Caroline Sabin’s research while affiliated with National Institute for Health and Care Excellence and other places

Objective In this study, we explore changes in the demographic and clinical characteristics of pregnant people living with HIV, and their post‐partum HIV outcomes between 2000 and 2018. Methods We described pregnancies resulting in a live birth from the linked UK CHIC/ISOSS dataset in three calendar periods (2000–2006; 2007–2012; 2013–2018). Thereafter, we explored median CD4 count change and viral rebound from delivery to 12 months post‐partum. Results In total, 4341 pregnancies were included. Maternal age increased from 31 (IQR: 28–35) years in 2000–2006 to 34 (IQR: 30–37) years in 2013–2018. Those conceiving in the most recent period had been diagnosed with HIV for longer (2000–2006: 3.0 years to 2013–2018: 7.5 years), had a higher median CD4 count (431–583 cells/mm³), and median nadir CD4 count (219–260 cells/mm³); they were also more likely to have initiated ART prior to estimated conception (70.1%–92.3%), and have a suppressed conception viral loads (VL) (56.6%–82.0%). There was no difference in median CD4 count change over the three calendar periods (2000–2006: +60 [IQR: −44, +179]; 2007–2012: +51 [−45, +172]; 2013–2018: +28 [−100, +175] cells/mm³; p = 0.12). The cumulative proportion of people with viral rebound at 12 months post‐delivery was reduced in 2013–2018 (8.2%) when compared with previous periods (2000–2006: 28.1%; 2007–2012: 19.5%). Conclusion Clinical management of pregnant people has changed over time, resulting in positive trends in this study both within pregnancy and post‐partum. Further work needs to understand what barriers remain for those who do not achieve optimal management of HIV in the post‐partum period.

Background Despite increasing multimorbidity among people with HIV, its impact on health outcomes over time remains uncertain. We explored how distinct multimorbidity patterns affect a broad range of health outcomes over a 3–5-year period. Methods Principal component analysis (PCA) was used to identify multimorbidity patterns at baseline. Burden z-scores were calculated for each individual/pattern at baseline and a follow-up visit, and the differences in scores over time were examined. Participants completed health assessments including questionnaires (physical/mental health (SF-36), depressive symptoms (CES-D and PHQ-9, falls, frailty and healthcare utilisation), cognitive testing and pain mannequins tests. Multivariable regression models assessed associations between changes in morbidity burden z-scores and health outcomes. Results Six multimorbidity patterns were identified in 1073 participants: “ cardiovascular disease” (CVD) , “ sexually transmitted infections” (STIs) , “ metabolic” , “ mental/joint” , “ neurological” , and “ cancer/other” . Subsequent analyses included 793 participants (median [interquartile range; IQR] age 53 [47–59] years; 86% male; 97% on ART) with follow up data. CVD and metabolic burden were associated with specialist appointments (CVD: β = 1.47; metabolic: β = 1.53, p < 0.01) and ED visits (CVD: β = 1.44; metabolic: β = 1.89, p < 0.01), mental/Joint and neurological burden with poorer physical and mental health, frailty and recurrent falls (p < 0.01), and cancer/other burden with higher depressive scores (β = 3.28, p < 0.001), widespread pain (OR = 2.20, p < 0.001), and hospital visits (OR = 2.31, p < 0.001). Conclusion Distinct morbidity patterns differentially affected health outcomes and healthcare utilisation over time, underscoring the need for targeted, integrated care to improve quality of life and address their complex needs.

Background The impact of long-term virological suppression and CD4 count recovery on non-AIDS defining cancers (NADC) is unclear. We determined whether poor immune recovery was associated with incident cancer risk in people with HIV with virological suppression (VS). Methods Participants from the D:A:D and RESPOND collaborations in Europe and Australia who achieved ≥2 years of VS on ART between Dec 1999 and Dec 2022 were included. Follow-up was from baseline (date of VS for two years) until the earliest of a first cancer event, virological failure, final follow-up, or administrative censoring date. Multivariable Poisson regression was used to assess associations between cancer incidence (overall, AIDS-defining cancer (ADC), NADC, infection-related, infection-unrelated) and time-updated CD4 count stratified by pre-ART nadir CD4 counts. Results Overall, 48,343 people with VS were included (median [IQR] baseline age 43 years [37-50], CD4 count 540 cells/µL [380-730], nadir CD4 count 245 cells/µL [121-394], 74% male). There were 1,933 incident cancers, median 6.2 years [2.9-9.5] (incidence rate (IR): 6.43 [95%CI 6.15-6.73]/1000 person-years). Higher time-updated CD4 count was associated with a reduced risk of overall cancer (adjusted incidence rate ratio [aIRR] for time-updated CD4 350-499: 0.45 [95%CI 0.39-0.51]; 500-749: 0.30 [0.27-0.34], and ≥750: 0.26 [0.23-0.30] vs. <350 cells/µL, p<0.0001). There was a significant reduction in all cancers risk by higher time-updated CD4 count regardless of nadir CD4 count, with higher pre-ART nadir CD4 count exhibiting lower risk. Conclusions Despite VS on ART for more than two years, people with poorer immune recovery experience a significantly higher incidence of cancer.

Background The COVID-19 pandemic disproportionately affected people of Black ethnicities, however, there are limited data on the post-acute sequelae of COVID-19 infection in these populations, and none in those with HIV. Methods We conducted a cross-sectional study in people of Black ethnicities with HIV in the UK. Participants were assessed for functional impairment, frailty, respiratory symptoms, anxiety and depression; they were also asked to rate aspects of their physical and mental health on a scale from 1 (poor) to 10 (excellent), both at enrolment and prior to the pandemic. We report associations with COVID-19 history and recovery status. Results We enrolled 183 participants between June 2021 and October 2022, 131 (72%) of whom reported COVID-19. A history of COVID-19 was associated with a reduced ability to carry out usual activities (OR 2.54 [1.03–6.21], p = 0.04), an increase in pain, tiredness and breathlessness, and overall decline in physical health. Of those with a history of COVID-19, 111 (85%) reported to have fully recovered. Those who had not fully recovered reported poorer functional status ( p < 0.001) and had higher generalised anxiety scores ( p = 0.02). Objective measures of physical function were similar in those who reported no COVID-19, COVID-19 with full recovery, and COVID-19 with incomplete recovery. Conclusions In this cohort of Black people with HIV, participants with a history of COVID-19 reported a reduced ability to carry out activities of daily living and various other health issues. Although most people reported full recovery from COVID-19, self-reported limitations in functional status and anxiety were common sequelae.

Objectives: There is no consensus definition for multimorbidity. We explored how different frameworks affect multimorbidity patterns and their associations with patient-reported outcomes using the prospective, observational Pharmacokinetic and clinical Observations in PeoPle over fiftY (POPPY) Study. Methods: Sixty-four conditions were classified into three frameworks: Framework-D (diseases), Framework-DCI (diseases and clinical indicators) and Framework-DCIS (diseases, clinical indicators and symptoms). Principal component analysis (PCA) identified five comparable patterns: Cardiovascular disease (CVD) , Sexually transmitted diseases , Metabolic/AIDS-related , Mental health/Other , and Cancer . A sixth pattern was identified using Framework-D ( Infections/Skin) and Framework-DCI/DCIS ( Cardiometabolic) . Using PCA loadings, burden z-scores were calculated for each individual/pattern, and their associations with functional impairment (Lawton Instrumental Activities of Daily Living <8), hospitalisation and SF-36 physical and mental health scores were assessed using logistic or linear regression. Results: The analyses included 1073 people with HIV (median [interquartile range; IQR] age 52 [47 - 59] years; 85% male; 97% on ART). Clinical indicators and symptoms were correlated with the CVD , Cardiometabolic and Mental health/Other patterns. While differences were marginal, Framework-DCI showed slightly stronger relationships between CVD and functional impairment, hospitalisation and physical health compared to Framework-D. Similarly, Framework-DCIS demonstrated somewhat stronger associations between Metabolic/AIDS-related and Mental health/Other patterns with certain outcomes. Conclusions: The inclusion of clinical indicators and symptoms were associated with some changes in the strength of associations between certain multimorbidity patterns and outcomes. Our findings suggest that their inclusion in multimorbidity frameworks should be guided by the specific research context and question, rather than solely by effect size on patient-important outcomes.

Background Inflammation and innate immune activation are associated with chronic human immunodeficiency virus (HIV) infection, despite effective treatment. Although gut microbiota alterations are linked to systemic inflammation, their relationship with HIV infection the relationships between the gut microbiome, inflammation, and HIV remains unclear. Methods The HIV UPBEAT Coronary Artery Disease sub-study evaluated cardiovascular disease (CVD) in people with and without HIV. Subclinical CVD was assessed using coronary computed tomography angiography (CCTA). Thirty-four biomarkers were measured using quantitative immunoassays. Stool samples underwent 16S rRNA sequencing. Differentially abundant species were identified by analysis of compositions of microbiomes with bias correction (ANCOM-BC) and correlated to biomarkers, diet, and CCTA outcomes using Spearman correlation. Results Among 81 participants (median age, 51 years; 73% male), people with HIV (n = 44) had higher rates of hypercholesterolemia (P < .025). Gut microbiome β-diversity differed significantly by HIV status. Enriched Bifidobacterium pseudocatenulatum, Megamonas hypermegale, and Selenomonas ruminantium correlated with lower plaque burden, while depleted Ruminococcus bromii correlated with higher plaque burden and fat intake. Depleted Bacteroides spp and Alistepes spp correlated with elevated biomarkers (D-dimer, CD40 ligand, C-reactive protein, and interferon-γ). Conclusions Gut microbiota differences in people with HIV were linked to subclinical CVD, diet, and inflammation, highlighting the microbiome’s role in cardiovascular risk in HIV infection.

Background Accurate assessment of cognitive impairment in low‐income settings may require consideration of complex psychosocial variables (PV). Failure to consider the association of PV with biological factors, such as HIV, could lead to false classification of cognitive impairment. We investigated the impact of PV on cognitive performance in people with HIV (PWH) and without in a low‐income area of Cape Town, South Africa. Method 273 (178 PWH) participants were recruited and 185 (145 PWH) followed‐up at 1‐year. We investigated the relationship between comprehensive cognitive testing (7 domains) and 12 PV: 5 current (income, employment, assets, accommodation, mood) and 7 recalled at childhood (assets, quality of education, exposure to trauma and violence, parental education and employment), as well as standard variables typically measured in cognition studies (age, sex, years of education). Univariable and multivariable linear regression models investigated relationships between PV (clustered using principal component analysis) and global cognitive performance (measured by global T‐score). Propensity score modelling adjusted for variables significantly associated with HIV status, to determine the direct effect of HIV status on global T‐score. Result PWH had significantly lower scores than people without HIV in 9/12 PV (ps<.015). At baseline, 8/12 PV significantly predicted global T‐score (ps<.007), which clustered into 3 components reflecting current PV, childhood PV and experience of childhood trauma. In addition to standard variables, residing in a shack remained in multivariable analysis at baseline (β = 2.40, p = .007), and lower childhood PV at follow‐up (β = 0.82, p = .035). Unadjusted, HIV status was associated with a reduction in global T‐score of 3.72 units at baseline and 2.46 at follow‐up. At baseline, after adjustment for standard variables, the HIV association was reduced by 25.8% to 2.76, and additional adjustment for PV reduced by 41.4% to 2.18. At follow‐up, adjustment for standard measures reduced by 41.9% to 1.43 and addition of PV by 56.1% to 1.08. Conclusion PWH in this setting have lower psychosocial indices, both now and in childhood, which are associated with lower cognitive test performance as an adult. This is incompletely mitigated by adjustments for standard variables which could result in misclassification of PWH as cognitively impaired if PV are not taken into account.