Search

Since the introduction of new growth charts in the mid 1990s, there has been confusion about which charts should be used, with many districts using more than one version. Because of this uncertainty, an expert working party, the Growth Reference Review Group, was convened by the Royal College of Paediatrics and Child Health to provide guidance on the validity and comparability of the different charts currently in use. This paper describes the technical background to the construction and evaluation of growth charts and outlines the group's findings on the validity of each growth reference in relation to contemporary British children. The group concluded that for most clinical purposes the UK90 reference is superior and for many measures is the only usable reference that can be recommended, while the original Tanner-Whitehouse and the Gairdner-Pearson charts are no longer reliable for use at any age. After the age of 2 the revised Buckler-Tanner references are still suitable for assessing height. There are presently no reliable head circumference reference charts for use beyond infancy. The group propose that apart from refinements of chart design and layout, the new UK90 reference should now be "frozen", with any future revisions only undertaken after careful planning and widespread consultation.

If a court fails to convict a defendant because of incomplete evidence, does that establish his innocence beyond doubt? Not necessarily. Indeed in Scotland, if sufficient uncertainty remains, the court can give a verdict of “not proven” instead of “not guilty”. If a randomised controlled trial (RCT) fails to show a significant difference between the treatment and the control group, does that prove that the treatment has no useful clinical effect? Again, not necessarily. The treatment may work, but the trial may have been unable to prove it.1 Despite this, many such “negative” trials,1 2 including many published in this journal, may wrongly be taken as evidence that the treatment is not clinically useful. For example, in an RCT of women at risk of preterm delivery that was not published as a full report,3 respiratory distress syndrome (RDS) occurred in three of 23 babies born to the treated group and three of 22 babies born to the untreated group. The difference is not significant (2p > 0.9). Had this been the first and only study of this treatment, many people might have decided that it was not effective and thus lost interest. In fact, overviews of this,3 and at least 14 other trials, eventually showed that the treatment—antenatal steroids—is highly effective because it reduced RDS and neonatal mortality in over 3500 preterm infants by about half.4 5 Note that the results of the single trial were quite consistent with this finding.3 The correct conclusion from that single trial …

SUMMARY This paper shows that the notion of minimizing the disagreement between parallel assessments of the same quantity unifies a number of current methods of scaling and leads to straightforward extensions to cover new problems. Applications are found in the assessment of physical maturity, the measurement of children's behaviour, and in longitudinal studies with sets of items at several occasions.

RESP-4818

Scientific computing in Great Britain has changed quite markedly since the Flowers' report appeared in December 1965, and one aspect of this change is the increased acceptance of FORTRAN. It is now taken for granted that any computer for scientific use will possess a FORTRAN compiler, and that this will be compatible with one or other of the ASA specifications (American Standards Association, 1964, 1965). As well as this, most FORTRAN IV compilers now offer a range of features over and above the minimum needed for ASA compatibility. If these extra features are examined a certain measure of agreement between the different compilers is apparent, and this leads to the idea that the time is ripe for a third level of 'standard FORTRAN' standing to FORTRAN IV much as this does to FORTRAN II or ASA Basic FORTRAN. This note aims at setting out some ideas for such a language. It is in many ways remarkable that a language as old as FORTRAN is still in use at all, and any 'extended FORTRAN' must be able to stand the competition of newer languages such as ALGOL and PL/I. One reason for survival seems to be the fact that its very limitations, while not too serious from the user's point of view, make it relatively easy to compile. Fast compiling is important for a language much used in teaching since students' exercises spend most of their time in the compiling phase, and indeed the same consideration is important in any research-based computer centre where the ratio of new programs to old will always remain high. It follows from this that an extended FORTRAN, while aiming to remove some of the limitations of FORTRAN IV, should remain simple in structure, and that it may be necessary to exclude features which, while occasionally useful, would complicate the compiler structure to an unacceptable extent.

1. The food intake of sixty-four infantry recruits was measured at six centres during 3 weeks of initial training. The daily energy expenditure was measured in thirty-five of these men.2. The mean daily consumption of the sixty-four subjects provided them with 3850 kcal (16110 kJ); the energy expenditure of the thirty-five subjects averaged 3750 kcal (15690 kJ).3. Serial auto- and cross-correlations of intake and expenditure were very small and there was no significant relationship between food intake and energy expenditure on the same day.4. The intakes and expenditures of different subjects at the same centre were not independent.5. There was a significant relationship between intake and expenditure for the whole period of the survey when results for all subjects were included. In three centres the correlation was high, +0.788 (P < 0.001), but was only + 0.083 (P < 0.5) in the remaining three centres.6. There was a positive but not significant correlation between body-weight and the average food intake of 6 d.7. There was a negative correlation between body-weight and calorie balance.8. Weight change and calorie balance over 1 week were related, the correlation averaging 0.40. There was a correlation of 0.32 between daily weight changes and calorie balance. A rather small amount of variation in calorie balance can be explained by contemporary changes in weight.

RESP-3804

When planning a factorial experiment, it is often desirable (as pointed out by Yates (1937)) to include certain extra treatments falling outside the usual factorial scheme. I discuss here the exact analysis of an experiment of this type. Only standard principles of experimental analysis are involved, and the full exact analysis is not always required in practice, but the example may illustrate techniques that are useful in investigating more complicated designs that are sometimes encountered. I am grateful to the late Dr E. M. Crowther and Miss B. Benzian, of the Chemistry Department at Rothamsted, for discussion of the problem and for providing data for the illustrative example.

In certain applications of multiple regression the coefficient matrix of the normal equations is singular and therefore does not possess an inverse. Recent theoretical developments make it possible to broaden the theory of least squares so as to include this situation without having to treat it as a special case, and these can be paralleled by appropriate numerical techniques. This paper gives a survey of these developments and outlines some of their possible applications.

RESP-3610A

Experiments using animal litters as experimental units usually require a weighted analysis to allow for variation in litter size. This paper describes methods of assessing appropriate weights for fully randomized and for randomized block experiments.

RESP-4727