Supplemental Digital Content is Available in the Text.
Key Words: HIV, lymphoma, intensive care, critical care, outcomes, survival
Abstract
Background:
Patients with lymphoma may require intensive care (ICU) because of disease- or treatment-related complications. The lymphoma–HIV interaction complicates management, but whether outcomes are worse in these patients, when critically ill, is unclear. A retrospective observational cohort study reviewed outcomes of patients admitted to ICU, subsequent 5-year survival, and prognostic factors.
Setting:
Academic ICU at the UK National Centre for HIV Malignancy.
Methods:
Records between 2007 and 2020 identified the following cohorts: HIV lymphoma, lymphoma alone, HIV alone, and patients without HIV/lymphoma. Patient demographics, lymphoma characteristics, ICU admission data, and survival outcomes were collected. Five-year survival outcomes were analyzed for the lymphoma cohorts. ICU outcomes were analyzed for all cohorts. Descriptive statistics summarized baseline characteristics and outcomes. Multivariate regression identified factors associated with ICU mortality.
Results:
Of 5929 patients admitted to the ICU, 63 had HIV lymphoma and 43 had lymphoma alone. Survival to ICU discharge was 71% and 72%, respectively. Adjusted log-odds ratio for ICU survival was significantly better in the comparator cohort. ICU survival between the HIV lymphoma and lymphoma-alone cohorts was not significantly different. Adjusted 5-year survival was not significantly different between lymphoma cohorts. Factors independently associated with a worse ICU survival prognosis were emergency admissions, Acute Physiology and Chronic Health Evaluation II score, initial lactate, and day requiring level 3 support. Mechanical ventilation and higher Acute Physiology and Chronic Health Evaluation II scores were independent risk factors for worse 5-year survival in the lymphoma cohorts.
Conclusions:
ICU outcomes and 5-year survival rates of patients with lymphoma were unaffected by HIV status, revealing favorable outcomes in patients with HIV-related lymphoma admitted to the ICU.
INTRODUCTION
Non-Hodgkin lymphoma (NHL) and Hodgkin lymphoma (HL) are common HIV-related malignancies in high-income countries and remain the major causes of mortality.1 The introduction of combined antiretroviral therapy (ART) in the 1990s improved the prophylaxis for opportunistic infections. Furthermore, the evolution of lymphoma treatment in the HIV population has led to improved survival of these patients,2 from less than 20% in the pre-ART era to up to 80% currently.3 The current standard treatment for patients with HIV lymphoma is intensive full-dose chemotherapy regimens, identical to those used for their HIV-negative counterparts.
Patients with lymphoma are at a high risk of critical illness owing to the aggressive nature of the malignancy and toxicities associated with chemotherapy. The prevalence of lymphoma in HIV-infected patients admitted to the ICU has been reported in the literature. Among the HIV-ICU population, Sowah et al4 reported NHL in 11 of 318 (3.5%) patients, Xiao et al5 reported lymphoma (NHL/HL not specified) in 3 of 122 (2.4%) patients, and Kanitkar et al6 described NHL in 11 of 221 (5%) patients, and HL in 1 of 221 (0.45%) patients. Although the widespread use of ART has improved short-term outcomes for patients with HIV admitted to the ICU7 and HIV serostatus itself is no longer a criterion for refusal to ICU admission, there are limited data specifically on ICU outcomes for patients with HIV-related lymphoma. One multicenter study in France that included all patients with HIV admitted to the ICU reported hospital mortality rates of 30.2% with AIDS-defining cancers (predominantly NHL) and 45.4% with non-AIDS-defining cancers, compared with a mortality rate of 12.4% in those without cancer.8
These patients have a unique immune profile owing to the intrinsic immunosuppression of lymphoma and its treatment, combined with varying levels of HIV-related immunodeficiency. Potential drug–drug interactions between ART and chemotherapeutic agents and overlapping toxicities increase patient risk. Whether HIV serostatus and the aforementioned considerations influence short-term outcomes for patients with lymphoma in the critical care setting is unclear, as are the long-term outcomes for ICU survivors.
This study aimed to compare the short-term ICU and 5-year survival rates of patients with HIV lymphoma and patients with non-HIV lymphoma admitted to the ICU of an Academic London Hospital, incorporating the National Centre for HIV Malignancy, and part of Imperial College London between 2007 and 2020. Thus, we sought to determine the differences in and predictors of short- and long-term outcomes in patients admitted to the ICU with lymphoma, and whether they were associated with HIV.
METHODS
This retrospective observational cohort study was performed. Electronic health records from an ICU database and separate HIV medicine and cancer databases were used to identify patients admitted to the ICU of an academic tertiary care referral hospital for HIV disease (the National Centre for HIV Malignancy) between January 1, 2007, and November 1, 2020. Referrals were made in line with established and integrated National Health Service care pathways. The medicosurgical ICU has staffing levels and expertise similar to any general ICU, with the addition of regular input from specialist multidisciplinary teams as part of the clinical pathway. This ICU has approximately 480 admissions per year, with approximately 19–20 admissions per year for patients with HIV, and 4–5 admissions per year for patients with HIV-associated malignancy. The following 4 cohorts were identified: HIV and lymphoma, lymphoma alone, HIV alone, and general ICU admission without HIV or lymphoma (comparator cohort).
The inclusion criterion was an age >16 years. Patients were excluded if they had conflicting data between the different databases used for data collection, had not had lymphoma for more than 5 years before ICU admission, or developed a new diagnosis of lymphoma 2 months after ICU admission (based on the expectation that lymphoma diagnosis would be made within this period after ICU admission). Readmissions were excluded to avoid duplication.
Data on patient demographics, lymphoma subtype, HIV status, indication for ICU admission [defined as part of the Intensive Care National Audit & Research Centre case mix data set9], severity of illness scores, organ failure, duration of ICU stay, and ICU survival outcomes were collected. The objectives of this study were to analyze in-ICU mortality across all 4 patient cohorts as well as-5-year survival outcomes for lymphoma patients with and without HIV infection. Definitions of the collected data are provided in Additional File 1, Supplemental Digital Content, http://links.lww.com/QAI/C354.
As this study involved the analysis of routinely collected, nonidentifiable clinical audit data, ethical approval and patient consent were waived by the local research and development office as per the United Kingdom policy framework for health and social care.10
Statistics
Descriptive statistics summarizing the baseline characteristics and outcomes were obtained. Multivariate regression was used to identify the factors associated with ICU mortality.
Two predominant outcomes were modeled: in-ICU mortality and 5-year survival.
After data extraction, missing data were imputed through multiple imputation by chained equation using classification and regression trees, culminating in 20 datasets of complete data. Continuous variables were standardized to a mean of 0 and a SD of 1 before further analysis. The models were thus fitted to each dataset independently, and estimates were pooled using a variance-aware algorithm11 (see the multiple imputation by chained equation package pool function documentation). Data analysis was performed using R version 4.3.1.12
For the ICU mortality models, multiple univariate models were fitted with the outcome variable of in-ICU mortality (see Additional File 2, Supplemental Digital Content, http://links.lww.com/QAI/C355). Variables with a median P < 0.05 across all imputed datasets were included in the multivariable analysis.13 A random intercept model was subsequently fitted with the upper grouping according to the patient diagnosis (comparator, HIV-, HIV-associated lymphoma, or non-HIV-associated lymphoma). A generalized linear model with a binomial distribution and logit link was fitted for each complete dataset. The mean coefficients and distributions are also presented.
To analyze the factors associated with 5-year mortality, HIV-associated lymphoma and non-HIV-associated lymphoma were compared. For each variable, a univariate Cox proportional hazards model was fitted, and variables with a median P < 0.0513 across all imputed datasets were subjected to a multivariate Cox proportional hazards model. The multivariate Cox proportional hazards model was then stratified according to the lymphoma type. Kaplan–Meier survival curves were plotted and the groups were compared using a log-rank test for dichotomous covariates.
RESULTS
The patient characteristics of the cohorts are given in Table 1 and Additional Files 3–5, Supplemental Digital Content, http://links.lww.com/QAI/C356, http://links.lww.com/QAI/C357, http://links.lww.com/QAI/C358 (variables with >20% missing data were excluded; see Additional File 6, Supplemental Digital Content, http://links.lww.com/QAI/C359).
TABLE 1.
Patient Characteristics
| Overall | Patient Groups | P | ||||
| Comparator | HIV | HIV Lymphoma | Lymphoma | |||
| Number of patients | 5929 | 5614 | 209 | 63 | 43 | |
| Age | <0.001 | |||||
| Mean (SD) | 59.62 (19.35) | 60.14 (19.44) | 48.48 (13.41) | 46.48 (11.51) | 66.21 (17.49) | |
| Gender (%) | <0.001 | |||||
| Female | 2852 (48.1) | 2795 (49.8) | 28 (13.4) | 15 (23.8) | 14 (32.6) | |
| Male | 3046 (51.4) | 2789 (49.7) | 180 (86.1) | 48 (76.2) | 29 (67.4) | |
| Missing data | 31 (0.5) | 30 (0.5) | 1 (0.5) | 0 (0.0) | 0 (0.0) | |
| Body Mass Index | <0.001 | |||||
| Mean (SD) | 26.37 (7.41) | 26.53 (7.50) | 23.53 (4.60) | 22.84 (4.94) | 24.10 (5.02) | |
| Ethnicity (%) | <0.001 | |||||
| Asian | 341 (5.8) | 323 (5.8) | 10 (4.8) | 5 (7.9) | 3 (7.0) | |
| Black | 404 (6.8) | 350 (6.2) | 35 (16.7) | 16 (25.4) | 3 (7.0) | |
| Mixed | 92 (1.6) | 86 (1.5) | 3 (1.4) | 2 (3.2) | 1 (2.3) | |
| White | 3771 (63.6) | 3579 (63.8) | 128 (61.2) | 36 (57.1) | 28 (65.1) | |
| Other | 858 (14.5) | 836 (14.9) | 11 (5.3) | 4 (6.3) | 7 (16.3) | |
| Missing data | 463 (7.8) | 440 (7.8) | 22 (10.5) | 0 (0.0) | 1 (2.3) | |
| Frailty score | 0.059 | |||||
| Mean (SD) | 1.66 (1.99) | 1.67 (2.00) | 1.36 (1.59) | 1.37 (1.52) | 2.36 (2.27) | |
Patient characteristics related to patient demographics. Analysis of variance was performed to compare all 4 cohorts, where a P < 0.05 indicated a statistically significant difference between the groups.
Figure 1 shows a consort diagram of the patient analysis.
FIGURE 1.
Consort chart demonstrating the distribution of patient cohorts.
Demographic Data
Table 1 shows the patient characteristics and demographic information. There was a significant difference in age between the cohorts, with mean ages of 60, 49, 47, and 66 years (P < 0.001) in the comparator, HIV, HIV lymphoma, and lymphoma cohorts, respectively. There was a greater proportion of male patients in the HIV, HIV lymphoma, and lymphoma cohorts than in the comparator group (86%, 76%, and 67%, respectively, vs. 50%).
Additional File 3, Supplemental Digital Content, http://links.lww.com/QAI/C356, shows the organ support and physiologic characteristics of patients admitted to the ICU. Acute Physiology and Chronic Health Evaluation II (APACHE II; range 0–71, increasing severity) scores were greater in the HIV, HIV lymphoma, and lymphoma cohorts than in the comparator cohort (mean 20, 22, and 21, respectively, vs. 15; P < 0.001). They also had higher emergency admission rates (86%–98% vs. 74%; P < 0.001). The HIV cohort had higher rates of intubation than the comparator and lymphoma-only cohorts (32%–34% vs. 23% and 19%, respectively; P < 0.001). The lymphoma cohort was found to have a higher lactate concentration at admission than the HIV only and comparator cohorts (3.2–4.3 vs. 2.4 and 2.2, respectively; P < 0.001), and the number of organs supported on the first day of ICU admission was significantly different; the HIV lymphoma group had the highest number of organs supported on day 1. The primary system affected, resulting in ICU admission, was most commonly the cardiovascular and respiratory systems across all cohorts. However, the gastrointestinal system was more common in the comparator group (30% vs. 8%–19% in the other cohorts), while hematologic or immunologic system-related admission indications were more common in the HIV lymphoma cohort (21% vs. 0.4%–3% in the other cohorts). Mean level 3 support days were highest in the HIV lymphoma cohort at 7 days, compared with 4, 1, and 3 (P = 0.001) days in the HIV, lymphoma, and comparator cohorts, respectively.
Additional File 4, Supplemental Digital Content, http://links.lww.com/QAI/C357, shows the characteristics of the patients with lymphoma, including histology and staging. Burkitt lymphoma was more prevalent in the HIV lymphoma cohort than in the lymphoma cohort (35% vs. 2%; P < 0.001). In the HIV lymphoma group, 92% of the patients were staged as 4 B, compared with 19% in the lymphoma-alone group. However, staging data were missing in 47% of the lymphoma cohort.
Outcomes
Additional File 5, Supplemental Digital Content, http://links.lww.com/QAI/C358, shows the unadjusted ICU and hospital discharge outcomes across all 4 cohorts. The overall ICU survival rate was 88%, with 76% of patients confirmed to be discharged from the hospital alive. ICU survival rates were greater in the comparator cohort than in the HIV, HIV lymphoma, and lymphoma cohorts (88% vs. 79%, 71%, and 72%, respectively; P < 0.001). This pattern was also observed for hospital survival (77% vs. 66%, 44%, and 58%, respectively; P < 0.001). Additional File 6, Supplemental Digital Content, http://links.lww.com/QAI/C359, shows a table of baseline comorbidities collected from patients starting in 2015 (when the Intensive Care National Audit & Research Centre data collection changes were implemented) across all 4 cohorts. These data were excluded from the analyses because approximately 50% of them were missing.
Figure 2 shows the variables and their estimates as odds ratios for in-ICU mortality. Each point per variable represents the estimated effect of the variable on in-unit mortality and spread (represented as a box plot), thus representing the uncertainty in the estimate of that variable given the imputations of missing data. Emergency admission, APACHE II score, initial lactate concentration, and number of level 3 days were significantly associated with worse ICU outcomes. The number of basic cardiovascular support days was significantly associated with improved ICU outcomes and less severe disease. This was also the case if the primary system affected at admission to the ICU was gastroenterologic, endocrine, or genitourinary; this could be attributed to presenting conditions associated with improved survival. The lymphoma subtypes were not included in this multivariate analysis because of the small sample size for each subtype, which resulted in wide estimation errors. The adjusted log-odds ratio for ICU survival showed a statistically significant improvement in ICU survival in the comparator cohort. Although not significantly different (P = 0.25), the HIV-associated lymphoma cohort seemed to favor a trend toward survival more than the lymphoma-only cohort.
FIGURE 2.
Unit outcome multivariable effect size. Asterisks represent statistically significant difference (P < 0.05).
Figure 3 shows the adjusted Kaplan–Meier survival plot of the 5-year outcome data, demonstrating nonsignificant differences between HIV-associated lymphoma and non-HIV-associated lymphoma in patients admitted to the ICU.
FIGURE 3.
Kaplan–Meier survival plot generated from the multivariable Cox-proportional hazards model demonstrating significant overlap between the 2 groups. The blue and red shaded areas represent the 95% confidence intervals for the lymphoma and HIV lymphoma groups, respectively.
Data related to 5-year survival originated from the patient's interaction with health care systems (eg, phlebotomy services, outpatient clinics, or inpatient admissions). The dataset was derived from interactions in which deceased patients were automatically updated to be deceased. The alive cohort was updated based on the date they were last seen to interact with the health care system and were then right censored.
Predictors of Outcome
Figure 4 shows the results of the Cox hazard model multivariate analysis for the 5-year mortality between the HIV lymphoma and non-HIV lymphoma cohorts. Each variable was estimated for each imputed dataset, resulting in 20 estimates for each pooled variable. In the HIV-lymphoma group (Fig. 4A), the severity of critical illness, as estimated by APACHE II scores, was significantly associated with worse outcomes at 5 years (P = 0.01). In contrast, in the lymphoma-only group (Fig. 4B), we observed a survival benefit for prolonged level 3 days despite worst survival in the high severity of illness and intubated subcohort of patients. This likely represents a group of patients who responded to treatment but suffered from tumor lysis syndrome and required renal replacement therapy and likely represents a treatable disease. However, similar to the HIV-lymphoma group, high severity of illness is still associated with increased 5-year mortality.
FIGURE 4.
Factors influencing the 5-year mortality split by HIV status. A, shows that in the HIV lymphoma cohort, intubation and severity of critical illness affect long-term outcomes, whereas in the lymphoma-only group (B), we observe a survival benefit for prolonged level 3 days, suggesting that patients who require renal replacement therapy for tumor lysis syndrome ultimately have a strong survival benefit owing to treatable disease. *Represents statistically significant difference (P < 0.05).
Additional Files 7 and 8, Supplemental Digital Content, http://links.lww.com/QAI/C360 and http://links.lww.com/QAI/C361, list the primary reasons for patients being admitted to the ICU in the HIV lymphoma and non-HIV lymphoma cohorts. Additional File 9, Supplemental Digital Content, http://links.lww.com/QAI/C362, also lists primary reasons for admission to the ICU from March 2020 (the start of the COVID-19 pandemic in the United Kingdom), in the non-HIV lymphoma cohort, with none being admitted with respiratory failure secondary to COVID-19. None of the patients with HIV lymphoma were admitted to the ICU, from March 2020.
DISCUSSIONS
This retrospective cohort study revealed no difference in 5-year mortality outcomes between patients admitted to the ICU with HIV-associated lymphoma and those admitted to the ICU with non-HIV-associated lymphoma. Moreover, there was no statistically significant difference in ICU outcomes between patients admitted to the ICU with HIV-associated lymphoma and those admitted to the ICU with lymphoma only. However, the non-HIV, nonlymphoma, and comparator groups were associated with increased survival. This study suggests that HIV infection is not an independent risk factor for short-term mortality in patients with lymphoma admitted to the ICU.
One retrospective study of >32,000 patients (excluding elective surgical patients) between 2002 and 2011 from 16 general intensive care units in Scotland reported in-ICU, in-hospital, 30-day, and 12-month mortality rates of 24%, 29%, 31%, and 41%, respectively; these findings are in comparison with our study, which revealed in-ICU and in-hospital mortality rates of 12% and 15.6%, respectively; however, our data included elective surgical admissions.14
Only 1 retrospective study has compared outcomes between patients with HIV and patients with non-HIV lymphoma admitted to the ICU. Moreover, there was no significant difference in overall survival.15 In this study, of the 48 patients, 12 had HIV disease, and overall survival up to a median of 53 months was 15% vs. 17%, and 34% vs. 40% 2-year survival was found for the non-HIV vs. HIV-related groups, respectively.15 This study included a small sample, and the mortality rates were greater than those observed in our study (approximate survival rate of 30% at 5 years in both cohorts). In addition, there may be differences in the sample population, treatment, and ICU eligibility criteria, which limits the generalizability of the results to patients with lymphoma in the United Kingdom. However, our study also showed good agreement with this study, suggesting that there is no long-term, statistically significant difference in outcomes between patients with HIV-associated lymphoma and patients with non-HIV-associated lymphoma admitted to the ICU.
The life expectancy of patients with HIV has improved remarkably since the advent of ART.16 More recent retrospective cohort studies of patients with HIV admitted to the ICU have revealed an in-hospital mortality rate of approximately 17%,4 more than 70% with undetectable viral loads, and reasons for ICU admission not being HIV-specific. This finding is in comparison with our study, which demonstrated an in-hospital mortality rate of at least 23%; however, viral load data were not collected. Morquin et al17 reported 37% ICU and 55% 1-year mortality rates in a French cohort of 82 patients with HIV admitted to the ICU. High viral load, use of vasopressors, and, similar to our study, higher ICU severity scores were poor outcome predictors. ART was protective if started before ICU admission, and was better for long-term outcomes when started in the ICU. In a large Spanish cohort study comparing ICU admissions in 1191 patients with HIV, 700 with HIV and HCV, and 7496 general patients, the incidence of sepsis was the highest (approximately 57%) in the HIV group, with ICU and 30-day mortality risk being the highest in the coinfected group, followed by the HIV group. There was no significant difference at 90 days.18 However, worse outcomes have been reported in 122 patients in China, with a 64%–66% ICU and hospital mortality rate, with more than 50% presenting because of respiratory failure,5 which is comparable with the approximately 31% reported in the HIV cohort in our study. A cohort study of 47 patients with certain HIV-related conditions, such as Kaposi disease, admitted to the ICU demonstrated 30% hospital mortality.19
Outcomes and predictors after lymphoma treatment after admission to the ICU were assessed in 190 patients between 2000 and 2010.20 The ICU, hospital, and 1-year mortality rates were 22%, 37%, and 51%, respectively. A total of 45% of the patients were mechanically ventilated. The predictors of higher risk were age >50 years, poor performance status, higher organ failure severity (Sequential Organ Failure Assessment) score, Burkitt lymphoma cerebral lymphoma, and hemophagocytic syndrome.20 Notably, these patients were admitted and did not deny treatment through the limits of care before admission. Of the 72 patients treated for diffuse large B-cell lymphoma, 45 were admitted to the ICU mainly because of respiratory failure (40%) or septic shock (33%), and the ICU mortality rate was 33%.21 Progression-free survival at 2 years was 32% vs. 60% in a matched cohort of patients treated for diffuse large B-cell lymphoma who did not require ICU admission. Essentially, these studies suggest that ICU admission for patients with lymphoma is associated with a significant mortality rate and poorer medium-term outcomes, whereas predictors are predominantly related to acute physiologic derangement, frailty, and the type and stage of the disease.
Limitations
The limitations of this study are as follows. First, as a retrospective observational study, there will be an inherent level of unmeasured confounding factors that could be resolved by a prospective study. Prudent to this study was the lack of staging data in the non-HIV lymphoma cohort. Furthermore, the sample size was relatively small, which potentially limited the statistical power. A larger sample size would reduce the rate of missing values at random and, therefore, shrink the confidence intervals around the estimates produced by the imputation procedure for missing values, which would facilitate greater confidence in the generalizability of our results to a larger audience. However, this study used the largest sample to directly compare ICU outcomes between patients with HIV-associated lymphoma and those with lymphoma only. Given the small sample size and the potential for confounding variables, multivariate regression analysis was used, with the aim of strengthening the statistical robustness of this study. In addition, because of the incomplete nature of using multiple datasets, imputation techniques were employed. This approach was propagated forward to the uncertainty of each estimate and limited the reliance on a single estimator. There were also inherent variations in ICU admission criteria throughout the course of the study, which may have influenced which patients were considered suitable candidates for ICU admission, as well as the timing of admission to the ICU, both of which can have significant effects on the recorded outcomes. Although this study was performed at a tertiary referral center for HIV, it was a single-center study and is likely not to account for regional variations in care and patient demographics. Other limitations include differences in age between the cohorts and missing data related to the stage of lymphoma, lactate dehydrogenase levels at admission (potential prognosticator), HIV viral load, and patient comorbidities (as shown in Additional File 6, Supplemental Digital Content, http://links.lww.com/QAI/C359), all of which can have key effects on survival outcomes.
These limitations demonstrate the need for caution when generalizing the study outcomes, and further studies are required to confirm the findings of this study.
CONCLUSIONS
There were no statistically significant differences in ICU or 5-year mortality outcomes between patients with HIV-associated lymphoma and patients with non-HIV-associated lymphoma admitted to the ICU. Furthermore, ICU outcomes did not differ between these cohorts and the HIV-only cohort. Severity of illness, as estimated by the APACHE II score, predicted worse 5-year survival outcomes in the patient groups with lymphoma, as did mechanical ventilation in the patient group with non-HIV lymphoma. HIV infection does not influence the short- or long-term survival of patients with lymphoma admitted to the ICU.
Footnotes
The authors have no funding or conflicts of interest to disclose.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (www.jaids.com).
Patient data contain identifiable information and thus, unfortunately, cannot be shared because of ethical, regulatory, and protocol restrictions. The data and code for the simulation and analysis of the visual search are freely available at: https://1drv.ms/f/s!AvRjS-Cmb5zPxDqJtSi1OXL5JqL_?e=TWzNZf.
F.E.-H., A.A.-H., and Sh.S. contributed to the conceptualization, data collection, statistical analysis, and writing of this study. Su.S. and M.B. contributed to the conceptualization, write-up, and review of this study.
Contributor Information
Fouad El-Hibri, Email: fouad.elhibri@nhs.net.
Ahmed Al-Hindawi, Email: ahmed.al-hindawi@nhs.net.
Shivani Singh, Email: shivani.singh2@nhs.net.
Mark Bower, Email: mark.bower@nhs.net.
Suveer Singh, Email: Suveer.singh@imperial.ac.uk.
REFERENCES
- 1.Re A, Cattaneo C, Montoto S. Treatment management of haematological malignancies in people living with HIV. Lancet Haematol. 2020;7:e679–e689. [DOI] [PubMed] [Google Scholar]
- 2.Ramaswami R, Chia G, Dalla Pria A, et al. Evolution of HIV-associated lymphoma over 3 decades. J Acquir Immune Defic Syndr. 2016;72:177–183. [DOI] [PubMed] [Google Scholar]
- 3.Yarchoan R, Uldrick TS. HIV-associated cancers and related diseases. N Engl J Med. 2018;378:1029–1041. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Sowah LA, George N, Doll M, et al. Predictors of in-hospital mortality in a cohort of people living with HIV (PLHIV) admitted to an academic medical intensive care unit from 2009 to 2014: a retrospective cohort study. Medicine (Baltimore). 2022;101:e29750. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Xiao J, Zhang W, Huang Y, et al. Etiology and outcomes for patients infected with HIV in intensive care units in a tertiary care hospital in China. J Med Virol. 2015;87:366–374. [DOI] [PubMed] [Google Scholar]
- 6.Kanitkar T, Dissanayake O, Bakewell N, et al. Changes in short-term (in-ICU and in-hospital) mortality following intensive care unit admission in adults living with HIV: 2000–2019. AIDS. 2023;37:2169–2177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Azoulay É, De Castro N, Barbier F. Critically ill patients with HIV: 40 years later. Chest. 2020;157:293–309. [DOI] [PubMed] [Google Scholar]
- 8.Szychowiak P, Boulain T, Timsit JF, et al. Clinical spectrum and prognostic impact of cancer in critically ill patients with HIV: a multicentre cohort study. Ann Intensive Care. 2023;13:74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Intensive Care National Audit & Research Centre . ICNARC Case Mix Programme Data Collection Manual, Version 4.0.5. London: Intensive Care National Audit & Research Centre; 2023. [Google Scholar]
- 10.UK Policy Framework for Health and Social Care Research. Health Research Authority. Available at: https://www.hra.nhs.uk/planning-and-improving-research/policies-standards-legislation/uk-policy-framework-health-social-care-research/. Accessed January 8, 2024. [Google Scholar]
- 11.Rubin DB. Multiple Imputation for Nonresponse in Surveys. 1st ed. New York: Wiley-Interscience; 1987. [Google Scholar]
- 12.Posit Team. RStudio: Integrated Development Environment for R; 2022. Available at: http://www.posit.co/. Accessed June 7, 2024. [Google Scholar]
- 13.Eekhout I, Van De Wiel MA, Heymans MW. Methods for significance testing of categorical covariates in logistic regression models after multiple imputation: power and applicability analysis. BMC Med Res Methodol. 2017;17:129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Simpson A, Puxty K, McLoone P, et al. Comorbidity and survival after admission to the intensive care unit: a population-based study of 41,230 patients. J Intensive Care Soc. 2021;22:143–151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ferrè C, De Guzmao BM, Morgades M, et al. Lack of impact of human immunodeficiency virus infection on the outcome of lymphoma patients transferred to the intensive care unit. Leuk Lymphoma. 2012;53:1966–1970. [DOI] [PubMed] [Google Scholar]
- 16.Morris A, Masur H, Huang L. Current issues in critical care of the human immunodeficiency virus-infected patient. Crit Care Med. 2006;34:42–49. [DOI] [PubMed] [Google Scholar]
- 17.Morquin D, Le Moing V, Mura T, et al. Short- and long-term outcomes of HIV-infected patients admitted to the intensive care unit: impact of antiretroviral therapy and immunovirological status. Ann Intensive Care. 2012;2:25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Medrano J, Álvaro-Meca A, Boyer A, et al. Mortality of patients infected with HIV in the intensive care unit (2005 through 2010): significant role of chronic hepatitis C and severe sepsis. Crit Care. 2014;18:475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hansen ME, Mangusan R, Lurain K, et al. Characteristics of patients admitted to the ICU with Kaposi sarcoma herpesvirus-associated diseases. AIDS. 2022;36:1969–1978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Algrin C, Faguer S, Lemiale V, et al. Outcomes after intensive care unit admission of patients with newly diagnosed lymphoma. Leuk Lymphoma. 2015;56:1240–1245. [DOI] [PubMed] [Google Scholar]
- 21.Zduniak A, Mihailescu S, Lequesne J, et al. Outcomes after intensive care unit admission in newly diagnosed diffuse large B-cell lymphoma patients: a real-life study. Eur J Haematol. 2021;106:788–799. [DOI] [PubMed] [Google Scholar]