Physical activity programs for promoting bone mineralization and growth in preterm infants

Schulzke SM, Trachsel D, Patole SK

Background - Methods - Results - Characteristics of Included Studies - References - Data Tables and Graphs

Dates

Date edited: 14/03/2007
Date of last substantive update: 28/12/2006
Date of last minor update: / /
Date next stage expected 01/01/2009
Protocol first published: Issue 3, 2006
Review first published: Issue 2, 2007

Contact reviewer

Dr Sven Schulzke
Senior Registrar
Neonatology
Women's and Children's Health Service
Princess Margaret Hospital for Children
Subiaco
Western Australia AUSTRALIA
6008
E-mail: sven.schulzke@unibas.ch

Contribution of reviewers

Sven Schulzke: Design and preparation of protocol; literature search, assessment of eligibility and quality of studies, data extraction and data analysis, writing of manuscript.
Daniel Trachsel: Contribution to design of protocol; literature search, assessment of eligibility and quality of studies, data extraction and data analysis, review of manuscript.
Sanjay Patole: Assessment of eligibility and study quality, guidance and supervision for planning and execution of the meta-analysis, review of manuscript.

Internal sources of support

Department of Neonatology, University Children's Hospital, Basel, SWITZERLAND
Department of Paediatric Intensive Care, University Children's Hospital, Basel, SWITZERLAND
Department of Neonatal Paediatrics, King Edward Memorial Hospital, Perth, AUSTRALIA

External sources of support

None

What's new

Dates

Date review re-formatted: / /
Date new studies sought but none found: / /
Date new studies found but not yet included/excluded: / /
Date new studies found and included/excluded: / /
Date reviewers' conclusions section amended: / /
Date comment/criticism added: / /
Date response to comment/criticisms added: / /

Text of review

Synopsis

Babies born too early (premature babies) are often cared for in a fashion that minimizes physical activity in order to reduce stress and stress-related complications. However, lack of physical activity might lead to poor bone development and growth, as seen in bed-ridden children and adults. It is believed that physical activity programs (moving and pressing all joints on all limbs for several minutes a day) may promote bone development and growth in premature babies. This review found that physical activity might have a small benefit on bone development and growth over a short term. There was inadequate data to assess long-term benefits and harms. Based on current knowledge, physical activity programs cannot be recommended as a standard procedure for premature babies.

Abstract

Background

Lack of physical stimulation may contribute to metabolic bone disease of preterm infants resulting in poor bone mineralization and growth. Physical activity programs in the presence of adequate nutrition might help to promote bone mineralization and growth.

Objectives

The primary objective of this review was to assess whether physical activity programs in preterm infants improve bone mineralization and growth and reduce the risk of fractures.

Search strategy

Following the standard search strategy of the Cochrane Neonatal Review Group, a search was conducted in September 2006 including PubMed, EMBASE, CINAHL, the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 3, 2006), cross-references and handsearching of abstracts of the Society for Pediatric Research and the International Journal of Sports Medicine. No language restrictions were applied.

Selection criteria

Randomized and quasi-randomized controlled trials comparing physical activity programs (extension and flexion, range-of-motion exercises for several minutes a day several days per week for at least two weeks) to no organized physical activity programs in preterm infants. Eligible studies included those that provided physical activity for the experimental group, with or without massage and/or tactile stimulation for both experimental and control groups, as well as information on at least one outcome of interest.

Data collection & analysis

Two review authors independently performed searches and extracted data. All three review authors were involved in selection and assessment of quality of studies. The statistical methods included relative risk (RR), risk difference (RD) and number needed to treat (NNT) for dichotomous outcomes and weighted mean difference (WMD) for continuous outcomes, reported with 95% confidence intervals (CI). Heterogeneity was estimated by the I² statistic. A fixed effect model was used to pool data for meta-analyses.

Main results

Six trials enrolling 169 preterm infants (gestational age 26 to 34 weeks) were included in this review. All were small (N = 20 - 49) single center studies evaluating daily physical activity for 3.5 to 4 weeks during initial hospitalization. The methodological quality and reporting of all trials was poor. None of them stated the methods of concealment of patient allocation, the method of randomization or attempted blinding of the intervention. Only two trials attempted blinding of outcome assessors for outcomes relevant to this review.

Two trials (N = 55) demonstrated moderate short-term benefits of physical activity on bone mineralization at completion of the physical activity program. Data was not pooled for meta-analyses due to methodological differences. The only trial (N = 20) assessing long-term effects on bone mineralization showed no effect of physical activity administered during initial hospitalization on bone mineralization at 12 months corrected age. Meta-analysis from three trials (N = 78) demonstrated an effect of physical activity on daily weight gain (WMD 2.77 g/kg/d, 95% CI 1.62, 3.92). Data from two trials (N = 58) showed no effect on linear growth (WMD -0.04 cm/week, 95% CI -0.19, 0.11) or head growth (WMD -0.03 cm/week, 95% CI -0.14, 0.09) during the study period. The I² statistic suggested heterogeneity on the analysis of linear growth (p = 0.006, I² = 86.9%). None of the trials assessed fractures or other outcomes relevant to this review. Data was insufficient for subgroup analyses based on birth weight and calcium/phosphorus intake.

Reviewers' conclusions

There is weak evidence from six small randomized trials of poor methodological and reporting quality that physical activity programs might promote moderate short-term weight gain and bone mineralization in preterm infants. The clinical importance of these findings is questionable given the small effect size and low baseline risk of poor bone mineralization and growth in study participants. Data is inadequate to assess harm or long term effects. Current evidence does not justify the standard use of physical activity programs in preterm infants. Further evaluation of this intervention in well designed trials incorporating extremely low birth weight infants who are at high risk of osteopenia is required. Future trials should report on adverse events and long term outcomes including fractures, growth, bone mineralization, skeletal deformities and neurodevelopmental impairment. These trials should address the possibility that nutritional intake (calories, protein, calcium, phosphorus) might modify the effects of physical activity.

Background

Very low birth weight (VLBW) infants are at risk of developing osteopenia of prematurity. The major etiological factor seems to be substrate deficiency, particularly of calcium and phosphorus, in the presence of low bone mass at birth (Steichen 1980). Immobilization may also contribute to osteopenia (Bishop 1999).

Diagnostic criteria of osteopenia vary considerably. Frequently used biochemical indicators of disturbed bone metabolism are low whole blood phosphate levels, increased urinary calcium/phosphate ratios, and high plasma alkaline phosphatase levels (Bishop 1999). In neonates, peak plasma alkaline phosphatase activities greater than five times the maximum adult normal range of 130 IU/ml are associated with reduced stature at 18 months corrected age in former preterm infants (Lucas 1989).

Osteopenia in preterm infants leads to impaired bone mineralization as measured by techniques such as single photon absorptiometry (SPA) or dual-energy x-ray absorptiometry (DEXA) (Salle 1992; Steichen 1980). As a result, growth velocity and long-term height may be reduced (Lucas 1989). In severe cases, fractures have been reported (Koo 1988). Although reduced bone mineralization is clearly associated with multiple skeletal deformities such as bowing of the legs, scoliosis and skull indentations (Juskeliene 1996; Oyemade 1981; Tubbs 2004), the prevalence of such deformities in former preterm infants with osteopenia has not yet been determined.

Common strategies for the prevention of osteopenia in VLBW infants include calcium and phosphorus supplementation of human milk/formula and physical activity programs. A systematic review of trials investigating the effect of fortification of human milk with multi-component fortifiers in nursery settings found that this intervention in VLBW infants was associated with short-term improvements in linear growth, head growth and weight gain (Kuschel 2004). In a randomized study, post-discharge multi-component fortified formula [compared to standard formula] led to enhanced linear growth at 18 months corrected age (Lucas 2001). Another randomized trial demonstrated increased bone mineral content and growth in preterm infants fed with an isocaloric, calcium and phosphorus enriched formula compared to controls receiving a conventional preterm formula (Lapillonne 2004). In this study, increasing calcium concentration from 80 to 100 mg/100 ml and increasing phosphorus from 42.5 to 60 mg/100 ml as soon as full enteral feedings were reached, was associated with higher bone mineral content and weight at term as measured by DEXA.

While mechanical strain on bones and joints stimulates bone formation and growth, inactivity leads to bone resorption (Larson 2000; MacKelvie 2004). Physical activity programs have been shown to reduce the risk of osteoporotic fractures and bone loss in adults (Bonaiuti 2002; Heinonen 1996; Kerr 2001). Observational studies in children beyond the neonatal age also suggest that physical activity might help to promote bone mineral density (Slemenda 1991). Minimal handling is frequently a routine policy for hospitalized preterm infants in order to facilitate stability and minimise stress. The resultant inactivity may lead to suboptimal stimulation of bone metabolism. Given the evidence from studies in older children and adults, regular physical activity programs (range-of-motion exercises) may provide a simple intervention for improving bone mineral content and skeletal growth in preterm infants. As range-of-motion exercises inevitably have an element of systematic holding and stroking, they may also promote general growth in preterm infants, because interventions solely consisting of systematic holding and stroking (e.g. massage/tactile stimulation) have been reported to promote growth (Vickers 2004). However, physical activity programs in preterm infants may have adverse effects such as fractures, or increase the risk/severity of complications of prematurity (e.g. apnea and bradycardia) with resultant altered blood flow to vital organs such as brain, and the possibility of long-term neurodevelopmental impairment.

Objectives

The primary objective was to assess whether physical activity programs in preterm infants improve bone mineralization and growth and reduce the risk of fractures.

The secondary objectives included other potential benefits in terms of length of hospital stay, skeletal deformities and neurodevelopmental outcome, and adverse events.

Subgroup analysis:
(1) Given that the smallest infants are most vulnerable for developing osteopenia (Bishop 1999) a subgroup analysis was planned for infants with a birth weight < 1000 g.

(2) Calcium and phosphorous intake may affect an infant's ability to increase bone mineral content (Kuschel 2004). Therefore, an additional subgroup analysis was planned for infants receiving different amounts of calcium and phosphorus along with full enteral feeds as follows:

Below 100 mg/60 mg calcium/phosphorus or equal/above 100 mg/60 mg calcium/phosphorus per 100 ml milk
Supplementation of calcium without phosphorus
Supplementation of phosphorus without calcium

Criteria for considering studies for this review

Types of studies

All randomized and quasi-randomized controlled trials in which the unit of allocation was the individual infant.

Types of participants

Preterm infants born at a gestational age < 37 completed weeks who did not receive physical therapy for any indication other than osteopenia of prematurity (e.g. severe contractures).

Types of interventions

Systematic physical activity programs consisting of extension and flexion, range-of-motion exercises of both the infant's upper and lower limbs, administered for several minutes at a time several times a week for at least two weeks, compared to no organized physical activity programs. Eligible studies included those which provided physical activity for the experimental group, with or without massage and/or tactile stimulation for both experimental and control groups.

Types of outcome measures

Trials had to assess at least one of the following outcomes:

Primary outcomes:
a) Bone mineralization
Bone mineral content, bone mineral density, and bone area measured by absorptiometric x-ray techniques

at completion of the physical activity program
at discharge
at term
at 12 - 24 months of corrected age

b) Fractures
Proportion of infants with one or more fractures

at completion of the physical activity program
at discharge
at term
at 12 - 24 months of corrected age

c) Somatic growth
Weight, length, head circumference

at completion of the physical activity program
at discharge
at term
at 12-24 months of corrected age

Secondary outcomes:
a) Complications of prematurity

Mortality at hospital discharge
Mean frequency of apnea during study period
Mean frequency of bradycardia during study period
Mean frequency of apnea and bradycardia during study period
Proportion of infants with one or more proven systemic infections diagnosed during the study period (positive culture from blood, urine, cerebrospinal fluid or other normally sterile body fluids)
Oxygen dependency at 28 days
Chronic lung disease defined as supplemental oxygen or ventilator support at 36 weeks postmenstrual age
Necrotising enterocolitis
Retinopathy of prematurity (all stages and severe stage 3 or greater)
Intraventricular haemorrhage (all grades and severe grades 3 or 4)
Periventricular leukomalacia

b) Length of hospital stay (days)

c) Proportion of infants with one or more secondary skeletal deformities (including skull, spine, limbs)

d) Neurodevelopmental abnormalities at 18 - 24 months of corrected age or later

Cerebral palsy
Developmental delay (assessed by standardized and validated test, e.g. Griffith or Bayley test, with abnormality being defined as more than two standard deviations below the mean)
Intellectual impairment (IQ more than two standard deviations below the mean as assessed by a standardized and validated test)
Blindness (vision less than 6/60 in both eyes)
Sensorineural deafness requiring amplification

Search strategy for identification of studies

The standard strategy of the Cochrane Neonatal Review Group was used for literature search in September 2005 and September 2006. The databases searched included PubMed (1966 to September 2006), EMBASE, and CINAHL. The MeSH headings included Infant, Newborn, Bone Diseases, Metabolic, Motor Activity or Movement or Exercise or Exercise Therapy, and the text words 'Physical activity' or 'Exercise'. The Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 3, 2006) was searched using words 'Newborn' and 'Physical activity' or 'exercise'. Previous reviews including cross-references, and references from identified studies were searched. No language restrictions were applied. Abstracts (1992 - 2005) of the Society for Pediatric Research, European Society for Pediatric Research, and the International Journal of Sports Medicine were also searched.

Methods of the review

The standard methods of the Cochrane Neonatal Review Group were used. Two review authors (SS and DT) independently conducted the literature search. All review authors assessed the trials for eligibility for inclusion and methodological quality. Trials were assessed according to allocation concealment (blinding of randomization), blinding of intervention, completeness of follow up and blinding of outcome measurement. Assessments were specified as "Yes", "No", "Can't tell". Two review authors (SS and DT) extracted data independently, differences were resolved by discussion involving all review authors. Additional information was requested from the authors if necessary. Statistical analyses were done using Review Manager software 4.2.8 (RevMan 2004). Analysis was performed using relative risk (RR), risk difference (RD), and number needed to treat (NNT) for categorical variables and weighted mean difference (WMD) for continuous variables. 95% confidence intervals were used for each statistic. Heterogeneity was estimated by the I² statistic. A fixed effect model was used to pool data for meta-analyses.

Description of studies

Twenty seven abstracts were identified using the prespecified search strategy in September 2006. Thirteen reports (seven full text publications and six abstracts) of nine potentially eligible studies were retrieved for detailed evaluation. No ongoing studies were found.

Included studies
Nine reports of six trials incorporating 169 preterm infants met inclusion criteria of this review (Eliakim 2002; Litmanovitz 2003; Moyer-Mileur 1995; Moyer-Mileur 2000; Moyer-Mileur 2000a; Nemet 2002). Abstracts with preliminary results of three studies (Moyer-Mileur 1995, Moyer-Mileur 2000 and Litmanovitz 2003) had been reported in Pediatric Research prior to full publication. Three included studies (Litmanovitz 2003; Moyer-Mileur 1995; Moyer-Mileur 2000) specified eligibility criteria for patient enrolment (see the Table, "Characteristics of Included Studies"). The following descriptions refer to enrolled rather than eligible patients. Moyer-Mileur 1995, Moyer-Mileur 2000 and Moyer-Mileur 2000a were single center studies in healthy preterm neonates (N = 49, 32 and 20, respectively) conducted at the University Hospital of Utah, UT, USA. Moyer-Mileur 1995 and Moyer-Mileur 2000 enrolled two to four week old preterm infants (mean gestation: 28 to 30 weeks) who were fed either fortified breast milk or preterm formula. The proportion (53 - 73%) of infants fed fortified breast milk did not differ significantly between treatment and control groups. Infants in treatment and control groups received well defined interventions applied by the same trained occupational therapist, described as follows: For the exercise group, range-of-motion exercises with gentle compression, extension and flexion of both upper and lower extremities. Each movement was done five times at each joint (wrist, elbow, shoulder, ankle, knee, and hip) for five times a week. For the control group, tactile stimulation was provided, i.e., a daily interactive period of holding and stroking but no range-of-motion activity. Both protocols were administered for 3.5 to 4 weeks. Outcomes included bone mineralization as measured by absorptiometric x-ray techniques, short-term growth, and biochemical markers of bone metabolism. Moyer-Mileur 2000a was a follow up study assessing bone mineralization and post discharge growth up to 12 months corrected age in infants who had been enrolled in an identical physical activity program during initial hospitalization.

Eliakim 2002, Litmanovitz 2003 and Nemet 2002 were single center studies in preterm neonates (N = 20, 24 and 24, respectively) performed at the Meir General Hospital, Sapir Medical Center, Israel. Eliakim 2002 and Nemet 2002 enrolled four to five week old preterm neonates (mean gestation: 28 to 29 weeks) fed either fortified breast milk or preterm formula. The proportion of infants with chronic lung disease in the study and control group of neonates was comparable in the studies by Eliakim 2002 (40 vs. 40%) and Nemet 2002 (42 vs. 42%). Litmanovitz 2003 enrolled infants of similar gestational age in the first week of life, including infants on parenteral nutrition accompanied by either fortified breast milk or preterm formula. The overall proportion of infants fed fortified breast milk in Eliakim 2002 was 50% and 46% in Litmanovitz 2003 with no significant difference between treatment and control groups within each trial (Eliakim 2002 5/10 vs. 5/10, Litmanovitz 2003 5/12 vs. 6/12). All three trials administered physical activity (treatment group) and tactile stimulation (control group) by a trained person for four weeks based on the Moyer-Mileur protocol (Moyer-Mileur 1995, see above). Outcomes included short-term growth, biochemical markers of bone and fat tissue metabolism, and bone ultrasound measurements.

Details of included studies are shown in the Table, "Characteristics of Included Studies".

Excluded studies
Four reports of three randomized trials (Aly 2004; Hassanein 2002; McIntyre 1992) were excluded from the review. One trial was excluded because the study population consisted of only term infants (McIntyre 1992). Two trials were excluded because the protocol involved physical activity programs plus additional massage in the intervention group, whereas the control group infants received neither of the two measures (Aly 2004; Hassanein 2002). The application of massage to infants only in the intervention group was considered to potentially affect the outcomes.

Details of excluded studies are summarised in the Table, "Characteristics of Excluded Studies".

Methodological quality of included studies

Overall, the methodological as well as reporting quality of the included trials was poor. None of the trials explicitly stated concealment of patient allocation and/or method of randomization. Additional information received after we contacted Dr Moyer-Mileur suggested that patient allocation was concealed in Moyer-Mileur 1995 and Moyer-Mileur 2000.

None of the trials attempted to blind the interventions, which in each trial were applied by the same trained staff person. Short-term follow up was complete (100%) in four trials (Eliakim 2002; Litmanovitz 2003; Moyer-Mileur 2000; Nemet 2002). Moyer-Mileur 1995 lost 23/49 (45%) participants due to hospital discharge or transfer before completing the four week study, leaving 26 (13 infants in treatment group and control group, respectively) in the study. Baseline data and results were available for only 26 infants completing the study. The abstract by Moyer-Mileur 2000a from the same research group, reporting on long term follow up of 20 former preterm neonates from hospital discharge to 12 months corrected age, did not include baseline data of the original cohort receiving the interventions and, therefore, completeness of follow up could not be determined. Assessors of bone mineralization in Moyer-Mileur 1995 and Moyer-Mileur 2000 were blinded. None of the other included trials reported on blinding of assessors for any other outcomes relevant to this review.

Results

PHYSICAL ACTIVITY PROGRAM VERSUS CONTROL (COMPARISON 01)

Primary outcomes

a) Bone mineralization

At completion of the physical activity program
Two trials (Moyer-Mileur 1995; Moyer-Mileur 2000) involving 55 infants reported on bone mineral content, bone mineral density and bone area at completion of the physical activity program. Because of differences in methodologies (SPA versus DEXA) and sites of measurement (radius versus forearm) data was not pooled for meta-analyses.

Bone mineral content at completion of the physical activity program (outcome 01-01):
Infants having physical activity in Moyer-Mileur 1995 (N = 23) had a higher mean radial bone mineral content compared to those in the control group as measured by SPA (mean difference (MD) 10.6 mg/cm, 95% CI 1.6, 19.6). In the study of Moyer-Mileur 2000 there was no significant difference in forearm bone mineral content between infants in physical activity versus control group as measured by DEXA (MD 103 mg, 95% CI -12.17, 218.17).

Bone mineral density at completion of the physical activity program (outcome 01-02):
There was no significant effect of physical activity on radial bone mineral density in Moyer-Mileur 1995 (MD 29.0 mg/cm², 95% CI -1.48, 59.48) as measured by SPA or forearm bone mineral density in Moyer-Mileur 2000 as measured by DEXA (MD -0.2 mg/cm², 95% CI -0.4, 0.0).

Bone area at completion of the physical activity program (outcome 01-03):
Infants in the physical activity group of Moyer-Mileur 2000 (N = 32) had a larger forearm bone area than those in the control group as measured by DEXA (MD 1.2 cm², 95% CI 0.26, 2.14).

At discharge
Not reported in any of the trials.

At term
Not reported in any of the trials.

At 12 to 24 months of corrected age
Bone mineral content and bone area at 12 months corrected age (outcomes 01-04 and 01-05):
One trial (Moyer-Mileur 2000a) with 20 infants who had received physical activity versus control tactile stimulation during their initial hospitalization reported on whole body bone mineral content and whole body bone area at 12 months corrected age as measured by DEXA. There was no difference in whole body bone mineral content (MD -17.3 g, 95% CI -68.95, 34.35) or whole body bone area (MD -21.0 cm², 95% CI -85.6, 43.6).

b) Fractures
Not reported in any of the trials.

c) Somatic growth

At completion of the physical activity program

Body weight gain during study period (outcome 01-06):
Four trials (Eliakim 2002; Moyer-Mileur 1995; Moyer-Mileur 2000; Nemet 2002) with a combined total of 102 infants reported a significant effect of physical activity on weight gain during the study period. Because of lack of data, pooling of results was only possible for three trials (Eliakim 2002; Moyer-Mileur 1995; Moyer-Mileur 2000) incorporating 78 infants. Meta-analysis showed a significant effect of physical activity on daily weight gain (WMD 2.77 g/kg/d, 95% CI 1.62, 3.92).

Body length gain and head circumference gain during study period (outcomes 01-07 and 01-08):
Three trials (Litmanovitz 2003; Moyer-Mileur 1995; Moyer-Mileur 2000) involving 82 infants reported on body length and head circumference during the study period, none reported a significant effect of physical activity on these outcomes. Because of lack of data, pooling of results was possible only for two trials (Moyer-Mileur 1995; Moyer-Mileur 2000) involving 58 infants. Meta-analyses showed no effect of physical activity on gain in body length (WMD -0.04 cm/week, 95%CI -0.19, 0.11) or head circumference (WMD -0.03 cm/week, 95% CI -0.14, 0.09) during the study period. The I² statistic suggested heterogeneity on the meta-analysis of gain in body length (p = 0.006, I² = 86.9%). Using a random effects model did not change results on gain in body length (WMD 0.11 cm/week, 95% CI -0.42, 0.63).

At discharge
Not reported in any of the trials.

At term
Not reported in any of the trials.

At 12-24 months of corrected age
Body weight at 12 months corrected age (outcome 01-09):
Only the trial by Moyer-Mileur 2000a (N = 20, physical activity vs. tactile stimulation during initial hospitalization) reported on body weight at 12 months corrected age. There was no significant effect of physical activity on body weight at 12 months corrected age (MD 200 g, 95% CI -799, 1199).

Secondary outcomes
None of the included trials reported on adverse effects of the interventions, or the length of hospital stay, skeletal deformities, and long term neurodevelopmental impairment.

Subgroup analyses
Pre-planned subgroup analyses based on birth weight and calcium/phosphorus supplementation could not be performed due to inadequate data/lack of data.

Discussion

Analysis of six small randomized trials indicates that daily physical activity programs of 5 to 10 minutes/day, administered for 3.5 to 4 weeks during initial hospitalization, might promote weight gain and improve bone mineralization in the short term in healthy preterm infants (gestation: 26 - 34 weeks) on full enteral feeds of fortified breast milk and/or preterm formula. The effects seem to be limited to the first few months of life. There are no trials reporting on adverse effects, length of hospital stay, fractures, skeletal deformities, long-term neurodevelopment, or other complications of prematurity.

The majority of included studies dealt with preterm infants of several weeks postnatal age who were not small for gestation, had no congenital abnormalities, were medically stable and on full enteral feeds (at least 100 kcal/kg/d). All trials involved nearly identical interventions based on the Moyer-Mileur protocol (Moyer-Mileur 1995) and reported similar beneficial effects of physical activity versus tactile stimulation on daily weight gain. This effect is somewhat surprising given that all studies reporting on weight gain during the study period showed similar nutritional intake (calories, protein, calcium, phosphate) in treatment and control groups. Enhanced bone and fat free mass (Moyer-Mileur 2000) as well as changes in growth hormones leading to an anabolic situation, as evidenced by a trend towards greater insulin like growth factor concentrations in the physical activity group (Eliakim 2002) could explain these findings. Three trials reported moderate short-term but no long-term benefits on bone mineralization. None of the secondary outcomes of interest were addressed. Statistically significant heterogeneity was noted on meta-analysis of the two trials (Moyer-Mileur 1995; Moyer-Mileur 2000) assessing the effect of physical activity on body length (outcome 01-07). One possible reason for heterogeneity might be the difficulty in obtaining accurate measurements of body length in preterm infants. Reproducibility of crown heel length measurements obtained with conventional methods (e.g. tape measurement or pencil marks made on a paper barrier) is low in newborn infants (Rosenberg 1992). Specific devices are hence required for accurate measurements in preterm infants (Lawn 2004). No other reasons for explaining this heterogeneity could be identified.

The results of this review need to be interpreted with great caution given the considerable methodological limitations of the six included trials. Though most trials were published several years after the first CONSORT statement (Begg 1996), the quality of reporting was poor. Recruitment bias cannot be excluded since none of the trials provided a patient flow chart or numbers of eligible patients in relation to enrolled, evaluated, and "lost to follow up" participants. None of the included trials clearly explained concealment of patient allocation and method of randomization. The majority of included trials (Eliakim 2002; Litmanovitz 2003; Moyer-Mileur 1995; Nemet 2002) reported that infants were "matched for gestational age, birth weight, gender, corrected age, and weight at start of the study and were then randomized to either treatment or control group" without further explanation. Given the small sample size (N = 20 - 49) and identical number of infants in the treatment and control group in all trials, the quality and adequacy of randomization in the majority of the included trials should be questioned. However, concealment of patient allocation seems to be adequate in Moyer-Mileur 1995 and Moyer-Mileur 2000 based on comments from the authors. None of the trials attempted blinding of the intervention. Follow up was incomplete in the largest study included in this review (Moyer-Mileur 1995, N = 49, follow up rate 55%) and impossible to assess due to lack of data in the only study reporting on long term effects (Moyer-Mileur 2000a). Apart from the evaluation of bone mineralization in Moyer-Mileur 1995 and Moyer-Mileur 2000, the assessment of outcomes relevant to this review were most likely not blinded in any of the trials. The lack of statistically significant heterogeneity in most meta-analyses in this review does not exclude heterogeneity given the small numbers. In relation to the clinical relevance of the results, it is important to realize that the baseline risk of osteopenia in participants was not high given their gestation and birth weight. Thus, the validity and general applicability of these results are very limited. Additionally, the clinical significance of neonatal unintentional physical activities (bathing, changing nappies, skin care) in relation to structured physical activity programs of a few minutes per day remains unclear.

Reviewers' conclusions

Implications for practice

Implications for research

Evaluation of the benefits and harms of physical activity programs for promoting bone mineralization and growth requires further testing in well designed randomized trials incorporating extremely low birth weight infants who are at high-risk for the condition. Such trials should aim at monitoring and reporting adverse events (e.g. apnea, sepsis, fractures), as well as long term growth, bone mineralization and neurodevelopmental outcomes, while addressing the possibility that nutritional intake (calories, protein, calcium, phosphorus) might modify the effect of physical activity.

Acknowledgements

We thank Bonny Specker, Hany Aly, and Laurie Moyer-Mileur who clarified existing data and provided us with additional information.

Potential conflict of interest

None

Characteristics of included studies

Study	Methods	Participants	Interventions	Outcomes	Notes	Allocation concealment
Eliakim 2002	Single center randomized controlled trial. Randomization and concealment of allocation: Can't tell. Intervention not blinded, both activity protocols provided by the same trainer. Follow-up complete (20/20). Blinding of outcome assessors: Can't tell.	Eligibility criteria unclear. Enrolled patients: 20 preterm neonates "matched for GA, BW, gender, CA, and weight" at initiation of study. Mean birth weight approx. 1050 g, mean GA approx. 28 to 29 wks, mean corrected age at enrolment 33 wks GA. Most infants healthy, but inclusion of several neonates with BPD and/or diuretics. Exclusion criteria: Severe IUGR, severe CNS disorders, suspected bone and/or muscular diseases or sepsis during study period. Infants had at least 100 kcal/kg/d oral intake and were fed either fortified breast milk or preterm formula.	Type of interventions: As described in Moyer-Mileur 1995 for both, intervention group (n=10) and control group (n=10). Duration of interventions: 4 wks.	Weight (g) at begin of study and completion of protocol (28 study days). Additionally, biochemical markers of bone and fat tissue metabolism (not subject of this review).		B
Litmanovitz 2003	Single center randomized controlled trial. Randomization and concealment of allocation: Can't tell. Intervention not blinded, both activity protocols provided by the same person. Follow-up complete (24/24). Partial blinding of outcome assessors (yes for ultrasound assessors, can't tell for anthropometric data). A preliminary report of this study had been published in abstract form.	Eligibility criteria: Birth weight < 1500 g, body size appropriate for gestational age and birthweight, postnatal age less than 1 wk, informed parental consent. Enrolled patients: 24 preterm neonates, mean birthweight 1135 g, mean GA approx. 28 to 29 wks. In addition to parenteral nutrition, infants were either fed fortified human breast milk or preterm formula. Exclusion criteria: intrauterine growth restriction, severe central nervous system disorder, major congenital anomalies.	Type of interventions: As described in Moyer-Mileur 1995 for both, intervention group (n=12) and control group (n=12). Duration of interventions: 4 weeks.	Weight (g), length (cm), head circumference (cm) at begin of study and completion of protocol (28 study days). Additionally, bone ultrasound measurements (not subject of this review).	This is the only trial involving infants in the first week of life.	B
Moyer-Mileur 1995	Single center randomized controlled trial. Randomization was accomplished by selecting sealed envelopes containing group codes that had been generated from a random numbers table. Intervention not blinded, as both activity protocols were administered by the same trained occupational therapist. Follow-up incomplete, 23/49 infants (45%) were unavailable for assessment of all outcomes because of discharge/transfer before completion of the exercise protocol. The drop-out rate was similar between groups. Blinding of outcome assessors: The technician who measured and analyzed bone density measurements was blinded to the subjects' group assignment. Anthropometric measurements: Can't tell. A preliminary report of this study had been published in abstract form.	Eligibility criteria: 26 to 34 wks GA, appropriate body size, able to tolerate enteral feeds of either fortified breast milk or preterm formula, both 24 kcal/oz, at or above 110 kcal/kg/d. Absence of medication other than vitamin supplements, informed parental consent. Enrolled patients: 49 healthy neonates, mean birth weight approx. 1200 g, mean GA approx. 28 to 29 wks, mean corrected age at enrolment 30 to 31 wks GA.	Type of interventions: Exercise group (analysed n=13): range-of-motion exercises with gentle compression and extension and flexion of both upper and lower extremities. Each movement was done five times at each joint (wrist, elbow, shoulder, ankle, knee, and hip). The program was applied 5 times a week. Control group (analysed n=13): Tactile stimulation, described as a daily interactive period of holding and stroking but no range-of-motion activity. Duration of interventions: 4 weeks.	Anthropometric and bone mineral data by SPA from right distal radius at begin of study and completion of protocol (28 study days): Weight (g), weight gain (g/kg/day), length (cm), length change (cm/week), head circumference (cm), head circumference change (cm/week). Bone mineral content (mg/cm), bone mineral content change (%), bone width (mm), bone width change (%), bone mineral density (mg/cm2), bone mineral density change (%). Additionally, several biochemical markers of bone metabolism (not subject of this review).	Bone mass data from two infants in the physical activity group and one infant in the control group were omitted from data analysis by the authors of this trial because they were > 2 standard deviations from the mean for bone mineral content and bone width. The calculation of mean gain in occipito frontal head circumference in the meta-analyses is based on manual calculation of head circumference at completion - head circumference at study entry divided by four and the standard deviation reported in the paper.	A
Moyer-Mileur 2000	Single center randomized controlled trial. Randomization by selecting sealed envelopes containing group codes. Intervention not blinded, principal investigator and occupational therapist were aware of the subjects' group allocation. Follow-up complete (32/32). Blinding of outcome assessors: Yes for pDEXA measurements, no for other outcomes. A preliminary report of this study had been published in abstract form.	Eligibility criteria: 26 to 32 wks GA, birth weight 800 to 1600 g, appropriate body size, tolerating enteral feeds of either fortified breast milk or preterm formula, both 24 kcal/oz, at or above 110 kcal/kg/d. Absence of medication other than vitamin supplements, informed parental consent. Enrolled patients: 32 healthy neonates, mean GA 28 to 29 wks, mean corrected age at enrolment 32 wks GA. Stratification: infants 800 to 1200 g and 1201 to 1500 g, infants 26 to 29 wks GA and 30 to 32 wks GA.	Type of interventions: As described in Moyer-Mileur 1995 for both, intervention group (n=16) and control group (n=16). Duration of interventions: Both protocols were continued until a body weight of 2.0 +- 0.15 kg was reached, resulting in a mean of 27 vs 24 study days for intervention vs control group.	Anthropometric and bone mineral data from right forearm (ulna and radius) by pDEXA at begin of study and completion of protocol (reaching body weight of 2000 g): Weight (g), weight gain (g/kg/day), length (cm), length change (cm), head circumference (cm), head circumference change (cm), bone mineral content (mg), bone mineral content gain (mg), bone area (cm2), bone area gain (cm2), bone mineral density (mg/cm2), bone mineral density gain (mg/cm2). Additionally, several biochemical markers of bone metabolism (not subject of this review).		A
Moyer-Mileur 2000a	Single center randomized controlled trial. Randomization, concealment of allocation, blinding of intervention, completeness of follow-up and blinding of outcome assessors: Can't tell. This is a follow-up study published in abstract form.	Eligibility criteria unclear. Enrolled patients:20 former preterm neonates, follow-up from initial hospital discharge until 12 months corrected age.	Type of interventions: As described in Moyer-Mileur 1995 for both, intervention group (analysed n=10) and control group (analysed n=10). Duration of interventions: Unknown (physical activity was stopped at hospital discharge).	Anthropometric measurements and bone mineral data by DEXA at hospital discharge, 6 months and 12 months corrected age: Weight (g), length (cm), total body bone mineral content (gm), total body bone area (cm2), total body bone density (g/cm2). Additionally, fat mass and fat free mass (not subject of this review).	The origin of the follow-up study population and the number of randomised (N=?) versus analysed patients (N=20) is unclear. Infants in treatment and control group did not differ in any reported outcome at discharge (begin of follow-up) and at 12 months (last follow-up).	B
Nemet 2002	Single center randomized controlled trial. Randomization and concealment of allocation: Can't tell. Intervention not blinded, both activity protocols provided by the same trainer. Follow-up complete (24/24). Blinding of outcome assessors: Can't tell.	Eligibility criteria unclear. Enrolled patients: 24 preterm neonates "matched for GA, BW, gender, CA, and weight" at initiation of study. Mean birth weight 1050 g, mean GA 28 to 29 wks, corrected age at enrolment 33 wks GA. Inclusion of several neonates with BPD and/or diuretics. Exclusion criteria: Severe IUGR, severe CNS disorders, suspected bone and/or muscular diseases or sepsis during study period. Infants had at least 100 kcal/kg/d oral intake and were fed either fortified breast milk or preterm formula.	Type of interventions: As described in Moyer-Mileur 1995 for both, intervention group (n=12) and control group (n=12). Duration of interventions: 4 wks.	Weight (g) at begin of study and completion of protocol (28 study days). Additionally, biochemical markers of bone metabolism (not subject of this review).		B

Characteristics of excluded studies

Study	Reason for exclusion
Aly 2004	Randomized controlled trial on physical activity and massage. Neonates in the intervention group were treated with physical activity and additional massage. Neonates in the control group didn't receive any of those measures. The trial was excluded because massage was provided to the intervention group only. A preliminary report of this study has been published in abstract form.
Hassanein 2002	This abstract reports on a randomized controlled trial on physical activity and massage, as stated by the authors when contacted in January 2006. Neonates in the intervention group were treated with physical activity and additional massage. Neonates in the control group didn't receive any of those measure. The trial was excluded because massage was provided to the intervention group only.
McIntyre 1992	This abstract reports on a randomized controlled trial in infants one to fifteen months old. The trial was excluded because there were no preterm infants in the study population, as clarified by the authors when contacted in January 2006.

References to studies

References to included studies

Eliakim 2002 {published data only}

Eliakim A, Dolfin T, Weiss E, Shainkin-Kestenbaum R, Lis M, Nemet D. The effects of exercise on body weight and circulating leptin in premature infants. Journal of Perinatology 2002;22:550-4.

Litmanovitz 2003 {published data only}

* Litmanovitz I, Dolfin T, Friedland O, Arnon S, Regev R, Shainkin-Kestenbaum R, et al. Early physical activity intervention prevents decrease of bone strength in very low birth weight infants. Pediatrics 2003;112:15-9.

Litmanovitz I, Friedland O, Dolfin T, Arnon S, Regev R, Eliakim A. Early physical activity intervention prevents decrease of bone strength in very low birth weight infants. Pediatric Research 2002;51:380A.

Moyer-Mileur 1995 {published data only}

Moyer-Mileur L, Leutkemeier MJ, Chan GM. Physical activity enhances bone mass in very low-birth weight (VLBW) infants. Pediatric Research 1995;37:314A.

* Moyer-Mileur L, Luetkemeier M, Boomer L, Chan GM. Effect of physical activity on bone mineralization in premature infants. The Journal of Pediatrics 1995;127:620-5.

Moyer-Mileur 2000 {published data only}

* Moyer-Mileur LJ, Brunstetter V, McNaught TP, Gill G, Chan GM. Daily physical activity program increases bone mineralization and growth in preterm very low birth weight infants. Pediatrics 2000;106:1088-92.

Moyer-Mileur LJ, McNaught TJ, Gurmail G, Chan GM. Physical activity and diet: key components for improved bone mass in premature, very-low-birth weight (VLBW: 1,500 gm) infants. Pediatric Research 1999;45:287A.

Moyer-Mileur 2000a {published data only}

Moyer-Mileur LJ, Bail SD, McNaught TP, Chan GM. Effect of physical activity on bone mineralization and body composition in preterm infants during the first year of life. Pediatric Research 2000;47:292A.

Nemet 2002 {published data only}

Nemet D, Dolfin T, Litmanowitz I, Shainkin-Kestenbaum R, Lis M, Eliakim A. Evidence for exercise-induced bone formation in premature infants. International Journal of Sports Medicine 2002;23:82-5.

References to excluded studies

Aly 2004 {published data only}

* Aly H, Moustafa MF, Hassanein SM, Massaro AN, Amer HA, Patel K. Physical activity combined with massage improves bone mineralization in premature infants: a randomized trial. Journal of Perinatology 2004;24:305-9.

Moustafa MF, Aly HZ, Hassanein SM, Nguyen AT, Amer HA, Patel K. Can a daily program of massage and physical exercise affect bone mineralization. Pediatric Research 2003;53:407A.

Hassanein 2002 {published data only}

Hassanein SM, Moustafa MF, Amer HA. Daily physical exercise stimulates growth in premature infants: a randomized controlled trial. Pediatric Research 2002;LB 2-10.

McIntyre 1992 {published data only}

McIntyre L, Hudson P, Smith L, Ho M, Specker B. Exercise increases bone mineral content in infants 1 to 15 months of age. Pediatric Research 1992;31:97A.

* indicates the primary reference for the study

Other references

Additional references

Begg 1996

Begg CB, Cho M, Eastwood S, Horton R, Moher D, Olkin I et al. Improving the quality of reporting of randomized controlled trials. The CONSORT statement. JAMA 1996;276:637-639.

Bishop 1999

Bishop NJ. Metabolic bone disease. In: Rennie JM, Roberton NRC, editor(s). Textbook of neonatology. 3rd edition. Edinburgh: Churchill Livingstone, 1999:1002-8.

Bonaiuti 2002

Bonaiuti D, Shea B, Iovine R, Negrini S, Robinson V, Kemper HC, et al. Exercise for preventing and treating osteoporosis in postmenopausal women. The Cochrane Database of Systematic Reviews 2002, Issue 2.

Heinonen 1996

Heinonen A, Kannus P, Sievanen H, Oja P, Pasanen M, Rinne M, et al. Randomised controlled trial of effect of high-impact exercise on selected risk factors for osteoporotic fractures. Lancet 1996;348:1343-7.

Juskeliene 1996

Juskeliene V, Magnus P, Bakketeig LS, Dailidiene N, Jurkuvenas V. Prevalence and risk factors for asymmetric posture in preschool children aged 6-7 years. International Journal of Epidemiology 1996;25:1053-9.

Kerr 2001

Kerr D, Ackland T, Maslen B, Morton A, Prince R. Resistance training over 2 years increases bone mass in calcium-replete postmenopausal women. Journal of Bone and Mineral Research 2001;16:175-81.

Koo 1988

Koo WW, Sherman R, Succop P, Oestreich AE, Tsang RC, Krug-Wispe SK, et al. Sequential bone mineral content in small preterm infants with and without fractures and rickets. Journal of Bone and Mineral Research 1988;3:193-7.

Kuschel 2001

Kuschel CA, Harding JE. Calcium and phosphorus supplementation of human milk for preterm infants (Cochrane Review). The Cochrane Database of Systematic Reviews 2001, Issue 4.

Kuschel 2004

Kuschel CA, Harding JE. Multicomponent fortified human milk for promoting growth in preterm infants (Cochrane Review). The Cochrane Database of Systematic Reviews 2004, Issue 1.

Lapillonne 2004

Lapillonne A, Salle BL, Glorieux FH, Claris O. Bone mineralization and growth are enhanced in preterm infants fed an isocaloric, nutrient-enriched preterm formula through term. American Journal of Clinical Nutrition 2004;80:1595-1603.

Larson 2000

Larson CM, Henderson RC. Bone mineral density and fractures in boys with Duchenne muscular dystrophy. Journal of Pediatric Orthopedics 2000;20:71-4.

Lawn 2004

Lawn CJ, Chavasse RJ, Booth KA, Angeles M, Weir FJ. The neorule: a new instrument to measure linear growth in preterm infants. Archives of Disease in Childhood Fetal and Neonatal Edition 2004;89:F360-3.

Lucas 1989

Lucas A, Brooke OG, Baker BA, Bishop N, Morley R. High alkaline phosphatase activity and growth in preterm neonates. Archives of Disease in Childhood 1989;64:902-9.

Lucas 2001

Lucas A, Fewtrell MS, Morley R, Singhal A, Abbott RA, Isaacs E, et al. Randomized trial of nutrient-enriched formula versus standard formula for postdischarge preterm infants. Pediatrics 2001;108:703-11.

MacKelvie 2004

MacKelvie KJ, Petit MA, Khan KM, Beck TJ, McKay HA. Bone mass and structure are enhanced following a 2-year randomized controlled trial of exercise in prepubertal boys. Bone 2004;34:755-65.

Oyemade 1981

Oyemade GA. The correction of primary knee deformities in children. International Orthopaedics 1981;5:241-5.

Rosenberg 1992

Rosenberg SN, Verzo B, Engstrom JL. Reliability of length measurements for preterm infants. Neonatal Network 1992;11:23-7.

Salle 1992

Salle BL, Braillon P, Glorieux FH, Brunet J, Cavero E, Meunier PJ. Lumbar bone mineral content measured by dual energy X-ray absorptiometry in newborns and infants. Acta Paediatrica 1992;81:953-8.

Slemenda 1991

Slemenda CW, Miller JZ, Hui SL, Reister TK, Johnston CCJ. Role of physical activity in the development of skeletal mass in children. Journal of Bone and Mineral Research 1991;6:1227-33.

Steichen 1980

Steichen JJ, Gratton TL, Tsang RC. Osteopenia of prematurity: the cause and possible treatment. The Journal of Pediatrics 1980;96:528-34.

Tubbs 2004

Tubbs RS, Webb D, Abdullatif H, Conklin M, Doyle S, Oakes WJ. Posterior cranial fossa volume in patients with rickets: insights into the increased occurrence of Chiari I malformation in metabolic bone disease. Neurosurgery 2004;55:380-3.

Vickers 2004

Vickers A, Ohlsson A, Lacy JB, Horsley A. Massage for promoting growth and development of preterm and/or low birth-weight infants. The Cochrane Database of Systematic Reviews 2004, Issue 2.

Comparisons and data

Comparison or outcome	Studies	Participants	Statistical method	Effect size
01 Physical activity program versus control
01 Bone mineral content at completion of the physical activity program			WMD (fixed), 95% CI	Subtotals only
02 Bone mineral density at completion of the physical activity program			WMD (fixed), 95% CI	Subtotals only
03 Bone area at completion of the physical activity program (cm2)	1	32	WMD (fixed), 95% CI	1.20 [0.26, 2.14]
04 Bone mineral content at 12 months corrected age (g)	1	20	WMD (fixed), 95% CI	-17.30 [-68.95, 34.35]
05 Bone area at 12 months corrected age (cm2)	1	20	WMD (fixed), 95% CI	-21.00 [-85.60, 43.60]
06 Body weight gain during study period (g/kg/d)	3	78	WMD (fixed), 95% CI	2.77 [1.62, 3.92]
07 Body length gain during study period (cm/wk)	2	58	WMD (fixed), 95% CI	-0.04 [-0.19, 0.11]
08 Head circumference gain during study period (cm/wk)	2	58	WMD (fixed), 95% CI	-0.03 [-0.14, 0.09]
09 Body weight at 12 months corrected age (g)	1	20	WMD (fixed), 95% CI	200.00 [-799.39, 1199.39]

01 Physical activity program versus control

01.01 Bone mineral content at completion of the physical activity program

01.01.01 Radial bone mineral content (mg/cm) measured by single photon absorptiometry (SPA)

01.01.02 Forearm bone mineral content (mg) measured by dual-energy x-ray absorptiometry (DEXA)

01.02 Bone mineral density at completion of the physical activity program

01.02.01 Radial bone mineral density (mg/cm2) measured by single photon absorptiometry (SPA)

01.02.02 Forearm bone density (mg/cm2) measured by dual-energy x-ray absorptiometry (DEXA)

01.03 Bone area at completion of the physical activity program (cm2)

01.04 Bone mineral content at 12 months corrected age (g)

01.05 Bone area at 12 months corrected age (cm2)

01.06 Body weight gain during study period (g/kg/d)

01.07 Body length gain during study period (cm/wk)

01.08 Head circumference gain during study period (cm/wk)

01.09 Body weight at 12 months corrected age (g)

Contact details for co-reviewers

Sanjay Patole, MD, DCH, FRACP
Staff Neonatologist, Senior Clinical Lecturer
Neonatal Paediatrics
King Edward Memorial Hospital for Women
Bagot Road
Subiaco
Perth AUSTRALIA
6008
Telephone 1: 61 08 93401260
Facsimile: 61 08 93401266
E-mail: Sanjay.Patole@health.wa.gov.au

Dr Daniel Trachsel
Consultant Paediatrician
Dept of Pediatric Intensive Care/Pulmonology
University Children's Hospital Basel
PO Box, CH-4005
Basel
SWITZERLAND
4005
Telephone 1: 41 61 6855472
Telephone 2: 41 61 6856565
E-mail: daniel.trachsel@ukbb.ch

This review is published as a Cochrane review in The Cochrane Library, Issue 2, 2007 (see http://www.thecochranelibrary.com for information). Cochrane reviews are regularly updated as new evidence emerges and in response to feedback. The Cochrane Library should be consulted for the most recent version of the review.