The Body Keeps Company

ALoneliness arrived in medical research late and by an awkward route. For most of the twentieth century the feeling was left to novelists and moralists; the clinical literature treated it, where it treated it at all, as a symptom of something else. The first instrument to measure it in a standard way, a twenty-item questionnaire published in 1978, changed that less by what it asked than by what it insisted on keeping apart. Loneliness, on that definition, is the felt gap between the company a person wants and the company that person has. Social isolation is a different quantity altogether: a count of contacts, conversations and household members that can be taken without asking anyone how they feel. The two were long assumed to rise and fall together. They do not do so as closely as the assumption requires. In a 2019 survey of 8,400 adults, only 31 per cent of those scoring in the top fifth for loneliness also fell in the bottom fifth for frequency of contact.

BIngrid Solvang, who directs the population health unit at the Marchmont Institute, treats the feeling as an alarm rather than a misfortune. On her account loneliness is the social equivalent of hunger: an unpleasant signal built to be uncomfortable, since a signal that could be ignored would not do its work. Her group followed 12,600 adults for nine years, recording loneliness scores at entry and deaths thereafter. Participants in the loneliest fifth were 26 per cent more likely to die during that period than those in the least lonely fifth, once smoking, body mass, household income and existing diagnoses had been taken into account. Solvang does not claim that the feeling kills anybody directly. She claims that it holds the body in a state of readiness — watchful, poorly rested, slow to recover — which served an animal separated from its group for a single night and becomes ruinous when it is sustained for years.

CAlarms can be examined, and Solvang's laboratory has examined this one. In a 2021 study, 180 volunteers slept three nights in a monitored room while sleep fragmentation and inflammatory markers were recorded. Compared with controls, the lonely volunteers woke more often in the night and showed higher readings on one inflammatory marker — but both differences were confined to the 74 participants who had reported the feeling for more than two years. Among the volunteers whose loneliness was of shorter standing, sleep and readings alike were indistinguishable from those of the controls. Solvang is careful about what this licenses. The finding, she writes, belongs to the chronic form, and cannot be carried across to the ordinary loneliness of a bad month or a first winter in a new city. A second team failed to reproduce the sleep result in 2023, using a sample drawn from night-shift workers whose rest was already broken.

DOscar Sandoval, an epidemiologist at Cawley University, accepts Solvang's measurements and rejects the arrow she draws through them. Illness, he argues, is where the sequence begins: pain, deafness, breathlessness and the surrender of a driving licence thin a person's social life long before any of them proves fatal, so the feeling that follows is a consequence of decline rather than a cause of it. Re-analysing 24,000 records in 2022, he added baseline mobility, hearing and depressive symptoms to the variables adjusted for. The excess risk of death fell from 26 per cent to 9 per cent, and the interval around the smaller figure was wide enough to include no effect at all. Sandoval volunteers two weaknesses of his own. The health measures taken at entry were self-reported; and adjusting for a condition that lies on the causal path, if loneliness does damage health, strips away part of the very effect under investigation.

EThe dispute has consequences in clinics. Since 2020 the Kelbrook health board has funded social prescribing: link workers refer patients who appear isolated to choirs, walking groups and gardening schemes rather than to a prescription pad. Here the expected positions are reversed. Sandoval, who doubts that loneliness causes much of anything, is the scheme's warmer supporter, because half the practices were assigned to it by lot and the arrangement will therefore yield the randomised comparison the field has never had. Solvang is the cautious one. Attendance, she observes, is a count of contacts — the very quantity her definition sets aside — and a patient may sit in a crowded room and still not feel wanted. Both object to the commercial screening apps now quoting the 26 per cent figure at their users, since that number came out of a nine-year cohort of adults and says nothing about what any individual should do this week.

FWhat the field has reached is a division of the question rather than a verdict on it. Where loneliness is brief and the circumstance is one a person can alter — a move, a bereavement, a first year at university — the feeling has not been shown to leave any mark on health, and it generally ends. Where it becomes chronic, the physiological picture Solvang describes appears with some consistency, although whether it appears because the feeling persists, or because the circumstances producing it also produce poor health, is exactly what Sandoval's re-analysis puts in doubt. Neither researcher will estimate what share of loneliness in the population is of the lasting kind. What both refuse is the framing common in public campaigns: that loneliness is an epidemic with a single cause and a single remedy, when the evidence turns on how long it lasts and on what else is happening to the person who reports it.