Thursday, December 16, 2010

A GP practice withdraws collaboration

click to enlarge

The letter copied above, with the name of the doctor who wrote it blocked out, was sent to this blog by its recipient. This letter was written by one of the GPs in Bendigo who, in the past, had been willing to support women choosing homebirths.  The main 'support' received in such cases is that the doctor orders routine blood tests and investigations, and facilitates access to the blood bank for women with Rhesus Negative blood to receive prophylactic Anti-D.  One would wonder what 'Duty of care' this doctor perceives is appropriate for these women once this facility is no longer available?

The writer makes reference to a paper comparing planned home births with planned hospital births, published in the American Journal of Obstetrics and Gynecology 2010, by Joseph R. Wax and colleagues.
[Maternal and Newborn Outcomes in Planned Home Birth Vs. Planned Hospital Births: A Meta-Analysis, Wax JR, Lucas FL, Lamont M, et al., Am J Obstet Gynecol 2010]

Using various statistical analyses, this author claimed that there is a three times higher neonatal mortality rate of babies in the planned home birth group than in the hospital birth group. This paper has a familiar ring to it: readers will recall the AMA publication of Kennare et al's retrospective analysis of data from South Australia, and the outrageous conclusions made by the authors. For links and comment, go to the MIPP blog.

A voice of reason has been published by Medscape Ob/Gyn & Women's Health, a respected online medical review site.

Medscape convened a 'Roundtable' in Ob/Gyn & Women's Health,
Experts Argue the Continuing Home Birth Issue
Perspectives on the Joseph R. Wax and Colleagues Home Birth Study in AJOG

The following excerpt is copied from Section 6: Eugene Declercq, PhD

After a long decline,[10] the number of home births has started increasing in the United States. ... Four years do not make a trend, but the rate of home birth is the highest it has been since 1994 and the total number of home births (26,667) is the highest since 1991.

It is in this context that we confront numerous problems faced by researchers trying to resolve the question of safety of home birth. Limited space prevents a full discussion of the barriers to such research, but 3 important ones are:
  1. Design. The gold standard -- a randomized trial -- is not feasible because women will not let themselves be randomly assigned to a given birth site. To argue that the safety of home birth can only be established by a randomized trial presumes that home births are unsafe. Because a randomized trial is impossible, then logically home birth can never be shown to be safe.
  2. Measurement. Because our focus is on planned home births, how do we define such births to make sure we exclude accidental home births (eg, precipitous labor) and include home births that result in transfers to the hospital? Current US data systems do not facilitate such tracking, although some US studies have tried to address this problem as have several recent studies from other countries.
  3. Statistical power. Because planned home births typically involve healthy women, poor outcomes are rare, and hence very large samples are needed to identify differences.

These limitations are among the constraints that Wax and colleagues[1] tried to overcome with their meta-analysis of studies that met certain criteria. The dearth of appropriate studies resulted in inclusion of some studies that are old (data from the 1970s),[25] studies that inferred rather than documented planning status, and studies that were primarily from other countries (only 2 of the 12 were US-based). The decisions that have drawn the most criticism have involved the largely unexplained inclusion and exclusion of certain studies. The inclusion of data from Pang and colleagues' study,[26] done in Washington state, drew fire because it was the source of most of the reported neonatal deaths and it could only infer planning status, which was defined as "home births of singleton newborns of at least 34 weeks' gestation [with] a midwife, nurse or physician listed as either the birth attendant or certifier..." The difficulty with that definition is seen in a recent 19-state study of births occurring at home in which planning status was indicated.[27] In that study, I and my colleagues found that 87% of the unplanned home births occurred after 34 weeks and 69% of the home births attended by physicians were unplanned. In addition, 22% of the planned home births were attended by "other" attendants; these would have been excluded by Pang and colleagues.[26] This means that Pang and colleagues probably both erroneously included physician-attended full gestation home births in the planned home birth category and excluded "other" attended planned home births. In fairness, our study was published in the same month as Wax and colleagues' study and years after Pang and colleagues' study, but one cannot now claim that the potential problems with selection bias simply cancel each other out.

Of greater consequence was Wax and colleagues' exclusion of the largest study of planned home births ever done: that by de Jonge and colleagues[7] in the Netherlands, with more than 300,000 planned home births. Wax and colleagues' decision to use neonatal mortality at 28 days as their primary outcome unfortunately led them to exclude that study, which reported on perinatal and neonatal mortality up to 7 days. After controlling for confounders, de Jonge and colleagues reported no added mortality risk in planned home births. The size of their study overwhelms that of Wax and colleagues' meta-analysis and was the basis on which the meta-analysis found no difference in perinatal mortality. Perinatal mortality, which includes fetal deaths, would seem a more appropriate outcome measure. Neonatal mortality is also widely used, but how important is the distinction between 7 and 28 days? Over the past decade in the United States, about 80% of all neonatal deaths up to 28 days have occurred in the first 7 days. Apparently, de Jonge and colleagues are now examining their data out to 28 days. Would anyone seriously suggest that such a great concentration of home birth-related deaths occurring between 7 and 28 days would alter their core finding of no added risk? If de Jonge and colleagues find no difference at 28 days, Wax and colleagues' meta-analysis would also most likely find no overall difference in neonatal mortality. What happens then -- are their findings withdrawn? Does the journal publish a retraction, even after it made Wax and colleagues' article an "editor’s choice"?

Currently under way is the Birthplace in England study, another large, well designed study of home, birth center, and hospital births. The fact remains, however, that the debate over home birth will not be resolved by this or any study, no matter how well designed. At its core, the home birth debate is ideological, centering on 2 diametrically opposed perspectives on birth held by groups that generally do not communicate with each other and unfortunately often hold each other in disdain. Perhaps the question we should be asking is not what is right or wrong about any study on this topic. Rather, why are increasing numbers of US women who are experienced in birth (80% with parity 2 or higher) choosing to reject hospital-centered systems of maternity care that so many well-meaning clinicians want to make better?

No comments: