Unsupervised machine learning predicts future sexual behaviour and sexually transmitted infections among HIV-positive men who have sex with men.

Machine learning is increasingly introduced into medical fields, yet there is limited evidence for its benefit over more commonly used statistical methods in epidemiological studies. We introduce an unsupervised machine learning framework for longitudinal features and evaluate it using sexual behaviour data from the last 20 years from over 3'700 participants in the Swiss HIV Cohort Study (SHCS). We use hierarchical clustering to find subgroups of men who have sex with men in the SHCS with similar sexual behaviour up to May 2017, and apply regression to test whether these clusters enhance predictions of sexual behaviour or sexually transmitted diseases (STIs) after May 2017 beyond what can be predicted with conventional parameters. We find that behavioural clusters enhance model performance according to likelihood ratio test, Akaike information criterion and area under the receiver operator characteristic curve for all outcomes studied, and according to Bayesian information criterion for five out of ten outcomes, with particularly good performance for predicting future sexual behaviour and recurrent STIs. We thus assess a methodology that can be used as an alternative means for creating exposure categories from longitudinal data in epidemiological models, and can contribute to the understanding of time-varying risk factors.

Decreasing Incidence and Determinants of Bacterial Pneumonia in People With HIV: The Swiss HIV Cohort Study.

BACKGROUND: Bacterial pneumonia is a leading reason for hospitalization among people with HIV (PWH); however, evidence regarding its drivers in the era of potent antiretroviral therapy is limited. METHODS: We assessed risk factors for bacterial pneumonia in the Swiss HIV Cohort Study using marginal models. We further assessed the relationship between risk factors and changes in bacterial pneumonia incidence using mediation analysis. RESULTS: We included 12927 PWH with follow-ups between 2008 and 2018. These patients had 985 bacterial pneumonia events during a follow-up of 100779 person-years. Bacterial pneumonia incidence significantly decreased from 13.2 cases/1000 person-years in 2008 to 6.8 cases/1000 person-years in 2018. Older age, lower education level, intravenous drug use, smoking, lower CD4-cell count, higher HIV load, and prior pneumonia were significantly associated with higher bacterial pneumonia incidence. Notably, CD4 cell counts 350-499 cells/µL were significantly associated with an increased risk compared to CD4 ≥ 500 cells/µL (adjusted hazard ratio, 1.39; 95% confidence interval, 1.01-1.89). Decreasing incidence over the last decade can be explained by increased CD4-cell counts and viral suppression and decreased smoking frequency. CONCLUSIONS: Improvements in cascade of care of HIV and decrease in smoking may have mediated a substantial decrease in bacterial pneumonia incidence.

Assessing the drivers of syphilis among men who have sex with men in Switzerland reveals a key impact of screening frequency: A modelling study.

Over the last decade, syphilis diagnoses among men-who-have-sex-with-men (MSM) have strongly increased in Europe. Understanding the drivers of the ongoing epidemic may aid to curb transmissions. In order to identify the drivers of syphilis transmission in MSM in Switzerland between 2006 and 2017 as well as the effect of potential interventions, we set up an epidemiological model stratified by syphilis stage, HIV-diagnosis, and behavioral factors to account for syphilis infectiousness and risk for transmission. In the main model, we used 'reported non-steady partners' (nsP) as the main proxy for sexual risk. We parameterized the model using data from the Swiss HIV Cohort Study, Swiss Voluntary Counselling and Testing center, cross-sectional surveys among the Swiss MSM population, and published syphilis notifications from the Federal Office of Public Health. The main model reproduced the increase in syphilis diagnoses from 168 cases in 2006 to 418 cases in 2017. It estimated that between 2006 and 2017, MSM with HIV diagnosis had 45.9 times the median syphilis incidence of MSM without HIV diagnosis. Defining risk as condomless anal intercourse with nsP decreased model accuracy (sum of squared weighted residuals, 378.8 vs. 148.3). Counterfactual scenarios suggested that increasing screening of MSM without HIV diagnosis and with nsP from once every two years to twice per year may reduce syphilis incidence (at most 12.8% reduction by 2017). Whereas, increasing screening among MSM with HIV diagnosis and with nsP from once per year to twice per year may substantially reduce syphilis incidence over time (at least 63.5% reduction by 2017). The model suggests that reporting nsP regardless of condom use is suitable for risk stratification when modelling syphilis transmission. More frequent screening of MSM with HIV diagnosis, particularly those with nsP may aid to curb syphilis transmission.

Cohort-Derived Machine Learning Models for Individual Prediction of Chronic Kidney Disease in People Living With Human Immunodeficiency Virus: A Prospective Multicenter Cohort Study.

BACKGROUND: It is unclear whether data-driven machine learning models, which are trained on large epidemiological cohorts, may improve prediction of comorbidities in people living with human immunodeficiency virus (HIV). METHODS: In this proof-of-concept study, we included people living with HIV in the prospective Swiss HIV Cohort Study with a first estimated glomerular filtration rate (eGFR) >60 mL/minute/1.73 m2 after 1 January 2002. Our primary outcome was chronic kidney disease (CKD)-defined as confirmed decrease in eGFR ≤60 mL/minute/1.73 m2 over 3 months apart. We split the cohort data into a training set (80%), validation set (10%), and test set (10%), stratified for CKD status and follow-up length. RESULTS: Of 12 761 eligible individuals (median baseline eGFR, 103 mL/minute/1.73 m2), 1192 (9%) developed a CKD after a median of 8 years. We used 64 static and 502 time-changing variables: Across prediction horizons and algorithms and in contrast to expert-based standard models, most machine learning models achieved state-of-the-art predictive performances with areas under the receiver operating characteristic curve and precision recall curve ranging from 0.926 to 0.996 and from 0.631 to 0.956, respectively. CONCLUSIONS: In people living with HIV, we observed state-of-the-art performances in forecasting individual CKD onsets with different machine learning algorithms.

Telomere Length, Traditional Risk Factors, Factors Related to Human Immunodeficiency Virus (HIV) and Coronary Artery Disease Events in Swiss Persons Living With HIV.

BACKGROUND: Leukocyte telomere length (TL) shortens with age and is associated with coronary artery disease (CAD) events in the general population. Persons living with human immunodeficiency virus (HIV; PLWH) may have accelerated atherosclerosis and shorter TL than the general population. It is unknown whether TL is associated with CAD in PLWH. METHODS: We measured TL by quantitative polymerase chain reaction (PCR) in white Swiss HIV Cohort Study participants. Cases had a first CAD event during 1 January 2000 to 31 December 2017. We matched 1-3 PLWH controls without CAD events on sex, age, and observation time. We obtained univariable and multivariable odds ratios (OR) for CAD from conditional logistic regression analyses. RESULTS: We included 333 cases (median age 54 years; 14% women; 83% with suppressed HIV RNA) and 745 controls. Median time (interquartile range) of TL measurement was 9.4 (5.9-13.8) years prior to CAD event. Compared to the 1st (shortest) TL quintile, participants in the 5th (longest) TL quintile had univariable and multivariable CAD event OR = 0.56 (95% confidence interval [CI], .35-.91) and OR = 0.54 (95% CI, .31-.96). Multivariable OR for current smoking was 1.93 (95% CI, 1.27-2.92), dyslipidemia OR = 1.92 (95% CI, 1.41-2.63), and for recent abacavir, cumulative lopinavir, indinavir, and darunavir exposure was OR = 1.82 (95% CI, 1.27-2.59), OR = 2.02 (95% CI, 1.34-3.04), OR = 3.42 (95% CI, 2.14-5.45), and OR = 1.66 (95% CI, 1.00-2.74), respectively. The TL-CAD association remained significant when adjusting only for Framingham risk score, when excluding TL outliers, and when adjusting for CMV-seropositivity, HCV-seropositivity, time spent with detectable HIV viremia, and injection drug use. CONCLUSIONS: In PLWH, TL measured >9 years before, is independently associated with CAD events after adjusting for multiple traditional and HIV-related factors.

Antimicrobial Prophylaxis for Postoperative Urinary Tract Infections in Transurethral Resection of Bladder Tumors: A Systematic Review and Meta-Analysis.

PURPOSE: The administration of antimicrobial prophylaxis for postoperative urinary tract infections following transurethral resection of bladder tumors is controversial. We aimed to systematically review evidence on the potential effect of antimicrobial prophylaxis on postoperative urinary tract infections and asymptomatic bacteriuria. MATERIALS AND METHODS: We conducted a systematic search in Embase®, Medline® and the Cochrane Central Register of Controlled Trials. Randomized controlled trials and nonrandomized controlled trials assessing the effect of any form of antimicrobial prophylaxis in patients with transurethral resection of bladder tumors on postoperative urinary tract infections or asymptomatic bacteriuria were included. Risk of bias was assessed using RoB 2.0 or the Newcastle-Ottawa Scale. Fixed and random effects meta-analyses were conducted. As a potential basis for a scoping review, we exploratorily searched Medline for risk factors for urinary tract infections after transurethral resection of bladder tumors. The protocol was registered on PROSPERO (CRD42019131733). RESULTS: Of 986 screened publications, 7 studies with 1,725 participants were included; the reported effect sizes varied considerably. We found no significant effect of antimicrobial prophylaxis on urinary tract infections: the pooled odds ratio of the random effects model was 1.55 (95% CI 0.73-3.31). The random effects meta-analysis examining the effect of antimicrobial prophylaxis on asymptomatic bacteriuria showed an OR of 0.43 (0.18-1.04). Risk of bias was moderate. Our exploratory search identified 3 studies reporting age, preoperative pelvic radiation, preoperative hospital stay, duration of operation, tumor size, preoperative asymptomatic bacteriuria and pyuria as risk factors for urinary tract infections following transurethral resection of bladder tumors. CONCLUSIONS: We observed insufficient evidence supporting routine antimicrobial prophylaxis in patients undergoing transurethral resection of bladder tumors for the prevention of postoperative urinary tract infections; our findings may inform harmonization of international guidelines.

High incidence of urinary tract infections after photoselective laser vaporisation of the prostate: a risk factor analysis of 665 patients.

PURPOSE: Although photoselective laser vaporisation of the prostate (PVP) is a recognised alternative to transurethral resection in treating benign prostatic obstruction, there is limited data on the incidence and determinants of postoperative urinary tract infections (UTI). We assessed patients subjected to PVP, evaluating incidence and potential determinants of postoperative UTIs. MATERIALS AND METHODS: Consecutive patients undergoing PVP between April 2010 and August 2018 were candidates for this retrospective cohort study. The primary outcome measure was microbiologically confirmed postoperative UTI. We fitted uni- and multi-variable Cox models to identify potential risk factors. RESULTS: Among the 665 included patients, 20% developed postoperative UTIs. The overall incidence rate per 100 patient-days was 0.65 (95% confidence interval [CI] 0.55-0.77). Risk factors for postoperative UTIs were end-stage renal failure (adjusted hazard ratio [aHR] = 14.10, 95% CI 2.08-64.58; p = 0.001) and presence of at least one of the following factors in the 3 months preceding PVP: (i) placement of urinary catheter, (ii) bacteriuria, (iii) UTI, or (iv) antimicrobial treatment (composite aHR = 1.99, 95% CI 1.22-3.24; p < 0.001). There was no apparent association between choice or duration of antimicrobial prophylaxis and incident UTIs. CONCLUSIONS: Our analysis revealed a high incidence of UTIs after PVP and served to identify certain preoperative risk factors. Neither the choice of antimicrobial regimen nor its duration affected the incidence of UTIs. Prolonged antimicrobials proved to be disproportionately high, warranting further scrutiny in randomised controlled trials.

Frequent Door Openings During Cardiac Surgery Are Associated With Increased Risk for Surgical Site Infection: A Prospective Observational Study.

BACKGROUND: Preliminary studies that analyzed surrogate markers have suggested that operating room (OR) door openings may be a risk factor for surgical site infection (SSI). We therefore aimed to estimate the effect of OR door openings on SSI risk in patients undergoing cardiac surgery. METHODS: This prospective, observational study involved consecutive patients undergoing cardiac surgery in 2 prespecified ORs equipped with automatic door-counting devices from June 2016 to October 2017. Occurrence of an SSI within 30 days after cardiac surgery was our primary outcome measure. Respective outcome data were obtained from a national SSI surveillance cohort. We analyzed the relationship between mean OR door opening frequencies and SSI risk by use of uni- and multivariable Cox regression models. RESULTS: A total of 301 594 OR door openings were recorded during the study period, with 87 676 eligible door openings being logged between incision and skin closure. There were 688 patients included in the study, of whom 24 (3.5%) developed an SSI within 30 days after surgery. In uni- and multivariable analysis, an increased mean door opening frequency during cardiac surgery was associated with higher risk for consecutive SSI (adjusted hazard ratio per 5-unit increment, 1.49; 95% confidence interval, 1.11-2.00; P = .008). The observed effect was driven by internal OR door openings toward the clean instrument preparation room. CONCLUSIONS: Frequent door openings during cardiac surgery were independently associated with an increased risk for SSI. This finding warrants further study to establish a potentially causal relationship between OR door openings and the occurrence of SSI.

Early versus late onset bloodstream infection during neutropenia after high-dose chemotherapy for hematologic malignancy.

PURPOSE: The length of neutropenia has a significant impact on the incidence of bloodstream infection (BSI) in cancer patients, but limited information is available about the pathogen distribution in late BSI. METHODS: Between 2002 and 2014, BSI episodes in patients with neutropenia receiving chemotherapy for hematologic malignancies were prospectively identified by multicenter, active surveillance in Germany, Switzerland and Austria. The incidence of first BSI episodes, their microbiology and time to BSI onset during the first episode of neutropenia of 15,988 patients are described. RESULTS: The incidence rate of BSI episodes was 14.7, 8.7, and 4.7 per 1000 patient-days in the first, second, and third week of neutropenia, respectively. BSI developed after a median of 5 days of neutropenia (interquartile range [IQR] 3-10 days). The medium duration of neutropenia to BSI onset was 4 days in Escherichia coli (IQR 3-7 days), Klebsiella spp. (2-8 days), and Staphylococcus aureus (3-6 days). In contrast, BSI due to Enterococcus faecium occurred after a median of 9 days (IQR 6-14 days; p < 0.001 vs. other BSI). Late onset of BSI (occurring after the first week of neutropenia) was also observed for Stenotrophomonas maltophilia (12 days, IQR 7-17 days; p < 0.001), and non-albicans Candida spp. (13 days, IQR 8-19 days; p < 0.001). CONCLUSIONS: Over the course of neutropenia, the proportion of difficult to treat pathogens such as E. faecium, S. maltophilia, and Candida spp. increased. Among other factors, prior duration of neutropenia may help to guide empiric antimicrobial treatment in febrile neutropenia.

Contribution of specific pathogens to bloodstream infection mortality in neutropenic patients with hematologic malignancies: Results from a multicentric surveillance cohort study.

Bloodstream infection (BSI) remains a serious complication in patients with hematologic malignancies and neutropenia. The risk factors for mortality after BSI and the contributions of BSI pathogens to mortality remain incompletely understood. We evaluated first BSI among adult neutropenic patients undergoing high-dose chemotherapy for hematologic malignancies in the setting of (a) an early disease stage of autologous (auto-HSCT) or allogeneic (allo-HSCT) hematopoietic stem cell transplantation or (b) for acute leukemia. Risk factors for intensive care admission and all-cause mortality were analyzed by multivariable logistic regression 7 and 30 days after onset of the first BSI in the first neutropenic episode. Between 2002 and 2015, 9080 patients met the study inclusion criteria, and 1424 (16%) developed BSIs, most of them during the first week of neutropenia. Mortality during neutropenia within 7 days and 30 days after BSI onset was 2.5% and 5.1%, respectively, and differed considerably between BSI pathogens. Both 7-day and 30-day mortalities were highest for Pseudomonas aeruginosa BSI (16.7% and 26.7%, respectively) and lowest for BSI due to coagulase-negative Staphylococcus spp. (CoNS) and Streptococcus spp. BSI pathogens were independently associated with 7-day mortality included P aeruginosa, Klebsiella spp., Enterobacter spp., Serratia spp., and enterococci. Only gram-negative BSI and candidemia were associated with admission to intensive care within 7 days after BSI onset. BSI caused by P aeruginosa continues to carry a particularly poor prognosis in neutropenic patients. The unexpected association between enterococcal BSI and increased mortality needs further study.

Extended-spectrum ß-lactamase-producing Enterobacterales in diverse foodstuffs: a prospective, longitudinal study in the city of Basel, Switzerland.

Background: The involvement of non-human-to-human transmission of extended-spectrum ß-lactamase-producing Enterobacterales (ESBL-PE) remains elusive. Foodstuffs may serve as reservoirs for ESBL-PE and contribute to their spread. Aim: We aimed to systematically investigate the presence and spatiotemporal distribution of ESBL-PE in diverse unprocessed foodstuffs of different origin purchased in a central European city. Methods: Chicken and green (herbs, salad, sprouts, vegetables) samples were collected monthly for two consecutive years, from June 2017 to June 2019, from large supermarket chains and small local food retailers, representing all ten postcode areas of the City of Basel (Switzerland), and the kitchen of the University Hospital Basel (Basel, Switzerland). After enrichment, presumptive ESBL-PE were isolated by selective culture methods and identified by Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. ESBL production was confirmed by phenotypic testing. Results: Among 947 food samples, 14.8% were positive for ESBL-PE isolate/s belonging to eight different ESBL-producing bacterial species. Escherichia coli and Serratia fonticola were predominant across samples (9 and 2%, respectively). Higher ESBL-PE prevalence was observed in chicken (25.9%) than in green (3.8%) samples (p < 0.001). Among greens, ESBL-PE were most frequently isolated from sprouts (15.2%). High ESBL-PE species diversity was observed among chicken samples, with E. coli as predominant (17.6%). ESBL-producing Enterobacter cloacae was detected among different greens. Yet, ESBL-producing Klebsiella pneumoniae was predominant in sprouts (12.1%). In total, 20.5% of samples from organic farming and 14.2% of samples from conventionally raised animals harbored an ESBL-producing isolate. Detection of ESBL-PE across samples differed between organic and non-organic when stratified by food source (p < 0.001), particularly among greens (12.5% organic, 2.4% conventional). High proportion of organic chicken samples was positive for ESBL-E. coli (33.3%), while the detection of several species characterized the conventional chicken samples. No significant differences in ESBL-PE frequences were detected between national (13.4%) and international samples (8.0%) (p = 0.122). Instead, differences were observed between regions of food production and countries (p < 0.001). No significant differences were found when comparing the proportion of ESBL-PE positive samples across districts, shop sizes and the hospital kitchen. The percentage of ESBL-PE positive samples did not differ monthly across the two-year sampling period (p = 0.107). Conclusion: Our findings indicate moderate dissemination of ESBL-PE in foodstuffs, especially between chicken products and sprouts. Chicken meat represents a source for several ESBL-producing Enterobacterales, especially E. coli, while greens are more prone to carry ESBL-K. pneumoniae and E. cloacae. We disclose the importance of food type, food production system and production origin when assessing the risk of contamination with different ESBL-PE species.

Spatiotemporal dissemination of ESBL-producing Enterobacterales in municipal sewer systems: a prospective, longitudinal study in the city of Basel, Switzerland.

Background: The contribution of community and hospital sources to the transmission of extended-spectrum ß-lactamase producing Enterobacterales (ESBL-PE) remains elusive. Aim: To investigate the extent of community dissemination and the contribution of hospitals to the spread of ESBL-PE by exploring their spatiotemporal distribution in municipal wastewater of the central European city of Basel. Methods: Wastewater samples were collected monthly for two consecutive years throughout Basel, Switzerland, including 21 sites across 10 postcode areas of the city collecting either community wastewater (urban sites, n = 17) or community and hospital wastewater (mixed sites, n = 4). Presumptive ESBL-PE were recovered by selective culture methods. Standard methodologies were applied for species identification, ESBL-confirmation, and quantification. Results: Ninety-five percent (477/504) of samples were positive for ESBL-PE. Among these isolates, Escherichia coli (85%, 1,140/1,334) and Klebsiella pneumoniae (11%, 153/1,334) were most common. They were recovered throughout the sampling period from all postcodes, with E. coli consistently predominating. The proportion of K. pneumoniae isolates was higher in wastewater samples from mixed sites as compared to samples from urban sites, while the proportion of E. coli was higher in samples from urban sites (p = 0.003). Higher numbers of colony forming units (CFUs) were recovered from mixed as compared to urban sites (median 3.2 × 102 vs. 1.6 × 102 CFU/mL). E. coli-counts showed moderate correlation with population size (rho = 0.44), while this correlation was weak for other ESBL-PE (rho = 0.21). Conclusion: ESBL-PE are widely spread in municipal wastewater supporting that community sources are important reservoirs entertaining the spread of ESBL-PE. Hospital-influenced abundance of ESBL-PE appears to be species dependent.

External Validation of the Revised Pretransplant Assessment of Mortality Score in Allogeneic Hematopoietic Cell Transplantation: A Cohort Study.

Pretransplant risk scores such as the revised Pretransplant Assessment of Mortality (rPAM) score help to predict outcome of patients receiving allogeneic hematopoietic cell transplantation (allo-HCT). Since the rPAM has not been validated externally in a heterogeneous patient population with different diseases, we aimed to validate the rPAM score in a real-world cohort of allo-HCT patients. A total of 429 patients were included receiving their first allo-HCT from 2008 to 2015. The predictive capacity of the rPAM score for 4-year overall survival (OS), nonrelapse mortality (NRM), and cumulative incidence of relapse (CIR) after allo-HCT was evaluated. Moreover, we evaluated the impact of the rPAM score for OS and used uni- and multivariable analyses to identify patient- and transplant-related predictors for OS. In rPAM score categories of <17, 17-23, 24-30, and >30, the OS probability at 4 years differed significantly with 61%, 36%, 26%, and 10%, respectively (P < 0.0001). In contrast to CIR, the NRM increased significantly in patients with higher rPAM scores (P < 0.001). Regarding the OS, the rPAM score had an area under the receiver operating characteristics curve of 0.676 (95% confidence interval [CI], 0.625-0.727) at 4 years. In the multivariable analysis, the rPAM score was associated with OS-independently of conditioning regimens (adjusted hazard ratio per 1-unit increase, 1.10; 95% CI, 1.06-1.10; P < 0.001). Additionally, forced expiratory volume in 1 second and the disease risk index were the components of the rPAM significantly associated with outcome. In our large real-world cohort with extended follow-up, the rPAM score was validated as an independent predictor of OS in patients with hematologic disorders undergoing allo-HCT.

A comparison of general and disease-specific machine learning models for the prediction of unplanned hospital readmissions.

Unplanned hospital readmissions are a burden to patients and increase healthcare costs. A wide variety of machine learning (ML) models have been suggested to predict unplanned hospital readmissions. These ML models were often specifically trained on patient populations with certain diseases. However, it is unclear whether these specialized ML models-trained on patient subpopulations with certain diseases or defined by other clinical characteristics-are more accurate than a general ML model trained on an unrestricted hospital cohort. In this study based on an electronic health record cohort of consecutive inpatient cases of a single tertiary care center, we demonstrate that accurate prediction of hospital readmissions may be obtained by general, disease-independent, ML models. This general approach may substantially decrease the cost of development and deployment of respective ML models in daily clinical routine, as all predictions are obtained by the use of a single model.

Handrub dispensers per acute care hospital bed: a study to develop a new minimum standard.

BACKGROUND: Accessibility to alcohol-based handrub (ABHR) dispenser is crucial to improve compliance to hand hygiene (HH), being offered as wall-mounted dispensers (ABHR-Ds), and/or pocket bottles. Nevertheless, information on the distribution and density of ABHR-Ds and their impact on HH have hardly been studied. Institutions such as the World Health Organisation or the Centers for Disease Control and Prevention do not provide guidance. The Robert-Koch-Institute (RKI) from Germany recommends an overall density of > 0.5 dispensers per patient bed. We aimed to investigate current conditions in hospitals to develop a standard on the minimal number of ABHR-D. METHODS: Between 07 and 09/2019, we applied a questionnaire to 178 hospitals participating in the Swissnoso National Surveillance Network to evaluate number and location of ABHR-Ds per bed in acute care hospitals, and compared the data with consumption and compliance with HH. RESULTS: 110 of the 178 (62%) hospitals provided data representing approximately 20,000 hospital beds. 83% hospitals provided information on both the total number of ABHR-Ds and patient beds, with a mean of 2.4 ABHR-Ds per bed (range, 0.4-22.1). While most hospitals (84%) had dispensers located at the room entrance, 47% reported also locations near or at the bed. Additionally, pocket-sized dispensers (100 mL) are available in 97% of hospitals. CONCLUSIONS: Swiss hospitals provide 2.4 dispensers per bed, much more than governmental recommendation. The first study on the number of ABHR-Ds in hospitals may help to define a minimal standard for national and international recommendations.

Diagnostic Accuracy of Imaging Findings in Pleural Empyema: Systematic Review and Meta-Analysis.

Computed tomography (CT) diagnosis of empyema is challenging because current literature features multiple overlapping pleural findings. We aimed to identify informative findings for structured reporting. The screening according to inclusion criteria (P: Pleural empyema, I: CT C: culture/gram-stain/pathology/pus, O: Diagnostic accuracy measures), data extraction, and risk of bias assessment of studies published between 01-1980 and 10-2021 on Pubmed, Embase, and Web of Science (WOS) were performed independently by two reviewers. CT findings with pooled diagnostic odds ratios (DOR) with 95% confidence intervals, not including 1, were considered as informative. Summary estimates of diagnostic accuracy for CT findings were calculated by using a bivariate random-effects model and heterogeneity sources were evaluated. Ten studies with a total of 252 patients with and 846 without empyema were included. From 119 overlapping descriptors, five informative CT findings were identified: Pleural enhancement, thickening, loculation, fat thickening, and fat stranding with an AUC of 0.80 (hierarchical summary receiver operating characteristic, HSROC). Potential sources of heterogeneity were different thresholds, empyema prevalence, and study year.

Bayesian networks to identify potential high-risk multimorbidity and intervention clusters in inpatients: an explorative data mining study.

AIMS OF THE STUDY: Based on large sets of routine hospital data from inpatient cases, we aimed to explore multimorbidity and intervention clusters showing high risks for in-hospital mortality and unplanned readmissions using data-driven analytical methods. METHODS: We performed an explorative, historical cohort study of consecutive inpatient cases at a tertiary care centre with an integrated platform for routine healthcare data in Switzerland. From January 2012 through to December 2017, all inpatients aged ≥18 years at hospital admission were eligible for study inclusion. We predefined all-cause in-hospital death and unplanned hospital readmission as co-primary outcomes. In a first step, we explored and visualised multimorbidity and intervention clusters using mutual information analysis. In a subsequent step, we trained multi-layer Bayesian networks to identify clusters associated with in-hospital death and/or unplanned hospital readmission. RESULTS: Among 190,837 inpatient cases, 7994 unique diagnoses and 6639 interventions were routinely recorded during the six-year study period. Based on the mutual information analysis, we identified 32 multimorbidity clusters and 24 intervention clusters – of which several were directly related to in-hospital mortality and/or unplanned readmission in the subsequent Bayesian network analysis. CONCLUSIONS: Bayesian network analysis may be used as a tool to mine large healthcare databases in order to explore intervention targets for quality improvement programmes. However, the resulting associations should be substantiated in consecutive investigations using specific causal models. (Trial registration no EKNZ 2016-02128.).