Abstract
Background:
Many clinical and population-based research studies pivoted from in-person assessments to phone-based surveys due to the COVID-19 pandemic. The impact of these transitions on survey response remains understudied, especially for people living with HIV. Given that there are gender-specific trends in alcohol and substance use, it is particularly important to capture these data for women.
Objective:
Identify factors associated with responding to an alcohol and substance use phone survey administered during the COVID-19 pandemic in the Women’s Interagency HIV Study, a multicenter US prospective cohort of women living with and without HIV.
Methods:
We used multivariable logistic regression to assess for associations of pre-pandemic (April–September 2019) sociodemographic factors, HIV status, housing status, depressive symptoms, alcohol use, and substance use with response to an early-pandemic (August–September 2020) phone survey.
Results:
Of 1,847 women who attended an in-person visit in 2019, 78% responded to a phone survey during the pandemic. The odds of responding were lower for women of Hispanic ethnicity (aOR 0.47 95% CI 0.33–0.66, ref=Black/African American) and those who reported substance use (aOR 0.63 95% CI 0.41–0.98). By contrast, the odds were higher for White women (aOR 1.64 95% CI 1.02–2.70, ref=Black/African American) and those with stable housing (aOR 1.74 95% CI 1.24–2.43).
Conclusions:
Pivoting from an in-person to phone-administered alcohol and substance use survey may lead to underrepresentation of key subpopulations of women who are often neglected in substance use and HIV research. As remote survey methods become more common, investigators need to ensure that the study population is representative of the target population.
Keywords: COVID-19 pandemic, HIV, women, alcohol consumption, substance use, survey methods
Introduction
The COVID-19 pandemic led to unprecedented changes in the way clinical care and research are conducted in the United States (US). Early in the pandemic, many clinical research and population studies pivoted from in-person survey administration to phone-based survey methods (1). The shift in survey methods offers an opportunity to examine the effect of rapidly adapting research study methods in the context of a widespread crisis as well as specific patterns in participant study engagement based on survey method.
Phone survey methods are convenient for collecting real-time data but have higher nonresponse rates than in-person visits, introducing a nonresponse selection bias if certain subgroups who are more or less likely to have the outcome do not respond (1,2). Survey nonresponse for alcohol and substance use has been described in studies prior to the COVID-19 pandemic and has variably resulted in biased prevalence estimates of alcohol and substance use related outcomes (3-10). Furthermore, self-reported alcohol consumption may vary by survey method. The National Alcohol Survey of 1990 found similar response rates between in-person and phone surveys. However, there was an underrepresentation of low-income participants and lower reported alcohol consumption quantities with phone surveys compared to in-person surveys (10). The 1985 National Household Survey on Drug Abuse also found lower reports of substance use by phone surveys compared to in-person surveys, especially among racial minorities (3). Since then, similar results have been replicated in other large epidemiology studies such as the National Survey on Drug Use and Health and the Canadian Addiction Survey (7,9).
Phone communication norms have shifted over the years with the decreasing use of landlines, and widespread use of cell phones, caller identification and text messages, creating an ongoing need to reevaluate phone survey modalities. Furthermore, widespread shifts in participant priorities due to economic insecurity, role strain, and health considerations early in the COVID-19 pandemic made data collection during this time period unique compared to pre-pandemic surveys. At the same time, substance use-related deaths and alcohol consumption have been increasing in the US, underscoring the importance of accurately estimating alcohol and substance use among at-risk subpopulations (11-14). This is particularly relevant for women, who may have been disproportionately impacted by the COVID-19 pandemic and had greater increases in alcohol consumption compared to men (12,13,15-18). However, the impact of changes from in-person assessments to phone survey methods on the response rate to alcohol and substance use questionnaires among women during the COVID-19 pandemic remains unexplored.
Leveraging data collected from the Women’s Interagency HIV Study (WIHS), a multicenter US prospective cohort of adult women living with and without HIV, we aimed to 1) determine the response rate to an alcohol and substance use phone survey administered early in the COVID-19 pandemic among women recently engaged in in-person assessments; 2) identify sociodemographic and clinical factors associated with response to an alcohol and substance use phone survey; and 3) assess the impact of phone survey nonresponse on COVID-19 pandemic prevalence estimates of alcohol and substance use.
Materials and methods
Data source
Established in 1994, WIHS is the longest prospective cohort study of women living with and without HIV enrolled from 10 US sites (19). In 2019, WIHS merged with the MACS (Multicenter AIDS Cohort Study), a long-standing cohort study of men living with and without HIV, to become the MACS/WIHS Combined Cohort Study (MWCCS) (20). At the time of the merger, 86% of participants from both cohorts had completed a visit within the previous year. There were 2,115 WIHS participants and the majority of participants had been enrolled for at least 5 years, as the last enrollment wave for WIHS was from 2011 to 2015. The WIHS is representative of women living with HIV in the US and also includes demographically similar women living without HIV for comparison. Recruitment, retention, and other study procedures are described in detail else-where (19-22). Briefly, women were recruited from a wide variety of venues including community-based organizations, support groups, substance use treatment programs, medical settings such as HIV and STI clinics, and other research studies. This was done primarily through flyers and word-of-mouth. Visits and phone surveys were conducted in English and Spanish, and participants were compensated. Study staff made multiple attempts to contact a participant. Contact methods (e.g., phone, text, e-mail) were site dependent and based upon preexisting modes of reaching participants for study visits. Data collection forms are available publicly (https://statepi.jhsph.edu/mwccs/). Clinical assessments and self-report surveys on medical, psychiatric, and social measures in WIHS were collected at in-person semiannual visits through February 2020. From April through September 2020, participants were invited to complete surveys by phone regarding COVID-19 symptoms, testing, and hospitalization, as well as mental health, alcohol, and substance use.
Ethics statement
The WIHS and now MWCCS are approved by the Institutional Review Board at each study site. Written consent was obtained from all participants. Participants were also verbally consented before proceeding with the phone survey.
Study sample
Women living with and without HIV who had a pre-pandemic visit from April 2019 through September 2019 and who remained enrolled in the study during 2020 were included. This pre-pandemic visit was considered the index visit for the current analysis. If index data were missing, data from the preceding visit (October 2018 to March 2019) were carried forward and only observations that ultimately had complete index data were retained in the analysis.
Alcohol and substance use measures
A survey on alcohol and substance use was conducted in-person at the index pre-pandemic visit and by phone during the pandemic from August to September 2020 (Table 1). In both instances, participants were asked questions adapted from the Alcohol Use Disorders Identification Test – Consumption (AUDIT-C), and responses were used to calculate the average number of drinks per week either since their last visit (during the pre-pandemic index visit) or since pandemic start (in the 2020 phone survey), respectively (23). Responses were used to assess risky drinking (>7 drinks/week or >3 drinks/day). Participants were also asked a single question about substance use since their last visit or the pandemic start for each substance (i.e., heroin, crack, cocaine, methamphetamine, sedatives, other non-prescribed drugs). Responses were dichotomized into any substance use versus none. Pandemic phone survey response was the primary study outcome. Participants were considered “responders” or “nonresponders” based on whether they had available alcohol and substance use data for this survey.
Table 1.
COVID-19 pandemic data collection timepoints in the Women’s Interagency HIV Study.
| Measures | Pre-pandemic Index Visit (April-Sept 2019*) |
Early Pandemic Timepoint (Aug-Sept 2020) |
|---|---|---|
| Study visit type | Semiannual in-person visit | Telephone survey |
| Sociodemographic | x | x |
| Clinical characteristics | x | x |
| Alcohol use | x | x |
| Substance use | x | x |
Data were obtained from a study visit that occurred between 10/2018–03/2019 when data for 04/2019–09/2019 visit were missing.
Sociodemographic measures
Age, study site region, race/ethnicity, education, employment, health insurance, annual household income, housing status, HIV serostatus, depressive symptoms (20-item Center for Epidemiologic Studies Depression Scale [CES-D], range 0–60, score >16 indicative of depression) (24,25), tobacco use, and cannabis use were captured from the index visit. These variables were selected based on previously published and hypothesized associations with phone survey nonresponse and retention in clinical research studies among women living with HIV (5,7,9,21,22,26-29). Study site region (categorized as Midwest, Southeast, Northeast, West) was included due to potential differences in demographic and clinical characteristics as well as the varying impact of the COVID-19 pandemic and related policies in different regions across the US. Homelessness and unstable housing was broadly defined as staying outdoors, in a transitional setting, or staying with someone else.
Statistical analysis
Descriptive statistics were used to assess index demographics, clinical characteristics, and alcohol and substance use measures. Univariable and multivariable logistic regression models were used to identify factors associated with phone survey response. Inverse response probability weights (IRPW) generated from the multivariable logistic regression model were used to obtain nonresponse-adjusted estimates of risky drinking and substance use prevalence in the COVID-19 pandemic. These were compared to the estimates from the unweighted sample to assess for the presence of selection bias due to participant nonresponse. For example, if the weighted estimates were higher than the unweighted estimates it would indicate that the sample was biased due to lower response rates among those with greater alcohol or substance use. There were no extreme weights on visual assessment, so all observations were retained. RStudio version 1.4.1717 was used for the analysis (30).
Results
There were 1,968 women with an index pre-pandemic visit. Missing data included educational attainment (2.8%), employment (2.9%), health insurance (3.2%), annual household income (6.8%), housing status (2.8%), drinks per week (3.1%), tobacco use (3.1%), cannabis use (3.1%), substance use (3.1%), depression (3.4%). These data were imputed by carrying forward data from the visit prior. After excluding participants who still did not have complete index visit data, or who were no longer followed in the study, 1,847 women qualified for this analysis, of whom 1,433 (77.6%) responded to the pandemic alcohol and substance use phone survey (Figure 1).
Figure 1.
Participant Flow Diagram.
The majority of participants were Black/African American (61.9%), had stable housing (88.2%), and were living with HIV (71.2%); nearly, half (46.0%) had an annual income <$12K (Table 2). The median age was 53 (IQR 46–59) years. Tobacco and cannabis use were relatively common (39.5% and 23.5% respectively), while risky drinking and substance use were less common (12.7% and 7.4% respectively).
Table 2.
Index visit sociodemographic and clinical factors for participants who did and did not respond to the alcohol and substance use phone survey early in the COVID-19 pandemic (August–September 2020).
| Total n = 1847 |
Responders n = 1433 (77.6%) |
Nonresponders n = 414 (22.4%) |
|
|---|---|---|---|
| N (%) | |||
| Age, years, median (IQR) | 53 (46–59) | 53 (47–60) | 52 (44–57) |
| Region | |||
| Northeast | 772 (41.8) | 678 (47.3) | 94 (22.7) |
| South | 612 (33.1) | 398 (27.8) | 214 (51.7) |
| Midwest | 223 (12.1) | 192 (13.4) | 31 (7.5) |
| West | 240 (13.0) | 165 (11.5) | 75 (18.1) |
| Race and Ethnicity | |||
| Black/African American | 1144 (61.9) | 884 (61.7) | 260 (62.8) |
| Hispanic | 263 (14.2) | 191 (13.3) | 72 (17.4) |
| White | 151 (8.2) | 126 (8.8) | 25 (6.0) |
| Multiracial | 257 (13.9) | 212 (14.8) | 45 (10.9) |
| Other/Unknown | 32 (1.7) | 20 (1.4) | 12 (2.9) |
| Educational attainment | |||
| High school incomplete | 606 (32.8) | 451 (31.5) | 155 (37.4) |
| High school complete | 565 (30.6) | 446 (31.1) | 119 (28.7) |
| Some college or above | 676 (36.6) | 536 (37.4) | 140 (33.8) |
| Employed | 716 (38.8) | 577 (40.3) | 139 (33.6) |
| Health Insurance* | 1777 (96.2) | 1392 (97.1) | 385 (93.0) |
| Annual household income | |||
| <$12,000/year | 850 (46.0) | 633 (44.2) | 217 (52.4) |
| $12,000–30,000/year | 560 (30.3) | 441 (30.8) | 119 (28.7) |
| >$30,000/year | 437 (23.7) | 359 (25.1) | 78 (18.8) |
| Stable Housing | 1629 (88.2) | 1296 (90.4) | 333 (80.4) |
| HIV Positive | 1315 (71.2) | 1014 (70.8) | 301 (72.7) |
| Number of drinks per week, mean (sd) | 2.6 (7.5) | 2.4 (7.2) | 3.1 (8.7) |
| Risky drinking | 234 (12.7) | 175 (12.2) | 59 (14.3) |
| Tobacco use | 730 (39.5) | 535 (37.3) | 195 (47.1) |
| Cannabis use | 434 (23.5) | 326 (22.7) | 108 (26.1) |
| Substance use** | 137 (7.4) | 85 (5.9) | 52 (12.6) |
| Depression (CESD ≥16***) | 545 (29.5) | 398 (27.8) | 147 (35.5) |
Includes the Ryan White HIV/AIDS Program.
Includes heroin, crack, cocaine, methamphetamine, sedatives, other non-prescribed drugs.
Center for Epidemiologic Studies Depression Scale (CESD), range 0–60, score > 16 suggestive of depression.
In univariable analysis, nearly all sociodemographic and clinical measures – except racial and ethnic groups, HIV serostatus, risky drinking, and cannabis use – had statistically significant associations with response to the pandemic alcohol and substance use phone survey (Table 3). In the adjusted model, the odds of responding were lower among women residing in the Western (aOR 0.35 95% CI: 0.21–0.57) and Southern (aOR 0.29 95% CI: 19–0.44) regions compared with Midwestern US regions; among women of Hispanic ethnicity (aOR 0.47 95% CI: 0.33–0.66, ref=Black/African American); and among those who reported pre-pandemic substance use (aOR 0.63 95% CI: 0.41–0.98) (Table 3). By contrast, the odds were higher for White women (aOR 1.64 95% CI: 1.02–2.70, ref=Black/African American) and those with stable housing (aOR 1.74 95% CI: 1.24–2.43). Unweighted versus IRPW prevalence estimates were 11.03% vs. 11.55% (standard error 0.89%) for risky drinking and 6.07% vs. 6.86% (standard error 0.73%) for substance use (Table 4). While the difference between the weighted and unweighted estimates of risky drinking represents a 4.77% increase, the unweighted estimate falls within the standard error of the weighted estimate. The difference between the weighted and unweighted estimates of substance use represents a 13.00% increase, with the unweighted estimate falling just outside of the standard error for the weighted estimate.
Table 3.
Logistic regression for response (vs. no response) to the alcohol and substance use phone survey early in the COVID-19 pandemic (August–September 2020).
| Unadjusted OR (95% CI) | p-value | Adjusted OR (95% CI) | p-value | |
|---|---|---|---|---|
| Age, years | 1.02 (1.01–1.03) | <.001 | 1.01 (1.00–1.03) | .066 |
| Region | ||||
| Midwest | REF | REF | ||
| Northeast | 1.17 (0.74–1.78) | .494 | 1.20 (0.75–1.86) | .440 |
| South | 0.30 (0.20–0.45) | <.001 | 0.29 (0.19–0.44) | <.001 |
| West | 0.36 (0.22–0.56) | <.001 | 0.35 (0.21–0.57) | <.001 |
| Race and Ethnicity | ||||
| Black/African American | REF | REF | ||
| Hispanic | 0.78 (0.58–1.06) | .110 | 0.47 (0.33–0.66) | <.001 |
| White | 1.48 (0.96–2.38) | .087 | 1.64 (1.02–2.70) | .046 |
| Multiracial | 1.39 (0.99–1.99) | .068 | 1.33 (0.92–1.94) | .138 |
| Other/Unknown | 0.49 (0.24–1.05) | .055 | 0.41 (0.18–0.93) | .028 |
| Educational Attainment | ||||
| High school incomplete | REF | REF | ||
| High school complete | 1.29 (0.98–1.69) | .069 | 1.23 (0.92–1.66) | .164 |
| Some college or above | 1.32 (1.01–1.71) | .039 | 1.17 (0.86–1.58) | .321 |
| Employed | 1.33 (1.06–1.68) | .014 | 1.25 (0.93–1.67) | .140 |
| Health Insurance* | 2.56 (1.56–4.15) | <.001 | 1.66 (0.96–2.87) | .069 |
| Annual household income | ||||
| <$12,000/year | REF | REF | ||
| $12,000–30,000/year | 1.27 (0.99–1.64) | .065 | 1.13 (0.85–1.50) | .416 |
| >$30,000/year | 1.58 (1.19–2.12) | .002 | 0.97 (0.67–1.39) | .852 |
| Stable Housing | 2.30 (1.70–3.10) | <.001 | 1.74 (1.24–2.43) | .001 |
| HIV Positive | 0.91 (0.71–1.16) | .442 | 0.79 (0.59–1.04) | .095 |
| Number of drinks per week | 0.99 (0.98–1.00) | .124 | 1.01 (0.99–1.02) | .309 |
| Tobacco use | 0.67 (0.54–0.84) | <.001 | 0.84 (0.65–1.09) | .186 |
| Cannabis use | 0.83 (0.65–1.08) | .159 | 1.09 (0.82–1.46) | .577 |
| Substance use** | 0.44 (0.31–0.64) | <.001 | 0.63 (0.41–0.98) | .037 |
| Depressive symptoms (CESD ≥16)*** | 0.70 (0.55–0.88) | .002 | 0 90 (0.70–1.17) | .425 |
Statistically significant values are bolded.
Includes the Ryan White HIV/AIDS Program.
Includes heroin, crack, cocaine, methamphetamine, sedatives, other non-prescribed drugs.
Center for Epidemiologic Studies Depression Scale (CESD), range 0–60, score > 16 suggestive of depression.
Table 4.
Weighted and unweighted prevalence of risky drinking and substance use early in the COVID-19 pandemic.
| Unweighted % |
Weighted % (SE %) |
Absolute Difference % |
Relative Difference % |
|
|---|---|---|---|---|
| Risky Drinking | 11.03 | 11.55 (0.89) | 0.53 | 4.77 |
| Substance Use | 6.07 | 6.86 (0.73) | 0.79 | 13.00 |
Discussion
The transition from in-person to remote survey methods in population and clinical research studies during the COVID-19 pandemic is an important opportunity for understanding the impact of rapid adaptations of remote research methods and to inform future best practices for study retention. Examining this question in a longitudinal cohort of socioeconomically disadvantaged US women living with and without HIV is critically important, because these women are underrepresented in research studies and may be more difficult to retain using traditional methods (32,33).
When we examined women from our cohort who were previously engaged in in-person assessments, over one-fifth did not respond to a phone-based alcohol and substance use survey. After adjusting for sociodemographic factors, HIV serostatus, and evidence of depression, we found that Black/African American and Hispanic women, those who were unstably housed or homeless, and women who reported pre-pandemic substance use had lower odds of responding to the phone survey. These findings raise concern that the transition to phone survey methods led to the underrepresentation of subgroups of women. Furthermore, these same subgroups were disproportionally burdened by COVID-19 illness and early pandemic policies, making data collection for these participants all the more important (34-36). Despite differences in survey response by sociodemographic and clinical subgroups, we found only small increases in prevalence estimates of risky drinking and substance use early in the COVID-19 pandemic after applying survey IRPW to the sample.
Our findings of differential survey response by sociodemographic factors and substance use are consistent with previous studies of survey nonresponse and study retention in both alcohol and substance use research as well as more broadly in cohort studies of men and women (3,5,21,22,27,28,37). A WIHS analysis among participants observed from 1994 through 2006 found that in-person study visit nonattendance was associated with temporary housing, substance use, and study site (22). Our findings in the context of these earlier WIHS data raise additional concern that a switch to phone survey methods could exacerbate low engagement rates in vulnerable subgroups given that we only included participants who were already engaged in the cohort. General population surveys including the U.S. Census Bureau Current Population Survey Annual Social and Economic Supplement (29,38) and National Health Interview Survey (26) also found similar results to our study when they switched to phone survey methods in the pandemic.
Underlying factors that mediate the lower response rates in these subgroups remain understudied. Members of racial and ethnic minority groups are more likely to be essential workers, experienced disproportionate job loss in the pandemic, had higher COVID-19 related hospitalization rates, and often have a higher number of medical comorbidities, making the navigation of early COVID-19 pandemic changes even more challenging (36). These factors may have contributed to lower response rates among these subgroups. While some of these circumstances were unique to the COVID-19 pandemic, persistent factors that underpin these disparities such as multilevel racism and the social marginalization of people who use drugs and people experiencing homelessness could continue to influence survey response moving forward. In the MACS/WIHS, in-person visits were resumed later in 2020, so it is unknown if lower response rates among these subgroups have persisted beyond the early phase of the COVID-19 pandemic with the use of phone surveys.
The use of phone surveys in research will continue to grow, underscoring the need for best practices and additional or alternative retention methods for phone surveys that are tailored toward specific participant populations, especially those who have been underrepresented in research historically and often face worse social and health outcomes (33). Current best practices for retention in longitudinal cohort studies focus on in-person study assessments (39). Future research should expand on the role of phone retention interventions such as calling at different times of day, providing participants with cell phones, and ensuring participants save study phone numbers at enrollment so that an unrecognized number is not ignored (2,40-42). Furthermore, having contingencies for contacting participants is important for future periods of social disruption and crisis.
Tradeoffs in the use of phone surveys in clinical research should be considered in the context of study populations and participant preferences. In addition to avoiding disease transmission, phone surveys can facilitate real-time data collection and reach of subgroups who have difficulty attending in-person study visits, such as people in rural locations, those with full-time work schedules, or those without access to transportation or childcare (43). However, these methods may exclude participants who do not have access to a personal phone or a private space to participate in a phone call, which is especially important in the context of discussing sensitive matters such as substance use (44). In a survey of participants enrolled in clinical research studies at one medical center early in the COVID-19 pandemic, about one-fifth to two-fifths preferred participating in study visits over the phone or internet, while others were neutral or preferred in-person visits, highlighting diverse participant preferences (31). In a discrete choice experiment of preferences for engaging in medical care early in the COVID-19 pandemic among PLWH who were experiencing homelessness/unstable housing, telehealth was not preferred to in-person visits even if patient navigation was available to help with technology barriers (45). Given that phone survey methods will continue to become more common, investigators should consider these tradeoffs, account for participant preferences, and try to mitigate the risk of selection bias.
After applying IRPW, we found small increases in the prevalence estimates of risky drinking and substance use that may be partially attributable to statistical imprecision and are therefore unlikely to have public health implications. However, true increases in alcohol and substance use among women living with HIV and their demographically similar peers could translate to important health and social consequences given that this population is already at greater risk for poor health outcomes. In the National Epidemiologic Survey on Alcohol and Related Conditions, where there were also no meaningful differences in prevalence estimates of alcohol use after adjusting for survey nonresponse (27). These findings contrast with previous studies such as the Canadian Addiction Survey, where there was a doubling of prevalence estimates for substance use after adjusting for nonresponse (9) and the Mental Health Surveillance Study, where there were differences in 53% of alcohol and substance use measures after applying response weights (7). The variability in these findings highlights the importance of testing missing data assumptions and assessing for bias in each data source.
Limitations
This study focuses on adult women in the US who are socioeconomically disadvantaged and have a high burden of medical comorbidities, limiting the generalizability of these findings. It is also limited to a single, early COVID-19 pandemic time point. The longitudinal nature of the WIHS introduces a selection bias toward response due to the participants’ demonstrated history of ongoing engagement in research. Furthermore, we excluded those with missing data at the index visit, including alcohol and substance use data, which could result in an overestimation of response rates, but an underrepresentation of subgroups in the overall sample for this study. Due to the sensitive nature of alcohol and substance use, these measures are also subject to social desirability bias, which can lead to underestimated alcohol and substance use prevalence. There may be factors associated with phone survey response and our outcome measures that were not included in our adjusted analysis, leading to inaccurate weighted prevalence estimates of risky drinking and substance use. Furthermore, the factors in the analysis were derived from a pre-pandemic index visit and may have changed in relation to the pandemic (e.g., employment, income, and housing changes).
Conclusion
Among a sample of women living with and without HIV who previously attended in-person study visits, participants of Hispanic ethnicity (compared to Black/African American race) and those who reported pre-pandemic substance use had lower odds of responding to an alcohol and substance use phone survey administered early in the COVID-19 pandemic, while those of White race (compared to Black/African American race) and those with stable housing had increased odds of responding, suggesting an underrepresentation of those most impacted by the COVID-19 pandemic. As remote survey methods become more common, best practices for using remote study methods are needed to ensure that data remain representative of the target population and investigators should assess for selection bias.
Acknowledgments
The authors gratefully acknowledge the contributions of the study participants and dedication of the staff at the MWCCS sites.
Funding
The contents of this publication are solely the responsibility of the authors and do not represent the official views of the National Institutes of Health (NIH). MWCCS (Principal Investigators): Atlanta CRS (Ighovwerha Ofotokun, Anandi Sheth, and Gina Wingood), U01-HL146241; Bronx CRS (Kathryn Anastos, David Hanna, and Anjali Sharma), U01-HL146204; Brooklyn CRS (Deborah Gustafson and Tracey Wilson), U01-HL146202; Data Analysis and Coordination Center (Gypsyamber D’Souza, Stephen Gange and Elizabeth Topper), U01-HL146193; Chicago-Cook County CRS (Mardge Cohen and Audrey French), U01-HL146245; Northern California CRS (Bradley Aouizerat, Jennifer Price, and Phyllis Tien), U01-HL146242; Metropolitan Washington CRS (Seble Kassaye and Daniel Merenstein), U01-HL146205; Miami CRS (Maria Alcaide, Margaret Fischl, and Deborah Jones), U01-HL146208; UAB-MS CRS (Mirjam-Colette Kempf, Jodie Dionne-Odom, and Deborah Konkle-Parker), U01-HL146192; UNC CRS (Adaora Adimora and Michelle Floris-Moore), U01-HL146194. The MWCCS is funded primarily by the National Heart, Lung, and Blood Institute (NHLBI), with additional co-funding from thirteen institutes of the NIH and in coordination and alignment with the research priorities of the Office of AIDS Research (OAR). MWCCS data collection is also supported by UL1-TR000004 (UCSF CTSA), UL1-TR003098 (JHU ICTR), UL1-TR001881 (UCLA CTSI), P30-AI-050409 (Atlanta CFAR), P30-AI-073961 (Miami CFAR), P30-AI-050410 (UNC CFAR), P30-AI-027767 (UAB CFAR), and P30-MH-116867 (Miami CHARM). HRT’s time on this project was supported by the Infectious Disease Society of America Foundation/HIV Medicine Association Grant for Emerging Researchers/Clinicians Mentorship Program, which had no role in the study design, analysis, manuscript preparation or approval. Support was also provided to JAH (NIH K24 AA022586) and PCT (NIH K24 AI108516) in the preparation of this manuscript.
Footnotes
Disclosure statement
RJD has received consulting fees from the Department of Defense, Morehouse School of Medicine, Benten Technologies, and Northwell Health. AAA has received consulting fees from Merck and Gilead; Merck and Gilead have provided her institution with funding for her research. AS has received consulting fees from Gilead; Gilead has provided her institution with funding for her research. Gilead, Merck, and Abbvie have provided JCP’s institution with funding for her research. Merck and Gilead have provided PCT’s institution with funding for her research. JAH has received consulting fees from Pear Therapeutics. The other authors have no conflicts of interest.
Data availability statement
Access to individual-level data from the MWCCS may be obtained upon review and approval of a MWCCS concept sheet. Links and instructions for online concept sheet submission are on the study website (http://mwccs.org/).
References
- 1.De Man J, Campbell L, Tabana H, Wouters E. The pandemic of online research in times of COVID-19. BMJ Open. 2021;11:e043866. doi: 10.1136/bmjopen-2020-043866. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Menon S, Sonderegger P, Totapally S. Five questions to consider when conducting COVID-19 phone research. BMJ Glob Health. 2021;6:e004917. doi: 10.1136/bmjgh-2020-004917. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Aquilino WS. Telephone versus face-to-face interviewing for household Drug use surveys. Int J Addict. 1991;27:71–91. doi: 10.3109/10826089109063463. [DOI] [PubMed] [Google Scholar]
- 4.Boniface S, Scholes S, Shelton N, Connor J, Schooling CM. Assessment of non-response bias in estimates of alcohol consumption: applying the continuum of resistance model in a general population survey in England. PLOS One. 2017;12:e0170892. doi: 10.1371/journal.pone.0170892. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.McCabe SE, West BT. Selective nonresponse bias in population-based survey estimates of drug use behaviors in the United States. Soc Psych Psych Epid. 2016;51:141–53. doi: 10.1007/s00127-015-1122-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Nwaru CA, Lövestad S, Gunnarsdóttir H, Sundh V, Hensing G. Determinants of non-response in a longitudinal study of participants in the women and alcohol in Gothenburg project. Women Health. 2021;61:452–60. doi: 10.1080/03630242.2021.1917482. [DOI] [PubMed] [Google Scholar]
- 7.Peytcheva E, Wang K, Bose J, Hedden S, Hunter D, Tice P. Nonresponse bias in demographic and survey estimates from the mental health surveillance study of the national survey on drug use and health. Rockville (MD): Substance Abuse and Mental Health Services Administration; 2017. [Google Scholar]
- 8.Rehm J, Kilian C, Rovira P, Shield KD, Manthey J. The elusiveness of representativeness in general population surveys for alcohol. Drug Alcohol Rev. 2021;40:161–65. doi: 10.1111/dar.13148. [DOI] [PubMed] [Google Scholar]
- 9.Zhao J, Stockwell T, Macdonald S. Non-response bias in alcohol and drug population surveys. Drug Alcohol Rev. 2009;28:648–57. doi: 10.1111/j.1465-3362.2009.00077.x. [DOI] [PubMed] [Google Scholar]
- 10.Greenfield TK, Midanik LT, Rogers JD. Effects of telephone versus face-to-face interview modes on reports of alcohol consumption. Addiction. 2000;95:277–84. doi: 10.1046/j.1360-0443.2000.95227714.x. [DOI] [PubMed] [Google Scholar]
- 11.Ahmad F, Rossen L, Sutton P. Provisional drug overdose death counts. Hyattsville (MD): National Center for Health Statistics; 2021. [Google Scholar]
- 12.Barbosa C, Cowell AJ, Dowd WN. Alcohol consumption in response to the COVID-19 pandemic in the United States. J Addict Med. 2020;15:341–44. doi: 10.1097/ADM.0000000000000767. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Pollard MS, Tucker JS, Green HD Jr. Changes in adult alcohol use and consequences during the COVID-19 pandemic in the US. JAMA Netw Open. 2020;3:e2022942. doi: 10.1001/jamanetworkopen.2020.22942. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Rodriguez LM, Litt DM, Stewart SH. Drinking to cope with the pandemic: the unique associations of COVID-19-related perceived threat and psychological distress to drinking behaviors in American men and women. Addict Behav. 2020;110:106532. doi: 10.1016/j.addbeh.2020.106532. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Connor J, Madhavan S, Mokashi M, Amanuel H, Johnson NR, Pace LE, Bartz D. Health risks and outcomes that disproportionately affect women during the covid-19 pandemic: a review. Social Sci Med. 2020;266:113364. 10.1016/j.socscimed.2020.113364 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Jones DL, Rodriguez VJ, Salazar AS, Montgomerie E, Raccamarich PD, Uribe Starita C, Barreto Ojeda IT, Beauchamps L, Vazquez A, Martinez T, et al. Sex differences in the association between stress, loneliness, and COVID-19 burden among people with HIV in the US. AIDS Res Hum Retroviruses. 2021;37:314–21. doi: 10.1089/aid.2020.0289. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Philpot LM, Ramar P, Roellinger DL, Barry BA, Sharma P, Ebbert JO. Changes in social relationships during an initial “stay-at-home” phase of the COVID-19 pandemic: a longitudinal survey study in the U.S. Social Sci Med. 2021;274:113779. doi: 10.1016/j.socscimed.2021.113779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Rodriguez LM, Litt DM, Stewart SH. Drinking to cope with the pandemic: the unique associations of COVID-19-related perceived threat and psychological distress to drinking behaviors in American men and women. Addict Behav. 2020;110:106532. doi: 10.1016/j.addbeh.2020.106532. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Adimora AA, Ramirez C, Benning L, Greenblatt RM, Kempf M-C, Tien PC, Kassaye SG, Anastos K, Cohen M, Minkoff H, et al. Cohort profile: the Women’s Interagency HIV Study (WIHS). Int J Epidemiol. 2018;47:393–4i. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.D’Souza G, Bhondoekhan F, Benning L, Margolick JB, Adedimeji AA, Adimora AA, Alcaide ML, Cohen MH, Detels R, Friedman MR, et al. Characteristics of the MACS/WIHS combined cohort study: opportunities for research on aging with HIV in the longest US observational study of HIV. Am J Epidemiol. 2021;190:1457–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hessol NA, Schneider M, Greenblatt RM, Bacon M, Barranday Y, Holman S, Robison E, Williams C, Cohen M, Weber K. Retention of women enrolled in a prospective study of human immunodeficiency virus infection: impact of race, unstable housing, and use of human immunodeficiency virus therapy. Am J Epidemiol. 2001;154:563–73. doi: 10.1093/aje/154.6.563. [DOI] [PubMed] [Google Scholar]
- 22.Hessol NA, Weber KM, Holman S, Robison E, Goparaju L, Alden CB, et al. Retention and attendance of women enrolled in a large prospective study of HIV-1 in the United States. J Women Health. 2009;18:1627–37. doi: 10.1089/jwh.2008.1337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Higgins-Biddle JC, Babor TF. A review of the alcohol use Disorders identification test (AUDIT), AUDIT-C, and USAUDIT for screening in the United States: past issues and future directions. Am J Drug Alcohol Abuse. 2018;44:578–86. doi: 10.1080/00952990.2018.1456545. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Devins GM, Orme CM, Costello CG, Binik YM, Frizzell B, Stam HJ, Pullin WM Measuring depressive symptoms in illness populations: psychometric properties of the Center for Epidemiologic Studies Depression (CES-D) Scale. Psychology & Health. 1988;2(2):139–56. doi: 10.1080/08870448808400349. [DOI] [Google Scholar]
- 25.Thomas JL, Jones GN, Scarinci IC, Mehan DJ, Brantley PJ. The utility of the ces-D as a depression screening measure among low-income women attending primary care clinics. Int J Psychiatry Med. 2001;31:25–40. doi: 10.2190/FUFR-PK9F-6U10-JXRK. [DOI] [PubMed] [Google Scholar]
- 26.Dahlhamer JM, Bramlett MD, Maitland A, Blumberg SJ. Preliminary evaluation of the nonresponse due to the COVID-19 pandemic on national health interview survey estimates, April-June 2020. Hyattsville (MD): Centers for Disease Control and Prevention; 2021. [Google Scholar]
- 27.Dawson DA, Goldstein RB, Pickering RP, Grant BF. Nonresponse bias in survey estimates of alcohol consumption and its association with harm. J Stud Alcohol Drugs. 2014;75:695–703. doi: 10.15288/jsad.2014.75.695. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Haley DF, Lucas J, Golin CE, Wang J, Hughes JP, Emel L, El-Sadr W, Frew PM, Justman J, Adimora AA, et al. Retention strategies and factors associated with missed visits among low income women at increased risk of HIV acquisition in the US (HPTN 064). AIDS Patient Care STDS. 2014;28:206–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Ward JM, Anne Edwards K. Assessing the link between survey interview method and survey outcomes: evidence from the CPS and the COVID-19 pandemic. Labour Econ. 2021;72:102060. doi: 10.1016/j.labeco.2021.102060. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Team R. RStudio: integrated development for R. In: Team R, editor. Boston, MA: RStudio, PBC; 2020. [Google Scholar]
- 31.Hsia DS, Williams KM, Beyl RA. Participant perspectives concerning resuming clinical research in the era of COVID-19. J Clin Med Res. 2022;14:165–69. doi: 10.14740/jocmr4670. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Falcon R, Bridge DA, Currier J, Squires K, Hagins D, Schaible D, et al. Recruitment and retention of diverse populations in antiretroviral clinical Trials: practical applications from the gender, race and clinical experience study. J Women Health. 2011;20:1043–50. doi: 10.1089/jwh.2010.2504. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Kelty M, Bates A, Pinn VW. Chapter 13 - National Institutes of Health policy on the inclusion of women and minorities as subjects in clinical research. In: Gallin J, Ognibene F, editors. Principles and practice of clinical research. 3rd ed. Boston: Academic Press; 2012. p. 147–59. [Google Scholar]
- 34.Melamed OC, Hauck TS, Buckley L, Selby P, Mulsant BH. Article commentary: covid-19 and persons with substance use Disorders: inequities and mitigation strategies. Subst Abus. 2020;41:286–91. doi: 10.1080/08897077.2020.1784363. [DOI] [PubMed] [Google Scholar]
- 35.Perri M, Dosani N, Hwang SW. COVID-19 and people experiencing homelessness: challenges and mitigation strategies. Can Med Assoc J. 2020;192:E716–E9. doi: 10.1503/cmaj.200834. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Tai DBG, Shah A, Doubeni CA, Sia IG, Wieland ML. The disproportionate impact of COVID-19 on racial and ethnic minorities in the United States. Clin Infect Dis. 2021;72:703–06. doi: 10.1093/cid/ciaa815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Link MW, Mokdad AH, Stackhouse HF, Flowers NT. Race, ethnicity, and linguistic isolation as determinants of participation in public health surveillance surveys. Prev Chronic Dis. 2006;3:A09–A. [PMC free article] [PubMed] [Google Scholar]
- 38.Rothbaum J, Bee A. Coronavirus infects surveys, too: survey nonresponse bias and the coronavirus pandemic. Washington (DC): U.S. Census Bureau; 2021. May 3. [Google Scholar]
- 39.Abshire M, Dinglas VD, Cajita MIA, Eakin MN, Needham DM, Himmelfarb CD. Participant retention practices in longitudinal clinical research studies with high retention rates. BMC Med Res Methodol. 2017;17. doi: 10.1186/s12874-017-0310-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Stewart C, Kopinski H, Liebschutz J, Holmdahl I, Keosaian J, Herman D, Anderson B, Stein M. The provision of cell phones as a recruitment and retention strategy for people who inject drugs enrolling in a randomized trial. Drug Alcohol Depen. 2018;184:20–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Teague S, Youssef GJ, Macdonald JA, Sciberras E, Shatte A, Fuller-Tyszkiewicz M, Greenwood C, McIntosh J, Olsson CA, Hutchinson D, et al. Retention strategies in longitudinal cohort studies: a systematic review and meta-analysis. BMC Med Res Methodol. 2018;18. doi: 10.1186/s12874-018-0586-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Pertl K, Petluri R, Wiest K, Hoffman K, McCarty D, Levander XA, Chan B, Martin SA, Korthuis PT. Recruitment challenges for a prospective telehealth cohort study. Contem Clin Trials Commun. 2023;31:101043. doi: 10.1016/j.conctc.2022.101043. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Tipre M, Scarinci IC, Pandya VN, Kim Y-I, Bae S,Peral S, Hardy C, Baskin ML. Attitudes toward telemedicine among urban and rural residents. J Telemed Telecare. 2022;1357633X2210942. doi: 10.1177/1357633X221094215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Nguyen M-L, Garcia F, Juarez J, Zeng B, Khoong EC, Nijagal MA, Sarkar U, Su G, Lyles CR. Satisfaction can co-exist with hesitation: qualitative analysis of acceptability of telemedicine among multi-lingual patients in a safety-net healthcare system during the COVID-19 pandemic. BMC Health Serv Res. 2022;22. doi: 10.1186/s12913-022-07547-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Imbert E, Hickey MD, Del Rosario JB, Conte M, Kerkhoff AD, Clemsenzi-Allen A, Riley ED, Havlir DV, Gandhi M. Heterogeneous preferences for care engagement among people with HIV experiencing homelessness or unstable housing during the COVID-19 pandemic. J Acquir Immune Defic Syndr. 2022;90:140–45. doi: 10.1097/QAI.0000000000002929. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Access to individual-level data from the MWCCS may be obtained upon review and approval of a MWCCS concept sheet. Links and instructions for online concept sheet submission are on the study website (http://mwccs.org/).