Randomized Control Trial of Unconditional versus Conditional Incentives to Increase Study Enrollment Rates in Participants at Increased Risk of Lung Cancer

Abstract

Introduction:

Understanding how incentives and their timing influence study enrollment rates is important to efficient study design and increasing the generalizability of findings. This 2-arm, parallel randomized trial evaluated how conditional versus unconditional mailed incentives of a $20 gift card affected study enrollment in a sample of participants screened for lung cancer screening.

Methods:

Eligible participants included Black and White adults who underwent lung cancer screening with low-dose CT and had negative screening results at two North Carolina imaging facilities in 2018. We used a stratified randomization scheme, by sex and race, to assign incentive type (conditional vs. unconditional). We used the Tailored Design Method with six points of mailed contact to engage participants. We compared study enrollment rates using chi-square tests and logistic regression analyses.

Results:

After adjusting for sex, race, age, smoking status, participant residence, and screening site, participants who received unconditional incentives were 74% more likely to enroll than those who received conditional incentives (adjusted OR= 1.74 (95% CI: 1.01, 3.00).

Conclusions:

Type of incentive can play a role in increasing study enrollment, especially mailed surveys that target individuals who currently or previously smoked. Unconditional incentives may be worth the initial cost to engage study participants.

1. Introduction

In recruiting individuals to participate in studies involving completion of surveys and questionnaires, high response and enrollment rates are essential to obtaining generalizable results and representative of populations of interest. Over the last two decades, there have been shifts in methods for recruitment from telephone and mailing approaches to online venues. However, online recruitment may not be appropriate for all populations. For example, several studies have reported higher response rates in older populations when using mailed versus online methods [1–3]. The use of survey incentives, such as gift cards, cash, and prepaid checks, has also been shown to increase study enrollment and response rates, in particular for mailed surveys [1, 2, 4, 5].

Prior studies reported that the type of incentive (conditional versus unconditional) has also been shown to influence participants’ behavior. Conditional incentives are incentives given upon completion of a survey while unconditional incentives are those given regardless of survey completion and often before or at the time of survey administration. Several studies have demonstrated that use of unconditional incentives results in an increased enrollment rate compared to conditional incentives that are given upon completion of surveys [5–7], while others have indicated minimal differences in response and enrollment rates based on incentive type [8–10]. Additionally, previous non-respondents appear to have higher enrollment rates when receiving multiple mailings and unconditional incentives at the time of questionnaire administration [11].

Of studies conducted on incentives and mailing methods, few studies focus on participants that formerly or currently smoke, dictating a need for review of survey methods in this population. As a stigmatized population, individuals who currently smoke may be less likely to engage in research studies than those who previously smoked [12, 13]. Unconditional incentives may encourage individuals who currently smoke to enroll in research studies and respond to health-related questionnaires at higher rates [14]. The target population of this study focuses on individuals who currently smoke or have previously smoked and who underwent lung cancer screening in North Carolina. To assess the effect of differing mailed incentive strategies, we conducted a stratified randomized trial to determine the effects of using conditional and unconditional incentives on study enrollment. Study enrollment included completion of three documents (a health history questionnaire, a consent form, and HIPAA waiver form), hereafter referred to as survey. The hypothesis was that an unconditional incentive included with the survey would result in a higher study enrollment rate than a conditional incentive, regardless of participant characteristics.

2. Methods and Materials

This study to evaluate mailing enrollment rate methods was part of a larger study to evaluate lung cancer screening patterns and outcomes through the North Carolina Lung Cancer Screening Registry (NCLSR). This study was approved by the University of North Carolina Institutional Review Board on October 30, 2018.

2.1. Sample Selection and Incentive Randomization

Study participants were selected from two imaging sites that are part of the NCLSR. All patients who underwent lung cancer screening with low-dose CT between January 1, 2018, and December 31, 2018, were Black or White, had a negative screening result, and were alive at the time of study randomization in January 2019 were included. The study sample included those who underwent LCDT for lung cancer screening and met the 2013 US Preventive Services Task Force (USPSTF) recommended criteria, which includes those ages 55 to 80 years, who currently smoke or previously smoked (quit within the last 15 years) with a 30 pack-year smoking history. We contacted potential participants after their lung cancer screening scan was complete, and potential participants were unaware of this study until we contacted them to participate.

The data manager used a random number generator to randomize participants to receive the conditional versus conditional incentive. Study personnel who were involved with creating and mailing study packets were not involved with the randomization but were aware of how participants were randomized since the mailing materials differed among the two groups. The mailing material packets were created in a bulk, uniform manner, and the mailing address was then included on each packet. When the packets were returned, study personnel entered this information blinded as to the randomization status of the participant. We limited to Black and White participants as only 4.1% of patients undergoing lung cancer screening at these two sites reported belonging to other race groups. Eligible participants were stratified into 4 groups based on sex (female, male) and race (Black, White) and then randomized within the 4 strata to receive either (1) a $20 gift card upfront as part of the survey (unconditional), or (2) a $20 gift card upon returning the survey (conditional). We stratified based on sex and race to ensure balance between the unconditional and conditional incentives types for these groups and to allow for examination of results within these subgroups [15].

2.2. Study Enrollment Procedures and Survey Administration

We used the Tailored Design Method to deploy the survey, with six points of contact to engage participants [16]. Eligible participants were mailed an announcement, pre-notification postcard (1^st point) in January 2019 to introduce the study. A study packet (2^nd point) containing four documents (health history questionnaire, informational brochure, HIPAA waiver, and consent form) was mailed one week after the pre-notification postcard, with the incentive included in this packet for those who were randomized to the unconditional incentive group. A first reminder postcard (3^rd point) was mailed one week after the initial study packet was sent to all participants. A second reminder postcard (4^th point) was sent three weeks after the initial study packet to those participants who had not responded. A second study packet (5^th point) was mailed seven weeks after the initial study packet to participants who had not yet responded and included another copy of the survey, informational brochure, HIPAA waiver, and consent form, with no incentive included. If participants requested a study packet in between the first study packet mailing and the second study packet mailing, a study packet was sent. A final reminder postcard (6^th point) was sent to participants. While the majority of the packets were expected to be returned within eight weeks, receipt continued through 11 weeks after the first notification was sent. If participants contacted study staff and indicated they did not want to participate or had moved away, the participants were not sent additional study packets or reminders. Upon return of the completed study packet, conditional incentives were mailed to participants randomized to the conditional incentive group.

2.3. Statistical Analyses

We excluded participants who opted out of the study, moved, or were deceased (n=17). For each participant, we collected race, sex, age (dichotomized into <65, 65+), smoking status (current, former), participant residence, and screening site (Site 1, Site 2). Participant residence was derived from the zip code of residence and categorized using the USDA Rural-Urban Commuting Area (RUCA) codes to determine if participants lived in urban or rural environments. RUCA codes of 1, 2, or 3 were classified as urban, while RUCA codes of 4, 5, 6, 7, 8, 9, or 10 were classified as rural. Smoking status was documented in the electronic medical record. We excluded two participants who had never smoked and two participants for whom smoking status was not assessed.

Our main outcome of interest was enrollment rate. We compared the demographic characteristics among the conditional and unconditional incentive groups using chi-square tests. We determined if study enrollment rates differed by incentive group, overall, and for each demographic subgroup using chi-square tests and univariate logistic regression. Finally, we used multivariate logistic regression to assess the odds of study enrollment based on incentive type, sex, race, age group, smoking status, participant residence, and screening site. All analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC).

3. Results

3.1. Study Population

Among the 223 participants in this study, 112 were randomized to the unconditional incentive group and 111 were randomized to the conditional incentive group (Table 1). Because of the stratified randomization, a similar proportion of Black and White participants, and male and female participants were in each incentive group – 18.8% of unconditional incentive participants and 19.8% of conditional incentive participants were Black. Those less than 65 years of age were more likely to be randomized to the conditional incentive group (59.5%) compared to the unconditional incentive group (40.2%) (p=0.0040). A similar proportion of participants were randomized to incentive type based on smoking status, participant residence, and screening site.

Table 1.

Demographic characteristics of the study participants

	Unconditional Incentive Group	Conditional Incentive Group
Characteristic	N (%)	N (%)	p-value
ALL	112	111
Race			0.8395
Black	21 (18.8)	22 (19.8)
White	91 (81.3)	89 (80.2)
Gender			0.5242
Female	45 (40.2)	40 (36.0)
Male	67 (59.8)	71 (64.0)
Age (years)			0.0040
<65	45 (40.2)	66 (59.5)
65+	67 (59.8)	45 (40.5)
Smoking status			0.6421
Former	43 (38.4)	46 (41.4)
Current	69 (61.6)	65 (58.6)
Participant Residence*			0.9624
Rural	37 (33.0)	37 (33.3)
Urban	75 (67.0)	74 (66.7)
Screening Site			0.7420
Site 1	48 (42.9)	50 (45.1)
Site 2	64 (57.1)	61 (55.0)

^*based on participant zip code and categorized using Rural-Urban Commuting Area (RUCA) codes

^**all p-values are chi-square

3.2. Enrollment Rates by Sociodemographic Characteristics

Participants in the unconditional incentive group tended to have higher enrollment rates compared to those in the conditional incentive group (54.5% vs. 41.4%, respectively), although this was not statistically significant (Table 2). Males in the unconditional incentive group were more likely to enroll than males in the conditional incentive group (56.7% versus 33.8%, p=0.0068). In contrast, the enrollment rate did not differ for female participants based on the type of incentive. Black participants tended to have higher enrollment rates with unconditional versus conditional incentives (52.4% vs. 31.8%; p = 0.1719), and while this was not statistically significantly different, it warrants additional investigation with larger studies. Unconditional versus conditional incentives resulted in higher enrollment rates among participants who used to smoke (69.8% vs. 41.3%, p=0.0070) but not among those who currently smoke (41.5% vs. 44.9%, p=0.6923). Additionally, urban residents were more likely to enroll if the incentive was unconditional versus conditional (58.7% vs. 39.2%, p=0.0174).

Table 2.

Demographic characteristics and study enrollment by type of incentive

Characteristic	Unconditional Incentive Group N= 112		Conditional Incentive Group N= 111
	Enrolled n (row %)	Did not Enroll n (row %)	Enrolled n (row %)	Did not Enroll n (row %)	p-value
ALL	61 (54.5)	51 (45.5)	46 (41.4)	65 (58.6)	0.0516
Race
Black	11 (52.4)	10 (47.6)	7 (31.8)	15 (68.2)	0.1719
White	50 (54.9)	41 (45.1)	39 (43.8)	50 (56.2)	0.1356
Gender
Female	23 (51.1)	22 (48.9)	22 (55.0)	18 (45.0)	0.7199
Male	38 (56.7)	29 (43.3)	24 (33.8)	47 (66.2)	0.0068
Age (years)
<65	25 (55.6)	20 (44.4)	28 (42.4)	38 (57.6)	0.1739
65+	36 (53.7)	31 (46.3)	18 (40.0)	27 (60.0)	0.1539
Smoking status
Former	30 (69.8)	13 (30.2)	19 (41.3)	27 (58.7)	0.0070
Current	31 (44.9)	38 (55.1)	27 (41.5)	38 (58.5)	0.6923
Participant Residence*
Rural	17 (45.9)	20 (54.1)	17 (45.9)	20 (54.1)	1.0000
Urban	44 (58.7)	31 (41.3)	29 (39.2)	45 (60.8)	0.0174
Screening Site
Site 1	27 (56.3)	21 (43.7)	20 (40.0)	30 (60.0)	0.1075
Site 2	34 (53.1)	30 (46.9)	26 (42.6)	35 (57.4)	0.2401

^*based on participant zip code and categorized using Rural-Urban Commuting Area (RUCA) codes

^**all p-values are chi-square

3.3. Multivariable Regression

Unadjusted odds ratios (OR) and 95% confidence intervals (95% CI) show similar results as above (Table 3). In the multivariate logistic regression model adjusting for sex, race, age, smoking status, participant residence, and screening site, participants who received the unconditional incentive versus the conditional incentive were 74% more likely to enroll in the study (adjusted OR= 1.74 (95% CI: 1.01, 3.00). With adjustment for other factors, differences by sex, race, age, smoking status, and participant residence were no longer statistically significant.

Table 3.

Multivariate logistic regression results of the association between incentive type and study enrollment

Characteristic	Jnadjusted Odds Ratio (95% Confidence Interval)	Adjusted Odds Ratio* (95% Confidence Interval)
Incentive: Unconditional vs. Conditional	1.67 (0.99, 2.83)	1.74 (1.01, 3.00)
Race: Black vs. White	0.74 (0.37, 1.44)	0.74 (0.37, 1.47)
Gender: Female vs. Male	1.42 (0.83, 2.44)	1.32 (0.76, 2.29)
Age: <65 vs. 65+	1.02 (0.61, 1.71)	0.88 (0.51, 1.53)
Smoking: Forme, vs current	1.53 (0.90, 2.61)	1.59 (0.91, 2.78)
Residence: Rural vs. Urban	0.85 (0.49, 1.48)	0.78 (0.42, 1.47)
Screening Site: Site 1 vs. Site 2	0.97 (0.57, 1.63)	1.07 (0.59, 1.93)

^*Adjusted for all other characteristics shown in Table 3.

4. Discussion

The results of this randomized trial show that unconditional incentives included in mailed study recruitment to participants at increased risk for lung cancer who are undergoing lung cancer screening, result in higher study enrollment compared to conditional incentives. Unconditional incentives increased enrollment in specific groups of participants such as men, those who used to smoke, and urban participants. Interestingly, our results suggest that unconditional incentives may increase study enrollment among Black participants (n=43; 52.4% enrollment rate in the unconditional incentive group vs. 31.8% enrollment rate in the conditional incentive group), although this finding was not statistically significant. Additional evaluation in a larger sample is needed and could be a strategy to increase study enrollment in Black participants at similar levels seen in White participants. In our study, unconditional incentives had similar enrollment rates in White and Black participants (54.9% vs. 52.4%, respectively).

Our results are consistent with previous studies that support the use of unconditional incentives to increase study enrollment and response rates compared to conditional incentives in postal surveys [1, 2, 10, 17] among participants residing in the Southeastern United States [6], and among non-white participants compared to white participants [10]. Edwards et al. found that the odds of response increased 61% when incentives were unconditional compared to conditional incentives [1], which is comparable to our findings of a 74% increase in enrollment. Wiant et al. found prepaid incentives increased absolute enrollment rates by about 15% compared to postpaid incentives for participants living in the Southeastern United States [6]. In our study, there was an absolute 13.1% increase in enrollment rates among those who received the unconditional versus the conditional incentives.

Similar to our findings, previous studies found that urban participants respond at higher rates to unconditional incentives than rural participants [18]. In studying the effect of an unconditional incentive, Parkes et al. observed a 20.4% absolute increase in enrollment rates in urban participants and a 1.0% absolute increase in rural participants [18]. We saw a similar 19.5% absolute increase in enrollment rates from unconditional and conditional incentives in urban participants but no increase in enrollment rates by incentive type among rural participants. While current literature supports various aspects of our results, few studies have been conducted on the effectiveness of unconditional incentives to increase enrollment for populations who currently smoke or have previously smoked in the United States, with the closest comparable study on those who smoke being conducted in Hong Kong [14]. In studying the effect of a mixed prepaid and postpaid incentive combination on those who smoke in Hong Kong, Cheung et al. found an enrollment rate of 40.8%, comparable to our enrollment rates of 54.5% and 41.4% in the unconditional and conditional incentive groups, respectively. Our results provide more insight into a population of older individuals who currently smoke or have previously smoked, often experiencing smoking-related stigma in the US and may thus have unique barriers to recruitment.

Unconditional incentives may be particularly effective in a population facing smoking-related stigma. Several studies have shown that the stigma associated with smoking is a barrier to engaging in healthcare and participating in research for individuals who smoke in the United States. Carter-Harris et al. [13] and Jonnalagadda et al. [12] reported that individuals who smoke are a stigmatized population in the United States who have less engagement with and a higher mistrust of healthcare systems, in particular, screening programs and research studies. Unconditional incentives and gifts included in surveys increase and establish trust in the social exchange process between participants and researchers, particularly for participants who identify with stigmatized groups [16, 19]. With increased trust, enrollment rates increase as participants are more likely to complete the incentivized task, according to Dillman [16]. Unconditional incentives can also increase norms of reciprocity between the participant and researcher, while conditional incentives do not increase these norms because of their compensational nature [19, 20]. Thus, unconditional incentives can result in higher enrollment rates than conditional incentives, as our results demonstrate.

Some limitations of this study lie in the specific sample represented and studied. Regarding race, only two categories (Black and White) were included due to the small number of participants who were other racial groups of Asian, American Indian, and Other. Additionally, Hispanic ethnicity was not evaluated due to the small number of Hispanic participants. Some subgroups of our study population were relatively small, suggesting the need for further research in larger samples. Although subgroup sample sizes were not large enough to observe a significant association, our findings are informative in understanding the impact of incentive type on enrollment rates in a population less likely to engage with health systems.

This study provides an evaluation of unconditional and conditional incentives used as part of study enrollment among individuals who currently smoke and have previously smoked, contributing to the knowledge of effectiveness in mailing and incentive methods. Unconditional incentives increased enrollment rates via postal methods and may be worthwhile if studies can bear the cost initially.

Figure 1:

Flow Diagram of Enrollment

What is new?

• Key findings:

  • When compared to conditional incentives, unconditional incentives may result in increased study enrollment rates in a lung cancer screened population.
  • In this study, unconditional incentives significantly increased enrollment in men, those who used to smoke, and urban residents, compared to conditional incentives.
• What it adds:

  • Given the lack of studies focused on best methods to increase study enrollment in individuals who smoke or who used to smoke, this study demonstrates the impact of different incentive types in a population that may be less engaged with the health system overall.
• Implications:

  • If studies can bear the cost, unconditional incentives may be worthwhile to increase engagement for difficult-to-reach populations, such as those who currently smoke or have previously smoked.