A systematic review and meta-analysis to revise the Fenton growth chart for preterm infants

The expected growth of the fetus describes the fastest human growth, increasing weight over six-fold between 22 and 40 weeks. Preterm infants, who are born during this rapid growth phase, rely on health professionals to assess their growth and provide appropriate nutrition and medical care.

In 2006, the World Health Organization (WHO) published their multicentre growth reference study, which is considered superior [1] to previous growth surveys since the measured infants were selected from communities in which economics were not likely to limit growth, among culturally diverse non-smoking mothers who planned to breastfeed [2]. Weekly longitudinal measures of the infants were made by trained data collection teams during the first 2 years of this study [3]. These WHO growth charts, although recommended for preterm infants after term age [4], begin at term and so do not inform preterm infant growth assessments younger than this age.

Optimum growth of preterm infants is considered to be equivalent to intrauterine rates [5‐7] since a superior growth standard has not been defined. Perhaps the best estimate of fetal growth may be obtained from large population-based studies, conducted in developed countries [8], where constraints on fetal growth may be less frequent.

A recent multicentre study by our group (the Preterm Multicentre Growth (PreM Growth) Study) revealed that although the pattern of preterm infant growth was generally consistent with intrauterine growth, the biggest deviation in weight gain velocity between the preterm infants and the fetus and infant was just before term, between 37 and 40 weeks (Fenton TR, Nasser R, Eliasziw M, Kim JH, Bilan D, Sauve R: Validating the weight gain of preterm infants between the reference growth curve of the fetus and the term infant, The Preterm Infant Multicentre Growth Study. Submitted BMC Ped 2012). Rather than demonstrating the slowing growth velocity of the term infant during the weeks just before term, the preterm infants had superior, close to linear, growth at this age. This finding has been observed by others as well [9‐11]. Therefore, there is evidence to support a smooth transition on growth charts between late fetal and early infant ages.

Several previous growth charts based on size at birth presented their data as completed age, which affects the interpretation and use of a growth chart [12]. The use of completed weeks when plotting a growth chart requires all the measurements to be plotted on the whole week vertical axes. However, the use of completed weeks in a neonatal unit may not be intuitive, as nursery staff and parents think of infants as their exact age, and not age truncated to previous whole weeks. The advent of computers in health care, for clinical care and health recording, allow the use of the computer to plot growth charts, daily and with accuracy. It would make sense to support plotting daily measurements continuously by shifting the data collected as completed weeks to the midpoint of the next week to remove the truncation of the data collection as completed weeks.

The objectives of this study were to revise the 2003 Fenton Preterm Growth Chart, specifically to: a) use more recent data on size at birth based on an inclusion criteria, b) harmonize the preterm growth chart with the new WHO Growth Standard, c) to smooth the data between the preterm and WHO estimates while maintaining integrity with the data from 22 to 36 and at 50 weeks, d) to derive sex specific growth curves, and to e) re-scale the chart x-axis to actual age rather than completed weeks, to support growth monitoring.

To revise the growth chart, thorough literature searches were performed to find published and unpublished population-based preterm size at birth (weight, length, and/or head circumference) references. The inclusion criteria, defined a priori, designed to minimize bias by restriction [13], were to locate population-based studies of preterm fetal growth, from developed countries with:

Six large population based surveys [14, 16‐20] of size at preterm birth from countries Germany, United States, Italy, Australia, Scotland, and Canada were located that met the inclusion criteria (Table 1). The literature search identified 2436 papers, of which 2373 were discarded as being not relevant or duplicates based on the titles (Figure 4). Reviewing reference lists identified another 12 studies. Seventy-five studies were examined in detail, however 27 of these did not meet the date criteria. Among the 48 studies that met the date of birth criteria, some did not meet the other inclusion criteria for the following reasons: Did not meet the criterion for more than 25,000 babies [21‐35], no low gestational age infants less than 25 weeks [31, 36‐41], insufficient number less than 30 weeks [34, 42‐45], no statistical correction for inaccurate gestational ages [46‐48], numerical data not available [49‐51], number of infants each week were not available [52], number of infants in the subgroups each week were not available [53], was not population based [54‐56], no direct measurements [27], some of the data [57] was also in one of the larger included studies [17].

Included in the meta-analyses were almost four million (3,986,456) infants at birth (34,639 less than 30 weeks) from six studies for weight (Table 2), and 173,612 infants for head circumference, and 151,527 for length [16, 18]. The World Health Organization data measurements were made longitudinally on 882 infants.

The individual datasets from the literature showed good agreement with each other, especially along the 50th and lower centiles (Figures 1, 2, and 3) and the meta-analysis curves had a close fit with the individual datasets up to 36 weeks and at 50 weeks (Figures 5, 6, 7). The final splined weight curves were within 3% of the meta-analysis curves for 24 through 36 weeks for both genders, except for a 3.8% difference for girls at 32 weeks along the 90th centile. None of the length measurements differed by more than 1.8% percent between the meta-analysis and the splined curves; all weeks of the head circumference curves were within 1.5%. The meta-analyses for head circumference and length for girls and boys were close enough to normal distributions that normal distributions were used to summarize the data. The measures at 50 weeks were within 0.5% of the WHOGS values.

Girl and boy charts were prepared (Figure 8 and 9), by shifting the age by 0.5 weeks to allow plotting by exact age instead of completed weeks. The LMS Parameters [15] were used to develop the exact z-score and percentile calculators for the new growth chart.

In the two graphical comparisons between the revised growth charts, one for each sex, with the 2003 Fenton preterm growth chart revealed that the curves were quite similar (Figures 10 and 11). Generally the new girls’ curves were slightly lower (Figure 10) and the new boys’ slightly higher (Figure 11) for all 3 parameters (weight, head circumference, and length) than the 2003 curves. The most dramatic visual and numerical difference between the new charts and the 2003 chart was the higher shift of the boys’ weight curves after 40 weeks compared to the 2003 chart, reaching a maximum difference at 50 weeks of 650, 580, and 740 grams at the 3rd, 50th, and 97th percentiles, respectively. The second biggest visual difference was the lower pattern of the girls’ length curves below 37 weeks; the difference in length reached a maximum numerical value of 1.7 centimeters at 24 weeks along the 97th percentile.

We used a strict set of inclusion criteria to include only the best data available to convert fetal and infant size data into fetal-infant growth charts for preterm infants. The revised sex-specific actual-age (versus completed weeks) growth charts (Figure 9 and 10), are based on birth size information of almost four million births with confirmed or corrected gestational ages, born in developed countries (See Features of the new growth chart). The revised charts are based on the recommended growth goal for preterm infants, the fetus and the term infant, with smoothing of the disjunction between these datasets, based on the findings of our international multicentre validation study (Fenton TR, Nasser R, Eliasziw M, Kim JH, Bilan D, Sauve R: Validating the weight gain of preterm infants between the reference growth curve of the fetus and the term infant, The Preterm Infant Multicentre Growth Study. Submitted BMC Ped 2012). These charts are consistent with the meta-analysis data up to and including 36 weeks, thus they can be used for the assessment of size for gestational age for preterm infants under 37 weeks of gestational age. This growth chart is likely applicable to preterm infants in both developed and developing countries since the data was selected from developed countries to minimize the influence from circumstances that may not have been ideal to support growth.

The inclusion of data from a number of developed countries increases the generalizability of the growth chart. The revised preterm growth chart, harmonized with the World Health Organization Growth Standard at 50 weeks, may support an improved transition of preterm infant growth monitoring to the WHO charts.