Strong families: a new family skills training programme for challenged and humanitarian settings: a single-arm intervention tested in Afghanistan

The Strong Families programme [16, 17] is a three sessions (7 components requiring 5-h of invested time by the families in total; Table 2) group intervention attended by children and their primary caregivers over a time span of 3 weeks (one session per week). In week one, a group of 12 caregivers meet for the 1-h caregiver pre-session. In weeks two and three, the same 12 caregivers meet again in one room, and their 12 children meet in parallel in another room for the child sessions. After these 1-h parallel sessions, all caregivers and children immediately meet in one room for another 1-hour for a joint family session. The session of week one (caregiver pre-session) explores parents challenges and develops ways to better deal with stress. In week two, caregivers discuss the means of showing love while at the same time having and enforcing limits and listening to children, while the children learn how to deal with stress. During the family session they practice positive communication and are encouraged to practice stress relief techniques together. In week three, parents learn to encourage good behaviour and discourage misbehaviour, while children explore rules and responsibilities and think about future goals in addition to the important roles their caregivers play in their lives. In the final family session, caregivers and children learn about family values and practice sharing appreciation to each other.

Cultural adaptation of the Strong Families programme to the Afghan context was assured through a 3-day working group with representatives from the Afghan Ministries of Public Health (MOPH), Labour and Social Affairs, Counter Narcotics (MCN), Education (MOE), Afghan NGOs (non-governmental organisations), the programme developers and UNODC.

The translation of the training material for facilitators and questionnaires for the participants into Dari was performed by a qualified and experienced translation company and reviewed by local researchers.

An open, pilot feasibility and acceptability trial was conducted with an embedded effectiveness evaluation including a prospective collection of outcome data assessing changes in children’s behaviour, parenting skills and family adjustment in caregivers.

Participants were selected based on being a female caregiver to a child between the age of 8–12 years. The cultural adaptation team, consisting of Afghan representatives who reviewed the materials, pointed out that religious and cultural norms do not allow for men and women, who are not related, to mix together. Respecting the cultural context, and given the stage and intent of the current pilot, it was decided to target either males or females caregivers at this time. As females are most often the primary caregivers in Afghan families, female caregivers and their children (of either gender) were chosen.

Sampling was opportunistic, using a ‘universal’ approach, in which facilitators recruited families from the general population, not targeting specifically those with a particular risk. Inclusion criteria included speaking Dari, willing to take part in the programme and being in the town for the duration of the whole study and measurement meetings. Families that had already taken part in another family skills training programme in the past 6 months or where the caregiver lived separately from the child were excluded from the programme.

The pilot study took place in three towns in Afghanistan, namely in Kabul, in Mazār-i-Sharīf (Balkh) and in Herat between July and October 2018. Based on UNODC’s mandate, and under the umbrella of the National Drug Control Strategy, UNODC approached MCN and MOPH to facilitate the implementation of the programme in Afghanistan who later expanded the collaboration with MOE, the Ministry of Women Affairs and the Ministry of Health. As displayed in Fig. 1, families were recruited via two high schools and one 100 beds Drug Treatment Centre (DTC) for women and children in Kabul, a 150 beds drug demand reduction (DDR) hospital in Balkh and a women DTC in Herat.

The schools were selected by the MOE based on the main criteria of having easy access of the families to the school and the provision of two class rooms for the programme to run. Caregiver information sheets were distributed to all children aged eight to 12, who attended the participating schools, to take home to their caregivers. Female caregivers were invited via a self-referral process to attend an information session where they were given further verbal and written information and questions were answered. Caregivers were asked to phone the school within the next 4 days if they wanted to participate. The first to call the school were invited to take part in the pilot study and attended a baseline measurement session the following week in which written informed consent was obtained prior to data collection.

In Afghanistan, DTC provide health education and services to the most affected families, are well connected with local communities and most families have access to them. The women DTC have four components of service delivery: home-based, outreach, outpatient and residential. The home-based part raises awareness and provides education and information to surrounding affected families whereas the outreach component raises awareness and provides screening and brief intervention in the communities, particularly focussing on the identification of the vulnerable and at-risk women and children and providing information and education program at home. The outpatient and residential parts of the DTCs are for drug treatment and rehabilitation of women suffering from drug use disorders. Further, the existing Afghan drug demand reduction policy and strategy recommends integration of the drug prevention in health services. DTC were included so as not to bias against families in which children were out of school. Again, female caregivers from the home-based or outreach component of DTC were given a caregiver information sheet and invited via the same self-referral process to attend the information session at a participating school, and the same recruitment process took place. Noting that the recruitment strategy was the same in both centres, all interested families were invited. The DTC were implicated given their interest to be involved in prevention responses. In general, we did not formally assess clinical diagnoses in caregivers or their children, however at the first session, caregivers were handed out leaflets with information as to where to get help in case they observe more severe stress reactions, severe trauma or other physical, mental or sexual health issues in themselves or their children.

Seventeen facilitators were selected through schools and health settings (13 females and four males), local NGOs, MCN and MOPH. Their backgrounds were mixed, some teachers, some caregivers who had previously taken part in family skills programmes, psychologists and social workers. This mix in backgrounds was important as the programme is designed to be run by lay people without a particular expertise. Facilitators took part in a 3-day training programme in New Delhi-India, delivered by the developers of the programme who are experienced international trainers. Clarification of questions and more practising of difficult parts was assured via a national refresher training prior to roll out to families, as well as through remote monitoring via weekly Skype calls between the developers from the UK, UNODC staff in Vienna and coordinators in Kabul to ensure coverage of all field questions and adherence to protocol.

This pilot study has been thoroughly reviewed and approved by the UNODC Drug Prevention and Health Branch in the Headquarters office of Vienna and the national field office in Kabul as well as the associated national ministries (Afghan Ministry of Counter Narcotics, Ministry of Public Health, Ministry of Labour and Social Affairs, Ministry of Education, Ministry of Women Affairs and Ministry of Health) and NGOs that supported the programme as alternative to ethics committee review. The Strong Families programme has been thoroughly analysed and, after approval, has been integrated in the National Drug Demand Reduction policy 2019–2023 for Afghanistan, as well as the National Drug Demand Reduction Strategy 2018–2022. Further, the donor to the programme development and implementation, US-INL (Bureau of International Narcotics and Law Enforcement Affairs) under the U.S. Department of State, had reviewed and approved the proposal before initiating the trial. The piloting was performed in accordance with the ethical standards of the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

The Chief Investigator and the research team assured the confidentiality of participants in accordance with the Data Protection Act 1998. Each participant was assigned a unique identification number to ensure matching of all questionnaires. All data collected as part of the trial were treated as confidential and were only be viewed by members of the trial team; anonymised data were used wherever possible.

All caregivers completed and signed a consent form at the first evaluation meeting, as all children were under the age of 16 years, they consented as a legal guardian for their children. No measures were taken directly from children, the reported results are from caregiver measure reports, thus children were not required to complete an assent form. In addition to written information being provided in the form of the Participant Information Sheets for the caregiver, participants were provided with a verbal explanation of the evaluation method at the first meeting and again when they attended the first data collection session.

Data on demographics, emotional and behavioural difficulties of children and parental skills and family adjustment measures were collected from caregivers through self-administered questionnaires.

Two outcome measures were completed by participants at baseline (i.e. 1 week before intervention delivery) (t1) and 2 weeks (t2) and 6 weeks (t3) after intervention delivery. These were the paper-based Strengths and Difficulties Questionnaire (SDQ) and a Parenting and Family Adjustment Scales (PAFAS). A standardized Family Background Questionnaire (FBQ) was completed at t1 to collect demographic characteristics.

The SDQ is a commonly used short screening tool to assess children’s behavioural, emotional, and social issues over the last six months. It is available in over 40 different languages and is frequently used for research purposes to examine children’s mental well-being. The advantages of the SDQ were its compact format (relative to the previously long-established and highly respected Rutter behavioural screening questionnaires) in covering the strengths as well as difficulties in inattention, peer relationships, and prosocial behavior [31]. It has shown good psychometric properties and has been used in Afghanistan previously [32]. The SDQ examines 25 attributes, each rated on a 3-point scale ranging from 0 (“Not True”) to 2 (“Certainly True”). The answers can be summed into five subscales (ranging from 0 to 10 points each), including items such as emotional symptoms (e.g. “Often unhappy”) and conduct problems (e.g. “Often fights with other children”). The Total Difficulties score (ranging from 0 to 40 points, with higher scores indicating higher levels of difficulties) is calculated from four of the subscales excluding prosocial behaviours [31]. We used the cut-points for the 4-banded categorization of the SDQ scores to classify continuous measures into “close to average”, “slightly raised/lowered”, “high/low” and “very high/low” risk [33].

The PAFAS is a brief 30-item, user-friendly questionnaire measuring parenting practices and family functioning, which are known risk or protective factors for child emotional or behavioural problems. PAFAS was developed to assess changes in parenting skills and family relationships before and after public health and individual or group parenting interventions. It comprises two scales: (i) Parenting, measuring parenting practices (e.g., descriptive praise, logical consequences) and the quality or parent-child relationship (e.g. level of reciprocal warmth, parental satisfaction with the relationship to the child) and (ii) Family Adjustment, measuring parental emotional adjustment (e.g. level of stress, depression and anxiety experienced by a parent in his or her role) and positive family relationships (e.g. supportive and conflict-free family environment) and parental teamwork (e.g. social support received from the partner in the parenting role) [34]. The PAFAS subscales have shown good internal consistencies in two Australian samples (ranging from .70 to .87) and satisfactory construct and predictive validity [34]. Furthermore, the measure has been validated in other cultures, such as Panama [35] and China [36]. In both studies, the factor structures in the original PAFAS measure were mostly retained with fewer items and adequate internal consistencies (ranging from .50 to .82 and from .65 to .95 for Chinese and Panamanian parents, respectively). PAFAS has been recently used with Arabic speaking families living in political conflict in the West bank [37]. To our knowledge there are no clinically relevant cut-off points available. For sub-analyses purposes, a cut off at the 75th percentile was assumed Participants with scores above the upper quartile (Q3) at baseline, 25% of participants, represent families with higher levels of difficulties.

The implementation process was evaluated based on the methods and fidelity assessment sheets previously described by Segrott et al. [38] In addition we provided questionnaires to be filled in by observers. Basic data on the number of sessions were provided from the coordinator in the field. In each session of the Strong Families programme, an independent observer assessed predefined indicators, and facilitators were asked to complete another questionnaire. Indicators to measure different process evaluation components were modified from the ones used for the SFP 10–14 in the UK [38] to match the Strong Families programme, as indicated in Table 6.

The data was entered in Epidata version 3.1 and analysed using GNU PSPP Statistical Analysis Software, version 1.0.1 and IBM SPSS Statistics version 25. Prior to analysis, data completeness and plausibility checking was assured.

Continuous variables are presented as mean, standard deviation (SD) and minimum and maximum scores. Categorical data was summarized by frequencies and proportions.

A visual inspection of the histograms, Q-Q plots and box plots as well as the calculation of the skewness and kurtosis z-values within a range of +/− 1.96 showed that data were approximately normally distributed.

In the event of normality, to compare means in demographic characteristics, a 2-sample t-test was used for continuous variables, while a chi-square test for categorical data.

For comparison of means or ranks at the different time points (pre-test, post-test and follow-up), a repeated measures ANOVA was used for normally distributed data with post-hoc tests using Bonferroni corrections. In case Mauchly’s Test of Sphericity indicated that the assumption of sphericity had been violated, a Greenhouse-Geisser correction was used. A Friedman’s ANOVA with a Wilcoxon’s Signed Rank test as post-hoc test were used as non-parametric tests.

For potential gender-based effects and to compare SDQ scores at each timepoint between boys and girls we used a 2-sample t-test for normally distributed data or a Mann-Whitney-U test if data were not assumed to be normally distributed. Participants with very high (20–40 points) scores at baseline on the total difficulty scale of the SDQ were analysed separately for each of the subscales to show potential effects on all subscales in people with most difficulties at start point.

For process evaluation data, proportions were calculated out of all data obtained from facilitators and observers.

Overall, 72 families were enrolled in the programme (Fig. 1), 48 in Kabul, 12 in Balkh and 12 in Herat. The five caregivers who were lost to follow up at t3 were significantly younger (mean age 28.2 compared to 36.7 years; t₆₉ = − 2.44; p = 0.017). No other difference in recruitment site, marital status, education or mean number of children was found.

Missing data on individual questions within the FBQ were regarded as minor and non-systematic. Individual PAFAS answers were missing in less than 5% of cases, apart from PAFAS question four (“I threaten something when my child misbehaves”) at time two (7/71 missing), for PAFAS question 15 (“I enjoy giving my child hugs and kisses”), where eight, 12 and 12 answers were missing at the 3 time-points. Not surprisingly, PAFAS questions 28–30 regarding parental teamwork (“I work as a team/disagree/ have a good relationship with my partner”) were mainly answered by persons in a relationship. No question in the SDQ at any timepoint was answered by less than 5% of participants. No imputation of the data was performed, participants with missing data were excluded from the respective analyses.

On average the caregivers were 36 years old, with an age range between 17 and 50 years. There was no difference in caregiver’s age between the three geographical sites. All 72 caregivers were female, 66% were married, 41% had primary school or less and 25% were working full time, as shown in Table 3. There was a significant difference in marital status and education between caregivers recruited through DTCs and high schools, but there were none regarding their partner’s education or their or their partner’s work status. Children from parents recruited through DTC were significantly older than those recruited through high schools and all caregivers recruited through DTC had experienced war or armed conflict, compared to 83% of those recruited through high schools (Table 3). All 72 participants identified most strongly with the Afghan ethnic or cultural group and reported Afghanistan as their country of origin.

Overall, the total difficulty score of the SDQ reduced significantly over time, from 17.8 at pre-test to 12.9 at post-test and 10.6 at follow-up (p < 0.001). Likewise, all SDQ subscales declined over time, as shown in Table 4.

There was no difference in SDQ scores at any time between participants recruited through DTCs or high schools. In children with very high (20+) scores on the total difficulty scale (n = 24; 12 boys, 12 girls) at baseline, highly significant reductions in all sub-scores as well as the total difficulty score could be found after the programme. There was no change between time 2 and time 3, however, overall, the effects were long-lasting, with highly significant differences at time 3 compared to baseline (Table 4).

Overall, parenting practice scores decreased significantly as assessed through all four PAFAS subscales between baseline and post-intervention and at follow-up compared to baseline. Across all parenting practices subscales, participants with scores above the 75 percentile at baseline had their scores more significantly reduced after the intervention compared to the rest of the sample (Table 5).

Likewise, family adjustment scores improved significantly after the programme, as seen in all three subscales. Again, those with the highest scores at baseline had their scores decline most after the intervention.

All process evaluation components and the respective performance indicators showed positive results ≥85%. Assessing fidelity, as previously described [38], overall 88–99% of our activities in all the sessions of the Strong Families programme were covered, with high consistency in facilitators attendance as well as group size. Facilitators reported very high doses received, measured through the interest of children, as well as their caregivers, being 100% in all sessions respectively. Assessing the reach of the programme, overall, 99% of families attended all three sessions of the programme. Inputs, such as the quality of childcare, travel arrangements, refreshments, rooms, materials and equipment were rated positively between 85 and 93% overall, as shown in Table 6.