On Not Building Blind.
I've been reading OHID's Commissioning Quality Standard closely this week. Not as a critic. As someone building a tool in this space who needs to understand what it's meant to sit alongside.
I’ve been reading OHID’s Commissioning Quality Standard closely over the past week, as one does, not as a critic and not as a commissioner, but as someone building a tool in this space who needs to understand the world it would eventually have to sit inside. Whatever I build, it either needs to be aware of this standard, work with it, or at minimum not get in its way. So I’ve been doing the reading properly, and wanted to think out loud about four things I noticed.
Some context on why. I’m building Sol, an AI companion for people concerned about their relationship with alcohol, mostly for people who haven’t yet engaged with a formal service. That’s a specific enough angle that I want to be upfront about it before saying anything else. I should also say I’m reading this as someone in recovery myself, and as someone who has worked inside a treatment service as well as used one. I’ve been on both sides of a caseload, the person waiting to be seen and the person doing the seeing. That doesn’t make my reading of this document neutral either. If anything it makes it more partial, in both senses of the word. But I’d rather say that plainly than pretend I’m coming to this cold.
OHID’s Commissioning Quality Standard, published in 2022 off the back of Dame Carol Black’s review of drugs, sets out what a well-run local alcohol and drug system is expected to look like. It’s not just aspirational language. It’s structured as standards with criteria, and criteria with evidence requirements. Partnerships are meant to be able to point to something and say, yes, we’re doing that.
As I say, four requirements really stood out to me. Not because they’re controversial, and not because I think I know what to do about them. They stood out because I couldn’t immediately picture how a caseload-capped, session-based treatment system meets them, and I wanted to ponder that honestly rather than assume I was missing something obvious.
The people who aren’t there
Standard 2.1 asks partnerships to build a needs assessment that incorporates the views of people who may benefit from support for problem drug or alcohol use, including people who are attending treatment and recovery services and those who are not.
Let me start here. Commissioners are explicitly required to understand the needs of people who have never engaged with a service. Not people on a waiting list. Not people who dropped out. People who haven’t come forward at all.
I don’t know how you’d go about that well. Surveys reach people who are already willing to be surveyed. Outreach work reaches people a worker happens to encounter. Neither seems designed to capture the person who’s noticed their drinking has become a problem, hasn’t told anyone, and has no current plan to. That person sits outside every dataset a needs assessment normally draws on. NDTMS can’t see them. Case files don’t exist for them. I’m genuinely unsure whether this is a gap because nobody’s had reason to solve it, or because the honest answer requires something structurally different from how services currently reach anyone. I don’t think I know the answer. I just noticed the standard asks for it.
Planning for relapse before it happens
Standard 3.3 asks recovery-oriented systems of care to ensure rapid and supported re-entry into treatment for people who have relapsed, and separately requires that services support people to identify their post-treatment support needs as early as possible from their initial contact with treatment.
Both halves of that I found fascinating. Post-treatment needs are meant to be identified from the very first contact, long before discharge is anywhere in view. Relapse, when it happens, is meant to trigger rapid re-entry, not a fresh referral process starting from zero.
I found myself asking what “rapid” is realistically meant to mean inside a system where caseloads are capped, or crammed, or both - and appointments are often booked weeks out. A person who relapses at 11pm on a Tuesday is encountering whatever’s available until the next working day, not a system built for rapid anything. I’ve been that person, and I’ve also been the worker who wanted to be rapid for someone and simply didn’t have the capacity in the diary to be. I want to be careful here, because it would be easy to read that as a criticism of services, and it isn’t one. Continuous availability was never really the design brief of a caseload-based model of care. The standard describes what the system ought to achieve. I’m not sure the structure it’s asking that of, was built for it.
Family support as a standing requirement
Two more separate standards return to this. Standard 3.3 says recovery planning must, at minimum, account for people’s social connections, friends, family, peers, colleagues, and that support after treatment should incorporate the needs of families, carers and dependants. Standard 4.2 goes further: family members and carers directly affected by another person’s problem drug or alcohol use can access support for their own needs, independent of the person in treatment.
That last point struck me as significant. This isn’t family support as an extension of someone else’s care plan. It’s a standing entitlement for the family member in their own right, whether or not the person they’re worried about has engaged with anything.
I went back to an older Public Health England document while reading this, their SROI guidance from 2015, which openly admitted it had no reliable way to estimate outcomes for families, because there’s no national dataset that captures what happens to them. Seven years later, the commissioning standard confirms family support isn’t optional. It’s required. I don’t have a tidy resolution to offer between those two documents. I just think it’s worth naming that a standing requirement and a measurement gap have been sitting next to each other for the best part of a decade.
Support when nobody can be face to face
For me, standard 4.3 is the most quietly practical of the four. Services are required to ensure business continuity and enable staff to continue offering support through unexpected challenges, including providing support when face-to-face interventions are not possible, for example during COVID-19 restrictions, or in rural council areas.
The pandemic reference dates the language, but I don’t think the requirement expires with it. Rural areas are named in the same breath as a global health emergency, which suggests “unable to be face to face” isn’t meant as a one-off crisis scenario. It reads more like an everyday reality for a meaningful slice of the population this system serves. A person forty minutes from the nearest service, without reliable transport, working shifts that don’t align with appointment slots, is living inside that phrase every week, not just during a lockdown. I don’t have a confident view on what continuity looks like for that person inside a system built primarily around a physical or scheduled appointment. It’s a genuine open question for me.
Sitting with all four
None of these four standards reads like a stretch target to me. They’re current, audited requirements, and none of them describe a problem with one obvious institutional owner. A needs assessment team can’t manufacture the views of people who’ve never engaged. A keyworker with a full caseload can’t be continuously available for rapid re-entry. A commissioner can fund family support services, but funding a service and reaching every affected family member are different problems. A rural service can extend its hours, but hours are still hours.
I don’t have a settled view on what closes any of these gaps, and I’d be wary of anyone who claimed they did after one close reading of a policy document. What I keep returning to, as someone building a tool that’s meant to be reachable outside appointment hours and present before someone’s ready for a service, is a genuinely open question rather than a pitch: is there a relevant role for something like that here, alongside a needs assessment team, a keyworker, a family support worker, a rural outreach service, in the specific hours and circumstances none of those can currently cover? I don’t know yet. It’s possible the honest answer is that reach isn’t the same as value, and that, for me is worth taking seriously rather than assuming away.
What I’m fairly sure about is this: the standard itself seems to accept that the current system, as designed, wasn’t built to close these four gaps alone. That’s not a criticism of the people running it. It’s possibly just an honest description of the limits of a model built around scheduled, in-person contact. I don’t think the interesting question is whether the standard is right to ask for these things. It clearly is. The question I’m still working through is what, if anything, could sit alongside that system to help meet them, and how you’d know honestly whether it was actually working.
I’m building Sol, an AI companion for people concerned about their relationship with alcohol, and this piece is part of me trying to understand the standards it would need to sit alongside. Sol is here if you want to see what that looks like in practice, or want to tell me where you think I’ve got this wrong.
David Henzell is the founder of Phenomenal Sobriety Limited, building Sol, an AI companion for people concerned about their relationship with alcohol.
Source: Office for Health Improvement and Disparities, Commissioning quality standard: alcohol and drug treatment and recovery guidance (2022). Available at: gov.uk/government/publications/commissioning-quality-standard-alcohol-and-drug-services