Search tips
Search criteria

Results 1-10 (10)

Clipboard (0)

Select a Filter Below

Year of Publication
Document Types
1.  Estimating Optimal Dynamic Regimes: Correcting Bias under the Null 
A dynamic regime provides a sequence of treatments that are tailored to patient-specific characteristics and outcomes. In 2004 James Robins proposed g-estimation using structural nested mean models for making inference about the optimal dynamic regime in a multi-interval trial. The method provides clear advantages over traditional parametric approaches. Robins’ g-estimation method always yields consistent estimators, but these can be asymptotically biased under a given structural nested mean model for certain longitudinal distributions of the treatments and covariates, termed exceptional laws. In fact, under the null hypothesis of no treatment effect, every distribution constitutes an exceptional law under structural nested mean models which allow for interaction of current treatment with past treatments or covariates. This paper provides an explanation of exceptional laws and describes a new approach to g-estimation which we call Zeroing Instead of Plugging In (ZIPI). ZIPI provides nearly identical estimators to recursive g-estimators at non-exceptional laws while providing substantial reduction in the bias at an exceptional law when decision rule parameters are not shared across intervals.
PMCID: PMC2880540  PMID: 20526433
adaptive treatment strategies; asymptotic bias; dynamic treatment regimes; g-estimation; optimal structural nested mean models; pre-test estimators
The annals of applied statistics  2014;8(2):703-725.
The PROmotion of Breastfeeding Intervention Trial (PROBIT) cluster-randomized a program encouraging breastfeeding to new mothers in hospital centers. The original studies indicated that this intervention successfully increased duration of breastfeeding and lowered rates of gastrointestinal tract infections in newborns. Additional scientific and popular interest lies in determining the causal effect of longer breastfeeding on gastrointestinal infection. In this study, we estimate the expected infection count under various lengths of breastfeeding in order to estimate the effect of breastfeeding duration on infection. Due to the presence of baseline and time-dependent confounding, specialized “causal” estimation methods are required. We demonstrate the double-robust method of Targeted Maximum Likelihood Estimation (TMLE) in the context of this application and review some related methods and the adjustments required to account for clustering. We compare TMLE (implemented both parametrically and using a data-adaptive algorithm) to other causal methods for this example. In addition, we conduct a simulation study to determine (1) the effectiveness of controlling for clustering indicators when cluster-specific confounders are unmeasured and (2) the importance of using data-adaptive TMLE.
PMCID: PMC4259272  PMID: 25505499
Causal inference; G-computation; inverse probability weighting; marginal effects; missing data; pediatrics
3.  Q-learning for estimating optimal dynamic treatment rules from observational data 
The area of dynamic treatment regimes (DTR) aims to make inference about adaptive, multistage decision-making in clinical practice. A DTR is a set of decision rules, one per interval of treatment, where each decision is a function of treatment and covariate history that returns a recommended treatment. Q-learning is a popular method from the reinforcement learning literature that has recently been applied to estimate DTRs. While, in principle, Q-learning can be used for both randomized and observational data, the focus in the literature thus far has been exclusively on the randomized treatment setting. We extend the method to incorporate measured confounding covariates, using direct adjustment and a variety of propensity score approaches. The methods are examined under various settings including non-regular scenarios. We illustrate the methods in examining the effect of breastfeeding on vocabulary testing, based on data from the Promotion of Breastfeeding Intervention Trial.
PMCID: PMC3551601  PMID: 23355757
Bias; confounding; dynamic treatment regime; inverse probability of treatment weighting; non-regularity; propensity scores
4.  Estimating the optimal dynamic antipsychotic treatment regime: Evidence from the sequential multiple assignment randomized CATIE Schizophrenia Study 
Treatment of schizophrenia is notoriously difficult and typically requires personalized adaption of treatment due to lack of efficacy of treatment, poor adherence, or intolerable side effects. The Clinical Antipsychotic Trials in Intervention Effectiveness (CATIE) Schizophrenia Study is a sequential multiple assignment randomized trial comparing the typical antipsychotic medication, perphenazine, to several newer atypical antipsychotics. This paper describes the marginal structural modeling method for estimating optimal dynamic treatment regimes and applies the approach to the CATIE Schizophrenia Study. Missing data and valid estimation of confidence intervals are also addressed.
PMCID: PMC3475611  PMID: 23087488
Adaptive treatment strategies; causal effects; dynamic treatment regimes; inverse probability weighting; marginal structural models; personalized medicine; schizophrenia
5.  Marijuana Smoking Does Not Accelerate Progression of Liver Disease in HIV–Hepatitis C Coinfection: A Longitudinal Cohort Analysis 
In a large human immunodeficiency virus–hepatitis C virus coinfection cohort, we found no evidence that marijuana smoking accelerated progression to significant liver fibrosis, cirrhosis, or end-stage liver disease. Previous studies reporting an association may have been biased by reverse causation due to self-medication.
Background. Marijuana smoking is common and believed to relieve many symptoms, but daily use has been associated with liver fibrosis in cross-sectional studies. We aimed to estimate the effect of marijuana smoking on liver disease progression in a Canadian prospective multicenter cohort of human immunodeficiency virus/hepatitis C virus (HIV/HCV) coinfected persons.
Methods. Data were analyzed for 690 HCV polymerase chain reaction positive (PCR-positive) individuals without significant fibrosis or end-stage liver disease (ESLD) at baseline. Time-updated Cox Proportional Hazards models were used to assess the association between the average number of joints smoked/week and progression to significant liver fibrosis (APRI ≥ 1.5), cirrhosis (APRI ≥ 2) or ESLD.
Results At baseline, 53% had smoked marijuana in the past 6 months, consuming a median of 7 joints/week (IQR, 1–21); 40% smoked daily. There was no evidence that marijuana smoking accelerates progression to significant liver fibrosis (APRI ≥ 1.5) or cirrhosis (APRI ≥ 2; hazard ratio [HR]: 1.02 [0.93–1.12] and 0.99 [0.88–1.12], respectively). Each 10 additional joints/week smoked slightly increased the risk of progression to a clinical diagnosis of cirrhosis and ESLD combined (HR, 1.13 [1.01–1.28]). However, when exposure was lagged to 6–12 months before the diagnosis, marijuana was no longer associated with clinical disease progression (HR, 1.10 [0.95–1.26]).
Conclusions In this prospective analysis we found no evidence for an association between marijuana smoking and significant liver fibrosis progression in HIV/HCV coinfection. A slight increase in the hazard of cirrhosis and ESLD with higher intensity of marijuana smoking was attenuated after lagging marijuana exposure, suggesting that reverse causation due to self-medication could explain previous results.
PMCID: PMC3739469  PMID: 23811492
HIV; HCV; cannabis; liver disease; cohort study
6.  Breastfeeding and Infant Size: Evidence of Reverse Causality 
American Journal of Epidemiology  2011;173(9):978-983.
Infants who receive prolonged and exclusive breastfeeding grow more slowly during the first year of life than those who do not. However, infant feeding and growth are dynamic processes in which feeding may affect growth, and prior growth and size may also influence subsequent feeding decisions. The authors carried out an observational analysis of 17,046 Belarusian infants who were recruited between June 1996 and December 1997 and who participated in a cluster-randomized trial of a breastfeeding promotion intervention. To assess the effects of infant size on subsequent feeding, the authors restricted the analysis to infants breastfed (or exclusively breastfed) at the beginning of each follow-up interval and examined associations between weight or length at the beginning of the interval and weaning or discontinuation of exclusive breastfeeding by the end of the interval. Smaller size (especially weight for age) was strongly and statistically significantly associated with increased risks of subsequent weaning and of discontinuing exclusive breastfeeding (adjusted odds ratios = 1.2–1.6), especially between 2 and 6 months, even after adjusment for potential confounding factors and clustered measurement. The authors speculate that similar dynamic processes involving infant crying, other signs of hunger, and supplementation/weaning undermine causal inferences about the “effect” of prolonged and exclusive breastfeeding on slower infant growth.
PMCID: PMC3390166  PMID: 21430194
body size; breast feeding; causal inference; evidence; infant
7.  The Impact of Antiretroviral Therapy in a Cohort of HIV Infected Patients Going in and out of the San Francisco County Jail 
PLoS ONE  2009;4(9):e7115.
Jails are an important venue of HIV care and a place for identification, treatment and referral for care. HIV infected inmates in the San Francisco County jail are offered antiretroviral treatment (ART), which many take only while in jail. We evaluated the effect of ART administration in a cohort of jail inmates going in and out of jail over a nine year period.
Methodology/Principal Findings
In this retrospective study, we examined inmates with HIV going in and out of jail. Inmates were categorized by patterns of ART use: continuous ART - ART both in and out of jail, intermittent ART - ART only in jail; never on ART - eligible by national guidelines, but refused ART. CD4 and HIV viral load (VL) were compared over time in these groups. Over a 9 year period, 512 inmates were studied: 388 (76%) on intermittent ART, 79 (15%) on continuous ART and 45(9%) never-on ART. In a linear mixed model analysis, inmates on intermittent ART were 1.43; 95%CI (1.03, 1.99) times and those never on ART were 2.89; 95%CI (1.71, 4.87) times more likely to have higher VL than inmates on continuous ART. Furthermore, Inmates on intermittent ART and never-on ART lost 1.60; 95%CI (1.06, 2.13) and 1.97; 95%CI (0.96, 3.00) more CD4 cells per month, respectively, compared to continuously treated inmates. The continuous ART inmates gained 0.67CD4 cells/month.
Continuous ART therapy in jail inmate's benefits CD4 cell counts and control of VL especially compared to those who never took ART. Although jail inmates on intermittent ART were more likely to lose CD4 cells and experience higher VL over time than those on continuous ART, CD4 cell loss was slower in these inmates as compared to inmates never on ART. Further studies are needed to evaluate whether or not intermittent ART provides some benefit in outcome if continuous ART is not possible or likely.
PMCID: PMC2744925  PMID: 19771176
8.  Is Antiretroviral Therapy Causing Long-Term Liver Damage? A Comparative Analysis of HIV-Mono-Infected and HIV/Hepatitis C Co-Infected Cohorts 
PLoS ONE  2009;4(2):e4517.
The effects of highly active antiretroviral therapy (HAART) on progression of hepatic fibrosis in HIV-hepatitis C virus (HCV) co-infection are not well understood. Deaths from liver diseases have risen in the post-HAART era, yet some cross-sectional studies have suggested that HAART use is associated with improved fibrosis rates. In a retrospective cohort of 533 HIV mono-infected and 127 HIV/HCV co-infected patients, followed between January 1991 and July 2005 at a university-based HIV clinic, we investigated the relationship between cumulative HAART exposure and hepatic fibrosis, as measured by the aspartate aminotransferase-to-platelet ratio index (APRI). We used a novel methodological approach to estimate the dose-response relationship of the effect of HAART exposure on APRI. HAART was associated with increasing APRI over time in HIV/HCV co-infected patients suggesting that they may be experiencing cumulative hepatotoxicity from antiretrovirals. The estimated median change (95% confidence interval) in APRI per one year of HAART intake was of −0.46% (−1.61% to 0.71%) in HIV mono-infected compared to 2.54% (−1.77% to 7.03%) in HIV/HCV co-infected patients. Similar results were found when the direct effect of HAART intake since the last visit was estimated on the change in APRI. HAART use associated is with increased APRI in patients with HIV/HCV co-infection. Therefore treatment for HCV infection may be required to slow the growing epidemic of end-stage liver disease in this population.
PMCID: PMC2637977  PMID: 19223976
9.  T-Cell Assays for Tuberculosis Infection: Deriving Cut-Offs for Conversions Using Reproducibility Data 
PLoS ONE  2008;3(3):e1850.
Although interferon-gamma release assays (IGRA) are promising alternatives to the tuberculin skin test, interpretation of repeated testing results is hampered by lack of evidence on optimal cut-offs for conversions and reversions. A logical start is to determine the within-person variability of T-cell responses during serial testing.
Methodology/Principal Findings
We performed a pilot study in India, to evaluate the short-term reproducibility of QuantiFERON-TB Gold In Tube assay (QFT) among 14 healthcare workers (HCWs) who underwent 4 serial QFT tests on day 0, 3, 9 and 12. QFT ELISA was repeated twice on the same sets of specimens. We assessed two types of reproducibility: 1) test-retest reproducibility (between-test variability), and 2) within-person reproducibility over time. Test-retest reproducibility: with dichotomous test results, extremely high concordance was noticed between two tests performed on the same sets of specimens: of the 56 samples, the test and re-test results agreed for all but 2 individuals (κ = 0.94). Discordance was noted in subjects who had IFN-γ values around the cut-off point, with both increases and decreases noted. With continuous IFN-γ results, re-test results tended to produce higher estimates of IFN-γ than the original test. Within-person reproducibility: when continuous IFN-γ data were analyzed, the within-person reproducibility was moderate to high. While persons with negative QFT results generally stayed negative, positive results tended to vary over time. Our data showed that increases of more than 16% in the IFN-γ levels are statistically improbable in the short-term.
Conservatively assuming that long-term variability might be at least twice higher than short-term, we hypothesize that a QFT conversion requires two conditions to be met: 1) change from negative to positive result, and 2) at least 30% increase in the baseline IFN-γ response. Larger studies are needed to confirm our preliminary findings, and determine the conversion thresholds for IGRAs.
PMCID: PMC2266993  PMID: 18365006
10.  The impact of antibiotics on growth in children in low and middle income countries: systematic review and meta-analysis of randomised controlled trials 
Objectives To determine whether antibiotic treatment leads to improvements in growth in prepubertal children in low and middle income countries, to determine the magnitude of improvements in growth, and to identify moderators of this treatment effect.
Design Systematic review and meta-analysis.
Data sources Medline, Embase, Scopus, the Cochrane central register of controlled trials, and Web of Science.
Study selection Randomised controlled trials conducted in low or middle income countries in which an orally administered antibacterial agent was allocated by randomisation or minimisation and growth was measured as an outcome. Participants aged 1 month to 12 years were included. Control was placebo or non-antimicrobial intervention.
Results Data were pooled from 10 randomised controlled trials representing 4316 children, across a variety of antibiotics, indications for treatment, treatment regimens, and countries. In random effects models, antibiotic use increased height by 0.04 cm/month (95% confidence interval 0.00 to 0.07) and weight by 23.8 g/month (95% confidence interval 4.3 to 43.3). After adjusting for age, effects on height were larger in younger populations and effects on weight were larger in African studies compared with other regions.
Conclusion Antibiotics have a growth promoting effect in prepubertal children in low and middle income countries. This effect was more pronounced for ponderal than for linear growth. The antibiotic growth promoting effect may be mediated by treatment of clinical or subclinical infections or possibly by modulation of the intestinal microbiota. Better definition of the mechanisms underlying this effect will be important to inform optimal and safe approaches to achieving healthy growth in vulnerable populations.
PMCID: PMC3988318  PMID: 24735883

Results 1-10 (10)