Authors: Vivienne Engel, Valentina Mauron-Papadimitriou, Lea Bührer, Laura Ebner, Lena Keller, Martina Gosteli, Sandra Siegfried, Beatrice Latal, Ursula Kiechl-Kohlendorfer, Ulrike Held
Categories: Paediatrics, Hospital to Home Transition, INTENSIVE & CRITICAL CARE, NEONATOLOGY, Protocol, 1719, 1506
Source: BMJ Open
Authors: Vivienne Engel, Valentina Mauron-Papadimitriou, Lea Bührer, Laura Ebner, Lena Keller, Martina Gosteli, Sandra Siegfried, Beatrice Latal, Ursula Kiechl-Kohlendorfer, Ulrike Held
Medical progress has significantly improved the survival rates of very preterm-born infants in recent decades. Nevertheless, these infants are still at increased risk for long-term impairments as compared with term-born infants. While the homecoming of a preterm-born infant is long-awaited and brings relief to families, it also marks the end of intensive monitoring and highly specialised professional care. This situation often leaves parents coping with anxiety and a sense of unpreparedness, as they navigate the transition to home. Correspondingly, the WHO advocates for additional support for parents of preterm infants following the transition to home. According to the WHO, preparation for this transition yields positive effects on crucial aspects such as nutrition, parent–child interaction and parental well-being. However, there is significant heterogeneity as to which interventions are in place and regarding their level of efficacy. Consequently, we aimed to provide an overview of existing transition-to-home interventions and their efficacy by conducting a systematic review of the literature.
We will perform a systematic review of interventions aiming at improving the transition to home process for very preterm-born infants and their parents and will search the following Cochrane, Medline, CINAHL, EMBASE and PsycINFO. Our main aim is to provide an overview of existing transition-to-home interventions and their efficacy in categories of intervention types. We will not predefine specific outcomes, but we will describe, assess and summarise the outcomes of the included studies. All reported outcomes will be recalculated to the scale of standardised mean differences. Meta-analysis will be performed with a random-effects model to account for expected large between-study heterogeneity.
No ethics approval was necessary for this study. The aim of our systematic review is to provide a comprehensive overview of current interventions designed to improve the transition from hospital to home. Subsequently, it will show the individual effect of these interventions and pool effect sizes if possible. The most promising interventions will then be combined and used as the basis of a subsequent randomised controlled trial of a new transition-to-home intervention. The results of the study will be published in a peer-reviewed journal.
The systematic review protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) on 30 August 2023. PROSPERO CRD42023455401.
The Federal Statistical Office of Switzerland states that 0.8% of births in 2021 were born very premature, that is, below 32 weeks of gestation.^1^ This rate is comparable with other countries in Europe.^2^ According to the WHO, preterm-born infants are at risk for severe morbidity and death.^3^ Their risk of mortality is 2–10 times higher compared with term-born infants.^4^ Consequently, the birth of such an infant and the following hospital stay lead to emotional distress for the parents.^5^
Very preterm-born infants have a higher risk of adverse outcomes during their development compared with moderate and late preterm-born and also term-born infants. These adverse outcomes affect different organ systems and can result in respiratory problems, hearing and visual impairment, gastrointestinal morbidities, neurodevelopmental disabilities and cognitive impairment.^6^ During neonatal hospital care, infants are constantly monitored and observed by medical staff. During this period, many interventions aim at improving parent–infant interaction (eg, Kangaroo care, creative music therapy and breastfeeding), and family-centred care is performed in many neonatal units. The transition from hospital to home constitutes a vulnerable period for parents and their preterm-born infants. While coming home is long-awaited and can be a relief for families, it is also associated with an interruption of continuous or regular monitoring and highly specialised professional care.
Accordingly, a study conducted by Aydon et al showed that parents of preterm-born infants felt anxious and not prepared regarding the transition to home.^7^ Furthermore, parents of very preterm-born infants often experience substantial stress which may also lead to symptoms of posttraumatic stress disorder.^8^ Correspondingly, in the ‘recommendations for care of the preterm or low birth weight infant’, the WHO states that parents of preterm-born infants should receive additional support. This support should continue during and after the transition to home. They also stated that the transition-to-home preparations had positive effects on nutrition, parent–child interaction and parental well-being.^4^
Due to the importance, complexity and heterogeneity of the topic, we aim to provide an overview of existing transition-to-home interventions for very preterm-born infants by conducting a systematic literature review and to summarise the efficacy of these interventions for different outcomes.
We aim to assess the existing literature of interventions focusing on improving the transition to home process for very preterm-born infants and their parents. These interventions may occur before, during or after hospital discharge and involve a subsequent follow-up assessment. If possible, we will compare these results with standard of care or no structured transition. Because our main goal is to work out an overview, we will not predefine or preselect possible outcomes.
The Preferred Reporting Items for Systematic review and Meta-Analysis Protocols (PRISMA-P) checklist was used as a basic framework for this protocol. The checklist is added as online supplemental file 1. The systematic review protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) on 30 August 2023 (registration CRD42023455401). As of today, the screening of titles and abstracts is finished, full-text screening has started and, in parallel, data extraction has started.
All studies that report transition-to-home interventions for very preterm-born infants (born before 32 0/7 weeks of gestation) and their parents.
Any randomised controlled trial (RCT), observational cohort study, prospective study, retrospective study, case series with more than 10 cases, case–control study and systematic review will be included. Studies on case series with less than 10 cases, conference abstracts and preprints will be excluded.
We will include studies published from 1990 onwards, as this marks the starting point of the OSIRIS trial, a pivotal event that significantly increased the likelihood of survival for very preterm-born infants.^9^
There will be no restrictions regarding language. During the full-text screening, studies in foreign languages will be translated with the translator program DeepL (www.deepl.com). The data extraction and quality assessment of these studies will be done with the help of a fluent speaker of the respective language and with a medical background.
Studies that report transition-to-home interventions for very preterm-born infants (born before 32 0/7 weeks of gestation) and their parents will be included. Studies on full-term infants and on very preterm-born infants who are still in the hospital will be excluded. Studies with a mixed population of infants born before 32 0/7 weeks and infants born afterwards will be included in the meta-analysis only if the subgroup of very preterm-born infants is reported separately, and intervention effects can be extracted in the subgroup of these infants. Studies focusing on infants with major congenital anomalies will be excluded. Studies on infants with low birth weight and other studies with heterogeneous populations where no specific information on very preterm-born infants can be retrieved will be excluded.
This includes any description of a structured hospital to home transition intervention for very preterm-born infants and their parents. Those transition interventions can begin during the period in which the infant is still hospitalised in the Neonatal Intensive Care Unit (NICU) or later, but they have to be carried on throughout the transition to home process.
Interventions will be categorised into different groups according to their focus. Thus, we will have categories focusing on parental advice and education. Those might include individualised care plans, such as family-centred care or parental education via booklet devices or mobile health applications, accompanying families when they are transitioning home. Another group might be about breastfeeding support, including regular face-to-face meetings between healthcare professionals and parents, or parental group sessions, aiming for a higher perseverance with breastfeeding once home. Other categories of interventions would target motor development via physiotherapy programmes, such as Coping and Caring for Infants with special needs.^10^ Finally, some interventions will focus on stimulating overall development, such as Supporting Play Exploration and Early Development^11^ programme.
The main goal of this review is to identify existing interventions improving the transition from hospital to home process for very preterm-born infants and their parents. Hence, we will not focus on or search for specific predefined outcomes, but we will describe, evaluate and, if possible, summarise the outcomes of the included studies for different interventions. Consequently, the search strategy did not include predefined outcomes.
Expected outcomes regarding very preterm-born infants are neurodevelopmental outcomes, commonly measured via development scales such as Bayley, including the cognitive and motor development scale, Griffith or Infant Motor Profile, or Alberta Infant Motor Scale. Other outcomes might include health status, growth and socioemotional competence.
Expected outcomes regarding parents of very preterm-born infants are parent–child interaction, which can be measured by standardised scales such as the Mother and Baby Interaction Scale,^12^ parameters of parental feelings of preparedness for discharge, the feeling of well-being or parental stress (Parental Stress Index Scale).
Descriptive or qualitative outcomes, such as parental interviews, will be considered in the review even though data extraction will not be possible for those outcomes. If outcomes are measured at discharge, we will include the intervention only if the aim is to enhance the transition to home. This could be the case with family-centred care, where parents benefit from an individualised programme aiming at preparing a smooth discharge-to-home process. Any health economic evaluations of transition to home will be considered as secondary outcomes.
We will include follow-up assessments of outcomes for a maximum duration of 2 years, but the intended follow-up time interval is 12 months. We aim to extract the follow-up time point closest to 12 months if multiple follow-up time points should be reported in a study. The different lengths of follow-up will be meta-analysed in subgroups of up to 12 months and more than 12 months.
We considered the challenge of extracting relevant outcomes from each publication due to the significant variation of reported outcomes between publications. Therefore, for every study that meets the inclusion criteria, we will extract the primary outcome and, if applicable, one secondary outcome.
In detail, the process can be described as follows. If a study has exactly one primary outcome and one secondary outcome reported, these will be extracted, independent of the type of outcome. If one of the outcomes of a study is the Bayley score, at 12- or 24-month follow-up, the cognition subscale is extracted as the primary outcome and the motor subscale is extracted as the secondary outcome as these scores represent standardised and clinically relevant outcome measures in the focused time frame of this systematic review. Alternatively, if the Griffith total score at 12 or 24 months or any other cognition scale is reported, it will be extracted. If information is available, parent–child-related interaction scores are extracted as the secondary outcome. The distinction between primary and secondary outcomes is made for data extraction purposes, but meta-analysis will be performed independently and the information on whether an outcome was primary or secondary in the individual study will be given as additional information.
We will search the following biomedical Cochrane Library, Medline (EBSCOhost), CINAHL (EBSCOhost), EMBASE (embase.com) and PsycINFO (EBSCOhost). We developed a search strategy for the following preterm AND transition to home AND parental support/need/interventions. For each concept, we will use a combination of subject heading terms and free text words. We will not apply any restriction to language or study types. Benchmark studies will not be used. To minimise the risk of bias, we will not use any search terms related to possible outcomes. We are not planning on re-running the search before the final analysis. Unpublished studies will not be sought.
The search strategy was developed by an experienced information specialist and librarian (MG) in collaboration with the other authors. The database search has been carried out by MG who is experienced in the literature search for systematic reviews. The individual search strings for all databases, including the number of hits, were included (online supplemental file 2). Furthermore, the search was registered on searchRxiv (https://searchrxiv.org/) and it is accessible via the following
The search was carried out on 5 April 2023. The time limit applied is from 1990 to 2023.
After searching the databases and deduplication, the identified references were uploaded to Covidence. Covidence is a software for managing systematic literature reviews (www.covidence.org) where the study selection will take place. The abstract screening will be done by two members of the research team (VMP and VE). The first 50 abstracts will be screened by both parties using Covidence. Afterwards, discrepancies will be revised, and a third party (BL, UKK and UH) will be involved if disagreements should occur. On satisfactory agreement, the remaining studies will be divided and screened by only one person to keep the workload on a manageable level. Otherwise, we will continue double screening and will have another revision after an additional 50 abstracts have been screened. Emerging questions and uncertainties will be discussed with each other and a third party. The full-text screening will be done by two members of our research team (VMP and VE). They will not be blinded to the names and affiliations of the authors and journal titles. Questions raised will be discussed with the research team to reach a consensus. In the case of a publication containing a narrative or systematic review in which not all included studies coincide with our inclusion criteria, we will check if the compatible studies are part of our data set. If the full text cannot be found, the corresponding author will be contacted to obtain the full text. If the authors do not reply within 1 month, the reference will be excluded. If the full text is not available and the contact information regarding the corresponding author cannot be found, the study will be excluded. We will manage multiple reports from the same study by documenting that there are overlapping publications. Data of these reports will be handled as stated in the Data extraction section. During full-text screening, excluded studies as well as their reason for exclusion will be listed in a document.
The data from the included studies will be extracted using a piloted data extraction form (data extraction form from Covidence will be exported into a comma-separated value file). The members of our research team with biostatistics/methodological expertise (UH, LB and SS) will pilot these forms on five studies and adapt them if necessary. The proposal of the draft of the data extraction form is added as online supplemental file 3. Regarding the study population, we will extract information on the study population, including gestational age, number of twins, the total number of children investigated in the study and further demographic information such as birth weight, age of the mother at birth, hospital length of stay, the age and the corrected age of the children at baseline at the time of intervention. We will record the described interventions and, if applicable, also information on the standard of care, control intervention or placebo. For more detailed information on the data extraction, we refer to our proposal of the draft of the data extraction form. The data extraction will be done independently by multiple members of the team (VMP, UH, UKK, LE, LK and LB). The data extraction will follow guidance from a recent paper.^13^ Any questions arising will be discussed within the research team until a consensus is reached. If we have multiple reports from the same study with similar outcomes, we will use the report that corresponds best to the 12-month follow-up duration for meta-analysis. If there are two reports with the same time interval as compared with the 12-month follow-up duration (eg, follow-up at 9 and 15 months), we will incorporate the report of the later time interval.
For data extraction, publications must report on a minimum set of information to be eligible for meta-analysis. The minimum set of information includes a definition of a follow-up time frame, and the outcome must have been operationalised. Therefore, narrative summaries of findings or qualitative interviews would not be considered for data extraction and meta-analysis.
If relevant data is missing in the full-text article, the corresponding authors will be contacted to obtain this information. If the authors do not reply within 1 month, we will use the published data and note that relevant data are missing.
Assessment of methodological quality and risk of bias (RoB) of included the quality assessment and RoB assessment will be performed by the members of the research team who extracted the data of the respective article (VMP, LB, UH, UKK, LE and LK). A second party will be involved if questions arise. It will be done at the study level. For observational studies and case series, the appropriate SIGN checklist (https://www.sign.ac.uk/what-we-do/methodology/checklists/) will be adapted by members of the research team with biostatistics/methodological expertise (UH, LB, SS) and will be used to perform the RoB assessment. The studies will be coded as ‘high quality’, ‘acceptable’ or ‘low quality’. The quality of the study (apart from ‘unacceptable/reject’) will not be used as weight in the meta-analysis, but it will be taken into consideration when discussing the results. For RCTs, Version 2 of the Cochrane risk-of-bias tool for randomised trials (RoB2) (https://methods.cochrane.org/bias/resources/rob-2-revised-cochrane-risk-bias-tool-randomized-trials) will be used to perform the RoB assessment. The RoB will be coded as ‘low’, ‘some concerns’ or ‘high’. The proposal of the draft of the RoB assessment form is added as online supplemental file 4.
On the study level, descriptive statistics will report mean and SD for continuous variables, as well as median and IQR for ordinal or skewed continuous variables. Frequencies and percentages of the total will be reported for categorical variables. The percentage of missing data will be reported with heat maps separately.
For evidence synthesis, a random-effects model will be assumed due to the large expected between-study heterogeneity. The organisation of the meta-analysis will be according to the type of intervention but will be flexible regarding the outcome measures. The time horizon of the outcome will be addressed in two ≤12-month follow-up and >12-month follow-up. The following categories of interventions have been identified a breastfeeding and nutrition, parental counselling, individualised family-centred care, physiotherapy, stimulation of global development, e-health, hospital-based intervention and home-based intervention.
Depending on the outcome distribution, different effect measures will have been reported in the individual studies, for example, risk ratios or risk differences for binary outcomes, and mean differences for continuous outcomes for studies including two groups. To be able to pool results across studies, the effect measures need to be unified before they can be pooled. The unification includes different outcome distributions, ranges of scales and direction of effect. We will use standardised mean differences (SMD) as the unifying effect measure, and the SMD will also be calculated for binary outcomes. Across meta-analyses, we will guarantee to unify the direction of the intervention effect, depending on the underlying outcome scale for valid interpretability of the direction of findings across interventions and outcome scales. Eventually, SMD>0 indicates favourable outcomes in the transition-to-home intervention group across all meta-analyses. If necessary, we will also summarise studies that report on pre- and postimplementation of specific interventions as well as single-group studies.
In the meta-analysis, the heterogeneity variance parameter τ^2^ will be estimated using restricted maximum likelihood, and DerSimonian-Laird will be used for sensitivity analysis. Results of the studies will be pooled if three or more studies report on the same intervention. If fewer studies report on the same intervention, we will report a narrative synthesis and avoid a summary estimate. The Hartung-Knapp-Sidik-Jonkman method will be applied to adjust the standard errors of the estimated coefficients to account for the uncertainty in the estimate of the amount of residual heterogeneity.^14^ If the number of studies in a meta-analysis is small, Bayesian methods may be used in an additional sensitivity analysis. Prediction intervals will be calculated. We do not plan to exclude studies because of the high risk of bias, but the risk of bias will be evaluated by meta-regression if the number of studies is large enough. The R programming language will be used for data analysis^15^ in combination with the metafor package for meta-analysis. All analyses will be done in a fully scripted way and using dynamic reporting for reproducibility.
For single-arm studies, descriptive results will be reported in graphical and tabular forms. For RCTs, or observational cohort studies with a comparator group and at least two arms, we will use recalculated SMD for binary outcomes, and SMD for continuous outcomes as effect measures.
If there is enough information, we plan to form and analyse the following parental sex, sex of the infant, different durations of follow-up, duration of hospital stay before the transition to home and adjusted age of the infant when transitioning to home. If possible, we will perform meta-regression for the duration of the hospital stay and parental sex. We expect variables describing the neonate during hospital stay (including major complications) to be balanced across treatment groups in randomised trials. In observational data, however, there could be imbalances across treatment groups. We will try to mitigate a potential bias using meta-regression to adjust the estimated treatment effect of the intervention regarding these variables. The analysis will be limited, however, by the granularity of the information across studies.
Funnel plots will be used for the evaluation of publication/reporting bias. Meta-regression will be applied if the number of studies is large enough to quantify potential sources of meta-bias.
The GRADE framework (https://bestpractice.bmj.com/info/us/toolkit/learn-ebm/what-is-grade/) will be used to present the summary of evidence in those instances where a pooled summary effect can be calculated for identical or similar interventions as compared with standard-of-care.
If we have to make amendments to the protocol, the date, the reasoning and an exact description of the change will be provided in the final report of the systematic review.
Before initiating the systematic review, a focus group was formed, including parents of very preterm-born children, physicians, psychologists, therapists, social workers and a statistician. The focus group was interviewed on 1 February 2022, for specific needs, difficulties and interventions that are offered across institutions. This expert meeting was conducted and recorded online via Zoom and resulted in a questionnaire to obtain further feedback and information on specific demands on an individual and institutional basis.
No ethics approval was necessary for this study. After finalisation of the study, every endeavour will be undertaken to publish the manuscript in a peer-reviewed journal. Data will be made available in a suitable repository.
We conclude that the transition to home process constitutes a critical period for the development of a very preterm-born infant. Consequently, various studies and reviews are performed regarding this subject. For example, Bedwell et al are planning to perform a systematic review regarding ‘Interventions to support parents, families and carers in caring for premature or low birth weight infants in the home’. In contrast to our review, they will not focus on interventions that target the transition process from hospital to home, but on interventions that are delivered when the infant is already in the home setting. Furthermore, they will also include low-birth-weight infants that constitute a different patient group than very preterm-born infants, which we will include in our review. They will focus on studies published in the past 20 years, whereas we will include studies published from 1990 onwards.^16^ Aagaard et al published a systematic review protocol focusing on the parental experiences regarding the transition to home after discharge from the NICU. They included all infants that were hospitalised in a NICU and defined a narrower time frame (2000–2014) in their search strategy compared with our systematic review.^17^
The results of this systematic review will aid and support clinical decision-making by providing an overview of existing interventions for the transition from hospital to home process for very preterm-born infants and their parents, and to obtain an estimate of their efficacy on a unified scale.
The initial screening has revealed significant heterogeneity across studies concerning the types of interventions, the outcome measures and the follow-up durations. The different interventions can be categorised into subgroups according to their focus areas, for example, breastfeeding, e-health devices, hospital-based or home-based interventions, including parental education, as well as parent–child interaction programmes. A challenge of this systematic review will be to find a balance between individual study details and reporting of groups of intervention effects on homogeneous outcome measures in similar follow-up time frames. For this, we strive for a global perspective while keeping the details of the individual publications in focus. The risk of bias assessments will also reveal the general level of bias, or quality in this area of research. It can be expected that through critical appraisal of all studies included in the systematic review, areas that need improvement will be identified, and these will be specific to the study design, that is, randomised trials or observational studies.
We expect the results of the systematic review to provide guidance for the effectiveness of different interventions on relevant outcomes for infants and their parents. The intervention(s) with the most promising effects, based on the GRADE evaluation of evidence, will then be the basis of an RCT which will be designed by members of our research team.