It is time, we are told, to rethink epidemiology. It is time to rethink our quest, our concepts, and the nature and origin of our commitments. That is a lot of thinking, stretching across the broad landscape of our professional practice. Even our metaphors—this blind quest for the Philosopher's Stone1—should be rethought, as well as our concept of cause, our much-maligned criteria of causation, our signature methods, and, it seems, our philosophical foundations. It is time, we are told, to step away from all that is associated with the natural laboratory of observational science, once the source of so many successes but now fraught with ‘profound and fundamental problems’, to use the precise language of Drs Buchanan, Weiss, and Fullerton.
This is no ordinary rethinking. We need innovative, creative, out-of-the-box thinking, for we have been, to borrow a few familiar phrases, barking up the wrong tree, beating a dead horse, dreaming an impossible dream.
It has been 5 years since the human genome was mapped. Long enough, it seems to Drs Buchanan, Weiss, and Fullerton, to determine that epidemiology cannot fulfil the promises that were made: the improvements in diagnosis, prevention, and therapeutics and a new kind of medicine that focuses on the ‘very essence’ of individual lives, guiding each of us along our unique trajectory, predicting precise risks for each and every one of us.2 Epidemiology, we are told, cannot even provide society with the kind of knowledge science itself expects: consistent, reproducible results about the causes of all kinds of important concerns, such as cancer, heart disease, stomach ulcers, colds, autism, schizophrenia, syphilis, health care costs, healthier Americans, even aging, and, in the words of Drs Buchanan, Weiss, and Fullerton, ‘you name it’ when it comes to the causal effects of genes.
Apparently, epidemiology was once a legitimate science of disease causation, but no longer. Who has not heard that we have found all the strong associations, with only the weak left to be discovered? To these dubious assertions about the past and not-yet-tested prognostications of the future, Drs Buchanan, Weiss, and Fullerton add a particularly chilling condition: even very large cohorts like those proposed for the Human Genome Project and by the investigators developing all those ‘biobanks’, cannot help the situation.3 The underlying structure of biology connecting genes to individual traits—the ‘hour glass’ with gravity-defying grains of sand—prevents us from understanding anything new about weak biological effects using epidemiological studies.
Small wonder we must rethink epidemiology, trying to understand how things could have gone so wrong in a present that is exploding in our faces. Let us begin by rethinking our quest: the metaphorical search for the Philosopher's Stone.
Rethinking the quest
We are not stuck, it seems to me, behind thick wooden doors in some medieval ivory tower, hunched over bubbling potions in an alchemist's hideaway, futilely experimenting with one substance after another. We have actually found what we were looking for, preventable causes of diseases, some with strong effects others more modest in their effects. Our critics have done us a disservice by not including a table of accepted risks (some manipulable, some not) for the so-called complex diseases. It would be a very large table indeed. Epidemiology has made (and will continue to make) many discoveries. Ironically, so did the alchemists. My favourite story is the discovery of a highly flammable powder—phosphorous—from the distillation of a certain excretory liquid. It did not turn metal into gold but it did wonders for the waging of wars and for fireworks.4 We should remember that discovery is as much an accident as it is a product of systematic inquiry.
So we are not searching for the Philosopher's Stone. If our critics are correct, things are much worse. We are more like dangling at the end of a rope, hanging on for dear life over a high seaside cliff, our only hope to climb back up from where we have come, retracing our steps. One of the best ways to understand why things are as bad as they are now, in the present, is to navigate history back to its source, ‘always struggling to get a better understanding of the events that have brought us where we are now. It's a lot of work, requires great perseverance, you have to keep the rope pulled tight all the time.’5
Epidemiology's rope is not only pulled tight, it is fraying and about to break. We must climb up and away from the promises made, although I have no recollection of making them. Someone promised a ready, universal explanation for why we sicken and die, all tied up in that spectacular staircase so carefully and completely described, base after base, and sitting precisely at the centre of our cellular universe: the genes that our methods cannot turn into the golden causes we seek. We have now been ‘properly advised’ by Drs Buchanan, Weiss, and Fullerton, to stop this approach and ‘seek our riches elsewhere’.
Once again, epidemiology has failed to live up to everyone's expectations. There is nothing gained by finding more inconsistent associations, creating statistical models of dubious distinction, and exploring the ever-expanding reach of technology. We must, we are told, recognize the hopelessness of our situation, the futility of blind commitments to concepts and methods that worked once but no longer, cannot work, except for diet and exercise, a peculiar and puzzling assertion by Drs Buchanan, Weiss, and Fullerton. But it does not really matter. We must climb or fall into the seething sea below.
Our climb from the present back to the past, from conclusion to premise of this extraordinary paper, begins.
Whether the complex disease of interest is of early onset or late, or whether it has a stable or an unstable prevalence, we must, for all diseases and all their causes, rethink our concepts and methods. And the usual suspects helping us to think about thinking—the philosophers—are not useful either. We have run through their ideas, one by one and are left with only one peg upon which we can hang up our alchemist's pointy cap, our blind ideological commitment to old and ineffective concepts and methods: the well-worn causal ‘criteria’, Bayesian inference, probabilistic causation, the sparse fallibility of deduction and falsifiability, induction, strict empiricism, and, even, classical intuitions. This briefest history of the philosophy of science, condensed from a few paragraphs to a string of labels, is cryptically displayed like graffiti on the cliff wall as we inch our way past. It is hard work this climb. I am grateful that diet and especially exercise can be relied upon, as I struggle up this fraying rope, one error at a time. My hands and head are aching from all this rethinking.
Biases, both analytical and from the vagaries of the publication process, face me next, then that old nemesis, the fallacy of ecological studies, the starting point of so many successful epidemiological stories.
Just in time, I have made it to the top. I have made it back to where we in epidemiology began this journey not so many years ago: at the crossroads.6 Exhausted from the exertion, I fall asleep and dream….
At the crossroads I am
scanning the horizon
What choice have I
with knowledge gained
Which way shall I
a prudent judge declare
To my left a hazy path
to my right a road well traveled
too clearly marked to miss the cells
they beckon us inside to see
where genes betray our hidden selves
and science reigns supreme
To my right biology
to my left society
shadow figures cram the path
their silent waltz a grey ether makes
the dust from shuffling feet
obscures our genes
still, science reigns supreme
In slow deep breaths
shall I proclaim
which road to take
or better yet
cast myself aloft
no rope to stop this daring dive
Return I will
to the sea of person time
upon whose waves I have
in pursuit of a golden tool
lost my ship
The sea upon whose shore
fair lady sits
on a bleached black box
weaving her stories
skeins of tangled lives at her feet
put not to the loom
a quilt her final aim
Each thread: knowledge
no slender knot it grows in time
then set loop by loop into patches
each a life in silent chromes
from blessed birth
to death's magnificence
She has sewn some for herself
for a mother destroyed
by the crab that lingers
a part yet not quite of self
Another for a young nephew
in fatal sleep his heart denied him
the wild thump of the chase
the love we cannot calm
the pain paid at the frame
where each life, sea-rinsed
In flight I am
merged with spray ascendant
jeweled patches piled high beside her
she waves me on
Rethinking our metaphors
The language of metaphors can be found in everyday conversation7 and in our professional practice. We read in the paper that once again an epidemiological study has grabbed the headline. We fight traffic on our way to the office, where we read in our books and journals that we are at the crossroads, black boxes in our heads, and now, on a quest to find the philosopher's stone. We might be tempted to ignore these literary devices as so much fluff if only they had not generated so much discussion over the years, serving as focal points for debate at best and as too simplistic labels at worst. To put it another way, there are good metaphors, useful, even inspiring, metaphors, and there are inappropriate and misleading metaphors. None are perfect; some are better than others. The ‘crossroads’ metaphor, for example, is a clever way of depicting the two directions epidemiologists can go: molecular or social, assuming of course that there are only two choices. Those who lean towards the ‘left’ and the development of a reinvigorated social epidemiology must be pleased with the quest metaphor, sealing the fate of molecular epidemiology, which was, in turn, a response to the complaints about epidemiology's black box mentality, risk minus mechanism, an epidemiology at the limits of its capacity.8–12 These are, then, connectable metaphors. They tell a story. But it is not everyone's story.
Some of us cannot choose either path at the crossroads. We see no advantage in forcing a decision between the study of biology and of sociology. Both ends of the spectrum (if these are, in fact, at the ends) may explain the distribution of disease, and it is explanations that we seek. It is a tall order (another metaphor) for epidemiology to expand its reach along both paths, but we have a broad mandate and a deep responsibility to find interventions within these explanations to be used for prevention and (as a last resort) for treatment.
We should not be troubled by black boxes. They are, for the conceptual engineers—the rethinkers among us—a useful concept at the heart of systems theory.13 If, for example, we let the black box stand for creativity, that essential human trait crossing over the border between science and the arts, seeking in both domains unity from variety,14 then we can allow it to reshape a metaphor for epidemiology worthy of the complexity and the mystery of the human condition. That, for me, would be the journey across the sea of person time, exploring and fishing its depths, diving and dying beneath its surface.15
Rethinking our concepts and our commitments
It is time to put aside the metaphors, the quest and the gold, and all the rest. It is time to rethink our concepts and our commitments.
The story we have been told is that the aim of science in this genomic era is to predict risk in the individual. The concept of cause best suited for this aim is that which characterized the aetiology of the infectious and Mendelian diseases, the highly specific, strong cause, which, we are told, is no longer discoverable, because (as so many of us repeat mantra-like) there are none left to find and because our genes and our diseases are connected by something like an ‘hour glass’ that prevents our methods from working, whether these are our analytic methods or our synthetic ones. The scientists, we are told, are so overcommitted to these methods (driven by the hope of finding yet another strong and specific cause) that they cannot see the forest for the trees, but Drs Buchanan, Weiss, and Fullerton can. They see the futility in intensifying the approach—the ever-larger cohort study—but the epidemiologists do not, our commitment arising from the way science as a social activity works: as a hotbed of ideological conformism, resistant to change, irrational in its exuberance, fed by ‘vanity, vested interest, hunches, experiences, politics, careerism, and (even) imperfections’.
All the pieces of this story can be found in our literature: the aim of predicting individual risk, the need for larger cohorts, and the inevitability and the challenge of weak associations. There are, as well, papers written by prominent practitioners outlining the imperfections of others, although I am reminded of sayings too trite to repeat of stones and glass houses and black kettles.
For commitment and conformism there are many sources in science and in philosophy. Forty years ago, Barber16 exhaustively categorized sources of resistance to change within science, although he also cautioned against exaggerating the extent to which it always occurs. Bartley,17 a Popperian protégé, writes of the retreat to commitment in science, offering a criticism-soaked rationalism as a way out. Blackburn18 warns us against the ideology that prohibits the kind of rethinking we are being asked to provide here.
But the story we have been told, of our outdated concepts and blind commitments, like all stories, can change in the retelling.
The aim of science is to explain the world, to gain an understanding of it, to generate a form of knowledge, not certain nor proven, but well-evidenced, carefully considered in an open forum by a rather specialized community of practitioners, and subject to public scrutiny. The aim of epidemiological science—a vital but incomplete characterization of the overarching aim of the profession—is not to predict individual risk nor is it to predict risk in a community; prediction serves the scientific aim of epidemiology as one component of our capacity to test our hypotheses, and these in turn arise from our explanations. We are not particularly good at making predictions nor building explanations, which look too much like lists of factors, some lifestyle, some demographic, others environmental or occupational, some physiologic, a few (but increasingly) genetic or molecular. Whatever, it seems, has made it through the methodological process that should and can be improved—a topic for another time.
My commitment, however, has nothing to do with our methods and everything to do with the overarching aim of the profession: to care for the public's health. My commitment to and my responsibility for the public's health do not emerge from ideological conformism, but rather from tradition, experience, and what I take to be an essential humanitarian ideal: that health is a fundamental social good.19
Perhaps, as I look back at this critique, we epidemiologists need to be goaded into one more ritual self-flagellation to achieve a greater level of insight, crawling on our knees up the rocky slope of scientific discovery. I doubt it. I cannot think of a better way to turn young creative minds away from the professional practice of epidemiology than to show them one more paper that declares our impotence. Had I known 25 years ago what a mess epidemiology would find itself in, perhaps I would have stayed in clinical medicine or engineering where I began. But then I remember the successes and the progress that has been made during those same years, for which I take no credit but have immense pride in, and I realize that there is no better science to understand the causes of human disease than epidemiology, partnering with all others needed to get done the job that needs to be done. There is no magic to it, no impossible dream, nothing more precious than our commitment to disease prevention and our willingness to do our best to achieve that aim. That is our promise to society and so our responsibility.
And that is what I think.
Thanks to Dr Graça M. Dores, for her insightful comments and judicious edits.