The 1972 Cuban national child growth study as an example of population health monitoring: design and methods.

A stratified 3-stage random sample of 50 360 children ages 0-19 drawn from the whole population of Cuba was measured in a large-scale growth study during 1972-73. Fifteen anthropometric measurements were taken and puberty stages and menarche status were assessed. Hand-wrist radiographs were done on 10% of the sample. Information regarding the social and education status of the parents was obtained and parental heights were measured. One year later, 30% of the sample were re-measured and 10% re-X-rayed. The overall response rate was 96% at pre-school and primary school age, and a lettle less later. Quality control sessions were held at which the nine measuring teams compared results. No significant differences were found between teams working different parts of the island. Differences between duplicate measurements of stature by individual measurers had standard deviations approximating 0.20 cm. Individual measurers' means differed from the grand means of all measurers by up to +/- 0.2 cm for stature, +/- 0.4 cm for sitting height and +/- 0.5 mm (7% of mean) for triceps skinfold. One or more measurements lay outside the 3rd-97th centile limits in 21% of individuals. Scrutiny of these individuals' records resulted in elimination of measurements in amounts ranging from 0.1% (triceps skinfold) to 1.1% (crown-rump length). Problems of planning and execution of growth surveys designed to set national standards are described, and solutions given or suggested.