Review of Randomized Controlled Trials of Massage in Preterm Infants

Preterm birth affects about 10% of infants born in the United States. Massage therapy is being used in some neonatal intensive care units for its potential beneficial effects on preterm infants. This article reviews published randomized controlled trials on the effects of massage in preterm infants. Most studies evaluating the effect of massage in weight gain in premature infants suggest a positive effect on weight gain. Increase in vagal tone has been reported in infants who receive massage and has been suggested as a possible mechanism for improved weight gain. More studies are needed on the underlying mechanisms of the effects of massage therapy on weight gain in preterm infants. While some trials suggest improvements in developmental scores, decreased stress behavior, positive effects on immune system, improved pain tolerance and earlier discharge from the hospital, the number of such studies is small and further evidence is needed. Further studies, including randomized controlled trials, are needed on the effects of massage in preterm infants.


Introduction
Preterm birth (birth at <37 weeks of gestation) affects about 10% of infants born in the USA [1]. Premature infants often spend weeks or months in intensive care unit due to immaturity and need for intensive medical care, with hospital stay often prolonged by feeding immaturity and slow weight gain. Massage therapy is being used in some neonatal intensive care units for its possible beneficial effects with minimal side effects. This article reviews published randomized controlled trials on the effects of massage in preterm infants.
A PubMed search was performed for randomized controlled trials (RCTs) of massage in preterm infants. RCTs in English and indexed in PubMed by January 2017 were included. Studies assessing the effects of massage on caregivers only were excluded. A total of 34 randomized controlled trials on the effects of massage in preterm infants are reviewed .
In the following sections the outcomes of these studies as well as limitations are briefly reviewed. For number of study participants, gestational age, age at the time of study entry, and duration of massage intervention see Table 1.
Most studies have demonstrated a significantly greater daily or overall weight gain during the study period in massage group compared to control group [2,9,10,12,17,19,21,26,28,30,[33][34][35] while some studies did not show a statistically significant difference in weight gain between groups [5,8,15,16,18,29,31]. Massaro et al. [16] evaluated the effects of massage or massage + kinesthetic stimulation on weight gain and length of hospital stay in a study of 59 infants who were randomized to massage only (n = 19), massage and kinesthetic stimulation (n = 20) or control group (n = 20). After controlling for covariates (birth weight, gestational age, caloric intake, etc.) the infants whose birth weight was >1000 g benefited from massage with a significantly higher average daily weight gain with more pronounced effects in infants who received massage + kinesthetic stimulation (p = 0.012) [16]. To study the possible contribution to weight gain of medium chain triglyceride (MCT) oil used in massage therapy (via absorption of oil through the skin) Saeidi et al. compared the effects of massage with MCT oil to massage without oil and controls (with no massage) and demonstrated a significanly higher weight gain during the 7-day study in the group that received massage with MCT oil compared to massage only (p = 0.002) and controls (p = 0.000) [2]. The mean weight gain during the study period in the MCT oil massage group was 105 ± 1.3 grams compared to a gain of 52 ± 0.1 grams in the massage only group and loss of 54 ± 1.3 grams in the control group [2].

Caloric Intake and Expenditure
Some studies have evaluated the effects of massage on caloric intake [5,8,17,19,30,[33][34][35] and its possible connection to weight gain in preterm infants. Most studies did not show a difference in caloric intake between infants who received massage compared controls despite statistically significant higher weight gains in infants who received massage [8,17,19,30,33,34]. A study by Diego et al. showed increased calorie intake in infants who received kinesthetic stimulation compared to those who received tactile stimulation only, no difference in weight gain between the two groups [5]. Scafidi et al. evaluated factors that predict higher weight gain in infants who receive massage and a higher caloric intake was one of the factors that predicted weight gain [35]. One study evaluated energy expenditure in preterm infants, which was reported to be lower after a 5-day study period in infants who received massage compared to controls [22] suggesting that a decrease in energy expenditure may be in part responsible for the enhanced growth caused by massage therapy.     Serum triglyceride values were significantly higher after massage in all groups compared to baseline (coconut oil p < 0.001, safflower oil p < 0.001, controls p < 0.05). There was a significant rise in linoleic and arachidonic acid in the safflower oil group (p < 0.001) and saturated fats in coconut oil group (p < 0.001). Massage group had greater average daily weight gain (p < 0.01) than controls. Factors that predicted higher weight gain were history of obstetric complications, higher caloric intake and longer stay in intermediate nursery.

Vagal Tone
Increased vagal activity induced by massage has been suggested as one potential mechanism for higher weight gain in infants who receive massage therapy. Vagal activity in most studies is assessed by measuring heart rate variability (HRV) with increased high frequency (HF) variability indicating higher vagal tone [37]. All studies that evaluated vagal activity demonstrated an increase in vagal tone in infants who receive massage [5][6][7]19,24,26,29,30]. However, in two studies, the increased vagal effect was more pronounced in male infants [6] or was demonstrated only in males [7]. In a study by Diego et al. comparing the effects of tactile to kinesthetic stimulation (the two components of TKS massage [36]), an increase in vagal tone measured by HRV was only noted in in the tactile group in group-by-time analysis (p = 0.01), whereas a decreased vagal tone was noted in infants receiving kinesthetic range of motion excercises only [5].

Gastric Motility
Massage-induced increases in vagal activity may lead to increased gastric motility, which has also been suggested as a possible mechanism for increased weight gain in infants who receive massage therapy. Gastric motility or number of stools have been assessed in three studies [3,24,30]. In infants who received massage therapy, significantly increased gastric motility during and immediately after massage and decreased tachygastria has been reported [24,30]. This increase in gastric motility was significantly correlated with increased weight gain and vagal tone [24,30]. During a 4-day study which evaluated the effects of massage on transcutaneous bilirubin levels and stool frequency, a significantly increased stool frequency was reported in the massage group compared to controls during the entire study period (day 1: p = 0.001, day 2: p = 0.02, day 3: p = 0.01 and day 4: p = 0.04) [3].

Immunological Effects
Two randomized controlled studies have reported on the effects of massage in immunological parameters and incidence of infections in preterm infants [12,18]. Absolute number of natural killer (NK) cells, white blood cells, B and T cells, and T-cell subsets were not statistically different between the groups. Number of infections did not differ between the groups. While the mean absolute NK cell number was not statistically different between the massage and control group, the adjusted mean NK cell cytotoxicity was higher in the massage group (p = 0.05) [12]. Mendes et al. reported on significantly lower incidence of late-onset sepsis in massage group (p = 0.005) [18]. The incidence of late-onset sepsis was 38.3% (18/47 infants) in controls compared to 10.9% (5/46 infants) in massage group. No serum immunological markers were reported in this latter study.

Bone Metabolism
Preterm infants have increased morbidity from osteopenia [38]. Physical activity stimulates bone formation. Two studies have evaluated the effects of massage therapy with physical activity on bone metabolism in preterm infants (Table 1) [11,32]. Physical activity was defined either as kinesthetic movement part of TKS massage [11] or a similar daily range of motion exercise, with gentle compression and extension/flexion to both upper and lower extremities [32]. Aly et al. reported an increase in serum type I collagen C-terminal propeptide (PICP, marker of bone formation) in the massage group (p < 0.01) compared to baseline, while PICP decreased in controls group (p < 0.01). This change differed significantly between the groups (p = 0.0001), suggesting increased bone formation in massage group. Furthermore, serum parathyroid hormone (PTH) level increased in the massage group while it decreased in the control group (p < 0.001).

Behavior, Sleep and Neurodevelopment
Yates et al. studied the effects of TKS massage on sleep efficiency in preterm infants in a crossover study [4] and found no significant difference between groups for sleep efficiency (p = 0.13). More infants were reported to sleep on the non-massage day (p = 0.026) [4]. Several studies have assessed the effects of massage on behavior and/or neurodevelopment of preterm infants [14,15,23,26,28,29,31,34,35]. Most studies have reported on immediate behavioral effects of massage [15,23,26,28,29,31,34,35] and only one randomized controlled trial has reported on long term neurodevelopmental outcome [14]. Massage groups were reported to have significantly less stress-related behaviors (crying, fidgeting) than controls in three studies [23,26,34], no difference between the groups in two [28,31], and one study reported increased crying and fidgeting in infants who received massage [29]. Ho et al. reported on a higher gain in motor performance score (Test of Infant Motor Performance, TIMP) in very low birth weight (VLBW) infants who received massage and who had a below-average pre-study score compared to a control group who received light touch (p = 0.043) suggesting that massage may have positive effects in motor outcomes in a subgroup of VLBW infants who have low motor performance [15]. Procianoy et al. compared neurodevelopmental outcomes (Psychomotor Development Index, PDI and Mental Development Index, MDI) at 2 years of age between preterm infants who had received massage intervention during hospital stay and controls [14]. Infants who had been randomly assigned to massage therapy group or control group during their hospital stay both received skin-to-skin care. Growth and neurodevelopmental outcome were evaluated in both groups at 2 years corrected age. Growth at 2 years corrected age did not differ between the groups. The massage group had slightly but not significantly higher PDI score (p = 0.072) and significantly higher MDI scores (p = 0.035) than controls [14].

Length of Hospital Stay
Length of hospital stay (LOS) was reported to be similar in the massage group and controls in one study [12], and in one there was a significant difference between groups [16] but this difference disappeared after controlling for gestational age, gender, sepsis and birth weight. Two studies have reported shorter duration of hospitalization in infants who received massage compared to controls [17,18]. A study by Gonzalez et al. evaluating the effects of Vimala massage in preterm infants reported a shorter hospital stay in infants of massage group (15.36 ± 5.41 days) compared to controls (19.33 ± 7.92 days (p = 0.03) [17]. A study by Mendez et al. on VLBW infants reported that those who received massage had a 1.85 times higher (95% confidence interval (CI): 1.09 to 3.13; p = 0.023) probability of earlier hospital discharge than control group [18].

Serum Markers
Levels of serum markers of growth and metabolism such as IGF-1 [8,19], adiponectin [8], serum triglycerides [9,27,31] have been studied in small randomized controlled trials. Moyer-Mileur et al. studied IGF-1, leptin, and adiponectin levels in infants who received massage and controls and correlated them with skin fold thickness, weight gain and ponderal index (PI) [8]. Serum leptin levels correlated significantly with weight gain, PI, triceps skin-fold thickness and mid-thigh skin-fold thickness [8]. Serum adiponectin level correlated with PI. There was no difference in weight gain however in this study between massage and control groups. In a study by Field et al. the massage group had a greater increase in weight gain (p = 0.02), and a greater increase in both insulin (p = 0.001) and IGF-1 (p = 0.05) levels [19] and weight gain correlated with increased insulin (p = 0.05) and IGF-1 (p = 0.02) levels [19]. Three randomized controlled trials have evaluated the effects of oil massage on serum triglycerides [9,27,31]. The primary objective of Solanki et al. was to assess transcutaneous absorption of massage oils by assessing serum lipid profiles in infants assigned to a massage without oil, or with safflower oil (rich in essential fatty acids, EFAs) or coconut oil (rich in saturated fats and medium chain triglycerides, MCTs) [27]. No weight gain was measured in this study. Serum triglyceride levels were significantly higher after massage compared to baseline in all groups (coconut oil p < 0.001, safflower oil p < 0.001, controls p < 0.05) though the rise was significantly higher in oil groups compared to controls (p < 0.05). In infants randomized to massage with safflower oil, a significant rise was seen in the essential fatty acids linoleic and arachidonic acid (p < 0.001) after massage. In those who received a massage with coconut oil an increase in serum saturated fats was reported (p < 0.001). In the control group who received a massage without oil a small but significant increase in serum linoleic (p < 0.05) and total saturated fats was seen (p < 0.05) [27]. The other two studies that assessed serum triglyceride levels used sunflower oil in the massage group evaluated serum triglyceride levels as a secondary outcome and reported no significant difference between massage group and controls [9,31].

Bilirubin levels, markers of brain maturation, pain
Basiri-Moghadam et al. reported significantly lower transcutaneous bilirubin levels (p = 0.003) in the massage group compared to controls after four days of massage [3]. Guzzetta et al. reported differences in EEG spectral power (an index of brain maturation) in massaged infants compared to controls suggesting a process of brain maturation in those who received massage [13]. Jain et al. reported in a cross-over study higher scores in Neonatal Infant Pain Scale (primary outcome) (p < 0.001) and heart rate (p = 0.03) in infants who had a blood sampling via a heel stick without a preceding 2-min massage of the ipsilateral leg compared to infants who had a massage prior to heel stick [25]. Respiratory rate, oxygen saturation and serum cortisol levels did not differ [25].

Discussion
Preterm birth (birth at <37 weeks of gestational age) is the leading cause of neonatal morbidity and mortality in the United States [39,40] and affects 10% of infants born in the USA [1]. Slow weight gain and poor oral-motor ability to feed are common factors delaying discharge from hospital in infants born prematurely. Interventions that improve either weight gain or oral-motor function may lead to shorter hospital stay and cost savings. Improved weight gain is the most commonly and consistently reported effect of massage in preterm infants with significantly greater daily or overall weight gain found in infants who received massage compared to controls [2,9,10,12,17,19,21,26,28,30,[33][34][35]. While one study reported increased gain in an infant motor performance score in VLBW infants who received massage suggesting positive effects on motor development [15], oral-motor coordination and ability to feed was not evaluated. To date there are no randomized controlled trials evaluating the effects of massage on oral-motor coordination or feeding ability offering an area of future studies.
A recent meta-analysis by Wang et al. concluded that massage improved daily weight gain by 5.32 g (95% CI 4.15, 6.49 g, p < 0.00001) and reduced length of hospital stay by 4.41 days (95% CI 2.81, 6.02 days, p < 0.00001) [41]. A Cochrane systematic review by Vickers et al. also concluded that massage improved daily weight gain by 5.1 g (95% CI 3.5, 6.7 g) and appeared to reduce length of hospital stay by 4.5 days (95% CI 2.4, 6.5 days) [42], consistent with meta-analysis by Wang et al. [41]. Furthermore, a review by Field et al also suggested that infant massage leads to 3-6 days shorter length of hospital stay and, consequently, significant cost-savings up to $10,000 per infant [43].
The mechanism by which massage improves weight gain in preterm infants is not well understood. One suggested mechanism is that massage leads to higher consumption of calories. However, studies have not shown a difference in caloric intake between infants who received massage compared controls [8,17,19,30,33,34]. Another possible mechanism is increased vagal tone which may promote food absorption and anabolism. Increase in vagal activity has indeed been consistently reported in infants who receive massage in all randomized controlled trials that evaluated vagal tone [5][6][7]19,24,26,29,30]. Levels of serum anabolic hormones insulin and IGF-1 have been reported to be higher in infants who had received massage and correlated with weight gain [19] suggesting a possible anabolic effect. Field et al. postulate in more detail potential mechanisms for improved weight gain in preterm infants who received massage, including increased vagal tone and IGF-1 levels [44].
Furthermore, increased weight gain in preterm infants who receive massage with oil may be due to transcutaneous absorption of oils and contribution to caloric intake. While Solanki et al. reported a significant increase in serum essential fatty acids and saturated fats in infants massaged with safflower and coconut oil, respectively, their study did not assess weight gain [27]. Thus, the possible effects of transcutaneous absorption of oil in weight gain could not be evaluated. Study by Saeidi et al. did demonstrate a significantly higher weight gain in infants who received massage with MCT oil compared to massage only or controls [2] suggesting that MCT oil, possibly via transcutaneous absorption, may partly contribute to weight gain.
Improved weight gain and decreased length of hospital stay are important measures on the effects of massage in preterm infants and may lead to significantly lower health care costs [43]. Yet little is known about the long-term effects of early massage on infant development. Only one small study (n = 73) has assessed the long-term effects of massage given during neonatal hospital stay and compared neurodevelopmental scores between infants who received massage therapy to controls [14] and reported on a slightly higher PDI score and significantly higher MDI scores in the massage group, suggesting massage may induce long-term effects in neural modeling and neurodevelopment.
Overall, several studies as well as meta-analyses have suggested beneficial effects of massage in preterm infants with regards to improved weight gain and shortening the length of hospital, the latter of which has potential for significant cost-savings. Limitations and challenges of current randomized controlled trials on the effects of massage in preterm infants include small sample size in most studies, lack of studies in extremely premature infants, and lack of studies in infants with congenital anomalies or congenital heart disease who often have poor weight gain. The provider of the massage is also not consistent between studies. Massage has been variably provided by licensed massage therapists, trained nurses, or by a mother trained in massage technique. Although one study assessed the effects of massage in weight gain in infants massaged by either the mother or a professional gained and reported more weight gain in both groups compared to controls (p = 0.03) suggesting that different providers provide equally effective massage. There is limited information available on side effects of massage therapy with most of the studies not reporting side effects or quoting massage as "safe" or "relatively safe". However, given lack of reported side effects it is likely safe to conclude that massage is "relatively safe" in the infant groups in which it has been studied. There is not enough evidence to recommend massage for all preterm infants but it may be beneficial in selected clinically-stable preterm infants with slow weight gain.

Conclusions
In summary, randomized controlled trials on the effects of massage in preterm infants suggest improved weight gain and shortened length of hospital stay. Improved weight gain may be mediated through increase in vagal tone but more studies are needed on the underlying mechanisms of the effects of massage. Further randomized controlled studies are needed on the effects of massage on outcomes such as neurodevelopment, stress behavior, immune system, and pain tolerance.

Conflicts of Interest:
The author declares no conflict of interest.