Intended for healthcare professionals


Problem based learning

BMJ 2008; 336 doi: (Published 01 May 2008) Cite this as: BMJ 2008;336:971
  1. Diana F Wood, director of medical education and clinical dean
  1. 1University of Cambridge School of Clinical Medicine, Addenbrooke’s Hospital, Cambridge CB2 2SP
  1. dfw23{at}

Time to stop arguing about the process and examine the outcomes

Problem based learning was developed in the late 1960s and has been the most influential innovation in medical education during the past 40 years. Essentially, problem based learning is a small group teaching method that combines the acquisition of knowledge with the development of generic skills and attitudes. Educationally, it is theoretically grounded in adult learning theory and constructivism and is predicted to produce a better learning environment and improved outcomes in terms of graduate knowledge, skills, and attitudes.

Implementation of problem based learning requires fundamental changes in the way educators conceive, design, deliver, and assess the curriculum.1 Despite the cost and resource implications, problem based learning has been introduced to varying degrees throughout the world—for example, it is used in most medical schools in the United States and many new medical schools in developing countries. Given this wide scale adoption, why is problem based learning still a controversial topic? The answer lies in the continuing lack of convincing evidence for its superiority over other teaching methods in terms of graduate outcomes. This is not for want of trying—the medical education literature abounds with publications on problem based learning, which have produced lively debate.2 3 4 5 6 But the question remains—does problem based learning produce better doctors?

A recent publication from the National University of Singapore sheds some light on this question.7 Koh and colleagues performed a systematic review of how problem based learning during medical school affected the competence of doctors after graduation. Only publications that included a control group of graduates from a “traditional” curriculum were included. The study population ranged from first year graduates to doctors who had been in practice for up to 20 years. Most of the studies were surveys, and an important feature in the final analysis was that doctors’ self assessments of the competencies in question and assessments by independent observers were considered separately. The level of evidence in favour of problem based learning over traditional learning was derived from previously published data coupled with the research team’s scoring system, which increased weighting for randomisation, sample size, objective assessment, and response rate. Thirteen studies were finally included and 38 competencies were identified, assessed, and categorised into eight dimensions—overall, technical, social, cognitive, managerial, research, teaching, and knowledge. Of these, the social dimension showed the strongest evidence in favour of problem based learning. In line with previously published data,8 little correlation was seen between self assessed and observer assessed competency. When both self reported and independently observed assessments were combined, four competencies had moderate to strong evidence in favour of problem based learning—coping with uncertainty (strong), appreciation of legal and ethical aspects of health care (strong), communication skills (moderate (self assessed), strong (observed), and self directed learning (moderate). Self assessment showed a strong level of evidence against problem based learning for possession of medical knowledge, but this was not confirmed by independent observation. The authors conclude that problem based learning has positive effects on graduate competencies in important social and cognitive domains.

This review confirms what most educators have come to believe on the basis of hundreds of less rigorous reports—that, compared with traditional learning, problem based learning has beneficial effects on some psychosocial outcomes of undergraduate medical education. Indeed, the argument seems somewhat stale. However, one important factor not acknowledged here or elsewhere in the medical education literature is the lack of definition of the “control traditional curriculum.”

The student cohorts reviewed by Koh and colleagues date from the 1980s and 1990s, when traditional control curriculums were probably based on a rigid divide between preclinical and clinical education, entirely lecture based programmes, and didactic clinical teaching. Since then, outcome based frameworks for medical education have focused on the competencies expected of graduates to meet the demands of patients in modern society. Crucially, the emphasis in medical education has moved from the process to the product.9 10 Features previously associated with problem based learning (fewer lectures, smaller groups, and vertical and horizontal integration) are now found in most undergraduate curriculums. Teaching and learning in communication skills and the psychosocial domains can be achieved in many ways, and working in small groups—coupled with timely and constructive feedback—may be just as effective as problem based learning.

Performing outcomes based research in education is difficult because of the large range of confounding factors. What has become clear, however, is that graduates from different medical schools perform very differently in postgraduate examinations, and some of this variance can be attributed to the undergraduate teaching programme.3 11 Surely it is time to stop arguing about the process and ensure that diversity in undergraduate educational provision is related to declared graduate outcomes and delivers doctors who have the required competencies for good medical practice.


  • Competing interests: None declared.

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