Is there any *actual* evidence for social prescribing yet?

18 February 2020

…..I was asked recently. In fact it’s a q that crops up a lot. My immediate answer is normally:

1) Yes

2) Define social prescribing

3) Define “work”

4) It’s tricky

5) Apply a level playing field, and a lot of common sense


My previous thoughts on evaluation in this thread... Here I unpack it a bit more. 

1. What evidence IS there?

A cynical (or pragmatic?) view would be that a “roll out” has shades of many other schemes that get rolled out …. Seems “intuitively a good idea but rolled out without proper resourcing or evaluation”. The former is definitely true.

With regard to the latter – It is CERTAINLY a concept that is under evaluated as the Bickerdike “rhetoric v reality” paper aptly demonstrates. What we have in terms of research based is a fairly wide set of smallish scale, short-ish term studies, often not published in peer reviewed journals.

There is plenty of stuff out there. I won’t re rehearse the evidence base here.

The evaluation of part of the Sheffield scheme is here c /o @JanetHarrisShef. It focused on the “link worker” bit (for us called Community Support Workers). Arguably this is the EASIEST bit of the SP system to research outcomes and impact. It wasn’t that easy. Getting the data sorted for all this was a royal trauma we did our best. The evaluation looked at the feasibility and utility of a city-wide, brief intervention provided by community support workers to link people with non-medical issues to a range of services.

The best quant analysis I ever saw was Luke Munford – talk here, paper here. @SoadyJohn in my team told me at the time that “Luke nailed it”. John is very battle hardened. If he is happy methodologically I’m not arguing with him. Key points – a) Participation in community assets is associated with substantially higher HRQoL but is not associated with lower healthcare costs once obvious confounding factors were controlled for, b) based on a threshold value of £20 000 per QALY, the net benefits of participation in community assets were £763 per participant per year, c) there is some interesting stuff (IMO) on demand shifting in the results section also.

Developing the evidence base – Does anyone want to invest in big studies?

TBF few seem interested in funding robust evaluations. Obviously there’s no commercial interest as there aren’t gizmos or drugs with a name ending in “mab” involved. This is exacerbated by commissioners having no funding to commission such research and / or don’t see it as their role to fund “proper” research.

There are multiple programmes of work being undertaken at the moment to improve the evidence base and develop a common outcomes framework. This is work in progress and something to be hopeful we will get the right note from it. The evidence for the types of intervention that SP links people to is relevant here – mostly with physical exercise and the benefit of programmes that combat loneliness and social isolation. We should be looking beyond the mechanism – there is no evidence for a paper prescription after all, just for the drug that is at the end of it.

2. However, there are a number of important buts in sorting out a robust conclusion on whether there is (or isn’t) any good evidence base supporting this field

a) What IS social prescribing, boundaries

Agreeing what the intervention “is” seems not uncontested.

The name remains problematic for many. It is not a term the NHS has actively created (building on work in Bromley By Bow – Everington / Mike Dixon and similar). The term SP has legitimised “social” as an intervention for the NHS, which is helpful in itself. The recent NHSE push is trying to find the sweet spot between medical model and full on ABCD, and provide a mechanism for NHS frontline staff to engage in what for many are called the “social determinants of health”, this seems a good thing. The aim is to systematise and scale it. We all know we need to take care re not about over medicalising or creating a paternalistic structure…. or simply a signposting service. We all know employing a link worker in itself does not constitute social prescribing, and that the concept would not exist without local councils/ Vol sector. It happened for decades without NHS

The blog by @BeckyMalby is excellent – 3 things you should know about social prescribing: – a) Don’t add Social Prescribing on as another project. b) Get out of the way. An asset-based approach generates masses of gifted time, energy, care and compassion. It’s not a service. c) When it works it’s not a service add-on; it’s a whole way of relating – redefining roles in the practice and re-shaping the way professional relate too and with people in communities. d) Count friendships.

Actually putting a definable “boundary” on the thing called social prescribing varies massively from place to place, and varies according to our beliefs, world views, who the funder is and whether funding is small and short term or more structural and long term. SP often misses (in my experience) big areas where there is good evidence of impact. For example a) physical activity (notwithstanding some very important wrinkles in the evidence of long term impact of exercise on prescription narrowly constructed), b) arts and culture See this recent WHO report and this helpful BMJ blog and c) welfare advice often don’t seem “included” in social prescribing.

What is now being called social prescribing has existed for considerably longer than the NHS has, and it is testing the boundaries of an asset based working v deficit model. It will be important to nail the opportunity to properly engage local government, communities & voluntary sector & genuinely relocate some authority and resources. It will be important to miss the opportunity; the NHS is clearly signalling that increasingly it sees “social” as a valid intervention when social is what is needed, not clinical. It is fair to say that the NHS is catching up. The £0.5bn NHSE investment in Link Workers over 5yrs is helpful but we all know (inc within NHSE) that link workers alone are not sufficient. It is reasonably that if NHSE do some investment there should be some investment to come from CCGs / local authorities. If NHSE make investment it is reasonable there is quid quo pro, and it is unreasonable that CCGs and local authorities see this as a reason to disinvest from the current funding. Of course it is a given that funding for all is exceptionally tight. Nothing comes for free.

Does / should the “evidence base” for social prescribing cover the notion of community development per se, link workers (between NHS or other statutory settings and community based assets) or interventions delivered. Or all of the above. To my mind there is a lot of muddled thinking on whether “social prescribing” covers some or all of these, and we probably all think about it differently.

 Social prescribing (or whatever you want to call it) doesn’t exist by itself. Asset based approaches are not new they are well established and we are well and we are continuing to build. Asset based community development is a precursor to the success of community infrastructure, which is a precursor to the success of what many now call social prescribing. We should think carefully on this and shift away from investing in activity towards investing in a community and an infrastructure. If we only invest in activity quite narrowly than the added value and spin-offs are lost.

b) Paradigm of evidence

There is a classic biomedical paradigm meets social paradigm territory. This makes it very, very tricky methodologically. We should be careful on the use and misuse of evidence in this one. I have blogged a lot on this one – see this blog series on the use and misuse of evidence in “public health” (term used advisedly). Esp blog 4 – special cases.

We often, in my experience, have woefully inadequate thinking around “evidence” in complex social paradigms, or an inadequate understanding conceptualization of what they label as SP.

This matters when asking questions about “what does the evidence say”!

c) What outcomes, what impact, return on investment and a level playing field.

I’ve consistently been of the view we need to be exceptionally careful to avoid getting into expectations and asks of what we now call SP.

And, of course we all want a nice, neat, clean and simple evidential answer on “impact”, or the answer to the “what outcomes will it deliver” question. The neat clean answer doesn’t exist. There is a similar thing to be said about say General Practice. As a perhaps inflammatory example, consider this study testing using patient-centred models for those with multiple long term conditions. The abstract of the study clearly says it made no difference. The subsequent cost effectiveness study said it wasn’t really a cost effective intervention (no surprise looking at the original study).

 But, I don’t hear many arguments for abandoning patient-centred models. For what it is worth, I think that the interpretation of this study should be far more nuanced than the last line in the abstract – a story for another day!). To me, measurable “impact” (especially in a way that might make finance teams happy) seems largely an exercise in impossibility given the nature of the system SP operates in. 

There is something to be said here about level playing field around the statutory sector  vs VCS (we all know the stat sector often make VCS sweat a lot about small short term contracts and outcomes, but really don’t know what outcomes we get for the £billions we put into many clinically oriented areas.

There is also something about NOT seeing investment in vol sector as only a mechanism for “managing demand for NHS and social care”, or worse “reducing admission to hospital”. If that happens then a) it is fabulous, b) it is a happy bi-product of an intervention c) there it’s a bi-product of the intervention or service under consideration or whether some broader construct. Demonstrating links and attributability is probably impossible.

We don’t have the sophisticated methods to be able to demonstrate it…. the evidential smoking gun from service to impact. There is some merit in mapping out some form of logic model…. but to really get from intervention to impact is probably impossible given all the other things that are also going on in that system.

This takes us back to the level playing field concept.

I’m not saying that outcomes don’t matter! Getting the right approach to “outcomes” remains an important deal. Of course it remains important to be practical and pragmatic on the extent to which we have an expectation of demand shifting or money saving, especially the ability to measure and attribute this in the complex system social prescribing operates in. 

NHSE have set out what is proposed as a common outcomes framework (see p28). On “outcome measures” I went to an excellent event a few weeks ago on how to best develop outcomes. The notes and live scribe illustrations here.  There is a need for a sensible discussion about proportionality here. We don’t ask statutory sector to collect large volumes of outcome data from every GP appointment or outpatient clinic. Whether we use data for outcomes / monitoring / simply celebrating what is good and happening / or research. We also discussed that if we collectively want “outcomes” data we will have to invest (probably quite heavily) in infrastructure to enable it to be collected, and skills development in VCS orgs and the statutory sector to enable different stakeholders to use it and to engage with it. The search for a single outcome measure to bind them all continues. Personally I don’t think it is that helpful, or at least it is probably a futile search. Finally the workshop had an interesting conversation about whether we should be collecting outcomes at individual level or whether we should focus on groups and places.

We SHOULD expect there to be some well-being benefit, back to the Munford study. Factoring these things into our evidence building model will be important. Then we get into what people want to / choose to believe in, that’s tuned according to our prior beliefs etc. I will settle for carefully thought through logic model, with well thought through process outcomes related to the service per se (are they delivering what they said they can deliver) rather than end user outcomes or outcomes expressed in terms of demand on other downstream services.

So for example our scheme in Sheffield (called People Keeping Well) operates in a complex space – far broader than say “primary care” (as defined medically). But when someone wants to know what “outcomes” are delivered from SP, I often ask what “outcomes” are delivered by primary care, or social care, or even renal or cardiovascular care (though the more specialised we get the more easy it is to have a good stab at outcomes. For cancer – 5yr survival or better site specific mortality, but for an entity that is rather more diffuse it just doesn’t work.

For example this modality of intervention doesn’t exist to saves “health” cashable savings if that happens then it’s a very happy bi-product it probably won’t be seeable, countable, never mind cashable.

On the concept of return on investment, this thread by @harryrutter is excellent. There is a lot of salient advice in this on ROI!

On opportunity cost

I would like to draw the reader’s attention to the cancer drug fund, I have lost count on number of drugs ending in nib or mab that NICE approval based on questionable economics, Proton Beam Therapy, Avastin in ophthalmology (of note NHSE largely silent on this), robotic surgery. I could go on, and on. There are many £bn tied up in these interventions of marginal (at best) benefit with significant opportunity cost. So let’s have a level playing field on the impact of SP please!

d) Funding

Finally on funding there remain big challenges with a fully funded model, imagine if there were a tariff for the activity that people are enabled to access in the same way there is a tariff for a rheumatology outpatient appointment. Imagine a big scheme to incentivise referrals to dermatologists but without any investment in dermatologists The Brit Assoc of Dermatology would go ballistic FTs up & down the land would refuse to participate RCGP & BMA GP committee might grumble about extra work. Imagine turning up to your local pharmacists for your scrip & assuming it’s staffed by VCS & you can take what you want for free.

Making it work requires a huge cultural change across all parts of the system, and that’s a big ask. Making it work at scale from the patchwork of what is there already – across local NHS, VCS and local government cultures – will not be easy, nor cheap. We shouldn’t expect the VCS to do this for free!

Thanks to James Sanderson and Emma Dickinson for helpful comments on an earlier draft. The views are mine.

This reference list for link worker related refs is also good -


Greg Fell is the Director of Public Health at Sheffield City Council.