Help ensure that the Global Internet services
of the SIDS Network continue to grow!

Donate directly to the SIDS Network securely with a major credit card.

Other ways to help can be found here.

Expert Group to Investigate Cot Death Theories: Toxic Gas Hypothesis

Abstracts of Chapters
bullet Chapter 2
bullet Chapter 3
bullet Chapter 4
bullet Chapter 5
bullet Chapter 6
bullet Chapter 7
bullet Chapter 8
bullet Chapter 9
bullet Chapter 10
bullet Chapter 11

Chapter 2 - The Toxic Gas Hypothesis

The toxic gas hypothesis of Richardson and Mitchell claims that the primary cause of SIDS is poisoning by the gases phosphine, arsine and stibine generated by deterioration of cot mattress materials by microorganisms, in particular, a fungus called Scopulariopsis brevicaulis. Richardson cites his own experimental work and that of others as supporting the hypothesis, as well as other factors such as the chemistry of these elements, the use of related compounds as fire retardants and plasticisers in cot mattress materials and the incidence of SIDS over time. Increases in SIDS have been linked by Richardson to the increased use of phosphorus and antimony compounds in cot mattress materials, particularly plasticised PVC cot mattress covers. The decrease in SIDS following the ‘Back to Sleep’ campaign launched in Autumn 1991, was suggested by Richardson to have started more than two years before and to be due to his advice, in June 1989, that a new mattress should be bought for every new infant, or old mattresses covered with polythene to protect the infant from gases generated from the mattress material. Further publicity was given to the hypothesis when new data were presented on television in two programmes in the Cook Report series in November-December 1994, and more recently by a book by J Sprott published in New Zealand in 1996 and in the UK in 1997.

Back to Report Contents page

Chapter 3 - Scope of this Investigation

This chapter summarises the approach of the Expert Group to the task of investigating whether toxic gases are generated from fire retardant chemicals present in cot mattress materials and whether these are either the primary cause of SIDS or pose any danger to infants.

Back to Report Contents page

Chapter 4 - The Turner Report and subsequent work on the toxic gas hypothesis (1991 to 1994)

In March 1990 the Chief Medical Officer established an expert working group chaired by the late Professor Paul Turner (the Turner Committee) to consider the toxic gas hypothesis. The Turner Committee commissioned work that replicated Richardson’s experiments using more sensitive analytical techniques. They concluded that there was no evidence to substantiate the claims that microorganisms on cot mattress material produce toxic gases from antimony or phosphorus compounds present in the mattresses or their covers, even though very sensitive and specific methods were used to detect the gases. A second study designed to replicate Richardson’s experiments also found no evidence to support the hypothesis. The Turner Report contained a number of recommendations concerning the necessity for fire retardants in cot furnishings, the chemical purity of fire retardants, and investigations of the microbial infestation of cot mattresses and their covers. Richardson’s criticisms of the Turner Report are outlined. The two Cook Reports on television showed Richardson’s experiments and presented data on concentrations of antimony in the tissues of SIDS infants and in the hair of live infants and their mothers.

Back to Report Contents page

Chapter 5 - Repetition and extension of Richardson’s experiments on microbiological gas generation

Microbial generation of phosphine and stibine from phosphorus and antimony compounds in PVC cot mattress materials is central to the toxic gas hypothesis. Richardson claims to have detected the production of these gases when mattress covers from SIDS infants, which were contaminated with the fungus Scopulariopsis brevicaulis, were incubated in plates with silver nitrate or mercuric bromide indicator papers. The Expert Group considered it essential to commission further experiments to investigate Richardson’s observations. This work, which was conducted with his cooperation and his presence at key stages, involved incubating small portions of mattress covers from SIDS infants on plates of nutrient media. Once good microbial growth had become established, indicator papers were added to the plates so that any gas production could be detected by the development of a colour change.

Richardson agreed that the experiments, which were conducted at the Mycology Reference Laboratory in Bristol, followed his procedures; that the organisms that grew were identical in colonial appearance to those that he had observed in his own experiments; and that the indicator papers underwent similar colour changes to those seen in his tests. In addition to replicating Richardson’s experiments, the work which the Expert Group commissioned went further in that the organisms recovered from the mattress covers were formally identified by expert microbiologists and the chemical composition of the exposed indicator papers was analysed. Several essential control experiments, not done during Richardson’s original work, were also performed. By replicating and extending Richardson’s work, it was demonstrated that his interpretation of his findings was incorrect.

The predominant microorganism, recovered from all mattresses tested, was a buff-coloured slime which Richardson identified (on the basis of its colonial appearance) as an alternative growth form of the fungus S. brevicaulis. Microbiological and biochemical tests, however, indicated that the slime consisted of a mixture of Bacillus species, commonly found in the normal domestic environment. Test paper colour changes occurred whenever bacterial growth was present, whether or not there was any mattress material present on the experimental plates. This suggests that the colour changes were due to bacterial growth on the cultivation medium, and that the presence of mattress materials on the plates was irrelevant.

Chemical analysis of exposed test papers showed that the colour reactions were not due to deposits of phosphorus, arsenic or antimony. More sulphur was found in test papers exposed in plates containing bacterial growth than in those without such growth. This suggests that the colour changes might have been caused by the generation of volatile sulphur-containing compound(s) during bacterial growth on the cultivation medium. Further investigations, recommended by the Expert Group, have identified the sulphur-containing compound(s) as dimethylsulphide (Chapter 6).

Mould contamination of cot mattress materials appears to be common, but the nature and extent of contamination on the mattress covers from SIDS infants is similar to that found on control covers. In marked contrast to Richardson’s contention that S. brevicaulis is ubiquitous in used cot mattress materials, several subsequent investigations have demonstrated that contamination with this mould is rare, and that it is no more common in mattress materials from SIDS infants than in other used mattresses.

Back to Report Contents page

Chapter 6 - Laboratory investigations of the ability of microorganisms to produce volatile products from antimony and phosphorus compounds

Fungal (i.e. Scopulariopsis brevicaulis) generation of antimony and phosphorus trihydrides (i.e. stibine and phosphine) from parent compounds in PVC cot mattress materials is a key requirement of the toxic gas hypothesis. Our own repetition (Chapter 5) of Mr Richardson’s work, did not substantiate his experimental evidence of such reactions. We considered it essential, however, to undertake further research to establish:

bullet whether there were laboratory conditions in which biovolatilisation of antimony and phosphorus compounds could be achieved and, if so, to identify the gases formed;
bullet whether cot mattress samples, similarly tested, would react in an analogous way;
bullet the relevance of the findings to the conditions in an infant’s cot.

In this chapter we have considered the chemistry of the relevant Group V compounds and methods for their detection and identification, surveyed the published work on the biovolatilisation of phosphorus, arsenic and antimony, and summarised new studies undertaken by several independent groups. Much fundamental research has been completed using laboratory conditions designed to maximise the potential yield of any gases produced, and a variety of analytical methods to identify them.

Fungal volatilisation of arsenic to yield trimet hylarsenic ((CH3)3As), but not arsine (AsH3), occurred in several experimental conditions. The analogous biovolatilisation of antimony was achieved with difficulty and only under laboratory conditions optimised to increase the efficiency of gas generation and its detection.

The generation of trimet hylantimony ((CH3)3Sb), but not stibine (SbH3), from added inorganic antimony compounds by both S. brevicaulis and unidentified anaerobes in enriched soil samples was demonstrated. Under the same conditions bacteria and other fungi isolated from cot mattresses did not volatilise antimony.

Volatilisation of antimony compounds encapsulated in mattress PVC samples did not occur under any laboratory conditions. Experimental conditions were found in which small amounts of antimony could be extracted from PVC samples. These conditions would either inhibit/decrease the germination or growth of S. brevicaulis or destroy microorganisms normally present on cot mattresses, including S. brevicaulis. Some evidence of antimony biomethylation was found if sufficient amounts of antimony constituents were first extracted from PVC e.g. by heating at 110C in an autoclave or at 80C for three days, and subsequently tested with added S. brevicaulis in optimised laboratory conditions.

Phosphine and trimet hylphosphorus were not produced under the test conditions described by S. brevicaulis from cot mattress samples containing phosphate fire retardants and plasticisers, or from other phosphate compounds added to or present in the culture medium. No phosphorus gases were unequivocally generated when PVC and foam mattress materials were incubated under the anaerobic conditions in which trimethylantimony was generated by enriched soil cultures grown in antimony-supplemented media.

Arsine and trimet hylarsenic have not been generated from cot mattress samples.

Sulphur-containing gases, such as dimet hylsulphide ((CH3)2S) and dimethyldisulphide ((CH3)2S2), were identified in the headspace of some cultures irrespective of whether the fungus, S. brevicaulis, or mattress samples, were present. They result from bacterial conversion of sulphur-containing compounds in the culture medium (see Chapter 5).

As a result of our investigations we have established that antimony compounds can be volatilised by the fungus (S. brevicaulis) under very specific conditions which are wholly unlike any to be found in an infant’s cot. Even under these conditions there was no evidence that phosphorus, arsenic, or antimony encapsulated in cot mattress PVC could be volatilised.

Back to Report Contents page

Chapter 7 - Toxins and the Sudden Infant Death Syndrome

Toxicological explanations for SIDS are not new and a wide array of drugs, metals, pesticides and environmental pollutants have been investigated although none has been clearly implicated. A critical component of the toxic gas hypothesis is that phosphine, arsine and stibine or their methylated derivatives, cause SIDS by inhibiting activity of cholinest erases (i.e. they are anticholinesterases) . The function of one of the cholinesterases, acetylcholinesterase (AChE) , is to split acetylcholine, a chemical which is of critical importance in the normal functioning of the nervous system. Thus, according to the hypothesis, inhibition of AChE results in increased levels of acetylcholine in the blood and subsequent cardiac failure in infants. The features of SIDS, however, are not compatible with those of acute poisoning with anticholinesterases such as organophosphate or carbamate pesticides, and there was no reduction in mean brain AChE activity in the brains of SIDS infants. Furthermore, exposure to stibine does not inhibit plasma cholinest erase or acetylcholinesterase activity at toxicologically relevant concentrations. Moreover, although phosphine inhibits AChE in vitro, there is no evidence that this is a significant contributor to human poisoning with this gas. SIDS babies also do not show evidence of haemolysis which is the hallmark of poisoning with arsine and stibine. The limited data for the trimethyl derivatives suggest that these are considerably less toxic than the parent compounds. Lethal phosphine poisoning is characterised by pulmonary oedema of an intensity that is not found in SIDS. We conclude therefore, that SIDS is not due to poisoning by phosphine, arsine and stibine, or their methylated derivatives. In addition, the toxicity data for the fire retardant chemicals themselves, namely, antimony trioxide and phosphate based plasticisers, do not suggest that these are a cause of SIDS, nor that they present any harm to infants.

Back to Report Contents page

Chapter 8 - The pathology of SIDS, normal infant developmental physiology and insights into possible pathophysiological mechanisms

Sudden deaths of unknown cause must be reported to the coroner (Procurator Fiscal in Scotland), who will almost invariably order that a post mortem examination be carried out. The common findings at post mortem examinations of such infants are reviewed. A feature of poisoning by arsine and stibine, gases implicated in the toxic gas hypothesis, is haemolysis. Significant haemolysis is not seen in post mortem examinations of SIDS.

We reviewed the developmental physiology of young infants with particular reference to changes with age in the pattern of sleep, temperature control and breathing, and the ways these changes might relate to the hypothesis. Complex changes in the relationships between these physiological control systems occur in the first few months after birth. The characteristic age distribution of SIDS deaths coincides with marked changes in several major physiological control systems (e.g. sleep, temperature, breathing) , and the ways in which these changes might be related to sudden unexpected deaths are considered. A number of features of infant development are put forward by Richardson in support of the hypothesis (see Appendix I). The published information on developmental physiology however, does not provide evidence to support his claims, particularly in relation to heat production, thermal environment, and the length of time babies are left unattended. Nor do the features of those infant deaths which have occurred during the course of detailed physiological recordings correspond with the sequence of events which would be expected to occur during acute anticholinesterase poisoning, which Richardson suggests is the mechanism by which death occurs in such infants.

Whilst some features of SIDS are compatible with the toxic gas hypothesis, a detailed review of infant developmental physiology, the pathophysiology of SIDS, and possible mechanisms for such deaths finds many features which cannot easily be accommodated within this hypothesis.

Back to Report Contents page

Chapter 9. Antimony and infant health

The detection of antimony in infants has been used as key evidence in support of the toxic gas hypothesis. In particular, the concentration of antimony in liver and in blood has been reported to be higher than ‘normal’ in SIDS infants and to correlate with the antimony content of their cot mattresses (Cook Report, November 1994). Subsequently, hair antimony concentrations in apparently healthy, live infants were said to be higher in infants than their mothers and to correlate with the antimony content in the infant’s cot mattress (Cook Report, December 1994). It was argued that the antimony found in infants indicated exposure to stibine (i.e. antimony trihydride) generated by fungal activity from antimony trioxide, used as a fire retardant, in PVC cot mattress covers.

Research studies were initiated by the Expert Group and by other organisations to examine these observations. These investigated the range of antimony concentrations in infants and compared it with that in SIDS infants. They also examined the relationship between antimony concentrations in infants and in their cot mattresses.

This work, undertaken by several independent groups, has shown that low amounts of antimony can be detected in samples from the majority of infants and not only in SIDS. The antimony concentration varies in individual infants and in different types of sample, such as liver and lung tissues, serum, urine and hair, a pattern commonly seen in adults with other non-essential metals.

Several studies have determined antimony concentrations in liver and lung tissue taken at post mortem examination from SIDS infants and from infants who have died of known causes. These have not found the concentrations of tissue antimony to be different in SIDS infants. Furthermore, the lung and liver antimony concentrations in SIDS and in the control infants are similar to those found in adults not occupationally exposed and in human fetal and placental tissues indicating that there is no exceptional accumulation in SIDS infants. The demonstration of antimony in tissue and body fluids taken before, at, and soon after birth indicates fetal exposure. Hence there is no evidence to support the suggestion that antimony in infant tissues is exclusively derived from post natal exposure.

As these measurements of antimony do not show any consistent difference between SIDS and non-SIDS controls, nor higher concentrations in the lungs of SIDS infants, we conclude that there is no evidence for inhalation or absorption of toxic gases containing antimony.

Other studies examined the relationship between liver, lung and hair antimony concentration in an infant, and both the infant’s age and the antimony content of his/her PVC cot mattress cover. Tissue and hair antimony concentrations were not higher in older infants nor did they correlate with the amount of antimony trioxide in the cot mattresses on which the infants had slept. Hence we found no evidence that either tissue or hair antimony is elevated in infants who had slept on cot mattresses for longer periods or had slept on PVC cot mattresses with a higher content of antimony. Furthermore, the detection of antimony at similar concentrations in infant hair and in fetal liver before antimony fire retardants were introduced into cot mattress PVC does not support the suggestion that its presence in infants is related to this source. A comparison between the antimony content in the hair of an infant and its mother has confirmed that higher concentrations are present in infant hair. While the reason for this has not been established there are compelling explanations other than exposure to toxic gas. Hair is known to accumulate metals. The higher amounts in infants may be due to placento-fetal transfer, most probably from maternal diet, and additionally to post natal exposure from the diet, and from physical contact with, for example, fabrics, car-seat , pushchairs, or household dust known to contain significant amounts of antimony. The detection of antimony in hair in the majority of live, apparently healthy infants examined suggests that the amounts found are not harmful to them.

Back to Report Contents page

Chapter 10 - Use of fire retardant (Fr) chemicals in cot mattress materials and other items

Richardson claims that increases in SIDS can be related to increased use of antimony and phosphorus compounds as fire retardants (FR) and plasticisers in PVC cot mattresses. Furthermore, elevated concentrations of antimony found in infants' hair compared to their mothers' was given on the Cook Report as evidence of infants' exposure to stibine generated by microbial action on the antimony in the cot mattress. A comparison of the data on the introduction of these fire retardants in domestic cot mattresses with changes in the SID rates over time does not support these claims. Antimony trioxide-containing fire retardants were not introduced in domestic cot mattresses until about 1988 and were used in many until 1994. This was a period during which the fall in the SID rate was the most rapid. With regard to the second claim, antimony is ubiquitous in the domestic environment, it is widely used in consumer products and is found at relatively high concentrations in household dust. Experimental work has shown that antimony in childcare articles can be physically transferred to human hair. This could be a source of the antimony found in infant hair.

Back to Report Contents page

Chapter 11 - Epidemiology of Sudden Infant Death Syndrome in relation to Mr Richardson’s hypothesis

In this chapter we address the question of whether the known epidemiological data relating to the sudden infant death syndrome, e.g. trends in incidence rates, geographical variations, and the social and other characteristics of the affected families, can be explained in terms of the hypothesis that the major cause is poisoning by toxic gases resulting from the action of fungi on antimony and phosphorus used as fire retardants in cot mattresses. If the epidemiological data cannot be so explained this casts doubt on the hypothesis. If, as Mr Richardson suggests, they can, it is necessary to consider whether other explanations are available and which explanation is more likely.

It is fairly widely accepted that the decrease in the cot death rate in the 1990s was due to the ‘Reduce the Risk of Cot Death’ / ‘Back to Sleep’ campaign in Autumn 1991 advising a change from prone to supine sleeping; the preceding rise may have been due to adoption of prone sleeping in the 1960s. A similar pattern of rise and fall has been observed in Denmark, Norway, Sweden and The Netherlands.

Richardson suggests that increases in the cot death rate before 1988 and the subsequent decrease can be explained by his hypothesis. However he makes a number of unsubstantiated statements about changes in SIDS rates and about changes in the use of fire retardant in cot mattresses. He stresses particularly the fact that the decrease in rate associated with supine sleeping pre-dated the ‘Reduce the Risk of Cot Death’ / ‘Back to Sleep’ campaign, and suggests that this decrease was a consequence of advice and publicity concerning the toxic gas hypothesis. There may indeed have been an increase in the cot death rate in the 1960s, but the decrease in 1989, before the introduction of the ‘Reduce the Risk of Cot Death’ / ‘Back to Sleep’ campaign, may have been due to the fact that advice to parents was already changing in at least some areas, since reports concerning the results of research into the effects of supine sleeping pre-dated the campaign.

Dr Sprott lays particular emphasis on the fact that SIDS rates are higher in later born children within a family, and claims that the only explanation for this is that they are sleeping on previously used mattresses. This is incorrect; the findings can equally well be explained in terms of epidemiological data relating to infectious diseases where similar patterns of risk in relation to birth order are found.

Richardson misinterprets information related to SIDS rates in service families. First, we have established that his assumptions concerning the constituents of mattresses supplied to service families are incorrect. Secondly, his interpretation of the data depends on analyses that fail to take into account possibly relevant aspects of the differences between service and civilian families.

Results from studies that directly tested Richardson’s hypothesis concerning the age and previous use of mattresses used by SIDS infants and healthy controls can be readily explained in terms of other theories concerning the aetiology of cot deaths. The CESDI/ SUDI studies (see Appendix III) show that ‘mattress wrapping’ was infrequent, and do not support the suggestion that it reduced the risk of SIDS.

We conclude that, although some features of the epidemiology of SIDS can be explained by Richardson’s hypothesis, the evidence is less convincing than he suggests or other explanations are available. Some predictions from the hypothesis are not supported by the data.

Back to Report Contents page

Published by the Department of Health
(c) Crown Copyright 1998
This site last updated 21 May 1998

Help ensure that the Global Internet services
of the SIDS Network continue to grow!

Donate directly to the SIDS Network securely with a major credit card.

Other ways to help can be found here.

new.gif (112 bytes) Now you can translate SIDS Network Web Site pages to/from English, Spanish, French, German, Italian & Portuguese

1995-2015, SIDS Network, Inc. <>
All rights reserved. Permission to use, copy, and distribute this document, in whole or in part, for non-commercial use and without fee,
is hereby granted, provided that this copyright, permission notice, and appropriate credit to the SIDS Network, Inc. be included in all copies.

The opinions and information provided here are not necessarily those of the author and are presented for educational purposes only.
The author accepts no responsibility for content, accuracy or use.

Privacy Policy

Please report any web site problems to sidsnet1-at-sids-network-dot-org
Web Design and maintenance by
CAM Consulting