Effect of Contingency Paradigm–Based Interventions on Developmental Outcomes in Young Infants: A Systematic Review

Abstract

Purpose: 

The purpose of this systematic review was to identify controlled trials evaluating the efficacy of contingency paradigm–based interventions to improve feeding, motor, or cognitive outcomes during the first year of life.

Summary of Key Points: 

Seventeen studies, including 10 randomized controlled trials, incorporating contingency paradigm–based interventions were identified. Three of 3 trials reported improvements in nutritive sucking using pacifier-activated lullaby in preterm infants before term age. Seven of 12 trials reported improvements in reaching, manual exploration, and kicking behaviors in term and preterm infants; and 6 of 10 trials reported gains in early cognition using sticky mittens and contingent toys.

Conclusion and Recommendations for Clinical Practice: 

Contingency paradigm–based interventions can improve feeding outcomes in the neonatal intensive care unit in very preterm infants, and increase reaching, and perceptual-cognitive behaviors in term infants. Future research is needed to establish contingency paradigms as an effective early intervention strategy.

What This Adds to the Evidence: 

This review synthesizes a body of literature on contingency paradigm-based interventions and highlights its potential paradigm–based interventions to improve developmental outcomes in infants.

INTRODUCTION

The period from birth to 12 months or infancy is a crucial period in development as it marks the onset of rapid growth in the physical, social, and neurocognitive domains.¹ These domains dynamically interact with each other such that a developmental change in any one domain causes a “cumulative and cascading” effect in other domains.² For instance, when infants learn to sit independently, they use their hands more frequently for communicative gestures with caregivers, supporting their language development.³ Likewise, the onset of independent locomotion allows infants to take control of play in terms of toy choices, supporting more mature social interactions with caregivers.⁴ Emerging evidence supports that pediatric therapeutic interventions (physical, occupational, and speech therapy) starting from birth can support development in multiple domains, ranging from feeding and neurobehavioral domains in the neonatal intensive care unit (NICU)⁵ to motor and cognitive domains in infancy.^6,7 However, the collective evidence for effective early intervention techniques, especially in children younger than 2 years, is heterogeneous and limited at best.⁸

The need for evidence-informed interventions in infancy calls for a more robust approach that explores complementary fields of development to identify theoretical models, measurement, and intervention approaches.^9,10 The developmental psychology literature is abundant with task-specific paradigms that examine learning and memory in infants, one of them being the “mobile paradigm.” In this paradigm, the infant's leg is tethered to an overhead mobile, such that when the infant kicks, the mobile moves. The original paradigm is based on the principle of operant conditioning, which suggests that behaviors that are reinforced are repeated, the reinforcer being the mobile movement.¹¹ However, a recent review proposed embodied cognition as a mechanism responsible for learning in contingency paradigms.¹² They stated that learning a contingency is not merely a result of reinforcement but rather occurs when the infant identifies the relationship between their self-initiated actions and its effect on the environment.¹² This form of contingency learning is dependent on variability in practice,¹³ age,¹⁴ and environmental constraints.¹⁵ In the last 2 decades, the mobile paradigm has been adapted to develop new contingency paradigms to evaluate motor-learning abilities¹³ in both term and preterm infants and to study motor adaptations to reinforcement.^14–16

Although these paradigms are typically used for evaluation purposes, they include critical elements required for effective early intervention.^17,18 First, these paradigms are inherently designed to be task-specific and require practice through trial and error to learn and retain the contingency. Such practice mimics typical motor development, during which infants repeatedly perform variable movements until they select the most successful strategy for function.¹⁹ This could be specifically beneficial in infants with or at risk for delays since they lack the movement variability necessary for learning functional skills.²⁰ Second, contingency paradigms provide opportunities for a more demanding practice while maintaining active infant participation. This is crucial as conventional early intervention services are not designed to be intensive.²¹ Moreover, repeated practice not only strengthens the developing musculature but also optimizes neuroplasticity.¹⁸ Finally, these paradigms can easily be adapted to home environment and may provide opportunities for caregiver involvement in therapy. Caregiver-delivered interventions are proven to be more effective in very young infants born preterm.⁵

Thus, the strong theoretical background, ease of application, and potential to improve developmental outcomes make contingency paradigms a promising new intervention strategy for infants. To the best of our knowledge, there is no review assimilating the studies and summarizing the effectiveness of contingency paradigms as an intervention, particularly as it relates to physical therapy and infant development. Thus, the aim of this systematic review was to quantify the efficacy of interventions based on contingency paradigms in infancy on feeding, motor, and cognitive outcomes. These specific outcome domains were selected as they form the primary focus of physical therapy practice for neonates and infants and capture therapy outcomes from birth to infancy.²² The contingency paradigms for this review included “a contingency learning component, active infant movement and a practice through trial and error to achieve a clearly defined outcome.” The interventions are abbreviated as CPBI, that is, contingency paradigm–based interventions throughout this article.

METHODS

Protocol and Registration

This systematic review was prospectively registered with PROSPERO and can be accessed at (https://www.crd.york.ac.uk/prospero/; identifier CRD42020163140). Review reporting is in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.²³

Eligibility Criteria

Full-text articles in English, published, or in press before our search, that is, January 2021; Population: Term or preterm infants with a mean age of 12 months and younger and no restriction on medical diagnosis; Intervention: Interventions incorporating task-specific activities based on contingency learning principles as either the primary component or one of the key components. Comparator: Noncontingency learning–based paradigms, usual care, or no treatment; Outcomes: feeding, motor, and cognitive outcomes. Outcomes had to be assessed pre- and postintervention (ranging from last day of intervention to 24 months post). The outcome had to measure “transfer of learning,” that is, application of the acquired skill to a new situation. For instance, an infant is trained to reach using a teether toy during therapy session and then uses this learned new skill to reach for a different toy during play; Studies: Randomized controlled trials (RCTs), non-RCTs (NRCTs), quasi–randomized controlled trials (QRCTs), multiarm pre-/postdesigns, and randomized crossover trials. Articles were excluded if mean age was more than 12 months at the time of intervention; interventions principles were based on other forms of associative learning such as classical conditioning; studies using principles of associative learning for evaluative or assessment purpose only; studies reporting only physiological outcomes (example: oxygen saturation, weight gain, heart rate, electroencephalography); abstracts, reviews, or conceptual papers; and study designs with no control group (single arm pre-/post, case series, or case study).

Search Strategy

A comprehensive literature search of articles from the earliest record to January 2021 was performed using the following electronic databases: PubMed, Cochrane, CINAHL, Web of Science, Psychology and Behavioral Sciences Collection, PsycINFO, and ERIC. The final search terms for PubMed database were as follows: infan*, “Association Learning“[Mesh]”, Conditioning, Operant”[Mesh], “Token Economy”[Mesh], “Reinforcement, Psychology”[Mesh], “contingent reinforcement”, “contingency learning”, “associative learning”, “learn*”, “motor learning.” These search terms were adapted according to the other database's command language, and no date or language restrictions were applied. Filters for age (infants or 0-12 months), population, and article type were applied to refine the search when necessary. Secondary search was performed using relevant references of the included articles or by searching for included interventions by name and author. Additional gray literature was searched using the following resources: Google Scholar, PsycEXTRA, and Proquest dissertation and thesis (see Supplemental Digital Content 1, available at: https://links.lww.com/PPT/A350).

Study Selection

Search results were imported in a Web-based systematic review management software called “Covidence” (https://www.covidence.org). After duplicates were removed, 2 reviewers independently performed the title and abstract screening based on the defined exclusion criteria. If the abstract was not excluded, full text of the selected article was retrieved and the screening process was repeated reviewing the eligibility criteria. If there was a lack of agreement between the 2 reviewers on the inclusion of any of the articles, the third reviewer was consulted for the final decision.

Data Extraction

Two independent reviewers extracted the following data from the included articles: study setting, sponsorship source, corresponding author details, study design, eligibility criteria, participant characteristics, intervention groups, outcomes used, statistical results, and study conclusion. The “Template for Intervention Description and Replication” (TIDieR) checklist was used to extract additional intervention-specific information as follows: provider, setting, dosage, fidelity, and compliance assessment.²⁴ If no agreement was reached, the third reviewer was consulted.

Risk of Bias and Quality of Evidence Assessment in Individual Studies

The risk of bias (ROB) for all the NRCTs or QRCTs was assessed using the ROBINS-I (“Risk of Bias in Non-randomized Studies of Interventions”). This tool comprises a 3-stage assessment and 7 specific domains. Each study was given an overall ROB judgment as having no information, low, moderate, serious, and critical ROB.²⁵ For the RCTs, ROB was assessed using Cochrane ROB 2.0 and it has 5 domains. The overall judgment for ROB can be low, high, or show some concerns.²⁶ Quality of evidence for each study was identified using the Sackett's level of evidence for group research designs (see Supplemental Digital Content 2, available at: https://links.lww.com/PPT/A351).²⁷

Data Synthesis

Because of the heterogeneity in CPBI and measured outcomes, a meta-analysis could not be completed. A narrative synthesis of results is presented, and the effectiveness of interventions is reported using P values at a level of significance of ≤.05.

RESULTS

Study Selection

The final search located 6025 studies. After duplicate removal, 4523 studies remained and were assessed for title and abstract screening and 4378 studies were excluded. The full text of 145 studies was examined and 125 studies were excluded as they did not meet the eligibility criteria. A total of 20 studies were identified for inclusion in the review.^28–47 Of the 20 studies, 6 studies shared a common data set^{32,33,37,38,43,44} and were combined to give a total of 17 unique studies (Figure).

Fig.: PRISMA flowchart of article selection. This figure is available in color online (www.pedpt.com).

Characteristics of Included Studies

A summary of individual study characteristics is shown in Table 1. The TIDieR checklist for each study is shown in Supplemental Digital Content 3, available at: https://links.lww.com/PPT/A352. Description of outcome measures for each study is shown in Table 2 and results for intervention effectiveness are summarized in Table 3.

TABLE 1 - Characteristics of Included Studies

			Participant Details
Study Details					Intervention Details
Study	Design	Sample Size	Gestational Age, wk	Age at Treatment Onset	Intervention Group	Control Group
Contingency learning–based interventions to improve feeding outcomes
Standley28 (2003)	RCT	32	24-36 (M = 31.8) Extremely to very preterm	36.1-wk AA	Pacifier activated lullaby for NNS	Usual care NNS
Standley et al29 (2010)	RCT	68	M = 29.2 Very preterm	32- to 36-wk AA	Pacifier-activated lullaby for NNS	Usual care NNS
Chorna et al30 (2014)	RCT	94	28-32 (M = 30) Very preterm	34.5-wk AA	Pacifier-activated mother's voice for NNS-3 trials and pacifier-activated mother's voice for NNS-1 trial	Usual care NNS
Contingency learning interventions to improve motor and cognitive outcomes
Needham et al31 (2002)	NRCT	32	>37 Full term	3-4 mo	Closed sticky mittens training	No training
Libertus and Needham32,33 (2010, 2011)	NRCT	58	>37 Full term	2-3 mo	Closed sticky mittens training	Nonsticky mittens observation training No treatment 3-mo-olds No treatment 5-mo-olds
Libertus et al34 (2016) Follow-up to Libertus and Needham32 (2010)	NRCT	40 (N = 25 were a subset of Libertus and Needham32 [2010])	>37 Full term	2-3 mo	Closed sticky mittens training	Nonsticky mittens observation training Control (no training)
Libertus and Needham35 (2014)	NRCT	72 (N = 36 were a subset of Libertus and Needham32 [2010])	>37 Full term	3 mo	Closed sticky mittens training	Nonsticky mittens observation training Encouragement experience Movement experience
Rakison and Krogh36 (2012)	RCT	40	>37 Full term	4.5 mo	Closed sticky mittens training	Nonsticky mittens movement training
Gerson and Woodward37,38 (2014)	NRCT	60	>37 Full term	3.5 mo	Closed sticky mittens training	Nonsticky mittens observation training
Williams et al39 (2015)	QRCT	37	>37 Full term	3 mo	Open sticky mittens	Nonsticky mittens movement training Control (no training)
Wiesen et al40 (2016)	RCT	32	>37 Full term	3 mo	Closed sticky mittens training	Nonsticky mittens observation training
Needham et al41 (2017)	RCT	38	>37 Full term	4-5 mo	Closed sticky mittens training	Nonsticky mittens observation training
Nascimento et al42 (2019)	RCT	24	34-36 mo (M = 36.08) Late preterm	4- to 4.5-mo AA	Open sticky mitten training	Social training
Heathcock et al43 (2008); Heathcock and Galloway44 (2009)	RCT	26	28-32 (M = 31) Very preterm	2-moAA	Contingent toys + Movement training	Social training
Needham et al45 (2014)	NRCT	38	>37 Full term	3 mo	Contingent toy training	Noncontingent passive observation
Campbell et al46 (2015)	RCT	13	23-32 (M = 27 wk) Extremely preterm with periventricular brain injury	2- to 4-mo AA	Contingent toy training	Control (no training)
Williams and Corbetta47 (2016)	RCT	35	>37 Full term	3 mo	Contingent toy training	Noncontingent toy movement training Control (no training)

Abbreviations: AA, adjusted age; M, mean; NNS, nonnutritive sucking; NRCT, non–randomized controlled trials; QRCT, quasi–randomized controlled trial; RCT, randomized controlled trials.

TABLE 2 - Description of Intervention Outcomes

Study	Outcome Name	Outcome Domain	Measurement Criteria	Purpose	Scoring Method	Psychometrics
Standley28 (2003)	Oral feeding rate	Feeding	Volume ingested through nipple feeding divided by minutes of feeding	Measures nutritive sucking abilities	Recorded from nurse's notes	None reported
Standley et al29 (2010)	Number of oral feeding days before discharge	Feeding	Number of nipple feeding days from the last day of nasogastric/oral-gastric feed to discharge date	Measures nutritive sucking abilities	Recorded from nurse's notes	None reported
	Length of gavage feeding days		Number of days from birth to the last day of nasogastric/oral-gastric feed	Measures days to transition to oral feeding. Indicates neural maturation
Chorna et al30 (2014)	Oral feeding rate	Feeding	Volume of nipple nutrition intake in milliliters divided by time in minutes for consumption	Measures nutritive sucking abilities	Recorded from nurse's notes	None reported
	Oral feeding volume		Amount of nipple nutrition measured in milliliters per kilogram per day
	Oral feeding frequency		Amount of nipple nutrition measured by the number of nipple feedings per day
	Days to oral feeding		Number of days to transition from gavage feeding to nipple feeding	Measures days to transition to oral feeding. Indicates neural maturation
Needham et al31 (2002)	Intentional swats	Motor	Number of times the infant attempted making manual contact with the toy while maintaining visual contact	Measures “intentionality” of swatting and early reaching abilities	Video coding using a joystick	One-third of sample recorded for interrater reliability: percent agreement was 87%-99% (mean = 93%)
	Visual attention to toy	Visual cognition	Percentage of time the infant visually explored the toy divided by total trial duration	Measures engagement in exploration
	Multimodal exploration	Motor and visual cognition	Adding the time infant spent mouthing and looking at the time and time spent switching between oral and visual exploration	Switching between 2 modalities requires more advanced cognitive and motor skills. Measures intersensory coordination
Libertus and Needham32,33 (2010, 2011)	Reaching	Motor	Proportion of time infant spent reaching toward the toy while looking at it, when the toy was within its reach divided by total trial duration	Measures early reaching abilities	Frame-by-frame video coding using StopFrameCoding software (Version 0.9)	Random sample was coded for interrater reliability: correlation (r = 0.90) Coding method was also correlated with previously used coding methods (r = 0.75-0.90)
	Grasping	Motor	Proportion of time infant spent touching the toy and bringing at least one corner of the toy off the testing table divided by total trial duration	Measures early object manipulation abilities
	Visual attention to toy	Visual cognition	Proportion of time infant spent looking at the toy when it was placed beyond its reach divided by total trial duration	Measures engagement in exploration
	Visual attention to experimenter	Visual cognition	Proportion of time infant spent looking at the experimenter when it was placed beyond its reach divided by total trial duration	Measures engagement in exploration and attention to task
	Shift in visual attention	Visual cognition	Proportion of time infant spent looking back and forth from the toy when it was placed in its hands divided by total trial duration	Measures the ability to identify relationship between object and environment
	Face preference	Social cognition	Difference in proportion of time infant spent looking at the photograph of a face (neutral expression of a White female) versus a toy (commercially available)	Measures infant's ability to identify itself as “intentional agent” and to discriminate social beings from inanimate objects	Remote eye tracker (Tobii 1750)	Not reported
	Face orienting	Social cognition	Face preference during the 1000-2000 ms time window following stimulus presentation
Libertus et al34 (2016) Follow-up to Libertus and Needham32 (2010)	Grasping	Motor	Duration of time the infant made a toy contact resulting in lifting at least 1 corner of the toy off the table, lifting internal parts of the toy, or by having fingers clearly curled around the object	Measures the effect of early reaching practice on later complex manual skills	Frame-by-frame video coding using StopFrameCoding software (Version 0.9)	30% of videos coded for interrater reliability: correlation analysis (r = 0.97)
	Toy rotations	Motor	Duration of time the infant touched the toy or its parts in such a way that the object turned around its own axis.	Measures the effect of early reaching practice on later complex manual skills
	Visual attention to toy	Visual cognition	Duration of time the infant spent looking at the toy	Measures engagement in exploration
	Distraction	Visual cognition	Duration of time the infant spent looking somewhere else	Measures engagement in exploration and attention to task
Libertus and Needham35 (2014)	Reaching and grasping combined	Motor	Duration of time the infant moved its arm toward the toy resulting in contact with the toy or partial/complete lift of the object off the table	Measures early object manipulation abilities	Frame-by-frame video coding using StopFrameCoding software (Version 0.9)	38% of videos coded for interrater reliability: correlation analysis (r = 0.88)
	Visual attention to toy	Motor	Frequency of the infant's looking episodes at the toy when the toy was placed in its hand	Motor, manual exploration
	Face preference	Social cognition	Difference in proportion of time infant spent looking at the photograph of a face (neutral expression of a White female) versus a toy (commercially available)	Measures infant's ability to identify itself as “intentional agent” and to discriminate social beings from inanimate objects	Remote eye tracker (Tobii 1750)	Not reported
Rakison and Krogh36 (2012)	Visual attention/looking at habituation experiment	Causal cognition	Duration of looking time at an animated event on a computer screen which resembled the contingent training (Training involved touching green ball with red mitten Test event displayed green ball being hit by a red ball)	Measures infant's ability to identify the goal structure in other's actions. Indicates change in information-processing abilities after self-produced actions	Video scoring from a live feed	Random videos coded for interrater reliability: correlation analysis (r > 0.97)
	Looking duration to switch in direction of action goal		Duration of looking time at an animated event on a computer screen which resembled the contingent training but with change in direction (Red ball being hit by green ball)	Measures infant's ability to discriminate the reversal in agent-recipient roles in other's actions
	Looking duration to switch in causality of action goal		Duration of looking time at an animated event on a computer screen which was displayed in same direction as training but there was no contingency/causality (Green ball moved towards red ball but did not hit it)	Measures infant's ability to discriminate noncausality in other's actions
Gerson and Woodward37,38 (2014)	Looking duration to change in action goal structure	Causal cognition	Duration of looking time at a staged event with a switch in action goal compared with a familiarization trial (Familiarization: Experimenter showed infant 2 toys, ie, bear and ball and touched the ball with mitten) (Test: Experimenter showed infant 2 toys, ie, bear and ball and touched the bear with mitten)	Measures infant's ability to discriminate goal structure in other's actions. Indicates change in information-processing abilities after self-produced actions	Video coding using the Java Habituation software (version 1.0.0)	100% double coding for interrater reliability: correlation analysis (r ≥ 0.94)
Williams et al39 (2015)	Total intentional toy contacts	Motor and visual cognition	Frequency of times the infant made hand toy contact while looking at it either prior or during the contact.	Measures “intentionality” during early reaching	Video coding using the Observer XT (Noldus Information Technology BV, Wageningen, The Netherlands)	30% of videos coded for interrater reliability: correlation analysis (r = 0.92)
	Total toy contacts	Motor	Frequency of times the infant made hand toy contact without consideration to the visual attention/looking	Measures early reaching abilities
	Visual attention to toy	Visual cognition	Duration of time the infant spent looking at the toy	Measures engagement in exploration		20% of videos coded for interrater reliability. Cohen κ values (0.70-0.95), average = 0.84
	Peak velocity of reaching	Motor	Average peak velocity of reaching was calculated by dividing the values of the velocity peaks by the total number of velocity peaks in the time series	Measures the kinematics of reaching, ie, speed of hand while approaching the target. Slower speed indicates better control	Collected using Mini Flock of Birds motion analysis system (Ascension Technology Corp, Burlington, Vermont) and processed using MATLAB	Not reported
Wiesen et al40 (2016)	Toy touches	Motor	Duration of time the infant made manual contact with the toy	Measures early reaching abilities	Frame-by-frame video coding using StopFrameCoding software (Version 0.9)	Random sample recoded for interrater reliability: Looking (ICC = 0.91), touching (ICC = 0.96), reaching (ICC = 0.85), bimanual (ICC = 0.90), and grasping (ICC = 0.99)
	Reaching		Duration of time the infant made a movement of the hand toward the toy
	Grasping		Duration of time the infant's finger encircled or gripped the toy	Measures early object manipulation abilities
	Bimanual exploration		Duration of time the infant touched the toy with both hands
	Visual attention to toy	Visual cognition	Duration of time the infant's gaze was directed toward the toy	Measures engagement in exploration
Needham et al41 (2017)	Total toy touches	Motor	Duration of time the infant spent with any of the hands or mouth in contact with the toy	Measures early object manipulation abilities	Frame-by-frame video coding using StopFrameCoding software (Version 0.9)	20% of videos were coded for interrater reliability: looking (ICC = 0.96), touching (ICC = 0.99)
	Visual attention to toy	Visual cognition	Duration of time the infant spent in visual contact with the toy	Measures engagement in exploration
Nascimento et al42 (2019)	Total number of reaches	Motor	Number of times the infant contacted the toy with one or both hands, regardless of grasping	Measures early reaching abilities	Video coding using the ReSpeedr software	20% of videos were coded for interrater
						reliability: Cohen's κ = .86 (95% CI ± 0.01)
	Total number of proximal hand adjustment, ie, bimanual reaches		Number of times the infant moved both hands simultaneously toward the toy for at least 50% of the trajectory, resulting in a touch	Measures motor adaptations during early reaching
	Total number and type of distal hand adjustment while reaching		Number of times the infant reached for a toy with hand closed versus hand open
	Grasping		Number of times the infant held/grabbed the toy with fingers using one or both hands after reaching	Measures early object manipulation abilities
Heathcock et al43 (2008); Heathcock and Galloway44 (2009)	Average hand-toy contacts	Motor	Average number of times the infant contacted the toy per assessment	Measures early reaching abilities	Video coding using custom program	20% of videos were coded for interrater reliability. Percent agreement was >90%
	Hand-toy contact duration		Proportion of time the infant contacted the toy for across all testing trials per assessment	Measures early object manipulation abilities
	Hand-toy contact type		Proportion of time the infant contacted the toy with dorsal aspect of hand versus ventral aspect and open hand versus closed hand, across all testing trials per assessment	Measures complexity of object manipulation. Contact with ventral surface and open hand represents a more functional grasp
	Foot-toy contact frequency		Number of times the infant contacted the toy with any part of its foot or toes	Measures early feet-reaching behaviors
	Foot-toy contact duration		The average amount of time infants spent touching the toy per foot-toy contact
Needham et al45 (2014)	Manual toy contact	Motor	Duration of time the infant spent touching the toy with hand or fingers when it was placed in its hand	Measures early object manipulation abilities	Frame-by-frame video coding using StopFrameCoding software (Version 0.9)	26% of videos coded for interrater reliability: Percent agreement was maintained at 94.28%
	Bimanual exploration		Duration of time the infant spent contacting the toy with both hands simultaneously, when it was placed in its hand	Measures early object manipulation abilities
	Reaching		Duration of time the infant moved its hand and arm in the direction of the toy. Contact with toy was not necessary	Measures early reaching abilities
	Visual attention to toy	Visual cognition	Duration of time the infant spent looking at the toy when it was placed out of its reach	Measures engagement in exploration
	Visual attention to experimenter		Duration of time the infant made contact with experimenter's body or face	Measures engagement in exploration and attention to task
Campbell et al46 (2015)	Average leg movement frequency	Motor	Frequency of times the hip was observed to move more than about 15° in either direction from the resting position in the slings of approximately 90° of hip and knee flexion		Video coding	Random videos coded for interrater reliability: percent agreement was 89%
	Interlimb coordination patterns	Coded in 3 ways (Reference leg = Right): 1. Ipsilateral leg movement 2. Alternate (left leg extending within 1 s of right hip flexion) 3. Synchronous (both hips flexing within 1 s of each other)
Williams and Corbetta47 (2016)	Visually attended toy contacts	Motor and visual cognition	Frequency of times the infant made hand-toy contact while looking at it either prior or during the contact	Measures “intentionality” during early reaching	Video coding using the with the ObserverXT9 (Noldus Information Technology, Wageningen, The Netherlands)	20% of videos were coded for interrater reliability. Percent agreement was 91%
	Total toy contacts	Motor	Frequency of times the infant made hand-toy contact without consideration to the visual attention/looking	Measures early reaching abilities
	Visual attention to toy	Visual cognition	Duration of time the infant spent looking at the toy	Measures engagement in exploration		20% of videos were coded for interrater reliability: percent agreement was 85%
	Mean peak velocity of reaching	Motor	Average peak velocity of reaching was calculated by dividing the values of the velocity peaks by the total number of velocity peaks in the time series	Measures the kinematics of reaching, ie, speed of hand while approaching the target. Slower speed indicates better control	Collected using Mini Flock of Birds motion analysis system (Ascension Technology Corp, Burlington, Vermont) and processed using MATLAB	Not reported

Abbreviations: CI, confidence interval; ICC, intraclass correlation coefficient (measure of reliability).

TABLE 3 - Results for Effectiveness of Included Studies

Study	Feeding Outcomes	Motor and Cognitive Outcomes	Favors Intervention Group (P)a	Favors Control Group/No Difference Between Groups (P)a
Standley28 (2003)	Oral feeding rate		<.05
Standley et al29 (2010)	Number of oral feeding days before discharge Reduction in length of gavage feeding days		<.05 (at 34-wk adjusted age and with 3 sessions of intervention)	>.05 <.05 (at 32-wk adjusted age)
Chorna et al 30 (2014)	Oral feeding rate and frequency Oral feeding volume Time/days to oral feeding		<.001 .001 .04 Average 7 d faster
Needham et al31 (2002)		Intentional swats Visual attention to toy Multimodal exploration (visual + oral)	<.05 <.001 <.05
Libertus and Needham32,33 (2010, 2011)		Reaching and grasping Visual attention to toy Visual attention to experimenter Face preference Face orienting	.009 .011 .03 .022 .004
Libertus et al34 (2016) Follow-up to Libertus and Needham32 (2010)		12-mo follow-up to Libertus and Needham32 (2010) Toy grasping Toy rotations Visual attention to toy Distraction	.012 .004 .004 <.001
Libertus and Needham35 (2014)		Reaching and grasping Visual attention to toy Face preference	.033 .019	>.121
Rakison and Krogh36 (2012)		Looking duration to switch in direction of action goal Looking duration to switch in causality of action goal	<.05	>.05
Gerson and Woodward37,38 (2014)		Looking duration to change in action goal structure	.021 .034
Williams et al39 (2015)		Total toy contacts Total intentional toy contacts Peak velocity of reaching Visual attention to toy	.010	>.05 .047 <.0001
Wiesen et al40 (2016)		Postintervention Toy touching and reaching Visual attention to toy 2-mo follow-up Toy touching Grasping Reaching Bimanual exploration Visual attention to toy	.025 .004 .005 <.001	>.05 >.05 .011
Needham et al41 (2017)		Toy touching Visual attention to toy	.008	.077
Nascimento et al42 (2019)		Postintervention Total number of reaches Number of bimanual reaches Open vs closed hand reaches Grasping Retention test (4 min later) Total number of reaches Number of bimanual reaches Open vs closed hand reaches Grasping	<.01 .01 <.01	>.63 >.05 .07 >.30 >.05
Heathcock et al43 (2008) Heathcock and Galloway44 (2009)		Average hand-toy contacts Hand-toy contact duration Number of open hand-toy contacts Number of ventral toy contacts Foot-toy contact frequency Hand-toy contact duration	<.01 <.05 <.01 <.002 .000 <.045
Needham et al45 (2014)		Manual toy-contact Bimanual exploration Reaching toward toy Visual attention to toy Visual attention to experimenter	.025	.798 .723 .957 .897
Campbell et al46 (2015)		Average leg movement frequency Interlimb coordination patterns, ie, more synchronous hip flexion movements		.121 .932
Williams and Corbetta47 (2016)		Total toy contacts Visually attended toy contacts Mean peak velocity of reaching Visual attention to toy	.044 <.05	.011 <.05

^aP ≤ .05 indicates significant results.

ROB and Quality Assessment Within Studies

Of the 17 studies selected, 10 were RCTs^{28–30,36,40–43,46,47} and 7 were QRCTs.^{31–35,37,39,45} Based on Sackett's grading, all the RCTs had level II evidence (sample size <100) and QRCTs had level III evidence. According to Cochrane ROB-2 for RCTs, 6 RCTs had threats,^{28,40–43,47} and 4 RCTs had a low ROB (Table 4).^29,30,46 Based on ROBINS-I, 3 QRCTs had serious ROB,^31,35,45 3 had moderate ROB,^32,37,39 and 1 had a low ROB (Table 5).³⁶

TABLE 4 - Cochrane Risk of Bias (ROB-2) for Randomized Controlled Trials

Study	Domain 1: Randomization Process	Domain 2: Deviation From Interventions	Domain 3: Missing Outcome Data	Domain 4: Measurement of Outcome	Domain 5: Selection of Reported Result	Overall Risk of Bias
Standley28 (2003)	Some concerns	Low	Low	Low	Some concerns	Some concerns
Standley29 et al (2010)	Some concerns	Low	Low	Low	Low	Low
Chorna et al30 (2014)	Low	Low	Low	Low	Low	Low
Rakison and Krogh36 (2012)	Low	Low	Low	Low	Low	Low
Wiesen et al40 (2016)	Some concerns	Some concerns	Low	Some concerns	Low	Some concerns
Needham et al41 (2017)	Low	Some concerns	Low	Some concerns	Low	Some concerns
Nascimento et al42 (2019)	Low	Low	Low	Low	Some concerns	Some concerns
Heathcock et al43 (2008)	Low	Some concerns	Low	Low	Low	Some concerns
Heathcock and Galloway44 (2009)	Low	Some concerns	Low	Low	Low	Some concerns
Campbell et al46 (2015)	Low	Low	Low	Low	Low	Low
Williams and Corbetta47 (2016)	Some concerns	Low	Low	Some concerns	Low	Some concerns

TABLE 5 - ROBINS-I Risk of Bias Assessment for Nonrandomized Controlled Trials

	Before Intervention		At Intervention	After Intervention
Study	Confounding	Selection of Participant	Classification of Intervention	Deviation From Intended Intervention	Missing Data	Measurement of Outcome	Selection of Reported Result	Overall ROB
Needham et al31 (2002)	Serious	Low	Low	Low	Moderate	Low	Moderate	Serious
Libertus and Needham32 (2010)	Moderate	Low	Low	Low	Moderate	Low	Moderate	Moderate
Libertus and Needham33 (2011)	Moderate	Low	Low	Low	Moderate	Moderate	Moderate	Moderate
Libertus et al34 (2016)	Low	Low	Low	Low	Low	Low	Low	Low
Libertus and Needham35 (2014)	Serious	Low	Low	Low	Low	Serious	Low	Serious
Gerson and Woodward37 (2014)	Moderate	Low	Low	Low	Low	Moderate	Low	Moderate
Gerson and Woodward38 (2014)	Moderate	Low	Low	Low	Low	Moderate	Low	Moderate
Williams et al39 (2015)	Moderate	Low	Low	Low	Serious	Low	Moderate	Moderate
Needham et al45 (2014)	Serious	Low	Low	Low	Low	Low	Serious	Serious

Abbreviation: ROB, risk of bias.

CPBI for Feeding Outcomes

Three studies used CPBI to improve feeding outcomes^28–30 and comprised a total of 194 preterm infants between 32 and 36 weeks of adjusted age (AA) at treatment onset. The CPBI used in all 3 studies was pacifier-activated lullaby (PAL), with 1 study specifically recording the lullabies in the mother's voice.³⁰ The PAL interventions positively reinforce the preterm infant's nonnutritive sucking (NNS) by associating the suck with a lullaby whenever the infant's suck reaches a threshold pressure. The control groups received usual care for NNS. The PAL intervention was provided once per day, each session lasting for 15 to 20 minutes. Frequency varied from 1 to 5 days. The PAL intervention was provided 30 to 60 minutes before the scheduled feeding time (see Supplemental Digital Content 3, available at: https://links.lww.com/PPT/A352).

Outcomes were assessed at baseline and postintervention. Primary outcomes measured nutritive sucking, that is, number of oral feeding days before discharge, oral feeding rate, volume, and frequency. Secondary outcomes included number of gavage feeding days and days to transition to oral feeding (Table 2).

The PAL intervention significantly improved oral feeding rate (P < .05),^28,30 frequency (P < .001),³⁰ and volume (P = .001)³⁰ in preterm infants compared with usual care. No differences between groups were reported for number of oral feeding days before discharge.²⁹ For the secondary outcomes, PAL trials significantly reduced gavage feeding days (P < .05) when initiated at 34 weeks of AA.²⁹ This reduction was significantly greater with 3 intervention sessions when compared with 1 session (P < .05). The PAL intervention provided at 32 weeks of AA had a negative outcome, that is, it lengthened the gavage feeding days significantly (P < .05). Finally, PAL intervention reported a faster transition to oral feeding (P = .04), that is, an average of 7 days earlier than the control group.³⁰

CPBI for Motor and Cognitive Outcomes

Fourteen studies used contingency paradigms as either the primary or one of the key components of intervention to improve motor and cognitive outcomes in infants.^{31,32,34–37,39–43,46,47} Eleven studies included term born infants^{31,32,34–37,39–41,45,47} (n = 421) and 3 studies included preterm infants^42,43,46 (n = 63) resulting in a total of 484 infants across the 14 studies. Age at treatment onset had a range of 2 to 5 months.

The CPBI comprised either sticky mittens/socks or contingent toys. In sticky mittens training, infants are fitted with Velcro mittens and the corresponding end of the Velcro is attached to light-weight toys. The toys stick to the mitten when infants try to contact them either accidently or on purpose providing positive reinforcement for reaching. Sticky socks are similar and are used to encourage lower extremity movements in infants. Contingent toys in this review refer to toys that provide reinforcement in response to the infant's active movement and include an overhead mobile attached via ribbon to infant's arms or legs, modified toys that made a sound, light up, or oscillated on touch, and socks with attached auditory reinforcements. Control groups included noncontingent interventions or no training (see Supplemental Digital Content 3, available at: https://links.lww.com/PPT/A352).

Outcomes were assessed at baseline and immediately postintervention in 12 studies^{31,32,35–37,39,41–43,46,47} and at 2 months⁴⁰ and 12 months³⁴ postintervention in 2 studies. Motor outcomes included (i) manual exploration behaviors such as touching, swatting, grasping, bimanual explores, and object rotations, and (ii) feet exploration behaviors such as kicking or foot-toy contacts. Cognitive outcomes included measures of early cognition such as (i) visual cognition, that is, visual attention to the toy, people, and environment, and shifts in visual attention, (ii) social cognition, that is, face preference, and (iii) causal cognition, that is, action goal understanding (Table 2).

Eight studies used a closed sticky mitten training protocol^{31,32,34–37,40,41} in term infants. Effectiveness of closed sticky mittens on motor outcomes was measured in 6 studies,^{31,32,34,40,41,48} and 4 of the 6 studies reported improvements in intentional swatting, reaching, and grasping (P < .05) in prereaching term infants.^31,32,35,41 An increase in bimanual exploration (P = .005), grasping (P = .004, P = .012), and object rotations (P = .004) was observed at 2-month⁴⁰ and 12-month³⁴ follow-up. Closed sticky mittens improved manual motor outcomes only when training was provided for minimum 10 sessions (total 100 minutes of training).

Eight studies measured the effectiveness of closed sticky mittens on 1or more cognitive outcomes.^{31,32,34–37,40,41} Six of the 8 studies measured visual cognition^{31,32,34,35,40,41} and 5 of the 6 studies reported an increase in visual attention and gaze shifts (P < .05) immediately postintervention^31,32,41 or at a follow-up in term infants.^34,40 Two of the 8 studies measured social cognition^33,35 and reported increased face preference in infants in the intervention group (P = .02 and P = .019). Closed sticky mittens improved visual and social cognition only when training was provided for minimum 10 sessions (total 100 minutes of training). Finally, 2 of the 8 studies measured causal cognition^36,37 and reported an increase in action-goal understanding in term infants in the intervention group after a single 3-minute training session (P < .05).

One study used an open sticky mitten training protocol in term infants³⁹ and reported that the control group (nonsticky mittens) improved significantly in reaching and visual attention compared with the open sticky mitten group (P = .047). Finally, 1 study⁴² used an individualized open sticky mitten training in late preterm infants and reported an immediate increase in reaching and bimanual exploration in the sticky mitten group following a single 4-minute session; however, this effect was transient and not retained.

Four studies used contingent toys^43–47 and 3 measured manual exploration.^43,45,47 Two of 3 studies reported improvements in intentional toy contacts in the contingent toy group in term and preterm infants, respectively (P = .04, P < .01).^43,47 In the study including preterm infants, intensive dosage of intervention was provided (total 600 minutes of training), and it included additional movement training components.⁴³ The remaining one study that did not report change in manual exploration included a single session of 9-minute training.⁴⁵

Two studies measured the effectiveness of contingent toy training on feet exploration in preterm infants,^44,46 with 1 study specifically including preterm infants with periventricular brain injury (PBI).⁴⁶ The study including preterm infants without PBI reported improvements in feet-toy contacts in very preterm infants.⁴⁴ The study including preterm infants with PBI reported no additional benefits of contingent toy training. However, a longitudinal follow-up revealed that higher number of infants (diagnosed as typically developing at 12 months) in the contingent toy group were walking independently at 12 months of age.⁴⁶ Two studies measured visual attention toward the toy and reported no effect of contingent toy training on visual cognition.^45,47

DISCUSSION

This systematic review identified 17 unique controlled trials using CPBI for improving feeding, motor, and cognitive outcomes in infants.^{28–32,34–37,39,40–43,45–47} Our review suggests that interventions based on contingency paradigms have potential for application to physical therapy practice, starting from NICU to early intervention in infancy. In the following sections, we will highlight the clinical implications for these findings based on each intervention, that is, PAL, sticky mittens, and contingent toys and its target outcome (feeding, motor, and/or cognitive).

Pacifier-Activated Lullaby

This CPBI was primarily applied in the NICU setting for infants born extremely to very preterm to improve feeding outcomes. Our results support that PAL intervention can reinforce NNS, improve nutritive sucking, and reduce number of gavage feeding days for extremely to very preterm infants.^28–30 This finding is important as nutritive sucking can be a challenge for very preterm infants due to underdeveloped oral-motor skills, poor sucking endurance, and inability to coordinate motor and autonomic systems.⁴⁹ The NNS experience during gavage feeding is crucial for facilitation of functional feeding. Moreover, ability to perform nutritive sucking with no cardiorespiratory comprise is one of the criteria for discharge to home for preterm infants.⁴⁹ Although contingency learning for feeding has been demonstrated in infants at as early as 32 weeks of gestational age, most of the studies refer to training-specific learning.⁵⁰ Our results support that reinforcement provided by CPBI during NNS can transfer to improved performance during functional feeding. We identified 3 important parameters that need to be considered for implementing CPBI for feeding in very preterm infants: (i) age of initiation, (ii) salience of reinforcer, and (iii) dosage of training. Most optimal feeding outcomes are achieved when CPBI are initiated at or after 34 weeks of AA.²⁹ This could be because coordinated suck-swallow function starts establishing around 34 to 36 weeks of gestational age in infants.⁴⁹ Moreover, preterm infants younger than 34 weeks may not tolerate CPBI due to poor readiness to oral feeding caused by intubation, fatigue, and impaired state regulation.⁴⁹ More significant gains in feeding were observed when mother's voice was used as an auditory reinforcer.³⁰ Biological sounds such as mother's voice are identified as strong auditory reinforcers in existing literature.⁵¹ Mother's voice can also simulate womb-like auditory experience for the infants and assist with state regulation, positively impacting feeding.⁵² Finally, a minimum of 3 days of training is required to see changes with CPBI for feeding in very preterm infants aged 34 weeks of AA and older. This finding corroborates with previous studies, which state that preterm infants need more practice than full-term infants to demonstrate learning in contingency learning paradigms.^13,53

Sticky Mittens

Sticky mittens constituted the majority of CPBI used for improving motor and cognitive outcomes in young infants. All but one of these studies were focused on term born infants aged 2 to 3 months.^{31,32,34–37,39–41} The results suggest that closed sticky mitten training, when provided in the “prereaching” stage, increases reaching, manual exploration, visual attention, gaze shifts, and face preference in term born infants.^{31,32,34,35,40,41} Several mechanisms have been proposed for the effectiveness of closed sticky mittens in term infants. First, sticky mittens reward the initial reaching attempts of infants allowing them to identify the effect of their self-generated actions on the environment, thus, motivating them to repeat these actions.⁴¹ Second, early interaction with objects during reaching allows infants to locate the position of their hand in space, contributing to the refinement of visual attention and development of complex visual skills such as gaze shifts.⁵⁴ Third, the ability to successfully act on the environment using closed sticky mittens may help young infants see themselves as intentional agents in the world, improving nonverbal communication with caregivers in the form of increased face preference.³³ Preliminary evidence from 2 studies also shows that closed sticky mittens training can help infants distinguish goal structure in other people's actions when tested in the laboratory setting.^36,37 Action-goal understanding is the foundation for social referencing, language learning, and imitative learning in the second year of life.³⁸ This finding should be tested in a real-world setting before making conclusive recommendations.

To summarize, the studies on closed sticky mittens provide promising results for improving manual motor skills and cognitive precursors in term infants. However, the overall effectiveness of the sticky mitten paradigm requires a more critical analysis. The 2 studies using the open sticky mittens paradigm reported no effect and a transient effect of sticky mitten training, respectively.^39,42 Our TIDieR analysis (see Supplemental Digital Content 3, available at: https://links.lww.com/PPT/A352) revealed crucial differences in intervention protocols for closed versus open sticky mittens, which make comparisons and overall clinical recommendations difficult for sticky mittens. Based on the TIDieR analysis, we identified the following prerequisites for the sticky mitten paradigm to be effective: (a) demonstration of the contingency at the beginning of training, (b) parental verbal encouragement during the training, and (c) a dosage of minimum 10 sessions across 2 weeks (total 100 minutes of reaching training).

Although only 1 study in this review tested effectiveness of sticky mittens training in preterm infants,⁴² this training has potential for improving reaching in preterm infants. The onset of reaching is delayed in infants born preterm and this delay has been attributed to both intrinsic (presence of brain injury, lower muscle tone, poor muscle strength and coordination, postural control, and perceptual deficits) and extrinsic factors (extrauterine environment and movement experiences).⁵⁵ Results for sticky mittens from term infants show that this training significantly increases reaching attempts. Replicating this finding in preterm infants can prove to be valuable, as increased number of reaches can positively predict neurodevelopment at 2.5 years in very preterm infants.⁵⁶ In addition, training prereaching preterm infants using sticky mittens may help minimize the gravitational demands of “successful reaching,” while allowing opportunities for refining arm control and improving proximal muscle strength. Finally, sticky mittens afford reaching success without demanding grasping precision. Thus, reducing constraints to reaching using sticky mittens may facilitate repetition and motor learning, especially in preterm infants.⁵⁷

Contingent Toys

Results from limited studies demonstrate that contingent toys can improve reaching and manual exploration in term and preterm infants aged 2 to 3 months, when provided in a dosage of minimum 10 sessions across 2 weeks (total 100 minutes of training).^43,47 Contingent toy intervention when provided in an intensive dosage (minimum 400 minutes of training across 8 weeks) may also improve feet exploration or kicking in preterm infants. However, this finding is based on 1 study,⁴⁴ which included preterm infants without brain injury and incorporated additional movement training components such as midline foot-toy contact and leg movement dissociation in training. Another study using similar dosage of “isolated” contingent toy training for kicking in preterm infants with brain injury did not report similar benefits.⁴⁶ We hypothesize that scaffolded movement training combined with reinforcements may be necessary for motor learning in preterm infants. Moreover, presence of brain injury may impact contingency learning abilities in preterm infants requiring either increased dosage or learning support for noticing gains. Although limited, the evidence of CPBI combined with movement training is of value as atypical selective motor control during kicking is predictive of delayed walking attainment in preterm infants. Thus, CPBI need to be further explored for training feet exploration.

LIMITATIONS

The inclusion of both term and preterm infants in this review could inflate the effectiveness of CPBI. However, this decision was made because of the lack of existing studies on CPBI in preterm infants. We outline the implications for results from term infants to design and modify interventions for preterm infants. Seven studies in this review were NRCTs with level III evidence, with majority having moderate to serious ROB.^{31–35,37,39,45} None of the 10 RCTs had a level I evidence, reducing the methodological rigor of the studies. All but 2 studies^34,40 in this review were cross-sectional, which limits the findings to infants aged 6 months and younger, and no conclusions on the long-term effect of CPBI on developmental outcomes can be drawn. Despite many studies in this review, we refrained from performing a meta-analysis due to inconsistent reporting of outcomes between studies. None of the outcomes in the studies in this review were standardized motor and cognitive measures. Although task-specific outcomes provide valuable information, they lack stringent psychometric properties that limit confidence in the effectiveness of contingency paradigms on global development. Although we planned on reporting effect sizes, the variation in number of behavioral outcomes would have impacted the interpretation of effect size. Hence, we report only the P values. Finally, our strict search criteria may have failed to include studies that have contingent paradigms built into the intervention^7,58,59 and relevant literature in other languages.

IMPLICATIONS FOR CLINICAL PRACTICE

The results of this review have several important clinical implications. The CPBI can be implemented by therapists in the NICU to promote optimal feeding outcomes in preterm infants and are most effective when initiated at 34 weeks of gestational age. Early reaching practice using closed sticky mitten protocol may increase reaching behaviors and set the stage for early onset of perceptual and cognitive milestones in term infants. Physical therapists can modify and use commercially available, inexpensive toys such as bells, rattles, and light up toy to set up a contingent response for training reaching and kicking in preterm infants, combined with intensive movement practice.^43,44 Infants with or at risk for delays often need more practice and may demonstrate lower levels of persistence. The CPBI can address both these barriers and have the potential to advance visual, social, and causal cognition in infants at risk for delays.

IMPLICATIONS FOR RESEARCH

Improvement in feeding outcomes after CPBI in preterm infants highlights their ability for contingency learning before term age. Additional research is needed to test the effectiveness of contingency paradigms for facilitating age-appropriate motor and cognitive skills in preterm infants in the NICU and its effect on later developmental outcomes. Our results show that CPBI can be an effective treatment strategy for improving reaching, visual attention, and face preference in term infants; however, its translation to at-risk infants is needed. Future research should also disentangle the role of movement practice and parental scaffolding from reinforcement training. The closed and open sticky mittens use different training protocols. More research is needed to determine whether one approach is better than the other. Objective documentation of other possible mechanisms for improvement in reaching with sticky mittens such as improved muscle strength and coordination, proprioceptive feedback, or sense of self-efficacy is needed. The CPBI may also have the potential to improve additional motor milestones such as head control development, prone position play,^60,61 rolling, sitting, and walking. High-quality RCTs are needed to evaluate these theoretical findings. Future studies also need to measure the effectiveness of CPBI using robust study designs, standardized motor measures, and objective early cognitive measurement tools such as eye trackers.⁶²