Intended for healthcare professionals

Science

The killing season—fact or fiction?

BMJ 1994; 309 doi: https://doi.org/10.1136/bmj.309.6970.1690 (Published 24 December 1994) Cite this as: BMJ 1994;309:1690
  1. Paul Aylin, senior registrar in public health medicine Department of Public Health Sciences, St George's Hospital Medical School, London SW17 0REa,
  2. F Azeem Majeed, lecturer in public health medicinea
  1. a Department of Health Statistics, Office of Population Censuses and Surveys, London WC2B 6JB
  1. Correspondence to: Dr Aylin.

    You come out of medical school knowing bugger all—no wonder August is the killing season. We all kill a few patients while we're learning.1

    In making the above statement, one of the principal characters in the controversial BBC television series Cardiac Arrest suggests that hospital patients are at increased risk of dying during the early days of August, when newly qualified docors start as house officers for the first time. The belief that hospital patients are at increased risk of dying during this period because of the inexperience of the doctors looking after them is fairly widespread and has led to the first week of August being dubbed the “killing season.” In view of the increased emphasis on the quality and outcome of clinical care and on the rights of patients,2 we thought that it would be appropriate to determine whether the idea of the killing season was fact or fiction.

    Patients, methods, and results

    We obtained from the Office of Population Censuses and Surveys details of all deaths occurring in NHS hospitals (except psychiatric institutions) in England and Wales for the last week of July (25-31 July) and the first week of August (1-7 August) from 1983 to 1992. Of the 107 576 deaths that occurred during the study period, 54 240 were in the last week of July and 53 336 in the first week of August. We found that for each major cause of death more deaths occurred during the last week of July than in the first week of August (table). Because we had no data on the number of admissions during the study period, we were not able to calculate rates of death. We did calculate, however, proportional mortality ratios3 for each cause of death; no significant differences occurred in the proportion of hospital deaths due to potentially avoidable causes such as asthma and ulcers between the last week of July and the first week of August (table).

    Number of deaths in hospital in England and Wales by cause during 25-31 July and 1-7 August from 1983 to 1992 and proportional mortality ratios by cause of death (July:August)

    View this table:

    Comment

    Our data do not support the hypothesis that hospital patients are more likely to die in the first week of August than at other times of the year. Our study shows that there are in fact fewer hospital deaths in the first week of August than in the last week of July. The main limitation of our study was that we had no data on the number of admissions during the study period and were not therefore able to calculate rates of death. If fewer patients were admitted in the first week of August then the expected number of deaths would be smaller, giving a favourable impression of the risk of death during this time.

    We did examine, however, the deaths that occurred in hospital by their cause and found no evidence that the number of deaths from conditions that usually require the patient to be admitted as an emergency (and that cannot usually be prevented)—such as myocardial infarction and road traffic accidents—was any higher during the first week of August. We also found no evidence that the proportion of hospital deaths due to potentially avoidable causes such as asthma increased in early August. A study in an American hospital also found no evidence that hospital deaths rose when newly qualified doctors started work.4 We therefore conclude that newly qualified house officers have been falsely accused of increasing the number of deaths in hospital and that the idea of the killing season is very much fiction.

    References

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