Intended for healthcare professionals

Education And Debate Complexity science

The challenge of complexity in health care

BMJ 2001; 323 doi: https://doi.org/10.1136/bmj.323.7313.625 (Published 15 September 2001) Cite this as: BMJ 2001;323:625

Let them eat complexity: the emperor's new conceptual toolkit

I admire the writing of Chris Manning, and even agree with a great
deal that he has to say. I admit I’m a little uneasy with his “gospel”,
but am happy to rally behind his call to “rise above [the] imposed
systematisation… burn out and disillusion so often occasioned by the dead
reckoning of centrism”. Great stuff. His enthusiasm for Plsek and
Greenhalgh’s article (BMJ 2001; 323:625-628)
is, however, misplaced. Rather than emancipating imagination and
creativity in health care, I fear that it is more likely to appeal to the
very “dead reckoning of centrism” that Manning opposes - though I
appreciate that this is not Plsek and Greenhalgh’s intention.

I suppose contemporary NHS managerialism has to have its own body of
knowledge and suite of ‘techniques’ to bolster a sense of expertise, but
it could do better than borrow from the wilder shores of pseudoscience.
Dr. Manning does not need to appeal to the natural sciences to make his
point. Plsek and Greenhalgh, on the other hand, mangle the legitimate
science of complex adaptive systems in an attempt to apply a veneer of
intellectual respectability to some ‘homey’ generalisations. To be fair,
this sort of treatment is not original to Plsek and Greenhalgh, and has
been regularly promulgated amongst US business management organisations
for at least a decade. Essentially, Plsek and Greenhalgh continue the
tradition of misusing scientific concepts by confusing technical terms
with everyday ideas, in the manner described (and exposed earlier in a
celebrated hoax in the journal ‘Social Text’) by Sokal and Bricmont
(1998). It has become commonplace to twist findings and theories generated
originally by the physical sciences in the last century (complexity
theory, chaos theory, catastrophe theory, non-linear dynamics and so) into
a caricature which serves political and careerist, rather than scientific,
ends. The anti-rationalist outcome has rather more in common with
nineteenth century romanticism than the sophisticated, post-modern
thinking that its proponents imagine they are practising. Plsek and
Greenhalgh are ostensibly a little more sober and cautious than some
writers in the field – they suggest in an early paragraph that the “new
science of complex adaptive systems may provide new metaphors (my
italics)”. Unfortunately, they continue in a vein that implies that the
potential contribution of complexity theory to healthcare is anything but
metaphorical.

More generally, the popular narrative [sic] goes as follows:
complexity science is a spooky revelation of the interconnectedness of
things. It demonstrates the limits of narrow-minded ‘conventional’
science, laughs at the foolishness of deterministic thinking, celebrates
unpredictability, and values imaginative ‘non-linear’ perspectives. It
spurns the empty clockwork thinking of the drab Newtonian universe and
promotes ‘intuitive thinking’. It recognises that reductionism destroys
rather than reveals (or as Wordsworth, an appropriately romantic source,
put it, ‘murder[s] only to dissect’). Best of all, has exciting ‘cool
sounding’ constructs like “strange attractor” to deploy.

In fact, complexity theory doesn’t hold ‘values’ as Plsek and
Greenhalgh suggest. It tells us about the limits to prediction, not the
limits to scientific method. These limits lie in the accuracy of
measurement and computational power, not the well-known limitations
inherent in the conceptual framework of Newtonian mechanics (which,
ironically, is where complexity theory is firmly grounded). Complexity
theory is about deterministic systems which are unpredictable because, as
Plsek and Greenhalgh state, of their “sensitive dependence on initial
conditions”, but not because there is something ‘paradoxical’ or
‘mysterious’ about them. The point is that small changes can have big
effects – the usual example given is that the atmospheric disruption
caused by the flap of a butterfly’s wing may result eventually in a
tornado on the other side of the world - as the effect on the atmosphere
ripples out and magnifies in a ‘non-linear’ fashion. Though the effect is
deterministic, Newtonian, and subject to scientific analysis, it is
unpredictable because it would be impossible to measure sufficiently
accurately the pre-flap situation, or muster the computational power
required to model the consequence and so on. Above all, the science of
complex adaptive systems does not argue in any sense against ‘reductionist
thinking’: the theory is, in fact, a product of it. Contrary to Plsek and
Greenhalgh’s view, formal ‘reductionism’ is perfectly capable of being
surprising and creative – and usually is (see, for example, Churchland,
1986). Unfortunately for Plsek and Greenhalgh’s thesis, it is also often
counter-intuitive (which is the one of the uncomfortable reasons why some
‘guidelines’ don’t ‘work’).

Moreover, Plsek and Greenhalgh’s example of “complexity in health
care” is absurd. Do they really encourage us to believe that if only Dr
Smith had some grounding in complexity theory, she would have been able to
understand why getting rid of lunch time upsets her staff and colleagues?
Even more amusing examples have been published recently in the BMJ. Here
is a short extract from “Bridging the Quality Chasm” (Kelly and Tucci,
2001), where the concept of the ‘Strange Attractor’ is misrepresented:

“In complex systems and chaos theory human behaviour is influenced by
"strange attractors," which are often hidden or poorly articulated values
or needs… To influence the elements of a complex adaptive system such as
health care, one must understand how such systems differ from machines.
Take the problem of throwing a rock and getting it to land where one
wishes. Understand the mass of the rock, the distance of the target, the
force of gravity, etc, and one can calculate the force and trajectory
needed. Try the same approach throwing a bird and the results will be
different. The complex behaviour of the bird becomes intelligible once we
know that birds are insatiable food seekers; we then know how to influence
their behaviour, for example, by placing food where we wish the bird to
land. We have used our knowledge of the "attractor" for this element of
our system to both understand and influence its complex behaviours.”

Of course hidden or poorly articulated values or needs influence
human behaviour, and can have large and even surprising effects – but this
need have nothing to do with chaos theory. There is nothing ‘strange’
about the attractiveness of food to birds. Interestingly, Kelly and Tucci
are under the impression that chaos theory has something to say about “how
such systems differ from machines”. Here, I suspect that the authors are
confusing naïve ideas about the nature of ‘reductionism’ (the ridiculous
but populist notion that reductionism is about ‘people’ being ‘reduced’ to
machines’) with metaphorical concepts about chaos theory gleaned from the
brand of business management seminar that perpetuates such poorly
articulated ideas. In fact very simple ‘machines’ (such as a ferrous
pendulum swinging over a magnet) produce complex, chaotic behaviour and
demonstrate the phenomenon of the ‘strange attractor’. Indeed, this is how
the concepts of chaos theory are usually introduced to schoolchildren.

This sort of writing could have some nasty side effects: one can just
imagine hapless NHS employees being shepherded into syndicate rooms after
a seminar entitled “Watch out – there’s a linear thinker about” haltingly
delivered by the visiting chief executive of some obscure NHS Trust, and
being forced to throw rocks, innocent birdlife and handfuls of ‘Swoop’
around the place in a ‘leadership’ exercise. “No, no, NO, Dr. Smith”, the
facilitator scolds, – “you are not allowed to throttle the bird before you
throw it – that’s exactly the sort of outdated reductionistic problem
solving that we are trying to discourage”.

We all have to tolerate, even celebrate, paradoxes on occasion and
some dilemmas do indeed have to be lived rather than solved. We do not,
however, have to appeal to the science of ‘complex adaptive systems’,
‘chaos theory’, ‘catastrophe theory’, ‘Einstein’s general theory of
relativity’ or even ‘quantum mechanics’ to appreciate this. While Plsek
and Greenhalgh’s aim may have been to make some fairly abstract science
more accessible, the result is misleading and potentially harmful. The
paper fails to articulate accurately the background and basis for the
(conventional) science of complex adaptive systems. The repeated switching
between metaphor and science is unhelpful. Useful, non-metaphorical
applications of chaos theory in the biosciences (analysis of cardiac
electrical rhythms, the EEG in epilepsy, sugar levels in diabetics) may be
swamped by the intellectual snake oil of ‘metaphorical complexity’, which
we can expect to see spattered across massed ranks of flip charts by
healthcare administration faddists.

Plsek and Greenhalgh appear to authorise, in a quasi-scientific way,
a means by which uncomfortable situations (e.g. ‘tension’ caused by poorly
run services) and expensive solutions (e.g. the application of
‘conventional’ science) may actually be dismissed as a ‘dilemma to be
lived’ rather than addressed and analysed – a handy conceptual toolkit
indeed for the credulous healthcare manager on an inadequate budget.

Churchland PS (1986) “Neurophilosophy: toward a unified science of
mind-brain” MIT Press

Kelley MA and Tucci JM (2001) “Bridging the quality chasm” BMJ 323:61
-62.

Plsek PE and Greenhalgh T (2001) “Complexity science: The challenge
of complexity in health care” BMJ 323:625-628.

Sokal A and Bricmont J (1998) “Intellectual Impostures” Profile
Books, London.

Competing interest: none declared

Competing interests: No competing interests

23 September 2001
Ian Reid
Professor of Psychiatry
University of Dundee