On KeyLIME this week, Jason discusses a study that takes a look at hospitalists in the US, as they set out to describe the association between clinician years in practice and 30-day and in-hospital mortality. Read on, and check out the podcast here (or on iTunes!)
KeyLIME Session 161:
Listen to the podcast
Read the episode abstract here.
Goodwin JS1,2, Salameh H1, Zhou J2, Singh S3, Kuo YF2,4, Nattinger AB3. Association of Hospitalist Years of Experience With Mortality in the Hospitalized Medicare Population. JAMA Intern Med. 2018 Feb 1;178(2):196-203.
Reviewer: Jason Frank (@drjfrank)
One of the biggest ongoing controversies in medical education is around the role of experience in practice on patient outcomes. KeyLIMErs will know that we have previously covered several previous papers looking at factors that influence graduates’ practice outcomes, including time since graduation. The literature on this topic is mixed and conflicting. Landmark studies by Asch suggest that graduates are set on an arc of practice the moment they leave residency, but at least their complication rates go down over time. However the majority of research says the opposite: the longer we are in practice, the more likely clinicians are to be out of date, less evidence-based in decision-making, and have a trend towards poor patient outcomes. This should give every clinician and clinician-educator pause. Should experience in seeing patients and learning from cases not improve us all? It seems counter-intuitive to say the least, and downright depressing as—ahem—many of us progress in our practice careers.
This study, by Goodwin et al, from the University of Texas Galveston, and University of Wisconsin, published in JAMA Internal Medicine this year, takes a new look at the perennial issue of the impact of experience on patient outcomes. Looking at hospitalists in the US, they set out to describe the association between clinician years in practice and 30-day and in-hospital mortality.
Type of Paper
Research: Observational; Administrative Database Study
Key Points on Method
The authors used a 5% sample of the US Medicare database to answer their question. They defined hospitalists as “generalist physicians” (geriatrics, internists, family physicians, or GPs) who focus on inpatient care (defined as >80% of billings were for inpatient services). This group has now done multiple studies on the US hospitalist population, which they note has high turnover (up to 20% per year) and sometimes includes IMGs who are using the job as a stepping stone to further career opportunities.
The sample of inpatients were at least 66 years old, had only Medicare coverage, had at least a 3-day inpatient stay, and had a hospitalist bill at least 50% of the charges related to the admission. This produced a sample of about 305,000 patients.
In their analysis, the authors looked at the database 2 ways:
1. They looked at a cross-sectional sample of 21,612 hospitalists, then looked back for their experience as a hospitalist over the prior 5 years; and
2. They also identified a cohort of 3860 hospitalists who entered practice after 2007 and worked as a hospitalist for at least 4 years.
They used logistic regression to model patient characteristics, hospitals, providers, and patient outcomes.
It is important to note the threats to validity of retrospective database studies, especially when physician coding practices can radically change outcomes. Furthermore, the heterogeneous nature of this group of hospitalists introduces multiple confounders.
Of the 21,612 hospitalists, about 25% had 1 year in practice or less. About 54% had >4 years experience.
For the primary outcome, 30-day mortality was 10.50% for patients of hospitalists in their first-year vs. 9.97% for patients of hospitalists in their second year. Other findings included:
The mortality odds for patients of second-year hospitalists were 0.90 (95%CI, 0.84-0.96) compared with patients of first-year hospitalists. Observed hospital mortality was 3.33% for patients cared for by first-year hospitalists vs 2.96% for second-year hospitalists (odds ratio, 0.84; 95%CI, 0.75-0.95). For both 30-day and hospital mortality, there was little change in odds of mortality between the second year and subsequent years of experience.
Shockingly, the authors estimated an excess of 3865 deaths in this 5% Medicare cohort due to MD inexperience.
The authors conclude that the first year as a hospitalist is a period of excess risk to patient mortality. After 1 year in the job, there was little change in mortality.
The authors suggest that more needs to be done to orient, mentor, monitor, and prepare physicians for their first year in practice.
Spare Keys – other take home points for clinician educators
1. This is the first paper I have seen that looks at early practice impact on patient outcomes in this way, and all of us as educators, leaders, administrators, and stewards of healthcare need to seriously consider these findings.
2. This is yet another meded-related study that yields important findings from existing databases of patient outcomes. All of us should look around to see what data already exists that allows us to ask key questions.
3. Beware the threats to validity in retrospective database studies—garbage in, garbage out rule applies.
Access KeyLIME podcast archives here
Check us out on iTunes