Predictors for Using a HIV Self-Test Among Migrant and Seasonal Farmworkers in North Carolina

Background: Approximately, two million migrant and seasonal farmworkers (MSF) work in the United States annually. Several factors, such as lack of access to healthcare services and health behaviors, contribute to risk of HIV transmission. Relatively few studies have explored MSF knowledge of HIV transmission and testing options. Methods: A 12-question, self-administered survey of farmworkers (n = 178) from 19 migrant camps was conducted. The survey assessed knowledge of factors related to HIV transmission, testing, and intention to use a HIV home-test kit. Results: Participants with knowledge of treatment for HIV (p = 0.03) and that condom use protects against HIV (p = 0.04) were more willing to express intent to use a home test kit than those with less knowledge. Concern among farmworkers that HIV was a very or somewhat serious problem in their community was associated with expressing intent to use a home test kit (OR = 2.3, 95% CI = 0.92–5.5). Respondents with less knowledge were less likely to use a home test kit. Conclusions: MSF were concerned about HIV in their community and would be willing to use to a home test kit. This pilot study provides a basis for additional research related to HIV testing within the MSF community.

by a community health center. The center had a prior program offering free HIV testing. Inclusion criteria for this study included Latino ethnicity and ≥18 years of age. There were no women farmworkers encountered, therefore, the study population was comprised of 100% males. All participants provided oral consent prior to participation. Upon completion of the survey, two condoms were distributed to participants.

Data Collection
Eighteen H-2A camps and 1 non-H-2A camp were visited. H-2A classification is one of many temporary, nonimmigrant worker classifications defined by the U.S. government. Workers residing at H-2A camps are defined as temporary, legal agricultural employees contracted with a farmer to provide services to his/her business [22]. A trained, local outreach worker, a male, native Spanish speaker, identified the dates and locations to conduct the survey. After arriving at the location, the outreach worker would informally explain the purpose, incentive and ask MSF about interest in completing the survey, typically in a group setting. Interested workers completed the survey. When literacy was an issue, the outreach worker would read questions and answer options aloud. After the survey was returned, respondents were offered two free condoms. No personal identifiers were collected. Data collection was conducted in the evening hours, typically after the worker had returned from the fields. The East Carolina University IRB approved this study prior to data collection.
The questionnaire was developed in English and then translated by a qualified Spanish speaking native Mexican. The questionnaire included 12 questions related to sexual behavior, concerns about HIV, sexuality and HIV knowledge. The survey was adapted from the 2000 National Survey of Teens on HIV/AIDS, Public Knowledge and Attitude about HIV/AIDS, sponsored by the Kaiser Family Foundation [23]. Our survey questions were adapted from this national survey because the questions were developed for individuals with middle-school to high-school education, similar to the MSF population. In addition, the survey was designed to be short and simple, given that our visit was unannounced to the workers.

Measures
The primary outcome for this study was assessing whether access to a HIV home test kit would increase HIV testing ("yes", "no", "maybe", "don't know"): "If you have access to a home HIV testing kit, would you increase your HIV testing?" This was operationalized as intention to use a home test kit [20]. Each question, except the one on HIV testing status, had an option to characterize the absence of knowledge or behavior, such as "Don't Know" or "Never", as this information was considered important for future education and outreach efforts.
A composite score was developed for five HIV knowledge questions [20]. Each answer selected from "yes", "no", or "don't know" was equal to 1, 2 or 3 points, respectively, with range of 5 to 15 points possible in the HIV knowledge section. HIV knowledge questions were: (1) "Are there drugs to treat HIV?" (2) "Is there a cure for HIV?" (3) "Can you become infected with HIV by having unprotected oral sex?" (4) "Can sex without a condom increase a person's risk of becoming infected with HIV?" (5) "Can needle sharing increase a person's risk of becoming infected with HIV?" Respondents with 1-5 points, 6-10, and 11-15 points, were categorized with sufficient knowledge, some knowledge, and little knowledge, respectively, in terms of HIV education and knowledge.
Likert scale questions included several related to behavior and beliefs. Sexual behavior was measured by "In general, how often do you use a condom?" with response options "all the time", "most of the time", "only sometimes", and "never". A question regarding concern about HIV was "Is HIV infection a serious problem for seasonal farmworkers?" with response options "very serious", "somewhat serious", "not serious" and "don't know." Belief was measured by "How likely are you to contract HIV?" with response options, "very likely", "somewhat likely", "not at all", and "don't know."

Analysis
Descriptive analysis, including two-way classification by primary outcome, were provided with Chi-square statistics. For logistic regression, "no" and "don't know" responses were collapsed into "no" for independent and dependent variables. Odds ratios and 95% confidence intervals (CI) were computed using two logistic regression models. Records with missing data for either response (n = 6) or predictor variables included in the model were dropped from the logistic regression procedure, leaving 149 out of 178 records to generate odds ratios. All variables from the survey were included in a logistic model to determine odds ratios after adjusting for each variable, respectively. Then to determine the best-fitting model predictors were included and backward elimination of variables (p = 0.20 and lower) was used to identify significant predictors. Analysis was conducted using Stata, Version 12 (Stata Corp LP, College Station, TX, USA).

Results
Of 199 surveys distributed at 19 camps (18 H2A and one non-H2A camp) between 11 April 2013 through 12 June 2013, 178 surveys were returned for an overall response rate of 89%. The lowest response rate for an individual question used in this analysis was 93% ("Can you become infected with HIV by having unprotected sex?") and the highest response rate was 98% ("Is HIV infection a serious problem for seasonal workers?"). Table 1 summarizes the survey results. The participants had a high degree of knowledge that condom use protected against HIV and sharing needles increased HIV risk. A majority of MSF (61%) had been tested for HIV in the past. A large majority responded that HIV was a somewhat or a very serious concern in the farmworker community (81%), and used condoms all the time (43%). A majority of farmworkers indicated they would use a home HIV testing kit (59%) and would be more inclined to use the kit if trained on how to use it (45%). Health clinics and providers (72%) were the dominate information source for health education material. Table 2 presents the intention to use home test kit by knowledge and behaviors. The knowledge composite variable (p = 0.03) and history of prior HIV test (p = 0.04) were significantly associated with the intent to use home test kits. However, there was no difference in intention to use a home test kit by condom use, concern for HIV in community, or likelihood to contract HIV. Many respondents, almost 20% on some questions, replied maybe or don't know to increasing HIV testing if provided access to a home test kit, particularly of interest, given the concern for HIV (81 serious or very serious) in the seasonal farmworker community and 40% of workers only sometimes or never using a condom.   Table 3 presents odds ratios for association between predictors and intention to use home test kit (yes/no). While in univariate analysis, respondents considered HIV a very or somewhat serious concern in seasonal farmworkers (OR = 3.2, 95% CI 1.5-7.2), after adjustment for composite knowledge measure, test history, and likelihood to contract HIV, the association with serious concern and intention to use a test kit was no longer significant (OR = 2.3, 95% CI 0.92-5.5). Both HIV test history and belief to contract HIV were elevated in the adjusted model, although not significantly. Beginning with a full model (education composite, seriousness, condom use, testing history, and belief), each predictor was removed leaving "seriousness" (concern that HIV is a serious or somewhat serious concern in seasonal farmworkers) as the only predictor for intention to use a home test kit in the model (OR = 2.7, 95% CI 1.1-6.2) (data not shown in table). When including each component of knowledge (instead of the composite) and other predictors in backward elimination, "seriousness" emerged as the only predictor significantly associated with intention to use a home test kit (OR = 2.5, 95% CI 1.1-5.8).

Discussion
The results of this study indicate that MSF consider HIV a serious concern in their community and would be motivated to use a home test kit to identify HIV status. Workers with sufficient knowledge of HIV risk factors or had been previously tested were more likely to express intent to use a test kit than those who were not as knowledgeable or had not been tested. Many farmworkers had prior HIV testing, which was also a motivator to use a HIV test kit. Those with little knowledge or prior HIV testing were least likely to intend to use a HIV home test kit.
Because HIV home test kits are relatively new, few studies were found investigating HIV testing behaviors in MSF and their intention to use a HIV home test kit. In a South Florida population of Hispanic farmworkers, Fernandez and colleagues surveyed intention to accept a free HIV test. Our findings were similar to Fernandez [20] and colleagues who found that farmworkers having been tested for HIV have intent to accept a free HIV test. In addition, "worrying" some or a lot about getting HIV indicated testing, similar to our measure of likelihood of "contracting" HIV and using a home test kit. While Fernandez did not have a measure of HIV in the community of farmworkers, our measure of community concern (that HIV is a serious concern in the farmworker community) may indicate some individual-level of "worrying". Unprotected sex was not a strong predictor for test use [20]. Our findings may be consistent with Fernandez in that HIV knowledge was a predictor of intent to accept a free HIV test, but our findings did not exclude the chance of no association. To the extent that urban day laborers in California have similar behaviors as rural farmworkers in North Carolina, predictors of intention to test for HIV within the following year included not having had a prior test and perceived risk for HIV infection, similar to our findings [19].
The findings of this exploratory study indicate that HIV is perceived as a serious problem in the farmworker community. Those that have been previously tested and have knowledge about HIV transmission were willing to use a home test kit. However, importantly, those that had the least knowledge and had not been tested did not have the same motivation to use a home test kit. A higher perceived risk for HIV in the community may be a prevention strategy, particularly if testing is offered in a residential setting. Opportunity to provide educational training on how to use a home test kit may engage those already knowledgeable as well as those that are not to avoid stigma associated with infection [24].
Before recommending the HIV self-test in this vulnerable population, healthcare and outreach workers should understand aspects of the validity and predictive value of the screening test. A recent meta-analysis of rapid HIV tests demonstrated that the oral fluid tests may be more appropriate in populations with a high prevalence of HIV [25], because the rapid test has a lower positive predictive value in low prevalence populations, which can result in more false positive screening results [26]. However, several studies showed that participants performed the home test correctly and were able to interpret the home test [27]. Further research is needed on training effectiveness, use of the test, and interpretation of results in a transient farmworker population. Additionally, more information is needed about the prevalence of HIV in seasonal and migrant farmworkers.
There are several limitations to this pilot study. First, women were not included, however, no MSF women were identified at the 19 camps on the days that we visited. The survey, which was not back-translated, consisted of 12 questions that was self-administered after the workday. We were unable to quantify the level of reading literacy for persons taking the survey, and some workers helped others with understanding the meaning of the questions. However, if a person could read, the survey was simple to understand. Participants self-selected themselves and there was an informal data collection process that allowed respondents to complete the survey in privacy but also opportunity for limited readers of Spanish to ask the outreach worker about the meaning of a question in a comfortable setting. While the outreach worker may have provided HIV testing and counseling to visiting MSF in the past, it is unlikely that this group of respondents received prior testing and counseling from the same outreach worker. Respondents were not routinely conversing with each other about the questions or their answers. For this reason, it is unlikely that results are due to information bias. Missing data may be problematic, but response rates for individual questions ranged from 93% to 98%, suggesting that missing data were distributed across individual questions. Residual confounding may be present; data were not collected on some key demographic and health behavior questions, such as history of STI, marital status, and number of sex partners. Cultural norms and values, frequency of cross-border migration, and contextual situations in the U.S. and their home countries could not be assessed, though would presumably impact a worker's intention to choose a self-test. The strengths include a relatively large sample size, validated questions, and a high response rate. Further research is needed to replicate these findings in other settings and better characterize perceptions about HIV and validity of self-testing as a HIV screening approach in MSF.

Conclusions
Rates of new HIV infection in Latino men continues to increase in the U.S. While limited current data exist on HIV incidence rates or prevalence in this vulnerable occupational group, MSF, the annual movement of thousands of documented and undocumented male (and female) workers between the U.S., Mexico, and other countries engaged in agricultural production sets the context for HIV spread into the U.S., Mexico and beyond. Migrant and seasonal farmworkers residing in a rural area of eastern North Carolina expressed concern about HIV in their community, which deserves further investigation. This pilot study identified the potential acceptability to use HIV home test kits to increase HIV screening in MSF. Increasing awareness of HIV risk in the MSF by promoting self-testing may be an effective prevention strategy for this population Culturally-acceptable community-based initiatives may be warranted to include self-testing for HIV that could result in better access to testing for other STIs, which would improve the health status of farmworkers and their partners wherever they live.