The value of mathematical modelling approaches in epidemiology for public health decisión making

It is discussed the relevance of quantitative approaches, specifically mathematical modelling in epidemiology, in the public health decision-making process. This topic is discussed here based on the experience of various experts in mathematical epidemiology and public health. First, the definition of mathematical modelling is presented, especially in the context of epidemiology. Second, the different uses and socio-political implications, including empirical examples of recent experiences that have taken place at the international level are addressed. Finally, some general considerations regarding the challenges encountered in the use and application of mathematical modelling in epidemiology in the decision-making process at the local and national levels.

Se trata sobre la importancia de los abordajes cuantitativos, específicamente la formulación de modelos matemáticos en epidemiología, dentro del proceso de toma de decisiones en salud pública. Esta importante temática se analiza basándose en la experiencia de algunos expertos en epidemiología matemática y salud pública. En primer lugar, se presenta la definición de modelación matemática, particularmente dentro del contexto de la epidemiología. En segundo lugar, se abordan los diferentes usos y las implicaciones socio-políticas, incluyendo ejemplos de experiencias recientes que han ocurrido a nivel internacional. Finalmente, se hace referencia a ciertas consideraciones generales respecto a los retos que representa el uso y la aplicación de modelos matemáticos en epidemiología para el proceso de toma de decisiones a nivel local y nacional.

Characterizing Trends in the Use of Food Donations and Other Food-Related Community-Based Social Assistance Programs in a Cohort of New Food Bank Users in Quebec, Canada.

Objective: To characterize 12-month trends in the use of food donations and other food-related community-based social assistance programs (CB-SAPs) during the first year following the enrollment of new food bank (FB) users in Quebec, Canada. Methods: A cohort of 1,001 newly registered FB-users in Quebec from the Pathways Study were followed-up during 12-month following baseline assessment. Outcomes were monthly use of food donations and other food-related CB-SAPs. Main predictors were alternative food source utilization (AFSU) profiles: 1) exclusive-FB-users; 2) FB+fruit/vegetable-market-users; and 3) Multiple/diverse-AFS-users. Covariates included sociodemographic characteristics, health status, and major life events. We fit Bayesian hierarchical mixed-effect models, accounting for spatial clustering, temporal correlation, and censoring. Results: We observed an overall downward trend of food donation use among study completers (n = 745). Each AFSU profile had a distinctive monthly trend of food donation use, but probabilities of use across the three profiles overlapped, between 44% and 55%. The use of other food-related CB-SAPs was low and not correlated with AFSU profiles. Conclusion: De novo FB-users use food donations in different ways over time according to specific contextual AFSU profiles.

How to plan and manage an individual participant data meta-analysis. An illustrative toolkit.

Individual participant data meta-analyses (IPD-MAs) have several benefits over standard aggregate data meta-analyses, including the consideration of additional participants, follow-up time, and the joint consideration of study- and participant-level heterogeneity for improved diagnostic and prognostic model development and evaluation. However, IPD-MAs are resource-intensive and require careful budgeting of time from data contributing groups, a dedicated management team, diversity of expertise, clearly documented data sharing and authorship agreements, and consistent and clear communication. We present a toolkit to facilitate the implementation and management of IPD-MAs, from study recruitment to retrospective harmonization. The toolkit was developed and refined over our work on multiple multinational IPD-MA projects over the last 13 years. The toolkit's budget and email templates, agreements, project management spreadsheets, and standard operating procedures are meant to facilitate routine IPD-MA tasks to expedite implementing and managing future IPD-MA projects.

The Pathways study: a cohort study of new food-aid users in rural, semi urban, and urban areas of Quebec, Canada.

BACKGROUND: While considerable research has been conducted on household food insecurity (HFI), little research has examined the effects of food donation programs on users' living conditions. The Pathways study was established to investigate the long-term effects of food donation programs on food insecurity as well as other critical outcomes, such as diet, health, and social support. Herein, we describe the design of the Pathways Study and the participants' characteristics at baseline. METHODS: The Pathways study is a prospective cohort study of 1001 food-aid users in Quebec (Canada). We recruited newly registered users of food donation programs from 106 community-based food-aid organizations that partnered with the study. Baseline data were collected through face-to-face interviews from September 2018 to January 2020, with planned follow-up interviews at 12 and 24 months after enrollment. Household food insecurity, diet, food competencies, food shopping behaviors, perceived food environment, health status, social support and isolation, sociodemographic characteristics, housing conditions, negative life events, and the impacts of COVID-19 were assessed with validated questionnaires. RESULTS: The cohort included 1001 participants living in rural (n = 181), semi-urban (n = 250), and urban areas (n = 570). Overall, household food insecurity was reported as severe among 46.2% and moderate in 36.9% of participants. Severe household food insecurity was more prevalent in rural (51.4%) and urban (47.8%) areas compared to semi-urban (39%) areas. Overall, 76.1% of participants reported an annual income below C$20,000. Half (52%) had low education levels (high school or lower), 22.0% lived in single-parent households, and 52.1% lived alone. Most (62.9%) experienced at least one major financial crisis in the preceding year. CONCLUSIONS: Results show that newly registered users of food donation programs often have low-income and severe food insecurity, with major differences across geographical locations. The Pathways study is the first study designed to follow, over a 2-year period, a cohort of newly registered users of food donation programs and to quantify their trajectories of service use. Findings from the Pathways study might help adapt the community response to the strategies used by food-insecure households to feed themselves.

Overlapping research efforts in a global pandemic: a rapid systematic review of COVID-19-related individual participant data meta-analyses.

BACKGROUND: Individual participant data meta-analyses (IPD-MAs), which involve harmonising and analysing participant-level data from related studies, provide several advantages over aggregate data meta-analyses, which pool study-level findings. IPD-MAs are especially important for building and evaluating diagnostic and prognostic models, making them an important tool for informing the research and public health responses to COVID-19. METHODS: We conducted a rapid systematic review of protocols and publications from planned, ongoing, or completed COVID-19-related IPD-MAs to identify areas of overlap and maximise data request and harmonisation efforts. We searched four databases using a combination of text and MeSH terms. Two independent reviewers determined eligibility at the title-abstract and full-text stages. Data were extracted by one reviewer into a pretested data extraction form and subsequently reviewed by a second reviewer. Data were analysed using a narrative synthesis approach. A formal risk of bias assessment was not conducted. RESULTS: We identified 31 COVID-19-related IPD-MAs, including five living IPD-MAs and ten IPD-MAs that limited their inference to published data (e.g., case reports). We found overlap in study designs, populations, exposures, and outcomes of interest. For example, 26 IPD-MAs included RCTs; 17 IPD-MAs were limited to hospitalised patients. Sixteen IPD-MAs focused on evaluating medical treatments, including six IPD-MAs for antivirals, four on antibodies, and two that evaluated convalescent plasma. CONCLUSIONS: Collaboration across related IPD-MAs can leverage limited resources and expertise by expediting the creation of cross-study participant-level data datasets, which can, in turn, fast-track evidence synthesis for the improved diagnosis and treatment of COVID-19. TRIAL REGISTRATION: 10.17605/OSF.IO/93GF2.

Traffic-Calming Measures and Road Traffic Collisions and Injuries: a Spatiotemporal Analysis.

Traffic-calming measures (TCMs) are physical modifications to the road network aimed at making the roads safer. Although studies have reported reductions in road crashes and injuries tied to the presence of TCMs, they have been criticized for their pre-post designs. This study aims to complement our knowledge of TCMs effectiveness by assessing their impact using a longitudinal design. The implementation of eight TCMs, including curb extensions and speed humps, was evaluated at the intersections and census tract levels in Montreal, Canada from 2012 to 2019. The primary outcome was fatal or serious collisions among all road users. Inference was performed using a Bayesian implementation of Conditional Poisson regression in which random effects were used to account for the spatiotemporal variation in collisions. TCMs were generally implemented on local roads, although most collisions occurred on arterial roads. Overall, there was weak evidence that TCMs were associated with study outcomes. However, subgroup analyses of intersections on local roads suggested a reduction in collision rates due to TCMs (median IRR: 0.31; 95% Credible Interval: 0.12 - 0.86). To improve road safety, effective counterparts of TCMs on arterial roads must be identified and implemented.

Heterogeneity of Zika virus exposure and outcome ascertainment across cohorts of pregnant women, their infants and their children: a metadata survey.

OBJECTIVES: To support the Zika virus (ZIKV) Individual Participant Data (IPD) Consortium's efforts to harmonise and analyse IPD from ZIKV-related prospective cohort studies and surveillance-based studies of pregnant women and their infants and children; we developed and disseminated a metadata survey among ZIKV-IPD Meta-Analysis (MA) study participants to identify and provide a comprehensive overview of study-level heterogeneity in exposure, outcome and covariate ascertainment and definitions. SETTING: Cohort and surveillance studies that measured ZIKV infection during pregnancy or at birth and measured fetal, infant, or child outcomes were identified through a systematic search and consultations with ZIKV researchers and Ministries of Health from 20 countries or territories. PARTICIPANTS: Fifty-four cohort or active surveillance studies shared deidentified data for the IPD-MA and completed the metadata survey, representing 33 061 women (11 020 with ZIKV) and 18 281 children. PRIMARY AND SECONDARY OUTCOME MEASURES: Study-level heterogeneity in exposure, outcome and covariate ascertainment and definitions. RESULTS: Median study sample size was 268 (IQR=100, 698). Inclusion criteria, follow-up procedures and exposure and outcome ascertainment were highly heterogenous, differing meaningfully across regions and multisite studies. Enrolment duration and follow-up for children after birth varied before and after the declaration of the Public Health Emergency of International Concern (PHEIC) and according to the type of funding received. CONCLUSION: This work highlights the logistic and statistical challenges that must be addressed to account for the multiple sources of within-study and between-study heterogeneity when conducting IPD-MAs of data collected in the research response to emergent pathogens like ZIKV.

Spatio-temporal clusters and patterns of spread of dengue, chikungunya, and Zika in Colombia.

BACKGROUND: Colombia has one of the highest burdens of arboviruses in South America. The country was in a state of hyperendemicity between 2014 and 2016, with co-circulation of several Aedes-borne viruses, including a syndemic of dengue, chikungunya, and Zika in 2015. METHODOLOGY/PRINCIPAL FINDINGS: We analyzed the cases of dengue, chikungunya, and Zika notified in Colombia from January 2014 to December 2018 by municipality and week. The trajectory and velocity of spread was studied using trend surface analysis, and spatio-temporal high-risk clusters for each disease in separate and for the three diseases simultaneously (multivariate) were identified using Kulldorff's scan statistics. During the study period, there were 366,628, 77,345 and 74,793 cases of dengue, chikungunya, and Zika, respectively, in Colombia. The spread patterns for chikungunya and Zika were similar, although Zika's spread was accelerated. Both chikungunya and Zika mainly spread from the regions on the Atlantic coast and the south-west to the rest of the country. We identified 21, 16, and 13 spatio-temporal clusters of dengue, chikungunya and Zika, respectively, and, from the multivariate analysis, 20 spatio-temporal clusters, among which 7 were simultaneous for the three diseases. For all disease-specific analyses and the multivariate analysis, the most-likely cluster was identified in the south-western region of Colombia, including the Valle del Cauca department. CONCLUSIONS/SIGNIFICANCE: The results further our understanding of emerging Aedes-borne diseases in Colombia by providing useful evidence on their potential site of entry and spread trajectory within the country, and identifying spatio-temporal disease-specific and multivariate high-risk clusters of dengue, chikungunya, and Zika, information that can be used to target interventions.

A joint spatial marked point process model for dengue and severe dengue in Medellin, Colombia.

The spatial distribution of surveillance-reported dengue cases and severity are usually analyzed separately, assuming independence between the spatial distribution of non-severe and severe cases. Given the availability of data for the individual geo-location of surveillance-notified dengue cases, we conducted a spatial analysis to model non-severe and severe dengue simultaneously, using a hierarchical Bayesian model. We fit a joint model to the spatial pattern formed by dengue cases as well as to the severity status of the cases. Results showed that age and socioeconomic status were associated with dengue presence, and there was evidence of clustering for overall cases but not for severity. Our findings inform decision making to address the preparedness or implementation of dengue control strategies at the local level.

Decomposition of socioeconomic inequalities in arboviral diseases in Brazil and Colombia (2007-2017).

BACKGROUND: We used surveillance data from Brazil and Colombia during 2007-2017 to assess the presence of socioeconomic inequalities on dengue, chikungunya and Zika at the neighborhood level in two Latin American cities. METHODS: To quantify the inequality, we estimated and decomposed the relative concentration index of inequality (RCI) accounting for the spatiotemporal distribution of the diseases. RESULTS: There were 281 426 arboviral cases notified in Fortaleza, Brazil, and 40 889 in Medellin, Colombia. The RCI indicated greater concentration of dengue cases among people living in low socioeconomic settings in both sites. The RCIs for chikungunya in Fortaleza covered the line of equality during their introduction in 2014, while the RCIs for Zika and chikungunya in Medellin indicated the presence of a small inequality. The RCI decomposition showed that year of notification and age were the main contributors to this inequality. In Medellin, the RCI decomposition showed that age and access to waste management accounted for 75.5%, 72.2% and 54.5% of the overall inequality towards the poor for dengue, chikungunya and Zika, respectively. CONCLUSIONS: Our study presents estimates of the socioeconomic inequality of arboviruses and its decomposition in two Latin American cities. We corroborate the concentration of arboviral diseases in low socioeconomic neighborhoods and identify that year of occurrence, age, presence of healthcare facilities and waste management are key determinants of the heterogenous distribution of endemic arboviruses across the socioeconomic spectrum.

Estimating force of infection from serologic surveys with imperfect tests.

BACKGROUND: The force of infection, or the rate at which susceptible individuals become infected, is an important public health measure for assessing the extent of outbreaks and the impact of control programs. METHODS AND FINDINGS: We present Bayesian methods for estimating force of infection using serological surveys of infections which produce a lasting immune response, accounting for imperfections of the test, and uncertainty in such imperfections. In this estimation, the sensitivity and specificity can either be fixed, or belief distributions of their values can be elicited to allow for uncertainty. We analyse data from two published serological studies of dengue, one in Colombo, Sri Lanka, with a single survey and one in Medellin, Colombia, with repeated surveys in the same individuals. For the Colombo study, we illustrate how the inferred force of infection increases as the sensitivity decreases, and the reverse for specificity. When 100% sensitivity and specificity are assumed, the results are very similar to those from a standard analysis with binomial regression. For the Medellin study, the elicited distribution for sensitivity had a lower mean and higher variance than the one for specificity. Consequently, taking uncertainty in sensitivity into account resulted in a wide credible interval for the force of infection. CONCLUSIONS: These methods can make more realistic estimates of force of infection, and help inform the choice of serological tests for future serosurveys.

Dengue, Severity Paradox, and Socioeconomic Distribution Among Afro-Colombians.

BACKGROUND: The clinical presentation of dengue ranges from self-limited mild illness to severe forms, including death. African ancestry is often described as protective against dengue severity. However, in the Latin American context, African ancestry has been associated with increased mortality. This "severity paradox" has been hypothesized as resulting from confounding or heterogeneity by socioeconomic status (SES). However, few systematic analyses have been conducted to investigate the presence and nature of the disparity paradox. METHODS: We fit Bayesian hierarchical spatiotemporal models using individual-level surveillance data from Cali, Colombia (2012-2017), to assess the overall morbidity and severity burden of notified dengue. We fitted overall and ethnic-specific models to assess the presence of heterogeneity by SES across and within ethnic groups (Afro-Colombian vs. non-Afro-Colombians), conducting sensitivity analyses to account for potential underreporting. RESULTS: Our study included 65,402 dengue cases and 13,732 (21%) hospitalizations. Overall notified dengue incidence rates did not vary across ethnic groups. Severity risk was higher among Afro-Colombians (risk ratio [RR] = 1.16; 95% Credible Interval [95% CrI] = 1.08, 1.24) but after accounting for underreporting by ethnicity this association was nearly null (RR = 1.02; 95% CrI = 0.97, 1.07). Subsidized health insurance and low-SES were associated with increased overall dengue rates and severity. CONCLUSION: The paradoxically increased severity among Afro-Colombians can be attributed to differential health-seeking behaviors and reporting among Afro-Colombians. Such differential reporting can be understood as a type of intersectionality between SES, insurance scheme, and ethnicity that requires a quantitative assessment in future studies.

Assessing the reporting of Dengue, Chikungunya and Zika to the National Surveillance System in Colombia from 2014-2017: A Capture-recapture analysis accounting for misclassification of arboviral diagnostics.

BACKGROUND: Chikungunya, dengue, and Zika are three different arboviruses which have similar symptoms and are a major public health issue in Colombia. Despite the mandatory reporting of these arboviruses to the National Surveillance System in Colombia (SIVIGILA), it has been reported that the system captures less than 10% of diagnosed cases in some cities. METHODOLOGY/PRINCIPAL FINDINGS: To assess the scope and degree of arboviruses reporting in Colombia between 2014-2017, we conducted an observational study of surveillance data using the capture-recapture approach in three Colombian cities. Using healthcare facility registries (capture data) and surveillance-notified cases (recapture data), we estimated the degree of reporting by clinical diagnosis. We fit robust Poisson regressions to identify predictors of reporting and estimated the predicted probability of reporting by disease and year. To account for the potential misclassification of the clinical diagnosis, we used the simulation extrapolation for misclassification (MC-SIMEX) method. A total of 266,549 registries were examined. Overall arboviruses' reporting ranged from 5.3% to 14.7% and varied in magnitude according to age and year of diagnosis. Dengue was the most notified disease (21-70%) followed by Zika (6-45%). The highest reporting rate was seen in 2016, an epidemic year. The MC-SIMEX corrected rates indicated underestimation of the reporting due to the potential misclassification bias. CONCLUSIONS: These findings reflect challenges on arboviruses' reporting, and therefore, potential challenges on the estimation of arboviral burden in Colombia and other endemic settings with similar surveillance systems.

Effectiveness of trauma centre verification: a systematic review and meta-analysis.

Background: There is a growing trend toward verification of trauma centres, but its impact remains unclear. This systematic review aimed to synthesize available evidence on the effectiveness of trauma centre verification. Methods: We conducted a systematic search of the CINAHL, Embase, HealthStar, MEDLINE and ProQuest databases, as well as the websites of key injury organizations for grey literature, from inception to June 2019, without language restrictions. Our population consisted of injured patients treated at trauma centres. The intervention was trauma centre verification. Comparison groups comprised nonverified trauma centres, or the same centre before it was first verified or re-verified. The primary outcome was in-hospital mortality; secondary outcomes included adverse events, resource use and processes of care. We computed pooled summary estimates using random-effects meta-analysis. Results: Of 5125 citations identified, 29, all conducted in the United States, satisfied our inclusion criteria. Mortality was the most frequently investigated outcome (n = 20), followed by processes of care (n = 12), resource use (n = 12) and adverse events (n = 7). The risk of bias was serious to critical in 22 studies. We observed an imprecise association between verification and decreased mortality (relative risk 0.74, 95% confidence interval 0.52 to 1.06) in severely injured patients. Conclusion: Our review showed mixed and inconsistent associations between verification and processes of care or patient outcomes. The validity of the published literature is limited by the lack of robust controls, as well as any evidence from outside the US, which precludes extrapolation to other health care jurisdictions. Quasiexperimental studies are needed to assess the impact of trauma centre verification. Systematic reviews registration: PROSPERO no. CRD42018107083.

Contexte: Le processus d'audit des centres de traumatologie gagne en popularité, mais ses effets concrets ne sont pas bien connus. La présente revue systématique a cherché à résumer les données probantes disponibles sur l'efficacité de l'audit des centres de traumatologie. Méthodes: Nous avons effectué des recherches systématiques dans les bases de données CINAHL, Embase, HealthSTAR, MEDLINE et ProQuest, de même qu'une recherche dans la littérature grise sur les sites Web d'organisations majeures du domaine des traumas, de leur création à juin 2019, sans restriction de langue. La population à l'étude était l'ensemble des patients blessés traités en centre de traumatologie. L'intervention était l'audit du centre de traumatologie. Les groupes de comparaison correspondaient aux centres de traumatologie n'ayant pas subi d'audit, ou le même centre, avant son premier audit ou un audit subséquent. Le principal résultat à l'étude était la mortalité en milieu hospitalier; les résultats secondaires étaient les événements indésirables, l'utilisation des ressources et les processus de soins. Nous avons calculé des estimations sommaires par méta-analyse à effets aléatoires sur données groupées. Résultats: Sur les 5125 citations retenues, 29 publications sur des études menées aux États-Unis répondaient à nos critères d'inclusion. La mortalité était le résultat le plus souvent à l'étude (n = 20), puis suivaient les processus de soins (n = 12), l'utilisation des ressources (n = 12) et les événements indésirables (n = 7). Le risque de biais était important ou critique dans 22 études. Nous avons observé une association imprécise entre l'audit et une baisse de la mortalité (risque relatif 0,74; intervalle de confiance à 95 % 0,52 à 1,06) chez les patients ayant subi un trauma grave. Conclusion: Notre revue a conclu qu'il y avait des associations mitigées et manquant d'uniformité entre l'audit et les processus de soins ou les issues pour les patients. La validité des données à l'étude était limitée par un manque de contrôles fiables, ainsi que par l'absence de données provenant d'autres pays que les États-Unis, ce qui empêche l'extrapolation à d'autres systèmes de santé. Des études quasi expérimentales devront être menées pour évaluer les effets de l'audit des centres de traumatologie. Enregistrement de la revue systématique: Registre PROSPERO, numéro CRD42018107083.

Estimating the Force of Infection for Dengue Virus Using Repeated Serosurveys, Ouagadougou, Burkina Faso.

Because of limited data on dengue virus in Burkina Faso, we conducted 4 consecutive age-stratified longitudinal serologic surveys, ≈6 months apart, among persons 1-55 years of age, during June 2015-March 2017, which included a 2016 outbreak. The seroconversion rate before the serosurvey enrollment was estimated by binomial regression, taking age as the duration of exposure, and assuming constant force of infection (FOI) over age and calendar time. We calculated FOI between consecutive surveys and rate ratios for potentially associated characteristics based on seroconversion using the duration of intervals. Among 2,897 persons at enrollment, 66.3% were IgG-positive, and estimated annual FOI was 5.95%. Of 1,269 enrollees participating in all 4 serosurveys, 438 were IgG-negative at enrollment. The annualized FOI ranged from 10% to 20% (during the 2016 outbreak). Overall, we observed high FOI for dengue. These results could support decision-making about control and preventive measures for dengue.

Spatiotemporal distribution and socioeconomic disparities of dengue, chikungunya and Zika in two Latin American cities from 2007 to 2017.

OBJECTIVE: To assess the presence, pattern and magnitude of socioeconomic inequalities on dengue, chikungunya and Zika in Latin America, accounting for their spatiotemporal distribution. METHODS: Using longitudinal surveillance data (reported arboviruses) from Fortaleza, Brazil and Medellin, Colombia (2007-2017), we fit Bayesian hierarchical models with structured random effects to estimate: (i) spatiotemporally adjusted incidence rates; (ii) Relative Concentration Index and Absolute Concentration Index of inequality; (iii) temporal trends in RCIs; and (iv) socioeconomic-specific estimates of disease distribution. The spatial analysis was conducted at the neighbourhood level (urban settings). The socioeconomic measures were the median monthly household income (MMHI) for Brazil and the Socio-Economic Strata index (SES) in Colombia. RESULTS: There were 281 426 notified arboviral cases in Fortaleza and 40 887 in Medellin. We observed greater concentration of dengue among residents of low socioeconomic neighbourhoods in both cities: Relative Concentration Index = -0.12 (95% CI = -0.13, -0.10) in Fortaleza and Relative Concentration Index = -0.04 (95% CI = -0.05, -0.03) in Medellin. The magnitude of inequalities varied over time across sites and was larger during outbreaks. We identified a non-monotonic association between disease rates and socioeconomic measures, especially for chikungunya, that changed over time. The Relative Concentration Index and Absolute Concentration Index showed few if any inequalities for Zika. The socioeconomic-specific model showed increased disease rates at MMHI below US$400 in Brazil and at SES-index below level four, in Colombia. CONCLUSIONS: We provide robust quantitative estimates of socioeconomic inequalities in arboviruses for two Latin American cities. Our findings could inform policymaking by identifying spatial hotspots for arboviruses and targeting strategies to decrease disparities at the local level.

Treatment Switch to Dolutegravir With 2 Nucleoside Reverse-Transcriptase Inhibitors (NRTI) in Comparison to Continuation With Protease Inhibitor/Ritonavir Among Patients With Human Immunodeficiency Virus at Risk for Prior NRTI Resistance: A Cohort Analysis of Real-World Data.

BACKGROUND: Switching antiretroviral regimens when human immunodeficiency virus (HIV) viremia is controlled for a new regimen is challenging when there is the potential for prior nucleoside reverse-transcriptase inhibitor (NRTI) resistance. The objective was to study virologic outcomes after switching to dolutegravir compared with remaining on a boosted protease inhibitor (protease inhibitor/ritonavir [PI/r]) regimen in people with HIV (PWH) with prior documented virologic failure and/or exposure to mono/dual NRTIs. METHODS: We used the Quebec HIV Cohort including 10 219 PWH whose data were collected at 4 sites in Montreal, Canada. We included all PWH with documented virologic failure or exposure to mono/dual NRTI therapy who were virologically suppressed on a PI/r-based regimen for at least 6 months on or after January 1, 2014 (n = 532). A marginal structural Cox model analysis was used to estimate the effect of the switch to dolutegravir on virologic outcome compared with remaining on PI/r. The outcome was defined as 2 consecutive viral loads (VLs) >50 copies/mL or 1 VL >50 copies/mL if it occurred at the last VL available. RESULTS: Among 532 eligible participants, 216 (40.6%) had their regimen switched to dolutegravir with 2 NRTIs, whereas 316 (59.4%) remained on the PI/r with 2 NRTIs. The weighted hazard ratio for the effect of dolutegravir switch on virologic failure compared with patients whose regimen remained on PI/r was 0.57 (95% confidence interval, 0.21-1.52). CONCLUSIONS: We did not find evidence of an increased risk for virologic failure after switching to dolutegravir from PI/r among patients with previous virologic failure or prior exposure to mono/dual NRTI.

Machine learning and dengue forecasting: Comparing random forests and artificial neural networks for predicting dengue burden at national and sub-national scales in Colombia.

The robust estimate and forecast capability of random forests (RF) has been widely recognized, however this ensemble machine learning method has not been widely used in mosquito-borne disease forecasting. In this study, two sets of RF models were developed at the national (pooled department-level data) and department level in Colombia to predict weekly dengue cases for 12-weeks ahead. A pooled national model based on artificial neural networks (ANN) was also developed and used as a comparator to the RF models. The various predictors included historic dengue cases, satellite-derived estimates for vegetation, precipitation, and air temperature, as well as population counts, income inequality, and education. Our RF model trained on the pooled national data was more accurate for department-specific weekly dengue cases estimation compared to a local model trained only on the department's data. Additionally, the forecast errors of the national RF model were smaller to those of the national pooled ANN model and were increased with the forecast horizon increasing from one-week-ahead (mean absolute error, MAE: 9.32) to 12-weeks ahead (MAE: 24.56). There was considerable variation in the relative importance of predictors dependent on forecast horizon. The environmental and meteorological predictors were relatively important for short-term dengue forecast horizons while socio-demographic predictors were relevant for longer-term forecast horizons. This study demonstrates the potential of RF in dengue forecasting with a feasible approach of using a national pooled model to forecast at finer spatial scales. Furthermore, including sociodemographic predictors is likely to be helpful in capturing longer-term dengue trends.