The Florida Cohort Study: Methodology, initial findings and lessons learned from a multisite cohort of people living with HIV in Florida (2024)

Abstract

In 2013, Florida had the highest rate of new HIV infections and only 56% of persons living with HIV (PLWH) were virally suppressed. In response, we initiated a new HIV cohort in Florida to better understand issues affecting HIV health outcomes including viral suppression. The objectives of this paper are to describe the study procedures for the Florida Cohort; summarize information regarding enrollment, follow-up, and findings to date; and discuss challenges and lessons learned regarding the establishment of a new, multisite cohort of PLWH. Florida Cohort participants were enrolled from clinics and community-based organizations in eight geographically diverse counties across Florida. Data were obtained from participant questionnaires, medical records, and state surveillance data that are updated annually. From 2014 to 2018, 932 PLWH (mean age 46.7±11.2 years, 64% male, 55% black, 20% Latinx) were enrolled; 61% (n=569) completed follow-up questionnaires >=6 months after enrollment. At baseline, 83% were retained in care and 75% had HIV viral suppression. Research findings to date have focused on outcomes such as HIV viral suppression, antiretroviral therapy adherence, HIV-related comorbidities, alcohol and drug use, and interest in mHealth interventions. Strengths of the study include the diversity of the sample and the linkage of participant surveys with existing surveillance data. However, the research data are affected by several challenges and decisions made during study planning and follow-up, including issues related to sampling strategy and recruitment location. The lessons learned from this study can be helpful when initiating a new longitudinal cohort study.

In 2013, Florida ranked first in new HIV infections in the United States. Currently, Florida ranks second nationally in new HIV diagnoses (22.1 cases in 100,000 persons) and third for persons currently living with HIV or AIDS (612.3 cases per 100,000 persons) (CDC, 2019; CDC, 2017). The Miami-Fort Lauderdale-Palm Beach metropolitan statistical area (MSA) continues to rank number one in the country for new HIV diagnoses (34.4 per 100,000 persons) (CDC, 2019). At end-of-year 2017, Florida accounted for approximately 11.0% of all adolescents and adults living with HIV in the U.S. (CDC, 2019). Compared to the general U.S. population of people living with HIV (PLWH), Florida has a greater proportion of women, African Americans, and persons older than 50 represented in the incident HIV population (CDC, 2019; FDOH, 2018).

Florida faces significant gaps in the HIV care continuum for those who are diagnosed. The HIV care continuum includes HIV testing, linkage to care, retention in care, and ultimately HIV viral load suppression (CDC, 2011; Colasanti et al., 2015; Gardner et al., 2011). Although definitions for the steps in the HIV care continuum continue to evolve, linkage to care refers to an individual visiting a medical provider within a certain time period (e.g., 1 - 6 months) after diagnosis. An individual is considered retained in care if they have had 2 or more medical visits at least 90 days apart in the last year. Viral suppression is established when the amount of the virus in the blood is less than the detectable amount of the diagnostic test assay (200 copies/mL at the time the study was initiated) (FDOH, 2018). In Florida, at the time the study was initiated, 64.4% were retained in care, and 56.3% virally suppressed (FDOH, 2017). Given the sheer number of HIV cases, the diversity of PLWH, and the gaps in HIV care, further HIV studies focusing on PLWH in Florida are warranted.

Prior to this study’s initiation, there had been no large cohort studies focusing on PLWH in Florida. Cohort studies allow researchers to understand what happens to persons over time, and to identify strategies to improve health outcomes. The Florida Cohort research team, comprised of investigators and clinicians from several Florida universities and the Florida Department of Health (FDOH), sought to create a cohort that would span the entire state of Florida and reflect the unique demographic characteristics of PLWH. The Florida Cohort recruited primarily from public health settings, in order to engage a population which typically has less access to healthcare. Employing a socio-ecological conceptual model, the research team hypothesized that variables affecting the HIV care continuum would reflect the individual (e.g., attitudes, substance use, mental health), society (e.g., social support, stigma), and healthcare system (e.g. insurance, access to care, types of provider). Alcohol consumption was a focus of the cohort, because it is common in PLWH and correlates with HIV care (Vagenas et al., 2015).

Establishing the Florida Cohort offered the promise of a vital rich database which could help inform policies or interventions to improve HIV viral suppression and reduce new HIV infections. Researchers also sought to create a participant registry to facilitate recruitment into future research studies and to create a de-identified database for future analyses. In the present study, the aims are to (1) describe the methodology utilized in the development of the Florida Cohort, (2) present baseline characteristics and summary of findings to date, and (3) discuss the challenges and lessons learned in initiating a diverse cohort of PLWH at multiple sites.

METHODS

RESULTS

The Florida Cohort study enrolled 932 PLWH. Of these, we obtained medical record abstraction for 90%, linked 98% to eHARS data, and 93% agreed to participate in a future study contact registry. Of those enrolled, 569 (61.1%) completed a follow-up survey 6-18 months after enrollment.

In table 2, we list the characteristics of our sample with the characteristics of Florida overall. The mean age of study participants was 46.7±11.2 years, 64% were male, and the racial/ethnic distribution was 55% black, 20% Latinx, and 21% white. While we did not conduct formal statistical comparisons of demographic characteristics between our sample and among PLWH in Florida in 2015 overall, in general a higher proportion of our sample reported retention in care (83% vs. 64%), being on ART (90% vs. 67%), and viral suppression (75% vs. 58%).

Table 3 describes additional background variables, mental health, and, alcohol/substance use. Approximately one-third (34%) had less than a high-school education, most reported currently drinking (71%) and almost a third reporting hazardous drinking (35%); however, only 9% reported heavy drinking. Over a third of participants reported marijuana use in the past 3 months (37%). In the past 12 months, 14% reported smoking crack/cocaine, 11% reported recreational use of pain medication, and 11% reported recreational use of sedatives.

As of early 2020, eleven peer-reviewed papers have been published using Florida Cohort data, at least 20 manuscripts currently in development. To date, the papers have focused on various outcomes among PLWH, including chronic disease (e.g. hypertension, cancer screening) (Xu et al., 2019; Wijayabahu et al. 2019), substance use (e.g. alcohol-related problems, drug use) (Mannes et al., 2019; Mannes et al., 2018; Sharpe et al., 2018), mobile app use and potential interest in mHealth interventions (Sharpe et al., 2018; Lucero et al., 2017), healthcare utilization (e.g. emergency room visits, hospitalizations) (Mannes et al., 2019a), HIV-related stigma (Algarin et al., 2019), PrEP awareness (Algarin et al., 2020), HIV medication adherence (Dunne et al., 2019; Mannes et al., 2018a), and HIV viral suppression (Cook et al., 2017). The findings have demonstrated clustering of risk factors and poor health outcomes, identified subgroups of the population who may be most vulnerable, and showed interest among PLWH in future mHealth intervention strategies. For example, one study found that heavy drinking was a risk factor for suboptimal anti-retroviral adherence and poor viral suppression (Cook et al., 2017). Another study found that those with moderate to severe generalized anxiety disorder had greater odds of missing their HIV care appointments and visiting an emergency room, making them vulnerable to poorer health outcomes (Mannes et al., 2019a). Additionally, our findings suggest that mental health, impulsivity, and recreational marijuana use are related to sub-optimal HIV care. Our data also shows that stigma is a nuanced construct among people living with HIV, and that the intersectionality of stigma with variables such as age, race, and ethnicity are important to consider in accessing HIV care. The Florida Cohort registry has also been successful in helping to recruit participants for future research studies.

DISCUSSION

We were able to successfully initiate The Florida Cohort study, which is, to our knowledge, the first HIV research cohort that specifically sought to span and focus upon an entire state. The Florida Cohort Study had several strengths. One of the main strengths of the study was the use of a multi-recruitment site approach that spanned an entire state, which included four of the counties with the highest rate of HIV incidence (FDOH, 2018). Recruitment from multiple locations offers a more generalizable sample of the state and could offer important insights on disparities by location such as urban and rural areas. By recruiting from public health department clinics, federally qualified health centers, and a food bank, we attempted to capture people who may have increased challenges to be engaged in HIV care. Another unique strength of the Florida Cohort is our ongoing collaboration with the FDOH and the ability to link participant data to HIV surveillance data such as eHARS. Linking our data with the Florida’s eHARS system allowed for the study to gain objective clinical health outcomes and to obtain follow up surveillance data on participants even after primary data collection is complete. The specific process to merge our data and the eHARS data securely from FDOH was worked out during several face-to-face discussions, required approvals by multiple IRBs, and involves a Data Sharing Agreement for this specific study to help ensure the protection of participant identity. Linkage to surveillance data will allow the research team to continue to monitor HIV viral load, CD4 counts, engagement in care, and survival for at least 5 years after enrollment. We also successfully linked questionnaire data to medical records for most participants; however, this required much additional labor to abstract the information, and medical records were harder to obtain in some settings than others.

Despite its strengths, the design and implementation of the Florida Cohort study presented several challenges and lessons learned, including issues related to sampling, data collection and management, and fiscal and regulatory processes. We sought a sample that was reasonably representative of the state of Florida, yet the available resources would require us to enroll from a limited number of locations. We sought to include participants from both urban and rural areas, but recruitment would be much more challenging in rural settings (e.g. Wildwood, Lake City) because the HIV clinic services were limited. In order to ensure sufficient numbers of persons from rural areas were included, we occasionally limited the pace of recruitment in larger clinics but not in the rural settings.

We considered the pros and cons of a cohort recruited from clinics and an interval cohort recruited from the general community (Lau et al., 2007). Recruitment from clinics would provide convenience and improved access to medical records but would include fewer individuals considered to be out of care, and HIV care engagement was of interest. Most of the clinical care for PLWH in Florida is provided throughout a network of HIV clinics based in the FDOH’s county health departments. Therefore, we chose to recruit most participants from county health department clinics but also included one recruitment setting that was not associated with a clinic (e.g. a community food bank for PLWH), and a community health center. It was more difficult to obtain medical record information from persons recruited outside of clinics.

During the design of the study questionnaire, the research team was also challenged to choose between measuring a wide range of domains related to the socio-ecological model, vs. the ability to have detailed measures of specific topics. We elected to use briefer, screening measures for most domains, rather than more extensive measures. The follow-up questionnaire maintained several “core” measures, but we also replaced some of the baseline measures with new measures.

Our goal was to try to enroll a sample that had a similar age, gender, and racial/ethnic distribution as the state of Florida overall. The surveys and questionnaires were translated to Spanish. Overall, the baseline demographic characteristics of the Florida Cohort appear to be similar to those of all PLWH in Florida in the same time period (FDOH, 2018). However, statistical testing to compare proportions between the Cohort data and the state-wide data was not possible. This is a limitation of the study. The Cohort did not enroll persons aged younger than 18 and included a lower proportion of older persons (aged >60 years). Yet, 583 (62%) participants reported that they had been in jail or prison at least once in their lifetime. We need to better understand how HIV care engagement and outcomes are affected as persons go into and out of the criminal justice system. Also, some groups of persons relevant to the HIV epidemic in Florida are underrepresented, including people who speak Haitian Creole, people under age 18, and people who are transgender. We intend to address these limitations in the next wave of the Florida Cohort study starting in 2020.

The follow-up rate for questionnaires was somewhat lower than desired (61%), with better follow-up rates at those clinics with the most consistent staffing over time. Although alcohol was to be a major focus of the cohort, only about 9% of participants met criteria for “heavy drinking” (>14 drinks per week for men and >7 drinks per week for women) (Cook et al., 2017), which was lower than expected based on prevalence of heavy drinking in other HIV cohorts (Conigliaro et al., 2006). The relatively low proportion of heavy drinkers limits our ability to understand potential mechanisms by which heavy drinking affects HIV-related health outcomes. It could be that heavy drinking among PLWH truly is lower overall, compared to the previous decades or other study populations, or it could be that the heaviest drinkers tended to avoid this research study or under-reported their true level of drinking. In the future, we plan to actively enroll a higher proportion of heavy drinkers, use more extensive alcohol assessments, include biomarkers of alcohol consumption, and/or use transdermal alcohol biosensors (Wang et al., 2018).

In summary, the Florida Cohort is a new HIV cohort that enrolled 932 PLWH from 2014 to 2018, with ongoing follow-up via HIV surveillance data. Data from the cohort has resulted in numerous publications and presentations that have helped to identify factors associated with the HIV care continuum in Florida. Our experience has provided lessons learned regarding the relative advantages and disadvantages of different recruitment settings and the importance of consistent staffing during follow-up. In 2018, additional research funding was obtained from NIAAA and NIDA to continue to follow a subset of the original cohort, while also including new enrollment that will include at least 200 PLWH with heavy drinking, over 300 PWLH who currently use marijuana, and up to 150 persons who are transgender. Collectively, the data from the Florida Cohort will continue to be used for research that seeks to improve the HIV care continuum and reduce HIV-related comorbidities.

Acknowledgements

REFERENCES