantee therapeutic engagement. Fewer than five percent of initiators in the program received counseling services before starting therapy, a strikingly low uptake given that discontinuation is consistently associated with rapid weight regain and loss of cardiometabolic benefit. The authors argue for expanded care management, patient education, behavioral counseling, and side effect support, alongside robust lifestyle programs that can mitigate regain among those who stop the drugs. The temporal improvement across cohorts also offers a hopeful signal: fixing structural access barriers, as the shortage resolution did, produces measurable gains in continuity. For the other states weighing coverage decisions, Mississippi’s data provide a rare, unvarnished look at what happens when expensive, effective medications meet a high-need population, and a reminder that the hardest part of obesity pharmacotherapy may begin not at the first injection, but at the second and every one after.

Subject of Research: Real-world adherence and persistence to GLP-1 receptor agonist anti-obesity medications among Medicaid enrollees

Article Title: Real-World Adherence Patterns and Predictors of GLP-1 Receptor Agonist Use for Obesity Management in a Medicaid Population

Article References: Bazzazzadehgan, S., Pittman, E., Lin, L.-Y., Qi, J., Ly-Ha, A., Smith, D., Kirby, T., Vest, T. A., Bentley, J. P., & Bhattacharya, K. (2026). Real-World Adherence Patterns and Predictors of GLP-1 Receptor Agonist Use for Obesity Management in a Medicaid Population. Advances in Therapy. https://doi.org/10.1007/s12325-026-03806-x

Image Credits: AI Generated

DOI: 10.1007/s12325-026-03806-x

Keywords: GLP-1 receptor agonists, anti-obesity medications, Medicaid, medication adherence, persistence, group-based trajectory modeling, obesity, health disparities, Mississippi, drug shortage, proportion of days covered, health policy