Reduction of postoperative mortality and morbidity with epidural or spinal anaesthesia: results from overview of randomised trials

Number needed to treat derived from a meta-analysis may be misleading

Rodgers and colleagues (1) estimated that neuraxial blockade would be
expected to result in about one fewer postoperative death for every 100
patients. Since this number needed to treat (NNT) of 100 may be the bottom
line for many clinicians from this study, we want to raise some comments
on the interpretation of this finding.

Whether the NNT of 100 is low or high is a matter of judgement and
depends on the severity of the endpoint considered and also on the
economical dimensions. If the NNT of 100 appeared to be valid in this
context, we agree that the benefit obtained by decreasing mortality would
be clinically important. However, the NNT, as estimated from the pooled
absolute mortality rates, is 104 with 95% CI of 78 to 313. Thus,
concluding more cautiously, the NNT to save one death ranges from 78 to
more than 300 patients.

NNT depends on both the absolute risk in the absence of treatment and
relative risk reduction by the treatment (2). NNT calculated from pooled
results from several studies with different baseline risks may be
misleading (2, 3). The NNT for the pooled data from the high-risk subgroup
of Rodgers et al. is 30 (95% CI 26 to 44). For the low-risk subgroup, the
NNT is 293 (95% CI 181 to infinity). Further, applying a similar relative
risk reduction as calculated by Rodgers et al. to patients at 17% risk of
operative mortality (4), gives an NNT of 20 patients. When the mortality
in these patients is decreased to 3.4% by perioperative beta-blockade (4),
NNT is increased to 98 patients. Thus, NNTs are dependent on the patient
population chosen. Absolute risks - and consequently NNTs - are sensitive
also to secular trends in incidence and case fatality of adverse events.
Within the recent decades, improvements in the perioperative care of the
surgical patient have had favourable effects on prognosis. This is also
reflected in the study of Rodgers et al. where the mortality in the two
studies in the high-risk subgroup from the early eighties has decreased
from 13% to 3% in the two most recent studies from the nineties.

The conclusion of Rodgers et al. for a more widespread use of
neuraxial blockade should be related to the baseline risk relevant to the
patient group considered. This recommendation may be clinically sound in
high risk patients, but large randomised trials are required to confirm
such practice.

Markku Hynynen

Senior Lecturer

Department of Anaesthesia and Intensive Care,
Helsinki University Central Hospital, Jorvi Hospital,
FIN-02740 Espoo, Finland


markku.hynynen@helsinki.fi

Timo Strandberg

Senior Lecturer

Department of Medicine, Geriatric Clinic, University of Helsinki,
PO Box 340, FIN-00029 HUS, Finland

Helena Varonen
MD,
Researcher

Stakes, National Research and Development Centre for Welfare and Health
PO Box 220, 00531 Helsinki, Finland

1. Rodgers A, Walker N, Schug S, McKee A, Kehlet H, van Zundert A, et
al. Reduction of postoperative mortality and morbidity with epidural or
spinal anaesthesia: results from overview of randomised trials. BMJ 2000;
321: 1493-7.

2. Chatellier G, Zapletal E, Lemaitre D, Menard J, Degoulet P. The
number needed to treat: a clinically useful nomogram in its proper
context. BMJ 1996; 312: 426-9.

3. Smeeth L, Haines A, Ebrahim S. Number needed to treat derived from
meta-analyses – sometimes informative, usually misleading. BMJ 1999; 318:
1548-51.

4. Poldermans D, Boersma E, Bax JJ, Thomson IR, van de Ven LLM,
Blankensteijn JD, et al. The effect of bisoprolol on perioperative
mortality and myocardial infarction in high-risk patients undergoing
vascular surgery. N Engl J Med 1999; 341: 1789-94.