Illingworth 1953.

Methods	Quasi‐randomised design.
Participants	210 mothers of full term infants, all babies were included.
Interventions	Forced fluids (104 women were advised to drink "at least six pints of fluid a day"). Control (106 women were not given any advice about fluid intake).
Outcomes	Breast milk production (measured by test feeds on day 8 combined with the volume of milk expressed after every feed throughout the day), gain in weight, duration of exclusive breastfeeding.
Notes	There were 2 floors and 2 wards on each floor. On each floor, in 1 ward the women were advised to drink at least 6 pints of fluids a day and in the other ward the women were not given any advice about fluid intake. After 50 sets of data were obtained in each ward the wards swapped conditions (i.e. extra fluids switched to no advice and vice versa) until another 50 data sets were obtained (involving different women). "The birth weights of the babies studied were shown to be comparable," "Parity of the mothers, the incidence of breastfeeding and the mount of fluid taken were comparable in the two groups," Infants were divided into 3 groups according to birthweight: group A: birthweight less than 6 lb 13 oz; group B: birthweight from 6 lb 13¼ oz to 7lb 13 oz; group C: birthweight greater than 7 lb 13 oz.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Wards were used for allocation, the women were not randomised.
Allocation concealment (selection bias)	High risk	Women knew which group they belonged to so there was no allocation concealment.
Blinding of participants and personnel (performance bias) All outcomes	High risk	There was no blinding as the personnel and women in intervention group knew they were to take additional fluids.
Blinding of outcome assessment (detection bias) All outcomes	High risk	There was no blinding.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The study design involved data collection until a set number of participants had been included. It is not clear from the report if any were missed.
Selective reporting (reporting bias)	Unclear risk	The study design involved data collection until a set number of participants had been included. Although unlikely, it is not clear from the report if any additional data items were collected and not reported, or if any results from participants were omitted from the analysis.
Other bias	High risk	The method of data collection is laden with bias.