Imbens and his co-winners were awarded the prize for their methodological contributions to the analysis of causal relationships.
Doctors often complain that patients don’t follow instructions. But when it comes to their own medical care, physicians do not necessarily do a better job than their patients.
In a new working paper published Monday by NBER, Petra Persson and Maria Polyakova — faculty fellows at the Stanford Institute for Economic Policy Research (SIEPR) — and their co-authors analyze data from Sweden and find that, on average, doctors and their close relatives adhere to medical guidelines just over 50 percent of the time. The average patient complies slightly more than 54 percent of the time.
Those compliance rates are low to begin with, but the 4-percentage point difference between the two groups is especially surprising, Persson and Polyakova say. Many health care experts think people don’t heed medical recommendations because they can be too complex, or patients don’t trust their doctors or want to avoid insurance co-pays. But the study’s evidence that doctors, too, deviate from protocols indicates something else might be happening as well.
“Doctors are the most informed patients,” says Persson, an assistant professor in the Department of Economics at Stanford’s School of Humanities and Sciences. “If medical recommendations are too complex, for example, or the patient-provider relationship is bad, this shouldn’t affect doctors themselves, so we would expect them to adhere at much higher rates. Instead, we find the opposite.”
The researchers say that the finding raises many new questions, including why doctors are more willing to veer from official guidelines and if such deviations generally benefit or harm patients.
Polyakova, an assistant professor in the School of Medicine and core faculty member at Stanford Health Policy, says the implications are especially important as health care evolves from a one-size-fits-all approach of the guidelines toward a “precision medicine” model where treatments are customized to each patient. Guideline deviations by experts and their family members could indicate that they are customizing the treatment decisions to their own circumstances.
Medical guidelines, which are usually created by physicians for physicians and grounded in evidence, function as best practices for health care. Until 2018, the federal government maintained a database with more than 1,400 recommendations for preventing, diagnosing and treating health conditions. During the COVID-19 pandemic, many of us have become intimately familiar with them: Guidelines aimed at preventing infections are the reason we have endured lockdowns and, in public, worn masks and stayed six feet apart.
Besides setting basic standards of care, they serve another crucial purpose, according to Persson.
“Think of them as one way to lift the lower end of healthcare distribution so that everyone is getting consistent, evidence-based care,” she says.
For Persson and Polyakova, investigating doctors and their family members as possible contributors to low adherence rates was not only novel, but also a natural next step in their research agenda. In previous collaborations, they have found that patients with a close relative who is a doctor have better health outcomes (and are more likely to become doctors themselves).
For this latest study, Persson and Polyakova — along with economists Amy Finkelstein of the Massachusetts Institute of Technology and Jesse Shapiro of Brown University — collected extensive administrative data on the entire population of Sweden from 2000 to 2017. Using these and other government records, they were able to identify doctors and their closest relatives and people who did not have direct access to medical advice. They then measured adherence to 63 government-issued prescription drug guidelines, including recommendations for antibiotic use, elder care, pregnant women and patients with specific diagnoses.
Their analysis — which controls for demographics, income and education — shows that the average patient followed medical standards 54.4 percent of the time. That average rate was 50.6 percent for people with access to a doctor in the family. Doctors themselves had even lower adherence rates.
Antibiotic use saw the largest deviation. Doctors and their immediate relatives followed guidance urging that infections first be treated with antibiotics targeting a narrow set of bacteria on average 6 percentage points less than other patients. Instead, they opted for antibiotics that attack larger numbers of microbes.
Persson and Polyakova say doctors are likely to deviate on antibiotics use because they know that the guidelines are also motivated by public health concerns over evidence that bacteria become resistant to antibiotics the longer and more often they are used. When it comes to the treatment of their own infections, or those of their family members, they may opt for the treatment that they know is most likely to cure them the fastest.
In the paper, Persson and Polyakova also look at reasons why this may be — not just for antibiotics but across most guidelines. They rule out the possibility of unobservable differences in income, for example, by showing that other high-earning professionals tend to follow recommendations. They show, too, that the gap isn’t because doctors and their close relatives are relatively more comfortable with taking medications not specified in guidelines; they are also less adherent to guidelines advising in favor of taking particular medications.
The conclusion that the researchers say makes the most sense is that doctors and their nearest relatives are better informed than most patients and adapt their care accordingly. Doctors know that medical protocols tend to be conservative and are based on patient averages, which is why they are more willing to ignore them. The study also finds that they are especially prone to deviating when the medical evidence behind a particular recommendation is weak.
In a separate paper released this spring, Polyakova showed how people’s perceptions of risk can influence how they respond to health mandates.
The results on antibiotic use support the idea that doctors respond based on what they know. Urging targeted medications as the first step in treating infections is rooted in public health concerns about antibiotic resistance in society generally — not what’s best for a particular patient. Doctors know this, so they are more likely to take a broader-spectrum antibiotic when sick and to prescribe one for family members.
The same dynamic could be at play in other instances when doctors steer clear of formal recommendations. “It’s possible that physicians act as discretionary decision makers for themselves and their families because they are very informed about their own situation and they know which guidelines make sense for them,” Polyakova says.
So, is low adherence by doctors healthy for patients? Further research is needed to answer that, say Persson and Polyakova. On its face, their finding suggests that guidelines are merely a default and that ignoring them is warranted at times. But it could also mean that without guidelines, less informed patients may get worse care; if that’s the case, it could exacerbate health inequities.
“Guidelines serve as a baseline,” Persson says. “If poor communities are not getting the same level of care as everyone else, that’s a reason to want to ensure that there is a universal minimum standard of care.”