Outcomes of planned home birth with registered midwife versus planned hospital birth with midwife or physician

Dear Dr. Cardwell: Thank you for your observation. The perinatal death rates in the text are a reflection of the column headings being reversed for planned hospital midwife vs. physician births in the last table of our paper which was the table in the appendix. The rate of perinatal death for the planned midwife-attended hospital group was 3/4723 or 0.64/1,000 as you point out and for the planned physician-attended hospital group was 3/5294 or 0.57/1,000. So the text with respect to perinatal death rates should read: 0.64 (95% CI 0.00-1.56) among the planned hospital births attended by a midwife and 0.57 (95% CI 0.00-1.43) among the planned hospital births attended by a physician. I apologize for the ongoing confusion this has caused – Patricia Janssen

Conflict of Interest:

None declared

There are inconsistencies in the statistics.

The numbers in the revised appendix (and Author's response to: Statistics Confusing) are not consistent with the calculated statistics in the article. For example, the rate of perinatal death in the planned hospital, midwife-attended group is indicated throughout the article to be 0.57 per 1000 births, but 3 deaths/4723 births is a rate of 0.64.

Conflict of Interest:

None declared

Dear Dr. Oyston: Thank you for your letter and persistence in pointing out your concerns about Table 1 in the Appendix. The transfer of this table from the main body of our table to an Appendix has caused considerable confusion. Newborn outcomes were analyzed within the planned home/hospital births as an "intention to treat" analysis. When we analyzed newborn outcomes, newborns with congenital anomalies were excluded. The number in the planned home birth group was 2882 (2899 minus 17 with anomalies); in the planned hospital, midwife-attended group was 4723 (4752 - 29) and in the planned hospital, physician - attended group 5294 (5331 minus 37). Unfortunately, the headings for the planned hospital birth with a midwife and physician were reversed, hence your confusion about the number of newborns in each group. This error was not caught because this table was initially deleted and then put back in as part of our discussions with editorial staff. The number of deaths was 1/2882 in the planned home birth group; 3/4723 in the planned hospital, midwife-attended group and 3/5294 in the planned hospital physician-attended group. I have sent this revised table to CMAJ. Thank you again for bringing this error to our attention - Patricia Janssen, PhD, author.

Conflict of Interest:

None declared

I appreciate the editors pointing out the online appendix which provides the actual raw numbers of deaths in each group.

When the "primary outcome measure was perinatal mortality" I would have expected the number of perinatal deaths in each group to be in the main body of the article.

Unfortunately the appendix confuses me more. There are three groups - Midwife/Home, Midwife/Hospital and MD/Hospital. The introduction says there were 2889, 4752 and 5331 women in each group, presumably based on intention to treat. Appendix 1, from which the published perinatal death rates are calculated, has 2882 women and one death, 5294 women and three deaths and 4723 women and three deaths in these group, presumably based on actual mode of delivery.

Somehow there are 608 fewer MD/Hospital births than planned and 542 more Midwife/Hospital births than planned. Although the main text states that 4604 of those who planned Midwife/Hospital births did so, the death rate for this group is calculated on 5294 births. Presumably 690 women who intended other birth types ended up in this group.

Surely the death rate should be calculated based on intention to treat? Shouldn't the authors let readers know what the original planned method of delivery was for each of the seven deaths, especially given the large movements between treatment groups?

Conflict of Interest:

None declared

I have had some difficulty understanding the statistics in this article. It is unfortunate that while the exact number of events is supplied for interventions and maternal outcomes (so that we know, for example, that 224 women in the planned home birth group received epidural analgesia) the neonatal outcomes are only reported as relative risk, which obscures the number of events. In particular, the number of perinatal deaths in each group is not clearly apparent, even though this was the main outcome. The authors quote a rate of 0.57 deaths per thousand in the 4752 planned midwife hospital births, but I calculate this means there were 2.7 deaths in this group. In the 5331 physician hospital births the death rate was 0.64 per thousand, which works out to 3.4 deaths. Surely there has to be a whole number of deaths? Reworking the statistics based on actual method and place of delivery rather than the planned ones does not resolve this dilemma. Surely the authors should have provided a table with the number of deaths, low apgar scores, and other outcomes in each group? Yours John Oyston

Conflict of Interest:

None declared

This is a population-based study which used data from a provincial perinatal database with no personal identifiers. The charts and therefore the clinical details surrounding each death were not available to the authors. In Canada, stillbirth is defined as intrauterine death after 20 weeks gestation. Early neonatal death is from birth to 7 days. Together this time period is referred to as perinatal death and it is a standard mortality rate in Canada. We also examined the time period between 8-28 days of life and there were no deaths during this period (late neonatal period) in any of the three comparison groups.

Following this first broad data search, the three groups were matched for comparison. All of the groups had to meet the eligibility requirements for home birth, regardless of planned place of birth. Only those births that occurred between 37+0 weeks and <42 weeks were included in the final analysis.

Conflict of Interest:

None declared