A Bayesian approach to estimating COVID-19 incidence and infection fatality rates.

Naive estimates of incidence and infection fatality rates (IFR) of coronavirus disease 2019 suffer from a variety of biases, many of which relate to preferential testing. This has motivated epidemiologists from around the globe to conduct serosurveys that measure the immunity of individuals by testing for the presence of SARS-CoV-2 antibodies in the blood. These quantitative measures (titer values) are then used as a proxy for previous or current infection. However, statistical methods that use this data to its full potential have yet to be developed. Previous researchers have discretized these continuous values, discarding potentially useful information. In this article, we demonstrate how multivariate mixture models can be used in combination with post-stratification to estimate cumulative incidence and IFR in an approximate Bayesian framework without discretization. In doing so, we account for uncertainty from both the estimated number of infections and incomplete deaths data to provide estimates of IFR. This method is demonstrated using data from the Action to Beat Coronavirus erosurvey in Canada.

The Frequency of Routine Blood Sampling and Patient Outcomes Among Maintenance Hemodialysis Recipients.

RATIONALE & OBJECTIVE: Surveillance blood work is routinely performed in maintenance hemodialysis (HD) recipients. Although more frequent blood testing may confer better outcomes, there is little evidence to support any particular monitoring interval. STUDY DESIGN: Retrospective population-based cohort study. SETTING & PARTICIPANTS: All prevalent HD recipients in Ontario, Canada, as of April 1, 2011, and a cohort of incident patients commencing maintenance HD in Ontario, Canada, between April 1, 2011, and March 31, 2016. EXPOSURE: Frequency of surveillance blood work, monthly versus every 6 weeks. OUTCOMES: The primary outcome was all-cause mortality. Secondary outcomes were major adverse cardiovascular events, all-cause hospitalization, and episodes of hyperkalemia. ANALYTICAL APPROACH: Cox proportional hazards with adjustment for demographic and clinical characteristics was used to evaluate the association between blood testing frequency and all-cause mortality. Secondary outcomes were evaluated using the Andersen-Gill extension of the Cox model to allow for potential recurrent events. RESULTS: 7,454 prevalent patients received care at 17 HD programs with monthly blood sampling protocols (n=5,335 patients) and at 8 programs with blood sampling every 6 weeks (n=2,119 patients). More frequent monitoring was not associated with a lower risk for all-cause mortality compared to blood sampling every 6 weeks (adjusted HR, 1.16; 95% CI, 0.99-1.38). Monthly monitoring was not associated with a lower risk for any of the secondary outcomes. Results were consistent among incident HD recipients. LIMITATIONS: Unmeasured confounding; limited data for center practices unrelated to blood sampling frequency; no information on frequency of unscheduled blood work performed outside the prescribed sampling interval. CONCLUSIONS: Monthly routine blood testing in HD recipients was not associated with a lower risk for death, cardiovascular events, or hospitalizations as compared with testing every 6 weeks. Given the health resource implications, the frequency of routine blood sampling in HD recipients deserves careful reassessment.

Bayesian Networks for Risk Prediction Using Real-World Data: A Tool for Precision Medicine.

OBJECTIVE: The fields of medicine and public health are undergoing a data revolution. An increasing availability of data has brought about a growing interest in machine-learning algorithms. Our objective is to present the reader with an introduction to a knowledge representation and machine-learning tool for risk estimation in medical science known as Bayesian networks (BNs). STUDY DESIGN: In this article we review how BNs are compact and intuitive graphical representations of joint probability distributions (JPDs) that can be used to conduct causal reasoning and risk estimation analysis and offer several advantages over regression-based methods. We discuss how BNs represent a different approach to risk estimation in that they are graphical representations of JPDs that take the form of a network representing model random variables and the influences between them, respectively. METHODS: We explore some of the challenges associated with traditional risk prediction methods and then describe BNs, their construction, application, and advantages in risk prediction based on examples in cancer and heart disease. RESULTS: Risk modeling with BNs has advantages over regression-based approaches, and in this article we focus on three that are relevant to health outcomes research: (1) the generation of network structures in which relationships between variables can be easily communicated; (2) their ability to apply Bayes's theorem to conduct individual-level risk estimation; and (3) their easy transformation into decision models. CONCLUSIONS: Bayesian networks represent a powerful and flexible tool for the analysis of health economics and outcomes research data in the era of precision medicine.

The risk of infective endocarditis among people who inject drugs: a retrospective, population-based time series analysis.

BACKGROUND: Infective endocarditis is an increasingly common complication among people who inject drugs. We conducted this study to determine whether the removal of traditional controlled-release oxycodone from the Canadian market would be associated with an increase in the use of hydromorphone and an increased risk of infective endocarditis. METHODS: We conducted a retrospective, population-based time series analysis using the linked health administrative databases of Ontario, Canada. We measured the quarterly risk of admissions for infective endocarditis related to injection drug use and changes in opioid prescription rates from 2006 to 2015. We set the intervention point at the fourth quarter of 2011, when traditional controlled-release oxycodone was removed from the Canadian market. RESULTS: We observed an increase in the risk of admissions for infective endocarditis related to injection drug use during the study period. Before the intervention point, we observed a mean of 13.4 admissions per quarter, and after the intervention, we observed a mean of 35.1 admissions per quarter. However, no significant change in this risk occurred at the intervention point. Rather, the risk of infectious endocarditis appeared to have increased earlier and in parallel with the rise in hydromorphone prescriptions. Hydromorphone represented 16% of all opioid prescriptions at the start of the observation period and 53% by the end. INTERPRETATION: The risk of infective endocarditis related to injection drug use is increasing and is temporally associated with increasing prescriptions for hydromorphone. This relation warrants further exploration.

BUGSnet: an R package to facilitate the conduct and reporting of Bayesian network Meta-analyses.

BACKGROUND: Several reviews have noted shortcomings regarding the quality and reporting of network meta-analyses (NMAs). We suspect that this issue may be partially attributable to limitations in current NMA software which do not readily produce all of the output needed to satisfy current guidelines. RESULTS: To better facilitate the conduct and reporting of NMAs, we have created an R package called "BUGSnet" (Bayesian inference Using Gibbs Sampling to conduct a Network meta-analysis). This R package relies upon Just Another Gibbs Sampler (JAGS) to conduct Bayesian NMA using a generalized linear model. BUGSnet contains a suite of functions that can be used to describe the evidence network, estimate a model and assess the model fit and convergence, assess the presence of heterogeneity and inconsistency, and output the results in a variety of formats including league tables and surface under the cumulative rank curve (SUCRA) plots. We provide a demonstration of the functions contained within BUGSnet by recreating a Bayesian NMA found in the second technical support document composed by the National Institute for Health and Care Excellence Decision Support Unit (NICE-DSU). We have also mapped these functions to checklist items within current reporting and best practice guidelines. CONCLUSION: BUGSnet is a new R package that can be used to conduct a Bayesian NMA and produce all of the necessary output needed to satisfy current scientific and regulatory standards. We hope that this software will help to improve the conduct and reporting of NMAs.

Capturing spatial dependence of COVID-19 case counts with cellphone mobility data.

Spatial dependence is usually introduced into spatial models using some measure of physical proximity. When analysing COVID-19 case counts, this makes sense as regions that are close together are more likely to have more people moving between them, spreading the disease. However, using the actual number of trips between each region may explain COVID-19 case counts better than physical proximity. In this paper, we investigate the efficacy of using telecommunications-derived mobility data to induce spatial dependence in spatial models applied to two Spanish communities' COVID-19 case counts. We do this by extending Besag York Mollié (BYM) models to include both a physical adjacency effect, alongside a mobility effect. The mobility effect is given a Gaussian Markov random field prior, with the number of trips between regions as edge weights. We leverage modern parametrizations of BYM models to conclude that the number of people moving between regions better explains variation in COVID-19 case counts than physical proximity data. We suggest that this data should be used in conjunction with physical proximity data when developing spatial models for COVID-19 case counts.

Assessment of SARS-CoV-2 Seropositivity During the First and Second Viral Waves in 2020 and 2021 Among Canadian Adults.

Importance: The incidence of infection during SARS-CoV-2 viral waves, the factors associated with infection, and the durability of antibody responses to infection among Canadian adults remain undocumented. Objective: To assess the cumulative incidence of SARS-CoV-2 infection during the first 2 viral waves in Canada by measuring seropositivity among adults. Design, Setting, and Participants: The Action to Beat Coronavirus study conducted 2 rounds of an online survey about COVID-19 experience and analyzed immunoglobulin G levels based on participant-collected dried blood spots (DBS) to assess the cumulative incidence of SARS-CoV-2 infection during the first and second viral waves in Canada. A sample of 19 994 Canadian adults (aged ≥18 years) was recruited from established members of the Angus Reid Forum, a public polling organization. The study comprised 2 phases (phase 1 from May 1 to September 30, 2020, and phase 2 from December 1, 2020, to March 31, 2021) that generally corresponded to the first (April 1 to July 31, 2020) and second (October 1, 2020, to March 1, 2021) viral waves. Main Outcomes and Measures: SARS-CoV-2 immunoglobulin G seropositivity (using a chemiluminescence assay) by major geographic and demographic variables and correlation with COVID-19 symptom reporting. Results: Among 19 994 adults who completed the online questionnaire in phase 1, the mean (SD) age was 50.9 (15.4) years, and 10 522 participants (51.9%) were female; 2948 participants (14.5%) had self-identified racial and ethnic minority group status, and 1578 participants (8.2%) were self-identified Indigenous Canadians. Among participants in phase 1, 8967 had DBS testing. In phase 2, 14 621 adults completed online questionnaires, and 7102 of those had DBS testing. Of 19 994 adults who completed the online survey in phase 1, fewer had an educational level of some college or less (4747 individuals [33.1%]) compared with the general population in Canada (45.0%). Survey respondents were otherwise representative of the general population, including in prevalence of known risk factors associated with SARS-CoV-2 infection. The cumulative incidence of SARS-CoV-2 infection among unvaccinated adults increased from 1.9% in phase 1 to 6.5% in phase 2. The seropositivity pattern was demographically and geographically heterogeneous during phase 1 but more homogeneous by phase 2 (with a cumulative incidence ranging from 6.4% to 7.0% in most regions). The exception was the Atlantic region, in which cumulative incidence reached only 3.3% (odds ratio [OR] vs Ontario, 0.46; 95% CI, 0.21-1.02). A total of 47 of 188 adults (25.3%) reporting COVID-19 symptoms during phase 2 were seropositive, and the OR of seropositivity for COVID-19 symptoms was 6.15 (95% CI, 2.02-18.69). In phase 2, 94 of 444 seropositive adults (22.2%) reported having no symptoms. Of 134 seropositive adults in phase 1 who were retested in phase 2, 111 individuals (81.8%) remained seropositive. Participants who had a history of diabetes (OR, 0.58; 95% CI, 0.38-0.90) had lower odds of having detectable antibodies in phase 2. Conclusions and Relevance: The Action to Beat Coronavirus study found that the incidence of SARS-CoV-2 infection in Canada was modest until March 2021, and this incidence was lower than the levels of population immunity required to substantially reduce transmission of the virus. Ongoing vaccination efforts remain central to reducing viral transmission and mortality. Assessment of future infection-induced and vaccine-induced immunity is practicable through the use of serial online surveys and participant-collected DBS.

Stability of diagnostic rate in a cohort of 38,813 colorectal polyp specimens and implications for histomorphology and statistical process control.

This work sought to quantify pathologists' diagnostic bias over time in their evaluation of colorectal polyps to assess how this may impact the utility of statistical process control (SPC). All colorectal polyp specimens(CRPS) for 2011-2017 in a region were categorized using a validated free text string matching algorithm. Pathologist diagnostic rates (PDRs) for high grade dysplasia (HGD), tubular adenoma (TA_ad), villous morphology (TVA + VA), sessile serrated adenoma (SSA) and hyperplastic polyp (HP), were assessed (1) for each pathologist in yearly intervals with control charts (CCs), and (2) with a generalized linear model (GLM). The study included 64,115 CRPS. Fifteen pathologists each interpreted > 150 CRPS/year in all years and together diagnosed 38,813. The number of pathologists (of 15) with zero or one (p < 0.05) outlier in seven years, compared to their overall PDR, was 13, 9, 9, 5 and 9 for HGD, TVA + VA, TA_ad, HP and SSA respectively. The GLM confirmed, for the subset where pathologists/endoscopists saw > 600 CRPS each(total 52,760 CRPS), that pathologist, endoscopist, anatomical location and year were all strongly correlated (all p < 0.0001) with the diagnosis. The moderate PDR stability over time supports the hypothesis that diagnostic rates are amendable to calibration via SPC and outcome data.

Risk of Hospital Encounters With Kidney Stones in Autosomal Dominant Polycystic Kidney Disease: A Cohort Study.

BACKGROUND: There is a perception that patients with autosomal dominant polycystic kidney disease (ADPKD) are more likely to develop kidney stones than the general population. OBJECTIVE: To compare the rate of hospital encounter with kidney stones and the rate of stone interventions between patients with and without ADPKD. DESIGN: Retrospective cohort study. SETTING: Ontario, Canada. PATIENTS: Patients with and without ADPKD who had a prior hospital encounter between 2002 and 2016. MEASUREMENTS: Rate of hospital encounter with kidney stones and rate of stone intervention. METHODS: We used inverse probability exposure weighting based on propensity scores to balance baseline indicators of health between patients with and without ADPKD. We followed each patient until death, emigration, outcomes, or March 31, 2017. We used a Cox proportional hazards model to compare event rates between the two groups. RESULTS: Patients with ADPKD were at higher risk of hospital encounter with stones compared with patients without ADPKD (81 patients of 2094 with ADPKD [3.8%] vs 60 patients of 1902 without ADPKD [3.2%]; 8.9 vs 5.1 events per 1000 person-years; hazard ratio 1.6 [95% CI, 1.3-2.1]). ADPKD was not associated with a higher risk of stone intervention (49 of 2094 [2.3%] vs 47 of 1902 [2.4%]; 5.3 vs 3.9 events per 1000 person-years; hazard ratio 1.2 [95% CI = 0.9-1.3]). LIMITATIONS: We did not have information on kidney stone events outside of the hospital. There is a possibility of residual confounding. CONCLUSION: ADPKD was a significant risk factor for hospital encounters with kidney stones.

CONTEXTE: Il existe une perception selon laquelle les patients atteints de polykystose rénale autosomique dominante (ADPKD) seraient plus susceptibles de développer des calculs rénaux que la population générale. OBJECTIF: Comparer les taux d'hospitalisations et d'interventions pour calculs rénaux entre des patients atteints ou non d'ADPKD. TYPE D'ÉTUDE: Étude de cohorte rétrospective. CADRE: Ontario, Canada. SUJETS: Des patients atteints ou non d'ADPKD qui avaient déjà été hospitalisés entre 2002 et 2016. MESURES: Les taux d'hospitalisations et d'interventions pour calculs rénaux. MÉTHODOLOGIE: Nous avons utilisé une pondération d'exposition à probabilité inverse fondée sur les scores de propension afin d'équilibrer les indicateurs de santé de base entre les patients atteints ou non d'ADPKD. Nous avons suivi chaque patient jusqu'à son décès, jusqu'à son émigration, jusqu'au résultat ou jusqu'au 31 mars 2017. Nous avons utilisé un modèle de risques proportionnels de Cox pour comparer les taux d'événements entre les deux groupes. RÉSULTATS: Les patients atteints d'ADPKD présentaient un risque plus élevé d'être hospitalisés pour calculs rénaux que les patients non atteints d'ADPKD (81 patients sur 2094 atteints d'ADPKD [3,8 %] contre 60 patients sur 1902 sans ADPKD [3,2 %]; 8,9 contre 5,1 événements pour 1 000 années-personnes; risque relatif: 1,6 [IC 95 %: 1,3 à 2,1]). L'ADPKD n'a pas été associée à un risque plus élevé d'interventions pour retirer des calculs rénaux (49 patients sur 2094 atteints d'ADPKD [2,3 %] contre 47 patients sur 1902 sans ADPKD [2,4 %]; 5,3 contre 3,9 événements pour 1 000 années-personnes; risque relatif: 1,2 [IC 95 %: 0,9 à 1,3]). LIMITES: Nous n'avions pas d'information sur les événements liés aux calculs rénaux à l'extérieur de l'hôpital. Il existe une possibilité de facteurs de confusion résiduels. CONCLUSION: L'ADPKD s'est avéré un facteur de risque important d'être hospitalisé pour des calculs rénaux.

Risk Factors and Outcomes of Early Hospital Readmission in Canadian Kidney Transplant Recipients: A Population-Based Multi-Center Cohort Study.

BACKGROUND: Early hospital readmissions (EHRs) occur commonly in kidney transplant recipients. Conflicting evidence exists regarding risk factors and outcomes of EHRs. OBJECTIVE: To determine risk factors and outcomes associated with EHRs (ie, hospitalization within 30 days of discharge from transplant hospitalization) in kidney transplant recipients. DESIGN: Population-based cohort study using linked, administrative health care databases. SETTING: Ontario, Canada. PATIENTS: We included 5437 kidney transplant recipients from 2002 to 2015. MEASUREMENTS: Risk factors and outcomes associated with EHRs. We assessed donor, recipient, and transplant risk factors. We also assessed the following outcomes: total graft failure, death-censored graft failure, death with a functioning graft, mortality, and late hospital readmission. METHODS: We used multivariable logistic regression to examine the association of each risk factor and the odds of EHR. To examine the relationship between EHR status (yes vs no [reference]) and the outcomes associated with EHR (eg, total graft failure), we used a multivariable Cox proportional hazards model. RESULTS: In all, 1128 kidney transplant recipients (20.7%) experienced an EHR. We found the following risk factors were associated with an increased risk of EHR: older recipient age, lower income quintile, several comorbidities, longer hospitalization for initial kidney transplant, and older donor age. After adjusting for clinical characteristics, compared to recipients without an EHR, recipients with an EHR had an increased risk of total graft failure (adjusted hazard ratio [aHR]: 1.46, 95% CI: 1.29, 1.65), death-censored graft failure (aHR: 1.62, 95% CI: 1.36, 1.94), death with graft function (aHR: 1.34, 95% CI: 1.13, 1.59), mortality (aHR: 1.41, 95% CI: 1.22, 1.63), and late hospital readmission in the first 0.5 years of follow-up (eg, 0 to <0.25 years: aHR: 2.11, 95% CI: 1.85, 2.40). LIMITATIONS: We were not able to identify which readmissions could have been preventable and there is a potential for residual confounding. CONCLUSIONS: Results can be used to identify kidney transplant recipients at risk of EHR and emphasize the need for interventions to reduce the risk of EHRs. TRIAL REGISTRATION: This is not applicable as this is a population-based cohort study and not a clinical trial.

CONTEXTE: Les réadmissions précoces à l'hôpital (RPH) sont fréquentes chez les receveurs d'une greffe rénale. Les données sur les facteurs de risque d'une RPH et sur les résultats qui y sont associés restent toutefois contradictoires. OBJECTIF: Définir les facteurs de risque et les effets associés à une RPH (soit une hospitalization dans les 30 jours suivant la sortie de l'hôpital après la transplantation) chez les receveurs de greffe rénale. TYPE D'ÉTUDE: Étude de cohorte représentative d'une population, réalisée à partir des bases de données administratives en santé. CADRE: Ontario, Canada. SUJETS: Ont été inclus 5 437 adultes receveurs d'une greffe rénale entre 2002 et 2015. MESURES: Les facteurs de risque et les résultats associés à une RPH. Nous avons évalué les facteurs de risque du donneur, du receveur et de la transplantation. Nous avons également évalué les résultats suivants : l'échec du greffon, l'échec du greffon censuré par le décès, le décès avec un greffon fonctionnel, la mortalité et les réadmissions tardives. MÉTHODOLOGIE: Nous avons utilisé la régression logistique multivariée pour examiner l'association de chaque facteur de risque et les probabilités de RPH. Un modèle multivarié des risques proportionnels de Cox a par ailleurs servi à examiner la relation entre le statut des RPH (oui vs non [référence]) et les résultats associés à celles-ci (p. ex., l'échec de la greffe). RÉSULTATS: Dans la cohorte étudiée, 1 128 receveurs d'une greffe rénale (20,7 %) ont été réadmis précocement à l'hôpital. Les facteurs de risque suivants ont été associés à un risque accru de RPH : âge plus avancé du receveur, provenance d'un quartier au quintile de revenu inférieur, présence de plusieurs comorbidités, hospitalization initiale plus longue pour la transplantation rénale et âge plus avancé du donneur. Après ajustement pour les caractéristiques cliniques, par rapport aux receveurs de greffe qui n'avaient pas été réadmis précocement, les patients avec une RPH présentaient un risque accru d'échec du greffon (risque relatif corrigé [RRc] : 1,46; IC 95 % : 1,29-1,65), d'échec du greffon censuré par le décès (RRc: 1,62; IC 95 % : 1,36-1,94), de décès avec un greffon fonctionnel (RRc: 1,34; IC 95 % : 1,13-1,59), de mortalité (RRc: 1,41; IC 95 % : 1,22-1,63) et de réadmission tardive au cours des premiers six mois de suivi (p. ex., entre 0 et moins de 0,25 an de suivi, le RRc était de 2,11; [IC 95 % : 1,85-2,40]). LIMITES: Nous n'avons pas été en mesure d'identifier les réadmissions qui auraient pu être prévenues et il existe un risque de facteurs de confusion résiduels. CONCLUSION: Ces résultats peuvent être employés pour identifier les receveurs d'une greffe rénale susceptibles d'être réadmis rapidement à l'hôpital. Ces résultats soulignent en outre la nécessité d'interventions pour réduire le risque de RPH. ENREGISTREMENT DE L'ESSAI: Sans objet puisqu'il s'agit d'une étude de cohorte basée sur la population et non d'un essai clinique.

Incidence, Treatment, and Implications of Kidney Stones During Pregnancy: A Matched Population-Based Cohort Study.

Purpose: To determine the incidence of kidney stones in pregnancy, the risk of adverse birth outcomes, and treatment trends. Methods: We performed a population-based matched cohort study using Ontario's health care databases. All pregnancies in Ontario from 2004 to 2014 were identified. The study exposure was hospital admission, emergency room visit, or intervention for kidney stones during pregnancy. Each pregnancy with a stone was matched to up to six pregnancies without a stone based on age, region of residence, income quintile, year of cohort entry, prior births, and multibirths. The primary outcome was adverse birth outcome defined as preterm birth, low birth weight, or infant death. Secondary outcomes included premature rupture of membranes (PROM), pre-eclampsia, and cesarean section (C/S), as well as the type/frequency of intervention for stones in pregnancy. Logistic regression models, with generalized estimating equations, were used to assess any differences in study outcomes across groups. Results: Of 1.39 million pregnancies identified, there were 2863 pregnancies with stones (0.2%), which were matched with 17,171 pregnancies without stones. Pregnancies with stones had an increased risk for adverse birth outcome compared with matched pregnancies without stones (odds ratio [OR] 1.62, confidence interval [95% CI] 1.43-1.82, p < 0.0001). Pregnancies with stones also had a greater risk for pre-eclampsia (OR 1.42, 95% CI 1.02-1.99, p = 0.04) and C/S (OR 1.39, 95% CI 1.27-1.51, p < 0.0001), but not PROM. Twenty-six percent of pregnant patients admitted for a stone had an intervention, most commonly a stent or ureteroscopy. Conclusion: Our study demonstrated an increased risk of adverse birth outcomes in pregnancies with kidney stones. These results will be important for counseling pregnant patients with kidney stones and women of reproductive age who are at risk of developing stones.

Hydromorphone and the risk of infective endocarditis among people who inject drugs: a population-based, retrospective cohort study.

BACKGROUND: The incidence of infective endocarditis related to injection drug use is increasing. On the basis of clinical practice and epidemiological and in-vitro data, we postulated that exposure to controlled-release hydromorphone is associated with an increased risk of infective endocarditis among people who inject drugs. METHODS: We used linked health administrative databases in Ontario, Canada, to assemble a retrospective cohort of adults (aged 18-55 years) who inject drugs for the period of April 1, 2006, to Sept 30, 2015. Cases of infective endocarditis among this cohort were identified using International Classification of Diseases 10 codes. We estimated exposure to hydromorphone and risk of infective endocarditis among this cohort in two ways. First, in a population-level analysis, we identified patients living in regions with high (≥25%) and low (≤15%) hydromorphone prescription rates and, after matching 1:1 on various baseline characteristics, compared their frequency of infective endocarditis. Second, in a patient-level analysis including only those with prescription drug data, we identified those who had filled prescriptions (ie, received the drug from the pharmacy) for controlled-release or immediate-release hydromorphone and, after matching 1:1 on various baseline characteristics, compared their frequency of infective endocarditis with that of patients who had filled prescriptions for other opioids. RESULTS: Between April 1, 2006, and Sept 30, 2015, 60 529 patients had evidence of injection drug use, 733 (1·2%, 95% CI 1·1-1·3) of whom had infective endocarditis. In the population-level analysis of 32 576 matched patients, we identified 254 (1·6%) admissions with infective endocarditis in regions with high hydromorphone use and 113 (0·7%) admissions in regions with low use (adjusted odds ratio [OR] 2·2, 95% CI 1·8-2·8, p<0·0001). In the patient-level analysis of 3884 matched patients, the frequency of infective endocarditis was higher among patients who filled prescriptions for hydromorphone than among those who filled prescriptions for non-hydromorphone opioids (2·8% [109 patients] vs 1·1% [41 patients]; adjusted OR 2·5, 95% CI 1·8-3·7, p<0·0001). This significant association was seen for controlled-release hydromorphone (3·9% [73 of 1895 patients] vs 1·1% [20 of 1895]; adjusted OR 3·3, 95% CI 2·1-5·6, p<0·0001), but not for immediate-release hydromorphone (1·8% [36 of 1989] vs 1·1% [21 of 1989]; 1·7, 0·9-3·6, p=0·072. INTERPRETATION: Among people who inject drugs, the risk of infective endocarditis is significantly higher for those exposed to controlled-release hydromorphone than to other opioids. This association might be mediated by the controlled-release mechanism and should be the subject of further investigation. FUNDING: Ontario Ministry of Health and Long-Term Care, Academic Medical Organization of Southwestern Ontario, Schulich School of Medicine and Dentistry (Western University), and Lawson Health Research Institute.

Complications in Patients With Autosomal Dominant Polycystic Kidney Disease Undergoing Ureteroscopy: A Cohort Study.

BACKGROUND: Ureteroscopy is a minimally invasive treatment option for upper tract stones. The distorted kidney anatomy in patients with autosomal dominant polycystic kidney disease (ADPKD) may place them at higher risk for ureteroscopic complications. OBJECTIVE: To compare the 30-day risk of ureteroscopic complications between patients with and without ADPKD. DESIGN: Retrospective cohort study. SETTING: Ontario, Canada. PATIENTS: Seventy three patients with ADPKD and 81 445 patients without ADPKD who underwent ureteroscopy for upper urinary tract stones between April 1, 2002, and March 1, 2018. MEASUREMENTS: A 30-day risk of (1) hospital presentation with ureteroscopic complications (which was a composite outcome of either emergency department visit or hospital admission with acute kidney injury, urinary tract infection, or sepsis); (2) all-cause hospital presentation; (3) all-cause hospital admission; and (4) all-cause emergency department visit. METHODS: We regressed outcomes on demographic variables, health care use in the prior 1-year, various procedures and comorbidities related to the outcome in the prior 5 years, and prescribed medications filled in the past 120 days using modified Poisson regression to compare the risk ratio (RR) of each outcome between patients with and without ADPKD. RESULTS: The median (interquartile, IQR) age was 44 (38-60 years) in the ADPKD group and 53 (42-64) in the control group. About 40% were women in both groups. The risk of ureteroscopic complications was not significantly different in patients with versus without ADPKD (8.2% vs 4.3%; adjusted RR = 1.5, 95% confidence interval [CI] = 0.7-3.2). Patients with versus without ADPKD were more likely to present to hospital after their procedure (35.6% vs. 20.0%; adjusted RR = 1.6, 95% CI = 1.2-2.2), which included a statistically significant increase in the risk of presenting to the emergency department (32.9% vs. 19.0%; adjusted RR = 1.6, 95% CI = 1.1-2.2) but not hospital admissions (10.9% vs. 5.0%; adjusted RR = 1.8, 95% CI = 0.9-3.4). LIMITATIONS: The low numbers of events led to imprecision around the estimates. CONCLUSION: Patients with ADPKD have a higher risk of return to the hospital within 30 days of ureteroscopy for stone disease. TRIAL REGISTRATION: We did not register this study.

CONTEXTE: L'urétéroscopie est une option minimalement invasive pour traiter les calculs des voies urinaires hautes. Cependant, les distorsions anatomiques des reins présentes chez les patients atteints de polykystose rénale autosomique dominante (ADPKD) exposent ces derniers à un risque accru de complications liées à la procédure. OBJECTIF: Comparer le risque de complications liées à l'urétéroscopie chez des patients atteints ou non d'ADPKD dans les 30 jours suivant la procédure. TYPE D'ÉTUDE: Étude de cohorte rétrospective. CADRE: Ontario, Canada. SUJETS: L'étude porte sur 73 patients atteints d'ADPKD et 81 445 patients témoins ayant subi une urétéroscopie entre le 1er avril 2002 et le 1er mars 2018 pour le traitement de calculs des voies urinaires hautes. MESURES: Le risque dans les 30 jours de: 1) visite à l'hôpital pour des complications liées à l'urétéroscopie (résultat composite d'une visite aux urgences ou d'une admission en raison d'une insuffisance rénale aiguë, d'une infection urinaire ou d'un sepsis); 2) toute autre cause de visite à l'hôpital; 3) toute autre cause d'admission; et 4) toute autre cause de visite aux urgences. MÉTHODOLOGIE: Une régression de Poisson modifiée a été employée pour l'analyze des résultats sur les variables démographiques, l'utilization des soins de santé au cours de l'année précédente, les différentes procédures et maladies concomitantes liées au résultat au cours des cinq années précédentes et les médicaments prescrits au cours des 120 derniers jours afin de comparer le rapport de risque (RR) de chaque résultat entre les patients atteints ou non d'ADPKD. RÉSULTATS: L'âge médian des sujets (écart interquartile) s'établissait à 44 ans (38-60 ans) dans le groupe de patients ADPKD et à 53 ans (42-64 ans) dans le groupe témoin; les femmes représentaient environ 40 % des sujets dans les deux groupes. Le risque de complications liées à l'urétéroscopie n'était pas significativement différent entre le groupe ADPKD et le groupe témoin (8,2 % vs 4,2 %; RR corrigé: 1,5; IC 95 %: 0,7 à 3,2). Tous les patients, avec ou sans ADPKD, étaient plus susceptibles de se présenter à l'hôpital après l'intervention (36,6 % vs 20,0 %; RR corrigé: 1,6; IC 95 %: 1,2 à 2,2). Ce résultat incluait un risque significativement plus élevé de se présenter aux urgences (32,9 % vs 19,0 %; RR corrigé: 1,6; IC 95 %: 1,1 à 2,2), mais pas d'être hospitalisé (11 % vs 5 %; RR corrigé: 1,8; IC 95 %: 0,9 à 3,4). LIMITES: Le faible nombre d'événements a mené à l'imprécision des estimations. CONCLUSION: Les patients atteints d'ADPKD présentent un risque accru de retourner à l'hôpital dans les 30 jours suivant une urétéroscopie pour traiter des calculs urinaires. ENREGISTREMENT DE L'ESSAI: L'étude n'a pas été enregistrée.

Interhospital Transfer and Outcomes in Patients with AKI: A Population-Based Cohort Study.

Background: Patients with AKI may require interhospital transfer to receive RRT. Interhospital transfer may lead to delays in therapy, resulting in poor patient outcomes. There is minimal data comparing outcomes among patients undergoing transfer for RRT versus those who receive RRT at the hospital to which they first present. Methods: We conducted a population-based cohort study of all adult patients (≥19 years) who received acute dialysis within 14 days of admission to an acute-care hospital between April 1, 2004 and March 31, 2015. The transferred group included all patients who presented to a hospital without a dialysis program and underwent interhospital transfer (with the start of dialysis ≤3 days of transfer and within 14 days of initial admission). All other patients were considered nontransferred. The primary outcome was time to 90-day all-cause mortality, adjusting for demographics, comorbidities, and measures of acute illness severity. We also assessed chronic dialysis dependence as a secondary outcome, using the Fine and Gray proportional hazards model to account for the competing risks of death. In a secondary post hoc analysis, we assessed these outcomes in a propensity score-matched cohort, matching on age, sex, and prior CKD status. Results: We identified 27,270 individuals initiating acute RRT within 14 days of a hospital admission, of whom 2113 underwent interhospital transfer. Interhospital transfer was associated with lower rate of mortality (adjusted hazard ratio [aHR], 0.90; 95% CI, 0.84 to 0.97). Chronic dialysis dependence was not significantly different between groups (aHR, 0.98; 95% CI, 0.91 to 1.06). In the propensity score-matched analysis, interhospital transfer remained associated with a lower risk of death (HR, 0.88; 95% CI, 0.80 to 0.96). Conclusions: Interhospital transfer for receipt of RRT does not confer higher mortality or worse kidney outcomes.

Probabilistic Graphical Modeling for Estimating Risk of Coronary Artery Disease: Applications of a Flexible Machine-Learning Method.

Objectives. Coronary artery disease (CAD) is the leading cause of death and disease burden worldwide, causing 1 in 7 deaths in the United States alone. Risk prediction models that can learn the complex causal relationships that give rise to CAD from data, instead of merely predicting the risk of disease, have the potential to improve transparency and efficacy of personalized CAD diagnosis and therapy selection for physicians, patients, and other decision makers. Methods. We use Bayesian networks (BNs) to model the risk of CAD using the Z-Alizadehsani data set-a published real-world observational data set of 303 Iranian patients at risk for CAD. We also describe how BNs can be used for incorporation of background knowledge, individual risk prediction, handling missing observations, and adaptive decision making under uncertainty. Results. BNs performed on par with machine-learning classifiers at predicting CAD and showed better probability calibration. They achieved a mean 10-fold area under the receiver-operating characteristic curve (AUC) of 0.93 ± 0.04, which was comparable with the performance of logistic regression with L1 or L2 regularization (AUC: 0.92 ± 0.06), support vector machine (AUC: 0.92 ± 0.06), and artificial neural network (AUC: 0.91 ± 0.05). We describe the use of BNs to predict with missing data and to adaptively calculate prognostic values of individual variables under uncertainty. Conclusion. BNs are powerful and versatile tools for risk prediction and health outcomes research that can complement traditional statistical techniques and are particularly useful in domains in which information is uncertain or incomplete and in which interpretability is important, such as medicine.