Review registration and reporting
The systematic review is submitted to PROSPERO (the International Prospective Register of Systematic Reviews) for registration on July 26, 2020. We have adhered to PRISMA-P (Preferred reporting items for systematic review and meta-analysis) guidelines to report this protocol (supplementary data file 1).
Eligibility criteria
Types of studies: Quasi-experimental trials and randomized controlled trials are eligible to be included. Non-randomized trials, observational studies, study protocols, editorials, reviews, conference abstracts, letters, and commentaries will be excluded.
Type of participants: Preterm infants born lesser than 32 weeks of gestation and their primary caregivers. Primary caregivers can be either mother, father, or grandparents.
Place of recruitment: This review does not impose any restrictions on the setting. It can be in the NICU, hospital, and home or community.
Time of recruitment and follow up: The studies should have recruited infants from birth in the NICU stay to six months, and followed up till two years of age.
Intervention: This review will include any FCC interventions involving the establishment of a collaborative relationship between the healthcare professional and the parent, mutually agreed-upon goal setting, creating the home program by selecting therapeutic activities that focus on accomplishing family objectives, supporting the implementation of the program through home visits, parent education and evaluating the outcomes. Interventions having at least two components of FCC will be included. The intervention will involve supervision and support from a clinician or professional such as a neonatologist, pediatrician, nurse, physiotherapist, occupational therapist, speech-language pathologists, and other rehabilitation team members.
Type of comparators: The studies should have compared the FCC to the therapist provided standard care interventions or usual care.
Type of exposure: such as age, time of recruitment and follow up, settings, FCC providers, intensity and frequency of intervention; parental behavior, responsivity, and parental satisfaction that may influence infant development.
Types of outcome measures: Primary outcomes - We will include studies that measure motor and neurobehavioral function.
Motor functions such as quality of movement, gross and fine motor skills, developmental milestones, visual-spatial, visual-motor integration, balance, and coordination are measured through any of the outcome measures. The outcome measures could be Prechtl’s General Movements Assessment (GMA), Test of Infant Motor Performance (TIMP), Alberta Infant Motor Scale (AIMS), Neuromotor Behavioral Assessment (NMBA), Hammersmith Infant Neurological Examination (HINE), Pediatric Evaluation of Disability Inventory (PEDI), Peabody Developmental Motor Scale (PDMS) and Bayley Scale of Infant and Toddler Development (BSID).
Neurobehavior is measured in terms of the sensory and autonomic nervous system, organization of state (calm, excited, irritable) and self-regulation (hand to mouth responses); language, attention, socio-emotional development, and executive function using any of the outcome measures. Neurobehavior could have measured using Assessment of Preterm Infants Behavior (APIB), Brazelton Neonatal Behavioral Assessment Scale (NBAS), Neurobehavioral Assessment of Preterm Infants (NAPI) and NICU Network Neurobehavioral Scale (NNNS).
Secondary outcomes are changes in parental behaviors or responsivity captured through videotaped interactions or observations and measured by any of the validated scales. Parental satisfaction will be measured by questionnaires and interviews. Factors such as age, time of recruitment and follow up, settings, FCC providers, intensity, and frequency of intervention which might influence infants development will be considered for the review.
Search methods for identification of studies
Electronic databases: We will search PubMed, Cochrane Central, Scopus, EMBASE, PsycINFO (Ovid SP), CINAHL and Web of Science. Articles that are written in English from January 2010 to August 15, 2020 will be included. The following keywords will be used; “family-centered care”, “family-centric approach”, “preterm infants”, “motor development”, “neurobehavior development”.
Searching other resources: We will search in ProQuest, ClinicalTrials.gov, EU Clinical Trials Register, metaRegister of Controlled Trials, and the World Health Organization (WHO) International Clinical Trials Registry Platform search portal. We will contact field experts and corresponding researchers of included studies to get relevant additional information. To find related studies, we shall search for reference lists and forward citations of included studies. Identified records will be exported to EndNote X7 for data management.
Selection of studies
Two reviewers will search and read the titles, abstracts of the listed sources individually, and exclude any studies depending on the eligibility criteria. If disparity arises between the authors, it will be resolved by discussions and will be reviewed as to their full text. After obtaining the full text of the included abstracts, three review authors will independently rank these as ‘include’ or ‘exclude’. If necessary, we will address any differences in agreement, with a senior review author. We will record the reasons for excluding the articles. Proposed screening protocol for abstracts and full texts has been attached as supplementary data file 2. Reference details, any knowledge accessible on current research and record details on similar publications such that each study will be considered as the unit of concern in the analysis. An adapted PRISMA will be implemented (Fig. 2).
Data extraction and management
The extraction of general characteristics such as study identifiers, location of the study, participants, study selection criteria, and outcomes from the included studies will be carried out. The TIDier (Template for Intervention Description and Replication) checklist will be used to summarize the list of intervention characteristics and assist in the replicability of interventions and comparability between the studies [28]. Independently, two review authors will abstract the data from the articles included in the review using the data extraction templates. From each of the included studies, the following information will be extracted: study identification (title and authors), the country of study, year of publication, sample size, features of intervention, outcomes, evaluation tools, results, and conclusions. In case of disagreement, we will discuss until consensus or with the help of the third reviewer, any disagreements will be resolved.
Addressing redundant and related publications
In the case of redundant papers, related records, or several primary research studies, we can optimize the data yield by gathering all the relevant details by utilizing the highly broad dataset collected through each of the documented papers. In correlation with our results, the publication which reports the longest follow-up will be considered as a priority.
Risk of bias assessment among included studies
The Revised Cochrane Risk of Bias Tool (RoB 2.0) would be employed to critically appraise the quality of the included articles. The RoB 2.0 tool assesses the randomization process, deviations from the intended protocol, measurement of the outcome, missing outcome data, and selective reporting. Two authors will appraise all the articles, independently. There should be agreement about the existence of certain inconsistencies. The risk will be categorized into low risk, high risk and some concern [30].
Addressing the missing data
From the authors missing information would be collected and significant empirical data such as screened, randomized, intention-to-treat, as-treated, and per-protocol population will be closely analyzed. If authors will not respond within 15 days of last communication (email), the study would be excluded from the review. We will critically appraise the concerns related to the missing data and imputation methods, e.g. Last Observation Carried Forward (LOCF). The attrition rates, e.g. drop-outs, losses to follow up and withdrawals will be investigated.
Data synthesis
The outcomes will be presented as the mean differences (MDs) / standardized mean differences (SMDs) with 95% confidence intervals (CI) unless otherwise stated for continuous variable. The odds ratios (ORs) or risk ratio (RRs) with 95% CI will be expressed as the dichotomous data. If there is strong evidence of homogeneous effects through findings, using a model of random effects we will mainly sum up a low probability of bias results. The meta-analysis of random effects will be viewed with due consideration of the full range of outcomes, preferably by providing an interval of predictions. In each study, the predicted range for the true treatment effect will be specified using the prediction interval. The quality of evidence of the included studies in the review will be assessed using Grading of Recommendations Assessment, Development and Evaluation (GRADE) system. This system grades quality of evidence at 4 levels: high (4), moderate (3), low (2), and very low (1) [31].
Subgroup analyses and investigation of heterogeneity
In the meta-analysis, we will not record the findings of the sample as the pooled effect estimate if there is any the event of major scientific, analytical, or empirical variability. Using a standard Chi-square test (significance level of \(\alpha\) = 0.1) we will recognize the heterogeneity. Using the I2 statistics, inconsistency across studies will be quantified. It is considered to be a high level of inconsistency if an I2 statistic is 75% or more. If there is significant variability, we will aim to identify potential explanations for this by analyzing the features of the specific study and other subgroups. Subgroup analyses of factors such as gender, age, country, type, and mode of intervention and outcome measures will be carried out to explore the interaction between them.
Sensitivity analysis
To understand how the following factors influence the effect sizes, we will perform the sensitivity analyses. The factors include, the risk of bias of included studies; large and long trials to understand the extent to which they influence the results. Different effect size measurements such as risk ratio and odds ratio along with various statistical models such as fixed-effect and random-effects models will be used to test the robustness of the results.
Narrative synthesis
When it is not possible to conduct the meta-analysis because of significant statistical, clinical or methodological heterogeneity, a narrative synthesis will be done. Studies would be narratively defined focusing on the intervention and outcomes. The subsequent results would be summarized using the tables and figures.