Poor, over 65 and rarely able to leave the house: more than four in ten Americans in that position got neither food stamps nor Meals on Wheels. Of those who did get help, more than a third still went without

Some older people hardly ever leave the house. They can't manage the steps, or the walk, or they need someone with them to do it. In the United States there are millions of them still living at home, and they are poorer, sicker and more often alone than other people their age. Getting food in is a daily problem that money only partly solves. A team at Johns Hopkins University School of Nursing, with a colleague in Arizona, asked how many of them are reached by the two best-known kinds of help: food stamps (SNAP, the Supplemental Nutrition Assistance Program) and delivered meals such as Meals on Wheels. Their paper came out in Health Affairs Scholar on 1 October 2026.

They used the National Health and Aging Trends Study, which interviews a national sample of people aged 65 and over on Medicare every year. From the round run between May 2023 and April 2024 they took everyone who lived in the community, was homebound (never or rarely went out, found it hard, or needed help to do it) and had a household income at or below 130% of the federal poverty line, the usual income limit for SNAP. That gave 1,091 people. With the survey's weights they stand for about 3.9 million Americans. Their average age was 78, seven in ten were women, a third lived alone and a quarter probably had dementia.

Just over half, 55.5%, had some food assistance: 35.1% had SNAP only, 10.9% had delivered meals only and 9.4% had both. The remaining 44.5%, about 1.75 million people, had neither. The paper's title puts this as the programmes serving 'the majority'. Seen from the other side, more than four in ten people who are poor enough to qualify and can barely get to a shop were getting nothing from either. And adding the two meal figures together, only about one in five had any delivered meals at all, the one service built for people who can't go out.

Then the result that looks backwards. Among people with no assistance, 24.0% were food insecure. Among people with assistance it was 36.4%. By programme it climbed in steps: 33.6% with SNAP only, 39.4% with delivered meals only and 44.1% with both. Nobody should read that as the programmes making people go hungry. The people on them were worse off to begin with: they needed help with more everyday tasks (4.3 basic activities such as bathing and dressing, against 3.7), and 41.8% lived alone against 21.8%. People in the most trouble are the ones who apply, or get signed up by a relative or a nurse.

The yardstick matters as much. This is not the US government's food security questionnaire, which is about money. The survey uses five questions, and a yes to any one counts: skipping meals because there wasn't enough money, doing so for more than a few days, and three about going without groceries, hot meals or eating because you couldn't manage it yourself and nobody was there to help. So a person can count as food insecure with enough money in the bank, because they can't stand at a stove. The authors say this is deliberate, and for this group it is the right thing to measure. It also means the measure and the delivered-meals question overlap. People were asked how often their hot meals came from Meals on Wheels or a similar programme, and people who can't cook are the ones who end up on it.

What the higher figure does show is that being enrolled is not the same as having enough. A food stamp card is no use to someone who can't get to the shop or cook what they buy. A delivered meal is one meal, and the rest of the day still has to be managed. The authors' argument is for more capacity and easier sign-up: SNAP paperwork and renewals are hard for someone with disabilities, and they cite a 2026 figure that one in three Meals on Wheels providers has a waiting list.

Among the 465 people with no help, 97 were food insecure. They needed help with more daily tasks than the rest (5.0 against 3.2) and were about twice as likely to live alone (35.0% against 17.8%). That second comparison only just passes the usual test for chance (p = 0.049), on 97 people. By our own arithmetic from the paper's weighted figures, that group stands for roughly 420,000 people nationally who are poor, homebound, short of food and on neither programme, and about 790,000 more are short of food while on one. The paper does not give those two totals.

The limits are mostly the authors' own. Everything is from interviews, some answered by a relative on the person's behalf. The delivered-meals questions lump Meals on Wheels with 'other food assistance' and count anyone who 'rarely' had such a meal, so that group is wider than the name suggests. Food stamp use is usually under-reported in surveys. It is one snapshot, so it can't show whether help reduced anyone's hunger; earlier studies the paper cites suggest it does. Other routes, such as Medicare Advantage food benefits or Medicaid programmes, were not counted. Two things we'd add: about six in ten of the people invited took part in that survey round, and the paper has no direct look at what anyone actually ate.

Our reading: the headline could have been written either way, and the gap is the more useful half. Roughly one in four of the homebound poor with no help said they were going without, and those people were frailer and more alone than their neighbours. If you have an older relative or neighbour who doesn't get out, the questions worth asking are plain ones: who does the shopping, who cooks, and what happens on the days nobody comes. In the US, the local Area Agency on Aging is a usual first call for delivered meals, and many can help with a SNAP application.