Efficacy of parent-infant psychotherapy compared to care as usual in children with regulatory disorders in clinical and outpatient settings: study protocol of a randomised controlled trial as part of the SKKIPPI project

Parent-Infant Psychotherapy (PIP) is a manualised focus-based short-term psychodynamic psychotherapeutic intervention that we have developed (Schlensog-Schuster F, Koch G, Ludwig-Körner C: SKKIPPI Study manual for Parent-Infant-Toddler Psychotherapy (PIP), Unpublished), which includes the children and their primary caregivers ([11, 25]; Schlensog-Schuster F, Koch G, Ludwig-Körner C: SKKIPPI Study manual for Parent-Infant-Toddler Psychotherapy (PIP), Unpublished). Generally, the therapist works with the parents (or other important caregivers) and the infant in joint sessions. In special situations (for example, if the mother has depressive symptoms) there may be a few sessions with one parent alone, in order to address important themes that cannot be addressed in joint sessions. The intervention focuses mainly on the meaning the parents ascribe to their child’s symptoms. The therapist observes the parent-child interaction and tries to link his/her observation to the parental beliefs and interpretations. The therapeutic focus includes the family history, parental thoughts, symptoms, and the therapeutic relationship that develops during the sessions. The aim of the therapist’s intervention is to help the parents to understand and to ‘contain’ the child’s affective states. The intervention builds on parental self-reflection and supports the parental ability to mentalise. PIP also aims to strengthen specific parental ego functions.

The focused psychodynamic psychotherapy is conducted in 12 sessions, twice a week over a period of 6 weeks. The course of PIP is divided into three phases: the initial diagnostic phase, the second phase of the focused intervention, and the third and final phase of integration. The therapists are trained to consider the parents’ couple relationship, the personalities of the parents, and possible conflicts in the mother-father-infant triads. The therapists try to create a holding environment for parents and infant within the therapeutic sessions. The holding therapeutic function helps participants to clarify problems, regulate affects, engage in self-reflection, and change their perspectives, and thus provides support for interpersonal resources in the parent-child triad. Psycho-education and the use of video feedback are additional therapeutic options for the intervention, which is oriented towards the mother’s psychodynamic, structural level. Video feedback can be used if there are maternal developmental deficits and a space of self-perception with the child cannot be created in inner presentations and phantasies.

At the end of the intervention, the therapist encourages the parents to reflect upon their experience of the therapeutic process and to develop strategies for using the therapeutic experience in their everyday life. The treatment technique is manualised and follows guidelines according to the setting. Every therapist undergoes specialised training according to the manual before the first session to ensure adherence to the treatment techniques described within it. All therapists are either psychodynamic psychotherapists (psychologists and medical doctors) or psychologists and medical doctors in advanced psychodynamic psychotherapy training. The therapies are supervised by experienced PIP therapists at regular intervals in order to secure treatment fidelity (Fig. 2).

Patients in the control group receive standard medical, pedagogical, and/or psychotherapeutic care for 6 weeks, as usually provided by the German health care system. Ethical considerations mean the control group could not be designed as a no treatment control. In the inpatient setting, more care might be provided than in outpatient settings. This could include counselling, physical or psychotherapy, and other psychiatric-therapeutic interventions. Outpatient settings may be either therapeutic or pedagogical care. Because of possible heterogeneities in the number and intensity of CAU treatments compared to PIP, all individual interventions will be documented in detail.

The primary outcome will be maternal sensitivity measured after 6 weeks of intervention. Videotaped dyadic sessions of free-play interaction will be coded using the subscale ‘sensitivity’ of the emotional availability scales (EAS) [30]. EAS assesses the quality of the mother-child interactions using a seven-point Likert scale for maternal sensitivity. Independent and reliable coders blind to treatment allocation rate all the 15- to 20-min video scenes.

The most important secondary outcome will be the child’s attachment style, measured using the Strange Situation Procedure (SSP) for children up to 20 months [31]. The SSP is a standardised procedure recording eight episodes to rate the child’s attachment quality to his or her caregiver. Attachment security is rated by independent coders blind to treatment allocation. The classification can be secure (B), avoidant (A), or ambivalent (C). Disorganised (D) attachment qualities will also be scored.

The attachment style of children older than 20 months is assessed using Attachment Q-Sort (AQS) [32]. The AQS evaluates the degree of attachment security for infants aged 12 to 48 months, and is a one- to two-hour procedure in which the child and its mother are observed in their home environment. The infant’s behaviour is evaluated by the observer for 90 standardised Q-Sort items, which are compared to a typical profile of securely attached children. The AQS varies on a scale between − 1 and 1, and does not reflect the variety of attachment categories, as the SSP does.

To quantify the child’s psychiatric symptoms, a multi-method approach is used, with reliable and valid questionnaires and a structured interview we have developed using the diagnostic criteria of the DC: 0–5 Axis 1 [5, 33]. DC: 0–5 interviews are conducted at 6 weeks follow-up, and 12 months follow-up. Mothers also complete the Child Behaviour Checklist (CBCL) [34] to detect the child’s emotional and behavioural problems. This 100-item screening instrument is validated at a child’s age of 18 months and re assessed at 6 weeks and 12 months follow-up. The child’s mental health symptoms are measured using the German questionnaire Crying, Feeding, Sleeping (SFS) (Gross S, Reck C, Thiel-Bonney C, Cierpka M: German version of the Questionnaire on Crying, Feeding, Sleeping, Unpublished). The SFS assesses problems in the child’s behavioural regulation and dysfunctional communication patterns, and the primary caregivers’ stress symptoms. The questionnaire is assessed at 6 weeks follow-up.

Maternal mental health problems are assessed using the Brief Symptom Checklist (BSCL) [35], a validated questionnaire to measure adult psychopathological symptoms, at 6 weeks follow-up. A further questionnaire assessing mental health status, the Anxiety Screening Questionnaire (ASQ-15) [36], covers all anxiety disorders diagnosed by DSM IV and ICD-10. Mothers complete the questionnaire at 6 weeks follow-up. The Scale for Impulsive Behaviour and Emotional Dysregulation of Borderline Personality Disorder (IES-27) [37] is a screening instrument to measure typical patterns of BPD. The IES-27 questionnaire is assessed at 6 weeks follow-up. Symptoms of maternal depression are measured using the Edinburgh Postnatal Depression Scale (EPDS) [38] at all measurement points. To explore psychiatric disorders, the Mini International Neuropsychiatric Interview (M.I.N.I.) [39], a structured diagnostic interview, which assesses 20 psychiatric disorders by DSM IV and ICD-10 diagnosis, is used. The M.I.N.I. assesses possible maternal clinical diagnoses and will be administered by trained interviewers at 12 months follow-up.

Maternal attachment style and reflective functioning, i.e. the capacity to mentalise, are considered essential moderators in the present design and are assessed using the Adult Attachment Interview (AAI) [40]. The AAI is a semi-structured interview with precise psychometric properties to evaluate the maternal ‘state of mind’ with respect to attachment. The AAI is conducted by trained interviewers, transcribed, and subsequently scored by reliable and independent coders blind to treatment allocation.

The standardised German developmental test ET 6–6 R [41] assesses the level of social-emotional, cognitive, motor and language development of children aged between 6 months and 6 years. The developmental status of the child is evaluated at all three measuring points. In addition, and at the Leipzig study centre subgroup only (due to availability reasons), the Bayley Scales of Infant and Toddler Development (Bayley-III) [42] is assessed. Bayley-III depicts the developmental level of infants aged between 0 and 42 months in a differentiated manner based on various subtests.

The Parental Stress Index (PSI) [43] is used to measure maternal parental stress, and the Parental Reflective Functioning Questionnaire (PRFQ) [44] to measure the mother’s self-reported ability to mentalise in the context of her relationship to her child. Both questionnaires are assessed at 6 weeks, and 12 months follow-up.

Additionally, the five further subscales of the EAS based on the 15-min videotaped mother-child interactions (maternal dimensions: structuring, non-intrusiveness and non-hostility; and child dimensions: responsiveness to parent, involvement with parent) are analysed as secondary outcomes. Free-play sessions are recorded at all measurement time-points and rated by reliable and independent coders blind to treatment allocation.

Treatment Adherence is assessed using an in-house questionnaire with 30 items to evaluate the methods and interventions applied by the PIP therapist after each session.

Another questionnaire we have developed also assesses the economic consequences of using health and social care, and requests for early intervention services. Mothers complete the questionnaire at 12 months follow-up. The utilisation documented will be analysed to assess the associated costs (Fig. 3).

Socio-demographic and clinical information are described at baseline using frequencies and percentages (e.g. categorical variables) or means and standard deviations (e.g. continuous variables), and treatment conditions compared to assess potential baseline differences. The statistical analyses are based on all mother-child dyads included in the study after randomisation (Fig. 1), according to ITT (Intention to Treat) principles. The data to be used in all analyses (unless otherwise stated) is defined as the Full Analysis Set (FAS) and includes all mother-child dyads after randomisation, irrespective of the treatment received. Data from mother-child dyads who drop out of the study are included in the FAS up to the date they withdraw. The Per Protocol (PP) population is a subset of the FAS and includes mother-child dyads who conform to the study plan (i.e. sufficient treatment adherence, no major protocol violations). The safety population is defined as mother-child dyads who have participated in at least one intervention session, and is analysed according to the treatment they received.

A secondary analysis of the primary endpoint will also be carried out using the ANCOVA model based on the PP population. Additional moderator analyses of the efficacy of PIP for a change in maternal sensitivity (EAS) will be conducted with treatment setting as the moderator, i.e. by additionally adding a ‘treatment group’ * ‘treatment setting’ interaction term to the model. Further moderator analyses regarding the efficacy of PIP will comprise maternal attachment type (AAI attachment) or maternal reflective functioning (AAI RF) as moderators.

In addition, models incorporating multiple imputation methods based on the PP population and a hierarchical regression model (including the therapist as a grouping factor and therapist’s adherence as covariate) will be used for sensitivity analyses of the primary outcome. To investigate the longitudinal change of the primary endpoint, linear mixed models (LMMs) incorporating both measurement points for each outcome (FAS population) will be computed. To explore further effects of additional predictors (such as the therapists carrying out the interventions), the models will be compared based on model fit values (AIC, the Akaike Information Criterion), with these predictors included in the model term.

Primary analyses of the secondary endpoints replicate the computation of the models for the primary endpoint, as either ANCOVAs or logistic regressions (depending on the continuous or dichotomous scale of the respective endpoint). Group differences in the children’s attachment organisation (SSP/ AQS, binarised evaluation: secure/ insecure attachment), as well as all responder analyses (EAS sensitivity, EPDS, PRFQ, PSI at T1 and T2; SFS at T1, CBCL at T1 and T2), will be computed as logistic regressions adjusted for the stratification variable treatment setting with presentation of odds ratios. Should no cut-off values be available, dichotomous response variables for these measures will be computed using the Reliable Change Index (RCI; RCI > 1.96) [45]. Linear mixed models will be used to analyse the change in the secondary endpoints (maternal stress level, reflective functioning (PRFQ), and infantile symptoms) during the course of the study (12-months follow-up) across all measurement points.

Additional sensitivity analyses will be performed for health economic parameters as secondary endpoints, varying unit-cost assumptions for documented health care utilisation within realistic ranges. All secondary analyses of the primary and secondary endpoints are considered exploratory.

After written informed consent and randomisation, participants can be excluded from the analyses only if there has not been any intervention, no data have been collected after randomisation, or at the participants’ express request after withdrawing from the study during the data collection phase. Primary statistical analyses of the primary and secondary endpoints are based on the FAS population without the imputation of missing data. For a sensitivity analysis, missing values for the primary endpoint will be imputed using Multiple Imputation methods for linear regression based on Markov chain Monte Carlo (MCMC) simulations. Imputation variables will be selected blinded for the intervention groups, and at least 5000 iterations and 100 imputations will be calculated.