Summary

Introduction

Behavioural weight loss treatment delivered via mobile technology is efficacious ¹, ², with roughly 25–44% of participants achieving clinically significant weight loss of 5% in interventions that are standalone (i.e. those without a counselling component) ³, ⁴, ⁵. Early weight loss is a robust predictor of clinically meaningful weight loss ⁶, ⁷, ⁸ and has been used as a trigger in stepped‐care behavioural interventions to implement rescue efforts for individuals not responding early to obesity treatment ⁶, ⁹, ¹⁰, ¹¹. A 12‐week internet‐based weight loss study found a strong positive relation of early weight loss at 1 month predicting weight loss at 3, 6 and 12 months ¹². However, no studies of mobile health (‘mHealth’) interventions – i.e. behavioural treatments delivered via modalities such as smartphone applications (‘apps’) or text messaging – have examined the ability of early weight loss success to predict longer‐term weight loss. Given the increasing evidence base for obesity treatment delivered via mHealth strategies ¹, ², research is needed that examines whether early weight loss success in mHealth interventions predicts future weight loss and intervention engagement.

Standalone mHealth interventions offer a unique treatment experience as their high portability enables self‐monitoring to occur at the time of actual food and beverage consumption, which can prompt real‐time tailored feedback, and they have high dissemination potential given that 77% of US adults own a smartphone ¹³. Further, mHealth interventions have unparalleled reach, providing treatment to those who are currently without care. The predictive utility of early weight loss is especially important in mHealth interventions: patients who successfully lose weight in these remotely delivered interventions can do so with lower healthcare costs, while those who need additional support can quickly be diverted to higher levels of care.

The GoalTracker trial compared self‐monitoring strategies for weight loss in a 12‐week commercial app‐based intervention that was remotely delivered ¹⁴. In intent‐to‐treat analyses, there were no differences between treatment arms in weight change at 1 (range: −0.80 to −1.76 kg), 3 (range: −2.43 to −2.75 kg) or 6 months (range: −1.88 to −3.05 kg).

The current study examines early weight loss and its predictive ability in a mobile app‐based intervention that emphasized self‐monitoring for weight loss (the GoalTracker trial). Hypotheses included that early weight loss (at 1 month) would predict greater weight loss at 3 and 6 months, as well as greater intervention engagement.

Methods

GoalTracker was a randomized controlled trial designed to promote weight loss via a 12‐week standalone intervention delivered via a smartphone app ¹⁴. Devised from self‐regulation theories, all three treatment arms received the following core intervention components: MyFitnessPal (a free commercial app) for self‐monitoring, a goal to lose 5% of initial weight by 12 weeks, a tailored daily calorie goal and weekly weight loss goal, and in‐app daily push reminders that would be automatically sent if tracking had not occurred by the evening. These reminders were programmed by the study staff in the MyFitnessPal app during the baseline visit.

Results

The 84 participants in the current analysis were predominantly female (81%) and college educated (85%), with a mean (SD) age of 43.7 years (11.6). A majority of the sample had obesity (58%). Roughly one‐third (36%; n = 30) of participants were categorized as early responders (i.e. achieving ≥2% weight loss at 1 month), while 64% (n = 54) were early non‐responders. Baseline characteristics did not differ by early weight loss status (Table 1).

Weight change

Per cent weight change at 1 month was positively associated with per cent weight change at 3 (r = 0.77) and 6 months (r = 0.51), with 58% and 25% of the variance accounted for by 1‐month per cent weight change, respectively. For every 1% weight loss at 1 month, there was a 1.50% and 1.38% increase in weight loss at 3 and 6 months, respectively.

Early responders had significantly greater per cent weight loss at 1, 3 and 6 months, compared with early non‐responders (ps < 0.0001; Table 2). The odds of achieving ≥3% and ≥5% weight loss at 3 months were 21.86 (95% confidence interval [CI]: 5.83, 81.99) and 9.72 (95% CI: 3.02, 31.34) times higher, respectively, among early responders, compared with early non‐responders. Similarly, at 6 months, the odds of achieving clinically significant weight loss were significantly greater among early responders than early non‐responders (for ≥3% weight loss: odds ratio, 6.26; 95% CI: 2.14, 18.28; for ≥5% weight loss: odds ratio, 6.93; 95% CI: 2.33, 20.62).

Table 2

Outcomes of early responders vs. early non‐responders

	N	Early responders ≥ 2% at 1 month (n = 30)	N	Early non‐responders < 2% at 1 month (n = 54)	P value of between‐group difference
% weight change from baseline, mean (95% CI), kg
1 month	30	−3.37 (−3.88%, −2.85%)	54	−0.43 (−0.81%, −0.04%)	<0.0001
3 months	28	−5.93 (−6.82%, −5.03%)	45	−1.45 (−2.15%, −0.75%)	<0.0001
6 months	26	−5.91 (−7.33%, −4.48%)	45	−1.28 (−2.37%, −0.19%)	<0.0001
Intervention engagement, median [IQR]
% action plans completed†	17	72.7 [27.3]	39	45.5 [72.7]	0.02
% of days tracked weight†
Weeks 1–4	17	96.4 [10.7]	39	85.7 [42.9]	0.0018
Weeks 5–12	17	92.7 [29.1]	39	67.3 [76.4]	0.0063
Weeks 1–12	17	90.4 [19.3]	39	73.5 [57.8]	0.0027
% of days tracked diet
Weeks 1–4‡	24	100.0 [5.4]	32	78.6 [71.4]	0.0006
Weeks 5–12	30	73.6 [50.9]	54	27.3 [78.2]	0.0027
Weeks 1–12§	24	82.5 [34.3]	32	46.4 [66.9]	0.008

Open in a separate window

^† Among participants in the Simultaneous and Sequential arms only (n = 56), because App‐Only arm was not asked to track body weight or complete action plans.

^‡ Among participants in the Simultaneous and App‐Only arms only (n = 56), because Sequential arm was not asked to track dietary intake during this period.

^§ Among participants in the Simultaneous and App‐Only arms only (n = 56), because Sequential arm was not asked to track dietary intake during the entire period.

CI, confidence interval; IQR, interquartile range.

Most (89.3%) early responders went on to achieve ≥3% weight loss at 3 months, compared with 24.4% of early non‐responders. In terms of ≥5% weight loss at 3 months, 57.1% of early responders and 11.1% of early non‐responders achieved this threshold. Similar patterns were observed at 6 months (see Figure 1).

Open in a separate window

Figure 1

Proportion of participants achieving clinically significant weight loss, by early weight loss status

Discussion

In a standalone app‐based intervention focused on self‐monitoring, over half of early responders met their goal of losing 5% of initial weight by 3 months and maintained this clinically significant weight loss at 6 months. In contrast, over 8 in 10 participants who did not respond early to treatment were unsuccessful in achieving 5% weight loss at 3 or 6 months. These findings illustrate the importance of early weight loss, which is consistent with past studies of behavioural interventions for weight loss ⁶, ⁷, ⁸. The GoalTracker trial is the first study to demonstrate this relation in the context of a standalone mHealth intervention, which holds promise as a less resource‐intensive and more scalable initial intervention strategy in stepped‐care approaches.

No gold standard exists for when to intervene with non‐responders in behavioural weight loss interventions. While later time points of assessing early weight loss (e.g. in months 2 and 3) are also predictive of overall weight loss ⁷, ²⁵, past trials have found that intervening with non‐responders at 3 months post‐randomization does not translate to improved outcomes ⁹, ²⁶, perhaps reflecting that waiting longer to intervene may be ineffective among those who feel discouraged about their poor early treatment response. Our findings support previous research that suggests 1 month may be an advantageous point at which to intervene ¹² and that engagement likely drives early weight loss success ¹², ²⁵, ²⁷. Future studies are needed that experimentally test the impact of intervening with non‐responders at this early time point.

The results of the present study suggest that practitioners consider the following strategies to optimize early weight loss for their patients who are interested in a standalone intervention: (i) discuss realistic expectations for weight loss in a standalone mHealth intervention, (ii) set both an overall weight loss goal (e.g. 3% or 5% weight loss by 3 months) and a weekly weight loss goal (e.g. 0.5–2 lb week⁻¹), (iii) decide whether to track diet, body weight or both in the initial month and commit to tracking daily, (iv) encourage patients to set reminders, such as automated in‐app push reminders, in order to prompt oneself to track each day, (v) check weight loss progress at the 1‐month mark and (vi) consider altering treatment if less than 2% weight loss was achieved by this point, discussing together other solutions (e.g. refer to a dietitian, use pre‐portioned meals, adhere to a ketogenic or DASH diet, join an in‐person weight loss group or meet with a psychologist to treat maladaptive eating behaviours), if needed. Given that the difference between responding and not responding to treatment in the first month was just a few extra days of self‐monitoring, clinicians should emphasize that daily tracking in the first month will likely prove meaningful and maintain accountability.

Strengths of our study include the collection of objective self‐monitoring data from a commercial mobile app and objective action plan completion data. A limitation of the data is that the smaller sample size yielded wide CIs in the odds of achieving 3% or 5% weight loss by early response status; future studies with larger sample sizes and longer follow‐up should replicate these analyses. Further, because of logistical constraints, in‐person weights at 6 months were unable to be collected, and instead, self‐report was used; relying on self‐report may underestimate weight values. Nevertheless, the response rate at 6 months was consistent with that at 3 months despite no intervention contact during this period, indicating that this remote weight collection method holds promise for use in research with real‐world dissemination potential as it reduces time and effort demands for both participants and staff.

Conclusion

Overall, early weight loss success in a standalone smartphone app‐based intervention predicted clinically significant weight loss and intervention engagement among adults with overweight or obesity. Interventionists and clinicians should discuss upfront with patients the importance of early weight loss in mHealth interventions and emphasize the utility of high levels of engagement (e.g. self‐monitoring daily or almost daily). In clinical settings, measuring patients' early weight loss response would allow providers and patients to make informed decisions together about whether to continue, stop or alter treatment.

Conflicts of Interest Statement

G. G. B. serves on the Scientific Advisory Board of Nutrisystem and Interactive Health and holds equity in Coeus Health. The remaining authors declare no competing interests.

Funding

The research described in this paper was supported by grants to the first author from the American Psychological Association, the Duke Interdisciplinary Behavioral Research Center and the Aleane Webb Dissertation Research Award provided by The Graduate School at Duke University.

Acknowledgements

Notes

Patel M. L., Hopkins C. M., and Bennett G. G. (2019) Early weight loss in a standalone mHealth intervention predicting treatment success, Obesity Science & Practice. 5, 231–237, 10.1002/osp4.329. [Europe PMC free article] [Abstract] [CrossRef] [Google Scholar]

References