Improvement in clinical work through feedback: intervention study

BMJ 1999; 318 doi: (Published 26 June 1999) Cite this as: BMJ 1999;318:1738
  1. Rolf Jorde, professor (rjorde{at},
  2. Arne Nord⊘y, professor
  1. Department of Internal Medicine, University Hospital of Troms⊘, 9038 Troms⊘, Norway
  1. Correspondence to: Professor R Jorde, Centre for Clinical Epidemiology and Biostatistics, David Maddison Clinical Science Building, Royal Newcastle Hospital, Newcastle, NSW 2300, Australia
  • Accepted 3 December 1998

We have frequently found an unacceptable number of hospital records and discharge reports lacking even the most basic information. To improve on this, we reviewed our hospital records and discharge reports on a regular basis, and we sent out questionnaires on quality of care to patients shortly after discharge. The hospital's medical staff received the results as a summarised report every other week. We deliberately disclosed only half of the variables studied. At the end of one year, the results were compared with those obtained before the intervention.

Subjects, methods, and results

Our study took place from September 1994 to October 1995 at the Department of Internal Medicine, University Hospital of Troms⊘ (120 beds and 45 doctors). We reviewed the hospital records for two sets of information: variables that were disclosed to the staff (past or present occupation, smoking habits, general physical condition, and blood pressure) and variables that were not disclosed (marital status, alcohol consumption, glandular enlargements, and pulse). We noted whether a copy of the initial discharge letter was in the hospital record and, although not disclosed to staff, whether the discharge letter had information on drugs prescribed and time and place for next follow up. Similarly, we reviewed the final discharge reports for information on drugs prescribed and, not disclosed, for time and place for next follow up.

In September 1994 and October 1995 we sent out 500 questionnaires to discharged patients. Questions included (a) at discharge, were you given time to speak with your doctor alone? (If so, for how many minutes?) (b) (did your doctor give you a discharge letter that included which drugs to use?) (c) considering politeness, respect, and humaneness, was the behaviour of the doctors and nurses excellent, very good, good, fair, or poor? Answers to the questions in parentheses were not disclosed to staff.

We analysed the data with two sided Pearson χ2 and Fisher's exact tests.

The table shows the effects of intervention on the hospital records, discharge letters, and reports, and on the doctors' and nurses' behaviour.

Number of variables found in hospital records, discharge letters, and final reports, and patients' evaluation of doctors' and nurses' behaviour, in September 1994 before intervention and again in October 1995

View this table:

Of the 500 patient questionnaires sent out in September 1994 and October 1995, 323 (65%) and 330 (66%) were returned respectively. There was no significant change between whether the patients had a chance to talk alone with their doctor before discharge (85.6% versus 88.4%) or the time allotted to them (<10 minutes in more than 50% of patients).


For an intervention to work, the methods used must be acceptable to the target group.1 We therefore gave feedback in such a way that individual doctors could not be identified, and the variables studied were such that no one could argue their relevance. When feedback was given to the doctors every other week, we found a highly significant improvement in almost all variables relating to hospital records, initial discharge letters, and final discharge reports.

The results, however, could not be considered satisfactory. Almost 15% of the final discharge reports had no information on drugs, and 12% of the patients had not talked to their doctor alone before they left. Furthermore, no significant change was seen in doctors' behaviour, although there was a potential for improvement. This shows the difficulty in inducing changes that are considered time consuming or that involve personal conduct.2

If substantial improvement in quality of clinical work is to be achieved then perhaps there is a need for information,3 administrative interventions, incentives, and penalties. 1 4 Furthermore, patients should be made aware of their rights about talking to their doctor and getting a proper discharge letter before they leave hospital.

We looked at only a small aspect of work done in our department. Despite this, the study was very time consuming. Therefore if improvements in the quality of clinical work are to be achieved the necessary investments must be made and costs must be met.


Contributors: RJ had the initial idea for the study. Both authors designed the protocol, collected and analysed the data, and wrote the paper. Both will act as guarantors for the paper.


  • Funding University Hospital of Troms⊘ and the Norwegian Medical Association.

  • Competing interests None declared.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
View Abstract