SUBJECTS AND METHODS

Subjects

Sixty eight preterm (less than 37 completed weeks gestation) infants were recruited from among 110 consecutive admissions of infants with a birth weight less than 2200 g to the Royal Hampshire County Hospital neonatal unit between February 1998 and July 1999.

Thirteen siblings of recruited twins or triplets were not enrolled in the study as they could not, for practical reasons, be randomised to a different feeding regimen from their siblings. The reasons for non-inclusion of the remaining infants were: consent was refused (eight); the baby died (two); consent was withdrawn (one); major congenital malformation (one); cleft palate (one); planning to move away (two); parents not approached as the researcher (KF) was on leave or failed to make contact with the parents (14; these tended to be the heavier more mature babies who had a shorter stay on the neonatal unit, so were more likely to be missed by the recruiter).

Ethical approval

Approval for the study was obtained from the advisory committees on ethics of the University of Surrey and Winchester local research ethics committee. All parents gave fully informed consent in writing.¹⁴

Randomisation

Infants were randomly assigned to either the preterm weaning strategy (PWS) group or the control group. The randomisation was stratified by sex and birth weight above and below 1500 g. Block randomisation within strata were used to achieve reasonable balance within strata. Random numbers were generated using Minitab. Everyone but the statistician was blind to the block sizes. Prior sample size calculations based on a two sample t test showed that 30 infants per treatment group were needed to detect a standardised effect size of 0.75 at 5% significance with 80% power.

Parents of babies in the PWS group were recommended to wean their babies according to the weaning strategy, whereas parents of babies in the control group were recommended to wean their babies according to current best advice.¹¹ Parents and all involved health professionals including the paediatricians following up the babies were blinded to the group allocation. A few parents did ask which group they were in but accepted the explanation that this was confidential until the end of the study. Members of families in friendship groups rarely asked about group allocation; any who did were reminded of the detailed information they were given about the study protocol and the need for random assignment. In addition, as study families were dispersed over a large area, even where families were friendly, opportunities to meet were infrequent. LM became aware of the group allocation after the first growth assessment and blood test at 0 months GCA.

The preterm weaning strategy

The strategy recommended the early onset of weaning, the use of foods with a higher energy and protein content than standard milk formula, and recommended foods that are rich sources of iron and zinc.

The feeding recommendations to the principal carers (parents) included:

• the early introduction to semisolid foods as soon after 13 weeks of postnatal age as an infant was deemed (by the parents) to be ready, provided that the infant weighed at least 3.5 kg;

• the use of solid foods with an energy density, whether home prepared or commercial foods, in the range 300–450 kJ/100 g (70–105 kcal/100 g);

• the use of solid foods with a protein content in the range 2.3–5.0 g/100 g for cereal and savoury foods, and 1.0–4.0 g/100 g for fruit puddings and other desserts;

• the use of a preterm formula milk (Nutriprem; Cow & Gate Ltd, Trowbridge, Wiltshire, UK) provided free of charge to parents for use as the liquefying agent for dried cereal foods and home prepared dishes.

Parents of babies in the PWS group were provided with a detailed set of guidelines, which specified the varieties and combinations of weaning foods appropriate for each stage of weaning.

The control group regimen

Parents of babies in the control group were provided with an alternative set of guidelines based on current recommendations for full term infants.¹¹ The Department of Health acknowledges that preterm weaning advice presents particular difficulties and offers no evidence based research for its suggestion that weaning these infants may begin after they reach 5 kg in weight and are able to eat from a spoon.¹¹

The feeding recommendations to the principal carers (parents) included:

• the introduction to semisolid foods as soon after 17 weeks of postnatal age as an infant was deemed (by the parents) to be ready, provided that the infant weighed at least 5.0 kg;

• the use of solid foods with an energy density, whether home prepared or commercial foods, in the range 250–450 kJ/100 g (60–105 kcal/100 g);

• the use of solid foods with a protein content in the range 1.5–5.0 g/100 g for cereal and savoury foods, and 0.5–4.0 g/100 g for fruit puddings and other desserts.

All recommended shop bought foods complied with the European Union directives governing baby food composition.¹⁵

Data collection

Energy and nutritional intake were assessed at 0, 6, and 12 months GCA by the recording of seven day weighed intake diaries, which is the standard method of dietary assessment.¹⁶ In addition, a food frequency data sheet was completed by parents, in which all types of weaning foods were listed, and parents recorded the date and name of the specific food of that type first fed to their infant. The dietary analyses of milk intake at 0 weeks GCA, and milk, food, and other drink intake at 6 and 12 months GCA were determined from the diaries completed by mothers whose babies received an infant formula milk/bottled expressed breast milk/cow’s milk as their main or only milk food for the duration of the diary.

Neonatal unit policy was to use breast milk fortifier for all breast fed babies admitted who were both preterm and weighed less than 2000 g at birth. It was always stopped before discharge as it was unit policy to ensure that babies were gaining weight on the regimen to be used at home before they left the unit. After discharge, formula fed infants were fed a standard infant formula of the parent’s choosing, except for one child, who was prescribed Nutripem 2 until 3 months of age. All preterm babies of birth weight < 2000 g were prescribed multivitamins (abidec) and iron (sodium iron edetate) from 2 weeks of age (provided that they were tolerating oral feeds) according to the unit policy.

The Diet 5 program was used for dietary analysis.

Blood samples (3 ml) were taken by venepuncture at 0 and 6 months GCA. Haemoglobin concentration and red cell indices (packed cell volume, mean corpuscular haemoglobin concentration) were assayed using Cell Dyn 3000. Serum ferritin was analysed by Microparticle Enzyme Immunoassay. Serum iron was measured as a coloured complex with ferrozine, and transferrin was measured by the antibody-antigen complex method.

Zinc, selenium, and manganese were assayed by flame atomic absorption spectrometry,¹⁷ inductively coupled mass spectrometry,¹⁸ and atomic absorption spectrometry¹⁹ respectively.

Anthropometry was performed on each infant by LM within two weeks of 0, 6, and 12 months GCA. Infants were weighed naked using a Seca 834 digital scale, accurate to 10 g from 0–10 kg and 20 g from 10–20 kg, maintained in accordance with the manufacturer’s instructions. Length was measured using a Harlow Healthcare Rollameter, accurate to 1 mm. Head circumference (occipitofrontal) was determined using Child Growth Foundation Lasso tapes, accurate to 1 mm. Whenever possible, the mean of two or three determinations of length and head circumference were obtained; intraobserver variability was less than 1%.

The mean standard deviation score (SDS) values for birth weight and for length, weight, and head circumference at the three measurement points were determined for all infants using the cross sectional stature and weight reference curves for the United Kingdom, 1990.²⁰

The social class of each study family was determined by comparing the job description of the main family breadwinner with the standard occupational classification.²¹

Statistical analysis

The SPSS program was used for statistical analysis. χ² and Student’s t tests or comparative non-parametric tests were used to compare ranked data and interval means for the study groups. Repeated measures analysis of variance was used to explore the effects of group affiliation on variations in growth, biochemical assays, and nutritional intake over time. Type 1 error probabilities at or below 5% were judged significant.

RESULTS

Subjects

Sixty eight infants were included in the study, with a mean (SD) birth weight of 1470 (0.43) g and mean (SD) gestational age of 31.4 (2.9) weeks. Sixty seven babies were white; the remaining infant was of Asian origin. Thirty seven infants were randomised to the preterm weaning strategy and 31 to the control strategy. Table 1 gives information on birth and discharge characteristics of the two groups.

The groups were similar in terms of medical interventions, feeding history, and nutritional support received in the neonatal unit, except for uptake of breast milk fortifier, which was used for 29 infants in the PWS group compared with 16 in the control group (p = 0.02).

The range of the social classes of the study families was broadly representative of the United Kingdom as a whole. There were no significant differences in the mean age, weight, or height of the parents of infants in the study groups or of the distribution of social classes represented by study families in each group.

During the study, three infants were withdrawn. Two were twins whose families were unable to comply with the protocol, and the third child’s family was concerned about a further blood test and food diary completion.

Dietary intake

Table 2 gives the daily mean (SE) intakes for energy, macronutrients, and iron at the three assessment points. Differences in the numbers of infants included at each assessment point arose from variations in the ratio of bottle fed to breast fed infants over time. Exclusively breast fed infants were excluded from the three analyses of dietary intake because of the inherent difficulties of accurately determining breast milk intake by suckling infants.

At 0 months GCA (term), there were no significant group differences in mean daily intakes of energy or nutrients except for iron, which was significantly higher for the control group infants. The range of intakes of energy and macronutrients was large, reflecting a wide variation in milk intakes.

At 6 months GCA, mean daily intakes of energy, protein, and carbohydrate were significantly higher for the PWS group than the control group. The mean protein energy ratio was 12.9% for the PWS group and 12.5% for the control group. The intervention group achieved a higher mean daily intake of the other measured nutrients (calcium, phosphorus, iron, zinc, copper, and vitamin C) than the control group, but these differences were not significant. The mean daily iron intake was higher for the PWS group than the control group, even though fewer of these babies were being iron supplemented (24% v 46%). Mean daily intakes of iron and vitamin C at 6 months GCA were significantly higher for all babies given vitamin/mineral supplements than for those not given supplements (15.5 v 10.6 mg (p = 0.001) and 145 v 105 mg (p = 0.003) respectively).

The mean postnatal age at which the first solid food was introduced was significantly earlier for babies in the PWS group (14.9 (0.55) weeks) than for those in the control group (17.8 (0.72) weeks) (p = 0.003). The corresponding mean (SE) GCAs for the introduction of the first solid food were 6.3 (0.55) weeks for babies in the PWS group and 9.9 (0.72) weeks for babies in the control group (p = 0.001).

At 12 months GCA, the dietary intakes were less group differentiated than at 6 months GCA. This reflects an expected convergence in feeding practices between groups towards the end of the study. However, the mean daily intake of iron was significantly higher for the intervention group than for the control group, even though fewer in the PWS group were being iron supplemented (3% v 11%). The mean protein energy ratio was significantly higher for the control group (15.6%) than for the intervention group (14.4%) (p = 0.03). This difference arose from the differences in the types and proportions of foods and milks used in feeding the infants in each group.

There were no significant sex differences in energy or nutrient intakes at any of the three assessment points.

Haematology and trace minerals

At 0 months GCA, there were no significant differences in iron indices between groups (table 3). There were no significant differences in any trace element analyte results between groups at 0 or 6 months GCA.

At 6 months GCA, mean haemoglobin and serum iron levels were significantly higher for the PWS group than the control group (table 3). In addition, the mean haemoglobin for the PWS group increased significantly between 0 and 6 months GCA, whereas the mean serum iron of the control group decreased significantly during this time.

Mean serum ferritin levels decreased significantly over time for both groups. However, more babies in the control group (50%) could be considered ferritin deficient (serum ferritin < 10 ng/ml) than could babies in the intervention group (29%). This difference was not significant.

There were no significant sex differences in blood test results at term or 6 months GCA except for the selenium analyses: girls had a significantly higher mean whole blood selenium concentration at 0 months GCA (0.93 v 0.80 μmol/l; p = 0.02) and a significantly higher mean red blood cell selenium concentration at 6 months GCA (1.38 v 1.27 μmol/l; p = 0.04) than boys.

Growth

The infants in the PWS group had a significantly greater mean rate of growth in length/week than babies in the control group (5.1 v 4.9 mm; p = 0.04) between 0 months and 12 months GCA (table 4). Boys had a significantly greater mean rate of growth in length/week than girls (5.1 v 4.9 mm; p = 0.04), but there was no difference in the ratio of boys to girls between groups. There were no significant differences in mean weight, length, or head circumference between the two groups at any measurement point or of SDS of weight, length, or head circumference between the two groups at any assessment point However, repeated measures analysis of variance showed a significant effect due to group (p = 0.008) on the change in length SDS over time (table 5).

There were significant sex effects on growth. The boys had significantly larger heads at all assessment points and were significantly longer but not heavier at 6 and 12 months GCA than girls.

DISCUSSION

Our study compared dietary intakes and haematological and trace element analytes with growth outcomes in preterm infants subjected to two different solid feeding regimens between hospital discharge and 1 year GCA in a blinded, randomised controlled trial. The subjects in this study were, in many respects, representative of preterm infants admitted to neonatal units in the United Kingdom. There was a low study drop out rate of 4%, and the PWS and control groups were similar with respect to birth weight, gestational age, and sex ratio. There were no significant differences in family characteristics, neonatal clinical management, and social class distribution between the groups. Thus differences in group outcome measures are more likely to have arisen from group assignment than from confounding variables.

Dietary intake

There were no significant differences in mean daily intakes of energy or nutrients between the groups at 0 months GCA, except for iron, which was higher in the control group. By 6 months GCA, the PWS group had achieved higher mean daily intakes of energy and all measured nutrients than the control group. In the case of energy, protein, and carbohydrate, these differences (13%, 16%, and 16% higher respectively) were statistically significant and confirmed the efficacy of the preterm weaning strategy in increasing nutrient intakes by 6 months GCA.

At the final assessment point (12 months GCA), mean daily intakes of energy, carbohydrate, iron, zinc, copper, and vitamin C were higher for the PWS group than the control group, but only the mean daily iron intake was significantly higher for the PWS group. It has been assumed that by 1 year GCA, the consumption of some habitual family foods by most of the babies reduced differences in energy and nutrient intakes between the groups.

The significantly higher mean iron intake by infants in the PWS group at 12 months GCA compared with infants in the control group can be attributed, at least in part, to the significant differences in milk feeding patterns at this age arising from the recommendations in the feeding guides. More infants in the control group had been switched to cow’s milk as a main drink at this stage than had infants in the PWS group (71% v 38%; p = 0.04). In addition, the greater emphasis placed on meat in the diet for the intervention group compared with the control group continued to contribute to increased iron intakes from foods in that group.

The dietary results show the effect of the combination of “early weaning” and the use of energy and nutrient dense solid baby foods on nutrient intakes during the study.

Haematology

As expected, there were no significant group differences in the mean values of any haematological or mineral level from the first blood sampling at term.

At 6 months GCA, a significant difference in iron status was apparent between the groups. We propose that the improved iron status of the PWS group arose in response to a combination of early weaning (from 3 months), the use of a preterm formula as a liquefying agent for dried cereal and home prepared dishes, and the increased use of foods with a high iron content, such as meat.

The improved iron status of the PWS group was particularly important because it is recognised that iron stores are quickly depleted by rapid growth,²² that preterm infants have a high iron requirement,⁶ and that iron deficiency and anaemia in infancy all have a long term negative influence on cognition.²³

There were no sex differences in blood results at term or 6 months GCA, except for some of the selenium analyses. Erythrocyte and whole blood selenium levels indicate selenium status over a longer time frame than plasma selenium levels.²⁴ It may be that boys have a higher physiological requirement for selenium than girls. This would explain why plasma selenium levels, which reflect current dietary intakes, were similar for both sexes but whole blood and erythrocyte selenium levels were lower for boys. Alternatively, preterm girls may begin life with larger selenium stores than preterm boys. Friel et al²⁵ similarly reported a better selenium status in girls than boys in a study of very low birthweight infants.

Growth

Infants in the intervention group achieved a faster mean rate of lengthwise growth between term and 12 months GCA than infants in the control group (which was already apparent at 6 months GCA). In the absence of significant differences in subject characteristics, family backgrounds, or sex ratio between groups, we conclude that the difference in feeding practices caused this accelerated rate of growth.

The lack of significant group differences in the rate of increase in weight or head circumference between term and 12 months GCA may indicate that growth in these areas is not as nutritionally sensitive as growth in length.

Over the period of the study, there was a significant effect due to group for length SDS but not for weight SDS or head circumference SDS. This finding provides further evidence to support the beneficial effects of the preterm weaning strategy for lengthwise growth.

However, there were no significant differences in absolute measures of weight, length, or head circumference at any assessment point. This may be because babies in the PWS group were shorter but older at the initial measurement point and younger at the final measurement point than those in the control group. The age differences arose because of the four week range during which each anthropometric determination could be performed. The differences in mean ages at measurement points and in mean initial growth determinations may have masked valid differences in group end points.

The significant difference between the groups in the number of babies who received fortified breast milk did not have any effect on final growth measurements or growth SDS. No confounding variables (other than sex) such as social class or birth weight, which may affect growth, were discovered during statistical analysis.

Study weaknesses

Our preterm infants were a heterogeneous cohort. However, they were stratified to include equivalent numbers of the smallest preterms in each group. Our resources precluded a larger subject group. Therefore it is possible that small or differential effects in the collected data were missed, but we were wary of performing essentially subgroup analyses in this relatively small study. We would expect that a larger study would confirm our results.

Conclusions

The study provides evidence of increased dietary intakes, improved iron status, and enhanced growth in length for infants fed in accordance with the preterm weaning strategy. The study regimen was well received by the parents, who correctly interpreted and carefully executed the feeding guidelines.

The beneficial effects of the strategy were apparent by 6 months GCA, when solid feeding was well established for the intervention group. At this stage, infants in the PWS group were consuming more than the age appropriate estimated average requirement for energy and in excess of the reference nutrient intake for all other nutrients.²⁶

The use of preterm formula after discharge would increase the energy and nutrient density of the diet of formula fed babies and would provide an easy, if expensive, alternative to early weaning. However, preterm breast fed babies would remain nutritionally disadvantaged, whereas the preterm weaning strategy benefits both breast fed and formula fed infants. In addition, the use of preterm formula for a limited period in infancy would not improve weaning practices for preterm infants or long term eating behaviours.

We advocate that, in economically developed countries, the recommendations in the preterm weaning strategy should be adopted for use as the feeding guidelines for preterm low birthweight infants after discharge up until 1 year GCA to promote optimum growth in length (without disadvantaging weight or head circumference growth patterns) and improved iron status. The recommendations are not appropriate for economically less developed countries where other considerations, such as risk of infection, must be given priority.²⁷