• Open Access

Cost-effectiveness of Weight Watchers and the Lighten Up to a Healthy Lifestyle program

Authors


Correspondence to:
Linda Cobiac, School of Population Health, The University of Queensland, Herston, QLD 4006, Australia. Fax: (07) 3365 5442; e-mail: l.cobiac@uq.edu.au.

Abstract

Objective: Intensive weight loss programs that incorporate dietary counselling and exercise advice are popular and are supported by evidence of immediate weight loss benefits. We evaluate the cost-effectiveness of two weight loss programs, Lighten Up to a Healthy Lifestyle and Weight Watchers.

Methods: Health gains from prevention of chronic disease are modelled over the lifetime of the Australian population. These results are combined with estimates of intervention costs and cost offsets (due to reduced rates of lifestyle-related diseases) to determine the dollars per disability-adjusted life year (DALY) averted by each intervention program, from an Australian health sector perspective.

Results: Both weight loss programs produced small improvements in population health compared to current practice. The time and travel associated with attending group-counselling sessions, however, was costly for patients, and overall the cost-effectiveness ratios for Lighten Up ($130,000/DALY) and Weight Watchers ($140,000/DALY) were high.

Conclusion: Based on current evidence, these intensive behavioural counselling interventions are not very cost-effective strategies for reducing obesity, and the potential benefits for population health are small.

Implications: It will be critical to consider other strategies (e.g. changing the ‘obesogenic’ environment) or explore alternative methods of intervention delivery (e.g. Internet) to see if they offer a more cost-effective approach by effectively reaching a high number of people at a low cost.

The prevalence of obesity is increasing in Australia, with 60% of adults now classified as either overweight or obese1 and at increased risk of cardiovascular disease, type 2 diabetes, osteoarthritis and cancer.2,3 In Australia, 7.5% of the total burden of disease and injury is due to high body mass.4 This was estimated to cost the economy $8.3 billion in 2008,5 and projections show that the expenditure for diabetes alone is likely to increase 400% by 2032, as a consequence of increasing body mass.6

Concerted action is needed from governments in providing the leadership, advocacy, funding and policy to combat the growing epidemic.7 Intensive weight loss programs that incorporate dietary counselling and exercise advice are popular, and are supported by evidence of immediate weight loss benefits for program participants,8 but can be costly for Governments to deliver. Studies in the US,9 Switzerland10 and the Netherlands11 suggest that these sorts of interventions can be cost-effective for the health sector. However, it is not clear if it is more cost-effective for Governments to develop and run their own intervention program or provide financial assistance for the overweight and obese to attend a commercial program such as Weight Watchers (http://www.weightwatchers.com.au).

In our research, we have investigated the relative population health benefits and value for money of these two policy approaches in Australia. We focus our analyses in the state of Queensland, where the Government currently offers a two-month ‘Lighten Up to a Healthy Lifestyle’ group counselling program, but could alternatively offer vouchers to the overweight and obese to pay for attendance at a commercial weight loss program instead. The research forms part of the large ACE-Prevention study, in which we are applying a standard set of cost-effectiveness methods to evaluate and compare 150 primarily preventive interventions for the Australian population.12 The evaluation of these two interventions, however, is also supported by a small grant from Queensland Health, which is keen to evaluate the benefits of economic evaluation in helping to make policy decisions that will maximise population health on a limited healthcare budget.

Methods

The interventions are evaluated using the Assessing Cost-Effectiveness (ACE) method of economic evaluation,12 which has been widely used for evaluating interventions and informing policy in an Australian healthcare context.13–15 In this analysis, we use the ACE methods to evaluate the cost-effectiveness of implementing each program for one year, compared to no weight loss intervention. We evaluate each intervention from an Australian health sector perspective, which includes population health gains from reduced rates of disease, subsequent changes in healthcare costs, Government costs of intervention delivery, and costs to patients for participation.

Lighten Up to a Healthy Lifestyle

Lighten Up to a Healthy Lifestyle aims to prevent weight gain and achieve weight loss by promoting changes in nutrition and physical activity behaviour. Initially developed as a weight management program for overweight adults in 1990,16 the program has since been revised with a broader emphasis on promoting healthy lifestyle changes in nutrition and physical activity behaviour, and providing guidance in addressing negative behaviours, managing stress and setting realistic goals. It is based on the trans-theoretical model of behaviour change.17 The two-month program includes six group workshops, facilitated by trained nurses or allied health professionals, and three individual appointments for measurement and personal goal setting. Further information about the program can be found at http://www.health.gov.au/lightenup/.

In 2004, the Lighten Up program was implemented in 20 Health Districts across Queensland, and 459 participants were enrolled in an evaluation study. Criteria for enrolment in the program included an age of at least 18 years and a desire to improve diet, exercise and/or weight; the majority of participants in the study were women (85%) of middle- or older-age (75% aged at least 40), with a high BMI (mean of 34). A survey of participants (n=238) after the two-month intervention showed a significant increase in the consumption of fruits and vegetables and a significant increase in weekly walking time, resulting in a significant reduction in BMI of 0.47 kg/m2 (95% confidence interval: 0.37 – 0.58) among overweight and obese participants (mean BMI of 34). 18,19 Follow-up at 12 months indicated that weight loss was, on average, sustained in the small number of participants who were followed up (n=23), with a mean BMI reduction of 0.60 kg/m2 (0.46 – 1.6) compared with the baseline measure.

We base our analysis of cost-effectiveness on the results of the Lighten Up study (Table 1). Since no control group was included in the study, we estimate a background trend in BMI during the intervention and follow-up period from the average change in BMI between the 1999-2000 and 2004/05 AusDiab surveys.20,21

Table 1.  Cost-effectiveness modelling input parameters.
ParameterMean (SD)Uncertainty distributionSources and assumptions
  1. Note: All costs adjusted to 2003 Australian dollars using Australian health price deflators,32 consumer price index45 and/or purchasing power parities46 where relevant.

Lighten Up
Change in fruit and vegetable consumption (grams/day)+87 (18)NormalDerived from change in daily serves (assuming one serve = 80 g) from data provided by Queensland Health.
Change in physical activity (MET-mins/wk)+120 (9.7)NormalDerived from change in minutes of walking from data provided by Queensland Health (assuming average intensity of 3.5 METs).
Change in body mass (BMI units)-0.87 (2.6)NormalData provided by Queensland Health. Includes trend in BMI of +0.17 per year.20,21
Cost of project management staff$91,000 per staff memberBased on one full-time equivalent staff member at Level PO4 (District Health Services Employees' Award - State, QLD Health Professional Stream Wage Rates) with additional 30% on-costs component.
Cost of intervention delivery$251 ($153) per participantGammaMean delivery cost reported from four sites (Cairns, Fraser Coast, Centacare and Mackay), including patient registration cost (see below).
Cost of patient registration$40 ($6) per participantUniformReported patient registration cost ranges from $30 to $50 across all sites.
Weight Watchers
Change in body mass (BMI units)-2.5 (2.0)NormalDerived from change in weight and mean height in UK randomised controlled trial.24
Cost of primary care referral$14.72 per participantIncludes 20 minutes of nurse and administration time.26
Cost of Weight Watchers attendance$212 ($283) per participantGammaMean of three valuations.22,23,37
Other parameters
Cost of patient time$17.44 per hourDerived from labour force participation38 and average weekly earnings.39
Cost of patient travel$7.45 per tripBased on average distance travelled to GP for urban (estimate), regional40 and remote41 populations, and Royal Automobile Club Victoria private vehicle reimbursement rate for medium 2-3 L vehicles.
Relative risks (RRs) of diseaseSee sourcesNormal (ln RR)Relative risks due to physical activity,42 type 2 diabetes,43 fruit and vegetable consumption44 and body mass.2,3
Disease cost offsetsSee sourcesUniformCost per case derived from Australian treatment cost data31 and rates of disease.4 Minimum/maximum assumed to be ±25% of point-estimate.

Weight Watchers

The major commercial weight loss programs (e.g. Weight Watchers, Jenny Craig) aim to achieve weight loss through a low-calorie diet and advice on physical activity.22 In a systematic review of commercial weight loss programs, only Weight Watchers was found to have sufficient evidence of efficacy.23 In the only independently sponsored randomised controlled trial, participants in a six-month Weight Watchers program in the United Kingdom (UK) achieved a 7.3% (standard deviation (SD): 6.1%) weight loss at the end of the program, compared to a 0.6% (SD: 2.7%) weight gain in the control group.24 Criteria for enrolment in the trial were aged 18 to 65 years and BMI 27 to 40; similar to the Lighten Up intervention. However, the majority of participants were women (70%) of middle-age (mean of 40), with a high BMI (mean of 31).24

To model cost-effectiveness, we estimate change in BMI (Table 1) from the reported weight change and the mean height of intervention and control participants in the UK trial at the end of the six-month intervention program. We base our analyses on the six-month intention-to-treat analysis, but also evaluate sensitivity of cost-effectiveness results to a potential loss of intervention effect between six and 12 months, assuming an increase in BMI of 0.01 units per month.25 Unlike the Lighten Up intervention, the Weight Watchers trial did not report change in fruit and vegetable consumption or physical activity, hence we could not include these measures in the cost-effectiveness analysis of this intervention.

Intervention recruitment

To enable comparison of weight loss intervention results with cost-effectiveness of other interventions evaluated using the ACE methods, we first estimate the number of people in Australia who could potentially benefit from the two interventions, taking rates of overweight and obesity and intervention-specific recruitment and intervention completion rates into account. Ninety-seven per cent of Australia's 15 million adults would be eligible to participate in a lifestyle intervention, because they have a BMI of 25 or more, do not eat at least seven serves of fruits and vegetables each day and/or do not achieve 30 minutes of moderate activity on at least five days of the week.21 However, we estimate that just 0.1% of the population at risk would be willing to enrol in a group-based counselling intervention each year, based on Queensland Government sales of Lighten Up intervention materials over three consecutive years. To ensure comparability of intervention results, we assume that a similar proportion of the population would be motivated to enrol in the Weight Watchers program.

For the Lighten Up intervention, it is assumed that 77% of those who enrol in the intervention begin the program and 69% of the program participants will complete the intervention.19 For the Weight Watchers intervention, it is assumed that 85% of those who enrol will begin the program26 and 81% of the program participants will complete the intervention.24

Health outcomes

We use a multi-state, multiple-cohort life-table approach27 to determine changes in mortality and morbidity in the population due to the intervention effects on behaviour (BMI, physical activity and fruit and vegetable consumption for Lighten Up and BMI for Weight Watchers). Modelled diseases include type 2 diabetes, ischaemic heart disease, hypertensive heart disease, ischaemic stroke, osteoarthritis and cancers of the breast (in women), colon, lung, stomach, oesophagus, endometrium, kidney and thyroid. All disease data are derived from the most recent study of Australia's burden of disease and injury.4

The effect of each intervention on the incidence of each disease is quantified by the potential impact fraction (PIF), which is a function of the relative risk of disease and prevalence of the risk factor.28 We use a modified version of the potential impact fraction equation in which the intervention modifies the relative risk of disease rather than the prevalence of the risk factor. This enables calculation of the PIF from continuous measures of intervention effect (i.e. grams per day for fruits and vegetables, MET-minutes per week for physical activity and BMI for body mass) rather than from broad categorical shifts in prevalence.

Where a disease has more than one risk factor (e.g. ischaemic heart disease), the intervention effect on incidence is determined multiplicatively29 with adjustment of relative risks where risk factors are not independent (e.g. physical activity and BMI). In estimating the joint contribution of physical activity and BMI to the global burden of cardiovascular disease, Ezzati et al.29 lowered excess risk estimates by 50% to account for mediated effects. In our analysis, we first reviewed the literature and identified six prospective cohort studies for which relative risks of disease due to physical activity were reported both before and after adjustment for BMI, or relative risks of disease due to BMI were reported both before and after statistical adjustment for physical activity (or physical fitness). Because studies used different categorical divisions and measures (e.g. <1,000 kcal/wk or ≥1,000 kcal/wk,30 <1 hr/wk, 1–3.4 hrs/wk or ≥3.4 hrs/wk40 for physical activity), we could not determine a combined effect of adjustment on relative risk, for each disease-risk factor combination. We therefore determined the value that most accurately adjusted the relative risks across all categories within each study, by least squares optimisation, then determined the average value of this adjustment factor across all studies (Table 2).

Table 2.  The average adjustment factors for adjusting relative risks of disease due to physical activity and body mass.
 Average adjustment factor for relative risk
 Physical activityBody mass
  1. Note: a) Studies based on measurement of physical fitness (e.g. treadmill endurance), due to a lack of studies reporting relative risk data before and after adjustment for physical activity.

  2. In (RRadjusted) = adjustment factor × In (RRunadjusted)

Ischaemic heart disease0.7747–500.8951,52,a
Ischaemic stroke0.7747–500.8951,52,a
Type 2 diabetes0.6930,53–551.0030
Breast cancer0.87470.9256,57,a
Colon cancer0.87470.9256,57,a

We model the effects of the interventions on behaviour in the first year based on the results of the two intervention studies, but the sustainability of these effects longer-term is unknown. A meta-regression of 46 dietary and dietary and exercise counselling trials found that weight lost was regained after five and a half years.25 Therefore, in our analyses we assume that the intervention effects on behaviour follow an exponential pattern of decay over the lifetime, with a decay rate of 50% per annum, which means that there is virtually no intervention effect after five and a half years.25 However, we also evaluate the sensitivity of results to variations in annual rate of decay between 0% (life-long behaviour change) and 100% (behaviour change reversed after the first year).

Costs

The cost of each intervention is evaluated as if it is being implemented at full capacity and in steady-state (i.e. with initial start-up costs excluded) for one year. The cost of Lighten Up includes the cost to Government of co-ordinating the project (one full-time equivalent (FTE) staff member per 500 intervention participants) and delivering the intervention, and the cost to each patient for time and travel in attendance. The patient cost of registration is included as a transfer cost to the Government. The cost of Weight Watchers includes the cost to Government of primary care referral and Weight Watchers attendance vouchers, and the cost to each patient for time and travel in attendance. We estimate a cost per person (Figure 1) from unit costs (Table 1) and determine total intervention costs based on the number of participants recruited.

Figure 1.

Cost breakdown of the total cost per person of Lighten Up (two-month program) and Weight Watchers (six-month program).

The cost of each intervention is offset by the changes in treatment costs for all modelled diseases. The cost per incident case for all cancers averted and the cost per prevalent case for all other disease averted are evaluated using data from the Australian Institute of Health and Welfare Disease Costs and Impacts Study 200131 and rates of disease in 2001.4 We then adjust costs to the year 2003 using the Australian Health Price Index.32 We do not include non-related healthcare costs in added years of life.

Cost-effectiveness analysis

We model all costs and population health outcomes over the lifetime of the Australian population in 2003, discounting future costs and health outcomes at a rate of 3% per year. The costs and health outcomes are summed over the lifetime to determine the incremental cost-effectiveness ratio (ICER), in dollars per disability-adjusted life year (DALY) averted, for each intervention. Monte Carlo analysis is used to derive 95% uncertainty intervals for all outcome measures and to determine the probability of intervention cost-effectiveness against a cost-effectiveness threshold of $50,000 per DALY.33,34 ICER results are displayed on a cost-effectiveness plane; values associated with an increase in DALYs and saving in costs (south-east quadrant of the plane) are ‘dominant’, and values associated with a decrease in DALYs and increase in costs (north-west quadrant) are ‘dominated’, while cost-effectiveness of all other values (north-east or south-west quadrants) depends on the value relative to the cost-effectiveness threshold.

Scenario analysis

For the Government-run Lighten Up program, we examine the cost-effectiveness consequences of reducing program costs or increasing recruitment. We evaluate the effect of reducing intervention delivery costs per participant from $251 to $200 and $100, and the effect of halving project management costs by increasing the number of participants per full-time equivalent co-ordinator from 500 to 1,000 participants.

We also examine the cost-effectiveness consequences of increasing Lighten Up program recruitment. We evaluate the effect of increasing the proportion of the at-risk population willing to enrol in the intervention, from 0.1% in baseline analyses up to and including the theoretical maximum of 100%.

Results

Implementation of the current Lighten Up program to the whole of the Australian population could avert 38 DALYs (95% uncertainty interval: -55 – 130), at a total cost of $5.3 million ($3.9 million –$7.7 million) to the health sector. Implementation of a Weight Watchers voucher system to the same population could avert 54 DALYs: -30 – 140), at a total cost of $8.3 million ($8.0 million –$8.7 million) to the health sector.

In terms of cost-effectiveness, Lighten Up has a median ICER of $130,000/DALY (95% uncertainty interval: $22,000/DALY – dominated) and Weight Watchers has a median ICER of $140,000/DALY (41,000/DALY – dominated). The ICERs are spread across the north-west (dominated) and north-east quadrants of the cost-effectiveness plane (Figure 2), illustrating the substantial uncertainty in the cost-effectiveness results. The overall probability of Lighten Up being under a cost-effectiveness threshold of $50,000 per DALY is only 17% and the probability for Weight Watchers is just 5% (Figure 3).

Figure 2.

Monte Carlo analysis results for the Lighten Up to a Healthy Lifestyle and Weight Watchers interventions on the cost-effectiveness plane (the x-axis measures the health gain in DALYs averted, the y-axis measures the associated net costs (including patient time and travel costs) in 2003 Australian dollars, and the diagonal line represents a cost-effectiveness threshold of $50,000 per DALY).

Figure 3.

Acceptability of the Lighten Up to a Healthy Lifestyle and Weight Watchers interventions, with and without the patient time and travel costs (the dashed lines indicate the probability of the interventions being under a cost-effectiveness threshold of $50,000 per DALY).

Without patient time and travel costs (Figure 3), the median ICER of Weight Watchers decreases to $27,000/DALY (dominant – dominated) and the median ICER of Lighten Up, which decreases to $64,000/DALY ($2,000/DALY – dominated). Probabilities of being under the cost-effectiveness threshold increase to 43% for Lighten Up and 66% for Weight Watchers, without the patient time and travel costs.

The results are sensitive to the assumed rate of decay in intervention effect, with increasing sustainability of intervention effects on behaviour leading to better cost-effectiveness ratios (Figure 4). Varying the sustainability of behavioural changes between no benefit on behaviour after the first year and full benefit sustained for life, leads to a probability of cost-effectiveness ranging from 1% to 75% for Lighten Up and from 0% to 83% for Weight Watchers.

Figure 4.

Sensitivity of cost-effectiveness to the assumed annual rate of decay in intervention effects on behaviour after the first year, illustrated for the Lighten Up to a Healthy Lifestyle intervention (including patient time and travel costs).

However, the cost-effectiveness results are not highly sensitive to potential decay in the Weight Watchers effect on BMI between six months (the time point for which we have data from the randomised controlled trial) and 12 months (the time point at which the effect is modelled). Adding a weight regain trend of 0.01 BMI units per month leads to an increase in ICER of just 3%.

Reducing the costs of intervention delivery or project management for the Government-run Lighten Up program does improve cost-effectiveness, but the overall probability of cost-effectiveness remains relatively low (31%) even under the best-case scenario for cost reduction in both project management and intervention delivery (Table 3). Without patient time and travel costs, the best-case scenario for Lighten Up is on a par with Weight Watchers, in terms of cost-effectiveness.

Table 3.  Cost-effectiveness of the Lighten Up program under a range of Government cost-reduction scenarios.
Costing scenarioIncluding time and travelExcluding time and travel
 Median ICER ($/DALY)Probability <$50,000/DALYMedian ICER ($/DALY)Probability <$50,000/DALY
Intervention delivery – cost per participant (including patient registration)
$251/participant$130,00017%$64,00043%
$200/participant$121,00018%$55,00047%
$100/participant$104,00024%$37,00057%
Project management – number of participants per 1 full-time equivalent employee
500 participants$130,00017%$64,00043%
1,000 participants$112,00021%$48,00051%
Intervention delivery + project management
$100/participant + 1,000 participants$88,00031%$21,00066%

Increasing recruitment to the Lighten Up program has little effect on intervention cost-effectiveness; the ICER becomes slightly less favourable (ICER increases by at most 4%). With no fixed component of the program costs, the total costs increase almost linearly with the number of participants, leading to little change in the ratio of costs and effects.

Discussion

Neither Lighten Up nor the Weight Watchers voucher system are cost-effective interventions for reducing obesity. While both interventions have the potential to reduce disease burden in Australia, the potential magnitude of health benefits at a population level is small. There can be a role for interventions that have only small population health benefit, if there is strong evidence of benefit for those targeted by the interventions,35 but either of these intervention programs would be best implemented as only one component of a suite of interventions to combat obesity and not as the mainstay of prevention or treatment.

It is important to keep in mind that the strength of evidence underlying the measures of effect for the interventions programs evaluated in this research is relatively weak. In the case of Lighten Up, the lack of control group in the Queensland study19 makes it difficult to determine the true effect of the intervention against unrelated changes in the population. We assume that, in the absence of intervention, participants would experience the average population increase in BMI. However, the increase in BMI is slightly lower than average for people at the upper end of the BMI scale (such as the overweight and obese in the Lighten Up study). While this difference in BMI trend is not significant, it could lead to a small over-estimate of the health gain from the Lighten Up intervention. In addition, the response rate to follow-up measurement in the Lighten Up study at 12 months was very low. If those who had completed the intervention program but did not respond at 12 months had lower weight loss than those who did respond, then it is again likely that we have over-estimated the health gain. Overall, it is probable that the Lighten Up program is even less cost-effective than our estimates suggest.

For the Weight Watchers analysis, although based on slightly stronger randomised controlled trial evidence of intervention effects on BMI,24 the number of trial participants was small (n=119) and participant adherence to the intervention may have been better than what could be achieved in the general population. Taking these factors into account may make the cost-effectiveness of Weight Watchers less favourable than predicted. On the other hand, the Weight Watchers effect on physical activity and fruit and vegetable consumption could not be included in our analyses because they had not been measured in the intervention trial. These two additional risk factors contribute approximately two-thirds of the health gain of the Lighten Up intervention (results not shown). Therefore, additional measurement of physical activity and fruit and vegetable consumption in the future may substantially improve the cost-effectiveness of Weight Watchers.

Our cost-effectiveness ratios for Lighten Up (median ICER: $130,000/DALY) and Weight Watchers (median ICER: $140,000/DALY) are less favourable than cost-effectiveness ratios reported for other adult weight loss interventions that incorporate dietary counselling and exercise, which range from dominant (i.e. cost-saving)10 to $18,100 (2003) per quality-adjusted life year gained.9 Differences in the specific characteristics of intervention programs evaluated and the different (Australian) context of evaluation may partly account for contrasting results. However, modelling methods and assumptions around discount rates, sustainability of intervention effects, costing perspectives and timeframe for measurement of outcomes, all differ substantially between studies and may play an important role. Our analyses highlight, in particular, how sensitive cost-effectiveness ratios can be to assumptions about the sustainability of interventions effects over time.

Comparison with Australian physical activity interventions36 that were previously evaluated using the same ACE modelling methods and assumptions (Table 4) suggests that Lighten Up and Weight Watchers are relatively costly interventions (both with and without the patient time and travel costs) when compared on a per person basis. This contributes to their high cost-effectiveness ratios. Substantial reductions in costs of both project management and intervention delivery for Lighten Up do improve its cost-effectiveness, but the probability of the intervention being cost-effective remains low. This is chiefly due to its relatively small BMI benefits (0.87 BMI units) compared to Weight Watchers (2.5 BMI units) and compared to the average effect seen in weight loss counselling interventions (2 BMI units).25

Table 4.  Comparison of intervention effect on physical activity, intervention cost per person and median ICER, between the Lighten Up and Weight Watchers interventions and the physical activity interventions evaluated in ACE-Prevention.36
InterventionIntervention effect (MET-mins/wk)Including time and travelExcluding time and travel
Cost per personMedian ICER ($/DALY)Cost per personMedian ICER ($/DALY)
  1. Notes: a) Effect on physical activity was not reported in the Weight Watchers trial.24

Pedometers574$27Dominant$27Dominant
Mass media148$1.40Dominant$1.40Dominant
TravelSmart57$36$18,000$36$18,000
GP prescription160$190$11,000$160$6,800
GP referral238$1,300$75,000$210Dominant
Internet129$53$2,000$53$2,000
Lighten Up120$600$130,000$320$64,000
Weight Watchersa$860$140,000$230$27,000

Conclusion

Based on current evidence, the Lighten Up to a Healthy Lifestyle program and voucher system for funding Weight Watchers attendance are not highly cost-effective strategies for reducing obesity, and potential benefits for population health are small. It will be critical to also consider other intervention strategies, such as community-wide campaigns and changes to the physical environment, or explore alternative methods of intervention delivery (e.g. Internet), to determine if they can offer a more cost-effective approach by effectively reaching a higher number of people at low cost.

Acknowledgements

This research was supported by a grant from Queensland Health. Queensland Health provided data from its study of the Lighten Up intervention, but played no role in developing the cost-effectiveness methods, analysing interventions or preparing the manuscript, and had no influence on the decision to publish.

Ancillary