Inverse probability weighting in sexually transmitted infection/human immunodeficiency virus prevention research: methods for evaluating social and community interventions

Assessing Direct and Spillover Effects of Intervention Packages in Network-randomized Studies

BACKGROUND

Intervention packages may result in a greater public health impact than single interventions. Understanding the separate impact of each component on the overall package effectiveness can improve intervention delivery.

METHODS

We adapted an approach to evaluate the effects of a time-varying intervention package in a network-randomized study. In some network-randomized studies, only a subset of participants in exposed networks receive the intervention themselves. The spillover effect contrasts average potential outcomes if a person was not exposed to themselves under intervention in the network versus no intervention in a control network. We estimated the effects of components of the intervention package in HIV Prevention Trials Network 037, a Phase III network-randomized HIV prevention trial among people who inject drugs and their risk networks using marginal structural models to adjust for time-varying confounding. The index participant in an intervention network received a peer education intervention initially at baseline, then boosters at 6 and 12 months. All participants were followed to ascertain HIV risk behaviors.

RESULTS

There were 560 participants with at least one follow-up visit, 48% of whom were randomized to the intervention, and 1,598 participant visits were observed. The spillover effect of the boosters in the presence of initial peer education training was a 39% rate reduction (rate ratio = 0.61; 95% confidence interval = 0.43, 0.87).

CONCLUSIONS

These methods will be useful for evaluating intervention packages in studies with network features.

Assessing Direct and Spillover Effects of Intervention Packages in Network-Randomized Studies

Intervention packages may result in a greater public health impact than single interventions. Understanding the separate impact of each component in the overall package effectiveness can improve intervention delivery. We adapted an approach to evaluate the effects of a time-varying intervention package in a network-randomized study. In some network-randomized studies, only a subset of participants in exposed networks receive the intervention themselves. The spillover effect contrasts average potential outcomes if a person was not exposed themselves under intervention in the network versus no intervention in a control network. We estimated effects of components of the intervention package in HIV Prevention Trials Network 037, a Phase III network-randomized HIV prevention trial among people who inject drugs and their risk networks using Marginal Structural Models to adjust for time-varying confounding. The index participant in an intervention network received a peer education intervention initially at baseline, then boosters at 6 and 12 months. All participants were followed to ascertain HIV risk behaviors. There were 560 participants with at least one follow-up visit, 48% of whom were randomized to the intervention, and 1,598 participant-visits were observed. The spillover effect of the boosters in the presence of initial peer education training was a 39% rate reduction (Rate Ratio = 0.61; 95% confidence interval= 0.43, 0.87). These methods will be useful to evaluate intervention packages in studies with network features.

Relationship Between Patient Portal Tool Use and Medication Adherence and Viral Load Among Patients Living with HIV

BACKGROUND

Patient portals play an increasingly critical role in engaging patients in their health care. They have the potential to significantly impact the health of those living with chronic diseases, such as HIV, for whom consistent care engagement is both critical and complex.

OBJECTIVE

The primary aim was to examine the longitudinal relationships between individual portal tool use and health-related outcomes in patients living with HIV.

DESIGN

Retrospective cohort study using electronic health record data to examine the relationship between patient portal tool use and key HIV-specific, health-related outcomes in patients engaged in care in the Veterans Health Administration (VA) through the application of marginal structural models.

PARTICIPANTS

A national sample of patients living with HIV (PLWH) active in VA care who were registered to use the VA's patient portal, My HealtheVet (MHV; n = 18,390) between 10/1/2012 and 4/1/2017.

MAIN MEASURES

The MHV tools examined were prescription refill (including prescription refill of an antiretroviral (ART) medication and any medication), secure messaging, view appointments, and view labs. Primary outcomes were viral load test receipt, viral load suppression, and ART medication adherence (measured as proportion of days covered).

KEY RESULTS

The use of prescription refill for any medication or for ART was positively associated with ART adherence. Secure messaging was positively associated with ART adherence but not with viral load test receipt or viral load suppression. The use of view appointments was positively associated with ART adherence and viral load test receipt but not viral load suppression. The use of view labs was positively associated with viral load suppression but not ART adherence or viral load test receipt.

CONCLUSIONS

These findings highlight the valuable role patient portals may play in improving health-related outcomes among PLWH and have implications for patients living with other types of chronic disease.

Social inequality in cancer survivorship: Educational differences in health-related quality of life among 27,857 cancer survivors in Denmark

BACKGROUND

With a growing population of cancer survivors in Denmark, the evaluation of health-related quality of life (HRQoL) has become increasingly important. We describe variations in HRQoL between educational groups in a national population of cancer survivors.

METHODS

We conducted a cross-sectional questionnaire study among breast, prostate, lung, and colon cancer survivors diagnosed in 2010-2019 in Denmark. We used the EORTC QLQ-C30 to assess HRQoL including physical, role, emotional, cognitive, social functioning, and symptoms (fatigue, nausea and vomiting, pain, dyspnea, insomnia, appetite loss, constipation, diarrhea, and financial difficulties). Information on educational level and clinical data were extracted from national registers and clinical databases. Levels of impaired functioning and severe symptoms were identified using newly established thresholds for clinical importance. Multivariate logistic regression was used to examine associations between education and HRQoL. All statistical tests were 2-sided.

RESULTS

In total, 27,857 (42%) participated in the study. Up to 72% and 75% of cancer survivors with short education (≤9 years) reported impaired functioning and severe symptoms, respectively. Cancer survivors with short compared to long education (>12 years) were more likely to report impaired functioning and severe symptoms, with for example significantly higher odds ratios (ORs) for impaired physical function (breast OR = 2.41, 99% CI = 2.01-2.89; prostate OR = 1.81, 99% CI = 1.48-2.21; lung OR = 2.97, 99% CI = 1.95-4.57; and colon cancer OR = 1.69, 99% CI = 1.28-2.24).

CONCLUSIONS

Cancer survivors with short education are at greater risk of impaired HRQoL than survivors with long education 2-12 years after diagnosis. This underscores the need for systematic screening and symptom management in cancer aftercare, in order to reach all cancer survivors, also cancer survivors with short education.

Risks of Opportunistic Infections in People With Human Immunodeficiency Virus With Cancers Treated With Chemotherapy

BACKGROUND

We ascertained incidence of opportunistic infections (OIs) in people with human immunodeficiency virus (PWH) with cancer undergoing chemotherapy with non-human immunodeficiency virus (HIV) comparators.

METHODS

We identified 2106 PWH and 2981 uninfected Veterans with cancer who received at least 1 dose of chemotherapy between 1996 and 2017 from the Veterans Aging Cohort Study. We ascertained incident OIs within 6 months of chemotherapy amongst zoster, cytomegalovirus, tuberculosis, Candida esophagitis, Pneumocystis jirovecii pneumonia (PCP), toxoplasmosis, Cryptococcosis, atypical Mycobacterium infection, Salmonella bacteremia, histoplasmosis, coccidioidomycosis, or progressive multifocal leukoencephalopathy. We used Poisson methods to calculate OI incidence rates by HIV status, stratifying for hematological and nonhematological tumors. We compared OI rates by HIV status, using inverse probability weights of HIV status, further adjusting for PCP prophylaxis.

RESULTS

We confirmed 106 OIs in 101 persons. Adjusted OI incidence rate ratios (IRRs) indicated higher risk in PWH for all cancers (IRR, 4.8; 95% confidence interval [CI], 2.8-8.2), hematological cancers (IRR, 8.2; 95% CI, 2.4-27.3), and nonhematological cancers (IRR, 3.9; 95% CI, 2.1-7.2). Incidence rate ratios were not significantly higher in those with CD4 >200 cells/mm3 and viral load <500 copies/mL (IRR, 1.8; 95% CI, 0.9-3.2). All PCP cases (n = 11) occurred in PWH, with 2 microbiologically unconfirmed cases among 1467 PWH with nonhematological cancers, no PCP prophylaxis, and CD4 counts >200/mm3.

CONCLUSIONS

Veterans with HIV undergoing chemotherapy had higher rates of OIs than uninfected Veterans, particularly those with hematological cancers, but not in PWH with HIV controlled disease. Our study does not support systematic PCP prophylaxis in solid tumors in PWH with HIV controlled disease.

Outcomes and costs of publicly funded patient navigation interventions to enhance HIV care continuum outcomes in the United States: A before-and-after study

BACKGROUND

In the United States, patients with HIV face significant barriers to linkage to and retention in care which impede the necessary steps toward achieving the desired clinical outcome of viral suppression. Individual-level interventions, such as patient navigation, are evidence based, effective strategies for improving care engagement. In addition, use of surveillance and clinical data to identify patients who are not fully engaged in care may improve the effectiveness and cost-effectiveness of these programs.

METHODS AND FINDINGS

We employed a pre-post design to estimate the outcomes and costs, from the program perspective, of 5 state-level demonstration programs funded under the Health Resources and Services Administration's Special Projects of National Significance Program (HRSA/SPNS) Systems Linkages Initiative that employed existing surveillance and/or clinical data to identify individuals who had never entered HIV care, had fallen out of care, or were at risk of falling out of care and navigation strategies to engage patients in HIV care. Outcomes and costs were measured relative to standard of care during the first year of implementation of the interventions (2013 to 2014). We followed patients to estimate the number and proportion of additional patients linked, reengaged, retained, and virally suppressed by 12 months after enrollment in the interventions. We employed inverse probability weighting to adjust for differences in patient characteristics across programs, missing data, and loss to follow-up. We estimated the additional costs expended during the first year of each intervention and the cost per outcome of each intervention as the additional cost per HIV additional care continuum target achieved (cost per patient linked, reengaged, retained, and virally suppressed) 12 months after enrollment in each intervention. In this study, 3,443 patients were enrolled in Louisiana (LA), Massachusetts (MA), North Carolina (NC), Virginia (VA), and Wisconsin (WI) (147, 151, 2,491, 321, and 333, respectively). Patients were a mean of 40 years old, 75% male, and African American (69%) or Caucasian (22%). At baseline, 24% were newly diagnosed, 2% had never been in HIV care, 45% had fallen out of care, and 29% were at risk of falling out of care. All 5 interventions were associated with increases in the number and proportion of patients with viral suppression [percent increase: LA = 90.9%, 95% confidence interval (CI) = 88.4 to 93.4; MA = 78.1%, 95% CI = 72.4 to 83.8; NC = 47.5%, 95% CI = 45.2 to 49.8; VA = 54.6, 95% CI = 49.4 to 59.9; WI = 58.4, 95% CI = 53.4 to 63.4]. Overall, interventions cost an additional $4,415 (range = $3,746 to $5,619), $2,009 (range = $1,516 to $2,274), $920 (range = $627 to $941), $2,212 (range = $1,789 to $2,683), and $3,700 ($2,734 to $4,101), respectively per additional patient virally suppressed. The results of this study are limited in that we did not have contemporaneous controls for each intervention; thus, we are only able to assess patients against themselves at baseline and not against standard of care during the same time period.

CONCLUSIONS

Patient navigation programs were associated with improvements in engagement of patients in HIV care and viral suppression. Cost per outcome was minimized in states that utilized surveillance data to identify individuals who were out of care and/or those that were able to identify a larger number of patients in need of improvement at baseline. These results have the potential to inform the targeting and design of future navigation-type interventions.

Identifying Opportunities for Collaboration Across the Social Sciences to Reach the 10-10-10: A Multilevel Approach

BACKGROUND

The national and global strategy to combat HIV, often referred to as the "90-90-90," aims to diagnose 90% of people living with HIV, get 90% of those diagnosed onto antiretroviral treatment (ART), and achieve viral suppression in 90% of those on ART. The remaining 10-10-10 who will be undiagnosed, not on ART, or not virally suppressed, include vulnerable persons and populations most affected by social determinants of health. Given their foci on the social determinants of health at the individual, social, and structural levels, social scientists are in a prime position to help reach the 10-10-10. A potentially effective way for social scientists to achieve this goal is to examine the issues that affect the 10-10-10 using a multilevel framework, to understand at what levels their own approaches fit within such a multilevel framework, and to seek intentional collaborations with other social scientists who may work at different levels but whose approaches may complement their own within multilevel collaborations.

APPROACH

The present article describes how a multilevel framework can guide collaboration across disciplines within the social sciences toward the common goal of reaching the 10-10-10.

CONCLUSIONS

Within a multilevel framework, social scientists can work collaboratively to address the needs of individuals among the 10-10-10 within the social and structural contexts (eg, social norms, stigma, poverty, and barriers to care) that affect their health. Such an approach draws on the unique strengths and approaches of different social-science disciplines while also building capacity for individuals most affected by social determinants of health.

Subfertility among HIV-affected couples in a safer conception cohort in South Africa

BACKGROUND

Subfertility among couples affected by HIV has an impact on the well-being of couples who desire to have children and may prolong HIV exposure. Subfertility in the antiretroviral therapy era and its determinants have not yet been well characterized.

OBJECTIVE

The objective of the study was to investigate the burden and determinants of subfertility among HIV-affected couples seeking safer conception services in South Africa.

STUDY DESIGN

Nonpregnant women and male partners in HIV seroconcordant or HIV discordant relationships desiring a child were enrolled in the Sakh'umndeni safer conception cohort at Witkoppen Clinic in Johannesburg between July 2013 and April 2017. Clients were followed up prospectively through pregnancy (if they conceived) or until 6 months of attempted conception, after which they were referred for infertility services. Subfertility was defined as not having conceived within 6 months of attempted conception. Robust Poisson regression was used to assess the association between baseline characteristics and subfertility outcomes; inverse probability weighting was used to account for missing data from women lost to safer conception care before 6 months of attempted conception.

RESULTS

Among 334 couples enrolled, 65% experienced subfertility (inverse probability weighting weighted, 95% confidence interval, 0.59-0.73), of which 33% were primary subfertility and 67% secondary subfertility. Compared with HIV-negative women, HIV-positive women not on antiretroviral therapy had a 2-fold increased risk of subfertility (weighted and adjusted risk ratio, 2.00; 95% confidence interval, 1.19-3.34). Infertility risk was attenuated in women on antiretroviral therapy but remained elevated, even after ≥2 years on antiretroviral therapy (weighted and adjusted risk ratio, 1.63; 95% confidence interval, 0.98-2.69). Other factors associated with subfertility were female age (weighted and adjusted risk ratio, 1.03, 95% confidence interval, 1.01-1.05 per year), male HIV-positive status (weighted and adjusted risk ratio, 1.31; 95% confidence interval, 1.02-1.68), male smoking (weighted and adjusted risk ratio, 1.29; 95% confidence interval, 1.05-1.60), and trying to conceive for ≥1 year (weighted and adjusted risk ratio, 1.38; 95% confidence interval, 1.13-1.68).

CONCLUSION

Two in 3 HIV-affected couples experienced subfertility. HIV-positive women were at increased risk of subfertility, even when on antiretroviral therapy. Both male and female HIV status were associated with subfertility. Subfertility is an underrecognized reproductive health problem in resource-limited settings and may contribute to prolonged HIV exposure and transmission within couples. Low-cost approaches for screening and treating subfertility in this population are needed.

Treating loss-to-follow-up as a missing data problem: a case study using a longitudinal cohort of HIV-infected patients in Haiti

Background

HIV programs are often assessed by the proportion of patients who are alive and retained in care; however some patients are categorized as lost to follow-up (LTF) and have unknown vital status. LTF is not an outcome but a mixed category of patients who have undocumented death, transfer and disengagement from care. Estimating vital status (dead versus alive) among this category is critical for survival analyses and program evaluation.

Methods

We used three methods to estimate survival in the cohort and to ascertain factors associated with death among the first cohort of HIV positive patients to receive antiretroviral therapy in Haiti: complete case (CC) (drops missing), Inverse Probability Weights (IPW) (uses tracking data) and Multiple Imputation with Chained Equations (MICE) (imputes missing data). Logistic regression was used to calculate odds ratios and 95% confidence intervals for adjusted models for death at 10 years. The logistic regression models controlled for sex, age, severe poverty (living on <$1 USD per day), Port-au-Prince residence and baseline clinical characteristics of weight, CD4, WHO stage and tuberculosis diagnosis.

Results

Age, severe poverty, baseline weight and WHO stage were statistically significant predictors of AIDS related mortality across all models. Gender was only statistically significant in the MICE model but had at least a 10% difference in odds ratios across all models.

Conclusion

Each of these methods had different assumptions and differed in the number of observations included due to how missing values were addressed. We found MICE to be most robust in predicting survival status as it allowed us to impute missing data so that we had the maximum number of observations to perform regression analyses. MICE also provides a complementary alternative for estimating survival among patients with unassigned vital status. Additionally, the results were easier to interpret, less likely to be biased and provided an alternative to a problem that is often commented upon in the extant literature.

Electronic supplementary material

The online version of this article (10.1186/s12889-018-6115-0) contains supplementary material, which is available to authorized users.

The Epidemiology of Chronic Critical Illness After Severe Traumatic Injury at Two Level-One Trauma Centers

OBJECTIVE

To determine the incidence and risk factors of chronic critical illness after severe blunt trauma.

DESIGN

Prospective observational cohort study (NCT01810328).

SETTING

Two level-one trauma centers in the United States.

PATIENTS

One hundred thirty-five adult blunt trauma patients with hemorrhagic shock who survived beyond 48 hours after injury.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Chronic critical illness was defined as an ICU stay lasting 14 days or more with evidence of persistent organ dysfunction. Three subjects (2%) died within the first 7 days, 107 (79%) exhibited rapid recovery and 25 (19%) progressed to chronic critical illness. Patients who developed chronic critical illness were older (55 vs 44-year-old; p = 0.01), had more severe shock (base deficit, -9.2 vs -5.5; p = 0.005), greater organ failure severity (Denver multiple organ failure score, 3.5 ± 2.4 vs 0.8 ± 1.1; p < 0.0001) and developed more infectious complications (84% vs 35%; p < 0.0001). Chronic critical illness patients were more likely to be discharged to a long-term care setting (56% vs 34%; p = 0.008) than to a rehabilitation facility/home. At 4 months, chronic critical illness patients had higher mortality (16.0% vs 1.9%; p < 0.05), with survivors scoring lower in general health measures (p < 0.005). Multivariate analysis revealed age greater than or equal to 55 years, systolic hypotension less than or equal to 70 mm Hg, transfusion greater than or equal to 5 units packed red blood cells within 24 hours, and Denver multiple organ failure score at 72 hours as independent predictors of chronic critical illness (area under the receiver operating curve, 0.87; 95% CI, 0.75-0.95).

CONCLUSIONS

Although early mortality is low after severe trauma, chronic critical illness is a common trajectory in survivors and is associated with poor long-term outcomes. Advancing age, shock severity, and persistent organ dysfunction are predictive of chronic critical illness. Early identification may facilitate targeted interventions to change the trajectory of this morbid phenotype.

A Case Study of the Impact of Data-Adaptive Versus Model-Based Estimation of the Propensity Scores on Causal Inferences from Three Inverse Probability Weighting Estimators

OBJECTIVE

Consistent estimation of causal effects with inverse probability weighting estimators is known to rely on consistent estimation of propensity scores. To alleviate the bias expected from incorrect model specification for these nuisance parameters in observational studies, data-adaptive estimation and in particular an ensemble learning approach known as Super Learning has been proposed as an alternative to the common practice of estimation based on arbitrary model specification. While the theoretical arguments against the use of the latter haphazard estimation strategy are evident, the extent to which data-adaptive estimation can improve inferences in practice is not. Some practitioners may view bias concerns over arbitrary parametric assumptions as academic considerations that are inconsequential in practice. They may also be wary of data-adaptive estimation of the propensity scores for fear of greatly increasing estimation variability due to extreme weight values. With this report, we aim to contribute to the understanding of the potential practical consequences of the choice of estimation strategy for the propensity scores in real-world comparative effectiveness research.

METHOD

We implement secondary analyses of Electronic Health Record data from a large cohort of type 2 diabetes patients to evaluate the effects of four adaptive treatment intensification strategies for glucose control (dynamic treatment regimens) on subsequent development or progression of urinary albumin excretion. Three Inverse Probability Weighting estimators are implemented using both model-based and data-adaptive estimation strategies for the propensity scores. Their practical performances for proper confounding and selection bias adjustment are compared and evaluated against results from previous randomized experiments.

CONCLUSION

Results suggest both potential reduction in bias and increase in efficiency at the cost of an increase in computing time when using Super Learning to implement Inverse Probability Weighting estimators to draw causal inferences.

Women's fertility desires and contraceptive behavior in three peri-urban communities in sub Saharan Africa

Background

Fertility desires and contraceptive behavior often change over time. This study examined the influence of change in fertility desires on change in modern contraceptive use over time in three peri-urban communities in sub-Saharan Africa.

Methods

This multi-site study includes baseline and follow up data from 3 sites in the Family Health and Wealth Study. Following a census in each site, a probability sample of at least 500 households was obtained. Generalized linear models were employed.

Results

Modern contraceptive use increased in Ipetumodu, Nigeria (29.4 % to 36.7 %), but declined slightly in Sebeta, Ethiopia (66.9 % to 61.3 %) and Asawase, Ghana (12.6 % to 10.8 %). Across sites, at baseline and follow up, women who wanted no more children reported more contraceptive use, compared with those who wanted more children, and were more likely to shift to being contraceptive users in Ipetumodu [aOR(95 % CI):1.55 (1.07,2.26)].

Conclusions

Women’s fertility desires influenced their contraceptive behavior, although there were cross-site differences. Changing contraceptive demand and program factors will be important to enable peri-urban women to frame and act on their fertility desires.

Understanding the black box: a systematic review of the measurement of the community mobilization process in evaluations of interventions targeting sexual, reproductive, and maternal health

Community mobilization (CM) interventions are often used to improve sexual, reproductive, and maternal health (SRMH). This systematic review provides an overview of CM indicators used in evaluation and then focuses on the use of linking constructs-those measures of the process of CM between programmatic outputs and outcomes. We identified 108 English-language articles evaluating 86 CM interventions that target SRMH. Content analysis was used to code CM indicators into five categories: qualitative descriptions of CM; participation, diffusion and community action indicators that measure tangible, programmatic outputs; and linking constructs that capture the process of moving from participation to empowerment. Fifty-five (64.0%) interventions include a CM indicator. Diffusion indicators are most common (56.4%); linking constructs are least common (20.0%). We found 23 linking constructs used in evaluations of 11 interventions, with limited information on psychometric properties available. Three evaluations report positive relationships between linking constructs and condom use, one of which was significant in multivariate analysis. To better understand how CM leads to improved outcomes, we recommend increasing the measurement of linking constructs in evaluations of CM interventions. Research should focus on developing and validating new linking construct indicators and better disseminating existing measurement tools.

A community empowerment approach to the HIV response among sex workers: effectiveness, challenges, and considerations for implementation and scale-up

A community empowerment-based response to HIV is a process by which sex workers take collective ownership of programmes to achieve the most effective HIV outcomes and address social and structural barriers to their overall health and human rights. Community empowerment has increasingly gained recognition as a key approach for addressing HIV in sex workers, with its focus on addressing the broad context within which the heightened risk for infection takes places in these individuals. However, large-scale implementation of community empowerment-based approaches has been scarce. We undertook a comprehensive review of community empowerment approaches for addressing HIV in sex workers. Within this effort, we did a systematic review and meta-analysis of the effectiveness of community empowerment in sex workers in low-income and middle-income countries. We found that community empowerment-based approaches to addressing HIV among sex workers were significantly associated with reductions in HIV and other sexually transmitted infections, and with increases in consistent condom use with all clients. Despite the promise of a community-empowerment approach, we identified formidable structural barriers to implementation and scale-up at various levels. These barriers include regressive international discourses and funding constraints; national laws criminalising sex work; and intersecting social stigmas, discrimination, and violence. The evidence base for community empowerment in sex workers needs to be strengthened and diversified, including its role in aiding access to, and uptake of, combination interventions for HIV prevention. Furthermore, social and political change are needed regarding the recognition of sex work as work, both globally and locally, to encourage increased support for community empowerment responses to HIV.

Effect of patient navigation on colorectal cancer screening in a community-based randomized controlled trial of urban African American adults

PURPOSE

In recent years, colorectal cancer (CRC) screening rates have increased steadily in the USA, though racial and ethnic disparities persist. In a community-based randomized controlled trial, we investigated the effect of patient navigation on increasing CRC screening adherence among older African Americans.

METHODS

Participants in the Cancer Prevention and Treatment Demonstration were randomized to either the control group, receiving only printed educational materials (PEM), or the intervention arm where they were assigned a patient navigator in addition to PEM. Navigators assisted participants with identifying and overcoming screening barriers. Logistic regression analyses were used to assess the effect of patient navigation on CRC screening adherence. Up-to-date with screening was defined as self-reported receipt of colonoscopy/sigmoidoscopy in the previous 10 years or fecal occult blood testing (FOBT) in the year prior to the exit interview.

RESULTS

Compared with controls, the intervention group was more likely to report being up-to-date with CRC screening at the exit interview (OR 1.55, 95 % CI 1.07-2.23), after adjusting for select demographics. When examining the screening modalities separately, the patient navigator increased screening for colonoscopy/sigmoidoscopy (OR 1.53, 95 % CI 1.07-2.19), but not FOBT screening. Analyses of moderation revealed stronger effects of navigation among participants 65-69 years and those with an adequate health literacy level.

CONCLUSIONS

In a population of older African Americans adults, patient navigation was effective in increasing the likelihood of CRC screening. However, more intensive navigation may be necessary for adults over 70 years and individuals with low literacy levels.

Cervical cancer precursors and hormonal contraceptive use in HIV-positive women: application of a causal model and semi-parametric estimation methods

OBJECTIVE

To demonstrate the application of causal inference methods to observational data in the obstetrics and gynecology field, particularly causal modeling and semi-parametric estimation.

BACKGROUND

Human immunodeficiency virus (HIV)-positive women are at increased risk for cervical cancer and its treatable precursors. Determining whether potential risk factors such as hormonal contraception are true causes is critical for informing public health strategies as longevity increases among HIV-positive women in developing countries.

METHODS

We developed a causal model of the factors related to combined oral contraceptive (COC) use and cervical intraepithelial neoplasia 2 or greater (CIN2+) and modified the model to fit the observed data, drawn from women in a cervical cancer screening program at HIV clinics in Kenya. Assumptions required for substantiation of a causal relationship were assessed. We estimated the population-level association using semi-parametric methods: g-computation, inverse probability of treatment weighting, and targeted maximum likelihood estimation.

RESULTS

We identified 2 plausible causal paths from COC use to CIN2+: via HPV infection and via increased disease progression. Study data enabled estimation of the latter only with strong assumptions of no unmeasured confounding. Of 2,519 women under 50 screened per protocol, 219 (8.7%) were diagnosed with CIN2+. Marginal modeling suggested a 2.9% (95% confidence interval 0.1%, 6.9%) increase in prevalence of CIN2+ if all women under 50 were exposed to COC; the significance of this association was sensitive to method of estimation and exposure misclassification.

CONCLUSION

Use of causal modeling enabled clear representation of the causal relationship of interest and the assumptions required to estimate that relationship from the observed data. Semi-parametric estimation methods provided flexibility and reduced reliance on correct model form. Although selected results suggest an increased prevalence of CIN2+ associated with COC, evidence is insufficient to conclude causality. Priority areas for future studies to better satisfy causal criteria are identified.

Targeted learning in real-world comparative effectiveness research with time-varying interventions

In comparative effectiveness research (CER), often the aim is to contrast survival outcomes between exposure groups defined by time-varying interventions. With observational data, standard regression analyses (e.g., Cox modeling) cannot account for time-dependent confounders on causal pathways between exposures and outcome nor for time-dependent selection bias that may arise from informative right censoring. Inverse probability weighting (IPW) estimation to fit marginal structural models (MSMs) has commonly been applied to properly adjust for these expected sources of bias in real-world observational studies. We describe the application and performance of an alternate estimation approach in such a study. The approach is based on the recently proposed targeted learning methodology and consists in targeted minimum loss-based estimation (TMLE) with super learning (SL) within a nonparametric MSM. The evaluation is based on the analysis of electronic health record data with both IPW estimation and TMLE to contrast cumulative risks under four more or less aggressive strategies for treatment intensification in adults with type 2 diabetes already on 2+ oral agents or basal insulin. Results from randomized experiments provide a surrogate gold standard to validate confounding and selection bias adjustment. Bootstrapping is used to validate analytic estimation of standard errors. This application does the following: (1) establishes the feasibility of TMLE in real-world CER based on large healthcare databases; (2) provides evidence of proper confounding and selection bias adjustment with TMLE and SL; and (3) motivates their application for improving estimation efficiency. Claims are reinforced with a simulation study that also illustrates the double-robustness property of TMLE.

Super learning to hedge against incorrect inference from arbitrary parametric assumptions in marginal structural modeling

OBJECTIVE

Clinical trials are unlikely to ever be launched for many comparative effectiveness research (CER) questions. Inferences from hypothetical randomized trials may however be emulated with marginal structural modeling (MSM) using observational data, but success in adjusting for time-dependent confounding and selection bias typically relies on parametric modeling assumptions. If these assumptions are violated, inferences from MSM may be inaccurate. In this article, we motivate the application of a data-adaptive estimation approach called super learning (SL) to avoid reliance on arbitrary parametric assumptions in CER.

STUDY DESIGN AND SETTING

Using the electronic health records data from adults with new-onset type 2 diabetes, we implemented MSM with inverse probability weighting (IPW) estimation to evaluate the effect of three oral antidiabetic therapies on the worsening of glomerular filtration rate.

RESULTS

Inferences from IPW estimation were noticeably sensitive to the parametric assumptions about the associations between both the exposure and censoring processes and the main suspected source of confounding, that is, time-dependent measurements of hemoglobin A1c. SL was successfully implemented to harness flexible confounding and selection bias adjustment from existing machine learning algorithms.

CONCLUSION

Erroneous IPW inference about clinical effectiveness because of arbitrary and incorrect modeling decisions may be avoided with SL.

Community empowerment among female sex workers is an effective HIV prevention intervention: a systematic review of the peer-reviewed evidence from low- and middle-income countries

We conducted a systematic review and meta-analysis of community empowerment interventions for HIV prevention among sex workers in low- and middle-income countries from 1990-2010. Two coders abstracted data using standardized forms. Of 6,664 citations screened, ten studies met inclusion criteria. For HIV infection, two observational studies showed a significantly protective combined effect [odds ratio (OR): 0.84, 95% confidence interval (CI): 0.709-0.988]. For STI infection, one longitudinal study showed reduced gonorrhoea/chlamydia (OR: 0.51, 95% CI: 0.26-0.99). Observational studies showed reduced gonorrhoea (OR: 0.65, 95% CI: 0.47-0.90), but non-significant effects on chlamydia and syphilis. For condom use, one randomized controlled trial showed improvements with clients (beta: 0.3447, p = 0.002). One longitudinal study showed improvements with regular clients (OR: 1.9, 95% CI: 1.1-3.3), but no change with new clients. Observational studies showed improvements with new clients (OR: 3.04, 95% CI: 1.29-7.17), regular clients (OR: 2.20, 95% CI: 1.41-3.42), and all clients (OR: 5.87, 95% CI: 2.88-11.94), but not regular non-paying partners. Overall, community empowerment-based HIV prevention was associated with significant improvements across HIV outcomes and settings.

Findings from Encontros: a multilevel STI/HIV intervention to increase condom use, reduce STI, and change the social environment among sex workers in Brazil

BACKGROUND

Sexually transmitted infection (STI)/HIV prevention programs, which do not modify social structural contexts that contribute to risk of STI/HIV may fail to bring about improvements in health, particularly among groups who experience discrimination and exclusion from public life. We conducted a multilevel intervention with sex workers, including improved clinical care and community-mobilizing strategies to modify social structural factors that shape sexual behavior, to improve condom use and reduce incident STI.

METHODS

We followed 420 sex workers participating in the Encontros intervention in Corumbá, Brazil, between 2003 and 2005. We estimated the effect of the intervention on incident chlamydia and gonorrhea infections and condom use using generalized estimating equations and inverse probability weighting by comparing those who actively engaged in the intervention activities (exposed) with those who were less engaged (unexposed). We also determined the association of participation on reported social cohesion and participation in networks.

RESULTS

Exposed participants had significantly higher odds of reporting consistent condom use with regular clients (odds ratio [OR]: 1.9, 95% confidence interval:1.1-3.3) and nonsignificantly increased odds with both new clients (OR: 1.6, 0.9-2.8) and nonpaying partners (OR: 1.5, 0.9-1.5). The odds of an incident STI were nonsignificantly reduced for exposed participants compared with unexposed (OR: 0.46, 0.2-1.3). Participation was significantly associated with increased perceived cohesion and participation in networks.

CONCLUSION

This prospective study provides evidence that multilevel interventions with mobilizing strategies to modify aspects of the social environment can improve condom use, reduce STIs, and increase social cohesion and participation in networks among sex workers.

Using culturally sensitive media messages to reduce HIV-associated sexual behavior in high-risk African American adolescents: results from a randomized trial

PURPOSE

To test the long-term effects of a mass media intervention that used culturally and developmentally appropriate messages to enhance human immunodeficiency virus (HIV)-preventive beliefs and behavior of high-risk African American adolescents.

METHODS

Television and radio messages were delivered for more than 3 years in two cities (Syracuse, NY; and Macon, GA) that were randomly selected within each of the two regionally matched city pairs, with the other cities (Providence, RI; and Columbia, SC) serving as controls. African American adolescents, aged 14-17 years (N = 1,710), recruited in the four cities over a 16-month period, completed audio computer-assisted self-interviews at recruitment and again at 3, 6, 12, and 18-months postrecruitment to assess the long-term effects of the media program. To identify the unique effects of the media intervention, youth who completed at least one follow-up and who did not test positive for any of the three sexually transmitted infections at recruitment or at 6-and 12-month follow-up were retained for analysis (N = 1,346).

RESULTS

The media intervention reached virtually all the adolescents in the trial and produced a range of effects including improved normative condom-use negotiation expectancies and increased sex refusal self-efficacy. Most importantly, older adolescents (aged 16-17 years) exposed to the media program showed a less risky age trajectory of unprotected sex than those in the nonmedia cities.

CONCLUSION

Culturally tailored mass media messages that are delivered consistently over time have the potential to reach a large audience of high-risk adolescents, to support changes in HIV-preventive beliefs, and to reduce HIV-associated risk behaviors among older youth.