April 2013 E-Journal Club


Spring was slow to come to Charlottesville, but warmer days have finally arrived.  Our trainees were greeted by our beautiful Dogwood and Redbud blossoms, and outside of some April showers, we had mild weather and a great week.  Our April trainees hailed from Beaver Dam, WI; Owensboro, KY; Bakersfield, CA and Rapid City, SD.  Our Journal Club article this month is unlike our usual articles, in that it deals with oral supplements.

April Citation:

Philipson TJ, Snider JT, Lakdawalla DN, et al.  Impact of oral nutritional supplementation on hospital outcomes.  Am J Manag Care. 2013 Feb;19(2):121-128.


This was a retrospective, observational study using a database containing the records of 44 million adult inpatients from 460 different sites over 10 years that investigated the associations between the ordering of oral nutrition supplements (ONS) and length of stay, costs and probability of re-admission within 30 days.  Investigators collected data on age, insurance, marital status, race, sex, comorbidities, and recent admissions.

The investigators used propensity score to match patients for whom ONS were ordered with similar patients who did not have ONS ordered.  Additionally, the investigators used instrumental variables analysis in an attempt to identify causal relationships with the use of oral supplements.

Inclusion and Exclusion Criteria were:

Inclusion criteria:

Adult patients not receiving tube feeding for whom ONSs were ordered.

Exclusion criteria:

Age ≤18 years old, terminal episodes, receiving tube feeding.

Major Results:

The investigators identified 810,589 episodes involving ONS; after removal of episodes that met exclusion criteria and patients with incomplete data, 724,027 ONS episodes were left.  The rate of ONS ordering was 1.6% of all adult inpatient episodes.

Patients that had ONS ordered compared to non-ONS patients had longer hospitalizations (12.5 days Vs. 4.8 days), were older (68.4 vs. 56.7 years), and had more admissions in the previous 6 months (42.2% Vs. 25.6%).  Ordinary least squares regression analysis showed that ONS use was associated with a 34.7% increase in LOS, 50.7% increase in episode cost, but a 0.9% decrease in the probability of a readmission.  However, after instrumental variables regression analysis was used to attempt to account for selection bias, ONS appeared to decrease LOS by 21.0%, decreased episode cost by 21.6%, and further decreased the probability readmission to 6.7%.

Author’s Conclusions:

“….the use of ONS led to statistically significant decreases in inpatient LOS, episode cost, and readmission.”


This study used a very large database of patients, and used propensity matching and instrumental variables analysis to try to overcome the inherent limitations of observational data.  Instrumental variables analysis is often used to estimate causal relationships from observational data when it is impossible or unethical to conduct randomized studies (for example, studies involving the use of tobacco products).

However, regardless of the methods used to reduce bias, observational data can never establish causation, because it is not possible to eliminate all of the unknown confounding factors.  In an observational study it is not possible to control for the reason why a particular patient may, or may not have ONS ordered.  My experience is that the decision to provide a patient with oral supplements is usually based on the individual circumstances of the patient, the engagement of the staff and caregiver, involvement of nutrition professionals, and is not based on an institutional inclination.  It is possible that an order for ONS may simply serve as a surrogate for patients that have nutritionally aware/attentive staff, or a nutrition consultant working with them.

There are a number of randomized studies of ONS in several different patient populations, so it is possible/ethical to conduct randomized studies with ONS.  A systematic review of 54 randomized studies of ONS reported support for the use of ONS in malnourished geriatric patients, so there is evidence from randomized studies that ONS may have benefits in clinically appropriate patients. (1)

We also discussed the fact that this study did not account for actual consumption of ONS, but only that it was ordered.  Anyone who has ever walked into a patient room to see cans of supplement untouched and stacked on the counter knows that ordering supplements and consumption of supplements are two very different things.  The study did not attempt to answer which patients may benefit from ONS.  The conclusions might seem to suggest that a wide variety of patients, regardless of nutrition status and oral intake, would benefit by simply ordering ONS.  However, it is likely that nutrition status, ability to eat and willingness to try and drink ONS will have an influence on its potential benefit, and it will require randomized, or a number of larger cohort studies, to determine the potential benefits of ONS in specific populations.

Our Take Home Message (s)

1.   Observations studies provide associations and theories that need to be tested in randomized studies.

2.   It is not possible to statistically control for all bias in an observational study.


  1.  Koretz RL, Avenell A, Lipman TO, et al.  Does enteral nutrition affect clinical outcome? A systematic review of the randomized trials.  Am J Gastroenterol. 2007 Feb;102(2):412-429.

Other News on the UVAHS GI Nutrition Website: (

Upcoming Webinars 2012/2013:

–May 14:  The Role of Fiber in Nutrition Support

–June 18:  Nutritional Management of Food Allergies in the Pediatric Patient

Check out What’s New:

–“Nutrition Support Blog”

–“ Resources for the Nutrition Support Clinician

Latest Practical Gastroenterology articles:

–Nouvenne  A, Ticinesi A, Meschi T. Nephrolithiasis and Gastrointestinal Tract Diseases: Can Diet Intervention Help?  Practical Gastroenterology 2013;XXXVII(4):27.

–Mccray S. FDA-Approved Pancreatic Enzyme Replacement Therapy.  Practical Gastroenterology 2013;XXXVII(4):36.


Joe Krenitsky MS, RD

Carol Rees Parrish MS, RD


PS – Please feel free to forward on to friends and colleagues.