Can targeted recruitment methods affect male participation in weight loss interventions and trials?

Abstract

Background

Behavioral weight loss studies often enroll more females than males despite similar obesity prevalence. There is a paucity of experimental evidence on recruitment strategies to attract male participants in behavioral trials broadly and in weight management trials, specifically.

Methods

We examined 4 theory and evidence-based advertising messages and 4 recruitment modalities (mailing, emailing, Facebook, and Spotify) in three iterative waves during a decentralized longitudinal randomized controlled trial comparing 2 behavioral weight loss interventions.

Results

The number of ad clicks and web screens completed did not appear to differ among the messages. Facebook was the most cost-effective overall ($32.86 per recruited person) with the highest proportion of male interest when the message was tailored and campaigns were set to target males (79%), but mailing produced the higher total rate of potential participants screened and garnered a higher proportion of male interest overall (23% average across all waves), albeit it at greater cost.

Conclusions

Social media is a cost-efficient recruitment avenue, although it may reduce the ability to assemble a representative sample. Future studies should formally experiment with recruitment to inform robust recommendations.

Introduction

While awareness of the health repercussions of obesity is growing, many states in the United States have an obesity prevalence greater than 30% according to the 2025 CDC report.¹ Although there is no significant difference in prevalence between males and females, females are more likely to participate in weight loss interventions.^2,3 Weight loss intervention trials are oversaturated with females and only 27% of participants that complete these interventions are male,⁴ a pattern evident in clinical trials testing behavioral weight loss strategies.⁵ Current literature indicates a disconnect between male and female perceptions of weight and appearance with males being less likely to perceive themselves as overweight and less likely to attempt weight loss.⁶ Females are often more concerned about weight loss and their physical appearance whereas males are more likely to focus on exercise and health⁷ and advertisements (ads) for both weight loss programs and behavioral trials tend to capitalize on the former rather than the latter.⁸ Thus, there is a pressing need to adjust and test advertising strategies to recruit males into behavioral weight loss trials.

Some research has shown that targeted messaging techniques can influence recruitment outcomes.⁹ For example, ads with images tend to get more attention¹⁰ and some studies have depicted images of males being active to successfully recruit more males.^11,12 Messages in ads, such as focusing on a threat of illness as opposed to an appeal to body image could increase an ad’s relevance for males, who are more likely to be concerned about health than body image.¹³ There are significant gaps in our knowledge because to date most conclusions are derived from observational studies and many recruitment factors have not been experimentally tested. Instead, most of what we know about trial recruitment comes from reports of what investigators have attempted during trials, without systematic investigation. Thus, randomizing aspects of an ad, such as message themes across modalities (eg, mail, social media, etc.) could inform best practices to recruit representative samples.

One promising development in the post-COVID-19 landscape is the rise of remote and decentralized trials which have a purported benefit of casting a wider recruitment net outside of the geographical boundaries of a trial site. Recruitment that leverages such benefits likely requires a multi-method strategy and few studies have sought to examine the ability of these strategies to recruit a diverse sample to fully realize the promise of decentralized trials as being more accessible to a wider range of the population. One review of decentralized trials noted that a wide range of modalities can be used and that a mix of online vs offline strategies is needed to reach a range of individuals, particularly those with limited access to online sources of media, yet little information exists about the mix that would be effective for recruiting a representative sample on other individual characteristics.² Specifically, we lack evidence supporting effective strategies to inform the selection of messages and modalities to successfully recruit males into weight loss trials, particularly for decentralized trials. Therefore, the purpose of our research was to systematically examine ad variations and their performance across recruitment modalities to determine their overall effectiveness in attracting male participants within the context of a decentralized trial comparing 2 behavioral weight management programs.

Methods

Overview

This study examined recruitment of males in Elements Vital to treat Obesity (EVO), a behavioral weight loss study. EVO was a 2-arm comparative effectiveness study that randomized participants to 1 of 2 active weight management programs, the protocol of which was previously published.¹⁴ We conducted 3 waves of ads, and randomized messages within modalities to understand responsiveness of males to each message type and modality. The primary goal was to improve recruitment for an ongoing trial which necessitated practical strategies, so we did not power a full trial or conduct statistical testing for significance. Rather, we provide a descriptive analysis that compares messages and modalities across 3 iterative waves of testing. We used QR codes and bit.ly links to determine study website visits from each unique advertisement. To participate in EVO, participants were provided with a link to a website that explained the study and included another link to an eligibility questionnaire (Web Screen) in REDCap, which asked how the participant heard about the study. A participant was considered “recruited” when they completed a web screen regardless of whether they were deemed eligible.

Modalities

We tested 4 recruitment modalities. We endeavored to acquire the most similar sample from each modality given the modality service provider’s constraints on segmenting its data, which provided some variability as described below.

Mailing lists

Paid mailing lists were acquired from targeted households who indicated obesity from the Complete Ailments & Illness Suffers list and were aged 16-80 years. We sent U.S. Postal Service 4 × 6 postcards to all individuals from this list, randomly selecting for each wave in which we used the list.

Email lists

Paid email lists were acquired from Direct Media Communications targeted individuals, aged 18-65 years, interested in weight loss. We sent emails to all individuals from this list.

Facebook

Paid ads on Facebook targeted users aged 18-65 years and older.

Spotify

Ads on Spotify targeted users aged 18-65 years and older.

Messages

We tested 4 unique messages focusing on different aspects of language manipulation to target males.

Threat of illness

The first message, “threat of illness,” used specific health threats to motivate enrollment and perceive weight loss interventions as useful.¹³ The language focused on excess weight leading to an increased risk for heart disease, cholesterol, and diabetes.

Health information

The second message, “health information,” included information about weight loss interventions that was more accessible by providing clear, concise health content related to weight.¹⁵ This content focused on the positive health outcomes of weight loss as opposed to diseases related to excess weight.

Research transparency

Emphasizing voluntariness and protection of rights when joining the clinical study, the third message aimed to alleviate potential fears of harm typically associated with clinical trials, particularly for those with a mistrust of institutions.¹⁵

Personalization

Lastly, the “personalization based on identity or interest” messaging focused on tailoring messages to individuals and increasing personal relevance.⁹ Because men are more likely to focus on physical health,⁷ the messages included pictures of men exercising and language focus on building strength.

All messaging, images, and advertisements directing to the EVO study website with an online prescreen survey were approved by the Northwestern University Institutional Review Board (IRB) before use. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.

Three waves of testing

Wave one

Wave one applied all 4 messages across all 4 modalities. Ads were randomized and sent to mail and email recipients. Ads were also set to be randomly and equally distributed for Facebook and Spotify, but both platforms have algorithms that engage audiences beyond a user’s control. All ads began on the same day. The target was to distribute 10 000 unique emails and postcards each and to reach 10 000 individuals from Facebook and Spotify each as noted by the dashboard for each platform.

Wave two

The second wave used only the paid mailing list and Facebook modalities based on their success in Wave 1. We incorporated an ad previously used by the EVO study team that appeared to attract more males as compared to Wave 1 results and used a “call to action” phrase as suggested by prior literature.¹³ For this wave, we randomized sending 2 messages, a threat of illness message and a call-to-action phrase like “Attention.” The target reach was 10 000 individuals from the mailing list, and a budget cap of $500 was used rather than impressions for Facebook.

Wave three

The final wave used the paid mailing list and Facebook based on their continued success. We tested 2 final versions of the advertisements. In 1 ad, we used a threat of illness message and the color and design from Wave 2. For the second ad we augmented it using strategies from the literature in addition to our “call to action.” The second ad used: (2) cool color tones rather than the previous ads’ purple scheme,¹⁶ (3) the study logo and font were altered to be bolder,¹⁷ and (4) a call-to-action phrase (“Take action today!”). The reach and cost settings stayed the same as in Wave 2, but Facebook was set to specifically target male users.

Statistical analysis

Descriptive statistics were used to produce results from each wave for each message and modality. Statistical tests were not appropriate due to the small amount of data available in some cases and the variety of restrictions and controls available across recruitment modalities. We experienced a wide range of controls and requirements, and we also witnessed changes to available advertising features on platforms during our tests making direct comparisons impossible. We examined reach primarily through unique link engagement with our study site, cost per click obtained, the number of completed web screens in total and by males, and price per completed web screen. Findings should be considered preliminary yet informative for future study development.

Results

Wave one

As shown in Table 1, to reach 10 000 individuals, Facebook ran for 9 days while Spotify ads ran for 7. The mailing list recruited the most male subjects, as indicated in Table 1, where 12 of the completed web screens out of 54 total were male (22.2%). The mailing list additionally had the highest count of completed web screens (n = 54). Facebook was the most cost-effective option at $0.55 per click and $4.96 per completed web screen. Spotify recruited the least subjects in terms of completed web screens (n = 1) and had the highest overall cost per web screen ($246.05), although it did generate 25 ad clicks and reached 11 237 listeners.

Table 1.

Results across all 3 waves by modality.

Wave	Modality	Reach	Clicks	Completed web screens	% of screens that were male	Total cost $USD	Cost per any completed screen $USD	Web screen/reach
1	Mailing	10 000	124	54	22.2	7764.00	143.78	0.54
1	Emailing	10 000	293	42	19	3000.00	71.43	0.42
1	Facebook	10 048	154	17	18	84.37	4.96	0.17
1	Spotify	11 237	25	1	0	246.05	246.05	0.0089
2	Mailing	10 000	106	82	22	8150.00	99.39	0.82
2	Facebook	57 648	902	60	5	500.00	8.33	0.10
3	Mailing	10 000	55	74	26	7762.00	104.89	0.74
3	Facebook	62 529	791	34	79	500.00	14.71	0.054

As shown in Table 2, the total number of clicks and the percentage of male clicks between the 4 message types appeared to be similar.

Table 2.

Total clicks by modality and message type in Wave 1.

Modality	Message	Reach	Clicks	Clicks/reach (%)
Mailing	Threat of illness	2500	32	1.28
	Transparency and privacy	2500	33	1.32
	Personalization	2500	29	1.16
	Health information	2500	30	1.20
Emailing	Threat of illness	2500	78	3.12
	Transparency and privacy	2500	73	2.92
	Personalization	2500	74	2.96
	Health information	2500	68	2.72
Facebook	Threat of illness	2594	42	1.62
	Transparency and privacy	3436	38	1.11
	Personalization	2627	49	1.87
	Health information	1631	25	1.62
Spotify	Threat of illness	2874	5	0.17
	Transparency and privacy	2907	5	0.17
	Personalization	2897	7	0.24
	Health information	2793	8	0.28
Total	Threat of illness	10 648	157	1.47
	Transparency and privacy	11 343	149	1.31
	Personalization	10 524	159	1.51
	Health information	9424	131	1.39

Wave two

Because the reach restriction was relaxed and a budget limit was set instead, Facebook reach in Wave 2 was higher than in Wave 1. However, Facebook’s increased reach did not generate more completed web screens between Wave 1 (0.17%) and Wave 2 (0.10%). Facebook continued to have the lowest cost out of the modalities used in Wave 2 (Table 1).

In comparing messages in Wave 2, the “threat of illness” message reached 801 clicks with 22% of total clicks being male and the call to action reached 101 clicks with 68% of those being male.

Wave three

The results for Wave 3 were similar to Wave 2, with Facebook recruiting more males that completed web screens overall (n = 27) as compared to the mailing list, (n = 19) resulting in lower cost per screen for Facebook (Table 1).

In testing the 2 ad versions, 63% of the total ad clicks in Wave 3 came from the threat of illness ad while 37% of clicks came from the call-to-action ad.

Discussion

In this brief report, we examined the performance of recruitment methods to engage males in a decentralized behavioral weight loss clinical trial. Facebook was the most cost-effective platform for advertising overall and for male recruitment ($32.86 per male), while generating the highest number of clicks in this study, particularly when campaigns were set to target men. Mailing lists generated more completed web screens (n = 210) and total male interest (23% average) across 3 waves, though at a higher cost ($238.67). In sum, while social media advertisements might offer an economical solution and high quantity of interested participants, traditional study recruitment methods could provide more interest per impression and a more representative sample. Investigators carrying out decentralized trials using such recruitment methods should carefully balance a variety of modalities that include analog and not online efforts even when all study protocols are conducted online or remotely.

It is worth noting that despite efforts to target males, female continued to enroll at higher rates than male, even when social media was set to target males. We posit that this is due to a greater overall interest in weight management for health and fitness overlaps among women and men, but other reasons for weight management may not be as salient for men.⁷ While this might lead to investigators using ads that specifically target males only when females will enroll regardless, it was our experience that females made note of feeling that ads were not welcoming to them. Thus, there should be careful consideration for executing a full recruitment plan that is open and welcoming to a variety of potential participants, although additional efforts may be needed for certain groups, such as when recruiting males.

Limitations and future directions

Many variables could contribute to responsiveness to the ads that were not part of this study due to the pragmatic nature. Additionally, many constraints such as budgetary, recruitment timeline for the trial, online ad service provider restrictions, and lack of ability to target specific audiences made execution of an experimental design impossible and clear conclusions elusive. Nonetheless, attempts to systematically examine ads, such as in this study, may serve as a starting point or template for thinking about approaching the study of recruitment as a primary endeavor rather than an afterthought. An additional limitation was being unable to verify whether the individual who clicked on the ad was indeed the account holder on Facebook or Spotify, making some of our conclusions open to error. Future work should be pursued to directly test recruitment strategies to provide more robust recommendations for behavioral trialists.

Author contributions

Ivelina Arnaoudova (Funding acquisition [equal], Investigation [supporting], Writing—original draft [lead], Writing—review & editing [equal]), Alice Pham (Visualization [supporting], Writing—review & editing [supporting]), Laura Scanlan (Conceptualization [supporting], Data curation [supporting], Investigation [supporting], Project administration [equal], Supervision [supporting], Writing—review & editing [supporting]), Charlie Olvera (Data curation [lead], Formal analysis [lead], Methodology [supporting], Validation [equal], Writing—review & editing [supporting]), Alexis Hernandez (Conceptualization [equal], Funding acquisition [equal], Writing—review & editing [equal]), and Angela Fidler Pfammatter (Conceptualization [lead], Funding acquisition [supporting], Methodology [lead], Project administration [lead], Supervision [lead], Validation [lead], Writing—original draft [supporting], Writing—review & editing [lead])

Funding

This work was supported by The Department of Preventive Medicine Mission Fund from Northwestern University Feinberg School of Medicine awarded to A.H. as well as a National Institute of Diabetes and Digestive and Kidney Diseases award (R01DK125749) awarded to A.F.P. The content is solely the responsibility of the authors and does not necessarily represent the official views of Northwestern University Feinberg School of Medicine or the National Institutes of Health.

Conflicts of interest

The authors declare none.

Open science statement

The study presented was not pre-registered. The clinical trial referenced in the manuscript is registered at clinicaltrials.gov under NCT04708769. The analytic plan for the study presented was not pre-registered. Data and analytic code for this study and the resulting clinical trial will be made available at a later date. Data and analytic code for this study may be requested from the corresponding author. Study materials are also available from the corresponding author.