Notes on the epidemiology of multi morbidity, and why it matters

Multi morbidity the new black.
Well, it’s not really. But it’s one of the defining features of modern healthcare. And if you stick multimorbidity in a social context – with all its social complexity, it’s one of the defining features of contemporary health and social care.
Given that, we know remarkably little about the epidemiology of it.
Arguably the ship that launched this was the Barnet et al study in the lancet. I use this study at least once a month. It’s a modern classic.

http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(12)60240-2/abstract

It’s featured in many of my blogs – to remind you – ITS NOT THE AGEING POPULATION (I suspect I’ll need to say that a few more times yet).

The Barnet study gives us amazing insight into some aspects of the epidemiology, but not all. I’ve seen multiple iterations of similar to this study – most famously the SYMPHONY project in Somerset,
 @toniwilliams01 did it in Bradford using the data that was in the risk stratification tool that GPs and others were using.

Key points  

  • The Bradford data was consistent with Barnet. In the most deprived areas avg age of developing MM was fifties….
  • whilst prevalence of multimorbidity increases with age and highest in older people – absolute numbers highest in working age….this data underscores need for whole population approach
  • people living in more deprived areas ‘developed’ multimorbidity on avg 10 years before those on least deprived areas.

C/o @toniwilliams01



Implications of this data – four areas to start

  1. We can use this data plus risk strat to model need for intermediate care.
  2. We can better understand the mediators where we can intervene? Literacy and health literacy?..
  3. We can use to develop new models of care delivery for multimobidity and disability, 
  4. We might even think about prevention 



How is the epidemiology changing over time.

One of the burning pieces of work that – to my knowledge – remains not yet done is trends over time.

Given my grumpiness that we inappropriately use “ageing” to forward project certain doom, when we should use morbidity…..I’m really hoping someone somewhere is doing it.
It’s beyond both my competence and resource to do this epi study.
I did ask the team that did the original Barnet et al study if it had been done. The answer was not, sadly. The prof who got back to me did make some very helpful points and put me in the direction of some very useful references.

To summarise the issues in the email trail:-

  • Not being done in the UK,
  • the problem being that the prevalence that you get critically depends on the way you measure MM (which conditions, how many) and the dataset you use.
  • So the first attachment is a cross section based on CPRD where they get 16% (using QOF conditions alone) or 58% (counting all ACGs) whereas we got 25% (counting 40 conditions).
  • So you would need consistent measurement of MM in the same dataset, but of course datasets change over time (better recording will make the prevalence go up). Doesn’t really matter how you count MM though in the more general sense that all studies show similar things (commoner in older people, commoner in the more deprived, mental health interactions with physical in various ways etc.

 

Studies of note

He pointed me towards some studies. These are worth bringing to your attention.

1)

European Journal of General Practice. 2008; 14(Suppl 1): 28 32

Multimorbidity in primary care: Prevalence and trend over the last 20 years

Uijen & Van De Lisdonk. https://www.ncbi.nlm.nih.gov/pubmed/18949641
“increasing age, female sex, and low socio-economic class are associated with an increasing number of patients with multimorbidity. The prevalence of chronic diseases doubled between 1985 and 2005. The proportion of patients with four or more chronic diseases increased in this period by approximately 300%.

CONCLUSION:
The increasing amount of multimorbidity in primary care as well as the increasing number of chronic diseases per patient leads to more complex medical care…….”

Points:-

  • uses a Dutch research registry,
  •  it’s a “patients consulting” rather than “registered population” analysis (both are wrong in different ways of course).
  • rise, both in crude rates (which could just be ageing populations) and in age-sex standardised rates (which could be because of better survival from acute events, so live to have an LTC; could also be improving recording but to my knowledge this is a pretty tightly managed registry,
  • so that is less likely than it would be in routine GP data in the UK over the same period).

2).  Epidemiology and impact of multimorbidity in primary care: a retrospective cohort study

Chris Salisbury, Leigh Johnson, Sarah Purdy, Jose M Valderas and Alan A Montgomery

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3020068/
“Sixteen per cent of patients had more than one chronic condition included in the Quality and Outcomes Framework, but these people accounted for 32% of all consultations. Using the wider ACG list of conditions, 58% of people had multimorbidity and they accounted for 78% of consultations. Multimorbidity was strongly related to age and deprivation. People with multimorbidity had higher consultation rates and less continuity of care compared with people without multimorbidity.
Conclusion

Multimorbidity is common in the population and most consultations in primary care involve people with multimorbidity. These people are less likely to receive continuity of care, although they may be more likely to gain from it.”

I’d encourage you to look at table 2

 

 

shows the independent relationships between consultation rate and age, sex, and deprivation before and after adjusting for multimorbidity. It demonstrates that the relationship between age and consultation rate is reduced, and that between deprivation and consultation rate almost disappears, after adjustment for multimorbidity. This suggests that the main reason that older patients and those in deprived areas consult more often is because they have more chronic health conditions.

 

For the epidemiology geeks (you know who you are…….) there’s also some useful content in the appendix on coding and classification

3) The rising tide of polypharmacy and drug-drug interactions: population database analysis 1995–2010

Guthrie et al. BMC Medicine 2015

http://bmcmedicine.biomedcentral.com/articles/10.1186/s12916-015-0322-7
This is a repeated cross-sectional analysis of community-dispensed prescribing data for all 310,000 adults resident in the Tayside region of Scotland in 1995 and 2010.

One might conclude changes in poly pharmacy as a proxy measure for multi morbidity. Of course that’s full of holes and flaws of all sorts (changes in prescribing predilections, over-diagnosis and over treatment etc). But it’s potentially useful data, and should be considered with the caveats.
See table 1

Numbers and class of drugs dispensed to adults in 1995 and 2010
As is Figure 1

Number of drug classes dispensed in the 84-day period in 1995 and 2010 by age of patient.


The conclusion is appropriately caveated. Go read it.







Can we do this analysis of epidemiology of multi morbidity over time – some thoughts.

It should be done. It can be done.

  • This could be done with CPRD or similar
  • Would need a v granular dataset with coded data for every consultation with a GP or primary care nurse, & external auditing/management of coding.
  • There would be a number of issues to contend with, most notably external factors such as QOF that changed the nature of general practice that may skew an analysis of morbidity over time per se (inbuilt incentives to case find “new” disease). Changes in QOF and other incentives over time will change the way we code things, and act. For example AF, CKD and the way in which we handle eGFR has changed the morbidity of CKD, whether this is a real change in epidemiology or “real” morbidity is moot. One might say the same about the QOF disincentivisation to record depression.
  • You could use the Electronic Frailty Index. I don’t know how far back the data has been crunched. And last time I talked to Andy Clegg on it he was clear it is a clinical tool, not an epidemiological one. I’ve heard one argument that the EFI isn’t really a frailty measure at all but rather a morbidity count which is a pretty good predictor of mortality, hospital admission and nursing home admission) and coprescribing,
  • We can also use the data that drives our risk stratification systems

Each has flaws and holes.
Profs guess is that multi morbidity is increasing, but that ageing populations are more important than increasing age-standardised rates. But he did say that is just a guess.




Why this matters

Someone somewhere definitely should do this. It is important in terms of framing (or more to the point reframing) the debate on “the ageing population demographic timebomb” myth.

The population is ageing, no doubt there.

If we think it’s all about something we can’t do anything about (ageing) we will simply prepare for a need for bigger shinier hospitals and the like. If we frame our narrative in terms of something we can do something about (i.e morbidity and multi morbidity) we sort out the risk and demand management first (like general practice and social care) and we might even give some Thought to prevention?

Take home messages:

Steve laitner later summarised pithily why this matters especially in our laziness around blaming the “ageing population” 

  1. really understand the segment needs & the individual needs 2) don’t do “bolt-on services” but complete care model
  2. Age (re health) is just a proxy measure of the amount of illness & disability we might have accumulated during our time on earth?
  3. AGE! – for when you don’t have the evidence or energy to really understand the changing health and care needs of society

 
THAT is why getting a handle on the epidemiology of multimorbidity is so very important!

Advertisements

3 Comments

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s