A. Data selection and combination

Major bibliographic databases were searched: MEDLINE (using PubMed) and CINHAL, by both authors back to year 1987 (given our 25 year limit), with no language restrictions, and foreign articles were translated. The following search terms as medical subject headings and textwords were used: (“Preterm infant” OR “Premature Birth”[Mesh]) OR (“Infant, Premature/classification”[Mesh] OR “Infant, Premature/growth and development”[Mesh] OR “Infant, Premature/statistics and numerical data”[Mesh] OR “Infant, very low birth weight”[Mesh]) AND (percentile OR *centile* OR weeks) AND (weight OR head circumference OR length). Grey literature sites including clinical trial websites and Google were searched in February 2012. Reference lists were reviewed for relevant studies.

All of the found data was reported as completed weeks except for the German Perinatal Statistics, which were reported as actual daily weights [14]. To combine the datasets, the German data was temporarily converted to completed weeks. A final step converted the meta-analyses to actual age.

B. Combine the data to produce weighted intrauterine growth curves for each sex

The located data (3rd, 10th, 50th, 90th, and 97th percentiles for weight, head circumference, and length) that met the inclusion criteria were extracted by copying and pasting into spreadsheets. The male and female percentile curves from each included data set for weight, head circumference and length were plotted together so they could be examined visually for heterogeneity (Figures 1, 2, and 3). The data for each gender were combined by using the weekly data for the percentiles: 3rd, 10th, 50th, 90th, and 97th, weighted by the sample sizes. The combined data was represented by relatively smooth curves.

C. Develop growth monitoring curves

D. Compared the revised charts with the 2003 version

The revised growth charts were compared graphically with the original 2003 Fenton preterm growth chart. To make the differences in chart values more apparent, the 2003 chart data was also shifted to actual weeks for these comparison figures.

Features of the new growth chart

It may be more intuitive to plot on growth charts using exact ages rather than on the basis of complete weeks. Several years ago, the WHO used completed age for growth chart development [12]. This recommendation was likely due to the way data had been collected in the past, that is all 26 0/7 through 26 6/7 week infants were included in the 26 week completed week category. However, with the use of computers to plot on growth charts comes the potential to more accurately plot measurements to the exact day of data collection. Thus the time scale of the horizontal axes of these new growth charts were re-scaled to actual age, for ease of use and understanding. For example, a baby at 25 3/7 can be intuitively plotted between 25 and 26 weeks.

Exact z-score and centile calculators for the revised charts are available for download: http://ucalgary.ca/fenton. Data is available for research upon request.

The data revealed that between 22 weeks to 50 weeks post menstrual age, the fetus/infant multiplies its weight tenfold, for example, the girls’ median weight increased from a median of 520 to 5360 grams. Using a fetal-infant growth chart allows clinicians to compare preterm infants’ growth to an estimated reference of the fetus and the term infant.

There was a remarkably close fit of the included preterm surveys for weight, head circumference and length from the 6 countries, especially at the 50th percentile, even though the data came from different countries.

The splining procedures we used have produced a chart that has integrity and good agreement with the original data. Smoothing of the LMS parameters is recommended since minor fluctuations are more likely due to sampling errors rather than physiological events [15]. Experts recommend that growth charts be developed based on smoothed L, M and S, to constrain the adjacent curves so that they relate to each other smoothly [15]. The World Health Organization set their L parameter to 1 for head circumference and length, while they maintained the exact L values for infants’ weights [58]. The data under study here revealed the same effect as the WHO data; we found that both head circumference and length were close enough to normal distributions that normal distributions could summarize the data, while the exact L’s were needed to retain the nuances of the weight curves.

The differences between the revised growth charts and the 2003 Fenton preterm growth chart may reflect improvements since the selected preterm growth references for the new versions are more likely globally representative of fetal and infant growth. Some of the differences between the current charts and the 2003 version are likely due to the separation into girl and boy charts, since the shifts of the girls’ curves tend to be downward and the boys’ curves upward. The weight shifts after 40 weeks were upward for both sexes, due to the higher values for the WHOGS compared to the CDC growth reference [59] at 10 weeks post term.

The ideal growth pattern of preterm infants remains undefined. These revised growth charts were developed based on the growth patterns of the fetus (as has been determined by size at birth in the large population studies) and the term infant (based on the WHO Growth Standard) [2]. Ultrasound studies and comparison of subgroups of prematurely born infants suggest that the fetal studies, such as those used in this development, may be biased by the premature birth since fetuses who remain in utero likely differ in important ways from babies who are born early [60, 61]. However, fetal size from these imperfect studies may be the best data available at this point in time for comparing the growth of preterm infants since the alternative, to compare to in utero infants requires extrapolation from ultrasound measurements. To use other premature infants as the growth reference for preterm infants may not be ideal since the ideal growth of preterm infants has not been defined, has been changing over time [62], and is influenced by the nutrition and medical care received after birth [63, 64].

Although the WHOGS is considered to be a growth standard, the infants in the population-based surveys of size at birth are more likely representative of the reference populations and were not selected to be healthy. Thus these growth charts are growth references and are not a growth standard. The INTERGROWTH study, currently underway, will rectify this problem, since their purpose is to develop prescriptive standards for fetal and preterm growth [65].