Re: Editorial; and Perinatal Death or Perineal Trauma?
Shorten and Shorten (1) and Gyte et al (2) have criticised some of
the features of the matching used in our recently reported study (3).
Shorten and Shorten state that there were differences in nutritional
status, smoking status, and alcohol consumption during pregnancy. This is
an assumption as we did not report any such differences, stating
explicitly that NHS data concerning these factors were not available. We
were able to report certain factors from the IMA database about these
factors, because they were available. These factors could, in fact, also
be similar in the NHS group. What we reported was a review of possible
factors that might explain the higher perinatal mortality rate in the IMA
group. However, even if these factors were available they would have to be
very strong confounders to explain such large differences in mortality.
Our design took account of all the strong confounding factors such as
socio-economic status, age, parity, breech presentation, twins, etc,
likely to impinge on the outcome and hence is a strength of the study
rather than a weakness.
The editorial is correct to note that there were differences between
the two groups in terms of obstetric risk from previous pregnancy, medical
complications during the current pregnancy, and incidence of breech
presentation; there were also differences in preterm birth rates, and the
incidence of low birth weight. The first three factors were controlled for
in our multivariable logistic regression; the other two, of course, are
clinical outcomes for the current pregnancy, and so could not be part of
any matching process.
We concede that there were difficulties in matching a self-selecting
atypical group with the general population, and were careful not to claim
that our results could be extrapolated. We disagree with Hassan (4): the
women in the IMA group were not, “by definition…low risk”; and we did
account for some of these self-selection factors in terms of socioeconomic
status, parity, age, and high risk factors. In any case, such self-
selection by this argument should have led to lower mortality, not higher.
Nevertheless, we did find some remarkable results. It is unlikely that
other factors that were not available could explain such large differences
in such a relatively large study. The case note review, as reported by
Gyte et al (2), is underway. We acknowledge Gyte et al’s support for our
call for urgent research into the apparently higher incidence of preterm
birth and low birth weight in the NHS cohort. This is potentially a very
serious public health and health economic issue.
1) Shorten A, Shorten B Independent midwifery care versus NHS care in
the UK (editorial). BMJ 2009;338:b2210
2) Gyte G et al Why do women book with independent midwives? Rapid
response
3) Symon A, Winter C, Inkster M, Donnan PT. Outcomes for births
booked under an independent midwife and births in NHS maternity units:
matched comparison study. BMJ 2009;338:b2060
4) Hassan S. Perinatal Death or Perineal Trauma? Rapid response
Competing interests:
AS, CW and PD are the original study authors
Rapid Response:
Re: Editorial; and Perinatal Death or Perineal Trauma?
Shorten and Shorten (1) and Gyte et al (2) have criticised some of
the features of the matching used in our recently reported study (3).
Shorten and Shorten state that there were differences in nutritional
status, smoking status, and alcohol consumption during pregnancy. This is
an assumption as we did not report any such differences, stating
explicitly that NHS data concerning these factors were not available. We
were able to report certain factors from the IMA database about these
factors, because they were available. These factors could, in fact, also
be similar in the NHS group. What we reported was a review of possible
factors that might explain the higher perinatal mortality rate in the IMA
group. However, even if these factors were available they would have to be
very strong confounders to explain such large differences in mortality.
Our design took account of all the strong confounding factors such as
socio-economic status, age, parity, breech presentation, twins, etc,
likely to impinge on the outcome and hence is a strength of the study
rather than a weakness.
The editorial is correct to note that there were differences between
the two groups in terms of obstetric risk from previous pregnancy, medical
complications during the current pregnancy, and incidence of breech
presentation; there were also differences in preterm birth rates, and the
incidence of low birth weight. The first three factors were controlled for
in our multivariable logistic regression; the other two, of course, are
clinical outcomes for the current pregnancy, and so could not be part of
any matching process.
We concede that there were difficulties in matching a self-selecting
atypical group with the general population, and were careful not to claim
that our results could be extrapolated. We disagree with Hassan (4): the
women in the IMA group were not, “by definition…low risk”; and we did
account for some of these self-selection factors in terms of socioeconomic
status, parity, age, and high risk factors. In any case, such self-
selection by this argument should have led to lower mortality, not higher.
Nevertheless, we did find some remarkable results. It is unlikely that
other factors that were not available could explain such large differences
in such a relatively large study. The case note review, as reported by
Gyte et al (2), is underway. We acknowledge Gyte et al’s support for our
call for urgent research into the apparently higher incidence of preterm
birth and low birth weight in the NHS cohort. This is potentially a very
serious public health and health economic issue.
1) Shorten A, Shorten B Independent midwifery care versus NHS care in
the UK (editorial). BMJ 2009;338:b2210
2) Gyte G et al Why do women book with independent midwives? Rapid
response
3) Symon A, Winter C, Inkster M, Donnan PT. Outcomes for births
booked under an independent midwife and births in NHS maternity units:
matched comparison study. BMJ 2009;338:b2060
4) Hassan S. Perinatal Death or Perineal Trauma? Rapid response
Competing interests:
AS, CW and PD are the original study authors
Competing interests: No competing interests