Monday, April 6, 2015

Harvard Study: Hospital Computerization Does Not Improve Performance (Updated 6 Apr 2015)

A republication of an earlier posting of this study. It's timely in the light of New York Times Opinion piece, "Why is health care tech still so bad?"

Title: Hospital Computing and the Costs and Quality of Care: A National Study
Authors: David U. Himmelstein, MD, Adam Wright, PhD, Steffie Woolhandler, MD, MPH Department of Medicine, Cambridge Hospital/Harvard Medical School, Cambridge, Mass;  Clinical Informatics Research and Development, Partners Healthcare System, Boston, Mass.


BACKGROUND: Many believe that computerization will improve health care quality, reduce costs, and increase administrative efficiency. However, no previous studies have examined computerization’s cost and quality impacts at a diverse national sample of hospitals.
METHODS: We linked data from an annual survey of computerization at approximately 4000 hospitals for the period from 2003 to 2007 with administrative cost data from Medicare Cost Reports and cost and quality data from the 2008 Dartmouth Health Atlas. We also compared hospitals included on a list of the “100 Most Wired” with others.
RESULTS: Hospitals on the “Most Wired” list performed no better than others on quality, costs, or administrative costs.
CONCLUSION: As currently implemented, hospital computing might modestly improve process measures of quality but does not reduce administrative or overall costs.
© 2009 Elsevier Inc. All rights reserved. • The American Journal of Medicine (2009)
KEYWORDS: Hospital costs; Hospital quality; Information systems

Summary and Critique of the Study

Researchers from Harvard University collected data over 5 years (2003 to 2007) from a variety of different hospital data repositories

The hospital computer applications considered for this study included:

Clinical applications subscore (8 applications)

  • Clinical data repository
  • Computerized practitioner order entry
  • Data warehousing and mining, clinical
  • Electronic medical record
  • Laboratory information system
  • Nursing documentation
  • Order entry
  • Physician documentation
Administrative applications (patient-related) subscore (4 applications)

  • Nurse acuity
  • Nurse staffing scheduling
  • Patient billing
  • Patient scheduling
Administrative applications (other) sub-score (12 applications)

  • Case mix management
  • Cost accounting
  • Credit collections
  • Eligibility
  • Data warehousing and mining, financial
  • Electronic data interchange
  • Executive information system
  • General ledger
  • Materials management
  • Personnel management
  • Staff scheduling 
The authors created a overall computerization score and three subscores on the basis of determining whether or not the hospital had computerized the areas listed above.  

Concerns regarding the construction and use of computerization score(s).
  • No details were disclosed on how they created the level of computerization score.  
  • Furthermore, no details were disclosed regarding the score distribution.  
  • >> This is particularly important because correlation depends on a wide distribution of values as well as a reasonably even distribution across the range of values.  
  • No information was forthcoming from the authors about these characteristics of the data set.   
  • If the computerization scores were tightly clustered, then the r values (Pearson product-moment correlations) would hover around zero - and they did.

Suggested that increases in computerization may not be having the desired effects in reducing costs and increasing productivity and performance.

With respect to clinical significance of hospital computerization, the authors concluded that:

  • Hospital computerization has not, thus far, achieved savings on clinical or administrative costs. 
  • More computerized hospitals might have a slight quality advantage for some conditions.
  • No reliable data support claims of cost savings or dramatic quality improvement from electronic medical records.

** Title/Author reference and abstract. Material below is edited for length and quoted. This study 
is available at no cost online. 

The following later post provides additional information about the effects of computerized medical records systems.


  1. I really like your blog. I really appreciate the good quality content you are posting here for free. May I ask which blog platform you are using?
    GE 8C