Limiting Variation in ED Admission Rates

No Comments

EPs too seldom follow guidelines when making this expensive, impactful decision. 

If you were to do a search on the word “variation” in the EM Abstracts database for the last five years you would find that there are 155 emergency medicine-specific citations with the word “variation” in the title – a huge number. Emergency medicine literature is full of papers telling us that the variation between physician practice is substantial – if not huge. Some physicians order lots of antibiotics, some much less. Some physicians order lots of CTs, some much less. Some physicians order lots of blood tests, some much less.

The bottom line – the emergency medicine literature indicates there are lots of opportunities to address variation and, ideally, create a narrow bell shaped curve focusing on the uniform practice of evidence-based medicine. The motivation to narrow variation seems obvious – to provide patients with consistent cost-effective, quality care. Large variations in care imply that many patients are not getting the care they should receive.

Every ED has a medical director (I was one for 25 years) and it would seem that one of the director’s main goals is to assure the quality of care provided by the physicians in the group. This means that the care provided needs to be measured and that variability must be determined and assessed in light of the best evidence available. In the past, measuring variability was virtually impossible because the data was hard, if not impossible to obtain. Quality assessment in the past required that charts be reviewed manually and was therefore limited to a small, often statistically insignificant, sample of charts.

But now things are different – we have EMRs. Although I believe that the net negative of EMRs far outweighs the net positives – at least for physicians (they’ve become super expensive data entry staff while inhibiting their ability to spend more time with patients), one of the few benefits of EMRs is that physician behavior can now be measured in an unprecedented manner.

Once meaningful data is obtained, we can see that there is a large variation in medical practice. While it is the ED director’s job to narrow variation in light of the medical evidence, comparing one physician’s practice with another is the “third rail” of ED directorship. The ED director will be assaulted with all manner of charges and excuses – “But my patients were sicker than his;” “I order lots of tests to avoid being sued;” “I don’t want to practice ‘cookbook medicine’;” “this is the way I was trained” – you name it. Narrowing physician practice variability is often one of the ED director’s most difficult jobs.

The EM Abstracts database has at least four papers looking at the variability between physicians in the evaluation of suspected PEs. All show that well established guidelines are often not followed. When this occurs, some patients are under-evaluated and many over-evaluated. CT pulmonary angiograms are not harmless tests. Performing CT angiograms subjects patients to radiation exposure and a recent study showed that there is a high likelihood for false-positive interpretations by community radiologists. To date, the literature regarding variation has largely focused on ED testing and treatment. But few studies have focused on which patients are admitted and which ones are not. Emergency physician recommendations have a significant impact on which patients stay in the hospital. Regardless of why an emergency physician recommends an admission, few on-call physicians would reject the EP’s recommendations for fear that the EP might be right in the marginal cases.

Unnecessary admissions costs huge dollars compared to an unneeded CT. The risks of an unnecessary admission may be significant to the patient. We all know that hospitals can be dangerous places – the patient gets more tests (that they don’t need); incidental findings noted on the unnecessary tests may precipitate even more tests; patients may then be subjected to potential treatments they don’t need, etc. Below are two papers that confirm what we all intuit – that variation in admission rates is huge.

The first looks at admissions for pneumonia and notes that treatment of admitted pneumonia costs 25% more than when treated in the outpatient setting. The study involved 2,069 patients at a single hospital. The focus of the study was variation in the decision to admit.

  • The overall admission rate was 58% (the low admitter was 38% vs the high at 79%) (variation was unrelated to illness severity)
  • High rates of admission were not associated with a decrease in mortality or secondary admission
  • Despite the development of multiple tools to help physicians make consistent dispositions in pneumonia, none were apparently used

Dean, N.C., et al, Ann Emerg Med 59(1):35, January 2012
BACKGROUND: Many patients with community-acquired pneumonia (CAP) can be safely managed without hospital admission, which has treatment costs 25 times higher than those of outpatient care.

METHODS: The authors of this observational study, from the University of Utah, performed an implicit chart review regarding variation between physicians in the decision whether or not to hospitalize 2,069 patients with CAP.
RESULTS: The admission rate was 58% overall, but varied nearly two-fold between individual emergency physicians (from a low of 38% to a high of 79%). This variability between physicians was unrelated to illness severity or patient/physician demographics. Patients managed as an outpatient were younger than those who were admitted (mean 46 vs. 63 years), and had a lower predicted mortality (0.93% vs. 2.65%); actual 30-day mortality was 0.34% vs. 6.8%, respectively. Hospital admission within seven days after discharge from the ED was documented for 10.7% of the outpatient group. Documentation of illness severity was noted for only 2.7% of the patients. Admission decisions by the emergency physician were in agreement with a “low severity” (suitable for outpatient management) rule constructed by the authors for only 69% of the patients. The disposition decisions of physicians with a higher admission rate were more likely than those of their lower-admission-rate counterparts to be discordant with this rule. Higher rates of hospital admission were not associated with a decrease in mortality or secondary hospital admission.
CONCLUSIONS: These findings demonstrate substantial variability between emergency physicians in the decision to hospitalize patients with CAP, without any relationship to morbidity or mortality.
30 references ( – no reprints) Copyright 2012 by Emergency Medical Abstracts – All Rights Reserved 9/12 – #26

This next entire paper should be required reading for every ED medical director, hospital CMO and health system administrator. Although technical in its approach and using lots of statistical analyses and adjustments, the authors analyzed almost 22 million ED visits to 964 EDs. The top 15 diagnoses of the admitted patients (20% of the ED patients were admitted) were compared with regard to admission variation. Although some diagnoses had narrow admission rate variability (e.g., AMI, septicemia and cerebrovascular disease) there was substantial variability in some other conditions. Specifically, of the top 15 admission diagnoses, five (mood disorders, chest pain, skin and soft tissue infections, UTIs and COPD) had variations in admission rates that were 3- to 5-fold higher than the conditions with little variation.

Venkatesh, A.K., et al, Med Care 53(3):237, March 2015
BACKGROUND: Emergency department (ED) admission rates in the US vary up to three-fold./ It is unclear whether decisions are influenced by the patient’s clinical condition and/or unrelated reasons.

METHODS: In a cross-sectional study coordinated at Yale University, these authors sought to characterize risk-standardized ED admission rates (RSARs). Data were gathered from the 2009 National Emergency Department Sample and included 21,885,845 adult ED visits (to 964 hospitals); visits were excluded if final disposition was unknown. Patient data (median age 44 years; 58% female), discharge diagnoses and hospital characteristics were recorded, as well as the top 15 clinical conditions resulting in hospital admission. The primary study outcome was the RSAR for each hospital ED and each clinical condition.
RESULTS: Of the total ED visits, 20% resulted in admission. The highest admission rates were for pneumonia, congestive heart failure, nonspecific chest pain, septicemia and pulmonary disease. Among the 15 conditions with the highest admission rates, the five diagnoses with the highest variation in rates were mood disorders, nonspecific chest pain, skin and soft-tissue infections, urinary tract infections and chronic obstructive pulmonary disease. Within-hospital correlations in RSARs for these five conditions were all statistically significant (P<0.001) and were three- to five-fold higher than the variation for other conditions (e.g., pneumonia, congestive heart failure). Analysis of hospital characteristics did not help explain these condition-specific variations.
CONCLUSIONS: US hospital EDs vary widely in condition-specific admission rates in ways not dependent on patient or hospital characteristics. Clearer guidelines are needed for “discretionary” medical conditions.
25 references ( for reprints) Copyright 2015 by Emergency Medical Abstracts – All Rights Reserved 8/15 – #34

The Venkatesh paper has an excellent discussion that addresses the many reasons why variability in admission of discretionary conditions may occur such as lack of guidelines, hospital culture, bed availability and the availability of other disposition alternatives (observation unit, home health). The bottom line is that we have to do better in addressing this huge opportunity to improve care. The authors conclude:

“Providers seeking to improve the efficiency of care delivery should develop clinical evidence and guidelines as well as hospital-level processes to safely reduce variation in admission rates for potentially discretionary conditions, whereas policymakers should develop quality measures to ensure that we achieve improvement. As inpatient hospital care comprises over one third of health care spending in the United States, the success of efforts to improve the quality of health care in the United States demand a better understanding of ED admissions decisions to ensure the optimal use of scarce hospital resources for the right patients at the right time.”

Although this study was published in 2015, it is based on 2009 data. In the last five years there have been changes in ED practice that may attenuate, at least to a modest degree, the conclusions of the authors regarding the magnitude of the variation problem (but I doubt it). Observation units are a recent phenomenon that should have an impact on admission rates (although, the difference in being admitted to the hospital and being admitted to an observation unit may be more a semantic differentiation than not). The development of hospitalist programs will also likely have some effect on admission decisions.

Taking on the huge challenge of narrowing the variability in admission rates for discretionary diagnoses must be an effort that involves multiple aspects of the hospital or health system leadership. The Venkatesh paper should serve as a wake-up call to both hospital executives and medical staff leaders to address a quality improvement opportunity associated with enormous medical and economic potential – but which can be expected to be extraordinarily difficult.


  2. (from 1999 but makes some interesting conclusions)

Dr. Bukata is the Editor of Emergency Medical Abstracts.

Leave A Reply