Original Article

Physical Activity Combined with Massage Improves Bone

Mineralization in Premature Infants: A Randomized Trial
Hany Aly, MD from 82.3±8.5 to 68.78±14.6 (p<0.01), while urinary Pyd increased
Mohamed F. Moustafa from 447.7±282.8 to 744.9±373.6 (p<0.01) indicating decreased bone
Sahar M. Hassanein formation and increased bone resorption, respectively. In the intervention
An N. Massaro group, serum PICP increased over time from 62.5±13.8 to 73.84±12.9
Hanna A. Amer ( p<0.01). Urinary Pyd also increased over time from 445.7±266.5 to
Kantilal Patel 716.8±301.8 (p<0.01). In a linear regression model including
gestational age and intervention, serum PICP increased significantly in
the intervention group (regression coefficient 18.8±4.6, p ¼ 0.0001)
while urinary Pyd did not differ between groups (regression
BACKGROUND: coefficient ¼ 5.6±114.3, p ¼ 0.961).
Osteopenia of prematurity is a known source for morbidity in preterm
CONCLUSIONS:
infants. Premature infants have shown favorable outcomes in response to
A combined massage and physical activity protocol improved bone
massage and physical activity. Whether such intervention can stimulate
formation (PICP) but did not affect bone resorption (Pyd). Pyd increased
bone formation or decrease bone resorption is yet to be determined.
over time in both groups, possibly due to continuous bone resorption and
OBJECTIVE: Ca mobilization.
To test the hypothesis that massage combined with physical activity can Journal of Perinatology (2004) 24, 305–309. doi:10.1038/sj.jp.7211083
stimulate bone formation and ameliorate bone resorption in premature Published online 8 April 2004
infants.

DESIGN/METHODS:
A prospective double-blinded randomized trial was conducted at the
Neonatal Intensive Care Unit of Ain Shams University in Cairo, Egypt. BACKGROUND
Thirty preterm infants (28 to 35 weeks’ gestation) were randomly
Despite recent advances in neonatal intensive care practices and
assigned to either control group (Group I, n ¼ 15) or intervention group
clearly defined nutritional goals for premature low-birth-weight
(Group II, n ¼ 15). Infants in the intervention group received a daily
infants, osteopenia of prematurity remains a well-identified source
protocol of combined massage and physical activity. Serum type I
of morbidity in preterm infants. 1,2 Bone mass in these infants does
collagen C-terminal propeptide (PICP) and urinary pyridinoline
not approach normal ranges until after the first year of life and
crosslinks of collagen (Pyd) were used as indices for bone formation and
may continue to be suboptimal into childhood.3 Nutritional
resorption, respectively. PICP and Pyd were measured at enrollment and
intervention alone did not produce the desired positive effect on
at discharge for all subjects. t-Test, ANOVA and linear regression analysis
postnatal bone mineralization in these infants.1,3 Bone formation
were used for statistical analyses.
and growth, however, can be stimulated by physical activity with
RESULTS: mechanical loading on bone and joints.4 In previous studies,
There was no difference between groups I and II in gestational age massage therapy increased weight gain and enabled earlier
(32.1±1.8 vs 31.5±1.4 weeks) or birth weight (1.429±0.148 vs discharge of premature infants.5,6 Recent studies have also
1.467±0.132 g). In the control group, serum PICP decreased over time demonstrated the ability of exercise therapy to attenuate the
decrease in bone strength associated with osteopenia of prematurity
and to promote growth and bone mineral density in this
population.7–9 Whether massage therapy combined with physical
Neonatology Department (H.A., A.N.M.), The George Washington University Hospital & Children’s
activity stimulates bone mineralization in premature infants and
National Medical Center, Washington, DC, USA; Pediatrics Department (M.F.M., S.M.H.), Ain prevents osteopenia of prematurity is yet to be determined.
Shams University School of Medicine, Cairo, Egypt; Clinical Pathology Department (H.A.A.), Ain
Shams University School of Medicine, Cairo, Egypt; and Biostatistics Department (K.P.), the
Skeletal development involves a delicate balance between two
Children’s National Medical Center, Washington, DC, USA. different bone activities. Osteoblastic activity constructs the organic
Address correspondence and reprint requests to Hany Aly, MD, The George Washington University bone matrix upon which mineral content is later incorporated.
Hospital, 900 23rd Street, N.W. Suite G2092, Room G132, Washington, DC 20037, USA. Osteoclastic activity is responsible for bone resorption and
Journal of Perinatology 2004; 24:305–309
r 2004 Nature Publishing Group All rights reserved. 0743-8346/04 $25

www.nature.com/jp 305
 Aly et al. Bone Mineralization in Infants

remodeling.10 Several biochemical markers that correlate to these Infants in the control group did not receive any of the above
two activities have been recently identified. C-terminal procollagen measures. For both groups, type of feeding was recorded and the
peptide or propeptide of type I collagen (PICP) correlates with amount of calories, calcium, phosphorus and protein given per day
collagen turnover and bone formation in premature infants.11,12 was calculated for all studied neonates. Body weight was recorded
Urine pyridinoline crosslinks of collagen (Pyd) is a marker for daily at a standard time each day, before feeding, with digital
bone resorption.12,13 These two markers provide a useful, electronic scale (Universal Weight Enterprise, Inc., Taiwan) and
noninvasive tool to monitor bone formation in premature infants. was approximated to the nearest gram.
The goal of this trial is to examine serum biochemical markers of
bone mineralization in preterm infants exposed to massage and Laboratory measurements
physical activity, and to test the hypothesis that this intervention Biomarkers of bone formation (serum PICP) and resorption (urine
can stimulate bone formation in preterm infants. Pyd), as well as serum calcium, alkaline phosphatase and
parathyroid hormone (PTH) were measured at study entry and at
1.8 kg. of body weight.
PATIENTS AND METHODS
This prospective double-blinded randomized trial was conducted at Serum level of calcium and alkaline phosphatase Sample
the neonatal intensive care unit of Ain Shams University Children’s was withdrawn without stasis (tourniquet), and immediately
Hospital in Cairo, Egypt. The institutional review board of the analyzed for calcium and alkaline phosphatase using the Hitachi
university approved the study and consents were obtained before 917s analyzer. The remainder of the serum was immediately
infants’ enrolment. Infants who were included in the study had a stored at 201C until use for assessment of other markers.
gestational age of 35 weeks or less, a postnatal age of less than 2
weeks, were tolerating full enteral feeds, and were not receiving any Serum PICP The procollagen-C assay is a sandwich immunoassay
medications except for the recommended vitamin supplementation. in a microtiter plate format utilizing a monoclonal anti-PICP
Infants were excluded from the study if they had a history of antibody coated on the plate, a rabbit anti-PICP antiserum, a goat
congenital malformations, asphyxia, or musculoskeletal, liver or anti-rabbit alkaline phosphatase conjugate and a pNPP substrate to
renal diseases. Assessment of maturity by date of last menstrual quantify PICP in human serum (Metra Biosystems, Inc., San
period, antenatal ultrasound when available and the new Ballard Diego, CA, USA).
scoring system was performed for all neonates.14
Urine Pyd This is competitive enzyme immunoassay in a
Randomization microtiter strip format utilizing a monoclonal antipyridinoline
Infants were randomized to either the control group (group I) or antibody to measure Pyd and Dpd in urine. The Pyd and Dpd in
the intervention group (group II) by selecting a closed envelope the sample compete for the antibody with Pyd coated on the strip.
containing a group code. The reaction is detected with a pNPP substrate. Pyrilinks results are
corrected for urinary concentration by correlating to creatinine
Intervention (Metra Biosystems, Inc., San Diego, CA, USA).
Infants in the intervention group received both physical activity
and massage. Physical activity was composed of daily range of PTH PTH hormone was measured by enzyme-linked immuno
motion, with gentle compression and extension/flexion to both assay (ELISA) technique using PTH-EASIA, which is a solid phase
upper and lower extremities. Five repetitions of each movement enzyme amplified sensitivity immunoassay performed on a
were performed at both wrists, elbows, shoulders, ankles, knees and microtiter plate. It allows the determination of the intact human
hips. Massage protocol was composed of gentle, slow stroking of PTH in serum or plasma. Sample binds to the affinity
each part of the body in turn. The infant was placed in a prone chromatography purified antibodies (Pabs, goat anti-1-34 PTH
position and stroked for a 1 minute period (12 strokes at fragment) coated on the inner surface of the microwell.
approximately 5 seconds per stroking motion) over each region in
the following sequence: (1) from the infant’s head and face to the Statistical Analysis
neck; (2) from the neck across the shoulders; (3) from the upper Data were analyzed using the SAS Systems Version 6.12.15
back to the waist; (4) from the thigh to the foot on both legs; and Demographic data were analyzed using t-test, Fisher’s exact test
(5) from the shoulder to the hand to the shoulder on both arms. and w2-test. For the five-biochemical markers of bone metabolism
Physical activity was performed daily by the same physician until (calcium, alkaline phosphatase, PICP, Pyd and parathyroid
the infant reached 1.8 kg. Vital signs and oxygen saturations were hormone), the absolute change over the study period was
constantly monitored 15 minutes before, during and 15 minutes determined for each variable. Percent changes in variables were
after each daily intervention. analyzed and compared between groups. The Kruskal–Wallis Test

306 Journal of Perinatology 2004; 24:305–309

Bone Mineralization in Infants Aly et al.

(w2 approximation) was used to determine significant differences higher in the activity group (19.5%) than in the control group
between the activity and control groups. A regression analysis was (4.3%), p ¼ 0.002.
also performed in order to control for gestational age since previous
reports have suggested its correlation with PICP levels.16 Serum alkaline phosphatase Serum alkaline phosphatase was
also similar for both groups at study entry (P ¼ 0.665). These
Power analysis We hypothesized that our intervention would levels did not change significantly for either group over the course
increase the group difference in PICP by three fold. When we used of the study.
the PICP baseline value from a previous study8 and assumed a
correlation coefficient (r ¼ 0.7) between pre- and postintervention Serum PICP Mean PICP concentration was higher in Group I
observations, a sample of 30 subjects (15 in each group) would (82.3±8.5 ng/ml) than in Group II (62.5±13.8 ng/ml) at the start
detect that difference with s ¼ 80, b ¼ 20 and a ¼ 0.05 level of of the study (p<0.001). In the control group PICP levels decreased
significance. during the study period while they increased in the activity group
(Figure 1). The percent change was significantly different for the
RESULTS two groups (p<0.001).

A total of 30 infants were enrolled in the study. They were 90

randomized blindly into control and activity groups, with 15
infants assigned to each group. The groups were similar in
PICP (ng/dL)
gestational age, birth weight and gender distribution. They also 60
had no significant differences in caloric, calcium or protein intake Control
over the study period (Table 1). Laboratory values at enrolment Intervention
and at conclusion of the study for both groups are shown in 30
Table 2.

0
Serum calcium Mean serum calcium at study entry was Start of Study End of Study
9.1±1.6 mg/dl. for Group I and 8.5±1.4 mg/dl for Group II Time
(p ¼ 0.3). Both groups had an increase in serum calcium levels Figure 1. Serum level of PICP in both groups at the start and at the
over the course of the study. The relative percent increase was end of the trial.

Table 1 Characteristics of the study subjects (n ¼ 30)

Control (n ¼ 15) Intervention (n ¼ 15) p-value

Gestational age (weeks) 32.13±1.85 31.53±1.46 0.328

Birth weight (grams) 1429.33±148.51 1465.67±132.83 0.430
Sex (% male) 53.3 53.3 0.984
Caloric intake (kcal/kg/day) 144.77±19.24 135.00±29.30 0.290
Calcium intake (mg/kg/day) 186.00±17.12 169.60±52.57 0.260
Protein intake (g/kg/day) 3.035±0.892 3.650±0.970 0.081

Table 2 Laboratory values at enrolment and at conclusion of the study

Control Intervention

Start End Start End

Serum calcium (mg/dl) 9.1±1.6 9.4±1.0 8.5±1.4 10.1±1.2

Alkaline phosphatase (U/l) 395.9±102.6 449.3±140.1 410.2±74.4 377.6±95.2
PICP (ng/ml) 82.3±8.5 68.8±14.7 62.5±13.8 73.8±12.9
Pyd (ng/ml) 447.7±282.8 744.9±373.6 445.7±266.6 716.8±301.8
PTH 33.0±15.8 25.7±20.6 8.3±3.3 15.1±3.5

Journal of Perinatology 2004; 24:305–309 307

 Aly et al. Bone Mineralization in Infants

1200 efficiency and increases growth hormone levels in humans, it is

plausible to expect physical activity to push the balance in favor of
bone formation.17,18 Tactile stimulation, provided by infant
800
massage, can also stimulate growth even without physical
Urine Pyd

Control
Intervention activity.5,6 Lack of stimulation due to maternal deprivation can
400 decrease growth hormone levels and impair metabolic efficiency.
Markers of tissue growth such as ornithine decarboxylase are
decreased with maternal deprivation.19,20
0
Osteopenia of prematurity occurs as a direct result of
Start of Study End of Study
insufficient bone deposition or increased resorption of organic bone
Time
matrix. Several factors can produce such imbalance of bone
Figure 2. Urinary Pyd in both groups at the start and at the end of the homeostasis in the ill premature infant. Medications often used,
trial. such as diuretics and corticosteroids, adversely contribute to this
imbalance. Other behavioral and environmental practices in the
50 nursery such as swaddling and immobilization can certainly
deprive premature infants from mechanical stimulation. Had they
40
continued their term in the uterus, persistent kicking against the
uterine wall would have provided a form of exercise to these
PTH (pg/mL)

30
Control
fetuses.
Intervention
20 Several methods have been used to evaluate bone growth and
mineralization. Critiques developed to the precision and clinical
10
applicability of each individual method. For example, markers of
0
bone mineralization such as PICP and Pyd are currently reported
Start of Study End of Study with a considerably wide distribution that can indeed jeopardize
Time claims of their precision. Anthropomorphic measurements, such as
Figure 3. Serum PTH level in both groups at the start and at the end forearm length, can generally vary when different individuals
of the trial. perform the measurement. Recent studies have looked at bone
speed of sound, single photon absorptiometry (SPA) and peripheral
dual-energy X-ray (pDEXA) as measurements of bone strength.
While these methods may provide useful information, they can be
Urine Pyd Pyd concentration was similar between the two groups affected not only by physical density but also by bone size and
at study entry (p ¼ 0.984). In both groups, Pyd levels increased geometry.7,9 For instance, in the first six months of life, total bone
over the study period (Figure 2). There was no difference between density of the femoral shaft decreases by 30% due to changes in
the two groups in the amount of percent increase. bone geometry, whereas bone strength increases by threefold and
absolute mineral content also increases during this time period.10
PTH PTH levels were higher in Group I (33.0±15.8 pg/ml). than Serum PICP concentration decreased in the control group.
in Group II (8.3±3.3 pg/ml) at the start of the study (p<0.001). Current standard NICU practices such as swaddling and
Serum PTH decreased in the control group while it increased in the immobilization contribute to the decreased bone formation
activity group (Figure 3). The percent change was significantly observed in the hospitalized infants. This finding was previously
different between the two groups (p<0.001). reported by Moyer-Mileur et al.7,8 in two separate studies. In both of
Differences between groups remained significant after controlling their reports, infants in the control group had decreased PICP
for gestational age using a regression analysis model. concentration, while infants in the physical activity group
maintained constant levels of PICP. Nevertheless, PICP
concentration in our study increased in the intervention group. Of
DISCUSSION note, infants of the intervention group in our study received a
This study showed that bone mineralization improved when combined massage and physical activity. We speculate that a
premature infants experienced a combined protocol for physical strategy of combined physical activity and massage could produce a
activity and massage. Bone growth and metabolism is a complex synergic effect on bone mineralization. This may explain why
process that involves deposition of bone matrix with incorporation massage therapy or physical activity individually could not increase
of mineral content, which is balanced with enzymatic resorption PICP concentration, while when combined together a significant
and remodeling. Considering the fact that it enhances metabolic increase was observed. We are currently conducting a randomized

308 Journal of Perinatology 2004; 24:305–309

Bone Mineralization in Infants Aly et al.

controlled clinical trial in order to evaluate the effects of massage 3. Helm I, Landin LA, Nilsson BE. Bone mineral content in preterm infants at
therapy alone and in combination with physical activity. age 4 to 16. Acta Paediatr Scand 1985;74:264–7.
We were surprised to notice that PICP levels were different at the 4. Yeh JK, Liu CC, Aloia JF. Effects of activity and immobilization on bone
start of this randomized double-blinded trial. Gestational age is formation and resorption in young rats. Am J Physiol 1993;264:E182–9.
known to correlate with PICP levels.16 The difference among the 5. Scafidi FA, Field TM. Massage stimulates growth in preterm infants: a
replication. Infant Behav Dev 1990;13:167–88.
two groups was maintained even after controlling for gestational
6. Scafidi F, Field T, Schanberg S. Factors that predict which preterm infants
age using a logistic regression model. Further studies are
benefit most from massage therapy. J Dev Behav Pediatr 1993;14:176–80.
recommended to explore other factors that affect PICP levels and 7. Moyer- Mileur L, Luetkemeier M, Boomer L, Chan GM. Effect of physical
could have explained the differences observed between the two activity on bone mineralization in premature infants. J Pediatr
groups of this study. Urinary Pyd did not differ among groups, 1995;127:620–5.
suggesting that bone resorption and remodeling are active 8. Moyer-Mileur LJ, Brunstetter V, McNaught TP, et al. Daily physical activity
processes that occur equally in all infants. program increases bone mineralization and growth in preterm very low
It is of interest that PTH levels increased significantly in infants birth weight infants. Pediatrics 2000;106:1088–92.
in the intervention group. Intuitively, this may suggest a higher 9. Litamanovitz I, Dolfin T, Friedland O, et al. Early physical activity
rate of bone resorption since PTH acts to mobilize calcium from intervention prevents decrease of bone strength in very low birth weight
bone in response to low serum calcium levels. However, PTH has infants. Pediatrics 2003;112:15–9.
been shown to enhance both the process of bone resorption and 10. Rauch F, Schoenau E. Skeletal development in premature infants: a review
of bone physiology beyond nutritional aspects. Arch Dis Child 2002;86:
bone formation, with more stimulation of the latter. In fact, PTH
F82–85.
administration has shown promise in the treatment of osteoporosis
11. Weiger K, Wollman HA, Ranke MB, Speer CP. Plasma concentrations of
and is associated with increased bone mass.21 The increase in PTH carboxyterminal propeptide of type I procollagen (PICP) in preterm
levels may be a normal physiologic response to activity since PTH neonates from birth to term. Pediatr Res 2000;48:104–8.
activity increases with physical activity in adults.22 It is difficult to 12. Crofton PM, Shrivastava A, Wade JC, et al. Bone and collagen markers in
separate this increase in PTH level from the increase in serum preterm infants: relationship with growth and bone mineral content over
calcium since higher PTH activity works to increase serum the first 10 weeks of life. Pediatr Res 1999;46:581–7.
calcium. The increase in serum calcium after activity may also be 13. Tsukahara H, Miura M, Hori C, et al. Urinary excretion of pyridium
attributable to lactic acidosis and/or reduction in plasma volume.23 crosslinks of collagen in infancy. Metabolism 1996;45:510–4.
It is important to recognize that infants in this study were 14. Ballard JL, Khoury JC, Wedig K. Assessment of gestational age using Ballard
provided with optimal nutritional intake. Growth and bone method. J Pediatr 1991;199:417.
formation are, of course, linked to adequate calories, protein and 15. SAS Institute Inc. SAS/STAT Software: Changes and enhancements through
release 6.12. Cary, NC; 1997, 831–43.
calcium in the diet of the preterm neonate. Otherwise, physical
16. Ogueh O, Khastgir G, Studd J, et al. The relationship of fetal serum markers
activity in the absence of adequate calcium intake decreases bone
of bone metabolism to gestational age. Early Hum Dev 1998;51:109–12.
mineral content.24 Thus, it is essential to provide sufficient 17. Torun B, Schutz Y, Viteri F, et al. Growth, body composition and heart rate/
nutritional intake in order for physical activity to produce favorable VO2 during the nutritional recovery of children in two different physical
effects on bone growth. activity levels. Bibl Nutr Dieta 1979;27:55–6.
18. VanWyk JJ, Underwood LE. Growth hormone, somatomedins and growth
failure. Hosp Pract 1978;68:57.
CONCLUSION 19. Schanberg SM, Evoniuk G, Kuhn CM. Tactile and nutritional aspects of
maternal care: specific regulators of neuroendocrine function and cellular
Physical activity combined with infant massage stimulates bone
development. Proc Soc Exp Biol Med 1984;175:135.
formation in premature infants as evidenced by an increase in
20. Field TM, Schanberg SM, Scafidi F, et al. Tactile/kinesthetic stimulation
PICP, a biochemical marker of bone formation, and an increase in effects on preterm neonates. Pediatrics 1986;77:654–8.
PTH activity, which may further stimulate bone growth and 21. Kurland ES, Cosman F, McMahon DJ, et al. Parathyroid hormone as therapy
mineralization. for idiopathic osteoporosis in men: effects on bone mineral density and bone
markers. J Clin Endocrinol Metab 2000;85:3069–76.
22. Ljunghall S, Joborn H, Roxin LE, et al. Increase in serum parathyroid
References hormone levels after prolonged activity. Med Sci Sport Exerc 1988;20:122–5.
1. Abrams SA, Schandler RJ, Tsang RC, et al. Bone mineralization in former 23. Ljunghall S. Effects of physical activity on serum calcium and parathyroid
very low birth weight infants fed either human milk or commercial hormone. Eur J Clin Invest 1984;14:469–73.
formula: one year follow up observation. J Pediatr 1989;144:1041–44. 24. Specker BL, Mulligan L, Ho M. Longitudinal study of calcium intake,
2. American Academy of Pediatrics, Committee on Nutrition. Nutritional needs physical activity and bone mineral content in infants 6–18 months of age.
of low birth weight infants. Pediatrics 1985;75:976–86. J Bone Miner Res 1999;14:569–76.

Journal of Perinatology 2004; 24:305–309 309