What the Household Pulse Survey Can and Cannot Tell Us About Loneliness

A study published February 1 in Community Mental Health Journal breaks down loneliness and social isolation prevalence in the United States by demographic and socioeconomic factors, using data from the 2024 Household Pulse Survey. It is a useful demographic breakdown. It is not, on its own, evidence that loneliness rose or fell relative to any other year, because the Household Pulse Survey was not designed to produce a stable loneliness time series in the first place.

The Household Pulse Survey is a Census Bureau instrument built for rapid-turnaround measurement, originally stood up to track the economic and social effects of the pandemic. It has added and dropped questions across its many rounds, changed its sampling approach, and adjusted question wording as it moved from a temporary tool to something closer to standing infrastructure. That flexibility is exactly what makes it useful for the purpose it was built for: fast reads on a fast-moving situation. It is a poor foundation for saying whether loneliness in the U.S. has moved by some number of points since a given prior year, because “loneliness” as asked in one round of the Pulse Survey is not guaranteed to be the same question, the same scale, or the same reference period as loneliness asked in another.

This matters because prevalence figures from different points get placed side by side constantly, as though they were readings from the same thermometer.

Three instruments, three numbers, one country

Consider what is already on record for the United States. The CDC’s Morbidity and Mortality Weekly Report, publishing in June 2024 but drawing on 2022 data, produced federal surveillance estimates of loneliness and lack of social and emotional support. The 2025 HINTS-6 analysis, using 2022 data as well but published in November 2025, found that roughly one in seven U.S. adults reported severe loneliness and one in five moderate loneliness, for a combined figure above 37% experiencing moderate-to-severe loneliness. Cigna’s 2025 Loneliness in America report, with fieldwork run in mid-2024, put self-reported loneliness at 57% of Americans. The Surgeon General’s 2023 advisory cited roughly half of U.S. adults experiencing loneliness, largely drawing on the Cigna loneliness index tradition and other survey work compiled around that time.

None of these figures is wrong. They are answers to different questions, asked to different samples, using different scales, in different years, published on different schedules relative to when the underlying data was actually collected. The CDC and HINTS-6 numbers both come from 2022 fieldwork but read as though separated by three years because of when each report happened to reach print. The Cigna figure comes from its own proprietary loneliness index, which has never used the UCLA Loneliness Scale that much of the academic literature treats as a baseline instrument. Layering the new Household Pulse Survey estimates on top of that pile does not resolve the disagreement. It adds a fourth instrument to a set that already cannot be reconciled.

The one clean comparison in the American literature is AARP’s survey of adults 45 and older, first fielded in 2018 and refielded for the 2025 follow-up, Disconnected, using the same 20-item UCLA Loneliness Scale in both rounds. That consistency is rare enough to be worth naming as an exception rather than the rule. AARP’s 2018 survey found one in three adults 45 and older reporting loneliness; the 2025 follow-up reports the figure has risen further in that population. Because the instrument did not change, the movement between the two rounds is more likely to reflect movement in the underlying condition than movement in the measurement.

What the new analysis actually offers

The Household Pulse Survey analysis is genuinely useful for what it does show: how loneliness and isolation prevalence break down by demographic and socioeconomic factors within a single 2024 dataset. Comparing groups measured at the same time, with the same instrument, is exactly the kind of comparison survey data supports well. Comparing that same 2024 snapshot against the 2022 HINTS-6 figures, the 2022-based CDC surveillance data, or Gallup’s global loneliness figures from a different survey instrument and a different sampling frame entirely, is a different exercise, and one the paper does not attempt to make.

Gallup’s own 2023 finding that roughly a quarter of people worldwide report feeling lonely came from yet another instrument, applied across 142 countries rather than within the U.S. alone. Treating any of these figures as a single moving needle, tracked over years, invites a comparison the underlying instruments were never built to support.

The fix is not complicated, and it is not happening

A body of literature already documents this problem in general terms; the state-of-the-field review published in BMC Public Health in 2023 named inconsistent measurement as a central barrier to comparing loneliness findings across studies. What would resolve it for a series like the Household Pulse Survey specifically is straightforward to describe, if not to execute: hold the loneliness question, its wording, its response scale, and its placement in the survey fixed across rounds, publish the exact instrument alongside every release, and resist adding or dropping items in response to whatever policy question is current that quarter. The Pulse Survey’s strength is speed. A prevalence figure worth tracking over time needs the opposite virtue: it needs to stay boring, on purpose, for years.