The likelihood of severe COVID‐19 outcomes among PLHIV with various comorbidities: a comparative frequentist and Bayesian meta‐analysis approach

Abstract

IntroductionThe SARS‐CoV‐2 virus can currently pose a serious health threat and can lead to severe COVID‐19 outcomes, especially for populations suffering from comorbidities. Currently, the data available on the risk for severe COVID‐19 outcomes due to an HIV infection with or without comorbidities paint a heterogenous picture. In this meta‐analysis, we summarized the likelihood for severe COVID‐19 outcomes among people living with HIV (PLHIV) with or without comorbidities.MethodsFollowing PRISMA guidelines, we utilized PubMed, Web of Science and medRxiv to search for studies describing COVID‐19 outcomes in PLHIV with or without comorbidities up to 25 June 2021. Consequently, we conducted two meta‐analyses, based on a classic frequentist and Bayesian perspective of higher quality studies.Results and discussionWe identified 2580 studies (search period: January 2020–25 June 2021, data extraction period: 1 January 2021–25 June 2021) and included nine in the meta‐analysis. Based on the frequentist meta‐analytical model, PLHIV with diabetes had a seven times higher risk of severe COVID‐19 outcomes (odd ratio, OR = 6.69, 95% CI: 3.03–19.30), PLHIV with hypertension a four times higher risk (OR = 4.14, 95% CI: 2.12–8.17), PLHIV with cardiovascular disease an odds ratio of 4.75 (95% CI: 1.89–11.94), PLHIV with respiratory disease an odds ratio of 3.67 (95% CI: 1.79–7.54) and PLHIV with chronic kidney disease an OR of 9.02 (95% CI: 2.53–32.14) compared to PLHIV without comorbidities. Both meta‐analytic models converged, thereby providing robust summative evidence. The Bayesian meta‐analysis produced similar effects overall, with the exclusion of PLHIV with respiratory diseases who showed a non‐significant higher risk to develop severe COVID‐19 outcomes compared to PLHIV without comorbidities.ConclusionsOur meta‐analyses show that people with HIV, PLHIV with coexisting diabetes, hypertension, cardiovascular disease, respiratory disease and chronic kidney disease are at a higher likelihood of developing severe COVID‐19 outcomes. Bayesian analysis helped to estimate small sample biases and provided predictive likelihoods. Clinical practice should take these risks due to comorbidities into account and not only focus on the HIV status alone, vaccination priorities should be adjusted accordingly.

The severity of COVID-19 across the spectrum of HIV

“…In this study, having diabetes, hypertension, chronic kidney disease, tuberculosis, and malignancy were all associated with an increased risk of in-hospital mortality. In addition, several meta-analyses have showed a high rate of comorbidities in PWH and an increase in risk of severe disease and mortality from COVID-19 as total number of comorbidities increased [44 ▪▪ ,45].…”

Long COVID in people living with HIV

“…Although initial small studies suggested that PLWH did not exhibit worse outcomes during acute SARS-CoV-2 infection [55][56][57][58][59], larger studies have shown that HIV status is a risk factor for severe outcomes even among the virologically suppressed [60][61][62][63][64][65][66][67]. This risk is most pronounced in those with comorbidities, viremia, and/or poor immune reconstitution [63,[68][69][70][71][72]. Second, PLWH commonly exhibit metabolic comorbidities such as diabetes and obesity, as well as cardiovascular disease [73][74][75][76][77][78][79][80][81], which are also risk factors for Long COVID.…”

Response to “Methodological issues in the designing and reporting of frequentist and Bayesian meta‐analysis to assess COVID‐19 outcomes among PLHIV with various comorbidities” by Ram Bajpai, Vivek Verma and Gyan Prakash Singh

“…Our colleagues Bajpai et al. have submitted a Letter to the Editor regarding our recently published meta‐analysis “The likelihood of severe COVID‐19 outcomes among PLHIV with various comorbidities: a comparative frequentist and Bayesian meta‐analysis approach [ 1 ].” We appreciate their feedback and request for clarification and would like to respond to their letter.…”

“…stressed the fact that their conditional predictive ordinates approach [ 4 ], compared to methods described by Gelfand and Ghosh [ 5 ], only evaluates the posterior predictive probability density but does not simulate from it, and both methods relied on applying posterior predictive distribution. Also, in our sensitivity analysis, we checked the prior predictive cumulative distribution function (CDF, as stated in Figure S5 of the Supporting Information of [ 1 ]). The CDF is a probability, and the probability density function (PDF) is the derivative of the CDF, and the predictive distribution function is a probability as well.…”