Ethics approval was obtained from the University of Melbourne Human Research Ethics Committee (Date of approval: 13 July, 2010; Ethics ID: 1033203).
Study setting
Tweddle child and family health service
Tweddle Child & Family Health Service (TCFHS) is a not-for-profit public sector early parenting centre, established in the 1920s and now funded through the Victorian Government Department of Human Services. Tweddle’s residential, day stay, home visiting, outreach programs aim to facilitate learning and parenting skill development in parents of babies and children up to 4 years old. TCFHS is part of the second tier of services, and programs are offered by experienced and highly skilled maternal and child health nurses, midwives, and early childhood professionals free-of-charge to parents from a wide range of socioeconomic and cultural backgrounds, living in metropolitan and rural areas. TCFHS prioritises provision of services to vulnerable or isolated families and to parents of infants and young children assessed as at risk.
Tweddle day stay program
The Day Stay Program (DSP) service was established in 1992. DSPs are managed by Tweddle or jointly with other organisations. These programs have good local recognition and are highly regarded and used [2]. A goal of the program is to link families to supports within their community including mental health, maternal and child health, general practice and family support agencies.
The objectives of the DSP are that parents explore alternative ways of managing their parenting concerns, receive coaching in interpreting their child’s cues; explore factors that have contributed to their loss of confidence and be referred to services in their community [12].
Parents self-refer to the DSP by telephoning the service directly. Experienced nurses conduct an assessment interview and triage to the appropriate Tweddle service, based on severity of presenting needs, age of child and personal and socioeconomic circumstances.
Theoretical model of DSP
Groups of 2 to 4 families are admitted together to the 7-hour program. Parents are assisted individually to establish short-and long-term goals for their child and themselves. A health promotion model of practice is used in which one-to-one and group learning opportunities and supported practice are provided. Educational strategies include discussion, self-directed learning, counselling, coaching, demonstration and didactic presentations. Participants learn about infant behaviour management strategies to promote settling and sleep, optimal infant nutrition and feeding practices, infant development and behaviour, managing parental fatigue and promoting emotional well being, parent - infant relationships, safety and play [13].
Study design
The study was a prospective longitudinal cohort, or single group pre-and post-test design. Participants were assessed twice: once prior to admission to the DSP and once 4 – 8 weeks after discharge from the program.
Participants
Clients were identified by the TCFHS intake team as being eligible for the study if they were 18 years old or over, with English proficiency sufficient to give informed consent to participate and complete a written questionnaire and a telephone interview, and had accepted a place in the DSP with an infant under 12 months old.
Sample size
Based on the normal distribution, a sample size of 97 at outcome, provides 95% confidence that the true population prevalence of study parameters will lie between ± 5% of the prevalence estimates observed in the study [14].
Data sources
Participant data
-
1.
Tweddle Client Record
As part of admission procedures, all clients provide personal and health information, which is held electronically in the computerised triage program Client Assessment and Intake System (CAIS) and in paper records of registration and maternal and infant histories.
-
2.
Study-specific questionnaire
This collected demographic information including postcode and participant and baby’s date of birth, general health, mental health and infant behaviour.
-
3.
Follow-up computer assisted telephone interview (CATI)
The interview assessed general and mental health, infant behavior, health service use, assessment of child’s behavior and parenting enjoyment since attending the DSP.
Standardised measures
Standardised, validated, published, self-report measures were used to enable comparison of the study sample with relevant population norms.
-
1.
General health
General self-reported health (GSRH) was assessed using the single question “In general, would you say your health is: excellent/ very good/ good/ fair/ poor?” derived from the SF-36 [15]. This question is a good predictor of mortality and healthcare utilization, comparable to other multi-item measures [16].
-
2.
Mental health
-
i)
The EPDS is a widely-used 10-item self-rating scale for screening for probable depression, using 4 response options (0–3) and yielding a total score of 0–30 [10]. It has been validated in Australia against diagnostic interviews. A score of greater than 12 yields a sensitivity of 100%, a specificity of 95.7% and a positive predictive value of 69.2% for depression [17].
-
ii)
The Kessler 6 is a six-item self-rating scale [18] used to detect serious mental illness. Five response options for each question (scored 1 – 5) generate a total score between 6 and 30. A higher score indicates more distress and a score of 19 and over signifies a high risk of mental disorder [19].
-
3.
Infant behaviour
The Baby Behaviour Scale (BBS) [20] is a parent self-report measure of average duration of infant crying, frequency of night time waking, ease of soothing and settling and number and length of day time sleeps in a 24-hour period during the last two weeks. Maternal confidence is assessed with a single question. These eight items yield good internal consistency (Cronbach Alpha = 0.7 [20]). Scores on 8 individual items are summed, total scores range between 0 and 21, higher scores indicate more unsettled infant behaviour.
Comparison population data
Four sources of comparison data were used.
-
1.
2009 Victorian Child Health and Wellbeing Survey
A Victorian statewide survey was conducted by the Data Outcomes and Evaluation Division, Department of Education and Early Childhood Development in 2009. This study had a response fraction of 75% and used a CATI to survey primary caregivers of 5025 randomly selected Victorian children aged under 13 years [21].
-
2.
Birth in Victoria 2007 and 2008
The most recent summary data available for all births in Victoria [22].
-
3.
Tweddle Residential Service in 2004
Presenting needs and outcomes of 79 women admitted to the Tweddle residential service [4, 23].
-
4.
Two community samples
-
a)
A survey assessing the mental health of 12,361 women recruited from 43 health services across Australia at 6–8 months postpartum [24].
-
b)
Mental health and infant behaviour reported by 800 women attending local government immunisation clinics in the Tweddle catchment area with their four month old infants [20].
Procedure
Recruitment
All eligible women who were offered and accepted a place at a Tweddle DSP were posted a letter of invitation to participate, a plain language statement, a consent form, a contact detail form and a brief study specific questionnaire with their standard pre-admission pack.
Participation and follow-up
Those agreeing to participate signed a standard consent form, provided contact details and completed the study specific questionnaire. Clients placed their sealed envelope (whether materials had been completed or not) in a box provided during attendance at the DSP. Trained telephone interviewers contacted participants within 4–8 weeks of attendance of the DSP to assess their continued willingness to participate and arrange a suitable time to conduct the interview. Where necessary, at least three repeated attempts were made to reach participants by telephone.
Participant compensation
To compensate for participants’ time and inconvenience, a shopping voucher to the value of AUD 25 was posted to those who completed all stages of the study.
Data extraction from client record
Data were extracted from participants’ client record and entered into a study-specific standardised data extraction tool by a member of Tweddle staff and the research team working together.
Data management and analysis
The sources of data, measures used and response options for sociodemographic characteristics, heath and circumstances, reproductive history, maternal mental health, partner relationship, social support, infant characteristics, health service use, parenting and infant behaviours are presented in Additional file 1.
Data analyses
Data were analysed in SPSS v 19 [25]. Total scores and proportions above relevant cut-off scores on standardised scales were computed for the EPDS [10] and Kessler 6 [18] measures.. Clinically significant symptoms, defined as EPDS scores over 12 (for clinical samples) and over 9 (for community samples) [10] and Kessler 6 scores over 19 [26] were computed and entered as categorical variables. Items on the BBS [20] were recorded on individual 3 or 4-point scales. Individual item scores were summed and a total score computed as a continuous variable. Individual items were reported as frequencies (n; %) and total score as a continuous variable.
Normality tests were conducted on continuous data. Internal consistencies of numerical measures were established and reported as Cronbach’s alpha statistic. Descriptive statistics were computed using mean (SD) for continuous variables and frequency distributions for categorical data. Non-parametric one-sample binomial and chi-square tests were used to establish all significant differences in baseline variables from population-based comparison data. Baseline characteristics of participants retained in the study were compared with those lost to follow-up using Pearson chi-square and Fisher’s exact test for multinomial and binomial categorical variables respectively; Mann–Whitney tests for non-normally distributed and t-tests for normally distributed continuous variables.
The primary outcomes were defined as changes in mean EPDS and BBS scores between baseline and follow-up (Time 2 – Time 1) and reported as mean differences (95% confidence interval for the difference). Employment status was re-coded into a binary variable as employed versus not in paid employment; mode of birth as caesarean or assisted versus spontaneous vaginal; birth weight as low (<2500grams) versus normal (>2500grams) in regression analyses. Multiple regression analyses were conducted in order to examine the factors associated with the primary outcomes. First, factors potentially associated with outcomes as hypothesised were checked in bi-variable analyses. Variables with p-values < 0.2 [27]were included in the models. Model statistics including regression coefficients, 95% CIs, and p-values are presented. Negative coefficients indicate greater improvements in the outcomes and positive coefficients indicate deteriorations in outcomes.