Intended for healthcare professionals

Views & Reviews Medical Classics

Dartmouth Atlas of Health Care

BMJ 2011; 342 doi: (Published 23 March 2011) Cite this as: BMJ 2011;342:d1756
  1. Richard Smith, director, UnitedHealth Chronic Disease Initiative, London
  1. richardswsmith{at}

It is not wholly fanciful to compare the Dartmouth Atlas of Health Care with On the Origin of Species. Both books resulted from a rigorous accumulation of data and fundamentally changed our world view. Darwin’s book showed our descent from apes. The atlas exploded the belief that medicine is based firmly on science.

The first Dartmouth Atlas was published in 1996, and it is now a rich website with well presented maps and publications on healthcare in the United States. “Jack” Wennberg (as he is known) is the main author but was not the first to identify variations. The atlas pays tribute to the British paediatrician J Alison Glover, who observed in the 1930s that the proportion of children who had received tonsillectomies varied from 10% to 50% in different school districts. Wennberg himself discovered big variations in numbers of hospital beds and physicians in Vermont in the 1970s.

The atlas grew out of failure of the Clinton health reforms when funds for studies to inform the reforms were diverted to creating the atlas. Wennberg and his team used Medicare and Medicaid data to produce the atlas but have shown a close correlation between all activities and those funded by Medicare and Medicaid.

Variation occurs in all of healthcare. The original atlas showed, for example, a twofold variation in numbers of hospital beds, a threefold variation in numbers of doctors, fourfold variation in rates of coronary bypass surgery, and eightfold variation in radical prostatectomy. Importantly, more hospitals and doctors did not mean better outcomes.

Politicians are fascinated by these data because of the huge savings that can be made if unwarranted variation can be eliminated. The first atlas calculated that if every region in the country was like Minneapolis, then 120 000 beds could be closed and $32.6bn (£20.2bn; €23.2bn, at current exchange rates) saved without any deterioration in outcomes.

Wennberg’s recent book, Tracking Medicine, tells the whole story and spells out the responses that are needed. He divides healthcare into the 15% that is effective and necessary, the 25% that is preference sensitive (do patients really want that operation?), and the 60% that is supply sensitive—driven by how many hospitals and doctors a region has. He thinks that the UK is a little different from the US. The NHS Atlas of Variation in Healthcare, published in 2010, shows, for example, a near 30-fold variation in the percentage of patients in primary care trusts who receive all nine key care processes recommended for people with diabetes (

The solution to these variations, argues Wennberg, is to promote organised systems of care; to establish fully informed patient choice as the legal standard; to improve the science of healthcare delivery; and to constrain undisciplined growth in capacity and spending. We have a long way to go.


Cite this as: BMJ 2011;342:d1756


  • Dartmouth Atlas of Health Care

  • By John E Wennberg and Megan McAndrew Cooper

  • First published 1996

  • Competing interests: RS is employed by the UnitedHealth Group and discovered for the first time while reviewing the atlas website that the company’s foundation is one of the atlas’ many sponsors.

  • Provenance and peer review: Commissioned; not externally peer reviewed.

  • Commentary: The Dutch approach to unwarranted medical practice variation (BMJ 2011;342:d1429, doi:10.1136/bmj.d1429); Analysis: Time to tackle unwarranted variations in practice (BMJ 2011;342:d1513, doi:10.1136/bmj.d1513); Review: Chris Ham on Wennberg’s latest book, Tracking Medicine (BMJ 2011;342:d1757, doi:10.1136/bmj.d1757); Podcast: A tale of two cycles: Fiona Godlee interviews Wennberg (