Lisa Gold

Complex interventions or complex systems? Implications for health economic evaluation.

Although guidelines exist for evaluating complex interventions, they may be of little help in dealing with the multiple effects of interventions in complex systems such as hospitals. Alan Shiell, Penelope Hawe, and Lisa Gold explain why it is important to distinguish the two types of complexity

Methods for exploring implementation variation and local context within a cluster randomised community intervention trial

Insignificant or modest findings in intervention trials may be attributable to poorly designed or theorised interventions, poorly implemented interventions, or inadequate evaluation methods. The pre-existing context may also account for the effects observed. A combination of qualitative and quantitative methods is outlined that will permit the determination of how context level factors might modify intervention effectiveness, within a cluster randomised community intervention trial to promote the health of mothers with new babies. The methods include written and oral narratives, key informant interviews, impact logs, and inter-organisational network analyses. Context level factors, which may affect intervention uptake, success, and sustainability are the density of inter-organisational ties within communities at the start of the intervention, the centrality of the primary care agencies expected to take a lead with the intervention, the extent of context-level adaptation of the intervention, and the amount of local resources contributed by the participating agencies. Investigation of how intervention effects are modified by context is a new methodological frontier in community intervention trial research.

Outcomes and costs of primary care surveillance and intervention for overweight or obese children: the LEAP 2 randomised controlled trial

Objective To determine whether ascertainment of childhood obesity by surveillance followed by structured secondary prevention in primary care improved outcomes in overweight or mildly obese children. Design Randomised controlled trial nested within a baseline cross sectional survey of body mass index (BMI). Randomisation and outcomes measurement, but not participants, were blinded to group assignment. Setting 45 family practices (66 general practitioners) in Melbourne, Australia. Participants 3958 children visiting their general practitioner in May 2005-July 2006 were surveyed for BMI. Of these, 258 children aged 5 years 0 months up to their 10th birthday who were overweight or obese by International Obesity Taskforce criteria were randomised to intervention (n=139) or control (n=119) groups. Children who were very obese (UK BMI z score ≥3.0) were excluded. Intervention Four standard consultations over 12 weeks targeting change in nutrition, physical activity, and sedentary behaviour, supported by purpose designed family materials. Main outcomes measures Primary measure was BMI at 6 and 12 months after randomisation. Secondary measures were mean activity count/min by 7-day accelerometry, nutrition score from 4-day abbreviated food frequency diary, and child health related quality of life. Differences were adjusted for socioeconomic status, age, sex, and baseline BMI. Results Of 781 eligible children, 258 (33%) entered the trial; attrition was 3.1% at 6 months and 6.2% at 12 months. Adjusted mean differences (intervention − control) at 6 and 12 months were, for BMI, −0.12 (95% CI −0.40 to 0.15, P=0.4) and −0.11 (−0.45 to 0.22, P=0.5); for physical activity in counts/min, 24 (−4 to 52, P=0.09) and 11 (−26 to 49, P=0.6); and, for nutrition score, 0.2 (−0.03 to 0.4, P=0.1) and 0.1 (−0.1 to 0.4, P=0.2). There was no evidence of harm to the child. Costs to the healthcare system were significantly higher in the intervention arm. Conclusions Primary care screening followed by brief counselling did not improve BMI, physical activity, or nutrition in overweight or mildly obese 5-10 year olds, and it would be very costly if universally implemented. These findings are at odds with national policies in countries including the US, UK, and Australia. Trial registration ISRCTN 52511065 (www.isrctn.org)

Outcome data from the LEAP (Live, Eat and Play) trial: a randomized controlled trial of a primary care intervention for childhood overweight/mild obesity

Objectives:To reduce gain in body mass index (BMI) in overweight/mildly obese children in the primary care setting.Design:Randomized controlled trial (RCT) nested within a baseline cross-sectional BMI survey.Setting:Twenty nine general practices, Melbourne, Australia.Participants:(1) BMI survey: 2112 children visiting their general practitioner (GP) April–December 2002; (2) RCT: individually randomized overweight/mildly obese (BMI z-score <3.0) children aged 5 years 0 months–9 years 11 months (82 intervention, 81 control).Intervention:Four standard GP consultations over 12 weeks, targeting change in nutrition, physical activity and sedentary behaviour, supported by purpose-designed family materials.Main outcome measures:Primary: BMI at 9 and 15 months post-randomization. Secondary: Parent-reported child nutrition, physical activity and health status; child-reported health status, body satisfaction and appearance/self-worth.Results:Attrition was 10%. The adjusted mean difference (intervention–control) in BMI was −0.2 kg/m2 (95% CI: −0.6 to 0.1; P=0.25) at 9 months and −0.0 kg/m2 (95% CI: −0.5 to 0.5; P=1.00) at 15 months. There was a relative improvement in nutrition scores in the intervention arm at both 9 and 15 months. There was weak evidence of an increase in daily physical activity in the intervention arm. Health status and body image were similar in the trial arms.Conclusions:This intervention did not result in a sustained BMI reduction, despite the improvement in parent-reported nutrition. Brief individualized solution-focused approaches may not be an effective approach to childhood overweight. Alternatively, this intervention may not have been intensive enough or the GP training may have been insufficient; however, increasing either would have significant cost and resource implications at a population level.

The effectiveness of the levonorgestrel‐releasing intrauterine system in menorrhagia: a systematic review

Objective To determine whether the levonorgestrel‐releasing intrauterine device (LNG‐IUS), licensed at present for contraceptive use, may reduce menstrual blood loss with few side effects. If effective, surgery could be avoided with consequent resource savings.

Effects of continuity of care by a primary midwife (caseload midwifery) on caesarean section rates in women of low obstetric risk: the COSMOS randomised controlled trial

Please cite this paper as: McLachlan H, Forster D, Davey M, Farrell T, Gold L, Biro M, Albers L, Flood M, Oats J, Waldenström U. Effects of continuity of care by a primary midwife (caseload midwifery) on caesarean section rates in women of low obstetric risk: the COSMOS randomised controlled trial. BJOG 2012;119:1483–1492.

A Parent-Focused Intervention to Reduce Infant Obesity Risk Behaviors: A Randomized Trial

OBJECTIVE: To assess the effectiveness of a parent-focused intervention on infants’ obesity-risk behaviors and BMI. METHODS: This cluster randomized controlled trial recruited 542 parents and their infants (mean age 3.8 months at baseline) from 62 first-time parent groups. Parents were offered six 2-hour dietitian-delivered sessions over 15 months focusing on parental knowledge, skills, and social support around infant feeding, diet, physical activity, and television viewing. Control group parents received 6 newsletters on nonobesity-focused themes; all parents received usual care from child health nurses. The primary outcomes of interest were child diet (3 × 24-hour diet recalls), child physical activity (accelerometry), and child TV viewing (parent report). Secondary outcomes included BMI z-scores (measured). Data were collected when children were 4, 9, and 20 months of age. RESULTS: Unadjusted analyses showed that, compared with controls, intervention group children consumed fewer grams of noncore drinks (mean difference = –4.45; 95% confidence interval [CI]: –7.92 to –0.99; P = .01) and were less likely to consume any noncore drinks (odds ratio = 0.48; 95% CI: 0.24 to 0.95; P = .034) midintervention (mean age 9 months). At intervention conclusion (mean age 19.8 months), intervention group children consumed fewer grams of sweet snacks (mean difference = –3.69; 95% CI: –6.41 to –0.96; P = .008) and viewed fewer daily minutes of television (mean difference = –15.97: 95% CI: –25.97 to –5.96; P = .002). There was little statistical evidence of differences in fruit, vegetable, savory snack, or water consumption or in BMI z-scores or physical activity. CONCLUSIONS: This intervention resulted in reductions in sweet snack consumption and television viewing in 20-month-old children.

Screening and counselling in the primary care setting for women who have experienced intimate partner violence (WEAVE): a cluster randomised controlled trial

BACKGROUND Evidence for a benefit of interventions to help women who screen positive for intimate partner violence (IPV) in health-care settings is limited. We assessed whether brief counselling from family doctors trained to respond to women identified through IPV screening would increase womens quality of life, safety planning and behaviour, and mental health. METHODS In this cluster randomised controlled trial, we enrolled family doctors from clinics in Victoria, Australia, and their female patients (aged 16-50 years) who screened positive for fear of a partner in past 12 months in a health and lifestyle survey. The study intervention consisted of the following: training of doctors, notification to doctors of women screening positive for fear of a partner, and invitation to women for one-to-six sessions of counselling for relationship and emotional issues. We used a computer-generated randomisation sequence to allocate doctors to control (standard care) or intervention, stratified by location of each doctors practice (urban vs rural), with random permuted block sizes of two and four within each stratum. Data were collected by postal survey at baseline and at 6 months and 12 months post-invitation (2008-11). Researchers were masked to treatment allocation, but women and doctors enrolled into the trial were not. Primary outcomes were quality of life (WHO Quality of Life-BREF), safety planning and behaviour, mental health (SF-12) at 12 months. Secondary outcomes included depression and anxiety (Hospital Anxiety and Depression Scale; cut-off ≥8); womens report of an inquiry from their doctor about the safety of them and their children; and comfort to discuss fear with their doctor (five-point Likert scale). Analyses were by intention to treat, accounting for missing data, and estimates reported were adjusted for doctor location and outcome scores at baseline. This trial is registered with the Australian New Zealand Clinical Trial Registry, number ACTRN12608000032358. FINDINGS We randomly allocated 52 doctors (and 272 women who were eligible for inclusion and returned their baseline survey) to either intervention (25 doctors, 137 women) or control (27 doctors, 135 women). 96 (70%) of 137 women in the intervention group (seeing 23 doctors) and 100 (74%) of 135 women in the control group (seeing 26 doctors) completed 12 month follow-up. We detected no difference in quality of life, safety planning and behaviour, or mental health SF-12 at 12 months. For secondary outcomes, we detected no between-group difference in anxiety at 12 months or comfort to discuss fear at 6 months, but depressiveness caseness at 12 months was improved in the intervention group compared with the control group (odds ratio 0·3, 0·1-0·7; p=0·005), as was doctor enquiry at 6 months about womens safety (5·1, 1·9-14·0; p=0·002) and childrens safety (5·5, 1·6-19·0; p=0·008). We recorded no adverse events. INTERPRETATION Our findings can inform further research on brief counselling for women disclosing intimate partner violence in primary care settings, but do not lend support to the use of postal screening in the identification of those patients. However, we suggest that family doctors should be trained to ask about the safety of women and children, and to provide supportive counselling for women experiencing abuse, because our findings suggest that, although we detected no improvement in quality of life, counselling can reduce depressive symptoms. FUNDING Australian National Health and Medical Research Council.

Improving infant sleep and maternal mental health: a cluster randomised trial

Objectives: To determine whether a community-delivered intervention targeting infant sleep problems improves infant sleep and maternal well-being and to report the costs of this approach to the healthcare system. Design: Cluster randomised trial. Setting: 49 Maternal and Child Health (MCH) centres (clusters) in Melbourne, Australia. Participants: 328 mothers reporting an infant sleep problem at 7 months recruited during October–November 2003. Intervention: Behavioural strategies delivered over individual structured MCH consultations versus usual care. Main outcome measures: Maternal report of infant sleep problem, depression symptoms (Edinburgh Postnatal Depression Scale (EPDS)), and SF-12 mental and physical health scores when infants were 10 and 12 months old. Costs included MCH sleep consultations, other healthcare services and intervention costs. Results: Prevalence of infant sleep problems was lower in the intervention than control group at 10 months (56% vs 68%; adjusted OR 0.58 (95% CI: 0.36 to 0.94)) and 12 months (39% vs 55%; adjusted OR 0.50 (0.31 to 0.80)). EPDS scores indicated less depression at 10 months (adjusted mean difference −1.4 (−2.3 to −0.4) and 12 months (−1.7 (−2.6 to −0.7)). SF-12 mental health scores indicated better health at 10 months (adjusted mean difference 3.7 (1.5 to 5.8)) and 12 months (3.9 (1.8 to 6.1)). Total mean costs including intervention design, delivery and use of non-MCH nurse services were £96.93 and £116.79 per intervention and control family, respectively. Conclusions: Implementing this sleep intervention may lead to health gains for infants and mothers and resource savings for the healthcare system. Trial registration: Current Controlled Trial Registry, number ISRCTN48752250 (registered November 2004).

Preference measurement using conjoint methods: an empirical investigation of reliability

The application of conjoint measurement to the field of health economics is relatively new, although there is growing interest and there have been a number of studies undertaken recently. Wider acceptance of the technique requires methodological issues concerning both reliability and validity to be addressed. This paper reports an empirical investigation of the test-retest reliability of the discrete choice conjoint measurement approach in health care. This investigation of conjoint reliability was framed using the clinical context of investigation and treatment of knee injuries. A high level of reliability at both the input data and results levels was demonstrated.