Abstract
Background:
During the COVID-19 pandemic, patients experienced significant care disruptions, including lab monitoring. We investigated changes in the time between viral load (VL) checks for people with HIV (PWH) associated with the pandemic.
Setting and Methods:
This was an observational analysis of VLs of PWH in routine care at a large subspecialty clinic. At pandemic onset, the clinic temporarily closed its onsite laboratory. The exposure was time period (time-varying): pre-pandemic (January 1st 2019-March 15th, 2020); pandemic lab-closed (March 16th-July 12th, 2020); and pandemic lab-open (July 13th-December 31st, 2020). We estimated time from an index VL to a subsequent VL, stratified by whether the index VL was suppressed (≤200 copies/mL). We also calculated cumulative incidence of a non-suppressed VL following a suppressed index VL, and of re-suppression following a loss of viral suppression.
Results:
Compared to pre-pandemic, hazard ratios for next VL check were: 0.34 (95% CI: 0.30, 0.37, lab-closed) and 0.73 (CI: 0.68, 0.78, lab-open) for suppressed patients; 0.56 (CI: 0.42, 0.79, lab-closed) and 0.92 (95% CI: 0.76, 1.10, lab-open) for non-suppressed patients. The 12-month cumulative incidence of loss of suppression was the same in the pandemic lab-open (4%) and pre-pandemic period (4%). The hazard of re-suppression following loss of suppression was lower during the pandemic lab-open versus the pre-pandemic period (hazard ratio: 0.68, 95% CI: 0.50, 0.92).
Conclusions:
Early pandemic restrictions and lab closure significantly delayed VL monitoring. Once the lab re-opened, non-suppressed patients resumed normal monitoring. Suppressed patients still had a delay, but no significant loss of suppression.
Keywords: Human Immunodeficiency Virus, Viral Load, COVID-19 Pandemic, Continuum of Care
Introduction
Viral suppression is critical to ending the HIV epidemic,1 as patients whose viral load is undetectable are unlikely to transmit infection.2,3 Viral suppression with antiretroviral therapy (ART) has also drastically reduced morbidity and mortality rates for people with HIV (PWH).4,5 Given the importance of viral suppression, viral load monitoring is recommended frequently at initiation of therapy, then every 3–6 months thereafter depending on the duration of durable suppression.
The COVID-19 pandemic has had a significant impact on the delivery of HIV care, including viral load monitoring. The pandemic precipitated a rapid shift of the bulk of care from in-person to telemedicine.6 Social distancing recommendations limited use of public transit during the pandemic,7 a barrier to lab collection for patients without access to private transportation.8 Surveys suggest Ryan White clinics saw a significant decline in lab monitoring frequency during early months of the pandemic,9 with some closing their onsite laboratories as part of social distancing measures. Additionally, early pandemic guidance also advised providers to delay lab monitoring for patients who have been suppressed.10
Herein, we aimed to describe time between viral load measurements for people living with HIV during the COVID-19 pandemic compared to pre-pandemic. We stratified analyses by whether the index (i.e., first) viral load in a pair of viral load measurements was suppressed or not, recognizing that the frequency of monitoring depends on prior suppression status even under normal conditions. We also identified groups at risk for longer gaps in viral monitoring during the pandemic. Finally, in secondary analyses, we estimated the cumulative incidence of having a non-suppressed viral load after a suppressed index viral load, and of time to viral load suppression among patients who were unsuppressed at enrollment in the clinic, and among established patients who lost viral suppression during follow-up, comparing pandemic to pre-pandemic periods.
Methods
Study Sample.
The John G. Bartlett Specialty Practice is a large subspecialty clinic in East Baltimore affiliated with the Johns Hopkins Hospital. The clinic provides comprehensive continuity care to people with HIV or Hepatitis C. The Johns Hopkins HIV Clinical Cohort enrolls clinic patients who consent to share their medical record data. The cohort has been described previously;11 data include self-reported age, gender, race, ethnicity, HIV acquisition risk factors, provider encounters, hospital admissions, lab data, prescribed treatment, and clinical diagnoses. Lab data collected includes both results from within the Johns Hopkins Health System as well as those from the two largest commercial laboratories in Baltimore. The unit of analysis was viral load samples, not individuals, and we included all samples collected between January 1st, 2019 and December 31st, 2020. In measuring the time between viral loads, each observation consisted of an “index” viral load and the subsequent “outcome” viral load (if a subsequent test was collected before censoring). All index viral loads also served as outcome viral loads for the prior index viral load, with the exception of each individual’s first viral load in the study period. Consequently, individuals who had multiple viral loads collected during the study period could contribute multiple observations to the analysis (Supplemental Figure 1).
Exposure of Interest.
In response to the COVID-19 pandemic, the John G. Bartlett Specialty Practice closed its onsite laboratory and converted most encounters from in-person to telemedicine on March 16th, 2020.12 Labs for the first 4 months of the pandemic had to be collected offsite. Then on July 12th, 2020 the onsite lab reopened. Our exposure was time-varying calendar period (i.e., probability of having a follow-up viral load test was a function of calendar period of follow-up, rather than as a function of calendar period in which the index viral load was collected). The pre-pandemic period was January 1st, 2019 to March 15th, 2020. The pandemic lab-closed period was March 16th, 2020-July 12th, 2020 and the pandemic lab-open period was July 13th, 2020-December 31st, 2020.
Outcome of Interest.
The outcome of interest was the time from each “index” viral load to a subsequent “outcome” viral load.
Covariates.
We were interested in risk factors for longer times between viral load tests and considered the following covariates: age category (20–39 years, 40–59 years, ≥60 years), gender, race, and ethnicity. Recent (within past 6 months) misuse of alcohol and use of cocaine or heroin were based on the findings of trained chart abstractors utilizing provider notes, toxicology testing, and treatment referrals. We also included insurance status (private or non-private, where non-private included Ryan White, Medicaid, or uninsured). We describe the study sample according to time since enrollment in the clinic and most recent CD4 cell count, however because the sample was well-established in care, we did not anticipate these variables would have sufficient heterogeneity to be associated with the frequency of viral load monitoring.
Statistical Analysis.
We reported hazard ratios (HR) and 95% confidence intervals (CI) from a Cox proportional hazards model and cumulative incidence curves for time from index viral load to subsequent viral load, comparing each of the two pandemic periods (lab closed and lab open) to the pre-pandemic period as our exposure. Cumulative incidence curves were estimated as the complement of the Kaplan-Meier function; we calculated and reported risk at 3, 6, and 12 months from the index viral load. Exposure was time-varying. If no subsequent viral load was collected by the end of a study period, that observation was censored and late-entered into the following period (Supplemental Figure 1, Observation 2). Because exposure (time period) was treated as time-varying, the crux of this analysis is the hazard of having a viral load measured – that is, among people with the same amount of time since their last viral load measurement, how does the probability they have a follow-up viral load measurement differ by time period. Additionally, because exposure was treated as time-varying, the x-axis on the incidence curves we plotted were longer than the exposure periods themselves. These curves are interpretable as the cumulative incidence of a subsequent viral load measurement we would expect to see if the lab and clinic conditions during that time period persisted for the entire year. We administratively censored time at 12 months or the cohort close date, December 31, 2020, whichever came first.
To describe risk factors for delays in viral load monitoring due to the pandemic, we report HRs for hazard of a subsequent viral load associated with time period from a Cox proportional hazards model within strata of each covariate above. We report p-values for an interaction term between each covariate and time period to identify whether the association between viral load monitoring and time period was different by patient covariate.
Recognizing that the frequency of viral load monitoring differs based on a patient’s suppression status, we stratified all analyses by whether the index viral load in each observation was suppressed (≤200 copies/ml) or not. The suppressed analysis was adjusted for duration of time a patient had previously been suppressed for, recognizing that recently suppressed patients may be monitored more frequently. This duration of suppression was measured as a backwards look from the observation’s index viral load to the patient’s earliest viral load on record without any intervening non-suppressed viral loads. Gaps greater than 12 months without monitoring were considered interruption of suppression. To account for possibly correlated outcomes within an individual (due to individuals being allowed to contribute multiple observations) we used the robust variance estimator for reporting all confidence limits and p-values from Cox proportional hazards models and Fine and Gray models.
Secondary Analyses: Discordant Viral Loads.
Our primary analysis was stratified by the index viral load, irrespective of whether the subsequent viral load was suppressed or not. In secondary analyses we aimed to investigate discordant viral load pairs. The goal of these analyses was to provide insight into (1) loss of suppression during the pandemic and (2) the time to suppression during the pandemic.
Suppressed to Non-suppressed.
For this analysis we restricted to observations with a suppressed index viral load. The outcome of interest was time to the subsequent viral load and whether it was non-suppressed. We treated subsequent viral loads that were suppressed as competing events. We report the cumulative incidence of non-suppression for each calendar period (estimated using the Aalen-Johansen estimator) and report sub-distribution hazard ratios for time to non-suppression by period estimated from Fine and Gray models.
Non-suppressed to Suppressed.
For this analysis, we restricted to observations with a non-suppressed index viral load. We stratified these into two clinically distinct groups: viral loads from (i) patients who were non-suppressed at enrollment and (ii) established patients who became non-suppressed after having been previously suppressed. Rather than time to subsequent viral load, we were interested in time to (i) initial suppression or (ii) re-suppression. The outcome of interest was time to first subsequent suppressed viral load (≤200 copies/ml) (interim viral loads that were not yet suppressed were disregarded). We report the cumulative incidence of suppression for each calendar period and report sub-distribution hazard ratios for time to suppression by period estimated from Cox proportional hazards models.
Results
Viral Load Observations.
There were 5,498 (70.9%) suppressed index viral loads pre-pandemic (1/1/2019–3/15/2020), 618 (7.9%) during the pandemic lab-closed period (3/16/2020–7/12/2020), and 1,644 (21.2%) during the pandemic lab-open period (7/13/2020–12/31/2020); 3,950 people contributed ≥1 observation to this analysis. There were 750 (73.2%) non-suppressed index viral loads pre-pandemic, 84 (8.2%) during the pandemic lab-closed period, and 191 (18.6%) during the pandemic lab-open period; 601 people contributed ≥1 observation to this analysis. Overall, individuals contributed a median of 3 (interquartile range: 2,5) index viral loads to this analysis. Viral loads were contributed by people that were predominantly age 40–59, male, Black and non-Hispanic (Table 1). The mean years enrolled in the cohort at the start of follow-up was 12.2 (interquartile range: 6.0, 19.4) and the mean CD4 cell count was 593 (interquartile range: 370–842).
Table 1:
Patient Characteristics for Viral Load Samples by Time Period for Non-Suppressed and Suppressed Index Viral Loads
| Non-Suppressed | Suppresseda | |||||
|---|---|---|---|---|---|---|
| Pre-Pandemic | Pandemic, Lab Closed | Pandemic, Lab Open | Pre-Pandemic | Pandemic, Lab Closed | Pandemic, Lab Open | |
| Individuals | n = 403 | n = 63 | n = 135 | n = 2,065 | n = 565 | n = 1,320 |
| Person-Time (Days) | 59,965 | 19,050 | 24,325 | 650,321 | 220,311 | 296,279 |
| Viral Loads Collected | 750 | 84 | 191 | 5,498 | 618 | 1,644 |
| Age | ||||||
| Age 20–39 | 247 (33%) | 30 (36%) | 63 (33%) | 820 (15%) | 69 (11%) | 206 (13%) |
| Age 40–59 | 352 (47%) | 34 (40%) | 85 (45%) | 2,753 (50%) | 323 (52%) | 798 (49%) |
| Age 60+ | 151 (20%) | 20 (24%) | 41 (23%) | 1,925 (35%) | 226 (37%) | 640 (39%) |
| Male | 454 (61%) | 51 (61%) | 110 (58%) | 3,516 (64%) | 395 (64%) | 1,029 (63%) |
| Race | ||||||
| Black | 649 (87%) | 72 (86%) | 164 (86%) | 4,313 (78%) | 453 (73%) | 1,266 (77%) |
| White | 79 (11%) | 11 (13%) | 21 (11%) | 967 (18%) | 142 (23%) | 218 (19%) |
| Other | 22 (3%) | 1 (1%) | 3 (3%) | 218 (4%) | 23 (4%) | 60 (4%) |
| Hispanic | 12 (2%) | 1 (1%) | 5 (3%) | 146 (3%) | 16 (3%) | 46 (3%) |
| Recent Smoking b | ||||||
| No | 219 (45%) | 27 (50%) | 61 (51%) | 2,945 (65%) | 331 (69%) | 848 (68%) |
| Yes | 266 (55%) | 27 (50%) | 58 (49%) | 1,616 (35%) | 151 (31%) | 392 (32%) |
| Missing | 265 | 30 | 72 | 937 | 136 | 404 |
| Recent Alcohol b | ||||||
| No | 380 (78%) | 43 (80%) | 100 (83%) | 4,069 (90%) | 435 (90%) | 1,135 (92%) |
| Yes | 105 (22%) | 11 (20%) | 21 (17%) | 480 (10%) | 48 (10%) | 105 (8%) |
| Missing | 265 | 30 | 70 | 953 | 135 | 404 |
| Recent Cocaine/Heroin b | ||||||
| No | 372 (77%) | 44 (81%) | 103 (85%) | 4,065 (89%) | 449 (93%) | 1,144 (92%) |
| Yes | 112 (23%) | 10 (19%) | 18 (15%) | 480 (11%) | 35 (7%) | 98 (8%) |
| Missing | 265 | 30 | 70 | 953 | 134 | 402 |
| Insurance c | ||||||
| Public | 392 (52%) | 40 (48%) | 99 (52%) | 1,807 (33%) | 168 (28%) | 511 (31%) |
| Private | 355 (48%) | 43 (52%) | 92 (48%) | 3,647 (67%) | 442 (72%) | 1,119 (69%) |
| Missing | 3 | 1 | 0 | 44 | 8 | 14 |
| Months Suppressed d | 0 | 0 | 0 | 34 (8,60) | 40 (8, 61) | 38 (9, 65) |
Suppressed if Viral Load ≤200 copies/mL on most recent check.
Smoking, Alcohol, Cocaine and Heroin use were obtained from medical record review of provider notes, toxicology screens and treatment referrals, conducted every 6 months by trained abstractors. Data were restricted to abstractions in the year prior to the viral load.
Non-Private Insurance defined as covered by Medicaid, Ryan White, or Uninsured.
Median (IQR)
Time to Next Viral Load for Suppressed Patients.
Pre-pandemic, the proportion of suppressed viral loads that had a follow-up viral load by 12 months, was 91%. Under pandemic lab-closed conditions, the proportion was 59%, and under pandemic lab-open conditions, it was 87% (Figure 1). Compared to pre-pandemic, the HRs for subsequent viral load collection were 0.34 (95% CI: 0.30, 0.37) during the pandemic lab-closed period and 0.73 (95% CI: 0.68, 0.78) during the pandemic lab-open period (Table 2). The hazard for viral load monitoring during the pandemic lab-open period relative to pre-pandemic was significantly lower for patients who were Black (HR: 0.71, 95% CI: 0.66, 0.76) compared to White (HR: 0.85, 95% CI: 0.74, 0.98), interaction p-value=0.02; patients who had recent cocaine or heroin use (HR: 0.56, 95% CI: 0.43, 0.73) compared to no recent substance use (HR: 0.76, 95% CI: 0.71, 0.82), interaction p-value=0.03; and patients who had non-private insurance (HR: 0.65, 95% CI: 0.58, 0.74) compared to patients with private insurance (HR: 0.77, 95% CI: 0.71, 0.83), interaction p-value=0.03.
Figure 1 -. Kaplan-Meier Curve of Time from Suppressed Viral Load to Subsequent Viral Load.
The proportion of suppressed viral loads with a follow-up viral load collected after 3, 6, and 12 months, comparing the pre-pandemic, pandemic lab-closed, and pandemic lab-open periods. Cumulative incidences and hazard ratios for subsequent viral load collection in each period are reported.
*Adjusted for duration of suppression prior to index viral load.
Table 2:
Hazard Ratios for Time Between Viral Loads Among Those with a Non-Suppressed and Suppressed Index Viral Load, Comparing Pandemic Periods to Pre-pandemic
| Pre-Pandemic | Pandemic, Lab Closed | Pandemic, Lab Open | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Non-Suppressed Hazard Ratio | Interaction p-valued | Suppressed Hazard Ratio | Interaction p-valued | Non-Suppressed Hazard Ratio | Interaction p-valued | Suppressed Hazard Ratio | Interaction p-valued | ||
| All Viral Loadsa | 1.0 | 0.57 (0.45, 0.72) | n/a | 0.34 (0.31, 0.37) | n/a | 0.92 (0.77, 1.09) | n/a | 0.73 (0.68, 0.78) | n/a |
| Age | |||||||||
| Age 20–39 | 1.0 | 0.47 (0.31, 0.71) | REF | 0.24 (0.19, 0.32) | REF | 0.83 (0.61, 1.12) | REF | 0.62 (0.52, 0.74) | REF |
| Age 40–59 | 1.0 | 0.63 (0.45, 0.89) | 0.29 | 0.37 (0.33, 0.42) | 0.01 | 0.92 (0.72, 1.20) | 0.57 | 0.73 (0.66, 0.79) | 0.11 |
| Age 60+ | 1.0 | 0.61 (0.38, 1.00) | 0.42 | 0.33 (0.29, 0.39) | 0.05 | 1.04 (0.73, 1.48) | 0.34 | 0.78 (0.71, 0.87) | 0.05 |
| Gender | |||||||||
| Female | 1.0 | 0.59 (0.42, 0.84) | REF | 0.34 (0.29, 0.39) | REF | 1.1 (0.78, 1.32) | REF | 0.79 (0.71. 0.88) | REF |
| Male | 1.0 | 0.56 (0.41, 0.75) | 0.79 | 0.34 (0.30. 0.38) | 0.97 | 0.86 (0.68, 1.08) | 0.34 | 0.70 (0.64, 0.76) | 0.06 |
| Race | |||||||||
| Black | 1.0 | 0.52 (0.40, 0.68) | REF | 0.31 (0.28, 0.34) | REF | 0.91 (0.75, 1.10) | REF | 0.71 (0.66, 0.76) | REF |
| White | 1.0 | 0.85 (0.48, 1.50) | 0.12 | 0.48 (0.40, 0.59) | <0.01 | 0.87 (0.51, 1.47) | 0.87 | 0.85 (0.74, 0.99) | 0.02 |
| Other | 1.0 | 0.80 (0.23, 2.78) | 0.52 | 0.32 (0.20, 0.51) | 0.92 | 1.23 (0.44, 3.42) | 0.57 | 0.65 (0.47, 0.90) | 0.62 |
| Ethnicity | |||||||||
| Non-Hispanic | 1.0 | 0.56 (0.44, 0.71) | REF | 0.34 (0.31, 0.37) | REF | 0.90 (0.75, 1.08) | REF | 0.73 (0.68, 0.78) | REF |
| Hispanic | 1.0 | 1.59 (0.40, 6.36) | 0.15 | 0.30 (0.16, 0.54) | 0.67 | 2.37 (0.67, 8.40) | 0.14 | 0.74 (0.51, 1.08) | 0.94 |
| Substance Useb | |||||||||
| No Recent Smoking | 1.0 | 0.52 (0.35, 0.78) | REF | 0.35 (0.31, 0.40) | REF | 0.85 (0.62, 1.17) | REF | 0.79 (0.73, 0.87) | REF |
| Recent Smoking | 1.0 | 0.50 (0.32, 0.76) | 0.88 | 0.27 (0.23, 0.33) | 0.02 | 1.00 (0.75, 1.34) | 0.44 | 0.66 (0.58, 0.74) | 0.01 |
| No Recent Alcohol Use | 1.0 | 0.51 (0.37, 0.72) | REF | 0.32 (0.29, 0.36) | REF | 0.96 (0.75, 1.22) | REF | 0.75 (0.69, 0.81) | REF |
| Recent Alcohol Use | 1.0 | 0.52 (0.27, 0.98) | 0.98 | 0.35 (0.26, 0.47) | 0.65 | 0.89 (0.56, 1.39) | 0.76 | 0.66 (0.53, 0.83) | 0.30 |
| No Recent Cocaine/Heroin | 1.0 | 0.51 (0.37, 0.71) | REF | 0.34 (0.30, 0.38) | REF | 0.91 (0.72, 1.15) | REF | 0.76 (0.71, 0.82) | REF |
| Recent Cocaine or Heroin | 1.0 | 0.50 (0.26, 0.97) | 0.94 | 0.23 (0.16, 0.33) | 0.04 | 0.98 (0.60, 1.58) | 0.79 | 0.56 (0.44, 0.71) | 0.03 |
| Insurancec | |||||||||
| Private | 1.0 | 0.56 (0.40, 0.77) | REF | 0.26 (0.22, 0.31) | REF | 0.92 (0.72, 1.17) | REF | 0.65 (0.58, 0.73) | REF |
| Non-Private | 1.0 | 0.58 (0.42, 0.81) | 0.85 | 0.37 (0.33, 0.41) | <0.01 | 0.90 (0.70, 1.16) | 0.91 | 0.77 (0.71, 0.83) | 0.03 |
Among Non-Suppressed patients: Index Viral Load > 200 copies/mL. Among Suppressed patients: Index Viral Load ≤200 copies/mL.
Suppressed analysis adjusted for duration of suppression prior to index viral load.
Smoking, Alcohol, Cocaine and Heroin use were obtained from medical record review of provider notes, toxicology screens and treatment referrals, conducted every 6 months by trained abstractors. Data were restricted to abstractions in the year prior to the viral load.
Non-Private Insurance defined as covered by Medicaid, Ryan White, or Uninsured.
P-value for the interaction between patient characteristic and the time period, to determine if there were statistically significant differences in hazard across subgroups.
Time to Next Viral Load for Non-suppressed Patients.
Pre-pandemic, the proportion of non-suppressed viral loads that had a follow-up viral load by 12 months, was 90%. Under pandemic lab-closed conditions, the proportion was 75%, while under pandemic lab-open conditions, it was 88% (Figure 2). Compared to pre-pandemic, the HRs for subsequent viral load collection were 0.56 (95% CI: 0.42, 0.79) during the pandemic lab-closed period and 0.92 (95% CI: 0.76, 1.10) during the pandemic lab-open period (Table 2). Among non-suppressed patients, once the lab re-opened, there were no statistically significant differences (compared to pre-pandemic) in time between viral loads across any patient characteristics.
Figure 2 -. Kaplan-Meier Curve of Time from Non-Suppressed Viral Load to Subsequent Viral Load.
The proportion of non-suppressed viral loads with a follow-up viral load collected after 3, 6, and 12 months, comparing the pre-pandemic, pandemic lab-closed, and pandemic lab-open periods. Cumulative incidences and hazard ratios for subsequent viral load collection in each period are reported.
Loss of Viral Suppression.
Following index viral loads that were suppressed, the 12-month cumulative incidence of a subsequent viral loads that was not suppressed was 4% pre-pandemic, 2% under pandemic lab-closed conditions, and 4% under the pandemic lab-open conditions (Supplemental Figure 2). The corresponding hazard ratios compared to pre-pandemic were 0.37 (95% CI: 0.33, 0.41) and 1.04 (95% CI: 0.98, 1.10) respectively.
Time from Non-Suppressed to (i) Initial Suppression or (ii) Re-Suppression:
There were 70 patients who enrolled into the cohort as non-suppressed. The cumulative incidence of initial suppression at 12 months was 66% pre-pandemic, 15% under pandemic lab-closed conditions, and 74% under pandemic lab-open conditions (Supplemental Figure 3). The corresponding HRs for initial suppression compared to pre-pandemic were 0.18 (95% CI: 0.02, 1.29) for pandemic lab-closed conditions and 1.37 (95% CI: 0.61, 3.11) for pandemic lab-open conditions. Among established previously suppressed patients, there were 608 instances of loss of suppression across 352 individuals. The 12-month cumulative incidence of re-suppression was 85% pre-pandemic, 46% under pandemic lab-closed conditions, and 70% under pandemic lab-open conditions (Supplemental Figure 4). The corresponding HRs for re-suppression compared to pre-pandemic were 0.38 (95% CI: 0.25, 0.59) for pandemic lab-closed conditions and 0.68 (95% CI: 0.50, 0.92) for pandemic lab-open conditions.
Discussion
At the onset of the pandemic, when the onsite lab was closed and pandemic restrictions were implemented, the probability of viral load monitoring was drastically reduced. Once the onsite lab reopened and restrictions loosened, non-suppressed viral loads were re-checked at approximately the same rate as pre-pandemic. For suppressed patients, the rate of re-checking viral loads was lower during the pandemic compared to pre-pandemic. The reduction in viral load monitoring of suppressed patients during the pandemic lab-open period compared to the pre-pandemic period was greatest for patients who were Black, had non-private insurance, or had recent substance use.
Longer time between viral loads during the early pandemic is unsurprising. With closure of the onsite lab, patients were reliant on third party commercial laboratories. These off-site labs were heavily strained by increasing COVID-19 testing demands during the pandemic, while limited by reduced in-person staffing as a means of social distancing, on top of pre-existing lab staffing shortages.13,14 Beyond lab-related issues, patients’ ability to reach commercial labs early in the pandemic may have been hindered by disruptions to public transportation during that period.7 Additionally, patients often forwent care during this early pandemic period for reasons including anxiety about contracting SARS-CoV-2, financial concerns related to job security during the pandemic, and insurance loss from job turnover during the recession.15 The onsite lab re-opening coincided somewhat with a better understanding of appropriate COVID-19 precautions and a gradual re-opening of other services (e.g., public transportation and more in-person clinical visits) and thus we might expect to see the associated shortening time between viral loads that was observed. However, our results may also suggest the importance of access to an onsite laboratory for our patients; certainly, when attending an in-person clinic visit, being able to get labs drawn onsite eliminated the need for additional travel and associated risk of COVID-19 exposure.
Once the lab reopened later in the pandemic, the duration between viral load checks shortened to pre-pandemic levels for patients who were not suppressed, but not for patients who are suppressed. The cause of the difference is uncertain. Informally, providers reported being more aggressive about getting viral load testing for patients who were not suppressed. Conversely, for patients who were known to be suppressed, interim guidance early in the pandemic suggested postponing lab monitoring to the extent possible for patients in otherwise stable health to mitigate risk of COVID-19 exposure.10
The longer duration between viral loads for suppressed patients during the pandemic was more prominent among patients who were Black, had non-private insurance, or had a history of substance use disorder. These differences in viral load monitoring frequency may be a reflection of disparities that exist in other dimensions of care, that have disproportionately impacted racial minorities both before and during the COVID-19 pandemic.8,16 Lack of private insurance is likely also a surrogate marker for socioeconomic factors that disproportionately impact PWH and negatively impact access to care, such as access to affordable and reliable transportation.6,8,17 Finally, patients with recent substance use also have historically been at risk for disengagement from care,18 due to factors including financial insecurity, housing insecurity, and the impact of active substance use on adherence.19,20 These groups are also at higher risk for severe COVID-19, thus while this difference could be interpreted as a disparity, it could also be interpreted as the result of efforts to limit COVID-19 exposure.
Despite the prolonged time between viral loads for suppressed patients during the pandemic, the cumulative incidence of loss of suppression was minimal. While this is reassuring, and may suggest that less frequent lab monitoring of suppressed patients is viable, it is crucial to note that our analysis was based on observed viral load values during a period when viral load monitoring was less frequent. We surely missed instances of non-suppressed viral loads that were not measured; the magnitude of this bias is unknown. Additionally, these findings apply to a unique period in patient care, the COVID-19 pandemic, and whether they apply under “normal” conditions merits further study.
The time to observed re-suppression was significantly impacted by the COVID-19 pandemic. This may be an artifact. Remote delivery of care through telemedicine and delayed lab monitoring might mean that patients resuppressed their viral load at the same rate pre-pandemic and during the pandemic, but we were not able to detect that re-suppression as quickly during the pandemic. However, we hypothesize that a more likely explanation is that during the pandemic, the rate of re-suppression was lower due to forgone care (that may have included ART adherence counseling) due to anxiety about COVID-19 exposure, financial insecurity, or loss of insurance during the pandemic recession; barriers to picking up anti-retroviral therapy (ART); and lower ART adherence due to increased anxiety, depression, substance use, or other factors associated with the COVID-19 pandemic. Our clinic worked to mitigate some of these barriers by offering ART delivery during the pandemic, telemedicine clinical visits, and telemedicine psychiatry visits.
Several limitations are worth noting in this analysis. Viral load monitoring requires the provider order a lab test and the patient get the lab test, and it is unclear from our findings alone whether delays in viral load monitoring were most influenced by less provider ordering or lower patient follow-through. In our analysis of loss of suppression and time to re-suppression, we are limited to viral loads that are observed. We may have underestimated the risk of both outcomes. Additionally, all viral load measurements apply only to a point in time, patients’ viral load between checks is unknown. Thus, we may have underestimated the probability of viral non-suppression due to loss of suppression (and subsequent re-suppression) that occurred during a prolonged, unmonitored period. Furthermore, while our two primary analyses had cumulative incidences of 87–88%, the remaining 12–13% censored patients with no follow-up may introduce some degree of informative censoring in our secondary analyses, as patients who are non-suppressed are presumably more likely to be lost to follow-up than those who are suppressed. Our analysis of time to initial suppression was limited by a very small sample size, reflected in wide confidence intervals. Finally, the data were collected during the pandemic, limiting their generalizability and whether they should be used to inform monitoring frequency guidelines moving forward.
Conclusion
In this study, the closure of onsite labs amid early pandemic restrictions resulted in less frequent monitoring of viral loads for both suppressed and non-suppressed patients. Once the lab re-opened and restrictions loosened, frequency of viral load monitoring for non-suppressed patients returned to pre-pandemic levels. Suppressed patients however continued to be monitored less frequently, with some subgroups affected more than others. The longer monitoring interval does not appear to have resulted in significant loss of suppression. However, patients who did experience a loss of viral suppression took longer to resuppress their viral load during the pandemic.
Supplementary Material
Acknowledgements & Funding:
All authors have contributed significantly to this work and have approved of the manuscript as submitted. This work was supported by grants from the National Institutes of Health [T32 AI007291, K24 AA027483, K01 AA028193, K08 MH118094, U01 DA036935 and P30 AI094189]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Conferences: Portions of this work were presented at IDWeek 2021 (Virtual Conference).
References
- 1.Eisinger RW, Dieffenbach CW, Fauci AS. HIV viral load and transmissibility of HIV infection undetectable equals untransmittable. JAMA - J Am Med Assoc. 2019;321(5):451–452. doi: 10.1001/jama.2018.21167 [DOI] [PubMed] [Google Scholar]
- 2.L J, T G, V O, et al. Risk of sexual transmission of human immunodeficiency virus with antiretroviral therapy, suppressed viral load and condom use: a systematic review. CMAJ. 2018;190(46):E1350–E1360. doi: 10.1503/CMAJ.180311 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.MS C, YQ C, M M, et al. Antiretroviral Therapy for the Prevention of HIV-1 Transmission. N Engl J Med. 2016;375(9):830–839. doi: 10.1056/NEJMOA1600693 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Mocroft A, Ledergerber B, Katlama C, et al. Decline in the AIDS and death rates in the EuroSIDA study: an observational study. Lancet. 2003;362(9377):22–29. doi: 10.1016/S0140-6736(03)13802-0 [DOI] [PubMed] [Google Scholar]
- 5.Poorolajal J, Hooshmand E, Mahjub H, Esmailnasab N, Jenabi E. Survival rate of AIDS disease and mortality in HIV-infected patients: a meta-analysis. Public Health. 2016;139:3–12. doi: 10.1016/j.puhe.2016.05.004 [DOI] [PubMed] [Google Scholar]
- 6.Wood BR, Young JD, Abdel-Massih RC, et al. Advancing Digital Health Equity: A Policy Paper of the Infectious Diseases Society of America and the HIV Medicine Association. Clin Infect Dis. 2021;72(6):913–919. doi: 10.1093/cid/ciaa1525 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Liu L, Miller HJ, Scheff J. The impacts of COVID-19 pandemic on public transit demand in the United States. PLoS One. 2020;15(11):e0242476. doi: 10.1371/JOURNAL.PONE.0242476 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lesko CR, Bengtson AM. HIV and COVID-19: Intersecting Epidemics with Many Unknowns. Am J Epidemiol. 2021;190(1):10–16. doi: 10.1093/aje/kwaa158 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Dawson L, Kates J. Delivering HIV Care Prevention in the COVID Era A National Survey of Ryan White Providers – Issue Brief – 9596 | KFF. Accessed September 1, 2021. https://www.kff.org/report-section/delivering-hiv-care-prevention-in-the-covid-era-a-national-survey-of-ryan-white-providers-issue-brief/
- 10.HHS Antiretroviral and Opportunistic Infections Guidelines Panel. Interim Guidance for COVID-19 and Persons with HIV Archive.; 2020. Accessed November 18, 2021. https://clinicalinfo.hiv.gov/en/guidelines/archived-guidelines/covid-19-and-persons-hiv-interim-guidance
- 11.Moore RD. Understanding the clinical and economic outcomes of HIV therapy: The Johns Hopkins HIV clinical practice cohort. J Acquir Immune Defic Syndr Hum Retrovirology. 1998;17(SUPPL. 1):S38–41. doi: 10.1097/00042560-199801001-00011 [DOI] [PubMed] [Google Scholar]
- 12.Nitkin K. Milestones in a Year of COVID-19 at Johns Hopkins Medicine. Published March 8, 2021. Accessed May 14, 2021. https://www.hopkinsmedicine.org/news/articles/covid-19-timeline-for-johns-hopkins-medicine
- 13.Nuñez-Argote L, Baker DP, Jones AP. Initial Clinical Laboratory Response to COVID-19: A Survey of Medical Laboratory Professionals. Lab Med. 2021;52(4):e115–e124. doi: 10.1093/LABMED/LMAB021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Tsai JM, Tolan NV., Petrides AK, et al. How SARS-CoV-2 Transformed the Clinical Laboratory: Challenges and Lessons Learned. J Appl Lab Med. 2021;6(5):1338–1354. doi: 10.1093/JALM/JFAB034 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Anderson KE, Mcginty EE, Presskreischer R, Barry CL. Reports of Forgone Medical Care Among US Adults During the Initial Phase of the COVID-19 Pandemic. JAMA Netw Open. 2021;4(1):e2034882–e2034882. doi: 10.1001/JAMANETWORKOPEN.2020.34882 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Bailey ZD, Feldman JM, Bassett MT. How Structural Racism Works — Racist Policies as a Root Cause of U.S. Racial Health Inequities. Malina D, ed. N Engl J Med. 2020;384(8):768–773. doi: 10.1056/nejmms2025396 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Pellowski JA, Kalichman SC, Matthews KA, Adler N. A pandemic of the poor: Social disadvantage and the U.S. HIV epidemic. Am Psychol. 2013;68(4):197–209. doi: 10.1037/a0032694 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Rebeiro P, Althoff KN, Buchacz K, et al. Retention Among North American HIV–infected Persons in Clinical Care, 2000–2008. J Acquir Immune Defic Syndr. 2013;62(3):356. doi: 10.1097/QAI.0B013E31827F578A [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Krüsi A, Wood E, Montaner J, Kerr T. Social and structural determinants of HAART access and adherence among injection drug users. Int J Drug Policy. 2010;21(1):4–9. doi: 10.1016/J.DRUGPO.2009.08.003 [DOI] [PubMed] [Google Scholar]
- 20.Knowlton A, Arnsten J, Eldred L, et al. Individual, interpersonal, and structural correlates of effective HAART use among urban active injection drug users. J Acquir Immune Defic Syndr. 2006;41(4):486–492. doi: 10.1097/01.QAI.0000186392.26334.E3 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.