On 2014 Jan 09, David Mage commented:

Randall et al. studied serotonin receptor binding in a non-probability (convenience) sample of case infants dying suddenly and unexpectedly without explanation (SIDS) and control infants dying of known causes, from San Diego County, CA. As pointed out previously to some of these same authors Mage DT, 2010 their results do not apply to any populations not probability sampled for this study and therefore statistical testing for significant differences between their cases and controls has no theoretical statistical basis and may lead to meaningless results. The authors' reply (op. cit.) leads one to believe that they think they are excused because they know of no "reason to believe that the levels of serotonin in brainstems of SIDS and non-SIDS in San Diego would be different than in infants elsewhere." Here are a few such reasons that may be relevant: 1) All cases and controls were autopsied at the San Diego County Medical Examiner's Office. It is well known that different pathologists independently looking at the same cases without any discovered obvious prima facie cause of death can give different causes of death by SIDS and non-SIDS such as positional asphyxia Emery JL, 1972 , Kim SY, 2012. Consequently their case definitions of SIDS may not apply elsewhere where different operative decisions for choice of cause of death are made by different medical examiners; 2) Serotonin does not cross the blood brain barrier, so all cerebral serotonin must come from its precursor dietary tryptophan (TRY). Because the infants' TRY in utero comes from their mothers' diet during pregnancy and after birth from their own diet it may make a big difference if they are breast fed or given milk formula as breast milk has higher TRY content than ordinary formula, that can be enriched in TRY (Young SN, 2007). The authors noted the higher percentage of Hispanic ethnicity and lower percentage of African Americans compared to the U.S. infant population. Consequently the frequency of breast feeding and the dietary carbohydrate intake of the Hispanic nursing mothers during and after pregnancy can be quite different in San Diego County compared to the rest of the U.S.; 3) Hispanic ethnicity mothers in San Diego may wean their infants at different ages than other non-Hispanic mothers and the infants' diets after weaning may also differ between Hispanic and non-Hispanic ethnicities; 4) Because the cause of SIDS is unknown, there may be yet unknown SIDS risk factors that vary between their cases and controls and all other cases and controls not sampled. In summary the authors' conclusions strictly apply to only those cases and controls they measured. The statistical testing and confidence intervals they report do not apply to all the other cases and controls that were not in the sample frame from which their cases and controls were chosen because they had no known finite probability of selection for this study.

On 2014 Jan 09, David Mage commented:

Randall et al. studied serotonin receptor binding in a non-probability (convenience) sample of case infants dying suddenly and unexpectedly without explanation (SIDS) and control infants dying of known causes, from San Diego County, CA. As pointed out previously to some of these same authors Mage DT, 2010 their results do not apply to any populations not probability sampled for this study and therefore statistical testing for significant differences between their cases and controls has no theoretical statistical basis and may lead to meaningless results. The authors' reply (op. cit.) leads one to believe that they think they are excused because they know of no "reason to believe that the levels of serotonin in brainstems of SIDS and non-SIDS in San Diego would be different than in infants elsewhere." Here are a few such reasons that may be relevant: 1) All cases and controls were autopsied at the San Diego County Medical Examiner's Office. It is well known that different pathologists independently looking at the same cases without any discovered obvious prima facie cause of death can give different causes of death by SIDS and non-SIDS such as positional asphyxia Emery JL, 1972 , Kim SY, 2012. Consequently their case definitions of SIDS may not apply elsewhere where different operative decisions for choice of cause of death are made by different medical examiners; 2) Serotonin does not cross the blood brain barrier, so all cerebral serotonin must come from its precursor dietary tryptophan (TRY). Because the infants' TRY in utero comes from their mothers' diet during pregnancy and after birth from their own diet it may make a big difference if they are breast fed or given milk formula as breast milk has higher TRY content than ordinary formula, that can be enriched in TRY (Young SN, 2007). The authors noted the higher percentage of Hispanic ethnicity and lower percentage of African Americans compared to the U.S. infant population. Consequently the frequency of breast feeding and the dietary carbohydrate intake of the Hispanic nursing mothers during and after pregnancy can be quite different in San Diego County compared to the rest of the U.S.; 3) Hispanic ethnicity mothers in San Diego may wean their infants at different ages than other non-Hispanic mothers and the infants' diets after weaning may also differ between Hispanic and non-Hispanic ethnicities; 4) Because the cause of SIDS is unknown, there may be yet unknown SIDS risk factors that vary between their cases and controls and all other cases and controls not sampled. In summary the authors' conclusions strictly apply to only those cases and controls they measured. The statistical testing and confidence intervals they report do not apply to all the other cases and controls that were not in the sample frame from which their cases and controls were chosen because they had no known finite probability of selection for this study.