Skip to Main Content

Physician training has long been notorious for marathon shifts, sleepless nights on call, and holidays worked. But that began to change in 2003, when the medical profession placed restrictions on work hours during residency. However, experts wondered, can we train residents in fewer hours and still make good doctors?

A new study in the BMJ says yes. The researchers, led by Dr. Anupam Jena, a professor of health care policy and medicine at Harvard Medical School, looked at the performance of internal medicine doctors in their first year of unsupervised medical practice after completing their training.


They compared the outcomes for patients of two groups of physicians: those trained before 2003, when the typical work week was 100 hours; and those trained later under the new rules, which capped weekly hours at a mere 80, with no individual shift exceeding 30 hours. For the three quality measures examined — mortality within 30 days of being hospitalized, readmissions, and hospital services used (a measure of efficiency) — they found no differences between the groups.

Looking at the data for just the sickest patients in the hospital, a “group of patients for whom the experience and training of a doctor is really important,” said Jena, the authors again found no difference in these three outcomes.

“If anything, it’s reassuring,” said Dr. Sanjay Desai, director of the internal medicine residency program at Johns Hopkins who was not involved in the study.


Jena and his team based their analysis on a random sampling of patient records from the Medicare database, including 450,000 hospitalizations at over 4,000 hospitals in the U.S., for each year from 2000 to 2012. 

Though the data are old, Desai said it’s a timely paper. Other reforms were made after 2003, but the 80-hour cap and 30-hour shift limits studied in this paper are similar to what’s in practice today, he said. Since 2017, when the latest reform was implemented by the Accreditation Council for Graduate Medical Education, the cap remains 80 clinical and educational work hours per week averaged over a four-week period, with no shift exceeding 28 hours.

Dr. Thomas J. Nasca, CEO of the accreditation group, said in a statement that the study supports its work rule, “which allows for flexibility within maximums while supporting patient safety, quality improvement, and physician well-being.”

Jena said he chose to study only the first year of a new doctor’s career because that’s when the effects of residency training should be most evident. “If doctors who just completed a residency had insufficient training during residency because they worked fewer hours,” he said, “then we should expect to see some differences really early on” before “real world” experience.

In 2006, co-author Dr. Jay Bhattacharya, professor of medicine and economics at Stanford University, looked at mortality of patients treated by residents before and after the changes implemented in 2003. He found that capping weekly hours decreased short-term mortality among high-risk patients. 

“Previous literature said that tired doctors hurt and kill patients. In 2006, short-term results showed that lower hours during residency reduced mortality in hospitals. But no one had assessed long-running quality of doctors,” said Bhattacharya.

“This study is unique because it studied outcomes after training,” said Desai.

The study also found that physicians with 10 years of experience provided higher quality care than the first-year doctors, as expected, but the gap in care was the same whether the new doctors trained before or after the cap was put in place. Over the 10 years studied, the study team also noticed a decrease in overall mortality, which accounts for nationwide trends that suggest hospitals are providing better care overall, said Jena. 

The authors noted that in the hospital setting, physicians work in big teams with specialists, nurses, social workers, and other advanced practitioners, which mutes the impact of any single physician on patient outcomes. 

What we need to know next, said Desai, is what residents are doing with their 80 hours. How much time is spent with patients? At the computer? Doing procedures?

“The real meaning in the hospital is where you spend those hours,” said Desai. “In addition to [examining] patient safety outcomes, we need to think very hard and study more rigorously how physicians are shaped” during residency. 

A limitation of the study, mentioned by both Desai and the authors, is that it looked only at one specialty — internal medicine doctors. Surgeons and other doctors who perform procedures may be affected differently by hour restrictions because it would limit the number of procedures they perform under supervision. 

  • I have seen some really bad research published and this one is one of the worst.

    Who would even consider that an impact of a change years before would impact outcomes? It also took time for the limitations to actually be set up and implemented.

    1. Social science usually explains very little of the variance or reason for an association, but this one takes the cake.
    2. Too many other influences have greater impact on outcomes

    The only reason for publication is because it is a popular topic – which has apparently pushed the journal, reviewers, and editor into accepting such and article and publishing it.

    BMJ has had a history of publishing such works, even estimates of medical error based on previous studies that made assumptions. BMJ says it needs to publish studies that may be controversial to give them air time. BMJ needs to be more rigorous.

    Outcomes (cost, quality, metrics, measurements) are about the patient population. Repeat this many times.

    Outcomes are about influences for many years or decades before an encounter, and shape who gets an encounter, and shapes the encounter, and shapes understanding of the encounter, and shapes what happens after the encounter.

    If the patient population does not change and is same or similar, the outcomes will be similar. This is seen in resident work hours studies with outcomes compared before and after limitations (Philadelphia). If the population is the same and the study compares MD vs nurse practitioner, the outcomes will be the same (this MD vs NP study was also a particularly bad study for other reasons). If the populations compared are different (rural hospital vs urban, higher volume vs lower, female vs male hospitalist, PCMH vs IPA practices, and many more – the outcomes will be different.

    When the authors, funding foundations, journals, editors, and reviewers want the publication to come out – then it will – regardless of low levels of correlation, despite the intent to demonstrate what is demonstrated (bias), despite many alternative explanations, despite variations within a group greater than differences between the groups compared, and despite numerous influences difficult to address in controls.

    If you care for populations with lesser outcomes, then you will have lesser outcomes. You will also be penalized by the financial design under pay for performance, value based, or readmission penalties – discrimination by design on top of being paid less.

    In the case of ratings by CMS or others, you will have lesser ratings. If you care for populations that are high cost, high risk, or low in social determinants shaping lower outcomes – this will pressure your employer or ACO to dump you.

    But it is not about you. It is about your high cost and poor outcome population that you are brave enough to serve.

    Metrics, measurements, and micromanagements hurt patient care, hurt team members, and distract from care. They benefit consultants, corporations, and CEOs.

    Studies that consider alternatives and include more of the data and control for changes in the populations matter. But they are uncommon. Researchers have learned it is better to go along with the bandwagon than be critical.

    One recent study demonstrated that Shared Savings programs for ACOs demonstrated no difference in outcomes – when controlling for the physicians and their populations driven away over time (orthopedic study).

    Of course the pro-micromanagement researchers had a cow over this one – rather than learning and attempting to do better studies.

    Readmissions, value based, pay for performance, and other penalties penalize those who care for the populations with lesser outcomes inherently. In general they are also paid less. Paid less, penalized more, and forced to pay more for micromanagement is killing generalist and general specialty practices in the US – about 70% of encounters and 90% where most Americans most need care. The workforce is shrinking where Americans are growing fastest in numbers, demand, and complexity.

    We need research and researchers to help guide health care design – not make it worse.

    What remains?

    Residents are abused by their employers. They are abused slightly less with regulations. Since they are physicians, their employers benefit from their work. Since they are professionals and are vulnerable in their positions – they can be abused. Residents need more support, regular sleep, and time for personal/family activities.

    And the abuse has only just begun in medical education. The debts are high and getting higher and there is a glut of workforce at a time with fewer and larger and more powerful employers. Nurse practitioners are expanding annual graduates at 14 times the annual population growth. Physician assistants and osteopathic graduates are being increased at 8 to 10 times. US MD designers for 100 years kept the growth of the MD population from exploding – until 2003. Since that times the growth has been at 6 times the population growth – up 30% already. International graduates remain about 20 – 25% of physicians.

    So what is all this rhetoric about shortages? Shortages are about the financial design, particularly payments too low and costs of delivery too high for generalists, general specialists, primary care, mental health, womens health, basic surgical services, and care where most Americans most need care.

    CMS and insurance “payers” pay less and penalize more – resulting in greater disparities.

    And by the way, as long as we allow the experts to promote more graduates as a solution for shortages – they will make the abuses even worse unless you are near to retirement.

  • Residence requirements continue to exist as a financial gift to hospitals in the form of low-cost labor. Physician training in the US has failed to keep the standards of care and outcomes in the top tier of world health care. I, for one, would be skeptical of the diagnosis from a doctor that has not slept for thirty hours.

  • The obvious question on this – were they looking at office based internists hospital admission outcomes, or those in hospitalist roles, or what? Internal medicine residents go out into a variety of work settings such that looking at the Medicare database and attempting to tie the noted markers to the individual physician is not at all a true statement of causation as opposed to the efficacy of the setting and system they are a cog in.

Comments are closed.