indicate that attitudes among the medical profession and patients are changing.There has been much discussion about clinical trials and informed consent,46 and, though we believe that informed consent is mandatory, we accept that a proportion of patients do not enter trials because they may dislike the idea of a random decision being made about treatment or because they refuse or request one part of the random option offered. A further group of patients are excluded because they do not fulfil the entry criteria for the trial, and this group may be larger than forecast at the time the trial was planned. These factors may be the reason why accrual to the Scottish breast conservation trial has been slower than anticipated. The planned total intake was 900 patients, and after four years 420 patients had been entered.In conclusion more than half of the patients thought initially to be suitable for conservation were excluded from our trials and one third of the remainder refused to take part. Those planning prospective clinical trials should therefore take into account the loss of patients through ineligibility and refusal when predicting the accrual rate and overall duration of any proposed trial and the possible effects of this selection on conclusions drawn from the results. Design-A study of all infants dying suddenly and unexpectedly and of two controls matched for age and date with each index case. The parents of control infants were interviewed within 72 hours of the index infant's death. Information was collected on bedding, sleeping position, heating, and recent signs of illness for index and control infants.Setting-A defined geographical area comprising most of the county of Avon and part of Somerset.Subjects-72 Infants who had died suddenly and unexpectedly (of whom 67 had died from the sudden infant death syndrome) and 144 control infants.Results-Compared with the control infants the infants who had died from the sudden infant death syndrome were more likely to have been sleeping prone (relative risk 8-8; 95% confidence interval 7.0 to 11-0; p<0-001), to have been more heavily wrapped (relative risk 1-14 per tog above 8 tog; 1-03 to 1-28; p<005), and to have had the heating on all night (relative risk 2-7; 1-4 to 5.2; p<0-01). These differences were less pronounced in the younger infants (less than 70 days) than the older ones. The risk of sudden unexpected death among infants older than 70 days, nursed prone, and with clothing and bedding of total thermal resistance greater than 10 tog was increased by factors of 15-1 (2.6 to 89.6) and 25-2 (3.7 to 169-0) respectively compared with the risk in

No abstract

for information about the thermal insulation of clothing 14 Nelson EAS, Taylor BJ, Weatherall IL. Sleeping position and infant bedding may predispose to hyperthermia and the sudden inf'ant death syndrome.

Bottle feeding is not a significant independent risk factor for the sudden infant death syndrome. Patterns of maternal smoking, preterm gestation, and parental employment status account for most of the apparent association with bottle feeding.

Objective-To determine whether signs of illness reported by parents can be used to identify babies at risk from the sudden infant death syndrome.Design-A two year prospective case-controlled study based in a geographically defined area.Setting-Four health districts in Avon and north Somerset.Subjects-Babies who had died suddenly and unexpectedly aged between 1 week and 2 years (index babies) and two control babies for each index baby selected from the same health visitor's list and matched for age, time of year of the interview, and area of residence.Main outcome measures-Major and minor signs of illness during two weeks before the index babies' death, or before the interview for control babies, and consultations with the general practitioner during the same period.Results Conclusion-Major and minor signs of illness are neither a sensitive nor a specific indicator of sudden