Reply to article Published online by the European Journal of Pediatrics in July 2007.

Mitchell, E A, Wrapping a cot mattress in plastic does not explain the
continuing fall in SIDS mortality, European Journal of Pediatrics (online,
20 July 2007). Biased sampling and invalid statistical analysis.

Preamble
--------

1. The above paper authored by Professor E A Mitchell (and published online by the European Journal of Pediatrics in July 2007) reports data sourced from a postal survey of 400 mothers conducted in 2005 in Auckland (New Zealand) regarding practices adopted by the mothers for the purpose of SIDS (cot death) prevention.

2. On the basis of 280 responses, Dr Mitchell reported the incidence of
mattress-wrapping for SIDS prevention as being 21.7%; and concluded that even if mattress-wrapping were 100% effective in preventing SIDS, that practice could have produced a reduction in SIDS mortality in New Zealand of only about 22% (as compared with an actual reduction in SIDS mortality of 63%).

3. However, both the mattress-wrapping statistic derived from the survey
and Dr Mitchell's above-stated conclusion are invalid, on the following
grounds:

(a) Biased research sample in survey; and

(b) Invalid statistical analysis of data sourced from survey.

4. As stated in the researchers' National Application Form for Ethical
Approval, the publication forum proposed by the research team was the New Zealand Medical Journal. It is highly noteworthy that despite publishing other results of the survey in comprehensive detail (refer: SIDS-protective infant care practices among Auckland, New Zealand mothers, NZ Med J 2006; 119(1247): U2365, 15 December 2006), the New Zealand Medical Journal did not publish any of results of the survey in relation to either:

(a) The incidence of mattress-wrapping for SIDS prevention as found in the survey; or

(b) Conclusion/s drawn from that incidence statistic.

5. As notified to the New Zealand Medical Association (the publisher of the New Zealand Medical Journal) on 1 November 2005, the research sample in Dr Mitchell's study was statistically baised. A copy of that notification to the NZMA is available on request.

Biased research sample
----------------------

6. Dr Mitchell's statistic for the incidence of mattress-wrapping for SIDS
prevention (21.7%) was derived from a postal survey titled “Survey of infant care practices that have been recommended for SIDS prevention” which purported to report (inter alia) the incidence of mattress-wrapping in the general community: National Application Form for Ethical Approval submitted by the researchers, Part III (Project Details: Aims of Project), Section 1.1(7).

7. The survey participants were sourced from the birth list at Auckland
City Hospital, a New Zealand Ministry of Health public hospital: Ibid,

Part III Project Details: Participants, Section 3.2.

8. The New Zealand Ministry of Health policy in respect of
mattress-wrapping is that there is no evidence of a link between mattresses and SIDS risk. Accordingly, sourcing research participants from the birth list at Auckland City Hospital (many of whom could be expected to have attended antenatal classes run by the hospital and at which the Ministry of Health policy on mattress-wrapping was presented) will have yielded unreliable statistics on the incidence of mattress-wrapping in the general community.

9. In order to obtain reliable statistics on the incidence of
mattress-wrapping in the New Zealand general community, it is necessary to source research participants from records of the New Zealand Registrar of Births - not records of a Ministry of Health hospital. Certainly, it would be possible to carry out a reliable study into the incidence of mattress-wrapping in the general community, but not by the means used in Dr Mitchell's study. The research sample was biased.

10. Prior to publication of Dr Mitchell's research by the European Journal
of Pediatrics on 20 July 2007, it was stated by a Board Member of the New Zealand Child Health Research Foundation that in respect of surveying the incidence of mattress-wrapping for SIDS prevention, the research may be subject to biased sampling; and the point was also noted by the New Zealand Ministry of Health.

11. As long ago as 2000, two studies published in the New Zealand Medical Journal found the incidence of mattress-wrapping for SIDS prevention to be 23%: NZ Med J 2000; 113: 8-10; NZ Med J 2000; 113: 326-327.

Invalid statistical analysis by Dr Mitchell
--------------------------------------

12. Even using Dr Mitchell's mattress-wrapping incidence statistic of
21.7%, his mathematical analysis of the data sourced in the survey is
invalid.

(a) In the context under consideration, since the commencement of
mattress-wrapping in New Zealand in 1995 there have been two groups of
babies in New Zealand:

Babies who slept on wrapped mattresses: "Case babies"

Babies who did not sleep on wrapped mattresses: "Control babies"

(b) The total number of live births in New Zealand during the years 1995 to 2007 (inclusive) was about 629,400

(c) Accordingly, on the basis of an incidence of mattress-wrapping of
21.7%, the number of Control babies was 629,400 x 0.783 = 492,820

(d) The number of Case babies was 629,400 x 0.217 = 136,580

(e) The number of SIDS deaths in New Zealand during the years 1995 to 2007
(inclusive) was 830 (approximately)

(f) There has been no reported SIDS death among the Case babies, i.e.
babies who slept on wrapped mattresses.

Determination of statistical efficacy of mattress-wrapping in preventing
SIDS
------------------------------------------------------------------

13. If mattress-wrapping had nil effect on the incidence of SIDS, the
number of SIDS deaths among the "Case" babies would have been:

830 / 492,820 x 136,580 = 230

Since there has been no SIDS death among Case babies, the efficacy of
mattress-wrapping in preventing SIDS was in the ratio 230/zero.

14. However there is no meaningful solution to the equation x = a/zero
where <a> is a real number. This situation can be overcome by positing that the number of SIDS deaths among "Case" babies was "very much less than 1, denoted by the conventional diacritic a = <<1. (The diacritic >>1 means very much greater than 1.)

Using this nomenclature the ratio of effectiveness of mattress-wrapping, in
the prevention of SIDS can be stated as:

Effectiveness = 230 divided by <<1, which equals >>230

This means that the efficacy of mattress-wrapping in preventing SIDS is much greater than 230 times the efficacy of all other SIDS prevention advice combined.

15. Accordingly, Dr Mitchell's conclusion that mattress-wrapping "cannot
account for the 63% decline in SIDS" in New Zealand is totally
unsubstantiated; and it is axiomatic that his contention (as reported in New Zealand on 11 December 2007) that mattress-wrapping has not contributed to the continuing decline in SIDS mortality in New Zealand is also invalid.

Conclusions
-------------

16. The data regarding mattress-wrapping for SIDS prevention among the New Zealand general community (as previously published in the New Zealand Medical Journal, refer no. 11 above) provides for mathematical analysis which proves conclusively that mattress-wrapping in accordance with a specified protocol prevents SIDS.

17. The European Journal of Epidemiology should have declined to publish Dr Mitchell's paper "Wrapping a cot mattress in plastic does not explain the continuing fall in SIDS mortality". The research sample was demonstrably biased; and the stated conclusion is statistically invalid. Furthermore, that conclusion constitutes product defamation of commercial products being marketed for mattress-wrapping for SIDS prevention (namely, BabeSafe products).

18. It is noteworthy that:

(a) All of the six researchers originally constituting Dr Mitchell's
research team for the "Survey of Infant Care Practices" (paediatricians Dr E Mitchell, Dr L Hutchison, Dr M Battin and Dr D Tipene-Leach; researcher Ms R Haretuku; and statistician Dr A Stewart - all of New Zealand) received the notification to the New Zealand Medical Journal dated 1 November 2005 pointing out statistical bias in the survey; and that

(b) Of that original research team, only three are listed as co-authors of
the paper "SIDS-protective infant care practices among Auckland, New Zealand mothers" published by the New Zealand Medical Journal in December 2006; and that

(c) Dr Mitchell himself is the sole author of the paper "Wrapping a cot
mattress in plastic does not explain the continuing fall in SIDS mortality"
published by the European Journal of Pediatrics on 20 July 2007.

19. By separate correspondence the Editors-in-Chief of the European Journal of Paediatrics have now been placed on legal notice of various matters relating to the "Survey of Infant Care Practices" (which were also notified to the New Zealand Medical Journal on 1 November 2005). It can be reliably assumed that the New Zealand Medical Journal elected not to publish the results of the "Survey of Infant Care Practices" in respect of mattress-wrapping on the grounds of (inter alia) flawed epidemiology by Dr Mitchell's research team by biased sampling; and potential legal risk of product defamation.

20. The publication "New Zealand Doctor" (which reported the publication of Dr Mitchell's research on 11 December 2007 in "New Zealand Doctor Online") has also been placed on legal notice of those matters; and a published retraction of various claims contained in their online report will be demanded.

T James Sprott OBE MSc PhD FNZIC
Consulting Chemist & Forensic Scientist
10 Combes Road
Remuera
Auckland 1050
NEW ZEALAND