We’re all setting up ACOs, ACPs, ACSs and similar. Outcome frameworks are, again, on the cards. Outcome based commissioning is, again, on the cards. This time it will (I hope) be a ‘different’ kind of “commissioning”.
I’ve written a lot on outcome based commissioning and on outcome frameworks for an ACO, mostly about specific metrics
See the references. I won’t go through those again.
This will be the last, I hope, time I commit thoughts to paper. I’ll cover some thoughts on population level outcomes across multiple organisations.
Here are 4 quick thoughts to (hopefully) finalise my thoughts on this one.
Accountable to who, for what, over what timeframe
An important consideration in developing a framework.
In this ACS thing, there’s the obvious danger this is all about the “C” and “S” in ACS, but nothing about the “A” (apart from assurance to boards, who remain sovereign).
Key to “A” is that the choice of outcomes / measures need to be meaningful and reflective of / sensitive to the intended transformation / system changes.
And also something about the “S” bit is that the partnership needs to collectively work together for the outcomes at population level.
There’s also a fundamental point about single accountability for the health of a population. Tricky across a range of outcomes (or more likely process measures that serve as proxy outcomes) across many sovereign organisations.
Service or population outcomes. What different stakeholders are accountable for.
Those that provide and often commission services are mostly (not always) happy to be ‘accountable’ for ‘outcomes’ in those that use their services. That leads to consideration of what ARE the outcomes that it is reasonable to expect of the investment of current set of services, individually, vs a whole system.
As we commission services – it is unreasonable to expect service xxxx to be responsible for the system level outcomes. Maybe.
There is a tricky mismatch between the population outcomes a system will be judged by and what ACTUALLY happens in the services in that system.
But as commissioners (currently) commission services from providers and providers (currently) provide those services, whats the incentive for a provider of xxxx services to take on responsibility for things that its not responsible for – ie broad range of health outcomes.
Few want to be accountable for population outcomes – i.e. being responsible for outcomes in service users and those that don’t use services. Or where “outcomes” cut across many different services and organisations, and are part of a complex causal chain, and tricky proximal and distal issues.
Anyone fancy being accountable for healthy life expectancy, and the 25 year gap between best and worst? ……… no, just me then. It’s a lonely club.
Outcomes considerations for leisure providers vs physical activity
Consider this, our lack of population and individual sweatiness leads to poor outcomes. Leisure providers have a big role here. Consider a leisure provider would happily be accountable for good user experience in members.
What about being accountable for population inactivity.
Obviously this is the key point about service users vs population, and also control of all bits of the causal chain that get to a more active population.
There’s a side discussion – but an important one – here about efficiency, and the cheapest way to achieve a goal. The service provider perspective, the commissioner perspective and the population perspective are all different. This needs to be considered in indicator selection.
Then apply the same mentality in say cancer.
Is the outcome we want cancer mortality, cancer 5yr survival, or some process measure – say the % of diagnosed cases commencing treatment in 62 days.
All matter, some matter more.
No provider is responsible for cancer mortality, it is the metric that matters most.
The tricky business of the gravitational pull of the current must do framework
How many indicators can we cope with. A limited number focuses the mind.
There’s the obvious danger that in an ACS, accountability is to statutory boards then the ACS would never escape the gravitational pull of statutory targets 4hr waits, 18 week, 2 week, DTOC etc etc since that’s what trust boards (rightly) obsess about.
Thus the risk here being that the ACP just becomes a vehicle for delivering that.
And if we must be accountable for the existing national must do framework (which is fair enough given parliament expectations of the NHS), then that’s a lot of indicators that will give little or no space to population outcomes and inertia will continue in the current direction.
The crunch is whether our indicators focusing minds on the strategic long term objectives and shifts, or are they focusing minds on status quo system shape wise. (For eg secondary to primary shift, locus of primary care and place, moving away from core targets that are hospital centric).
Of course, pragmatically and innthe relay world we need to ensure we’ve the right mix of indicators around
- Must do indicators, some of which are plain perverse and drive the wrong sort of system, but are linked to parliament expectation of NHS
- Sustaining the performance of the system
- Transforming – what would the indicators be that would be suggestive of transformation. Two specific off the cuff examples – 4hr wait for GP appointment, % of the NHS budget spent in primary care (demand management).
- Outcomes – individual level or population.
Mainon & Davies provide an excellent commentary on some of the considerations re outcome selection (admittedly related to payment).
My take on those points is below.
What is it that you want your indicator to focus on – Quality / performance (against what standard) / efficiency / equity / volume
Performance standards – Absolute. Thresholds. Relative to peers. rolling programme of performance, retirement of indicators that have maxed out. half life targets. trajectories vs threshold
What are the Units of assessment
Process v outcome indicators. Timelag between process and outcome may be an issue
Complexity. Is it possible to attribute outcome to the input of the individual, a service, or are there other factors at stake.
Financial rewards linked to targets – Too small – not strong enough incentive. Too large – too risky, unpredictable results. Gaming. Tunnel vision on some areas / adverse selection / erosion of professional and ethical basis.
Can the system be gamed – overtreating / cream skimming? How easily? How can this be verified. Does the compensation system reward efficiency at the expense of quality. A constant trade off.
Investment for monitoring and verification – has costs. Other non non quantifiable aspects can also have an impact on behaviour – promotion, prestige, social context.
Methods of risk adjustment – So as not to penalise unfairly clinicians with more complex patients – poor, multiple comorbidities, complex health and social issues, BME.
Adequate sample for performance measure – No of patients included in data sample – random variability will be > in a small pool & risk of year on year variation, random effects
Patients with multiple providers – how does this look in indicator terms. Must be more nuanced that aggregating up individual level performance data – great performance in some areas might mask unacceptable in others
Literacy and health literacy – impact on outcome indicators can be considerable.
Propensity to seek health care varies between populations. An impact on performance
The effort needed to achieve clin meaningful change in some pt groups is considerably greater than others.
Data availability – Directly addressing the issue of the measurable with routine data vs the important stuff that may not be so readily measurable with easily available data.
Development work on metrics and leading indicators that will be tangible
Levels of granularity at which indicator data is available.
Get the clinical indicators right – eg from national clinical audits and similar? Always more nuanced than administrative data.
Where does CQC & regulator fit fits into this.
5) lastly some specifics
The assiduous will note I’ve avoided specifics….
There’s one reason, you’ve got to think about it, or if you aren’t prepared to think about it trust the view of someone else who has. I have. The National Academy of Medicine also have. They by are cleverer than me
Their suggestions for a streamlined set of 15 standardized measures can be found here
I’ve had a few go’s at outcome measures for ACO over last 6m or so
What outcome measure for an Accountable Care System – how will we know its working https://gregfellpublichealth.wordpress.com/2017/03/31/what-outcome-measure-for-an-accountable-care-system-how-will-we-know-its-working/
Outcomes based commissioning – https://gregfellpublichealth.wordpress.com/2016/09/10/on-being-an-outcome-based/ ……if you’re properly commissioning for outcomes, esp pop level, you’d commission something different. Very different.
Value in pathways versus value in populations – Don’t forget population health in your efforts to improve “efficiency”. Its important but remarkably easy to forget – https://gregfellpublichealth.wordpress.com/2016/02/21/value-in-pathways-versus-value-in-populations-dont-forget-population-health-in-your-efforts-to-improve-efficiency-its-important-but-remarkably-easy-to-forget/……If you want to improve stroke outcomes, prevention and management of population risk factors is the best bet.
Mainon & Davies – Payment for Performance. BMJ. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2234517/