A statistical approach to improving health care systems has recently become popular internationally. This technocratic approach to health is flawed when we do not consider the underlying political and social realities that undergird different communities and nation states.
One wintry night on the Gulf Coast, a biomedical engineer told me about a new god in health care. As the car perilously skated over icy bridges, the engineer told me of a mammography programme that assisted radiologists’ ability to detect cancer. “The computer can look at thousands of images, more images than any radiologist can ever see in his or her life. Then, given a certain n amount of data and repeat iterative loops, new images can teach the program to be able to predict with 99% accuracy which breast lumps are cancer and which are not.” But who gets screened? Who does not? And whose images make the programme? The engineer did not understand.
One sticky Texas summer day, my cousin from Bangalore repeated his faith in data. The IBM programmer’s focus was singular. “Algorithms are my passion,” he stated flatly. What about the world? What about society? “Nonsensical and illogical! India should be run meritocratically and competently like Google.” But how can you wish away thousands of years of culture, caste, and religion? “Simple. Assume a society with the following characteristics….,” he droned on.
Imagine a society? This was not some John Lennon, a man very much in the world but not of it. A sheltered and inexperienced young graduate student should not look at the world as a blank computer screen to be coded correctly. Neither India nor the world can be wiped into a clean slate. But my cousin is one in a long line of missionaries for the gospel of the data, a Veda of algorithms, and submission to the histogram.
Many important people in medicine think such things. Barack Obama once candidly declared health to be a human right; now audiences snooze as he rambles about “bending the curve of health care cost increases.” The blame for shifting the conversation from democracy and human rights to technocracy and management may go to his budget director, Peter Orzag. Or it may go to a Harvard surgeon writing about McAllen, Texas.
Using a database called the Dartmouth Atlas, Atul Gawande introduced Middle America to the bell curve of health care spending and quality. The two often do not overlap. Based on Medicare spending (not all insurance programmes), he stated that this poor town on the US-Mexico border had the highest spending per beneficiary in the United States, twice the national average, but had no comparable outcomes to show for it. It was an indictment of the culture of American medicine (at least out in the provinces). Gawande emphasized that by bringing low-quality, high-spending regions in line with high-quality, low-spending regions (like the Mayo Clinic), the US could get better care with less money.
Soon a spate of articles appeared in the press and in medical journals about cost curves and quality. Rankings of who did the best efficiently and who did worst inefficiently became popular. This parallels the development of rankings seen outside of medicine to determine the ‘best’ university, medical school, employer, or K-12 teacher. Everyone wants the best (according to the data).
But sometimes wanting the best all the time is not the best for everyone. Particularly when we do not really know what the best is.
The instability of ‘ranking’ systems is rarely mentioned by the curve pushers. Counterproductive consequences occur when targeting narrow measures. It is widely recognized by university presidents that the US News and World Report league table of universities has belittled the purpose of higher education by targeting endowment sizes and fictional staff to student ratios. But the ranking hysteria has seeped out of the ivory tower and gone international.
The Commonwealth of Massachusetts, a state with a public-private universal health care system that President Obama has introduced nationally, recently ordered a study to find the fairest way to rank hospitals by their (risk-adjusted) mortality rates. When four different methods of calculating hospital-wide mortality were found, the variations were striking. A top-ranked hospital in one method could end up in the bottom tier with another method and vice versa. What is the meaning of pushing the quality curve when we cannot agree on which curve to use?
A similar controversy erupted when the World Health Organization attempted to rank the world’s health care systems in 2000. They emphasized 5 factors (overall level of health, health equity, overall level of responsiveness, distribution of responsiveness, and the distribution of financial contribution). Critics argued that these five factors could just as well have been replaced by five other criteria.
Even the controversy about McAllen, Texas shows the danger of using incomplete data to make snap judgments about ‘good and bad’ spending regions. A less-publicized follow up article noted that data on the region coming from a private insurer revealed normal levels of spending. Your output is only as good as your inputs. As the research adage goes, garbage in leads to garbage out.
This would all be an academic parlour game if serious policy was not being based on it. And what ranking hysteria hides is as important as what it focuses on. Data fetish ignores two crucial factors: culture and politics.
The technocrats and curve pushers will simply say, “Study the positive deviants.” Even if we ignore the complexities of the curve, putting these positive deviants under the knife reveals more questions than answers. It brings out the importance of culture, an anthropology of often accidental excellence that the spreadsheet wielders fail to appreciate. Take the Mayo Clinic, the well-reputed Minnesota hospital held up as an example for the rest of America to follow.
But what is Mayo Clinic but a cultural creation? It came out of a combination of geography (favourable railway access but no local competition), an environment of honest Midwestern Protestant work ethic, and a very charismatic and far-sighted founding family[i]. It recruits the brightest physicians and the hardest-working support staff. The hospital expects its physicians to be team players, a rarity anywhere. It puts all its physicians on salary (anathema for twentieth century American medicine) and pays everyone in the same specialty the same after five years of experience. And even their patients are special; they are educated and motivated enough to travel hundreds or thousands of miles to see Mayo Clinic.
What lessons can the average general hospital manager learn from this? Not as much as the ranking gurus hope. Not everyone can get cream of the crop talent (medical, nursing, or otherwise). Not everyone can get enlightened administrators, dedicated to living up their institution’s historic legacy. And not everyone can get such solid profit margins either. With such top-flight inputs, how could the Mayo Clinic fail to be a top tier and efficient institution? Knowing what they have and we do not, it is also clearer why the rest of medicine fails.
Ranking evangelists tend to ignore how concentrations of power shape the world of health care. They seem to believe that an apolitical excellence will lead the way forward. I disagree. The Rockefeller Foundation cited four positive deviants in the developing world as models for good health care at low costs in 1984: Sri Lanka, Costa Rica, China, and the Indian state of Kerala. All four of these had significant accomplishments in basic primary health care because of strong political commitments to health and health care.
For example, the spectacular health accomplishments of Kerala are more due to the particularities of its society and politics than anything else. A strong commitment to participation in health, education, and nutrition by the general public did more than any particular programme to make Kerala the most socially developed state of India. A very active and organized Communist Party with sections in every village watches over the society, ensuring that education and food rations are equitably delivered to all religions, castes, and sexes. Generalizing this to the rest of India would have significant political and cultural obstacles even if all of the exact programmes, ideas, and leaders were stolen. Similarly, the social democratic culture of Costa Rica and the Maoist focus on rural medicine in China all had huge impacts on health but cannot numerically fit into any algorithm.
We ignore the power of institutions when we try to apolitically manage our way to better health care and costs. Massachusetts’s universal health care system runs huge bills because some basic political questions have not been addressed. Brigham & Women’s Hospital and Massachusetts General Hospital make up the health care giant Partners HealthCare. The attorney general for the state has been investigating Partners HealthCare for monopolistic and anti-competitive practices of price fixing. Collusion with local insurance companies allows the two hospitals to be paid twice as much as other hospitals for the same procedures, tests, and hospitalizations. That no one mentions that Gawande’s own hospital is a known culprit in health care inflation while citing potentially misleading high costs in South Texas is a bit rich. The power and prestige of two of the nation’s most famous hospitals warp the views of the technocrats. Better management and pushing cost curves will not fix this issue. Better politics will.
Philosophy and professionalism
The gospel of data fails to answer the following question: what is the weighting factor for justice, humanity, and compassion? The data remains silent when we speak of professionalism and philosophy.
None of this can be measured, and measuring itself is not the point. Foolish benchmarking systems have spread from university rankings to teacher ‘performance’ to health care. An industry of measurement has sprouted up to tell us what is good from bad.
But we cannot leave it up to the experts. What we measure is what we value.
We as societies have to decide philosophically what we want from our health systems. Any measurements thereafter have to be based on our collective vision as a society or on sound scientific practice. If the goals of the health care institutions are not in line with the health needs or expectations of the society it serves, the numbers do not matter. A health system’s performance comes not from satisfying the moving targets of quality and cost rankings. We do not need to pay someone to tell us right from wrong.
And where should physicians stand? When many in health care are conflicted about their goals we in medicine need to argue for the primacy of the patient and the dignity of all. We should re-commit ourselves to our professional ethics and moral code. The communal culture of academia and medicine must be protected from outsiders while professional bans on commercialism need to be more vigorously enforced. We should focus on the public interest and emphasize cooperation not competition when it comes to improving care. Expanding our traditional focus on the patient into systemic questions of power inequalities and social determinants of health (at least part of the time) would bring more meaning to the health care debate than any spreadsheet. We cannot assume away our past nor fit it into any algorithm. The gospel of data is a false god.
Martin Luther King once said, “The arc of history is long but it bends towards justice.” And that, dear readers, is the only curve worth pushing towards.