I said I’d never write on this again.
I tried to summarise what I’ve done across umpteen blogs into a single slide set. It is here
But have been persuaded to summarise further. Here goes (often ideas shamelessly nicked from others, I can’t recall who so apologies for the plagiarism where it happens)
1. Core elements. For me the core elements are
use of data linkage -better understanding of risks and service use patterns
Build own capability or buy it in
ability to predictive model
How to segment / what structure. Population focused segments not service design focused.
Personally I’m a fan of a model built around Bridges to Health.
2. design of service response to manage risk in segments.
Decision needed on what / how to segment
Keep it simple and meaningful to the frontline.
3. whole pop approach, clue in the title. POPULATION
not just those top 5%, and not those with disease x or in care of organisation y
all parties should consider where is the investment in primary care, focused investment around the practices with most deprived populations.
4). What might it look like every day
Steve Laitner summed it up recently….
Imagine everyone in a certain locality was working together and pooling resources and paid based on shared outcomes for the populations they served and not on activity through their services.
Incentives for preventing and delaying loss of health
We need that system to be able to
Data – Collect – the right info, including citizen generated data. (Big cultural change needed).Aggregate – that info, make it useable. Understand – Describe the need at neighbourhood level. Outcomes, map treatment pathways and monitor performance. Segment and risk stratify the population into groups
Care model Care management / Workflow optimisation / QA and monitoring
Operations – Process redesign / Payer risk contracting / Provider education and coaching / Top of licence staffing procedures / Reimbursement based on population value
5). Specific interventions
If I had to do specific interventions, I’d focus on MM and systems for multi morbidity.
Target interventions :
Over diagnosis / treatment. Overall burden of treatment. Problem prioritisation, goal setting, and shared decision making
Person centered care & consultations. Training and competency at person level (patient oriented and clinician oriented) and systems designed to enable person centred as default.
geriatric assessments & personalised care planning.
Exercise / muscle mass / CV fitness / frailty
Don’t be all medical model. Think in terms of how communities really operate, focus on asset based approaches to well being as well as “disease management”.
6). Clusters and prevention models focused on clusters of conditions
if I had to think about PHM in a preventive mindset it would be about clusters of multi morbidity
Thinking in a prevention frame…. not just primary prevention. The delay & prevent mindset.
what’s the most likely LTC to develop first – presumably diabetes. If you have diabetes what’s the next likely LTC you’ll get? – presumably CHD.
If you develop COPD first what’s the next thing you are going to get? – probably heart failure.
If you get CKD first … etc etc.
we all know what the risk factors for some of this stuff. Make sure we achieve good and equitable coverage of the right risk reduction / asset building interventions. Prof Bentley has written eloquently on this and given us a methodology.
7). What to focus on
I’ve heard a line that we will take a PHM approach to condition x,y, or z.
The clue is in the title – population, health, management. ALL three elements matter.
For me PHM should focus on populations not topics or diseases of interest
we needs to be careful we don’t end up dicing and slicing by diseases or groups of not pop and person focused risk segmenting
8). Other stuff
segmentation and Care models based around neighbourhoods. It, as a concept, MUST be owned by places and rooted there, not just around the NHS as an institution.
There’s an ask of the centre that one bit of NHS is relentlessly flogging RTT and 4 hr wait, DTOC etc…. which is inconsistent with PHM. Both together does not compute
Thanks to Nigel Slone, Rupert Dunbar Rees, Steve Laitner, John Soady, Rupert Suckling and all others that I can’t recall stealing specific ideas from.