Tekin, Aysun

OBJECTIVES: There is a paucity of literature regarding administrative approvals required for clinical studies during a pandemic. We aimed to evaluate variation in duration of administrative approvals within the Viral Infection and Respiratory illness Universal Study (VIRUS): A Global COVID-19 Registry. DESIGN, SETTING, AND SUBJECTS: Survey analysis of 188 investigators who participated in the VIRUS: COVID-19 registry, a prospective, observational global registry database of 287 sites. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: For each study site approved through December 8, 2020, we assessed the duration in days: 1) from institutional review board (IRB) submission to IRB approval, 2) from IRB approval to Research Electronic Data Capture (REDCap) access, 3) from REDCap access to first patient data entry in REDCap, and 4) total duration from IRB submission to first patient data entry in REDCap. Analysis of variance and Wilcoxon rank-sum test were used to compare time durations. Of 287 sites, 188 sites (United States = 155, non-United States = 33) provided complete administrative data. There was considerable variability in duration from IRB submission to first patient data entry with median (interquartile range) of 28 days (16–50 d), with differences not significantly different by country (United States: 30 [17–50] vs non-United States: 23 d [8–46 d]; p = 0.08) or previous “multisite trial experience” (experienced: 27 [15–51] vs not experienced: 29 d [13–47 d]; p = 0.67). The U.S. sites had a higher proportion of female principal investigators (n = 77; 50%), compared with non-U.S. sites (n = 7; 21%; p = 0.002). Non-U.S. sites had a significantly shorter time to first patient data entry after REDCap access: 7 (1–28) versus 3 days (1–6 d) (p = 0.02). CONCLUSIONS: In this Society of Critical Care Medicine global VIRUS: COVID-19 Registry, we identified considerable variability in time from IRB submission to first patient data entry with no significant differences by country or prior multicenter trial experience. However, there was a significant difference between US and non-U.S. sites in the time from REDCap access to first data entry.

Pulmonary auscultation is essential for detecting abnormal lung sounds during physical assessments, but its reliability depends on the operator. Machine learning (ML) models offer an alternative by automatically classifying lung sounds. ML models require substantial data, and public databases aim to address this limitation. This systematic review compares characteristics, diagnostic accuracy, concerns, and data sources of existing models in the literature. Papers published from five major databases between 1990 and 2022 were assessed. Quality assessment was accomplished with a modified QUADAS-2 tool. The review encompassed 62 studies utilizing ML models and public-access databases for lung sound classification. Artificial neural networks (ANN) and support vector machines (SVM) were frequently employed in the ML classifiers. The accuracy ranged from 49.43% to 100% for discriminating abnormal sound types and 69.40% to 99.62% for disease class classification. Seventeen public databases were identified, with the ICBHI 2017 database being the most used (66%). The majority of studies exhibited a high risk of bias and concerns related to patient selection and reference standards. Summarizing, ML models can effectively classify abnormal lung sounds using publicly available data sources. Nevertheless, inconsistent reporting and methodologies pose limitations to advancing the field, and therefore, public databases should adhere to standardized recording and labeling procedures.

AbstractBackgroundPneumocystis pneumonia (PCP) is a growing concern as the immunocompromised population expands. Current laboratory approaches are limited. This systematic review aimed to evaluate metagenomic next-generation sequencing (MNGS) tests’ performance in detecting PCP.MethodsFive databases were searched through December 19, 2022, to identify original studies comparing MNGS with clinically diagnosed PCP. To assess the accuracy, symmetric hierarchical summary receiver operating characteristic models were used.ResultsEleven observational studies reporting 1442 patients (424 with PCP) were included. Six studies focused exclusively on recipients of biologic immunosuppression (none with HIV-associated immunosuppression). Six were exclusively on bronchoalveolar lavage, while 1 was on blood samples. The sensitivity of MGNS was 0.96 (95% CI, 0.90–0.99), and specificity was 0.96 (95% CI, 0.92–0.98), with negative and positive likelihood ratios of 0.02 (95% CI, 0.01–0.05) and 19.31 (95% CI, 10.26–36.36), respectively. A subgroup analysis of studies exclusively including bronchoalveolar lavage (BAL) and blood samples demonstrated a sensitivity of 0.94 (95% CI, 0.78–0.99) and 0.93 (95% CI, 0.80–0.98) and a specificity of 0.96 (95% CI, 0.88–0.99) and 0.98 (95% CI, 0.76–1.00), respectively. The sensitivity analysis on recipients of biologic immunosuppression showed a sensitivity and specificity of 0.96 (95% CI, 0.90–0.98) and 0.94 (95% CI, 0.84–0.98), respectively. The overall confidence in the estimates was low.ConclusionsDespite the low certainty of evidence, MNGS detects PCP with high sensitivity and specificity. This also applies to recipients of biologic immunosuppression and tests performed exclusively on blood samples without the need for BAL. Further studies are required in individuals with HIV-associated immunosuppression.

Background Although corticosteroids have become the standard of care for patients with coronavirus disease-2019 (COVID-19) on supplemental oxygen, there is growing evidence of differential treatment response. This study aimed to evaluate if there was an association between biomarker-concordant corticosteroid treatment and COVID-19 outcomes. Methods This registry-based cohort study included adult COVID-19 hospitalized patients between January 2020 and December 2021 from 109 institutions. Patients with available C-reactive protein (CRP) levels within 48 h of admission were evaluated. Those on steroids before admission, stayed in the hospital for <48 h, or were not on oxygen support were excluded. Corticosteroid treatment was biomarker-concordant if given with high baseline CRP ≥150 mg/L or withheld with low CRP (<150 mg/L) and vice-versa was considered discordant (low CRP with steroids, high CRP without steroids). Hospital mortality was the primary outcome. Sensitivity analyses were conducted using varying CRP level thresholds. The model interaction was tested to determine steroid effectiveness with increasing CRP levels. Results Corticosteroid treatment was biomarker-concordant in 1778 (49%) patients and discordant in 1835 (51%). The concordant group consisted of higher-risk patients than the discordant group. After adjusting for covariates, the odds of in-hospital mortality were significantly lower in the concordant group than the discordant (odds ratio [95% confidence interval (C.I.)] = 0.71 [0.51, 0.98]). Similarly, adjusted mortality difference was significant at the CRP thresholds of 100 and 200 mg/L (odds ratio [95% C.I.] = 0.70 [0.52, 0.95] and 0.57 [0.38, 0.85], respectively), and concordant steroid use was associated with lower need for invasive ventilation for 200 mg/L threshold (odds ratio [95% C.I.] = 0.52 [0.30, 0.91]). In contrast, no outcome benefit was observed at CRP threshold of 50. When the model interaction was tested, steroids were more effective at reducing mortality as CRP levels increased. Conclusion Biomarker-concordant corticosteroid treatment was associated with lower odds of in-hospital mortality in severe COVID-19.