Title: An updated meta-analysis on the association between HIV infection and COVID-19 mortality
Abstract: Recently, Mellor et al.[1] published a paper titled 'Risk of adverse coronavirus disease 2019 outcomes for people living with HIV' in the journal of AIDS. The authors found that people living with HIV had a higher risk of coronavirus disease 2019 (COVID-19) mortality [hazard ratio = 1.95, 95% confidence interval (CI): 1.62–2.34] compared with people without HIV based on a meta-analysis of five studies [1]. This is an extremely interesting study. However, this study had limited sample sizes. To our knowledge, a considerable number of articles on this topic have emerged since then. Therefore, it is necessary to clarify the association between HIV infection and the risk of mortality among COVID-19 patients by a meta-analysis on the basis of updated data. This meta-analysis was reported following the guideline of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) [2]. We systematically search EMBASE, PubMed and Web of Science to identify the potential articles from 15 December 2019 to 12 May 2021. Predefined search terms were applied as follows: 'COVID-19' OR 'SARS-CoV-2' OR 'coronavirus disease 2019' and 'HIV' OR 'AIDS' OR 'human immunodeficiency virus'. All peer-reviewed articles written in English were included if they reported the mortality of COVID-19 patients with and without HIV infection. Accordingly, we excluded case reports, reviews, duplicates, errata, comments and studies with insufficient data. R software (Version 3.6.3) was used to calculate the pooled relative risk using a random-effects meta-analysis model [3]. The heterogeneity was estimated by I2 statistic and Cochran's Q statistic [4,5]. Leave-one-out sensitivity analysis was performed to evaluate the stability of our results [6]. Begg's test and Egger's test were conducted to quantitatively assess the potential publication bias [7–9]. Subgroup analyses were also carried out. Two-sided P values of less than 0.05 were considered a statistically significant result. Initially, 6616 studies were identified through our search strategy. According to the selection criteria, 84 studies with 816 678 cases were eligible for this quantitative meta-analysis. There are 46 studies from Americas (37 from USA, five from Brazil, three from Mexico and one from Latin America), 25 studies from Europe (11 from Spain, six from UK, four from Italy, two from France, one each from Sweden and the Netherlands), six studies from Asia (three from Korea, one each from China, Iran and Yemen), five studies from Africa (two from South Africa, one each from Congo, Ghana and Nigeria) and two studies from international centre. Sixty-three studies are retrospective, 11 are prospective and 10 are other study designs. The basic characteristics of the included studies are summarized in Table 1. Table 1 - Main characteristics of the included studies. Nonsurvivors Survivors Author Country/region Study design Cases Male (%) Age HIV Non-HIV HIV Non-HIV Effect size (95% CI) Aloisio E Italy Retrospective study 409 69 61 (50–73) 0 71 9 329 0.24 (0.01–4.22) Rodríguez-Molinero A Spain Observational cohort 418 56.9 65.4 ± 16.6 1 78 2 337 2.29 (0.07–28.64) Hadi YB USA Retrospective study 50167 45.2 48 20 1585 384 48178 1.33 (0.69–2.51) Berenguer J Spain Retrospective study 4035 61 70 (56–80) 6 1096 20 2840 0.78 (0.31–1.94) Nachega JB Congo Retrospective study 766 65.6 46 (34–58) NR NR NR NR 1.23 (0.30–4.99) Gavin W USA Retrospective study 140 51.4 60 (48–72) 1 17 2 116 3.41 (0.29–39.69) Yadaw ASa USA Cohort 5051 55.4 56.2 6 410 74 4561 1.05 (0.48–2.31) Parker A South Africa Retrospective study 113 39 84 ± 14 6 22 18 67 1.0 (0.4–2.9) Chand S USA Retrospective study 300 60.7 58.2 ± 12.6 2 155 3 140 0.76 (0.26–2.24) Deiana G Italy Retrospective study 1223 40.8 NR NR NR NR NR 1.8 (0.2–13.5) Lamure S France Retrospective study 89 66 67 (19–92) 0 30 2 57 0.38 (0.02–8.11) Bahl A USA Retrospective study 1461 52.7 62.0 (50.0–74.0) 1 326 7 1127 0.55 (0.08–4.00) Wang Z USA Retrospective study 3273 57.3 65 (53–77) 10 732 42 2489 0.81 (0.4–1.62) Tartof SY USA Retrospective study 6916 45 49.1 ± 16.6 0 206 11 6699 1.41 (0.08–24.02) Geretti AM UK Prospective study 47592 57.3 51.75 (40.25–61.42) NR NR NR NR 1.69 (1.15–2.18) Garibaldi BT USA Retrospective study 832 53 63 (49–75) 1 130 8 693 0.67 (0.08–5.37) Chilimuri S USA Retrospective study 375 63 63.0 (52.0–72.0) 14 146 8 207 2.48 (1.02–6.07) Kim D USA Retrospective study 817 54.5 57.1 1 120 21 675 0.27 (0.04–2.01) Miyashita H USA Retrospective study 8912 55.2 52.84 23 1235 138 7516 1.01 (0.69–1.48) Boulle A South Africa Population cohort study 22308 31.6 NR 115 510 3863 17820 2.14 (1.70–2.70) Cortes-Telles A Mexico Prospective study 200 69 55 (41–65) 2 75 2 121 1.61 (0.22–11.7) Mikami T USA Retrospective study 3708 57 66 (55–78) 11 795 53 2849 0.74 (0.39–1.43) Al-Waleedi AA Yemen NR 469 75 47 1 110 0 358 9.73 (0.39–240.62) Miller J USA Retrospective study 2316 51.8 64.5 ± 16.3 8 425 18 1865 1.95 (0.84–4.52) Torres-Macho J Spain Retrospective study 1968 56.4 67 0 325 11 1632 0.22 (0.01–3.71) Chatterjee A The Netherlands Retrospective study 2337 62.9 66.8 ± 14.4 0 568 10 1759 0.15 (0.01–2.52) Jimenez E Spain Retrospective study 1549 57.5 69 (55.0–81.0) 0 295 8 1083 0.22 (0.01–3.75) Nascimento IJBD Brazil Retrospective study 17631 NR NR 4 2324 5 15298 5.27 (1.41–19.62) McPadden J USA Retrospective study 2152 52.1 66.2 (53.7–79.9) 4 301 28 1819 0.86 (0.3–2.48) Chamorro-de-Vega E Spain Retrospective study 130 70.8 64 (53–72) 1 53 0 76 4.29 (0.17–107.32) El-Solh AA USA Retrospective study 1634 95 68.8 ± 13.4 10 465 22 1137 1.11 (0.52–2.37) Cabello A Spain Retrospective study 7061 NR NR 1 900 30 6130 0.23 (0.03–1.67) Vu CA USA Retrospective study 60 66.7 54 (26–87) 0 9 1 50 1.77 (0.07–46.85) Goodman KE USA Retrospective study 66646 52.9 62.83 ± 17.89 36 12352 180 54078 0.88 (0.61–1.25) Brandao Neto RA Brazil Prospective study 506 57.3 60.1 ± 15.1 3 150 4 349 1.75 (0.39–7.89) Supady A International center Retrospective study 127 79 59 (53–66) 1 57 0 69 3.63 (0.14–90.72) Burke E USA Prospective study 166 58 63 ± 14 2 58 3 103 1.18 (0.19–7.29) Silva DL Brazil Retrospective study 212 54.72 61.5 4 103 4 101 0.98 (0.24–4.03) Incerti D USA Retrospective study 13658 51.9 62.0 (49.0–75.0) 12 2151 89 11406 0.71 (0.39–1.31) Correal JCD Brazil Retrospective study 98 51 71 (56–86) 1 43 0 54 3.76 (0.15–94.57) Avila-Nava A Mexico Cohort study 38 73.7 49.3 1 13 2 22 0.85 (0.07–10.27) Vrotsou K Spain Retrospective study 14197 38.9 53.7 ± 17.4 4 761 19 13413 3.71 (1.26–10.93) Karmen-Tuohy S USA Retrospective study 63 91 61.12 NR NR NR NR 1.20 (0.50–2.85) Tang O USA Retrospective study 752 39.9 75.5 NR NR NR NR 0.59 (0.08–4.36) Lorente L Spain Prospective study 143 42 65.0 ± 11.8 0 25 1 117 1.54 (0.06–38.79) King JT Jr USA Retrospective study 3681 92.6 64.8 (53.7–73.4) NR NR NR NR 1.13 (0.59–2.16) Ken-Dror G UK Prospective study 429 56.4 70 ± 18 2 135 0 292 10.79 (0.51–226.36) Lee SG Korea Retrospective study 7339 40.1 47.1 ± 19 1 226 3 7109 5.54 (4.21–7.30) Filardo TD USA Retrospective study 270 67.4 58 (50–67) NR NR NR NR 1.39 (0.47–4.15) Rosenthal N USA Retrospective study 64781 49.3 56.1 ± 19.9 37 7318 215 57211 1.35 (0.95–1.91) Rechtman E USA Retrospective study 8770 54.3 60 (44–72) 16 1098 123 7533 0.89 (0.53–1.51) Bhaskaran K UK Retrospective study NR NR NR NR NR NR NR 2.30 (1.55–3.41) Lundon DJ USA Cross-sectional study 8928 46.2 58.0 ± 18.8 16 1163 124 7625 1.37 (0.91–2.05) Gude-Sampedro F Spain Retrospective study 10454 39.9 58.0 ± 20.0 0 544 9 9901 0.96 (0.06–16.47) Cui N China Retrospective study 836 52.5 64 (51–71) 0 137 2 697 1.01 (0.05–21.25) Kim SW Korea Retrospective study 2254 35.8 58.0 (42.0–70.0) 1 178 5 2070 2.33 (0.27–20.02) Oh TK Korea Population-based cohort 7780 NR NR NR NR NR NR 1.43 (0.11–19.37) Rustgi V USA Retrospective study 403 57.6 62.1 0 90 4 309 0.38 (0.02–7.12) Santeusanio AD USA Retrospective study 38 65.8 53.8 ± 13.6 3 8 0 27 22.65 (1.06–483.68) Hobbs ALV USA Retrospective study 476 55.3 62 (49–71) 0 76 7 393 0.34 (0.02–6.07) Marcolino MS Brazil Retrospective study 2054 52.6 59 (47–71) 8 431 19 1534 1.5 (0.65–3.45) Marmarchi F USA Retrospective study 288 55 63 ± 16 NR NR NR NR 3.873 (0.635–23.620) Orlando V Italy Retrospective study 3497 55.6 NR 11 424 57 3005 1.37 (0.71–2.63) Mohamed NE USA Case series 7624 54.6 46.78 10 846 111 6657 0.71 (0.37–1.36) Hendra H UK Retrospective study 148 56.8 64.13 ± 14.6 3 33 3 109 3.3 (0.64–17.15) Olivas-Martinez A Mexico Prospective study 800 61 51.9 ± 13.9 1 240 9 550 0.25 (0.03–2.02) Harrison SL USA Retrospective study 31461 45.5 50 (35–63) 17 1279 209 29956 1.91 (1.16–3.13) Mollalo Aa USA Retrospective study NR NR NR NR NR NR NR 1.32 (0.55–3.14) Geriatric Medicine Research Collaborative International centre Prospective/Retrospective study 5711 55.2 74 (54–83) 0 1596 16 4099 0.08 (0–1.3) Rodríguez A Spain Prospective study 2022 70.4 64 (55–71) 1 659 4 1358 0.52 (0.06–4.62) Abdollahpour I Iran Retrospective study 513 58.3 58.7 0 60 2 461 1.49 (0.07–31.45) Lorente L Spain Prospective study 53 35.8 NR 0 11 1 41 1.2 (0.05–31.54) Laracy J USA Retrospective study 340 71 58.4 ± 12.7 NR NR NR NR 1.4 (0.7–2.8) Navaratnam AV UK Retrospective study 91541 55.5 71.51 26 28174 123 63218 0.47 (0.31–0.72) Bergman J Sweden Nationwide study 84658 42.8 47.7 NR NR NR NR 0.85 (0.38–1.89) Osibogun A Nigeria Retrospective study 1798 65.1 43 (33–55) 2 71 5 1720 9.69 (1.85–50.81) Bhargava A USA Retrospective study 565 52 64.4 ± 16.2 0 172 3 390 0.32 (0.02–6.29) Mendizabal M 11 Latin American countries Prospective study 2211 60.6 54.3 ± 17.3 NR NR NR NR 1.1 (0.3–3.3) Alser O USA Prospective study 235 66.4 60 (47–70) 3 55 1 176 9.6 (0.98–94.17) Dulery R France Retrospective study 111 63 65 (19–92) 0 24 2 85 0.7 (0.03–15.03) Andreano A Italy Retrospective study 18286 47.6 71 10 3489 55 14732 0.77 (0.39–1.51) Guerson-Gil A USA Retrospective study 3530 55.27 65 (55–76) NR NR NR NR 0.91 (0.55–1.53) Gray WK UK Retrospective study 149725 55.8 60.6 24 32263 185 117253 0.47 (0.31–0.72) Adjei P Ghana Retrospective study 50 50 53 (36–65) 1 5 2 42 4.2 (0.32–55.06) The value of age (years) was presented as mean ± standard deviation (SD) or median with interquartile range (IQR).CI, confidence interval; NR, not clearly reported.aIndicates the combined values. Our meta-analysis demonstrated that COVID-19 patients with positive HIV infection had a significantly increased risk for mortality compared with those without HIV infection (pooled relative risk = 1.23, 95% CI: 1.02–1.48, Fig. 1a). Sensitivity analysis indicated that our results were stable and robust (Fig. 1b). Subgroup analyses revealed that age (pooled relative risk = 1.48, 95% CI: 1.10–1.98 for <60 years old and pooled relative risk = 0.86, 95% CI: 0.72–1.02 for ≥60 years old), regions (pooled relative risk = 5.28, 95% CI: 4.04–6.91 for Asia, pooled relative risk = 2.03, 95% CI: 1.14–3.63 for Africa, pooled relative risk = 1.13, 95% CI: 1.00–1.28 for Americas, and pooled relative risk = 1.00, 95% CI: 0.68–1.48 for Europe) and study design (pooled relative risk = 1.61, 95% CI: 1.21–2.14 for prospective study and pooled relative risk = 1.20, 95% CI: 0.95–1.51 for retrospective study) might modulate the association between HIV infection and COVID-19 mortality. No potential publication bias was observed in neither Egger's test (P = 0.106) nor Begg's test (P = 0.088).Fig. 1: (a) The forest plot indicating the pooled effect size with 95% confidence interval (CI) on the association between HIV infection and the risk of mortality among coronavirus disease 2019 (COVID-19) patients on the basis of 84 studies with 816 678 cases; (b) Leave-one-out sensitivity analysis demonstrated that omitting each study once had no obvious effects on the overall results, which suggests that our results are stable and robust.Our study had several limitations. First, the majority of the included articles were retrospective, well designed studies with more prospective articles are warranted to verify our present findings. Second, most of the included studies were from North America and Europe. Third, data on HIV medications were lacking in most of the included studies, hence we could not address the impact of medications on the association between HIV infection and COVID-19 mortality. Fourth, most of the included studies provided crude effect estimates, thus future study based on adjusted effect estimates should be performed to verify our findings when more data are available. In conclusion, our updated meta-analysis indicated that HIV infection was significantly associated with an increased risk for COVID-19 mortality, which might be modulated by age, regions and study design. We hope that the updated data will contribute to more accurate elaboration and substantiation of the findings reported by Mellor et al.[1]. Acknowledgements We greatly appreciate the following individuals for their kind help with the literature search and data collection: Hongjie Hou, Yang Li, Peihua Zhang, Jian Wu, Xuan Liang, Wenwei Xiao and Ying Wang (All of them are from Department of Epidemiology, College of Public Health, Zhengzhou University, Zhengzhou 450001, China). The data supporting the findings of this study are available within this article. This research was funded by National Natural Science Foundation of China (No. 81973105), Joint Construction Project of Henan Medical Science and Technology Research Plan (No. LHGJ20190679), and Key Scientific Research Project of Henan Institution of Higher Education (No. 21A330008). The funders have no role in collecting data, analysing data, writing manuscript and making submission decision. Yadong Wang, Haiyan Yang and Huifen Feng performed study design. Ruo Feng, Li Shi, Jie Xu and Yadong Wang performed article search and data extraction. Jie Xu, Haiyan Yang and Yadong Wang performed data analysis. Yadong Wang and Ruo Feng wrote this manuscript and performed careful review. The final version of this submitted manuscript was approved by all of the authors. Conflicts of interest There are no conflicts of interest.