GLP-1 treatment reduces sickness absence

GLP-1 drugs have transformed the treatment of type 2 diabetes and obesity. As governments and insurers debate whether and how to cover GLP-1 drugs, much of the discussion focuses on health spending. But obesity and diabetes also impose costs through the labour market. Using linked Danish register data, this column examines whether people are more likely to work after beginning treatment and spend less time away because of illness. The findings reveal that GLP-1 treatment reduces long-term sickness leave, as well as emergency department visits and cardiovascular medication use.

US gross spending on GLP-1 receptor agonists such as Ozempic and Wegovy reached $27.5 billion in Medicare and $8.6 billion in Medicaid in 2024, even with restrictions on coverage for obesity (KFF 2026). These figures capture only part of a larger fiscal question facing governments, insurers, and employers worldwide: these drugs have transformed the treatment of type 2 diabetes and obesity, but their high cost and rapid uptake have made coverage decisions consequential well beyond the health budget. In Europe, the debate forms part of a wider challenge over how to price and procure innovative medicines while preserving access and incentives for innovation (Dubois 2026).

A central issue in these debates is whether the benefits of GLP-1 drugs extend beyond clinical outcomes and medical spending. Obesity and type 2 diabetes impose costs well outside the health system. They are associated with lower employment, reduced productivity, and more sickness absence (OECD 2019, Sawers and Dobbs 2014). The estimated annual cost of treating obesity-related diseases is $425 billion (US dollar purchasing power parity) across OECD, G20, and EU28 countries, while the economic cost of diagnosed diabetes in the US reached $413 billion in 2022 (OECD 2019, Parker et al. 2024). If GLP-1 treatment improves people’s capacity to work or reduces absenteeism, evaluations based only on pharmacy and medical spending will miss part of its value (Hwang et al. 2025, Bock et al. 2026).

Using treatment timing to identify labour market effects

In new research, we bring evidence to this debate using population-wide Danish administrative data (Daysal et al. 2026). Denmark is a useful setting because Ozempic was introduced relatively early and administrative registers link prescription histories to medically certified sickness leave, healthcare use, income, and labour market status over several years. This allows us to examine not only whether people are more likely to work after beginning treatment, but also whether those who are already working spend less time away because of illness.

We compare people who started Ozempic during its first 17 months in Denmark with similar people who began treatment four years later. Everyone in the study eventually initiates treatment, but the later starters provide a comparison group for what would likely have happened to the early starters before they themselves received a GLP-1 drug. We match the groups on age, sex, diabetes and obesity status, income, education, region, marital status, and prior healthcare use. The resulting sample contains 7,011 early initiators and 7,011 comparable later initiators. 

Treatment is persistent in Denmark

Figure 1 shows that exposure to GLP-1 treatment remains high among early Danish Ozempic initiators. Four years after initiation, 75% are still filling prescriptions for a GLP-1 drug. This persistence matters because many potential economic effects are unlikely to appear immediately. Improvements in cardiometabolic health and work capacity may take time to translate into fewer prolonged absences.

Figure 1 Persistence in GLP-1 treatment after Ozempic initiation

Note: The figure reports the estimated difference in semaglutide use between early Ozempic initiators and matched patients who initiate four years later.
Source: Daysal et al. (2026).

Persistence is also relevant to coverage policy. Among commercially insured patients in the US, estimated adherence falls to roughly 40%–60% by the end of the first year (Wing et al. 2026), well below the 75% four-year adherence we document in Denmark. Prior work often attributes discontinuation to side effects, but our results suggest that prescription drug costs may be a further explanation. If affordability shapes whether patients remain on treatment as well as whether they start it, expanding coverage may be necessary for GLP-1 drugs to deliver their full long-run labour market benefits.

Sickness absence falls gradually

Our main finding is that Ozempic initiation reduces medically certified long-term sickness leave, defined as spells lasting more than 30 days. Averaged over four years, treatment lowers long-term sickness leave by 0.95 percentage points, or 17.3% relative to the pre-treatment mean. Figure 2 shows that the reduction develops gradually. It is about 0.8 percentage points in the first two years and around 1.1 percentage points in years three and four.

Figure 2 Effects of Ozempic initiation on long-term sickness leave

a) Share of months in a quarter with any sickness spell lasting more than 30 days

b) Municipality-paid sickness benefits after the first 30 days

Note: Shaded areas are 95% confidence intervals.
Source: Daysal et al. (2026).

The decline in sickness absence is accompanied by reductions in emergency department visits and cardiovascular medication use. This pattern is consistent with randomised trial evidence that semaglutide lowers the incidence of major adverse cardiovascular events (Marso et al. 2016), events that plausibly generate prolonged work absences. 

Better attendance does not lead to higher employment or income

Despite the sizeable reduction in sickness leave, we find no meaningful effects on employment, labour force participation, or total income over four years. The estimates are sufficiently precise to rule out employment gains larger than 0.4 percentage points and income gains larger than 0.7%. The results are similar across groups defined by sex, age, and education.

Why does better attendance not translate into higher measured income? The Danish institutional setting provides part of the answer. Employers cover the initial period of sickness absence, while municipalities take over for longer spells. For many workers, sickness benefits replace most or all lost earnings. A reduction in absence therefore need not raise the worker’s annual income. Instead, it lowers the cost borne by employers and the public sector.

We estimate that reduced sickness absence generates combined employer and public savings of DKK 4,766 per employed treated individual per year, equivalent to $722 and to 1.3%–1.5% of annual labour income among employed Ozempic initiators.

These savings are not large enough for GLP-1 drugs to pay for themselves through reduced sickness absence at current prices. But they provide a meaningful offset to the substantial cost of GLP-1 treatment, one that would be missed entirely by an accounting exercise focused only on drug costs and healthcare spending.

Institutions shape where the gains appear

The broader lesson is that the economic value of medical innovation depends on institutions. In a country with less generous sickness insurance, similar improvements in health could appear as higher earnings, more hours worked, less presenteeism, or better job retention. In Denmark, where income losses during illness are partly insured, the clearest labour market gain is less time spent too sick to work and lower spending by employers and municipalities.

Early Danish Ozempic users were mostly patients with a clear medical need: 84% had evidence of diabetes before treatment. Our estimates are therefore best interpreted as bounding the labour market gains among patients with a clear clinical indication, rather than the marginal patient who might initiate treatment under a broader coverage expansion. Effects may differ for younger populations, for people using GLP-1 drugs mainly for weight management, or in settings where excess weight creates different barriers to work. Previous research, for example, finds an obesity-related wage penalty that is particularly pronounced for women (Dolado and Minale 2023), while recent US evidence suggests that GLP-1-induced weight loss increases employment among women who were initially out of work (Diamond 2026). These margins may matter less in our Danish sample, or could take a different form.

As GLP-1 use expands, coverage decisions will hinge partly on which benefits enter the calculus alongside the cost of the drugs themselves. Our evidence indicates that labour market benefits are part of the answer, although not through higher employment or earnings. In Denmark, the relevant margin is a reduction in sickness absence, worth an estimated 1.3%–1.5% of annual labour income per employed patient. Coverage assessments that ignore this channel will understate the value of GLP-1 treatment.

Source: cepr.org