How are weight-loss drugs changing our food shopping?

American households with at least one user of GLP-1 weight-loss medications reduce their spending on groceries by 5.3% in the first six months of starting the treatment, with the biggest declines in snacks and sweets. But buying habits largely return to normal once consumers stop taking the drugs.

The growing popularity of GLP-1 weight-loss medications like Mounjaro and Ozempic is already changing people’s food shopping baskets. According to research that analyses data on US households, those with at least one GLP-1 user reduce their grocery spending by 5.3% in the first six months after adoption – equivalent to about $390 (just under £300) a year, or roughly three weeks’ worth of groceries (Hristakeva et al, 2025).

The same study shows that spending at fast food restaurants, coffee shops and other establishments is also affected, falling by 8% among GLP-1 users. These reductions are broad and happen across a wide range of categories, but they are largest for savoury snacks, sweet bakery products and biscuits.

Approximately 34% of adopting households in the study’s sample discontinue GLP-1 use during the period of observation. Among those households in which people stop taking the medication, the average duration of use is six months, and after they discontinue, their grocery spending tends to returns close to its earlier level.

These patterns are important to know for retailers, restaurants and food producers, although they do not yet tell us how diets or spending will change over the longer run.

Who is taking GLP-1 medications?

GLP-1-based medications include semaglutide, sold as Ozempic and Wegovy, and tirzepatide, sold as Mounjaro and, in the United States, Zepbound. They were developed or first approved to treat type 2 diabetes, and some brands are now approved specifically for weight management. These medications regulate blood sugar and reduce appetite, with users commonly reporting weaker cravings and feeling full sooner.

From October 2023 to July 2024, the fraction of surveyed households that include at least one person using a GLP-1 medication increased from 11.5% to 16.3%, reaching approximately one in six households and 8.3% of the total US population. About half of these households reported that the user was taking the drug for weight loss.

Compared with those using the drugs to manage diabetes, users for weight loss were younger, had higher incomes and were more likely to be paying out-of-pocket for the drug. In a follow-up survey, users reported losing an average of 21kg (although this was self-reported).

Wider adoption of GLP-1-based treatments could have a substantial health benefit. In 2021-23, over 40% of adults (aged 20+) in the United States were living with obesity (CDC: Centers for Disease Control and Prevention, 2024).

In England, NHS access to tirzepatide for weight management began rolling out through primary care from June 2025. Initially, the rollout is prioritising patients with the highest clinical need (NHS England, 2026). The gradual expansion in NHS access is likely to emerge first among these priority groups and grow as access and uptake increase. At the same time, GLP-1 use outside the NHS is already substantial. So, the rollout does not capture the whole British market (UCL, 2026).

What happens to the weekly shop?

In our research, we link four waves of survey data on GLP-1 use to verified transactions from a panel of about 150,000 US households. Our main analysis follows 2,602 adopting households. Each GLP-1 household is matched to a non-adopting household with similar demographics, diabetes status and pre-adoption shopping patterns. The analysis compares how spending changes across the two groups.

We find that grocery spending falls by 5.3% during the first six months after adoption and remains about 4% lower in months 7-12. Item counts fall by a similar amount, which suggests that households mainly buy fewer items rather than simply substituting towards cheaper products.

Although the average is informative, there are also meaningful differences in spending patterns across household income levels. Households that earn over $125,000 (about £93,000) per year cut their grocery spending by 8.2% – equivalent to about $700 (£520) per year. For those in lower income brackets, grocery spending falls by 3.9% – equivalent to about $260 (£195) per year.

The motivation for uptake also matters. People using GLP-1s for weight loss reduce their grocery spending by 5.6%, while those using them for diabetes cut spending by 4.6% (although this difference is not statistically different from one another).

Grocery spending declines gradually after GLP-1 adoption

Figure 1. Monthly changes in grocery spending relative to the period before adoption

Rectangles show ±1 standard error and whiskers show 95% confidence intervals.
Source: Hristakeva et al (2025).

People’s food habits are often hard to change substantially. For example, while the calories purchased by households initially decrease after a diabetes diagnosis, the changes in the average household are quite small and diminish over time (Oster, 2018). Similarly, the quality of diets changes little on average after individuals experience a diagnosis of a disease, the birth of a child, a job loss or retirement (Hut and Oster, 2022).

Even so, policies that change the price of specific food products (such as a sugar tax) can change purchases of these products (Allcott et al, 2019). Posting calorie information at Starbucks decreased calories per transaction by 6%, with the reduction persisting throughout the ten-month follow-up (Bollinger et al, 2011). Since GLP-1s target appetite, they may affect purchases across a wide range of foods, although these calorie and spending estimates are not directly comparable.

Which foods change most?

The largest decreases among the top categories are for crisps and other savoury snacks – down by more than 10% in the first half year. This is followed by sweet bakery products (down by 8.8%) and biscuits (6.5%). The decreases also apply to categories such as meat, eggs, bread and fresh vegetables, suggesting a broad reduction in food purchases.

Only four of the 40 largest food categories have positive estimated effects: nutrition bars, fresh fruit, meat snacks and yogurt. The estimated increase for yogurt is 3.6% – the only statistically significant positive estimate.

Overall, our results show that GLP-1 households are buying fewer items. Any evidence that the remaining composition of their baskets is significantly healthier is less clear.

Do the changes last?

Households reduce spending less in months 7-12 than in months 1-6 (from 5.3% to approximately 4%) compared with their baseline prior to taking GLP-1 weight-loss medications. The spending effect among those using the drug for diabetes appears roughly unchanged, while the effect among those using the drug for weight loss is smaller and indistinguishable from zero.

One important factor is that a large share of users in the survey stops taking the drug. In the household data, about one-third of adopters stop using the drug during the observation period. Other research shows that about 37% of people stop taking GLP-1s within a year of starting, with the rate of discontinuation higher among people with obesity but without diabetes (about 50%).

For people who stop using the drug in the purchasing data, total grocery spending rebounds back to almost the original level within six months of discontinuation. On a related note, a randomised withdrawal trial finds significant weight regain among participants who switched from tirzepatide to a placebo (Aronne et al, 2024).

At the same time, the composition of the basket does not fully revert to its original form after discontinuation. Relative to spending on sweets and chocolate prior to adoption, spending on these items is about 7% higher in months 7-12 after adoption, and over 11% higher after discontinuation of the drug (though this should be interpreted with caution given the number of hypotheses tested).

Similarly, the overall healthiness of the grocery basket decreases by approximately 1.6 percentile points after discontinuation of the drug. In short, these results suggest that households may eat less healthy foods compared with their pre-GLP-1 habits after they stop using GLP-1s.

How category spending changes across three stages

Figure 2. Changes in monthly category spending relative to matched non-adopting households and the pre-adoption period.

Asterisks indicate estimates that are statistically significant at the 5% level. Values are calculated from Web Appendix Table E2.
Source: Hristakeva et al (2025).

What does this mean for the food industry?

Extrapolating these results can help us to predict wider sector effects. If we apply the 5.3% figure to the 16.3% of households in our July 2024 survey that reported using GLP-1s, this would imply a reduction of around 0.9% in total US grocery spending – amounting to around $9 billion (about £6.7 billion).

A similar calculation for limited-service restaurants suggests another $7 billion (£5.2 billion) of reduced spending, using 2024 market spending totals from USDA ERS.

Of course, these are back-of-the-envelope calculations, and certainly not a forecast: we assume that the estimates at the household level are the same at the national level and that the percentage reduction in spending of adopting households remains the same over time. An important consideration here is that adopting households tend to be higher-income and spend around 20% more on groceries than the average household does pre-adoption, so their share of grocery spending is larger than their share of households.

That being said, our estimates are broadly in line with what some in the industry have been saying and with other evidence that exists. In October 2023, Walmart’s US chief executive reported ‘a slight pullback in overall basket’ among GLP-1 users (Bloomberg, 2023).

Looking ahead, Morgan Stanley projects that there will be 55 million GLP-1 users in the United States by 2035 – roughly equivalent to the entire population of England. Such a figure could reduce total US calorie consumption by 1.6%, a modest amount in part because of assumptions about discontinuation and weight regain.

Separate analysis shows that half of GLP-1 users report cutting their calorie consumption by 20% or more (Food Business News, 2026). A different report estimates that increased use of GLP-1s could reduce the growth of the US snack market by up to $12 billion (£9 billion) over a decade (EY-Parthenon, 2025). Indeed, several snack manufacturers have closed plants in recent years amid weak volumes: GLP-1s may add to these pressures, but the evidence does not yet establish their contribution to these closures (Bakery and Snacks, 2026).

The continued growth of GLP-1 use won’t affect all food producers equally: a firm selling sweet bakery products or cheese, for example, is more exposed to category-level declines than a more diversified company selling across the entire grocery basket. Conversely, yogurt producers may benefit from it being the only category to show a statistically significant increase. It’s worth emphasising again that these differences reflect exposure to different category-level changes, rather than being estimates of the effect on individual firms.

Which categories combine falling spending with specialised producers?

Figure 3. Estimated change in category spending during the first six months after adoption, plotted against the average share of parent-company US grocery revenue that comes from that category.

Larger circles are categories with a larger share of grocery spending; labelled categories are discussed in the text.
Source: Hristakeva et al (2025).

We’ve already seen some companies adapting to these changes. For example, Nestlé has launched a range of meals designed to match the portion sizes and protein requirements of GLP-1 users, although they are designed to appeal to anyone looking to manage their weight (Nestlé, 2024).

Across the sector, food manufacturers and restaurant chains are increasing the protein and fibre content of their products, as well as looking at smaller portion sizes and less sugary drinks (CNBC, 2026). Whether these innovations represent an expansion in the market for these firms or just a reshuffling of spending is unclear at this point.

A key question for the long-term impact of these drugs on the food market is likely to be their uptake, which in turn will be driven by their cost. Recent US evidence suggests that GLP-1s have not paid for themselves so far through savings on other healthcare spending: a study using commercial insurance claims data from 2016 to 2023 finds no reduction in other medical spending over five years after someone started taking a GLP-1. In fact, other spending increased (Wing et al, 2026).

Much of this evidence predates widespread use of the drugs for obesity. In 2024, the Congressional Budget Office (CBO) estimated that an illustrative policy extending Medicare coverage to anti-obesity medicines would increase net federal spending by around $36 billion (£27 billion) between 2026 and 2034 (CBO, 2024).

Some of the benefits may be felt elsewhere. A study using Danish registry data finds that people who begin taking semaglutide have 17% lower rates of long-term sickness absence, with estimated savings for employers and the public sector of around 1.4% of annual labour income for employed individuals (Daysal et al, 2026).

In the United States, the ‘Medicare GLP-1 Bridge’ began in July 2026, providing access to certain GLP-1 drugs for eligible beneficiaries at a $50 monthly co-payment. This is a temporary programme with specific eligibility criteria, and not a general Medicare coverage for everyone with obesity (CMS, 2026).

In England, meanwhile, the National Institute for Health and Care Excellence (NICE) estimates that 3.4 million people meet its criteria for tirzepatide, although access is being phased in for around 220,000 people in the first three years. The guidance allows a maximum implementation period of 12 years, with a review within the first three years that could shorten it (NICE, 2024; House of Commons Library, 2025).

How reliable is the evidence?

Household-level transaction data are a strong feature of our research, although purchases differ from consumption. We look at household purchases while medication is taken at an individual level. It is possible that food may be shared differently within the household once a household member starts taking GLP-1s.

Second, as with many questions in economics, there is a selection issue: individuals choose whether and when to take GLP-1s. We are able to allay some of these concerns by showing that we can match households based on observed characteristics, that pre-adoption trends are similar across groups and that the results are robust to different estimation methods. But we are not able to rule out changes in unobserved factors, such as motivation to lose weight, that affect both the decision to start medication and food purchases.

Effects such as weight loss and weight regain if the drug is stopped are shown in randomised controlled trials (Jastreboff et al, 2022; Aronne et al, 2024). Consistent with this medical evidence, our estimates of the effect on household purchases are likely to reflect, at least in part, a causal effect of the medication. But strictly speaking, we cannot fully disentangle this effect from other potentially confounded changes.

Third, our medium-run estimates and our estimates regarding discontinuation should be interpreted with caution. People select into continuing or discontinuing use of the drug. So the households that we observe in months 7-12 post-adoption may differ from those that have discontinued use of the medications.

Finally, our estimates are for the United States at a point in time when the drug was being adopted rapidly for the first time. Effects in the UK may differ.

What else do we need to know?

We do not know what will happen in the long run: will there be a new equilibrium level of food expenditure among those who continue to use the drugs? How will the effectiveness and ease of use of these drugs evolve? What will happen if individuals continue to stop and start using the drugs? What will be the impact on nutrition and health?

As access to the drugs increases around the world, it will be important to see how our results compare across countries and across different types of users. In the United States, there is a very large difference in estimates across income, so who has access to the drug will be a key determinant of the impact on aggregate food demand.

Where can I find out more?

The no-hunger games: how GLP-1 medication adoption is changing consumer food demand: Sylvia Hristakeva, Jūra Liaukonytė and Leo Feler present the household purchasing evidence discussed in this article.
How can we reduce obesity and improve diets? This Economics Observatory article reviews evidence on information, taxes and other policies intended to improve diets.
Changes in household diet: determinants and predictability: Stefan Hut and Emily Oster examine why large dietary changes are unusual.
Continued treatment with tirzepatide for maintenance of weight reduction: The SURMOUNT-4 trial studies weight change after continued treatment or withdrawal.
Weight loss medicines in England: This House of Commons Library briefing explains NHS access rules and the tirzepatide rollout.
The labor market impacts of GLP-1s: Meltem Daysal and co-authors use Danish registry data to measure effects on sickness absence and earnings.
Do GLP-1 medications pay for themselves? Coady Wing and co-authors examine whether the drugs reduce other medical spending.
Silver bullet or sticking plaster? Weight-loss drugs and the UK's obesity crisis: Nesta report by Charlie Meyrick and Parita Doshi.