Article Text
PDF
Abstract
Objective Interventions to reduce health inequalities for young children and their mothers are important: involving peers is recommended, but evidence of value for this approach is limited. The authors aimed to examine the effect of an innovative tailored peer-mentoring programme, based on perceived needs, for first-time mothers in socio-economically deprived communities.
Design Randomised controlled trial; parallel qualitative study with purposive samples using semistructured interviews.
Setting Socio-economically disadvantaged areas, Belfast.
Participants Primigravidae, aged 16–30 years, without significant co-morbidity.
Intervention Peer-mentoring by a lay-worker fortnightly during pregnancy and monthly for the following year, tailored to participants' wishes (home visits/telephone contacts), additional to usual care.
Main outcome measures Infant psychomotor and mental development (Bayley Scales of Infant Development (BSID-II)) at 1 year, assessed by an observer blinded to group allocation. Mothers' health at 1 year postnatal (SF-36).
Results Of 534 women invited, 343(64%) participated; 85%, with their children, completed outcome assessments (140 of 172 intervention; 152 of 171 controls). Intervention and control groups did not differ in BSID-II psychomotor (mean difference 1.64, 95% CI −0.94 to 4.21) or mental (−0.81, −2.78 to 1.16) scores, nor SF-36 physical functioning (−5.4, −11.6 to 0.7) or mental health (−1.8, −6.1 to 2.6). Women valued advice given in context of personal experience of child-rearing. Mentors gained health-related knowledge, personal skills and new employment opportunities.
Conclusions Despite possible longer-term social advantage, this peer-mentoring programme showed no benefit for infant development or maternal health at 1 year. Further rigorous evaluation of important outcomes of complex interventions promoting health for children in socially disadvantaged communities is warranted.
Trial registration no ISRCTN 55055030
Statistics from Altmetric.com
Reducing health inequalities is a key government priority.1 Despite recent initiatives, challenges remain in meeting 2010 targets for reducing gaps in health outcomes.1 Mothers and young children in socially disadvantaged areas are an important group whose healthcare needs should be addressed.2 Socio-economic deprivation is associated with poor health,3 social isolation, and adverse lifestyle choices, with many adverse childhood outcomes continuing into adulthood.4
What is already known on this subject
▶ Despite many government initiatives, challenges remain in meeting targets for reducing health inequalities, particularly for mothers and young children living in socially disadvantaged areas.
▶ Involving peers as lay health workers in interventions to reduce health inequalities is recommended, but evidence of the value of this approach is limited.
What this study adds
▶ This tailored peer-mentoring programme for first-time mothers living in socio-economically deprived areas showed no effects on infant development or maternal health at 1-year postnatal but may show later benefits—mothers valued health advice given, and mentors gained personal skills.
▶ Interventions to improve health outcomes for socially disadvantaged communities require further evaluation with careful design and long-term follow up of both medical and social outcomes.
Successful initiatives aimed at reducing health inequalities have included multidisciplinary approaches and peer involvement in delivering interventions.5 However, many studies involved small samples and lacked detail.5 A Cochrane review found benefits for interventions involving lay health workers in promoting breast-feeding and immunisation but inconclusive evidence of other value.6 Possible explanations for inconclusive findings included poor uptake of interventions, lack of clarity of the lay-worker's role, high staff turnover and insufficient professional support.7,–,9
We aimed to address these problems in a randomised controlled trial, determining the effect of an innovative peer-mentoring programme, for first-time mothers living in socio-economically deprived areas, on infant psychomotor and mental development and maternal health at 1 year postnatal. We explored participants' experiences in a parallel qualitative evaluation.10 The programme was designed in response to the community's perceptions of their health and social care needs and, if found to have value, could be translated easily into routine health service care.
Methods
Study design and participants
Midwives recruited primigravidae, aged 16–30 years, with postcodes in the lowest tertile of deprivation scores11 in Northern Ireland (NI), less than 20 weeks' gestation, without co-morbidity requiring ongoing healthcare and who spoke English, at their first hospital antenatal visit (November 2003–February 2005). A home visit or telephone call by a mentor, a non-professional peer who would provide health-related information, was offered twice monthly during pregnancy and monthly for the following year. After completing questionnaires, including quality of life (SF-36) and “Wantedness of Pregnancy”,12 women who consented were randomly allocated to intervention and control groups by an independent individual using a computer-generated programme, with alternate blocks of 20 and 40. Mentors subsequently contacted intervention group women, matched in age and locality, and recorded details of all contacts. They were supervised by two research midwives; a project manager co-ordinated the trial.
Mentor recruitment, training and support
Mentors, aged less than 40 years, with at least one child less than 10 years old, were recruited through local advertisements. Successful applicants attended three 2-h training workshops initially and further training sessions, six to eight weekly, throughout the trial. Information was given about how and when to contact appropriate services, ethical, practical and safety issues in home-visiting and health-related advice, including written materials which could be shared with women. Mentors could telephone the midwives for advice at any time and were paid £6.00 per hour, plus travel and telephone expenses. Further details of training were reported previously.10
Outcome measurements
At 1 year, primary outcomes (infants' neurodevelopmental performance (Bayley Scales of Infant Development II)13; mothers' physical and mental well-being (SF-36)) and infant growth were assessed by a researcher blinded to group allocation. Secondary outcomes, assessed during routine hospital visits, included parenting stress,14 parental efficacy,15 maternal attachment16 and self-completed lifestyle questionnaires. Information about feeding and use of health and social services was collected at a home visit at 9 months postnatal and checked in primary care records.
The experiences, within the trial, of research midwives and purposive samples of mentors (n=11) and women (n=11) were explored in semistructured interviews.
Sample size
We estimated that 144 in each group would allow detection of 5 points difference between groups in Bayley scores13 and, based on previous data,17 a 10% difference in mental health and physical function (SF-36 components), with 80% power and 0.05 α. To allow for 15% attrition, we aimed to recruit 170 in each group.
Data analysis
Intention-to-treat analyses for primary outcomes used t tests within SPSS (v15), comparing differences between groups, initially excluding cases with missing data and then, for those missing, imputing values, derived from all observed values, using a single expectation-maximisation algorithm.18 χ2 testing compared categorical measures between groups. To account for lack of independence between twins, within analyses we randomly selected one from each set and ignored the other (The study included six sets of twins—three control, three intervention). A Complier Average Causal Effect (CACE) analysis examined potential bias in interpreting effects by comparing outcomes for intervention “compliers” (received at least 10 telephone contacts or three visits) and “expected compliers” in the control group. CACE models for primary outcomes and growth were estimated19 from covariates which predicted compliance (age, smoking at booking, education, wantedness of pregnancy, deprivation, maternal self-efficacy, antenatal attachment) using multiple imputation methods (n=10) to handle missing data.
The analysis of interview data was reported previously.10 A questionnaire relating to qualitative findings was posted to all mentors.
Ethical approval
The Central Office for Research Ethics Committees (Ref 124/03) approved the study in April 2003.
Results
Of 675 women identified as potentially eligible, 141 (21%) did not fulfil inclusion criteria: 534 were invited to participate (fig 1). Of these, 64% (343 of 534) consented, completed questionnaires and were allocated to intervention (n=172) or control (n=171) groups. Mean gestation (14 weeks) and age (22 years) for participants and non-participants did not differ significantly.
Recruitment, allocation and follow-up of participants.
Demographic characteristics, educational qualifications, wantedness of pregnancy, lifestyle habits and maternal attachment of both groups were similar; socio-economic indicators reflected greater deprivation among participants than the NI population (table 1).
Comparison of baseline characteristics of intervention and control groups
Intervention group women received a mean of 8.5 mentor contacts (SD 9.3); 29% received more than 12 contacts, but 16% received none (table 2). General childcare and diet were the most frequently discussed topics (table 3).
Numbers and types of mentor contacts received by women in intervention group
Mentoring topics and frequency of discussion in mentoring contacts
Primary and secondary outcome measures
Analysis using non-imputed data showed no significant differences between the groups in infant Bayley scale scores or maternal physical or mental health (table 4). Analysis using imputed data indicated that the control group's physical health was better than that of the intervention group, but this was of borderline statistical significance and probably of no clinical significance.
Primary outcome measures (BSID-II; SF-36) at 1 year for intervention and control groups
No differences were found between groups in infant growth or other secondary outcomes (tables 5 and 6). Since infants were assessed at slightly different ages (11–13 months), we converted actual growth measurements into z scores based on population mean scores for age at assessment,20 using the LmsGrowth programme (v2.4).21 Overall, relative to their age, our infants were shorter, with smaller head circumferences, but heavier than the reference population.
Secondary outcome measures for intervention and control groups: percentages (numbers) and 95% CI of percentages reporting achievement of categorical variables
Secondary outcome measures for intervention and control groups (mean (SD) values for gestation, birth weight and 1 year growth measures; Parenting Stress Index (PSI), maternal self-efficacy and alcohol consumption)
There were no significant differences between groups in changes in smoking, alcohol or recreational drug use. At booking, 40.8% (140 of 343) reported cigarette smoking; 16% of these (23 of 140) reported having stopped at 9 months postnatal. Regular alcohol consumption was reported by 44% (151 of 343) at booking, though many reported decreasing their intake; mean weekly consumption at 22 weeks' gestation was less than at 9 months postnatal (1.3 vs 5.2 units). At booking, 27.9% (95 of 343) reported ever having used recreational drugs; few used them during or after pregnancy.
“Compliance” with mentoring programme
We defined arbitrarily as “non-compliant” those who received less than 10 telephone contacts or three visits (88/172, 51.2%). “Compliers” and “non-compliers” did not differ significantly in primary outcomes or infant growth, although CACE models suggested that more visits (six or more) were associated with lower levels of well-being.
Mentors
During the study, we appointed 32 mentors; six resigned before contacting women, and 16 others resigned, after varying intervals, for reasons including new employment opportunities (n=9), insufficient time due to personal/family circumstances (n=8) or disillusionment (n=3). All women whose mentor resigned were offered a replacement: 33% (28 of 85) declined. Of those who accepted (n=57), approximately half experienced further resignation(s), at which stage many were over 6 months postnatal, considered they had sufficient support and declined further replacements. Overall, 56% (48 of 85) declined continued mentoring until 1 year postnatal. Midwives asked women if there were any problems with mentors: none was identified.
Qualitative findings
Our previous report10 focused on difficulties encountered in the trial, primarily related to making initial contacts, developing relationships and mentors' time constraints. Additionally, we identified themes of successful mentoring, the mentor's role and training. Successful mentors were perceived to have good communication skills, with experience of motherhood and time to share personal experiences—“I would ask her things, like what did you do?” Women appreciated their mentor's social support—“It was just nice to have someone to talk to”. Some mentors perceived their role primarily as providing friendship, saying “I think of her as a friend”, while sharing health-related information. The impact of this information extended beyond the trial participants—“Some of the stuff that (mentor) gave me, I shared it with my sister too.” Within their training, mentors appreciated opportunities to learn about health and discuss health-related news items with midwives. Their questionnaire responses confirmed the qualitative findings.
After 1 year follow-up (81%; 140 of 172), women were invited to complete an anonymous questionnaire about their experiences. Responses indicated 38 of 140 (27%) liked the mentor visits, 136 of 140 (97%) the 1 year assessment and 67 of 140 (48%) “everything” about the programme.
Discussion
Following a randomised controlled trial of a tailored peer-mentoring programme for primigravidae living in socio-economically deprived areas, we found no effects on infant development or maternal health at 1 year postnatal. However, mothers appreciated mentors' support, and there may be longer-term benefits for the community through mentors' dissemination of health-related information and their development of skills.
This is the first randomised controlled trial of a peer-mentoring programme delivered both antenatally and for a year postnatally in the UK. The intervention design was informed by the local community and underpinned by a theoretical framework, based on the social cognitive theory of health promotion, providing information about risk and benefit of behaviour change and taking account of social support in influencing behaviour.22 We recruited mentors from the same locality, with similar social backgrounds as the women they were to mentor, provided mentor training and support, and endeavoured to clarify their role, thus seeking to address some limitations of previous studies.7 9 23 24
We believe our results have external and internal validity. Although only 64% of those identified as eligible participated, reasons for non-participation, such as being too busy, having enough support and not being interested reflect those of other communities.9 The socio-economic indicators we examined confirmed our study sample's disadvantage compared to the general population. Random allocation, performed at a remote location to avoid selection bias, yielded groups well-matched in socio-economic indicators. Measurement bias was minimised by using an assessor blinded to treatment allocation. Descriptions of mentor recruitment, selection, and training allow comparisons with other programmes.
Our primary outcome measures are well-validated but may not be sufficiently sensitive to detect intervention effects in the UK healthcare system, which is accessible and free to all. Also, the impact of lifestyle advice on health-related behaviour,25 may only become apparent as the child develops.26 Tailoring the intervention, allowing individuals to choose the frequency and nature of mentoring contacts, may have reduced its potential effect, but we wished to encourage women's participation. Also, we did not assess mentors' lifestyles nor “quality assure” their mentoring, as such assessments may have deterred community engagement. Lay-workers' own needs for support are relevant to future work.
The frequency of peer-mentoring contact was less than in studies in Dublin27 and the developing world28 but compares favourably with other UK studies.7 23 UK women may accept additional support less readily because their “usual” care involves more frequent healthcare contacts. Nevertheless, 85% of our participants attended 1-year follow-up. Within the context of a well-developed public healthcare system, a “ceiling” effect may limit the potential for benefit from lay-worker mentoring. However, while we did not conduct an economic analysis, health advice received in the context of personal experience and improved employment opportunities in socially disadvantaged communities may have important benefits in terms of “social capital”.
In conclusion, we consider that our negative findings indicate that our intervention was ineffective: its design was tailored in content and frequency of contact to the perceived needs of individual women, and it was delivered as planned. Effective interventions involving peer mentoring may require a more prescriptive design. Careful scientific evaluation of health service innovations29 30 is important.31 Limitations in the design of such evaluations result in a lack of clear evidence to inform major policy decisions.32 While our study provides limited evidence of short-term benefits from a lay-worker intervention, it shows potential for longer-term social advantage. Further evaluation of interventions to reduce health inequalities for socially disadvantaged children is warranted and should include a range of relevant social and medical outcomes.
Acknowledgments
The authors wish to thank all the midwives and nurses who contributed to the delivery of the study, the participants, mentors and collaborating general practices.
References
Footnotes
-
Funding The study was funded by the Research and Development Office, Northern Ireland, under the “Targeting Social Need” initiative. MEC was co-funded by the Centre of Excellence for Public Health (Northern Ireland), a UKCRC Public Health Research Centre of Excellence. Funding from the British Heart Foundation, Cancer Research UK, Economic and Social Research Council, Medical Research Council, Research and Development Office for the Northern Ireland Health and Social Services and the Wellcome Trust, under the auspices of the UK Clinical Research Collaboration, is gratefully acknowledged.
-
Competing interests None.
-
Ethics approval This study was conducted with the approval of the Central Office for Research Ethics Committees (Ref 124/03), Northern Ireland, April 2003.
-
Provenance and peer review Not commissioned; externally peer reviewed.