Randomised study of long term outcome after epidural versus non-epidural analgesia during labour
BMJ 2002; 325 doi: https://doi.org/10.1136/bmj.325.7360.357 (Published 17 August 2002) Cite this as: BMJ 2002;325:357All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
We read the interesting paper by Howell and colleagues about long-term back pain after labour analgesia.1 However, we were concerned about some methodological aspects.
The Authors used an intention-to-treat analysis to determine whether epidural analgesia is associated with backache, or not. Intention-to-treat analysis preserves the effects of randomisation, analysing noncompliant patients in the groups they were randomised into, and provides an estimate of the practical impact of the treatment under "real world" conditions, particularly with treatments difficult to tolerate.
Although it is generally the best choice for the primary analysis, sometimes intention-to-treat could answer a different question that the one you are interested in. In fact, we believe that this is the case as the real question is: does epidural labour analgesia cause long-term back pain? Actually, using intention-to-treat the backache rate in the non-epidural group is diluted by the 28% of cases who received epidural and, vice versa, in the epidural group the backache rate is diluted by a 33% of women who had not. This way we obtain the incidence of back pain in the two arms of the study but we cannot really answer the underlying question. We think that reporting also a per-protocol analysis (i.e. comparing the outcomes of women who really had epidural with those of women that had not) would have added relevant information, making this study comparable with most previous research, as well.
In our opinion, also the statistical power might be a concern, as the overall number of women who had epidural was 144, slightly less than the original 150 calculated by the Authors. Moreover, to maintain the same level of statistical power, the sample size should have been increased proportionally to non-compliers rate. Using the formula in the appendix 2, we estimate that with non-compliers rates of 28% (non-epidural) and 33% (epidural) respectively, the two study arms should have contained 289 and 334 cases. Therefore, if our calculation is correct, a type II error cannot be definitely excluded comparing the incidences of long-term back pain.
Appendix2
Adjustment of sample size according to proportion of non-compliers.
Nc = N * (1/(1-R)2)
were:
N = Sample size required for 100% compliers
Nc= Sample size to maintain same level of power with non-compliers
R = Proportion of non-compliers
References
1. Howell CJ, Dean T, Lucking L, Dziedzic K, Jones PW, Johanson RB. Randomised study of long term outcome after epidural versus non-epidural analgesia during labour. BMJ 2002; 325(7360): 357.
2. Lachin JM. Statistical consideration in the intention-to-treat principle. (Accessed September 10, 2002 at http://www.bsc.gwu.edu/~jml/download/ittansl2.doc)
Competing interests: No competing interests
There is no proof that wearing a gown is necessary to minimize
neuraxial infection when siting an epidural or other neuraxial block.
However, common sense and experience demonstrates that an epidural cather
can easily flip beyond the sterile field and touch the anesthesiologist's
arms. Furthermore, extending bare and hairy arms over one's sterile
equipment and field is not an acceptable aseptic technique - especially
when many neuraxial blocks are sited outside the OR.
(See the picture in the BMJ 24 MAR 1999; 318. This is similar to the ones
published, and has been used by this author and others to illustrate most
features of a 'Septic Technique'.)
Does one see improperly clad neurosurgeons inserting drains or other
equipment into the vertebral column, skull or other areas in intimate
contact with the CNS? I suggest not, because they are only too aware of
the catastrophic effects of infection.
When one looks at infections associated with neuraxial blocks, poor
aseptic technique is usually a significant factor. Other associated
factors are:
Procedure done outside the OR;
Organism almost always from patient's skin or the operator;
Difficulty with the procedure;
Epidural catheters left in situ a long time (> 48 hours)
END
Competing interests: No competing interests
Intention to treat is an appropriate analysis because it is closest
to what happens in the real world. Say for example those women randomised
to recieve epidurals all chose to discontinue them because they got back
ache during insertion. If not included in the epidural group the results
would be terribly skewed.
Competing interests: No competing interests
Re: need for gloves AND GOWN when placing epidural: it is my usual
practice to wear sterile gloves, mask but not gown when placing epidurals,
and I believe most of my colleagues follow suit. As far as I am aware,
our institution does not have an abnormally high rate of epidural abscess.
Do you have any actual EVIDENCE to back your statements regarding proper
sterile technique, or is it just your opinion?
Competing interests: No competing interests
Anaesthetists have reason to be grateful to the BMJ for being willing
to publish yet another trial, albeit a small one, demonstrating that
epidural analgesia is not associated with long term backache. On each
occasion this outcome appears to come as a surprise, so the finding is
worth repeating. The editor published with alacrity the retrospective
studies that gave the erroneous result that epidurals did cause
backache,[1,2] but took more persuading to publish prospective studies
with negative results [3] (and flatly refused to publish one showing
epidurals were good for babies - good news is no news).
Readers may find it useful to know that further clinical details of
this study can be found in an earlier publication by Howell and her
colleagues [4] in which 184 women were randomised to epidural and 185 to
non-epidural analgesia, primary outcome backache 3 and 12 months after
delivery.
The authors state that crossover between treatment groups is
inevitable in such trials. Not so. Researchers in the University of Texas
Southwestern Medical Center, Dallas, have published a series trials in
which to date a total of 3727 women were randomised to receive either
epidural or systemic analgesia. They made progressively more successful
efforts to improve analgesia in the non-epidural arm by the use of
generous patient controlled analgesia regimens. In the latest study [5]
the crossover amounted to 3.1%.
The cover picture relating to this article is misleading, not only
because, as other correspondents indicate, the epiduralist should be
wearing mask (most important) and gown, but also in the caption "Do
epidurals cause long term backache? No more than other forms of pain
relief in labour." It is not the pain relief that causes backache, it is
having a baby!
Felicity Reynolds, Emeritus Professor of Obstetric Anaesthesia, St
Thomas’ Hospital, London.
References
1. MacArthur C, Lewis M, Knox EG, Crawford JS. Epidural anaesthesia and
long term backache after childbirth. BMJ 1990; 301: 9-12.
2. Russell R, Groves P, Taub N, O’Dowd J, Reynolds F. Assessing long term
backache after childbirth. BMJ 1993; 306: 1299-1303.
3. Russell R, Dundas R, Reynolds F. Long term backache after childbirth:
prospective search for causative factors. BMJ 1996; 312: 1384-8.
4. Howell CJ, Kidd C, Roberts W, Upton P, Jones PW, Johanson RB. A
randomised controlled trial of epidural compared with non-epidural
analgesia in labour. Br J Obstet Gynaecol 2001,108:27-33.
5. Sharma SK, Alexander JM, Messick G et al. A randomized trial of
epidural analgesia versus intravenous meperidine analgesia during labor in
nulliparous women. Anesthesiology 2002;96:546-51.
Competing interests: No competing interests
I must agree with the comments of Dr Austin of 21 August 2002
regarding the paper "Randomised study of long term outcome after epidural
versus non-epidural analgesia during labour" BMJ 2002;325:357-9.
With a crossover of 33% in the epidural group and of 28% in the non-
epidural group, an intention-to-treat analysis is significantly weakened.
This becomes clearest when the crossover approaches 50%, at which time the
treatment and control groups become identical with regard to the treatment
received.
Scientifically, an intention-to-treat analysis is generally the
prefered method, but in this study it would have been useful if the
authors had also published the results of a received-treatment analysis.
Then it would be possible to compare the backpain outcome for patients who
actually received an epidural with those who did not.
Competing interests: No competing interests
James Austin makes a good point. This study does not tell us whether
epidural results in more back pain. It tells us whether a woman who plans
to have an epidural ends up with more back pain than one who doesn't. What
women will want to know is whether the epidural itself results in more
back pain. If available, this information will have a major influence on
whether a woman plans to have an epidural or not. Would it be possible for
the authors of the paper to present us with an alternative analysis
according to treatment actually received, rather than the intention-to-
treat analysis?
Competing interests: No competing interests
Where was the gown asks Professor Rosen?
I agree but where also was the mask - for the patient
and the operator's protection?
Competing interests: No competing interests
The study by Howell and collegues does not provide information on
certain variables within the epidural group. For example, was the epidural
drug regimen standardised or left up to the individual anaesthetist?
Although motor block after an edidural is not thought to be a significant
causative factor in backache [1] further details would be welcome. Were
all the patients lying or sitting for the epidural to be sited and might
this be significant? Details about the ease of epidural siting may also be
of benefit as several attepts to position a "difficult" epidural may
possibly result in postpartum backache.
1. Longterm backache after childbirth:prospective search for
causative factors. BMJ 1996 Jun 1;312(7043):1384-8
Competing interests: No competing interests
Re: Re: Why 'intention to treat'?
It has been suggested by several of the correspondents that the
authors of this trial ought to have published an analysis on the basis of
treatment received alongside the intention to treat one. If one keeps
anaylysing the same data in different ways, one will eventually arrive at
a significant finding. If two different ways of anaylsing data are planned
from the outset than this has to be taken into account when calculating
the sample size.
The description of sample size calculation for this study or rather
its predecessor study is contained in [1]. As another respondent has
already argued the actual sample size is insufficient to exclude a type II
error. More importantly, however, is the difference between the groups this
study could have picked up. The authors chose a twofold increase in the
incidence of back pain in the epidural group. Many obstetric anaesthetists
and even more parturients would think a much lesser increase in the
incidence clinically important.
[1] Howell CJ, Kidd C, Roberts W, Upton P, Lucking L, Jones PW, et
al. A randomised controlled trial of epidural compared with non-epidural
analgesia in labour. Br J Obstet Gynaecol 2001; 108: 27-33.
Competing interests: No competing interests