The study included two components. The first was a survey evaluating the acceptability of HB-PrEP and use of the Tasso device for specimen collection. The second involved having a subset of participants self-collect blood specimens using the Tasso device and answer questions regarding their experience. We also compared RPR, HIV antigen/antibody, and creatinine test results from clinician-collected venipuncture and self-collected Tasso specimens.
Participants were patients at either the Public Health-Seattle & King County (PHSKC) Sexual Health Clinic (SHC) or the Max Clinic, both of which are located on the Harborview Medical Center campus in Seattle, WA, USA. The SHC provides comprehensive medical services related to STIs and is the largest single provider of PrEP in Washington State, with approximately 650 current PrEP patients. The Max Clinic is a walk-in clinic for persons living with HIV .
Recruitment and enrollment
The study population comprised two groups enrolled from May 2020 to February 2021. The first group was recruited online from a subset of current or eligible PrEP users (approximately 75% of all SHC PrEP patients) who had agreed to receive communication from clinic staff through a secure text messaging platform. At study initiation and three months following initiation, we sent patients a recruitment text that contained a brief introduction to the study and a link to complete an online consent form.
The second group included English-speaking patients receiving in-person clinical services at the Max Clinic or SHC who had venipuncture specimens taken as part of their routine care. Additionally, patients had to be either eligible for or currently using PrEP, have a known or suspected new syphilis infection, or be living with HIV. Since our study began shortly after COVID-19 restrictions significantly reduced the number of face-to-face visits allowed at University of Washington (UW) and PHSKC facilities, the SHC population was comprised primarily of persons with symptoms suggestive of STI. Inclusion criteria for this second group were designed to ensure that the study included persons with positive HIV and syphilis tests to allow for evaluation of the agreement between venipuncture and Tasso device-collected specimens. Persons taking anticoagulant medications were excluded from performing the blood self-collection. Online consent forms and survey responses were collected and managed using secure, web-based REDCap (Research Electronic Data Capture) tools  hosted at the UW’s Institute of Translational Health Sciences. Survey instruments for both groups are available as appendices (Additional files 1, 2). Study procedures were approved by the University of Washington’s Human Subjects Division and institutional review board (study #00009004) and Public Health-Seattle & King County’s Research Administrative Review Committee (study #20-680).
After enrollment, participants were asked to view a brief video introducing the Tasso device and concept of remote PrEP monitoring and then complete an online survey on their interest in HB-PrEP. Online participants did not receive compensation for study participation.
Eligible persons were referred to the study by medical providers after completing their clinic visit. Following enrollment, participants watched a training video  and received detailed instructions on how to obtain blood samples using the Tasso device (Additional file 3: Fig. S3). To simulate unsupervised home collection conditions and comply with UW COVID-19 distancing restrictions, study staff were not physically present during the blood self-collections but remained available as needed for assistance via telephone from offices adjoining the clinic rooms. The study ordered serum tests on self-collected specimens to match tests ordered during the clinical visit, allowing for paired comparisons between the clinician-collected venipuncture (gold standard) and self-collected Tasso samples. As such, the number of tests performed for each participant varied (e.g., a person attending a routine month 15 PrEP monitoring visit would typically have HIV and syphilis testing but not a creatinine test). In the first months after study initiation, we discovered that use of only one Tasso device often resulted in inadequate blood volumes to run all three tests. Thus, we revised study procedures to direct all participants to use two devices for sequential collections on either the same or different arms. After specimen collection, participants completed a survey on the acceptability of self-collecting specimens and likelihood of participating in a HB-PrEP program. In cases where collection was unsuccessful, we attempted to determine the likely cause by reviewing the collection technique and asking questions about recent activity in the day leading up to collection. All enrollees who attempted the blood collection and completed a post-procedure survey were offered a $10 gift card.
Outcome and sample size considerations
We defined thresholds for acceptability and feasibility prior to initiation of the study. Acceptability was defined as ≥ 85% of participants reporting that the device was not difficult to use and that they could envision being able to self-collect a specimen at home. We defined feasibility as ≥ 80% of participants successfully self-collecting samples, without direct supervision, of volume sufficient to perform at least one of the three serum tests for PrEP monitoring.
Given the significant clinical implications of misdiagnosing HIV, we determined a priori that paired HIV specimens should be 100% concordant. For syphilis test accuracy, we specified that qualitative results would be concordant for ≥ 75% of cases and that paired quantitative titers would be different by ≤ 1 dilution, based on the fact that a two-titer difference is usually defined to represent a clinically significant change . We defined the acceptable range of variation for creatinine specimens to be ± 0.15 mg/dL. Based on data from our SHC PrEP cohorts and previous evaluations of capillary vs. venous creatinine values that suggest venous sample results are on average 0.15 mg/dL higher , we hypothesized that the average creatinine estimates would be 1.1 and 0.95 mg/dL for venipuncture and Tasso specimens, respectively. Thus, we determined 10 or more paired samples would yield at least 80% power to evaluate whether the mean creatinine difference between specimens was within a clinically acceptable range.
All sample processing and testing was performed in Clinical Laboratory Improvement Amendments-certified facilities. Clinician-collected venipuncture samples were tested for HIV, creatinine and syphilis according to routine SHC protocols. Tasso-collected whole blood specimens were first delivered to the PHSKC laboratory for centrifugation, after which serum was used for testing. Sera were tested for HIV using the 4th generation GS HIV Combo antigen/antibody enzyme-linked immunosorbent assay (EIA) (Bio-Rad, Hercules, CA, USA) and for syphilis using the ASI RPR card (Arlington Scientific Inc. Springville, UT, USA). All positive RPR tests from Tasso samples were quantified up to a 1:16 titer; values beyond this were determined if residual specimen volume permitted. The CAPTIA™ syphilis IgG EIA (Bio-Rad, Hercules, CA, USA) was used for confirmation of discordant venipuncture/Tasso RPR specimens. HIV and syphilis tests were internally validated by the PHSKC laboratory prior to this study. Creatinine testing was performed at the UW laboratory using the automated Beckman Coulter AU analyzer (Beckman Coulter Inc., Brea, CA, USA).
We used descriptive statistics to characterize participant demographics, survey responses, and the proportion of participants who attempted and completed self-collections. For HIV and qualitative RPR values, we determined the percent positive agreement (PPA) and negative percent agreement (NPA) between venipuncture and self-collected specimen results. We also used descriptive statistics to characterize the means, standard deviations and differences for paired quantitative RPR titers and creatinine values. Correlation between the paired sample results was analyzed using Pearson’s correlation coefficients and linear regression. Quantitative RPR titers were plotted on the binary logarithm scale (e.g., value of 3 represents a titer of 1:8). All analyses were conducted using R version 3.5.1 (R Foundation for Statistical Computing, 2018) and used a two-sided alpha level of 0.05 for statistical significance.