Truth or Fiction?:
A review of a new medical
"study" on home birth in Dublin

© Marie O’Connor

Erroneously, this article, published in the Irish Medical Journal by P MacKenna and T Matthews, is entitled, "Safety of Home Delivery Compared with Hospital Delivery in the Eastern Regional Health Authority (sic) in Ireland in the Years 1999-2002". A close reading of the "study" shows the authors' claim to be hollow.

The abstract, all that appears in most databases, and the only text read by many, is inaccurate and misleading. The summary claims, falsely, that "If the intended place of birth is home, the chance of dying due to intrapartum hypoxia is 1: 70 (5 in 346). If the intended place of birth is hospital, the chance of dying is 1: 3600 (17 in 61, 215)".

The authors set about constructing these rates imaginatively, apparently uninhibited by the need to adhere to scientific norms, counting selected births and deaths from 1999 - 2001 and/or 2002, in selected locations, varying from virtual to actual, in an expanding time frame (before, during, or any time after birth). Bias runs through the text like a giant faultline, embedded deep in the layers of the "study's" foundations, and culminating in the authors' wholly unsubstantiated conclusion that "planned home birth is a less safe procedure than hospital births (sic) in Dublin".

Bizarrely, home birth is defined as "a birth that at onset of labour is intended to occur at home, with the assistance of an independent midwife". But excluding home births attended by hospital midwives reduces the size of the home birth pool, inflating the so-called death rate. The smaller the group, the more vulnerable it is to distortion.

Curiously, the authors define home births as including hospital transfers during labour: no statistics exist for these births, however, as the authors themselves readily admit.

"Mortality rates" have been constructed through a series of statistical manouevres, attributing births and deaths to home or hospital, depending, in some cases, on where the mother intended to give birth, or, in others, on where the baby was actually born, or, in a further highly select group, on the outcome of the birth.

Moving the goalposts like this results in death rates that deceive: hospital mortality rates, for example, are staggeringly under-reported for normal-weight babies, by a factor of 9-15, while community rates are inflated by a multiple of three, at least.

These imaginative mortality rates have been produced by using the shifting sands of overlapping classifications; cherrypicking the data, then manipulating it still further; splitting key categories, such as stillbirths and hospital transfers, then subdividing them; coding some births on the basis of intentionality, and others on the basis of actuality; assigning births to home, then re-assigning them to hospital, and then back to home. Or vice versa: the permutations in this deeply skewed piece of work are endless.

The "study's" focus on 1999-2002 has produced a massively distorted picture of birth in the community. Over 1,000 babies, for example, are estimated to have been born at home under the care of independent midwives in the ERHA Region between 1979 - 1998: excluding babies who died from congenital problems, there was one stillbirth, and no neo-natal death. These results show that independent midwives' outcomes in the Region compare very well with those published in the scientific literature on home birth.

The authors based their work, not on a random sample derived from a scientifically sound database, but on five selected deaths that occurred around the time of birth, or at some point thereafter, either at home or in hospital, between 1999 and 2002, where the mother is said to have planned a home birth. These five cases form the scaffolding for this spurious work: the rest amounts to what can only be described as statistical invention, concocted from a mish-mash of secondary, or, in the case of home births, non-existent sources.

Hospital data were taken from the main Dublin maternity hospitals' annual reports: these reports do not use comparable definitions, making comparisons invalid and misleading.

Worryingly, the source quoted for home births, the "regional interactive child health system in the Southern Health Board", does not exist, as the second author confirmed on RTE Radio One's "Morning Ireland" on 11 September. While this system did not exist in that Health Board, he said, it existed in another health board, the South-Western Area Health Board: efforts to locate the "system" in that health board have proved fruitless, however.

But the imaginary roots of the "study" go even deeper. The abstract claims that it is based on where mother intended to give birth, not on where their babies were actually born: no official information exists on mothers' intentions in this area.

Another troubling feature of the study is that hospital death rates are so seriously under-reported: in contrast to the 17 deaths claimed by the authors over 1999-2002 for the Rotunda, Holles Street and the Coombe, the total number of stillbirths and first week-of-life neonatal deaths estimated to have occurred in these hospitals, in babies weighing 2,500g or over, is 150-250, depending on the deaths selected. Remember also that these figures understate the actual number of babies who died in these hospitals during the perinatal period: they exclude babies, some on ventilators from birth, who died after 7 days.

How did the authors whittle down the number of hospital deaths so much? The answer is by focussing on selected deaths and ignoring others. Splitting the stillbirth category twice, for example. Including only deaths said to have occurred during labour, and within this artificially reduced group, making a further selection, counting only those said to have been caused by lack of oxygen. This left a total of 17 deaths under the heading "intrapartum hypoxia".

But this category is doubly suspect: hypoxia, or oxygen deficiency, is a classification that lends itself to creativity, and intrapartum is a term that expand and contracts at will. Oxygen deficiency is common to all deaths: to classify deaths by hypoxia, as the authors have done, is to play with statistics. And the "intrapartum" time zone is far from watertight, as when labour begins is a matter of dispute. Hospital records on this point are creative, as they refuse to recognise time spent by a woman in labour at home as labour (!): on admission to the labour ward, the graph is marked zero. While this Alice-in-Wonderland approach results in inaccurate and misleading statistics, it is advantageous: all deaths, or most, occurring before a mother is admitted to the labour ward can now be classified as antepartum (before labour begins). For to classify deaths as intrapartum is to raise the spectre of doctor-induced death.

How did the the domiciliary death rate come to be so grossly inflated? Double standards ruled, and in such a tiny group, this led to severe distortions. Of the five deaths allegedly due to oxygen deficiency during labour, at least three are misrepresented. The first one, for example, is classified in the National Maternity Hospital's annual report for 2000 as a hospital death due to infection. In the second case, due to a very rare maternal condition undiagnosed by hospital tests, the baby died before labour began. The third was a stillbirth that, in hospital, would have been classified as a cord death. Excluding these three deaths reduces the mortality rate that the authors attributed to the community by three-fifths, putting it within the accepted range for planned home births.

One detail: the cord-related stillbirth took place in 2002: this may explain why the authors decided to include 2002 for domiciliary births, while omitting equivalent figures for hospital births. In such a tiny group, including 2002 increased by 20 per cent the mortality rate that could be now be attributed to it, while hospital deaths for that year remained off-camera.

Other worrying questions remain. Despite the authors' stated focus on planning, for example, they excluded planned hospital births that took place outside hospital (BBAs). Births such as as these carry very high mortality rates: including them would have significantly worsened the hospital death rates. Then there is the geographic base claimed by the study: how can a geographic region properly be said to apply to what is, in effect, a virtual study, based on mothers’ intentions?

But perhaps the most disquieting feature of this deeply disturbing work is the authors' implicit suggestion - and it is one that runs through the entire article - that the outcome for these babies would have been different in hospital. Babies die from infection, maternal conditions and cord complications in hospital every week, as do babies who die from unknown causes. To pretend otherwise is disingenuous.

Nevertheless, the authors' assertion that place of birth may be an important factor in determining outcome is not without foundation, as a study done by the second author at his own hospital shows. A study of cerebral palsy published in the Irish Medical Journal in 1992 analyses the history of 28,655 term and post-term babies, born alive at the Rotunda. The research reveals that there were 13 deaths due to intrapartum asphyxia, while a further 32 babies suffered seizures: of these, 19 per cent subsequently developed cerebral palsy. Syntocinon, or oxytocin, was used in 44 per cent of infants who seized, and in 31 per cent of those who died. The significance of this finding was not analysed by the authors: oxytocin is central to "active management" in Irish hospitals.

To return to home birth: this deeply misleading piece of work will doubtless be hailed by home birth opponents as a major contribution to the "debate" on the "place" of birth. But this debate has always cloaked another, even more fundamental issue, namely, the service provider question, or who should control (and profit) from birth. In this "study", however, the provider issue emerges as central to the authors' concerns. This may ultimately represent their real contribution to the ongoing debate on home birth.

To redress the balance between truth and fiction in home birth, what is needed now is a 20-year retrospective analysis of independent midwives' outcomes that, unlike this exercise in data manipulation, conforms to established standards in scientific research.

© Marie O’Connor

Marie O'Connor is a sociologist, health correspondent and author of: Women and Birth: A National Study of Intentional Home Birth in Ireland, and Birth Tides; turning towards home birth (Pandora, London 1995). A frequent contributor to scientific journals and international conferences, her work has been translated into a number of European languages.

The author wishes to thank Dr Krysia Rybaczuk, of Trinity College, Dublin, for her valuable insights, comments and refutations in the development of this critique.


© National Birth Alliance
An Chomhghuallaiocht Naisiunta Breithe