This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on https://formative.jmir.org, as well as this copyright and license information must be included.
Evidence suggests that smartphone apps can be effective in the self-management of weight. Given the low cost, broad reach, and apparent effectiveness of weight loss apps, governments may seek to encourage their uptake as a tool to reduce excess weight in the population. Mass media campaigns are 1 mechanism for promoting app use. However, the cost and potential cost-effectiveness are important considerations.
The aim of our study was to use modeling to assess the health impacts, health system costs, cost-effectiveness, and health equity of a mass media campaign to promote high-quality smartphone apps for weight loss in New Zealand.
We used an established proportional multistate life table model that simulates the 2011 New Zealand adult population over the lifetime, subgrouped by age, sex, and ethnicity (Māori [Indigenous] or non-Māori). The risk factor was BMI. The model compared business as usual to a one-off mass media campaign intervention, which included the pooled effect size from a recent meta-analysis of smartphone weight loss apps. The resulting impact on BMI and BMI-related diseases was captured through changes in health gain (quality-adjusted life years) and in health system costs. The difference in total health system costs was the net sum of intervention costs and downstream cost offsets because of altered disease rates. An annual discount rate of 3% was applied to health gains and health system costs. Multiple scenarios and sensitivity analyses were conducted, including an equity adjustment.
Across the remaining lifetime of the modeled 2011 New Zealand population, the mass media campaign to promote weight loss app use had an estimated overall health gain of 181 (95% uncertainty interval 113-270) quality-adjusted life years and health care costs of –NZ $606,000 (–US $408,000; 95% uncertainty interval –NZ $2,540,000 [–US $1,709,000] to NZ $907,000 [US $610,000]). The mean health care costs were negative, representing overall savings to the health system. Across the outcomes examined in this study, the modeled mass media campaign to promote weight loss apps among the general population would be expected to provide higher per capita health gain for Māori and hence reduce health inequities arising from high BMI, assuming that the intervention would be as effective for Māori as it is for non-Māori.
A modeled mass media campaign to encourage the adoption of smartphone apps to promote weight loss among the New Zealand adult population is expected to yield an overall gain in health and to be cost-saving to the health system. Although other interventions in the nutrition and physical activity space are even more beneficial to health and produce larger cost savings (eg, fiscal policies and food reformulation), governments may choose to include strategies to promote health app use as complementary measures.
The obesogenic food environment and unhealthy dietary patterns have led to overweight and obesity becoming a critical public health problem [
Although modifying the food and physical activity environment is critical, addressing unhealthy dietary patterns and insufficient physical activity (and therefore overweight and obesity [
The use of health apps is increasing, with a reported 50% of smartphone users having ever downloaded a health app [
Reviews have found that mHealth interventions can be more effective than non-mHealth interventions at inducing weight loss and improving diet and physical activity [
Given the low cost, broad reach, and apparent effectiveness of apps at promoting weight loss, governments may seek to encourage the uptake of such apps as an opportunity for reducing excess weight among the population. For example, in the United Kingdom, the National Health Service has developed a free 12-week diet and exercise plan that is available as an app [
The cost and potential cost-effectiveness are important considerations when governments are selecting among obesity reduction interventions, including the use of mass media campaigns. Research by Cleghorn et al [
Since the modeling conducted by Cleghorn et al [
We used an established proportional multistate life table model [
The modeled intervention was a one-off mass media campaign among the New Zealand population stimulating the uptake and use of a smartphone app for weight loss that effectively promotes weight loss. The pathway from the intervention’s implementation to impact is detailed in
Flowchart of intervention conceptualization. NZ: New Zealand.
We quantified the effectiveness of weight loss apps using the results of a recent meta-analysis by Islam et al [
where
In our modeling, the business-as-usual baseline encapsulated the existing levels of dietary health promotion in New Zealand, including current promotion of weight loss apps, and the continuation of the current low or no app promotion environment. The costs of implementing the intervention are reported in
Intervention parameters and uncertainty distributions.
Parameter | Value | Distribution | Source |
Adult New Zealand population who own a smartphone, % (SD) | 81 (5) | Beta | As reported by DataReportal based on Google Consumer Barometer data [ |
Adult New Zealand population who are assumed to be aware of a relevant mass media campaign, % (SD) | 45 (20) | Beta | On the basis of an evaluation of an Australian obesity-prevention mass media campaign that measured the proportion of survey respondents who recognized the campaign [ |
Adult New Zealand population who were assumed to download and use a promoted weight loss app, % (SD) | 14 (20) | Beta | On the basis of the proportion of survey respondents who reported |
Intervention BMI reduction for those who used the app (kg/m2; 95% CI) | –0.400 (–0.858 to 0.051) | Normal | The weighted results of studies included in the Islam et al [ |
Assumed weight regain after delivery of the intervention (kg/m2 per month; SD %) | 0.03 (20) | Log-normal | Meta-analysis of weight loss decay evidence from Dansinger et al [ |
Estimated cost of one-off 1-year national-level mass media campaign, NZ $ (US $; SD %) | 2,883,000 (1,940,000; 20) | Gamma | As used in the previous published work by Cleghorn et al [ |
The multistate life table model consists of a main life table organized by age, sex (male or female), and ethnicity (Māori or non-Māori) and populated with all-cause mortality and morbidity rates for the 2011 New Zealand adult population. Parallel to this are life tables for each BMI-related disease where proportions of the simulated population are also modeled. Although the model includes a wide array of diet-related diseases [
Within the model, the proportions of the population in each disease table are a function of past and current rates of disease incidence, case fatality, and, for cancers only, remission, which are calculated at each annual time step. The model is populated with mean BMI values according to age, sex, and self-identified ethnicity measured in person during New Zealand’s most recent available national nutrition survey (New Zealand Adult Nutrition Survey 2008-2009) [
Our model included the health system costs associated with changing disease prevalence and population longevity, which were calculated using an established protocol [
In addition to the main base case intervention, we conducted an equity analysis where an
The model was built in Microsoft Excel and run using Ersatz (version 1.34; EpiGear International). Uncertainty around health gains and cost-effectiveness was quantified using a Monte Carlo analysis. The parameters were sampled independently 2000 times from each of their respective probability distributions. The presented results are the mean values, with 95% uncertainty intervals (UIs). The exception to this is the expected values in the scenario and sensitivity analyses, which did not include uncertainty analysis.
Across the remaining lifetime of the modeled 2011 New Zealand population, the mass media campaign to promote weight loss app use in the base case analysis had an estimated overall health gain of 181 (95% UI 113-270) QALYs and health care costs of –NZ $606,000 (–US $408,000; 95% UI –NZ $2,540,000 [–US $1,709,000] to NZ $907,000 [US $610,000]). The mean of the health care costs is negative, representing an overall savings to the health system and a cost-saving intervention. However, the 95% UI spans positive values, indicating that there remains a possibility that the intervention is not cost saving, albeit still cost-effective (ie, below the threshold of NZ $45,000 [US $30,000] per QALY gained, approximately the gross domestic product per capita for New Zealand that we use in our modeling [
In the first 10 years after the intervention was implemented (2011-20), the mean health gain was 56 QALYs and net health system expenditures averaged NZ $850,000 (US $572,000) because of the cost of implementing that mass media campaign and few savings to the health system (although still below the cost-effective threshold of NZ $45,000 (US $30,000) per QALY gained). After 20 years (2011-30), the total health gain was 112 QALYs and the mass media campaign became cost-saving (–NZ $176,000 [–US $118,000]) for the health care system. Most of the health gain (62%) occurred between the years 2011 and 2030 (first 20 years after implementation), whereas most of the health system savings (71%) occurred after 2030 (ie, 20 years after implementation). The delayed health system savings was due to the initial up-front cost of implementing the intervention, which was also relatively high compared with the eventual reductions in downstream cost offsets because of altered disease rates.
Health gains and cost-effectiveness of a mass media campaign to promote smartphone apps for weight loss in New Zealand by age, sex, and ethnicity (lifetime impacts and 3% discount rate).
Sex, ethnicity, and age group (years) | Health gain in QALYsa (95% UIb) | Health gain in QALYs per 1000 population (95% UI) | Health system costsc, NZ $ (US $; 95% UI) | |||||
All | 181 (113 to 270) | 0.041 (0.026 to 0.061) | –606,000 (2,540,000 to –907,000); 408,000 (–1,709,000 to 610,000) | |||||
Non-Māori, all ages | 148 (85 to 231) | 0.040 (0.023 to 0.062) | –491,000 (2,310,000 to 921,000); –330,000 (–1,555,000 to 620,000) | |||||
Māori, all ages | 33 (18 to 53) | 0.049 (0.027 to 0.079) | –115,000 (–494,000 to 158,000); –77,400 (–332,000 to 106,000) | |||||
|
97 (48 to 170) | 0.045 (0.022 to 0.079) | –436,000 (–1,872,000 to 554,000); –293,000 (–1,260,000 to 373,000) | |||||
|
|
|||||||
|
|
25-44 | 19 | 0.038 | –85,000 (–57,200) | |||
|
|
45-64 | 46 | 0.094 | –556,000 (–374,000) | |||
|
|
≥65 | 16 | 0.063 | –127,000 (–85,500) | |||
|
|
|||||||
|
|
25-44 | 6 | 0.080 | –51,000 (–34,300) | |||
|
|
45-64 | 9 | 0.171 | –118,000 (–79,400) | |||
|
|
≥65 | 1 | 0.078 | –13,000 (–8750) | |||
|
84 (44 to 141) | 0.037 (0.020 to 0.063) | –170,000 (–1,370,000 to 698,000); –114,000 (–922,000 to 470,000) | |||||
|
|
|||||||
|
|
25-44 | 16 | 0.031 | –45,000 (–30,000) | |||
|
|
45-64 | 35 | 0.069 | –369,000 (–248,000) | |||
|
|
≥65 | 17 | 0.055 | –77,000 (–52,000) | |||
|
|
|||||||
|
|
25-44 | 6 | 0.066 | –49,000 (–33,000) | |||
|
|
45-64 | 9 | 0.143 | –106,000 (–71,000) | |||
|
|
≥65 | 1 | 0.079 | –11,000 (–7000) |
aQALY: quality-adjusted life year.
bUI: uncertainty interval.
cA negative cost indicates that the intervention is cost-saving to the health system.
dThe 95% uncertainty intervals for QALY and health system costs were not calculated for these subgroups.
Results for Māori with equity adjustment applied (lifetime gains and 3% discount rate).
Population | Health gain in QALYsa (95% UIb) | Health gain in QALYs per 1000 population (95% UI) | Health system costsc, NZ $ (US $; 95% UI) |
All | 40 (23 to 65) | 0.060 (0.034 to 0.097) | –132,000 (–513,000 to 152,000); –89,000 (–345,000 to 102,000) |
Men | 20 (9 to 39) | 0.062 (0.026 to 0.118) | –67,000 (–341,000 to 115,000); –45,000 (–229,000 to 77,000) |
Women | 20 (9 to 37) | 0.058 (0.025 to 0.109) | –65,000 (–331,000 to 111,000); –44,000 (–223,000 to 75,000) |
aQALY: quality-adjusted life year.
bUI: uncertainty interval.
cA negative cost indicates that the intervention is cost saving to the health system.
The modeled lifetime health gains among adults experiencing overweight or obesity were 0.065 (95% UI 0.041-0.097) QALYs per 1000 target population. By ethnicity, the gains for non-Māori were 0.064 (95% UI 0.037-0.100) QALYs and for Māori 0.070 (95% UI 0.038-0.113) QALYs.
A range of results for scenario and sensitivity analyses are presented in
Sensitivity and scenario analyses for a mass media campaign to promote weight loss smartphone apps by age, sex, and ethnicity (expected value analysis, lifetime perspective, and 3% discount rate, unless otherwise stated).
Sensitivity and scenario analysesa | Health gain in QALYsb | Difference in QALYs from base case, % | Health system costsc, NZ $ (US $) | Difference in health system costs from base case, % |
Base case analysis | 183 | —d | –625,000 (–421,000) | — |
1. Mass media campaign: higher recognition at 68% | 276 | 51 | –2,414,000 (–1,620,000) | 286 |
2. Increase effect size of app use by 50% | 274 | 50 | –2,375,000 (–1,600,000) | 280 |
3. 100% of population use the app for more time | 278 | 52 | –2,454,000 (–1,650,000) | 293 |
4a. Delaying weight regain by 1 year | 203 | 11 | –2,400,000 (–1,620,000) | 284 |
4b. Delaying weight regain by 5 years | 1,261 | 589 | –21,271,000 (–14,300,000) | 3305 |
4c. No weight regain | 14,727 | 7948 | –286,465,000 (–193,000,000) | 45,762 |
5. Value from the previous Cleghorn et al [ |
69 | –62 | 1,549,000 (–1,040,000) | –348 |
6a. 0% discount rate | 334 | 83 | –1,892,000 (–1,270,000) | 203 |
6b. 6% discount rate | 114 | –38 | 186,000 (–125,000) | –130 |
aExpected values given for all scenarios.
bQALY: quality-adjusted life year.
cA negative cost indicates that the intervention is cost saving to the health system.
dBase case is the reference with which scenarios are compared.
The results from this updated health economic simulation modeling suggest that a hypothetical government-initiated mass media campaign to promote use of smartphone weight loss apps would result in modest health gains over the remaining lifetime of the New Zealand adult population. There was an estimated net saving to the health care system because of reductions in BMI-related diseases, although the UIs included estimates that were cost-effective (rather than cost saving). The intervention would be expected to generate greater per capita health gain for Māori and therefore potentially reduce health inequities attributable to BMI differences between Māori and non-Māori, assuming that the intervention would be as effective for Māori as it is for non-Māori.
Several key characteristics contributed to this modeled mass media intervention being cost-effective. First, and perhaps most importantly for this study, recent evidence from a meta-analysis of randomized controlled trials and case-control studies shows that the use of smartphone weight loss apps largely results in some degree of weight loss, even when accounting for variations in the duration of app use [
We found that there was very limited research examining whether mass media campaigns stimulate the specific action of adopting use of a smartphone app. Therefore, we relied on a UK evaluation of a mass media campaign that encouraged the use of a physical activity app [
This paper provides new evidence using updated parameters on the potential health gain and cost-effectiveness of this health intervention. The previous Cleghorn et al [
A strength of this research is that it builds upon an established model [
As part of a wide range of interventions to address the obesogenic environment and unhealthy dietary patterns, governments should consider investing in promoting such weight loss apps, along with funding research that improves their effectiveness and uptake in the community. But all such interventions should also be well evaluated, particularly given the large potential for scalability. Laws and taxes can create a less obesogenic environment (and have been shown to be more cost-effective than nutrition mass media campaigns [
Using recent evidence on the effectiveness of smartphone weight loss apps, a modeled mass media campaign to encourage the adoption of smartphone apps to promote weight loss among the New Zealand adult population is expected to yield an overall gain in health and to be cost saving to the health system. This is an update of previous modeling that showed a smaller health gain and that the intervention was not cost-effective. Although other interventions in the nutrition and physical activity space are even more beneficial to health and cost savings (eg, pricing policies and food reformulation [
Calculating the weighted pooled intervention effect size.
mobile health
New South Wales
quality-adjusted life year
uncertainty interval
This research was funded by the New Zealand Ministry of Business Innovation & Employment (project number UOOX1406). The funding body had no role in the design of the study; the collection, analysis, and interpretation of data; or in the writing of the manuscript.
ACJ led the conceptualization of the intervention, modeled the intervention, and wrote the first draft of the paper (excluding the introduction). LG wrote the first draft of the introduction. NW contributed to the study and intervention conceptualization and led the research grant that funded the study. NN provided foundational work in the costing of the modeled intervention. CC led the development of the diet model, contributed to the study and intervention conceptualization, and provided modeling expertise. All authors contributed to drafts of the manuscript and read and approved the submitted manuscript.
None declared.