Semiparametric marginal regression for clustered competing risks data with missing cause of failure.

Clustered competing risks data are commonly encountered in multicenter studies. The analysis of such data is often complicated due to informative cluster size (ICS), a situation where the outcomes under study are associated with the size of the cluster. In addition, the cause of failure is frequently incompletely observed in real-world settings. To the best of our knowledge, there is no methodology for population-averaged analysis with clustered competing risks data with an ICS and missing causes of failure. To address this problem, we consider the semiparametric marginal proportional cause-specific hazards model and propose a maximum partial pseudolikelihood estimator under a missing at random assumption. To make the latter assumption more plausible in practice, we allow for auxiliary variables that may be related to the probability of missingness. The proposed method does not impose assumptions regarding the within-cluster dependence and allows for ICS. The asymptotic properties of the proposed estimators for both regression coefficients and infinite-dimensional parameters, such as the marginal cumulative incidence functions, are rigorously established. Simulation studies show that the proposed method performs well and that methods that ignore the within-cluster dependence and the ICS lead to invalid inferences. The proposed method is applied to competing risks data from a large multicenter HIV study in sub-Saharan Africa where a significant portion of causes of failure is missing.

Correcting mortality estimates among children and youth on antiretroviral therapy in southern Africa: A comparative analysis between a multi-country tracing study and linkage to a health information exchange.

OBJECTIVES: The objective of this study is to assess the outcomes of children, adolescents and young adults with HIV reported as lost to follow-up, correct mortality estimates for children, adolescents and young adults with HIV for unascertained outcomes in those loss to follow-up (LTFU) based on tracing and linkage data separately using data from the International epidemiology Databases to Evaluate AIDS in Southern Africa. METHODS: We included data from two different populations of children, adolescents and young adults with HIV; (1) clinical data from children, adolescents and young adults with HIV aged ≤24 years from Lesotho, Malawi, Mozambique, Zambia and Zimbabwe; (2) clinical data from children, adolescents and young adults with HIV aged ≤14 years from the Western Cape (WC) in South Africa. Outcomes of patients lost to follow-up were available from (1) a tracing study and (2) linkage to a health information exchange. For both populations, we compared six methods for correcting mortality estimates for all children, adolescents and young adults with HIV. RESULTS: We found substantial variations of mortality estimates among children, adolescents and young adults with HIV reported as lost to follow-up versus those retained in care. Ascertained mortality was higher among lost and traceable children, adolescents and young adults with HIV and lower among lost and linkable than those retained in care (mortality: 13.4% [traced] vs. 12.6% [retained-other Southern Africa countries]; 3.4% [linked] vs. 9.4% [retained-WC]). A high proportion of lost to follow-up children, adolescents and young adults with HIV had self-transferred (21.0% and 47.0%) in the traced and linked samples, respectively. The uncorrected method of non-informative censoring yielded the lowest mortality estimates among all methods for both tracing (6.0%) and linkage (4.0%) approaches at 2 years from ART start. Among corrected methods using ascertained data, multiple imputation, incorporating ascertained data (MI(asc.)) and inverse probability weighting with logistic weights were most robust for the tracing approach. In contrast, for the linkage approach, MI(asc.) was the most robust. CONCLUSIONS: Our findings emphasise that lost to follow-up is non-ignorable and both tracing and linkage improved outcome ascertainment: tracing identified substantial mortality in those reported as lost to follow-up, whereas linkage did not identify out-of-facility deaths, but showed that a large proportion of those reported as lost to follow-up were self-transfers.

Characterizing participants who respond to text, email, phone calls, or postcards in a SARS-CoV-2 prevalence study.

INTRODUCTION: Multiple modalities and frequencies of contact are needed to maximize recruitment in many public health surveys. The purpose of this analysis is to characterize respondents to a statewide SARS-CoV-2 testing study whose participation followed either postcard, phone outreach or electronic means of invitation. In addition, we examine how participant characteristics differ based upon the number of contacts needed to elicit participation. METHODS: This is a cross-sectional analysis of survey data collected from participants who were randomly selected to represent Indiana residents and were invited to be tested for Covid-19 in April 2020. Participants received invitations via postcard, text/emails, and/or robocalls/texts based upon available contact information. The modality, and frequency of contacts, that prompted participation was determined by when the notification was sent and when the participant responded and subsequently registered to participate in the study. Chi square analyses were used to determine differences between groups and significant findings were analyzed using multinomial logistic regression. RESULTS: Respondents included 3,658 individuals and were stratified by postcards (7.9%), text/emails (26.5%), and robocalls/text (65.7%) with 19.7% registering after 1 contact, 47.9% after 2 contacts, and 32.4% after 3 contacts encouraging participation. Females made up 54.6% of the sample and responded at a higher rate for postcards (8.2% vs. 7.5%) and text/emails (28.1 vs. 24.6%) as compared to males (χ2 = 7.43, p = 0.025). Compared to males, females responded at a higher percentage after 1 contact (21.4 vs. 17.9%, χ2 = 7.6, p = 0.023). Those over 60 years responded most often after 2 contacts (χ2 = 27.5, p < 0.001) when compared to others at younger age groups. In regression analysis, participant sex (p = 0.036) age (p = 0.005), educational attainment (p = < 0.0001), and being motivated by "free testing" (p = 0.036) were correlated with participation in the prevalence study. DISCUSSION: Researchers should be aware that the modality of contact as well as the number of prompts used could influence differential participation in public health studies. Our findings can inform researchers developing studies that rely on selective participation by study subjects. We explore how to increase participation within targeted demographic groups using specific modalities and examining frequency of contact.

Bayesian estimation of SARS-CoV-2 prevalence in Indiana by random testing.

From 25 to 29 April 2020, the state of Indiana undertook testing of 3,658 randomly chosen state residents for the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus, the agent causing COVID-19 disease. This was the first statewide randomized study of COVID-19 testing in the United States. Both PCR and serological tests were administered to all study participants. This paper describes statistical methods used to address nonresponse among various demographic groups and to adjust for testing errors to reduce bias in the estimates of the overall disease prevalence in Indiana. These adjustments were implemented through Bayesian methods, which incorporated all available information on disease prevalence and test performance, along with external data obtained from census of the Indiana statewide population. Both adjustments appeared to have significant impact on the unadjusted estimates, mainly due to upweighting data in study participants of non-White races and Hispanic ethnicity and anticipated false-positive and false-negative test results among both the PCR and antibody tests utilized in the study.

Same-Day Antiretroviral Therapy Initiation as a Predictor of Loss to Follow-up and Viral Suppression Among People With Human Immunodeficiency Virus in Sub-Saharan Africa.

BACKGROUND: Treat-All guidelines recommend initiation of antiretroviral therapy (ART) for all people with HIV (PWH) on the day of diagnosis when possible, yet uncertainty exists about the impact of same-day ART initiation on subsequent care engagement. We examined the association of same-day ART initiation with loss to follow-up and viral suppression among patients in 11 sub-Saharan African countries. METHODS: We included ART-naive adult PWH from sites participating in the International epidemiology Databases to Evaluate AIDS (IeDEA) consortium who enrolled in care after Treat-All implementation and prior to January 2019. We used multivariable Cox regression to estimate the association between same-day ART initiation and loss to follow-up and Poisson regression to estimate the association between same-day ART initiation and 6-month viral suppression. RESULTS: Among 29 017 patients from 63 sites, 18 584 (64.0%) initiated ART on the day of enrollment. Same-day ART initiation was less likely among those with advanced HIV disease versus early-stage disease. Loss to follow-up was significantly lower among those initiating ART ≥1 day of enrollment, compared with same-day ART initiators (20.6% vs 27.7%; adjusted hazard ratio: .66; 95% CI .57-.76). No difference in viral suppression was observed by time to ART initiation (adjusted rate ratio: 1.00; 95% CI: .98-1.02). CONCLUSIONS: Patients initiating ART on the day of enrollment were more frequently lost to follow-up than those initiating later but were equally likely to be virally suppressed. Our findings support recent World Health Organization recommendations for providing tailored counseling and support to patients who accept an offer of same-day ART.

The Effect of HIV Treatment Interruption on Subsequent Immunological Response.

Recovery of CD4-positive T lymphocyte count after initiation of antiretroviral therapy (ART) has been thoroughly examined among people with human immunodeficiency virus infection. However, immunological response after restart of ART following care interruption is less well studied. We compared CD4 cell-count trends before disengagement from care and after ART reinitiation. Data were obtained from the East Africa International Epidemiology Databases to Evaluate AIDS (IeDEA) Collaboration (2001-2011; n = 62,534). CD4 cell-count trends before disengagement, during disengagement, and after ART reinitiation were simultaneously estimated through a linear mixed model with 2 subject-specific knots placed at the times of disengagement and treatment reinitiation. We also estimated CD4 trends conditional on the baseline CD4 value. A total of 10,961 patients returned to care after disengagement from care, with the median gap in care being 2.7 (interquartile range, 2.1-5.4) months. Our model showed that CD4 cell-count increases after ART reinitiation were much slower than those before disengagement. Assuming that disengagement from care occurred 12 months after ART initiation and a 3-month treatment gap, CD4 counts measured at 3 years since ART initiation would be lower by 36.5 cells/µL than those obtained under no disengagement. Given that poorer CD4 restoration is associated with increased mortality/morbidity, specific interventions targeted at better retention in care are urgently required.

Joint modeling of longitudinal and competing-risk data using cumulative incidence functions for the failure submodels accounting for potential failure cause misclassification through double sampling.

Most of the literature on joint modeling of longitudinal and competing-risk data is based on cause-specific hazards, although modeling of the cumulative incidence function (CIF) is an easier and more direct approach to evaluate the prognosis of an event. We propose a flexible class of shared parameter models to jointly model a normally distributed marker over time and multiple causes of failure using CIFs for the survival submodels, with CIFs depending on the "true" marker value over time (i.e., removing the measurement error). The generalized odds rate transformation is applied, thus a proportional subdistribution hazards model is a special case. The requirement that the all-cause CIF should be bounded by 1 is formally considered. The proposed models are extended to account for potential failure cause misclassification, where the true failure causes are available in a small random sample of individuals. We also provide a multistate representation of the whole population by defining mutually exclusive states based on the marker values and the competing risks. Based solely on the assumed joint model, we derive fully Bayesian posterior samples for state occupation and transition probabilities. The proposed approach is evaluated in a simulation study and, as an illustration, it is fitted to real data from people with HIV.

Semiparametric regression on cumulative incidence function with interval-censored competing risks data and missing event types.

Competing risk data are frequently interval-censored, that is, the exact event time is not observed but only known to lie between two examination time points such as clinic visits. In addition to interval censoring, another common complication is that the event type is missing for some study participants. In this article, we propose an augmented inverse probability weighted sieve maximum likelihood estimator for the analysis of interval-censored competing risk data in the presence of missing event types. The estimator imposes weaker than usual missing at random assumptions by allowing for the inclusion of auxiliary variables that are potentially associated with the probability of missingness. The proposed estimator is shown to be doubly robust, in the sense that it is consistent even if either the model for the probability of missingness or the model for the probability of the event type is misspecified. Extensive Monte Carlo simulation studies show good performance of the proposed method even under a large amount of missing event types. The method is illustrated using data from an HIV cohort study in sub-Saharan Africa, where a significant portion of events types is missing. The proposed method can be readily implemented using the new function ciregic_aipw in the R package intccr.

Factors that differentiate COVID-19 vaccine intentions among Indiana parents: Implications for targeted vaccine promotion.

Given low rates of uptake of the COVID-19 vaccine for children 12-17 and 5-11 years old, research is needed to understand parental behaviors and behavioral intentions related to COVID-19 vaccination for their children. In the state of Indiana, we conducted a non-random, online survey of parents or caregivers (N = 10,266) about their COVID-19 vaccine intentions or behaviors, demographic characteristics, and potential motivating reasons for getting the vaccine. In terms of behaviors/intentions, 44.8% of participants indicated they were vaccine acceptors (i.e., had already had their children vaccinated or would as soon as it was possible), 13.0% indicated they were vaccine hesitators (i.e., wanted to wait and see), and 42.2% indicated they were vaccine rejecters (i.e., would not vaccinate or only would if mandated). Compared to vaccine rejecters, vaccine hesitators were more likely to be motivated by perceptions of vaccine safety and efficacy, normative influences such as close friends/family who had been vaccinated and a recommendation from a provider, as well as if they were vaccinated themselves. These findings have implications for the development of targeted vaccine promotion strategies, such as social norms messaging and a focus on vaccine safety, in order to increase COVID-19 vaccination for eligible children.

Effects of National Adoption of Treat-All Guidelines on Pre-Antiretroviral Therapy (ART) CD4 Testing and Viral Load Monitoring After ART initiation: A Regression Discontinuity Analysis.

BACKGROUND: The World Health Organization's Treat-All guidance recommends CD4 testing before initiating antiretroviral therapy (ART), and routine viral load (VL) monitoring (over CD4 monitoring) for patients on ART. METHODS: We used regression discontinuity analyses to estimate changes in CD4 testing and VL monitoring among 547 837 ART-naive patients enrolling in human immunodeficiency virus (HIV) care during 2006-2018 at 225 clinics in 26 countries where Treat-All policies were adopted. We examined CD4 testing within 12 months before and VL monitoring 6 months after ART initiation among adults (≥20 years), adolescents (10-19 years), and children (0-9 years) in low/lower-middle-income countries (L/LMICs) and high/upper-middle-income countries (H/UMICs). RESULTS: Treat-All adoption led to an immediate decrease in pre-ART CD4 testing among adults in L/LMICs, from 57.0% to 48.1% (-8.9 percentage points [pp]; 95% CI: -11.0, -6.8), and a small increase in H/UMICs, from 90.1% to 91.7% (+1.6pp; 95% CI: 0.2, 3.0), with no changes among adolescents or children; decreases in pre-ART CD4 testing accelerated after Treat-All adoption in L/LMICs. In L/LMICs, VL monitoring after ART initiation was low among all patients in L/LMICs before Treat-All; while there was no immediate change at Treat-All adoption, VL monitoring trends significantly increased afterwards. VL monitoring increased among adults immediately after Treat-All adoption, from 58.2% to 61.1% (+2.9pp; 95% CI: 0.5, 5.4), with no significant changes among adolescents/children. CONCLUSIONS: While on-ART VL monitoring has improved in L/LMICs, Treat-All adoption has accelerated and disparately worsened suboptimal pre-ART CD4 monitoring, which may compromise care outcomes for individuals with advanced HIV.

Pregnancies among women living with HIV using contraceptives and antiretroviral therapy in western Kenya: a retrospective, cohort study.

BACKGROUND: Preventing unintended pregnancies is paramount for women living with HIV (WLHIV). Previous studies have suggested that efavirenz-containing antiretroviral therapy (ART) reduces contraceptive effectiveness of implants, but there are uncertainties regarding the quality of the electronic medical record (EMR) data used in these prior studies. METHODS: We conducted a retrospective, cohort study of EMR data from 2011 to 2015 among WLHIV of reproductive age accessing HIV care in public facilities in western Kenya. We validated a large subsample of records with manual chart review and telephone interviews. We estimated adjusted incidence rate ratios (aIRRs) with Poisson regression accounting for the validation sampling using inverse probability weighting and generalized raking. RESULTS: A total of 85,324 women contributed a total of 170,845 women-years (w-y) of observation time; a subset of 5080 women had their charts reviewed, and 1285 underwent interviews. Among implant users, the aIRR of pregnancy for efavirenz- vs. nevirapine-containing ART was 1.9 (95% CI 1.6, 2.4) using EMR data only and 3.2 (95% CI 1.8, 5.7) when additionally using both chart review and interview validated data. Among efavirenz users, the aIRR of pregnancy for depomedroxyprogesterone acetate (DMPA) vs. implant use was 1.8 (95% CI 1.5, 2.1) in EMR only and 2.4 (95% CI 1.0, 6.1) using validated data. CONCLUSION: Pregnancy rates are higher when contraceptive implants are concomitantly used with efavirenz-containing ART, though rates were similar to leading alternative contraceptive methods such as DMPA. Our data provides policymakers, program staff, and WLHIV greater confidence in guiding their decision-making around contraceptive and ART options. Our novel, 3-phase validation sampling provides an innovative tool for using routine EMR data to improve the robustness of data quality.

How Many SARS-CoV-2-Infected People Require Hospitalization? Using Random Sample Testing to Better Inform Preparedness Efforts.

CONTEXT: Existing hospitalization ratios for COVID-19 typically use case counts in the denominator, which problematically underestimates total infections because asymptomatic and mildly infected persons rarely get tested. As a result, surge models that rely on case counts to forecast hospital demand may be inaccurately influencing policy and decision-maker action. OBJECTIVE: Based on SARS-CoV-2 prevalence data derived from a statewide random sample (as opposed to relying on reported case counts), we determine the infection-hospitalization ratio (IHR), defined as the percentage of infected individuals who are hospitalized, for various demographic groups in Indiana. Furthermore, for comparison, we show the extent to which case-based hospitalization ratios, compared with the IHR, overestimate the probability of hospitalization by demographic group. DESIGN: Secondary analysis of statewide prevalence data from Indiana, COVID-19 hospitalization data extracted from a statewide health information exchange, and all reported COVID-19 cases to the state health department. SETTING: State of Indiana as of April 30, 2020. MAIN OUTCOME MEASURES: Demographic-stratified IHRs and case-hospitalization ratios. RESULTS: The overall IHR was 2.1% and varied more by age than by race or sex. Infection-hospitalization ratio estimates ranged from 0.4% for those younger than 40 years to 9.2% for those older than 60 years. Hospitalization rates based on case counts overestimated the IHR by a factor of 10, but this overestimation differed by demographic groups, especially age. CONCLUSIONS: In this first study of the IHR based on population prevalence, our results can improve forecasting models of hospital demand-especially in preparation for the upcoming winter period when an increase in SARS CoV-2 infections is expected.

Population Point Prevalence of SARS-CoV-2 Infection Based on a Statewide Random Sample - Indiana, April 25-29, 2020.

Population prevalence of persons infected with SARS-CoV-2, the virus that causes coronavirus disease 2019 (COVID-19), varies by subpopulation and locality. U.S. studies of SARS-CoV-2 infection have examined infections in nonrandom samples (1) or seroprevalence in specific populations* (2), which are limited in their generalizability and cannot be used to accurately calculate infection-fatality rates. During April 25-29, 2020, Indiana conducted statewide random sample testing of persons aged ≥12 years to assess prevalence of active infection and presence of antibodies to SARS-CoV-2; additional nonrandom sampling was conducted in racial and ethnic minority communities to better understand the impact of the virus in certain racial and ethnic minority populations. Estimates were adjusted for nonresponse to reflect state demographics using an iterative proportional fitting method. Among 3,658 noninstitutionalized participants in the random sample survey, the estimated statewide point prevalence of active SARS-CoV-2 infection confirmed by reverse transcription-polymerase chain reaction (RT-PCR) testing was 1.74% (95% confidence interval [CI] = 1.10-2.54); 44.2% of these persons reported no symptoms during the 2 weeks before testing. The prevalence of immunoglobulin G (IgG) seropositivity, indicating past infection, was 1.09% (95% CI = 0.76-1.45). The overall prevalence of current and previous infections of SARS-CoV-2 in Indiana was 2.79% (95% CI = 2.02-3.70). In the random sample, higher overall prevalences were observed among Hispanics and those who reported having a household contact who had previously been told by a health care provider that they had COVID-19. By late April, an estimated 187,802 Indiana residents were currently or previously infected with SARS-CoV-2 (9.6 times higher than the number of confirmed cases [17,792]) (3), and 1,099 residents died (infection-fatality ratio = 0.58%). The number of reported cases represents only a fraction of the estimated total number of infections. Given the large number of persons who remain susceptible in Indiana, adherence to evidence-based public health mitigation and containment measures (e.g., social distancing, consistent and correct use of face coverings, and hand hygiene) is needed to reduce surge in hospitalizations and prevent morbidity and mortality from COVID-19.

A pseudo-likelihood method for estimating misclassification probabilities in competing-risks settings when true-event data are partially observed.

Outcome misclassification occurs frequently in binary-outcome studies and can result in biased estimation of quantities such as the incidence, prevalence, cause-specific hazards, cumulative incidence functions, and so forth. A number of remedies have been proposed to address the potential misclassification of the outcomes in such data. The majority of these remedies lie in the estimation of misclassification probabilities, which are in turn used to adjust analyses for outcome misclassification. A number of authors advocate using a gold-standard procedure on a sample internal to the study to learn about the extent of the misclassification. With this type of internal validation, the problem of quantifying the misclassification also becomes a missing data problem as, by design, the true outcomes are only ascertained on a subset of the entire study sample. Although, the process of estimating misclassification probabilities appears simple conceptually, the estimation methods proposed so far have several methodological and practical shortcomings. Most methods rely on missing outcome data to be missing completely at random (MCAR), a rather stringent assumption which is unlikely to hold in practice. Some of the existing methods also tend to be computationally-intensive. To address these issues, we propose a computationally-efficient, easy-to-implement, pseudo-likelihood estimator of the misclassification probabilities under a missing at random (MAR) assumption, in studies with an available internal-validation sample. We present the estimator through the lens of studies with competing-risks outcomes, though the estimator extends beyond this setting. We describe the consistency and asymptotic distributional properties of the resulting estimator, and derive a closed-form estimator of its variance. The finite-sample performance of this estimator is evaluated via simulations. Using data from a real-world study with competing-risks outcomes, we illustrate how the proposed method can be used to estimate misclassification probabilities. We also show how the estimated misclassification probabilities can be used in an external study to adjust for possible misclassification bias when modeling cumulative incidence functions.

Semiparametric regression and risk prediction with competing risks data under missing cause of failure.

The cause of failure in cohort studies that involve competing risks is frequently incompletely observed. To address this, several methods have been proposed for the semiparametric proportional cause-specific hazards model under a missing at random assumption. However, these proposals provide inference for the regression coefficients only, and do not consider the infinite dimensional parameters, such as the covariate-specific cumulative incidence function. Nevertheless, the latter quantity is essential for risk prediction in modern medicine. In this paper we propose a unified framework for inference about both the regression coefficients of the proportional cause-specific hazards model and the covariate-specific cumulative incidence functions under missing at random cause of failure. Our approach is based on a novel computationally efficient maximum pseudo-partial-likelihood estimation method for the semiparametric proportional cause-specific hazards model. Using modern empirical process theory we derive the asymptotic properties of the proposed estimators for the regression coefficients and the covariate-specific cumulative incidence functions, and provide methodology for constructing simultaneous confidence bands for the latter. Simulation studies show that our estimators perform well even in the presence of a large fraction of missing cause of failures, and that the regression coefficient estimator can be substantially more efficient compared to the previously proposed augmented inverse probability weighting estimator. The method is applied using data from an HIV cohort study and a bladder cancer clinical trial.

Changes in rapid HIV treatment initiation after national "treat all" policy adoption in 6 sub-Saharan African countries: Regression discontinuity analysis.

BACKGROUND: Most countries have formally adopted the World Health Organization's 2015 recommendation of universal HIV treatment ("treat all"). However, there are few rigorous assessments of the real-world impact of treat all policies on antiretroviral treatment (ART) uptake across different contexts. METHODS AND FINDINGS: We used longitudinal data for 814,603 patients enrolling in HIV care between 1 January 2004 and 10 July 2018 in 6 countries participating in the global International epidemiology Databases to Evaluate AIDS (IeDEA) consortium: Burundi (N = 11,176), Kenya (N = 179,941), Malawi (N = 84,558), Rwanda (N = 17,396), Uganda (N = 96,286), and Zambia (N = 425,246). Using a quasi-experimental regression discontinuity design, we assessed the change in the proportion initiating ART within 30 days of enrollment in HIV care (rapid ART initiation) after country-level adoption of the treat all policy. A modified Poisson model was used to identify factors associated with failure to initiate ART rapidly under treat all. In each of the 6 countries, over 60% of included patients were female, and median age at enrollment ranged from 32 to 36 years. In all countries studied, national adoption of treat all was associated with large increases in rapid ART initiation. Significant increases in rapid ART initiation immediately after treat all policy adoption were observed in Rwanda, from 44.4% to 78.9% of patients (34.5 percentage points [pp], 95% CI 27.2 to 41.7; p < 0.001), Kenya (25.7 pp, 95% CI 21.8 to 29.5; p < 0.001), Burundi (17.7 pp, 95% CI 6.5 to 28.9; p = 0.002), and Malawi (12.5 pp, 95% CI 7.5 to 17.5; p < 0.001), while no immediate increase was observed in Zambia (0.4 pp, 95% CI -2.9 to 3.8; p = 0.804) and Uganda (-4.2 pp, 95% CI -9.0 to 0.7; p = 0.090). The rate of rapid ART initiation accelerated sharply following treat all policy adoption in Malawi, Uganda, and Zambia; slowed in Kenya; and did not change in Rwanda and Burundi. In post hoc analyses restricted to patients enrolling under treat all, young adults (16-24 years) and men were at increased risk of not rapidly initiating ART (compared to older patients and women, respectively). However, rapid ART initiation following enrollment increased for all groups as more time elapsed since treat all policy adoption. Study limitations include incomplete data on potential ART eligibility criteria, such as clinical status, pregnancy, and enrollment CD4 count, which precluded the assessment of rapid ART initiation specifically among patients known to be eligible for ART before treat all. CONCLUSIONS: Our analysis indicates that adoption of treat all policies had a strong effect on increasing rates of rapid ART initiation, and that these increases followed different trajectories across the 6 countries. Young adults and men still require additional attention to further improve rapid ART initiation.

Nonparametric estimation of the cumulative incidence function under outcome misclassification using external validation data.

Misclassification of outcomes or event types is common in health sciences research and can lead to serious bias when estimating the cumulative incidence functions in settings with competing risks. Recent work has shown how to estimate nonparametric cumulative incidence functions in the presence of nondifferential outcome misclassification when the misclassification probabilities are known. Here, we extend this approach to account for misclassification that is differential with respect to important predictors of the outcome using misclassification probabilities estimated from external validation data. Moreover, we propose a bootstrap approach in which the observations from both the main study data and the external validation study are resampled to allow the uncertainty in the misclassification probabilities to propagate through the analysis into the final confidence intervals, ensuring appropriate confidence interval coverage probabilities. The proposed estimator is shown to be uniformly consistent and simulation studies indicate that both the estimator and the standard error estimation approach perform well in finite samples. The methodology is applied to estimate the cumulative incidence of death and disengagement from HIV care in a large cohort of HIV infected individuals in sub-Saharan Africa, where a significant death underreporting issue leads to outcome misclassification. This analysis uses external validation data from a separate study conducted in the same country.

NONPARAMETRIC INFERENCE FOR MARKOV PROCESSES WITH MISSING ABSORBING STATE.

This paper deals with the issue of nonparametric estimation of the transition probability matrix of a non-homogeneous Markov process with finite state space and partially observed absorbing state. We impose a missing at random assumption and propose a computationally efficient nonparametric maximum pseudolikelihood estimator (NPMPLE). The estimator depends on a parametric model that is used to estimate the probability of each absorbing state for the missing observations based, potentially, on auxiliary data. For the latter model we propose a formal goodness-of-fit test based on a residual process. Using modern empirical process theory we show that the estimator is uniformly consistent and converges weakly to a tight mean-zero Gaussian random field. We also provide methodology for simultaneous confidence band construction. Simulation studies show that the NPMPLE works well with small sample sizes and that it is robust against some degree of misspecification of the parametric model for the missing absorbing states. The method is illustrated using HIV data from sub-Saharan Africa to estimate the transition probabilities of death and disengagement from HIV care.

AUCTSP: an improved biomarker gene pair class predictor.

BACKGROUND: The Top Scoring Pair (TSP) classifier, based on the concept of relative ranking reversals in the expressions of pairs of genes, has been proposed as a simple, accurate, and easily interpretable decision rule for classification and class prediction of gene expression profiles. The idea that differences in gene expression ranking are associated with presence or absence of disease is compelling and has strong biological plausibility. Nevertheless, the TSP formulation ignores significant available information which can improve classification accuracy and is vulnerable to selecting genes which do not have differential expression in the two conditions ("pivot" genes). RESULTS: We introduce the AUCTSP classifier as an alternative rank-based estimator of the magnitude of the ranking reversals involved in the original TSP. The proposed estimator is based on the Area Under the Receiver Operating Characteristic (ROC) Curve (AUC) and as such, takes into account the separation of the entire distribution of gene expression levels in gene pairs under the conditions considered, as opposed to comparing gene rankings within individual subjects as in the original TSP formulation. Through extensive simulations and case studies involving classification in ovarian, leukemia, colon, breast and prostate cancers and diffuse large b-cell lymphoma, we show the superiority of the proposed approach in terms of improving classification accuracy, avoiding overfitting and being less prone to selecting non-informative (pivot) genes. CONCLUSIONS: The proposed AUCTSP is a simple yet reliable and robust rank-based classifier for gene expression classification. While the AUCTSP works by the same principle as TSP, its ability to determine the top scoring gene pair based on the relative rankings of two marker genes across all subjects as opposed to each individual subject results in significant performance gains in classification accuracy. In addition, the proposed method tends to avoid selection of non-informative (pivot) genes as members of the top-scoring pair.

Diagnostic and Prognostic Value of JC Virus DNA in Plasma in Progressive Multifocal Leukoencephalopathy.

Background: Progressive multifocal leukoencephalopathy (PML) is a severe demyelinating disease caused by the polyomavirus JC (John Cunningham; JCV) that affects patients with impaired immune systems. While JCV-DNA detection in cerebrospinal fluid (CSF) is diagnostic of PML, the clinical significance of plasma JCV-DNA is uncertain. Methods: We retrospectively analyzed plasma samples from PML patients that were drawn close to disease onset and from controls without PML. In PML patients, we compared plasma JCV-DNA detection and levels to clinical and laboratory parameters, and patient survival. Results: JCV-DNA was detected in plasma of 49/103 (48%) patients with PML (20/24, 83%, human immunodeficiency virus [HIV] negative; 29/79, 37%, HIV-positive) and of 4/144 (3%) controls without PML (0/95 HIV-negative; 4/49, 8%, HIV-positive), yielding a diagnostic sensitivity and specificity of 48% and 97% (83% and 100% in HIV-negative; 37% and 92% in HIV-positive), respectively. Among 16 PML patients with undetectable CSF JCV-DNA, 4 (25%) had detectable plasma JCV-DNA. Plasma JCV-DNA levels were independently associated with CSF levels (P < .0001) and previous corticosteroid treatment (P = .012). Higher plasma JCV-DNA levels were associated with disease progression in HIV-negative patients (P = .005); in HIV-positive patients, there was an increased risk of progression only in those treated with combination antiretroviral therapy (cART; P < .0001). Conclusions: Testing JCV-DNA in plasma might complement PML diagnosis, especially when CSF is unavailable or JCV-DNA not detectable in CSF. In addition, JCV-DNA plasma levels could be useful as a marker of disease progression in both HIV-negative and cART-treated, HIV-positive PML patients.