To maximize the public health impact of diabetes lifestyle interventions, programs must attract their intended audience, including ethnically diverse and high-risk participants. One way of assessing program reach is to examine differences between eligible enrollees and non-enrollees. Our analyses revealed few differences between participants and nonparticipants in ¡Viva Bien!, suggesting that a representative sample of Latinas was obtained. Other studies have reported few significant differences between Latino participants and eligible nonparticipants2,3
, noting only differences in education and age2
, and gender and language preference.
The reach of ¡Viva Bien! was likely enhanced by attractive program characteristics such as the lack of a fee, bilingual staff, culturally appropriate materials in Spanish and English, and affiliation with one’s regular health-care system. The program attracted (nonsignificantly) more eligible Salud than KP patients by percentage, suggesting that it reached lower-income, ethnically diverse, and high-risk populations as well as those with more economic resources. Latina participants represented a range of acculturation, and were drawn from both low- and higher-income levels. Language preference did not play a significant role in ¡Viva Bien! participation, nor in the Escobar-Chaves et al2
trial, but significantly more participants than nonparticipants preferred to speak Spanish in Eakin et al.3
The ¡Viva Bien! telephone recruitment strategy, as part of a multi-faceted approach, was highly effective. Most women (61%) reached by telephone who were eligible for the study agreed to participate, and a representative sample was achieved (ie, there were few significant differences between nonparticipants and participants). The greatest barrier to participation was not refusal to take part, but ineligibility due to non-Latino ethnicity (89%) or not having type 2 diabetes (5%), and this was expected given that the patient lists used for recruitment were known to consist mostly of non-Latinas. Our findings suggest that future studies with Latinas might benefit from a similar multi-faceted recruitment strategy, in which the project initiates contact with potential participants, rather than a self-selection strategy, which requires potential participants to initiate contact. While more costly, this approach produces a less-biased sample.
Numerous similarities were found in the results between ¡Viva Bien! and the MLP, the study with mostly Anglo women upon which ¡Viva Bien! was based. Women in the 2 studies gave similar reasons for nonparticipation, primarily being too busy or too time-consuming, with a few saying they were too ill to participate. Reasons for declining participation unique to ¡Viva Bien! included inability to attend the retreat or weekly meetings, and objections to being labeled Hispanic, although the latter objections were raised by very few of the women. These obstacles to participation were similar to those found in the other studies24
. The most common reason given by Latinos for nonparticipation in Robertson25
was lack of information; a review of studies by Ford et al24
of under-represented populations noted that mistrust of research and the medical system was paramount. Trust may have been bolstered in ¡Viva Bien! by endorsements from the women’s health-care providers.
Samples recruited in the ¡Viva Bien! and MLP studies were similar in employment status, smoking status, percent prescribed oral medications, percent prescribed insulin, and waist/hip ratio. Compared to the mostly Anglo MLP women, the ¡Viva Bien! Latinas on average weighed less, had lower BMIs, had higher hemoglobin A1c values, were younger, were diagnosed with diabetes at a younger age, had been taking diabetes medications longer, and had higher incomes (adjusted for inflation) and less formal education. Compared to the MLP participants, more ¡Viva Bien! participants took no diabetes medications or took a combination of insulin and oral medications. ¡Viva Bien! participants were more likely than MLP women to live with their spouse and children, or with children and others; a greater percentage of MLP participants lived alone. Since there were very few differences between participants and nonparticipants in either study, the above differences between MLP and ¡Viva Bien! samples are most likely due to underlying differences in the Latina/Denver and largely Anglo Oregon target populations.
¡Viva Bien! had a higher rate of women who agreed to participate completing baseline assessment (68%) than the MLP (51%), and much higher than other, less-intensive intervention programs with similar ethnically diverse samples. Participation rates were 14% in the Preferences of Women Evaluating Risks of Tamoxifen (POWER) study4
and 27% to 53% in a cancer prevention study of mostly Latina women26
. As in the MLP, ¡Viva Bien! participants and nonparticipants did not differ significantly on many variables.
The cost of recruitment for ¡Viva Bien! was $262.54 per randomized participant, compared to the $714 for the MLP27
(with a 14% inflation adjustment). ¡Viva Bien! recruitment cost less per unit than the MLP because of the efficient use of electronic medical record and administrative databases, and because the MLP required an additional step of recruiting individual physician practices prior to initiating patient recruitment. Reflecting the well-recognized challenges of recruiting Latinas for clinical trials6
, ¡Viva Bien! recruitment was moderately expensive. Most of the expense was due to telephone screening, which should be reported in more studies. KP Colorado is now undertaking a major effort to obtain information on race and ethnicity; had this information been available, it would have greatly reduced recruitment costs.
Given the importance of including ethnically diverse populations in research, recruitment lessons learned from the ¡Viva Bien! study may be informative for research and practice. Similar to the experiences described in Rodríguez et al5
, ¡Viva Bien! had varying levels of difficulty in recruiting Latino women across 4 waves. Especially challenging was wave 2, which required a 63% larger sampling pool than the other waves. After sluggish initial recruitment, strategic changes made in procedures resulted in more favorable participation rates. Especially important was the switch to in-house recruiters, who were more invested, more familiar with the study, more likely to be bilingual, and able to ask project managers to immediately address concerns. ¡Viva Bien! recruiters tried to complete calls without deferring to another time, and to make multiple call attempts instead of leaving messages. Using these measures, ¡Viva Bien! nearly achieved the recruitment goal of 300 participants.
This study has both methodological strengths and limitations. Strengths include a defined population to assess program reach, assessment of sample representativeness on demographic and medical characteristics, identification of reasons for nonparticipation, analysis of patient factors related to participation, and multi-faceted recruitment procedures to limit bias in the sample. An intentional but somewhat specialized feature of the study is that much of the sample (KP) represented a largely employed population in one health maintenance organization. As with all studies conducted within one geographical area or organization, generalizability is reduced. However, the subsample recruited from Salud provided additional diversity.
Health research could benefit from more widespread reporting and evaluation of recruitment efforts8
. This is especially true of studies involving under-represented populations, such as Latinos, who have poorer health outcomes. The ¡Viva Bien! project provides a useful case study of how participation rates may be enhanced by addressing cultural characteristics, and systematically removing obstacles to participation. Future researchers could expand on this work by exploring the relative effectiveness of different contact points (eg, churches, schools, social groups, health fairs, and the Internet) for recruiting minority samples.