Background

Tobacco smoking during pregnancy is the most important preventable risk factor for pregnancy complications and adverse birth outcomes [1]. Smoking during pregnancy increases the risk of adverse infant outcomes, including stillbirth, preterm birth, and low birth weight [1]. Maternal death is five times more likely in women who smoke during pregnancy than those who do not [1]. Smoking during pregnancy can have lifelong consequences for infants. Low birth weight is associated with childhood respiratory infections, asthma, high blood pressure, heart disease, type 2 diabetes [2,3,4], and being overweight or obese as a child [5,6,7] or adult [8, 9]. Exposure to secondhand smoke may occur during pregnancy [10, 11], or through passive exposure following birth [11,12,13]. Both forms of exposure can contribute to a number of health conditions in children [14].

In Australia, while the rate of smoking during pregnancy has been consistently falling in the last decade, 9.2% of women reported smoking at any time during pregnancy in 2020 [15]. Of those who smoked in the first 20 weeks of pregnancy (8.8%), the majority (78%) continued to smoke after 20 weeks [15]. The gap between Aboriginal and non-Aboriginal women who smoke during pregnancy is stark, with Aboriginal women almost six times more likely to smoke at any time during pregnancy (43.4%) than non-Indigenous women (7.5%) [15]. Higher rates of smoking during pregnancy are also seen among women aged less than 20 (34%), women aged 20–24 (21%), and those in very remote areas [15].

Globally, it is estimated that healthcare expenditure attributable to smoking-related diseases totalled US$422 billion in 2012, equivalent to 5.7% of global health expenditure [16]. Australian health care expenditure attributable to smoking was estimated to be $6.8 billion in 2015–16 [17], and tobacco use is responsible for 9.3% of the total burden of disease, greater than any other contributing risk factor [18]. Health care service usage and costs are consistently found to be higher for smokers than for non-smokers [19]. For women who smoke during pregnancy, costs associated with birth complications are 66% higher than those for non-smoking women [20], and healthcare costs during childhood have also been found to be higher for infants and children of women who smoke during pregnancy, primarily due to increased in-patient hospital care [21].

Evidence from systematic reviews indicates that psychosocial interventions to support pregnant women to quit are effective [22, 23]. Evidence-based international [24] and Australian [25, 26] guidelines recommend routine, repeated smoking cessation support (SCS) for all pregnant women using brief interventions. This can be undertaken in a few minutes during routine care [24], however provision of recommended SCS to pregnant women in Australia has remained persistently poor [27, 28]. The missing link is a failure to consistently implement effective smoking cessation care for pregnant women.

Generalisable evidence of the cost-effectiveness of interventions to improve smoking cessation is necessary to inform policy change at a system-level. A recent review of the cost-effectiveness of smoking cessation interventions for pregnant women found that interventions, such as cognitive behavioural therapy and nicotine replacement therapy, for both pregnant women and the wider population may be cost-effective from both a health system and a societal perspective [29]. However, these interventions targeted the smokers themselves and, despite evidence that system change interventions for smoking cessation can also be effective [30], few studies have investigated the cost-effectiveness of systems change approaches in the general population or in maternity services [30,31,32].

Here we describe in detail the protocol for an economic evaluation of the MOHMQuit intervention. MOHMQuit is a systems-change intervention being tested in a pragmatic trial undertaken in a real-world setting. The intervention includes focused training for maternity service leaders and clinicians to achieve culture change and increase prioritisation of support for smoking cessation, and provides key resources to deliver effective and appropriate SCS to pregnant women [33]. The intervention is hypothesised to increase the rates of smoking cessation in pregnancy by improving the provision of guideline-recommended SCS to pregnant women through enhancing clinicians’ knowledge, skills and confidence to provide SCS to pregnant women. Evidence of the effectiveness and cost-effectiveness of smoking cessation interventions of this kind is required to implement evidence-based, health system-wide SCS innovations at scale. The aim of this economic evaluation is to explore, from an Australian healthcare system perspective, the cost-effectiveness of MOHMQuit to increase the quit rate of pregnant women who smoke compared to usual care.

Methods/Design

MOHMQuit trial

The details of the MOHMQuit intervention and the protocol for the implementation trial are described in detail elsewhere [ANZCTR #382491] [33, 34]. In summary, the MOHMQuit trial is a pragmatic stepped-wedge cluster-randomised controlled trial to be implemented in nine public hospitals providing maternity services in New South Wales (NSW). Pregnant smokers will attend a maternity service that, based on their randomised start date for the MOHMQuit intervention, has either received the MOHMQuit intervention or is still in the baseline ‘control’ period. All sites will receive the intervention, with continuous data collection occurring throughout the baseline, intervention, washout and follow-up periods for the three years of the trial [33]. The primary outcome is the 7-day point prevalence abstinence among pregnant smokers at the end of pregnancy, confirmed by salivary cotinine testing. Secondary outcomes include cost-effectiveness of the MOHMQuit intervention, rates of documentation of SCS provided, changes in clinicians’ knowledge, confidence, and attitudes to providing SCS, women’s views on SCS received, and intervention fidelity. The MOHMQuit trial is due to be completed in December 2024. See Fig. 1 for a diagram of the stepped-wedge design and timeframe.

Fig.1
figure 1

Stepped-wedge design of the MOHMQuit trial [33]

Full size image

Economic evaluation overview

The economic evaluation will be conducted from an Australian healthcare system perspective and include costs in Australian dollars (AUD) for the year 2022. The economic evaluation will be undertaken by adapting the Economics of Smoking in Pregnancy (ESIP) Model [35, 36] to use NSW or Australian data where possible. Two primary analyses will be carried out: a) a within-trial cost-effectiveness analysis with an 8-month time horizon, with results presented as the incremental cost per additional quitter; and b) a lifetime cost-utility analysis, with results presented as incremental cost per quality-adjusted life-year (QALY) gained for mother and child, presented both separately and as a combined ‘per pregnancy’ measure of cost-effectiveness. The results of the cost-utility analysis will be assessed against a commonly used benchmark of $50,000 AUD per QALY gained [37,38,39]. A secondary analysis of cost–benefit over the lifetime will also be conducted.

The robustness of the cost-effectiveness and cost-utility analyses will be explored using probabilistic sensitivity analysis and expressed in a cost-effectiveness acceptability curve. Return on investment estimates will also be produced for maternal and infant health care (separately and combined) for the lifetime time horizon. Cost–benefit ratios (defined as incremental health care savings divided by incremental intervention cost) will be produced. The economic evaluation will adhere to current best practice guidelines [40, 41] and analysis will be undertaken using Microsoft Excel 2010 [42].

Within-trial analysis

Patient-level data on resource use and outcomes will be routinely collected during the trial and used in the economic evaluation. The time horizon for the within-trial analysis is 8 months, representing the maximum time each woman is in contact with the health service, from early pregnancy to immediately after birth. No discounting of costs and outcomes is required as this time horizon is less than one year. The analysis will be undertaken based on an intention-to-treat approach, and participants lost to follow-up will be assumed to be continuing smokers.

Measurement and valuation of resource use (within-trial analysis)

The three resource use groups of relevance to the within-trial analysis are: delivery of the intervention; clinician and leader time to engage with the intervention; and neonatal care. A summary of the sources of measurement and valuation for intervention costs are shown in Table 1. Intervention implementation will be monitored using project management logs, training logs, eMaternity data, and participant (clinician, manager, and patient) self-report. Clinician and clinical leader time to engage with MOHMQuit activities will be collected through self-report. The child’s birth weight, date of delivery, gestational age (allowing calculation of preterm status), and whether the baby is admitted to the special care nursery or neonatal intensive care unit will be collected postpartum via the trial data management system.

Table 1 Measurement and valuation of intervention costs
Full size table

Unit costs will be obtained from routine sources, including NSW Award Wage scales (to value time of MOHMQuit implementation team and clinician/leader engagement with the intervention), trial data management system (for consumables), and IHACPA (Independent Health and Aged Care Pricing Authority) National Hospital Cost Data Collection (for neonatal costs) [43]. Both mean and median costs will be presented [44]. Where appropriate, mean cost estimates will be used with confidence intervals generated through bootstrapping. All costs will be presented in AUD and updated to a standard reference year (2022) for analysis. Unit cost information will be combined with the resource use data to estimate the total cost per pregnant smoker who received maternity care either with or without the MOHMQuit intervention. These per person total costs will be aggregated to estimate the overall total cost of MOHMQuit and standard care and subsequently the average cost per pregnant smoker for each. Costs associated with providing the intervention will be reported separately. The proportion of leaders, clinicians and patients who participate in the MOHMQuit intervention, and the proportion of women who report receiving SCS, will be presented. Clinician time to implement SCS strategies is incorporated into standard care and will not have a cost associated with it.

Measurement and valuation of benefit (within-trial analysis)

The primary benefit of the intervention will be measured by comparing the 7-day point prevalence abstinence at the end of pregnancy, confirmed by salivary cotinine testing, among women who report current smoking or quitting since becoming pregnant at antenatal booking in the intervention period to the control (baseline) period. This will inform the incremental cost per additional quitter analysis. Seven-day point prevalence abstinence with biochemical verification is recommended as an outcome measure [45, 46], and is commonly used in pregnancy smoking cessation trials, as longer timeframes are not relevant to benefits to the foetus [47, 48]. It is recognised that this outcome has limitations, in that it is an intermediate outcome and not a measure of health. However, collecting broader outcomes is not practicable in the context of this trial as data will only be collected directly from women in postpartum interviews. Additionally, the myriad physical and social changes that occur with birth and in the early postnatal period [49, 50] make it difficult to assess whether changes in quality of life outcomes would be due to smoking cessation, the postpartum stage, or other factors.

Analysis and sensitivity analysis (within-trial analysis)

The cost-effectiveness analysis will use the estimates of cost and effect as described above to estimate an incremental cost-effectiveness ratio (ICER) of cost per quitter. Where significant levels of missing data occur (5% or greater of the observations), approaches to account for missingness will be undertaken, including multiple imputation to account for data missing at random or missing completely at random [51] and other relevant best practice approaches [52].

Sensitivity analyses will examine uncertainty around the primary endpoint, costs, the fidelity of implementation, and the impact of economies of scale if the intervention was rolled out at the population level. Sensitivity analysis will be undertaken using non-parametric bootstrapping to provide the confidence ellipse, which reflects the uncertainty in the estimate of the ICER. The ellipse provides a region on the cost-effectiveness plane that should contain 95% of the uncertainty [53]. Uncertainty regarding the cost-effectiveness of the intervention will be summarised using a cost-effectiveness acceptability curve (CEAC) and will represent the likelihood of the intervention being cost-effective at a range of ceiling willingness to pay thresholds for an additional woman quitting.

Lifetime model

The time horizon for the lifetime economic model is lifetime of both the mother and the child. Costs and effects will be discounted at 5% annually in line with recommendations from NSW Treasury [54]. While the within-trial analysis considers the costs and effects of the pregnant women and neonatal costs only across the eight months follow-up, the lifetime model expands the evaluation to include the healthcare costs and health outcomes of smoking behaviour and any changes across the mother and child’s lifetime. This includes the impacts of the child’s exposure to secondhand smoke, and that they are more likely to smoke themselves, increasing the likelihood of future smoking-related diseases [55].

Model structure (lifetime model)

To conduct the lifetime analysis we will use an adapted version of a published decision analytic model – the ESIP model, modified for Australian populations [35, 36]. ESIP predicts the impact that smoking both in pregnancy and after can have on the lifetime healthcare costs and health outcomes for the mother and her offspring. Using ESIP enables us to expand the evaluation to include the benefits of stopping smoking which occur beyond the short time horizon of the within-trial cost-effectiveness analysis.

The structure of the original ESIP model is shown in Fig. 2. The model is divided into mother and infant, and within-pregnancy (outcomes and costs during pregnancy), childhood (outcomes and costs for the infant from birth to 15 years) and lifetime (lifetime outcomes and costs for both mother and infant) time periods. The mother’s component of the model will include the costs of SCS, smoking-related diseases (coronary heart disease [CHD], chronic obstructive pulmonary disease [COPD], lung cancer and stroke), data on quality of life, relapse rates and transition to smoking-related diseases if the mother relapses. The childhood component of the model will include costs associated at birth for children with smoking mothers if relevant (for premature birth and low birth weight newborns), outcomes related to secondhand smoke exposure in the home, quality of life, smoking uptake rates for children of smokers and associated smoking-related morbidities (asthma, CHD, COPD, lung cancer and stroke). Resource use and outcomes collected during the trial will be extrapolated and supplemented with published data from the literature.

Fig. 2
figure 2

Original ESIP model structure [36] LBW: low birth weight; NBW: normal birth weight

Full size image

ESIP will be modified for the Australian population by using Australian and NSW data sources where possible. For example, the Cancer Institute NSW Smoking and Health and Tobacco Tracking Surveys (maternal post-partum smoking behaviour and maternal and partner lifetime smoking behaviour), NSW Perinatal Data Collection (maternal morbidities and outcomes, and infant birth outcomes), Australian Bureau of Statistics National Health Survey (prevalence of morbidities among general population [CHD, stroke, COPD, lung cancer, childhood asthma]), Australian Institute of Health and Welfare National Drug Strategy Household Survey (infant exposure to secondhand smoke in the home).

Analysis and sensitivity analysis (lifetime model)

The lifetime analysis results will be presented as incremental cost per QALY gained for both the mother and child, separately and combined. Cost–benefit analysis ratios (defined as incremental health-care savings divided by incremental intervention cost) will be produced for the lifetime time horizon. Deterministic and probabilistic sensitivity analyses will be undertaken for the range of areas of uncertainty (informed by the 95% confidence interval or standard error of the mean input value where available) to explore underlying model assumptions. One-way sensitivity analysis will be undertaken to demonstrate the impact of varying specific inputs for: self-reported outcomes; alternative discount rates (3% and 7%); varying relapse rates and changing adherence rates. Two-way sensitivity analysis will be carried out on variables found to substantially increase the ICER in one-way sensitivity analysis, and those which are correlated. Probabilistic sensitivity analysis will be undertaken using model parameter distributions for ESIP’s 390 input variables. Methods used for fitting distributions have been described elsewhere [35]. Ten thousand Monte Carlo simulations will be performed, and cost-effectiveness acceptability curves produced.

Discussion

While many pregnant women are highly motivated to quit smoking [56], they face significant challenges including a lack of effective support from clinicians [57]. This study represents one of the few analyses conducted of the cost-effectiveness of smoking cessation interventions that target antenatal healthcare providers for pregnant women and conducted under ‘real world’ conditions. There is increasing demand for evidence of cost-effectiveness of implementation of healthcare interventions to provide policy makers with critical information for the best value-for-money spend on finite budgets. Randomised controlled trials of the clinical effectiveness of interventions and their implementation can provide good opportunities to conduct an economic evaluation alongside the trial, provided the appropriate steps are taken from the outset to ensure that the design of the trial is fit-for-purpose [51]. Decreasing antenatal smoking to reduce the effects on pregnancy and newborn outcomes is a government priority [58, 59]. Evidence of the effectiveness and cost-effectiveness of smoking cessation interventions is required to implement evidence-based, health system-wide SCS innovations. This study will provide evidence of MOHMQuit’s costs and benefits to inform decisions on scalability.

The MOHMQuit trial economic evaluation has been planned using available national and international guidelines for conducting economic evaluations [51, 60], promoting greater transparency in the methods undertaken and increasing the rigor and validity of the findings. While it will draw on the MOHMQuit trial results for model inputs, the economic evaluation will adopt its own methods and analysis and results will be reported separately from the main trial without duplicating information. The ESIP model provides a comprehensive approach to estimating costs, outcomes, and long-term cost-effectiveness and cost-utility, of smoking cessation interventions in pregnancy [35]. Its ability to provide common outcome measures (for example, incremental cost per QALY) allow comparisons between smoking cessation and other healthcare interventions. Where possible, ESIP will be populated with NSW and Australian data, however initial investigations have found that local level data may not be available for all model parameters. The analysis might be strengthened through adaptation of the ESIP model to allow for comparison between population sub-groups with higher rates of smoking in pregnancy. The study will provide essential policy-relevant information for decision makers on the value of evidence-based implementation of support for antenatal healthcare providers delivering SCS for pregnant women.