BMJ 1994;309:1015-1016 (15 October)

Letters

Non-randomised studies cannot be ignored

EDITOR, - James G Thornton and Richard J Lilford's assessment of the components of active management of labour is equivalent of taking a car to pieces and finding that a gearbox left lying in the road goes neither very far nor very fast.1 The strict diagnosis of labour is not the final component of the management of labour - that is delivery of the placenta; rather it is the first component, both in chronological sequence and in importance.2 The fact that 40% of women diagnosed as not being in labour returned promptly in unequivocal labour implies that the remaining 60% were spared inadvertent and unnecessary induction of labour at that time.

Artificial rupture of membrances is done to confirm the presence of clear liquor as oxytocin is dangerous if no liquor can be seen or if meconium is present. Speeding up established labour has never been claimed to confer more than marginal benefit. Care is taken not to use this method of inducing labour without good indication. Thornton and Lilford discuss use of oxytocin with amniotomy on the basis of three trials, for which meta-analysis gives inconclusive results. Of the two peer reviewed trials, the first concluded, after stepwise logistic regression, that oxytocin is effective. The second3 used oxytocin in a dose so low that 20 hours would be needed to reach the Dublin hospital's target dose, intended to ensure delivery within 12 hours.

Turner et al observed over 1000 consecutive labours at Northwick Park Hospital managed actively.4 Changes in rates of caesarean section and normal delivery had significance values of between P<0.05 and P<0.0001. The implication that changes of this order of significance arose as the result of poor randomisation or some factor other than change in practice when an entire obstetric population was studied strains credulity to its limits.

The National Maternity Hospital in Dublin has records of labour in over 200 000 consecutive women having their first baby. These data may be observational and non-randomised but cannot easily be dismissed, certainly not without explanation.

Thornton and Lilford's meta-analysis is seriously flawed. Meta-analysis may be useful for searching through piles of chaff, looking for missed grains of wheat, and thereby for examining issues on which trials have been inconclusive In this meta-analysis the chaff seems to have been added back to the wheat.5

T H Bloomfield 

Department of Obstetrics and Gynaecology, West Wales General Hospital, Carmarthen, Dyfed SA31 2AF

  1. Thornton JG, Lilford RJ. Active management of labour: current knowledge and research issues. BMJ 1994;309:366-9. (6 August.) [Abstract/Free Full Text]
  2. O'Driscoll K, Meagher B, Boylan P. Active management of labour. London: Mosby, 1993.
  3. Cohen GR, O'Brien WF, Lewis L, Knuppel RA. A prospective randomised study of the aggressive management of early labor. Am J Obstet Gynecol 1987;157:1174-7. [Medline]
  4. Turner MJ, Brassil M, Gordon G. Active management of labor associated with a decrease in the cesarean section rate in nulliparos. Obstet Gynecol 1988;71:150-4. [Abstract/Free Full Text]
  5. Hawkins DF. Clinical trials - meta-analysis, confidence limits and "intention to treat" analysis. J Obstet Gynecol 1990;10:259-60.

Add to CiteULike CiteULike   Add to Complore Complore   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us   Add to Digg Digg   Add to Reddit Reddit   Add to Technorati Technorati    What's this?

Relevant Article

Active management of labour: current knowledge and research issues
J G Thornton and R J Lilford
BMJ 1994 309: 366-9. [Abstract] [Full Text]

Access all current jobs at BMJ Group
Whats new online at Student BMJ
Listen to the latest BMJ Interview