Framework
Count Who You Missed, Not Just Who Improved
Community connection programmes almost always report outcomes for the people who turned up and nothing about the people who did not. A practical method for building a reach denominator you can defend to a funder, and why outcome data alone will mislead you.
Institute for Social Connection

Your quarterly report probably says something like: 24 participants, mean UCLA loneliness score down four points, 91% would recommend the group. What it almost certainly does not say is how many people in your catchment could have come, how many knew the group existed, and how many of those who enquired never showed up once.
That missing number is the more consequential one, and you can start collecting it this quarter.
Outcome-only reporting will flatter almost any programme
The social prescribing literature makes the problem visible. The 2021 systematic review in Perspectives in Public Health found that all nine included studies reported positive individual impacts on loneliness, with three also reporting reductions in GP, emergency, social worker or inpatient service use. A run of nine studies with nine positive results is not, on its own, evidence that the intervention works. It is equally consistent with a field that evaluates the people who self-selected in and stayed. The parallel review in the International Journal of Environmental Research and Public Health said as much more directly: limited trial evidence, high heterogeneity across programmes.
The American Heart Association’s 2022 scientific statement, which established that social isolation and loneliness carry roughly a 30% increased risk of heart attack, stroke or death from either, named the absence of intervention evidence as the central research gap. The 2024 HEAL-HOA dual randomised controlled trial, which tested prosocial engagement and volunteering against a control among lonely older adults in Hong Kong, is notable mostly for being a randomised trial at all in a field otherwise dominated by small uncontrolled evaluations.
You are not going to fix that with your programme’s pre-post survey. But you can stop your pre-post survey from being actively misleading, and reach data is how.
The flattering denominator
Here is the failure mode to name and watch for. The flattering denominator is what happens when the only population you can describe is the population that showed up.
It works like this. You recruit through a partner GP surgery, a library noticeboard and a WhatsApp group. Fourteen people attend the first session. Nine complete the eight weeks. You measure those nine, twice, and report the change. Every number in your report has a denominator of nine or fourteen — never of the 4,000 adults over 65 in the catchment, of whom the National Academies’ 2020 consensus report would suggest roughly a quarter are socially isolated. Your reported effect might be real. Your reported effect is also, structurally, a description of people who were already willing to walk into a room full of strangers.
The consequence is not just that funders eventually catch on. It is that you cannot tell whether to spend next year’s money on improving the group or on getting different people into it.
Build a reach denominator in five steps
- Name one denominator in a single sentence. “Adults aged 65+ registered at the three practices we take referrals from: 4,100.” Or “residents of the two wards the funding covers: 11,200.” It does not need to be the theoretically perfect population. It needs to be stable, countable and the same one you use next quarter.
- Estimate the in-scope share using published prevalence, and label it as an estimate. The AARP Foundation’s 2018 national survey, which used the 20-item UCLA scale rather than a bespoke instrument, found one in three US adults aged 45 and older were lonely. Harvard’s Making Caring Common survey put serious loneliness at 36% of Americans and 61% of young adults aged 18–25. Applying a national rate to your ward is crude. Say so, and use it anyway — a crude denominator beats no denominator.
- Count four stages, not one. Aware of the offer; enquired or was referred; attended once; attended four or more times. Most programmes only have the third and fourth numbers. The gap between referred and attended-once is usually where the programme is actually losing people, and it is invisible without stage two.
- Collect demographics at the referral stage, not the attendance stage. Otherwise your equity analysis describes your regulars. KFF’s 2024 survey is one of the few that breaks loneliness and social support down by race and ethnicity, and connects experiences of discrimination to network size; the Survey Center on American Life reported 15% of men with no close friends, a fivefold rise since 1990. If groups like these are absent from your attendance list, you need to know whether they were absent from your referral list too.
- Report reach and outcomes on the same page, for the same period. “Nine of an estimated 1,000 isolated older residents completed the programme; their mean loneliness score fell four points” is an honest sentence. It is also the sentence that gets a conversation about outreach funded.
What this means in practice: if you can only add one field to your intake form this year, add the referral date and source for everyone referred — including the people who never attend. That single field converts your attendance list into a funnel, and a funnel is the smallest unit of reach measurement a funder will accept.
Why reach cannot replace outcomes
Reach on its own justifies volume, and volume is not the goal. The 2022 qualitative meta-synthesis in BMC Health Services Research found participants describing benefit that went beyond social contact to restored meaningful participation and purpose, and suggested structured, purposeful group activity outperforms contact alone. A programme that touches 400 people with unstructured drop-in contact may well produce less than one that touches 40 with something purposeful. You need both numbers to see that trade-off at all.
| Claim | Evidence status |
|---|---|
| Isolation and loneliness raise cardiovascular and mortality risk | Strong. Consistent across meta-analyses and the AHA statement |
| Social prescribing improves individual loneliness outcomes | Weak-to-moderate. Uniformly positive but mostly uncontrolled; two 2021 reviews flag limited trial evidence |
| Structured, purposeful activity beats contact alone | Moderate, qualitative. Consistent participant accounts, no trial comparison |
| Volunteering reduces loneliness in lonely older adults | Under test. One dual RCT exists; treat as promising, not settled |
| Loneliness and isolation are distinct constructs needing separate measures | Reasonably strong. Supported by 2024 work in Scientific Reports on how the two relate differently by age |
| Reach data predicts population-level impact | No direct evidence. It is an arithmetic prerequisite, not a finding |
One more measurement trap: BMC Public Health’s 2023 review of the field identified inconsistent measurement as the main barrier to comparing findings across studies. That applies to your reach metrics as much as your outcome scales. Pick your four stages, write down the definitions, and do not redefine them when the numbers look bad.
What this does not solve
A reach denominator tells you the size of the hole. It does not tell you how to fill it, and the honest position is that nobody has a well-evidenced method for pulling in people who will not self-refer. Klinenberg’s argument that libraries, parks and other social infrastructure shape rates of social contact is the most plausible route — reach people where they already are rather than asking them to come to you — but it is a case for a different kind of investment than a weekly group, and it is not something a programme manager can commission.
Measuring reach also cannot fix the deeper selection problem. Everyone your funnel captures at stage one is someone a referrer noticed. The people no referrer notices stay outside the denominator entirely, and they are likely the people at highest risk.
Sources
- Understanding Loneliness: A Systematic Review of the Impact of Social Prescribing Initiatives on Loneliness
- Can Social Prescribing Foster Individual and Community Well-Being? A Systematic Review of the Evidence
- Do People Perceive Benefits in the Use of Social Prescribing to Address Loneliness and/or Social Isolation? A Qualitative Meta-Synthesis
- The Effects of Volunteering on Loneliness Among Lonely Older Adults: The HEAL-HOA Dual Randomised Controlled Trial
- Effects of Objective and Perceived Social Isolation on Cardiovascular and Brain Health: A Scientific Statement From the American Heart Association
- Loneliness and Social Connections: A National Survey of Adults 45 and Older
- Social Isolation and Loneliness in Older Adults: Opportunities for the Health Care System
- Loneliness in America: How the Pandemic Has Deepened an Epidemic of Loneliness
- Loneliness and Social Support Networks: Findings from the KFF Survey of Racism, Discrimination and Health
- The State of American Friendship: Change, Challenges, and Loss
- The State of Loneliness and Social Isolation Research: Current Knowledge and Future Directions
- Understanding the Interplay Between Social Isolation, Age, and Loneliness During the COVID-19 Pandemic
- Palaces for the People: How Social Infrastructure Can Help Fight Inequality, Polarization, and the Decline of Civic Life