Sunday, August 23, 2009

New Studies on Subsequent Children: Revisiting the 'Replacement Child Syndrome': The Studies did not find 'no effect'.



Recently in the news, there was widespread coverage of a research study by Dr. Penelope Turton in the Journal of Child Psychology and Psychiatry regarding the effects on the next child following a stillbirth.

A summary article appears at Stillbirth Not an Issue for Next Born in Forbes.com

Some are misreading both the study and misapplying the findings to suggest that there is no risk in the pregnancy subsequent to a stillbirth for the child born later. This seemed to negate the observation of hundreds of researchers and counselors who have noted for 50 years the existence of what has been termed the 'replacement child syndrome', a phenomenon in which many feelings about the 'ideal' child that died are overlaid on the feelings regarding the next child who survives. Many historical examples have been noted of pathological examples in which the 'new' child cannot live up to the expectations of what the previous child would have been like.

The actual findings in the new 2009 study in Journal of Child Psychology and Psychiatry, and a previous article in Attachment and Human Behavior, included the following statements in the two abstract:

We report that in this population there was no significant correspondence between U and PTSD scores or caseness and no association between maternal PTSD and infant D. We discuss possible interpretations of these findings.

Recent Study: Results: There were no significant between-group differences in child cognitive or health assessments, or in teacher-rated child difficulties. However, mothers with history of stillbirth (the index group) reported increased child difficulties, in particular peer problems, and more adverse interaction was observed in respect of higher levels of maternal criticism of the child's actions, more overall controlling behaviour by the mother, a less harmonious emotional atmosphere and a lower level of maternal engagement with the child. Some of these effects appeared to be mediated by maternal perinatal psychological symptoms and family breakdown.


Two major issues of are deep concern to the research and comfort community in Perinatal bereavement:

1. The study in no way says there are no differences at all: the study says there are no differences in teacher rated child difficulties, but that there are a number of differences in maternal reports! Serious differences contrary to the way the study was reported.

2. A fundamental error that is made in the social sciences among those who are doing field research is that when we find no significant difference between groups or measures, in this case, no difference between subsequent children following stillbirth vs regular children in teacher-ratings, we cannot say at all that there are no differences, only that we cannot support the 'Null' hypothesis.


In research especially field research (research that takes place in a natural setting, and not in a lab), countless variables and possible errors may confound our results. These natural 'errors' may exaggerate our data, or cause it to look like there are no differences when there really are. It is a common and deadly error over the years in Perinatal Bereavement research and must be heeded. Even those in their first quarter of college statistics learn that when we do not see a statistically significant difference, it does not mean that there was not one, only that it did not appear given our methodology and choice of statistical test. There can be many reasons that there appears to be no difference:

1. Raw Error
2. Unknown factors
3. Faulty methods such as poorly designed rating scales or assessments
4. The wrong statistical test, or the right one, lacking power or enough data
5. Confounding factors such as nonrandom selection, a common problem in field research, historical factors, meaning that either during the study an event happens which changes the data, or for example a difference in parity, maternal age, or time since loss which might not have randomized out.
6. Design facts

and others.

For this reason, virtually all Statisticians and trained researchers know that they cannot draw conclusions and say there are 'no differences' just because a statistically significant difference was not found. Two studies which did the same thing, concluded 'no differences' when they should not have, were that of Peppers and Knapp in the early 1980s who claimed to have found no difference in grief between women who lost babies in the first, second or third trimester, (which is not logical to those who have worked with many mothers) or a study by Kellner et. al[: Links
Kellner KR, Donnelly WH, Gould SD. Parental behavior after perinatal death: lack of predictive demographic and obstetric variables.Obstet Gynecol. 1984 Jun;63(6):809-14. ]

in which a less powerful statistic found 'no differences' when all they should have said was that they were unable to detect one. They used a chi square instead of a multivariate design, and seemed to contradict the earlier study of Kirkley-Best (1981), the first prospective and controlled study, which yield a significant difference with the highest predictive variable being length of gestation, followed by maternal parity. The Kirkley Best study has been replicated twice, by an extensive Swedish study [5], and another stateside.

The reason for this caution is that

1. The study actually did find a difference in maternal report, supporting all earlier observations (See The Forgotten Grief: A Review of the Psychology of Stillbirth-section on 'Replacement Child Syndrome' and
2. Designs out in the field can almost never conclude 'no difference'.


The reason this is so significant is because the study rapidly hit the news services and appeared to negate the warning of so many researchers and counselors, including myself of the insidious effects of perinatal mourning on subsequent children and family relations. It is not that healthy families cannot compensate for their feelings in later births, nor does it indicate that the effects are always severe, but even 'face validity' tells that the feelings about a child that died will affect feelings about the child who comes next! The example is often given of Vincent Van Gogh, whose mother even named him after the deceased sibling and had him visit the grave on a regular basis. While that is an extreme and pathological example, it indicates that the process does attend future children, and needs to be addressed.

This is not to say it is a poor or unimportant study, but the more appropriate question to be asked, how to get at the variables affecting the severity of the 'carry over' feelings, so as not to hamper the child's development later, and to avoid such traumatizing issues as using the same name or 'suspending' a name for a later child, or any thing which causes each child, living or dead not to retain a God given identity.

A last note quickly so that I do not slip into my old moronic psychology professor self: I have no vested interest any longer in research issues or diagnostic issues,as I left the research a few years back since in this particular field, research yield almost nothing that careful observation could not, with out reducing grief and sorrow over the loss of an infant to yet one more problem of clinical pathology. My greater concern is that healthy happy wholesome children and families find their way through despair and mourning, while retaining the dignity and unique 'glory' of each child, including the one who died. Just a passing note:

1Cr 15:41 [There is] one glory of the sun, and another glory of the moon, and another glory of the stars: for [one] star differeth from [another] star in glory.



________________________________
1. Forbes.com
2. Penelope Turton 1 , William Badenhorst 1 , Susan Pawlby 2 , Sarah White 1 , and Patricia Hughes "Psychological vulnerability in children next-born after stillbirth: a case–control follow-up study" Division of Mental Health, St George's University of London, UK ; Institute of Psychiatry, King's College London, UK
Correspondence to Penelope Turton, Division of Mental Health, St George's University of London
3.Penelope Turton; Patricia Hughes; Peter Fonagy; David Fainman
journal Attachment & Human Development, Volume 6, Issue 3 September 2004 , 241 - 253
4. Kellner KR, Donnelly WH, Gould SD. Parental behavior after perinatal death: lack of predictive demographic and obstetric variables.Obstet Gynecol. 1984 Jun;63(6):809-14.
5.Ingela Radestad, Gunnar Steineck, Conny Nordin, Berit Sjogren, associate BMJ 1996;312:1505-1508 (15 June)Papers Psychological complications after stillbirth--influence of memories and immediate management: population based study
6 Peppers LG, Knapp RJ. Psychiatry. 1980 May;43(2):155-9.
Maternal reactions to involuntary fetal/infant death.
7. Kirkley Best E. Grief in response to prenatal loss: An argument for the earliest maternal attachment EK Best - 1981 -Doctoral Dissertation, Department of Psychology, University of Florida

No comments: