Causally estimating the effect of YouTube's recommender system using counterfactual bots.

In recent years, critics of online platforms have raised concerns about the ability of recommendation algorithms to amplify problematic content, with potentially radicalizing consequences. However, attempts to evaluate the effect of recommenders have suffered from a lack of appropriate counterfactuals-what a user would have viewed in the absence of algorithmic recommendations-and hence cannot disentangle the effects of the algorithm from a user's intentions. Here we propose a method that we call "counterfactual bots" to causally estimate the role of algorithmic recommendations on the consumption of highly partisan content on YouTube. By comparing bots that replicate real users' consumption patterns with "counterfactual" bots that follow rule-based trajectories, we show that, on average, relying exclusively on the YouTube recommender results in less partisan consumption, where the effect is most pronounced for heavy partisan consumers. Following a similar method, we also show that if partisan consumers switch to moderate content, YouTube's sidebar recommender "forgets" their partisan preference within roughly 30 videos regardless of their prior history, while homepage recommendations shift more gradually toward moderate content. Overall, our findings indicate that, at least since the algorithm changes that YouTube implemented in 2019, individual consumption patterns mostly reflect individual preferences, where algorithmic recommendations play, if anything, a moderating role.

Prevalence and Outcomes of Congenital Heart Disease in Very Low Birth Weight Preterm Infants: An Observational Study From the Brazilian Neonatal Network Database.

OBJECTIVES: To evaluate the prevalence of congenital heart disease and their outcomes in a Brazilian cohort of very low birth weight preterm infants. DESIGN: Post hoc analysis of data from the Brazilian Neonatal Network database, complemented by retrospective data from medical charts and a cross-sectional survey. SETTING: Twenty public tertiary-care university hospitals. PATIENTS: A total of 13,955 newborns weighing from 401 to 1,499 g and between 22 and 36 weeks of gestational age, born from 2010 to 2017. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: The prevalence of congenital heart disease was 2.45% (95% CI, 2.20-2.72%). In a multivariate regression analysis, risk factors associated with congenital heart disease were maternal diabetes (relative risk, 1.55; 95% CI, 1.11-2.20) and maternal age above 35 years (relative risk, 2.09; 95% CI, 1.73-2.51), whereas the protection factors were maternal hypertension (relative risk, 0.54; 95% CI, 0.43-0.69), congenital infection (relative risk, 0.45; 95% CI, 0.21-0.94), and multiple gestation (relative risk, 0.73; 95% CI, 0.55-0.97). The pooled standardized mortality ratio in patients with congenital heart disease was 2.48 (95% CI, 2.22-2.80), which was significantly higher than in patients without congenital heart disease (2.08; 95% CI, 2.03-2.13). However, in multiple log-binomial regression analyses, only the presence of major congenital anomaly, gestational age (< 29 wk; relative risk, 2.32; 95% CI, 2.13-2.52), and Score for Neonatal Acute Physiology and Perinatal Extension II (> 20; relative risk, 3.76; 95% CI, 3.41-4.14) were independently associated with death, whereas the effect of congenital heart disease was spotted only when a conditional inference tree approach was used. CONCLUSIONS: The overall prevalence of congenital heart disease in this cohort of very low birth weight infants was higher and with higher mortality than in the general population of live births. The occurrence of a major congenital anomaly, gestational age (< 29 wk), and Score for Neonatal Acute Physiology and Perinatal Extension II (> 20) were significantly and independently associated with death, whereas the association of congenital heart disease and death was only evident when a major congenital anomaly was present.

Tele-electrocardiography and bigdata: The CODE (Clinical Outcomes in Digital Electrocardiography) study.

Digital electrocardiographs are now widely available and a large number of digital electrocardiograms (ECGs) have been recorded and stored. The present study describes the development and clinical applications of a large database of such digital ECGs, namely the CODE (Clinical Outcomes in Digital Electrocardiology) study. ECGs obtained by the Telehealth Network of Minas Gerais, Brazil, from 2010 to 17, were organized in a structured database. A hierarchical free-text machine learning algorithm recognized specific ECG diagnoses from cardiologist reports. The Glasgow ECG Analysis Program provided Minnesota Codes and automatic diagnostic statements. The presence of a specific ECG abnormality was considered when both automatic and medical diagnosis were concordant; cases of discordance were decided using heuristisc rules and manual review. The ECG database was linked to the national mortality information system using probabilistic linkage methods. From 2,470,424 ECGs, 1,773,689 patients were identified. After excluding the ECGs with technical problems and patients <16 years-old, 1,558,415 patients were studied. High performance measures were obtained using an end-to-end deep neural network trained to detect 6 types of ECG abnormalities, with F1 scores >80% and specificity >99% in an independent test dataset. We also evaluated the risk of mortality associated with the presence of atrial fibrillation (AF), which showed that AF was a strong predictor of cardiovascular mortality and mortality for all causes, with increased risk in women. In conclusion, a large database that comprises all ECGs performed by a large telehealth network can be useful for further developments in the field of digital electrocardiography, clinical cardiology and cardiovascular epidemiology.

Evaluation of mortality in bundle branch block patients from an electronic cohort: Clinical Outcomes in Digital Electrocardiography (CODE) study.

BACKGROUND: Left bundle branch block is recognized as a marker of higher risk of death, but the prognostic value of the right bundle branch block in the general population is still controversial. Our aim is to evaluate the risk of overall and cardiovascular mortality in patients with right (RBBB) and left bundle branch block (LBBB) in a large electronic cohort of Brazilian patients. METHODS: This observational retrospective study was developed with the database of digital ECGs from Telehealth Network of Minas Gerais, Brazil (TNMG). All ECGs performed from 2010 to 2017 in primary care patients over 16 years old were assessed. The electronic cohort was obtained by linking data from ECG exams (name, sex, date of birth, city of residence) and those from national mortality information system, using standard probabilistic linkage methods (FRIL: Fine-grained record linkage software, v.2.1.5, Atlanta, GA). Only the first ECG of each patient was considered. Clinical data were self-reported, and ECGs were interpreted manually by cardiologists and automatically by the Glasgow University Interpreter software. Hazard ratio (HR) for mortality was estimated using Cox regression. RESULTS: From a dataset of 1,773,689 patients, 1,558,421 primary care patients over 16 years old underwent a valid ECG recording during 2010 to 2017. We excluded 17,359 patients that didn't have a valid QRS measure from the Glasgow program and 11,091 patients from the control group that had QRS equal or above 120 ms and were not RBBB or LBBB. Therefore, 1,529,971 were included (median age 52 [Q1:38; Q3:65] years; 40.2% were male). In a mean follow-up of 3.7 years, the overall mortality rate was 3.34%. RBBB was more frequent (2.42%) than LBBB (1.32%). In multivariate analysis, adjusting for sex, age and comorbidities, both patients with RBBB (HR 1.32; CI 95% 1.27-1.37) and LBBB (HR 1.69; CI 95% 1.62-1.76) had higher risk of overall mortality. Women with RBBB had an increased risk of all-cause death compared to men (p < 0.001). Cardiovascular mortality was higher in patients with LBBB (HR 1.77; CI 95% 1.55-2.01), but not for RBBB. CONCLUSIONS: Patients with RBBB and LBBB had higher risk of overall mortality. Women with RBBB had more risk of all-cause death than men. LBBB was associated with higher risk of cardiovascular mortality.

Early Empiric Antibiotic Use Is Associated With Delayed Feeding Tolerance in Preterm Infants: A Retrospective Analysis.

The causative factors of neonatal feeding intolerance are poorly understood, but potentially related to clinical practices such as empiric antibiotic usage. The objective of this study was to evaluate whether early empiric antibiotic exposure negatively affects preterm infants' enteral feeding tolerance. Data from infants without risk factors for sepsis, 500 to 1499 g birth weight and 24 to 34 weeks gestational age were analyzed. The primary outcomes were the empiric antibiotic exposure effects on the infants' total parenteral nutrition usage duration and prevalence of necrotizing enterocolitis (NEC). Among the 901 infants included, 67 were exposed to early empiric antibiotic. A 50% increase in parenteral nutrition usage duration and a 4-fold greater prevalence of NEC was seen in the early empiric antibiotic-exposed neonates, when compared with control infants (P < 0.01). Early empiric antibiotic exposure appears to negatively influence preterm infant feeding tolerance and possibly contributes to NEC.

High HIV Prevalence among Men who have Sex with Men in a Time-Location Sampling Survey, São Paulo, Brazil.

We conducted a time-location sampling sero-behavioral surveillance survey of men who have sex with men (MSM) in São Paulo, Brazil, the largest city in Latin America and the Southern Hemisphere (N = 1,217 interviewed with serological results for 771). HIV prevalence was 15.4 % (95 % CI 11.6-20.1), with only 45.8 % previously aware of their infection. HIV prevalence achieved 6.4 % among youth 18-24 years and was higher among MSM with lower socio-economic status. In multivariate analysis, correlates of HIV were older age, gay identity, lower socio-economic status, social networks with HIV-positive MSM, receptive anal sex, and internet sex partners. Policy change towards increasing avenues for HIV testing coupled with antiretroviral treatment regardless of CD4 count or stage of disease stand to benefit the MSM community if scaled up fast enough.

Bronchopulmonary dysplasia: temporal trend from 2010 to 2019 in the Brazilian Network on Neonatal Research.

OBJECTIVE: To evaluate the temporal trend of bronchopulmonary dysplasia (BPD) in preterm infants who survived to at least 36 weeks' post-menstrual age (PMA) and BPD or death at 36 weeks' PMA, and to analyse variables associated with both outcomes. DESIGN: Retrospective cohort with data retrieved from an ongoing national registry. SETTING: 19 Brazilian university public hospitals. PATIENTS: Infants born between 2010 and 2019 with 23-31 weeks and birth weight 400-1499 g. MAIN OUTCOME MEASURES: Temporal trend was evaluated by Prais-Winsten model and variables associated with BPD in survivors or BPD or death were analysed by logistic regression. RESULTS: Of the 11 128 included infants, BPD in survivors occurred in 22%, being constant over time (annual per cent change (APC): -0.80%; 95% CI: -2.59%; 1.03%) and BPD or death in 45%, decreasing over time (APC: -1.05%; 95% CI: -1.67%; -0.43%). Being male, small for gestational age, presenting with respiratory distress syndrome, air leaks, needing longer duration of mechanical ventilation, presenting with treated patent ductus arteriosus and late-onset sepsis were associated with an increase in the chance of BPD. For the outcome BPD or death, maternal bleeding, multiple gestation, 5-minute Apgar <7, late-onset sepsis, necrotising enterocolitis and intraventricular haemorrhage were added to the variables reported above as increasing the chance of the outcome. CONCLUSION: The frequency of BPD in survivors was constant and BPD or death decreased by 1.05% at each study year. These results show some improvement in perinatal care in Brazilian units which resulted in a reduction of BPD or death, but further improvements are still needed to reduce BPD in survivors.

Deplatforming did not decrease Parler users' activity on fringe social media.

Online platforms have banned ("deplatformed") influencers, communities, and even entire websites to reduce content deemed harmful. Deplatformed users often migrate to alternative platforms, which raises concerns about the effectiveness of deplatforming. Here, we study the deplatforming of Parler, a fringe social media platform, between 2020 January 11 and 2021 February 25, in the aftermath of the US Capitol riot. Using two large panels that capture longitudinal user-level activity across mainstream and fringe social media content (N = 112, 705, adjusted to be representative of US desktop and mobile users), we find that other fringe social media, such as Gab and Rumble, prospered after Parler's deplatforming. Further, the overall activity on fringe social media increased while Parler was offline. Using a difference-in-differences analysis (N = 996), we then identify the causal effect of deplatforming on active Parler users, finding that deplatforming increased the probability of daily activity across other fringe social media in early 2021 by 10.9 percentage points (pp) (95% CI [5.9 pp, 15.9 pp]) on desktop devices, and by 15.9 pp (95% CI [10.2 pp, 21.7 pp]) on mobile devices, without decreasing activity on fringe social media in general (including Parler). Our results indicate that the isolated deplatforming of a major fringe platform was ineffective at reducing overall user activity on fringe social media.

Measuring adherence to antiretroviral treatment: the role of pharmacy records of drug withdrawals.

This study aimed to evaluate adherence to antiretroviral treatment (ART) among HIV + adults, assess its association with HIV viral load (VL) and identify factors associated to adherence. A survey involving a random sample of adults followed at a HIV/AIDS reference center in São Paulo city, Brazil, from 2007 to 2009 was done. A questionnaire was applied and data were retrieved from the pharmacy and medical records. The study involved 292 subjects: 70.2% men; median age: 43 years; median duration of ART: 8 years. 89.3% self-reported taken all prescribed pills in the last 3 days but only 39.3% picked up ≥95% of the prescribed ART from the pharmacy in the last 12 months. At the multivariate analysis having symptoms prior to ART, taking fewer ART pills, and not missing medical appointments were independently associated to higher adherence. Adherence was strongly associated with undetectable HIV VL. Rates of undetectable HIV VL did not differ from 80 to ≥95% of adherence.

Volunteer contributions to Wikipedia increased during COVID-19 mobility restrictions.

Wikipedia, the largest encyclopedia ever created, is a global initiative driven by volunteer contributions. When the COVID-19 pandemic broke out and mobility restrictions ensued across the globe, it was unclear whether contributions to Wikipedia would decrease in the face of the pandemic, or whether volunteers would withstand the added stress and increase their contributions to accommodate the growing readership uncovered in recent studies. We analyze [Formula: see text] million edits contributed from 2018 to 2020 across twelve Wikipedia language editions and find that Wikipedia's global volunteer community responded resiliently to the pandemic, substantially increasing both productivity and the number of newcomers who joined the community. For example, contributions to the English Wikipedia increased by over [Formula: see text] compared to the expectation derived from pre-pandemic data. Our work sheds light on the response of a global volunteer population to the COVID-19 crisis, providing valuable insights into the behavior of critical online communities under stress.

Popular knowledge of stroke in São Paulo: a cross-sectional study within the World Stroke Campaign.

BACKGROUND: Stroke is the second leading cause of death in Brazil and the main cause of disability. Inability to identify alarm signals causes delays in seeking emergency services, thereby leading to a worse prognosis. OBJECTIVES: To assess the population's knowledge of how to recognize and prevent stroke. DESIGN AND SETTING: Prospective cross-sectional study on data derived from a questionnaire that was administered during the 2016 World Stroke Campaign, launched in the city of São Paulo, Brazil. METHODS: Data on 806 interviewees were evaluated using descriptive statistics and univariate and multivariate analyses. RESULTS: Among all the interviewees, 52.1% knew how to conceptualize stroke; 70.07% knew someone who had suffered a stroke; and 29.03% listed three or more risk factors. Only 27.5% mentioned controlling high blood pressure as a preventive measure. In the event of witnessing a stroke, 57.8% would call the emergency service and 2.9% would check the timing. Less educated individuals were 5.6 times more likely (95% confidence interval, CI 3.45-9.02) to have poor knowledge of stroke, compared with the more educated group. Knowing someone who had had a stroke reduced the chances of not knowing the terms relating to the disease (odds ratio, OR = 0.56; 95% CI 0.4-0.78). CONCLUSIONS: Despite the severity and prevalence of stroke, the population still has little information on this disease. In this context, the importance of mounting campaigns to improve prevention and treatment and to contribute to healthcare policies becomes evident.

Deep neural network-estimated electrocardiographic age as a mortality predictor.

The electrocardiogram (ECG) is the most commonly used exam for the evaluation of cardiovascular diseases. Here we propose that the age predicted by artificial intelligence (AI) from the raw ECG (ECG-age) can be a measure of cardiovascular health. A deep neural network is trained to predict a patient's age from the 12-lead ECG in the CODE study cohort (n = 1,558,415 patients). On a 15% hold-out split, patients with ECG-age more than 8 years greater than the chronological age have a higher mortality rate (hazard ratio (HR) 1.79, p < 0.001), whereas those with ECG-age more than 8 years smaller, have a lower mortality rate (HR 0.78, p < 0.001). Similar results are obtained in the external cohorts ELSA-Brasil (n = 14,236) and SaMi-Trop (n = 1,631). Moreover, even for apparent normal ECGs, the predicted ECG-age gap from the chronological age remains a statistically significant risk predictor. These results show that the AI-enabled analysis of the ECG can add prognostic information.

Health-related factors correlate with behavior trends in physical activity level in old age: longitudinal results from a population in São Paulo, Brazil.

BACKGROUND: Physical inactivity in leisure time is common among elderly in Brazil and this finding is particularly alarming considering that this population is greatly affected by chronic diseases. The identification of health factors that influence physical activity level (PAL) will help in the development of strategies for increasing PAL older adults. The current research aimed to identify variables that independently affect behavior trends in PAL over the course of two years among elderly. METHODS: A survey entitled the Epidoso Project ("Epidemiology of aging") studied 1,667 community-based older individuals in São Paulo city, Brazil over the course of two years. Physical activity level was determined through questions about frequency and duration of physical activities. Body Mass Index was calculated; functional capacity was assessed through the ADL (activities of daily living) scale; cognition was assessed by Mini-Mental State Examination; and mental health was assessed through the Dysthymia Screening. Experiences of falls and fractures were also assessed. Subjects were divided into three groups according to their self-report of Physical Activity Level: a--Regularly Active; b--Insufficiently Active and c--Physically Inactive. Behavior trends in PAL were also measured after two years. Multivariate regression model methodology was used to test associations longitudinally. RESULTS: Results from the final model demonstrated that the risk of a not favorable behavior trend in PAL, which included the group who remained physically inactive and the group that displayed decreased PAL, in this cohort of older adults was significantly increased if the individual was female (OR = 2.50; 95% CI = 1.60-3.89; P < 0.01), older (80 y vs. 65 y, OR = 6.29, 95% CI = 2.69-14.67; P < 0.01), dependent on help from others for activities in the ADL scale (moderate-severe = 4-7+ vs. 0 ADLs) (OR = 2.25, 95% CI = 1.20-4.21; P < 0.011) or had experienced a history of falls with consequences (OR = 6.88, 95% CI = 0.91-52.01; P < 0.062). CONCLUSIONS: Age, gender, ADL scores and falls were associated with a not favorable behavior trend in PAL. Promotion programs should target these factors, reducing barriers to achieve desired changes in PAL.

TEMPORAL TREND OF HOSPITALIZATIONS FOR ACUTE BRONCHIOLITIS IN INFANTS UNDER ONE YEAR OF AGE IN BRAZIL BETWEEN 2008 AND 2015.

OBJECTIVE: To evaluate the trend of hospitalization for acute bronchiolitis in infants under one year of age, in the past eight years and after the implementation of the palivizumab immunization program in Brazil. METHODS: The study is a retrospective analysis of data on infants younger than one year of age, who were hospitalized with acute bronchiolitis between 2008 and 2015 in Brazil. The Brazilian National Health System database was used. The rates of hospitalization in the pre-implementation (2008-2012) and post-implementation (2014-2015) periods of the palivizumab immunization program were evaluated. The total number of admissions in the same period was used as a comparison. RESULTS: Between January 2008 and December 2015, 263,679 hospitalizations for bronchiolitis were recorded in infants younger than one year of age, 60% represented by boys. The incidence of hospitalization for bronchiolitis increased by 49% over this period (8.5 to 12.7 per 1,000 inhabitants per year). Between 2013 and 2014, the incidence rate of hospitalization for acute bronchiolitis decreased by 8% (12.5 to 11.5 per 1,000 inhabitants per year). However, in the second year of the program, hospitalization rate increased again by 10% (12.7 per 1,000 inhabitants per years). CONCLUSIONS: Acute bronchiolitis presented increasing rates of hospitalization over the study period. Hospitalization incidence for acute bronchiolitis declined one year after the implementation of palivizumab but increased again in the second year of the program.

Automatic diagnosis of the 12-lead ECG using a deep neural network.

The role of automatic electrocardiogram (ECG) analysis in clinical practice is limited by the accuracy of existing models. Deep Neural Networks (DNNs) are models composed of stacked transformations that learn tasks by examples. This technology has recently achieved striking success in a variety of task and there are great expectations on how it might improve clinical practice. Here we present a DNN model trained in a dataset with more than 2 million labeled exams analyzed by the Telehealth Network of Minas Gerais and collected under the scope of the CODE (Clinical Outcomes in Digital Electrocardiology) study. The DNN outperform cardiology resident medical doctors in recognizing 6 types of abnormalities in 12-lead ECG recordings, with F1 scores above 80% and specificity over 99%. These results indicate ECG analysis based on DNNs, previously studied in a single-lead setup, generalizes well to 12-lead exams, taking the technology closer to the standard clinical practice.

Author Correction: Automatic diagnosis of the 12-lead ECG using a deep neural network.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Evaluation of Mortality in Atrial Fibrillation: Clinical Outcomes in Digital Electrocardiography (CODE) Study.

Aims: Atrial fibrillation (AF) is a public health problem and its prevalence is increasing worldwide. Electronic cohorts, with large electrocardiogram (ECG) databases linked to mortality data, can be useful in determining prognostic value of ECG abnormalities. Our aim is to evaluate the risk of mortality in patients with AF from Brazil. Methods: This observational retrospective study of primary care patients was developed with the digital ECG database from the Telehealth Network of Minas Gerais, Brazil. ECGs performed from 2010 to 2017 were interpreted by cardiologists and the University of Glasgow automated analysis software. An electronic cohort was obtained linking data from ECG exams and those from a national mortality information system, using standard probabilistic linkage methods. We considered only the first ECG of each patient. Patients under 16 years were excluded. Hazard ratios (HR) for mortality were adjusted for demographic and self-reported clinical factors and estimated with Cox regression. Results: From a dataset of 1,773,689 patients, 1,558,421 were included, mean age 51.6 years; 40.2% male. There were 3.34% deaths from all causes in 3.68 years of median follow up. The prevalence of AF was 1.33%. AF was an independent risk factor for all-cause mortality (HR 2.10, 95%CI 2.03-2.17) and cardiovascular mortality (HR 2.06, 95%CI 1.86-2.29). Females with AF had a higher risk of overall and cardiovascular mortality compared with males (p < 0.001). Conclusions: AF was a strong predictor of cardiovascular and all-cause mortality in a primary care population, with increased risk in women. Condensed abstract: To assess risk of mortality in AF patients, an electronic cohort was obtained linking data from ECG exams of Brazilian primary care patients and a national mortality information system. From 1,558,421 patients, AF (prevalence 1.33%) carried a higher risk of overall and cardiovascular mortality, with increased risk in women. What's New: This is the first study with a large Brazilian electronic cohort to evaluate the risk of mortality linked to AF in primary care patients.AF patients from a Brazilian primary care population had a higher risk of death for all causes (HR 2.10, 95%CI 2.03-2.17) and cardiovascular mortality (HR 2.06, 95%CI 1.86-2.29).Female patients with AF had an increased risk of overall and cardiovascular mortality compared with male patients (p < 0.001).

Rapid diagnosis of respiratory distress syndrome by oral aspirate in premature newborns.

OBJECTIVE: The stable microbubble test on gastric aspirate and on amniotic fluid has been used for the diagnosis of respiratory distress syndrome in the newborn. However, no study has performed this test on oral aspirates from premature infants. The objective of this study was to evaluate the performance of the stable microbubble test on oral aspirates from preterm newborns to predict respiratory distress syndrome. METHOD: This study included infants with gestational age <34 weeks. Oral fluids were obtained immediately after birth and gastric fluids were collected within the first 30 minutes of life. The samples were frozen and tested within 72 hours. RESULTS: The sample was composed of paired aspirates from 64 newborns, who were divided into two groups: respiratory distress syndrome group (n=21) and control group (n=43). The median (interquartile range) of the stable microbubble count in the oral samples of infants with respiratory distress syndrome was significantly lower than that of infants who did not develop respiratory symptoms: respiratory distress syndrome group=12 (8-22) stable microbubbles/mm2; control group=100 (48-230)microbubbles/mm2 (p<0.001). The correlation between microbubble count in gastric and oral aspirates was 0.90 (95% confidence interval=0.85-0.95; p<0.001). Considering a cut-off point of 25microbubbles/mm2, the sensitivity and the specificity of the stable microbubble test were 81.4% and 85.7%, respectively. CONCLUSION: The study suggests that the stable microbubble test performed on oral aspirate is a reliable alternative to that performed on gastric fluid for the prediction of respiratory distress syndrome in the newborn.

Author Correction: Resistance training regulates gene expression of molecules associated with intramyocellular lipids, glucose signaling and fiber size in old rats.

A correction to this article has been published and is linked from the HTML and PDF versions of this paper. The error has been fixed in the paper.

Using sexually transmitted infection biomarkers to validate reporting of sexual behavior within a randomized, experimental evaluation of interviewing methods.

This paper examines the reporting of sexual and other risk behaviors within a randomized experiment using a computerized versus face-to-face interview mode. Biomarkers for sexually transmitted infection (STI) were used to validate self-reported behavior by interview mode. As part of a parent study evaluating home versus clinic screening and diagnosis for STIs, 818 women aged 18-40 years were recruited in 2004 at or near a primary care clinic in São Paulo, Brazil, and were randomized to a face-to-face interview or audio computer-assisted self-interviewing. Ninety-six percent of participants were tested for chlamydia, gonorrhea, and trichomoniasis. Reporting of STI risk behavior was consistently higher with the computerized mode of interview. Stronger associations between risk behaviors and STI were found with the computerized interview after controlling for sociodemographic factors. These results were obtained by using logistic regression approaches, as well as statistical methods that address potential residual confounding and covariate endogeneity. Furthermore, STI-positive participants were more likely than STI-negative participants to underreport risk behavior in the face-to-face interview. Results strongly suggest that computerized interviewing provides more accurate and reliable behavioral data. The analyses also confirm the benefits of using data on prevalent STIs for externally validating behavioral reporting.