Funding Applications When the Evidence Base Is Thin

Most funding applications for social connection work follow the same template: state the problem in epidemic terms, cite a mortality statistic, promise an outcome the evidence cannot actually support, and hope no one checks. This works until it doesn’t — until a commissioner asks for the underlying trial evidence for a claim like “reduces GP attendance by 20%,” and the applicant discovers that number came from a single small study that the systematic review flagged as low quality.

This is not a rare situation. It is close to the default one. The evidence base for social connection interventions is real but patchy: strong on the health consequences of isolation, weak on which interventions reliably reverse it. Organisations writing funding applications sit in that gap constantly. The question is what to do when a funder’s ask outruns what the evidence will bear.

The gap, named specifically

The American Heart Association’s 2022 scientific statement is unusually direct about this. It documents a roughly 30% increased risk of heart attack, stroke, or death associated with social isolation and loneliness, and then states plainly that the absence of intervention evidence is the central research gap. The association between disconnection and poor health outcomes is well established. The evidence that any particular programme reverses that association is not.

The National Academies’ 2020 consensus report on older adults makes a related point from the clinical side: it calls for routine assessment of isolation in health care settings but stops well short of endorsing specific interventions as proven. A 2023 review of the loneliness and isolation literature more broadly found inconsistent measurement across studies to be a persistent barrier — different tools, different definitions, different time windows — which makes it hard to compare one programme’s claimed effect against another’s, let alone build a cumulative evidence base.

Social prescribing sits in the middle of this gap, and it is worth looking at closely because it is the intervention most often funded on the strength of claims the evidence doesn’t fully carry. A 2021 systematic review of social prescribing and loneliness found that all nine included studies reported positive individual impacts, and three reported reductions in service use — GP visits, emergency attendance, social worker contact. That sounds like exactly what a commissioner wants to hear. But nine studies, self-reported outcomes, and no control group in most of them is a thin foundation for a claim like “reduces NHS costs.” A parallel 2021 review of social prescribing and wellbeing reported gains in self-esteem and confidence, while explicitly noting limited trial evidence and heterogeneity across programmes. A 2022 qualitative synthesis added a genuinely useful and more modest finding: participants describe benefit less in terms of contact itself and more in terms of restored purpose and participation. Structured activity with a role in it appears to do more than unstructured contact.

None of this supports “will reduce loneliness by X%.” All of it supports a more careful claim about the kind of intervention design associated with reported benefit, and about the mechanism — purpose and participation, not contact volume — that seems to matter.

The evidence-status table a funder should see

Applicants who write around the evidence gap, rather than through it, tend to separate claims by how well supported they actually are, and say so.

Claim	Evidence status
Loneliness and isolation are associated with elevated mortality and cardiovascular risk	Strong. Multiple large meta-analyses and a formal AHA scientific statement converge on this.
Structured, purposeful group activity is associated with better reported outcomes than unstructured contact	Moderate. Consistent across qualitative syntheses, not yet tested in controlled trials at scale.
Social prescribing reduces loneliness for individual participants	Moderate, self-reported. Positive in every included study in the relevant systematic review, but the review base is small and heterogeneous.
This specific programme will reduce loneliness by a stated percentage	Not supported by anything currently published. No intervention literature exists at that level of specificity.
This specific programme will reduce downstream health service use	Weak. A minority of studies in the 2021 social prescribing review found this; it was not the majority finding and was not tested with controls.

Handing a funder something like this table, rather than a single confident number, does two things. It shows the applicant understands the literature well enough to know its limits, which is itself a credibility signal. And it forecloses the awkward later conversation where an evaluator finds the same gaps and starts wondering what else in the application was overstated.

What to do instead of overclaiming

Reframe the ask around the mechanism, not the outcome statistic. Instead of promising a reduction in loneliness scores, propose to build the thing the qualitative evidence associates with benefit — structured roles, recurring participation, purpose within the group — and commit to measuring participation and retention as leading indicators. This is a defensible claim because it rests on what the evidence actually shows, rather than what a mortality meta-analysis implies at several removes.

Separate the population-level case from the programme-level case. The population-level argument — that disconnection carries the mortality and cardiovascular risk documented by Holt-Lunstad’s 2015 meta-analysis and the American Heart Association — is strong and worth stating up front to establish why the funder should care at all. But it says nothing about whether this specific programme, run this specific way, will move the needle. Conflating the two is the most common overclaim in the sector, and it is also the easiest one for a diligent reviewer to catch.

Propose a smaller, honest evaluation instead of a large, unfalsifiable one. A programme that commits to measuring retention past the point where most people drop out, and reporting honestly if that number is bad, is more fundable to a sophisticated funder than one promising a wellbeing score improvement it cannot demonstrate causally. The 2023 Surgeon General’s advisory frames social connection as a public health priority requiring sustained infrastructure, not a single programme’s outcome metric — that framing gives applicants room to propose infrastructure-building goals rather than inflated individual-level claims.

Name the limitation in the application itself. Funders who work across health and social programmes have generally seen enough inflated logic models to be suspicious of ones with no acknowledged weaknesses. Stating explicitly that intervention-level evidence for this category of programme is still developing, and that the proposed monitoring plan is designed to contribute to that evidence rather than assume it, reads as more credible than a claim with no edges.

What this means in practice: if a funder’s request for proposals asks for a projected reduction in loneliness or isolation, do not manufacture a number to fit the ask. Offer the strongest evidence-backed claim available — usually about mechanism (structured, purposeful activity) rather than magnitude — and propose to measure retention and participation as the proxy that is actually defensible. If the funder insists on an outcome number with no basis in the literature, that is useful information about the funder, not a problem to be solved by inventing evidence.

The failure mode: borrowed certainty

Call it the borrowed-certainty problem. It happens when an application takes a well-evidenced population statistic — the AHA’s 30% cardiovascular risk figure, the Holt-Lunstad mortality odds ratios — and attaches it, implicitly or explicitly, to a specific programme’s projected impact, as though the strength of the population evidence transfers to the intervention. It doesn’t. The statistic describes the risk of disconnection. It says nothing about whether a twelve-week walking group in a particular town reverses it. Reviewers who know the literature will notice the substitution; reviewers who don’t will fund it anyway, and then the programme will be evaluated against a number it was never going to hit.

What this does not solve

Being honest about evidence gaps in a funding application does not close those gaps. The intervention literature for social connection programmes remains thin relative to the size of the problem it addresses, and no amount of careful framing in a single application changes that. It also does not solve the structural problem that funders sometimes want a specific number for their own reporting requirements, regardless of what the literature supports — in which case the honest option may simply be turning down that particular funding line rather than manufacturing the figure it demands.