Implementing machine learning methods with complex survey data: Lessons learned on the impacts of accounting sampling weights in gradient boosting

Despite the prominent use of complex survey data and the growing popularity of machine learning methods in epidemiologic research, few machine learning software implementations offer options for handling complex samples. A major challenge impeding the broader incorporation of machine learning into epidemiologic research is incomplete guidance for analyzing complex survey data, including the importance of sampling weights for valid prediction in target populations. Using data from 15, 820 participants in the 1988-1994 National Health and Nutrition Examination Survey cohort, we determined whether ignoring weights in gradient boosting models of all-cause mortality affected prediction, as measured by the F1 score and corresponding 95% confidence intervals. In simulations, we additionally assessed the impact of sample size, weight variability, predictor strength, and model dimensionality. In the National Health and Nutrition Examination Survey data, unweighted model performance was inflated compared to the weighted model (F1 score 81.9% [95% confidence interval: 81.2%, 82.7%] vs 77.4% [95% confidence interval: 76.1%, 78.6%]). However, the error was mitigated if the F1 score was subsequently recalculated with observed outcomes from the weighted dataset (F1: 77.0%; 95% confidence interval: 75.7%, 78.4%). In simulations, this finding held in the largest sample size (N = 10,000) under all analytic conditions assessed. For sample sizes <5,000, sampling weights had little impact in simulations that more closely resembled a simple random sample (low weight variability) or in models with strong predictors, but findings were inconsistent under other analytic scenarios. Failing to account for sampling weights in gradient boosting models may limit generalizability for data from complex surveys, dependent on sample size and other analytic properties. In the absence of software for configuring weighted algorithms, post-hoc re-calculations of unweighted model performance using weighted observed outcomes may more accurately reflect model prediction in target populations than ignoring weights entirely.

Serum antioxidant vitamins and respiratory morbidity and mortality: a pooled analysis

BACKGROUND

Oxidative stress plays a key role in the pathogenesis of respiratory diseases; however, studies on antioxidant vitamins and respiratory outcomes have been conflicting. We evaluated whether lower serum levels of vitamins A, C, D, and E are associated with respiratory morbidity and mortality in the U.S. adult population.

METHODS

We conducted a pooled analysis of data from the 1988-1994 and 1999-2006 National Health and Nutrition Examination Survey (participants aged ≥ 20 years). We estimated covariate-adjusted odds ratios (aOR) per interquartile decrease in each serum vitamin level to quantify associations with respiratory morbidity, and covariate-adjusted hazard ratios (aHR) to quantify associations with respiratory mortality assessed prospectively through 2015. Vitamin supplementation and smoking were evaluated as potential effect modifiers.

RESULTS

Lower serum vitamin C increased the odds of wheeze among all participants (overall aOR: 1.08, 95% CI: 1.01-1.16). Among smokers, lower serum α-tocopherol vitamin E increased the odds of wheeze (aOR: 1.11, 95% CI: 1.04-1.19) and chronic bronchitis/emphysema (aOR: 1.13, 95% CI: 1.03-1.24). Conversely, lower serum γ-tocopherol vitamin E was associated with lower odds of wheeze and chronic bronchitis/emphysema (overall aORs: 0.85, 95% CI: 0.79-0.92 and 0.85, 95% CI: 0.76-0.95, respectively). Lower serum vitamin C was associated with increased chronic lower respiratory disease (CLRD) mortality in all participants (overall aHR: 1.27, 95% CI: 1.07-1.51), whereas lower serum 25-hydroxyvitamin D (25-OHD) tended to increase mortality from CLRD and influenza/pneumonia among smokers (aHR range: 1.33-1.75). Mortality from influenza/ pneumonia increased with decreasing serum vitamin A levels in all participants (overall aHR: 1.21, 95% CI: 0.99-1.48). In pooled analysis, vitamin C deficiency and 25-OHD insufficiency were associated with mortality from influenza/pneumonia, increasing mortality risk up to twofold.

CONCLUSIONS

Our analysis of nationally representative data on over 34,000 participants showed that lower serum levels of vitamins A, C, D, and α-tocopherol vitamin E are associated with increased respiratory morbidity and/or mortality in U.S. adults. The results underscore the importance of antioxidant vitamins in respiratory health.

Poor clinical guideline adherence and inappropriate testing for incident lower urinary tract symptoms associated with benign prostatic hyperplasia

BACKGROUND

The American Urological Association makes recommendations for evaluation and testing for lower urinary tract symptoms associated with benign prostatic hyperplasia (LUTS/BPH) to help primary care providers and specialists identify LUTS/BPH and harmful related conditions including urinary retention and prostate or bladder cancer. Our understanding of provider adherence to these Guidelines is limited to single-site or nonrepresentative settings.

METHODS

We analyzed two insurance claims databases: the Optum^® de-identified Clinformatics^® Data Mart database for privately insured males aged 40-64 years (N ≈ 1,650,900 annually) and the Medicare 5% Sample for males aged ≥65 years (N ≈ 546,000 annually). We calculated the annual prevalence of LUTS/BPH and comorbid bladder cancer and bladder stones from 2004 to 2013. We additionally examined LUTS/BPH incidence and adherence to testing guidelines in a cohort of men newly diagnosed with LUTS/BPH in 2009.

RESULTS

While LUTS/BPH prevalence and incidence increased with increasing age, evaluation testing became less common. Urinalysis was the most common testing type but was performed in <60% of incident patients. Serum prostate-specific antigen (PSA) was the second most common test across age groups (range: 15-34%). Prevalence of comorbid bladder cancer (range: 0-4%), but not bladder stones (range: 1-2%), increased with increasing age.

CONCLUSIONS

Although older men were at greater risk of LUTS/BPH than younger men, they were less likely to undergo testing at diagnosis. Recommended testing with urinalysis was poor despite higher prevalence of bladder cancer in older men and a standard recommendation for urinalysis since 1994. Providers should be more cognizant of AUA Guidelines when assessing LUTS/BPH patients.

National Trends in Neuromodulation for Urinary Incontinence Among Insured Adult Women and Men, 2004-2013: The Urologic Diseases in America Project

OBJECTIVE

To examine US trends in neuromodulation for urinary incontinence (UI) treatment from 2004 to 2013.

METHODS

This study utilized 2 data sources: the Optum© de-identified Clinformatics® Data Mart Database for privately insured adults aged 18-64 years with a UI diagnosis (N≈40,000 women and men annually) and the Medicare 5% Sample for beneficiaries aged ≥65 years with a UI diagnosis (N≈65,000 women and men annually). We created annual cross-sectional cohorts and assessed prevalence of UI-related neuromodulation procedures among men and women separately from 2004 to 2013. Analyses were conducted overall and stratified by age, race/ethnicity, and geographic region.

RESULTS

Nearly all neuromodulation procedures occurred in outpatient settings. Sacral neuromodulation (SNM) procedures for UI in both women and men grew steadily from 2004 to 2013, with more procedures performed in women than men. Among women with UI, SNM prevalence grew from 0.1%-0.2% in 2004 to 0.5%-0.6% in 2013. Posterior tibial nerve stimulation (PTNS) experienced growth from 2011 to 2013. Chemodenervation of the bladder with onabotulinumtoxinA (BTX) combined with other injectable procedures (including urethral bulking) remained stable over time.

CONCLUSIONS

From 2004 to 2013, SNM procedures remained relatively uncommon but increased consistently. PTNS experienced growth starting in 2011 when PTNS-specific insurance claims became available. BTX trends remain unclear; future studies should assess it separately from other injectable procedures. Neuromodulation has a growing role in UI treatment, and ongoing trends will be important to examine.

Pathogen burden and leukocyte telomere length in the United States

BACKGROUND

Prior studies in humans have suggested that telomere shortening may be accelerated by infection, but research on multiple pathogens and use of large population-based study samples has been limited. We estimated cross-sectional associations between seropositivity to five persistent pathogens (Herpes Simplex Virus Type-1 (HSV-1), Herpes Simplex Virus Type-2 (HSV-2), cytomegalovirus (CMV), Helicobacter pylori (H.pylori), and Hepatitis B) as well as total pathogen burden and leukocyte telomere length. Data were derived from the National Health and Nutrition Examination Survey (1999-2000) for individuals 20-49 years of age, N = 1708. We analyzed the influence of each pathogen separately, a pathogen count score and a latent class model of pathogen burden on log telomere length using linear regression models, adjusted for covariates.

RESULTS

Individuals in a latent pathogen burden class characterized by high probabilities of infection with HSV-1, CMV, and H. pylori, had significantly decreased log telomere length (- 0.30 [95% CI: - 0.36, - 0.24]) compared to those in a latent class characterized by low probabilities of all five infections. There were limited significant associations using other pathogen measures.

CONCLUSIONS

These results suggest that infection with specific combinations of pathogens may be one mechanism contributing to accelerated cellular senescence with possible origins early in the life course.

Use of National Asthma Guidelines by Allergists and Pulmonologists: A National Survey

BACKGROUND

Little is known about specialist-specific variations in guideline agreement and adoption.

OBJECTIVE

To assess similarities and differences between allergists and pulmonologists in adherence to cornerstone components of the National Asthma Education and Prevention Program's Third Expert Panel Report.

METHODS

Self-reported guideline agreement, self-efficacy, and adherence were assessed in allergists (n = 134) and pulmonologists (n = 99) in the 2012 National Asthma Survey of Physicians. Multivariate models were used to assess if physician and practice characteristics explained bivariate associations between specialty and "almost always" adhering to recommendations (ie, ≥75% of the time).

RESULTS

Allergists and pulmonologists reported high guideline self-efficacy and moderate guideline agreement. Both groups "almost always" assessed asthma control (66.2%, standard error [SE] 4.3), assessed school/work asthma triggers (71.3%, SE, 3.9), and endorsed inhaled corticosteroids use (95.5%, SE 2.0). Repeated assessment of the inhaler technique, use of asthma action/treatment plans, and spirometry were lower (39.7%, SE 4.0; 30.6%, SE 3.6; 44.7%, SE 4.1, respectively). Compared with pulmonologists, more allergists almost always performed spirometry (56.6% vs 38.6%, P = .06), asked about nighttime awakening (91.9% vs 76.5%, P = .03) and emergency department visits (92.2% vs 76.5%, P = .03), assessed home triggers (70.5% vs 52.6%, P = .06), and performed allergy testing (61.8% vs 21.3%, P < .001). In multivariate analyses, practice-specific characteristics explained differences except for allergy testing.

CONCLUSIONS

Overall, allergists and pulmonologists adhere to the asthma guidelines with notable exceptions, including asthma action plan use and inhaler technique assessment. Recommendations with low implementation offer opportunities for further exploration and could serve as targets for increasing guideline uptake.

0367 Social Determinants of Black-White Disparities in the Work-Sleep Relationship by Occupational Class: A Sequential Mixed Methods Approach

Introduction

Although Black adults disproportionately work in lower-wage, lower-skilled jobs and experience short sleep (&lt;7 hours), which has been shown to vary by employment industry and occupation, there is scant literature regarding the influence of the work-sleep relationship on racial/ethnic sleep disparities. Our prior quantitative research based on nationally-representative data revealed a novel finding that the prevalence of short sleep was generally highest at professional occupational classes among Black adults but was the least prevalent among their White counterparts.

Methods

To identify reasons for short sleep generally increasing with increasing professional occupations among blacks but decreasing among whites, we conducted a qualitative study using a sequential mixed methods design among Black and White workers across a range of industries and occupations. Occupations were classified as “professional” (e.g., doctors; lawyers) or “non-professional” (e.g., retail; food service). Race-matched trained facilitators conducted 36 focus groups that were homogenous in terms of race-sex/gender-occupational class and 63 one-on-one interviews (N=334 overall participants) using semi-structured interview guides. NVivo software was used to identify themes/patterns.

Results

Participants were a mean age of 41 ± 11 years, 42% were men, 58% had an annual income of ≥$50,000, and 57% were professionals. Black professionals overwhelmingly described less informational and emotional support as well as needing to “work twice as hard to get half as far” (i.e., John Henryism) compared to coworkers as potential explanations for work-sleep disparities. Both Black and White professionals identified longstanding social structures, interpersonal discrimination, income disparities, and familial or self-imposed pressures to succeed. White professional women frequently reported experiences with gender discrimination, which - they reported - may intersect with and amplify the effects of racial discrimination among Black women. Regardless of occupational class, Black men additionally described unique stressors (e.g., political climate; finances; police). White men frequently avoided discussing disparities, and the existence of disparities was often denied/questioned by non-professionals across race and sex/gender.

Conclusion

Our findings inform future research and interventions designed to illuminate and/or address sleep disparities emanating from the workplace.

Support

This work was funded by the Intramural Program at the National Institute of Environmental Health Sciences (Z1AES103325-01).

Association of urinary levels of bisphenols F and S used as bisphenol A substitutes with asthma and hay fever outcomes

BACKGROUND

Bisphenols F (BPF) and S (BPS) are bisphenol A (BPA) analogs used as substitutes in consumer products. Despite previous reports of BPA's association with asthma, no studies have examined its structural analogs in relation to asthma and allergy outcomes.

OBJECTIVE

To examine the association of urinary BPF, BPS, and BPA with asthma and hay fever in a US representative sample.

METHODS

We analyzed data from 3,538 participants aged 12 years or older in the 2013-2016 National Health and Nutrition Examination Survey (NHANES). Children aged 6-11 years (N = 738), who did not have all covariate data available, were analyzed separately. Covariate-adjusted logistic regression was used to assess the association of the exposures with the outcomes.

RESULTS

BPF, BPS, and BPA were detected in 57.1%, 88.4%, and 94.8% of the urine samples, respectively. Urinary BPF detection was positively associated with current asthma (odds ratio [OR]: 1.54, 95% confidence interval [CI]: 1.16-2.04) and hay fever (OR: 1.66, 95% CI: 1.12-2.46). Urinary BPS was associated with increased odds of current asthma in men (OR: 1.64, 95% CI: 1.13-2.40) and urinary BPA was associated with increased odds of asthma without hay fever in children aged 6-11 years (OR: 2.65, 95% CI: 1.05-6.68).

CONCLUSION

Our nationally-representative findings document that BPF and BPS exposure is common in the US and that exposure to these BPA analogs is associated with asthma and/or hay fever. Our results suggest that BPF and BPS may not be safe alternatives to BPA; however, prospective studies should be conducted to confirm these results.

Everyday and major experiences of racial/ethnic discrimination and sleep health in a multiethnic population of U.S. women: findings from the Sister Study

BACKGROUND

Perceived racial/ethnic discrimination and poor sleep occur across all races/ethnicities in the U.S., although both are most common among racial/ethnic minorities. Few studies have investigated associations between perceived racial/ethnic discrimination and various sleep dimensions in a multiethnic population.

METHODS

We analyzed cross-sectional associations among 40,038 eligible Sister Study participants (enrollment: 2003-2009) who reported ever/never experiencing specific types of everyday (eg, treated unfairly at a store or restaurant) or major (eg, unfairly stopped, threatened, or searched by police) discrimination attributed to their race/ethnicity during a follow-up survey in 2008-2012. Participants also reported short sleep duration (<7 h), sleep debt (≥2-h difference between longest and shortest sleep duration), frequent napping (≥3 times/week), and insomnia. Poisson regression with robust variance estimation, adjusted for sociodemographic and health characteristics, estimated prevalence ratios (PRs) and 95% confidence intervals (CIs) for the association between each type of racial/ethnic discrimination and each sleep dimension, overall and by race/ethnicity.

RESULTS

Mean age was 55 ± 8.9 years, 89% were Non-Hispanic (NH)-white, 8% NH-black, and 3% Hispanic/Latina. NH-black participants were the most likely to report everyday (76% vs. 4% [NH-whites] and 36% [Hispanics/Latinas]) and major racial/ethnic discrimination (52% vs. 2% [NH-whites] and 18% [Hispanics/Latinas]). Participants who experienced both types versus neither were more likely to report short sleep duration (PR = 1.17 [95% CI: 1.09-1.25]) and insomnia symptoms (PR = 1.10 [1.01-1.20]) but not other poor sleep dimensions.

CONCLUSIONS

Racial/ethnic minority women were most likely to experience racial/ethnic discrimination, which was associated with certain poor sleep dimensions among women of all races/ethnicities.

Synergistic Association of House Endotoxin Exposure and Ambient Air Pollution with Asthma Outcomes

Rationale: House endotoxin and ambient air pollution are risk factors for asthma; however, the effects of their coexposure on asthma are not well characterized.Objectives: To examine potential synergistic associations of coexposure to house dust endotoxin and ambient air pollutants with asthma outcomes.Methods: We analyzed data of 6,488 participants in the National Health and Nutrition Examination Survey 2005-2006. Dust from bedding and bedroom floor was analyzed for endotoxin content. The Community Multiscale Air Quality Modeling System (CMAQ) and Downscaler Model data were used to determine annual average particulate matter ≤2.5 μm in aerodynamic diameter (PM_2.5), ozone (O₃), and nitrogen dioxide (NO₂) exposures at participants' residential locations. The associations of the coexposures with asthma outcomes were assessed and tested for synergistic interaction.Measurements and Main Results: In adjusted analysis, PM_2.5 (CMAQ) (odds ratio [OR], 1.12; 95% confidence interval [CI], 1.07-1.18), O₃ (Downscaler Model) (OR, 1.07; 95% CI, 1.02-1.13), and log₁₀ NO₂ (CMAQ) (OR, 3.15; 95% CI, 1.33-7.45) were positively associated with emergency room visits for asthma in the past 12 months. Coexposure to elevated concentrations of house dust endotoxin and PM_2.5 (CMAQ) was synergistically associated with the outcome, increasing the odds by fivefold (OR, 5.01; 95% CI, 2.54-9.87). A synergistic association was also found for coexposure to higher concentrations of endotoxin and NO₂ in children (OR, 3.45; 95% CI, 1.65-7.18).Conclusions: Coexposure to elevated concentrations of residential endotoxin and ambient PM_2.5 in all participants and NO₂ in children is synergistically associated with increased emergency room visits for asthma. Therefore, decreasing exposure to both endotoxin and air pollution may help reduce asthma morbidity.

Racial/ethnic disparities in sleep duration and sleep disturbances among pregnant and non-pregnant women in the United States

Sleep disturbances among pregnant women are increasingly linked to suboptimal maternal/birth outcomes. Few studies in the USA investigating sleep by pregnancy status have included racially/ethnically diverse populations, despite worsening disparities in adverse birth outcomes. Using a nationally representative sample of 71,644 (2,349 pregnant) women from the National Health Interview Survey (2004-2017), we investigated relationships between self-reported pregnancy and six sleep characteristics stratified by race/ethnicity. We also examined associations between race/ethnicity and sleep stratified by pregnancy status. We used average marginal predictions from fitted logistic regression models to estimate prevalence ratios (PRs) and 95% confidence intervals (CIs) for each sleep dimension, adjusting for sociodemographic and health characteristics. Pregnant women were less likely than non-pregnant women to report short sleep (PR_Overall  = 0.75; 95% CI, 0.68-0.82) and more likely to report long sleep (PR_Overall  = 2.06; 95% CI, 1.74-2.43) and trouble staying asleep (PR_Overall  = 1.34; 95% CI, 1.25-1.44). The association between pregnancy and sleep duration was less pronounced among women aged 35-49 years compared to those <35 years. Among white women, sleep medication use was less prevalent among pregnant compared to non-pregnant women (PR_White  = 0.45; 95% CI, 0.31-0.64), but this association was not observed among black women (PR_Black  = 0.98; 95% CI, 0.46-2.09) and was less pronounced among Hispanic/Latina women (PR_{Hispanic/Latina}  = 0.82; 95% CI, 0.38-1.77). Compared to pregnant white women, pregnant black women had a higher short sleep prevalence (PR_Black  = 1.35; 95% CI, 1.08-1.67). Given disparities in maternal/birth outcomes and sleep, expectant mothers (particularly racial/ethnic minorities) may need screening followed by treatment for sleep disturbances. Our findings should be interpreted in the historical and sociocultural context of the USA.

Perceived discrimination and depressive symptoms among US Latinos: the modifying role of educational attainment

OBJECTIVE

Despite growing evidence that discrimination may contribute to poor mental health, few studies have assessed this association among US Latinos. Furthermore, the interaction between discrimination and educational attainment in shaping Latino mental health is virtually unexplored. This study aims to examine the association between perceived discrimination and depressive symptoms and the modifying role of education among a population of Mexican-origin adults.

DESIGN

We utilized population-based data from 629 Mexican-origin adults (mean age = 52.8 years) participating the Niños Lifestyle and Diabetes Study (2013-2014). Perceived discrimination was defined as responding 'sometimes' or 'often' to at least one item on the 9-item Everyday Discrimination Scale. High depressive symptoms were defined as scoring ≥10 on the CESD-10. We used log-binomial and linear-binomial models to estimate prevalence ratios (PR) and prevalence differences (PD), respectively, of high depressive symptoms for levels of perceived discrimination. Final models were adjusted for age, sex, education, cultural orientation, and nativity. General estimating equations were employed to account for within-family clustering.

RESULTS

Prevalence of perceived discrimination and high depressive symptoms were 49.5% and 29.2%, respectively. Participants experiencing discrimination had higher depressive symptom prevalence than those never or rarely experiencing discrimination [PR = 1.94, 95% confidence interval (CI): 1.46-2.58; PD = 0.19, 95% CI: 0.12-0.27]. The strength of this association varied by education level. The association between discrimination and depressive symptoms was stronger among those with >12 years of education (PR = 2.69; PD = 0.24) compared to those with ≤12 years of education (PR = 1.36; PD = 0.09).

CONCLUSION

US Latinos suffer a high burden of depressive symptoms, and discrimination may be an important driver of this burden. Our results suggest that effortful coping strategies, such as achieving high education despite high perceived discrimination, may magnify discrimination's adverse effect on Latino mental health.

The legacy of redlining in the effect of foreclosures on Detroit residents' self-rated health

Historical practices, such as housing discrimination in Detroit, have been shown to have lasting impacts on communities. Perhaps the most explicit example is the practice of redlining in the 1930s, whereby lenders outlined financially undesirable neighborhoods, populated by minority families, on maps and prevented residents from moving to better resourced neighborhoods. Awareness of historical housing discrimination may improve research assessing the impacts of current neighborhood characteristics on health. Using the Detroit Neighborhood Health Study (DNHS), we assessed the association between two-year changes in home foreclosure rates following the 2007-2008 Great Recession, and residents' five-year self-rated health trajectories (2008-2013); and estimated the confounding bias introduced by ignoring historical redlining practices in the city. We used both ecological and multilevel models to make inference about person- and community-level processes. In a neighborhood-level linear regression adjusted for confounders (including percent redlined); a 10%-point slower foreclosure rate recovery was associated with an increase in prevalence of poor self-rated health of 0.31 (95% CI:-0.02 to 0.64). At the individual level, it was associated with a within-person increase in probability of poor health of 0.45 (95% CI:0.15-0.72). Removing redlining from the model biased the estimated effect upward to 0.38 (95% CI:0.07-0.69) and 0.56 (95% CI:0.21-0.84) in the neighborhood and individual-level models, respectively. Stratum-specific foreclosure recovery effects indicate stronger influence in neighborhoods with a greater proportion of residents identifying as white and a greater degree of historic redlining. These findings support earlier theory suggesting a historical influence of structural discrimination on the association between current neighborhood characteristics and health, and suggests that historical redlining specifically may increase vulnerability to contemporary neighborhood foreclosures. Community interventions should consider historical discrimination in conjunction with current place-based indicators to more equitably improve population health.

The Great Recession and Immune Function

The Great Recession precipitated unprecedented home foreclosures increases, but documentation of related neighborhood changes and population health is scant. Using the Detroit Neighborhood Health Study (N = 277), we examined associations between neighborhood-level recession indicators and thymic function, a life course immunological health indicator. In covariate-adjusted multilevel models, each 10 percentage point increase in abandoned home prevalence and 1 percentage point increase in 2009 home foreclosures was associated with 1.7-year and 3.3-year increases in thymic aging, respectively. Associations attenuated after adjustment for neighborhood-level social cohesion, suggesting community ties may buffer recession-related immune aging. Effects of neighborhood stressors were strongest in middle-income households, supporting theory of excess vulnerability in this group. Future research should assess whether ongoing foreclosure and blight reduction efforts improve health for residents of recession impacted neighborhoods.

Income and Markers of Immunological Cellular Aging

OBJECTIVE

Socioeconomic disadvantage may contribute to poor health through immune-related biological mechanisms. We examined the associations between socioeconomic status, as measured by annual household income, and T-cell markers of aging, including the ratios of CD4 and CD8 effector cells to naïve cells (E/N ratio) and the CD4/CD8 T-cell ratio. We hypothesized that participants with a lower income would have higher E/N ratios and lower CD4/CD8 ratios compared with participants with a higher income, and that these associations would be partially mediated by elevated cytomegalovirus (CMV) IgG antibody levels, a virus implicated in aging and clonal expansion of T cells.

METHODS

Data were from 79 individuals who participated in the population-based Detroit Neighborhood Health Study. We used linear regression to quantify the association between a $10,000 decrease in income and each ratio outcome.

RESULTS

After adjustment for age, sex, race, smoking, medication use, and lifetime history of mental health conditions, lower income was associated with a 0.41 (95% confidence interval = 0.09-0.72) log-unit increase in the CD4 E/N ratio and a 0.20 (95% confidence interval = 0.02-0.39) log-unit increase in the CD8 E/N ratio. CMV immunoglobulin G antibody level partially mediated these associations.

CONCLUSIONS

Our study suggests that low socioeconomic status is associated with immunological aging as measured by the E/N ratio and that impaired immune control of CMV may partially mediate these associations.

Population Distributions of Thymic Function in Adults: Variation by Sociodemographic Characteristics and Health Status

The thymus is critical for mounting an effective immune response and maintaining health. However, epidemiologic studies characterizing thymic function in the population setting are lacking. Using data from 263 adults in the Detroit Neighborhood Health Study, we examined thymic function as measured by the number of signal joint T-cell receptor excision circles (sjTREC) and assessed associations with established indicators of physiological health. Overall, increasing age and male gender were significantly associated with reduced thymic function. Adjusting for covariates, individuals with elevated levels of the pro-inflammatory biomarkers C-reactive protein (β: -0.50 [95% CI: -0.82, -0.18] for moderate elevation, β: -0.29 [95% CI: -0.59, 0.00] for high elevation) and interleukin-6 (β: -0.60 [95% CI: -0.92, -0.28] for moderate elevation, β: -0.43 [95% CI: -0.77, -0.08] for severe elevation) also had lower thymic function. Compared to individuals with a BMI < 25, individuals who were overweight (β: 0.36 [95% CI: 0.07, 0.64]) or obese (β: 0.27 [95% CI: -0.03, 0.56]) had higher thymic function. Differences by self-rated health were not statistically significant. Our findings underscore demographic- and health-related gradients in thymic function among adult residents of Detroit, suggesting thymic function may be an important biomarker of health status in adults at the population level.

Implementation and Operational Research: Decentralization Does Not Assure Optimal Delivery of PMTCT and HIV-Exposed Infant Services in a Low Prevalence Setting

BACKGROUND

The consequences of decentralizing prevention of mother-to-child HIV transmission and HIV-exposed infant services to antenatal care (ANC)/labor and delivery (L&D) sites from dedicated HIV care and treatment (C&T) centers remain unknown, particularly in low prevalence settings.

METHODS

In a cohort of mother-infant pairs, we compared delivery of routine services at ANC/L&D and C&T facilities in Kinshasa, Democratic Republic of Congo from 2010-2013, using methods accounting for competing risks (eg, death). Women could opt to receive interventions at 90 decentralized ANC/L&D sites, or 2 affiliated C&T centers. Additionally, we assessed decentralization's population-level impacts by comparing proportions of women and infants receiving interventions before (2009-2010) and after (2011-2013) decentralization.

RESULTS

Among newly HIV-diagnosed women (N = 1482), the 14-week cumulative incidence of receiving the package of CD4 testing and zidovudine or antiretroviral therapy was less at ANC/L&D [66%; 95% confidence interval (CI): 63% to 69%] than at C&T (88%; 95% CI: 83% to 92%) sites (subdistribution hazard ratio, 0.62; 95% CI: 0.55 to 0.69). Delivery of cotrimoxazole and DNA polymerase chain reaction testing to HIV-exposed infants (N = 1182) was inferior at ANC/L&D sites (subdistribution hazard ratio, 0.84; 95% CI: 0.76 to 0.92); the 10-month cumulative incidence of the package at ANC/L&D sites was 89% (95% CI: 82% to 93%) versus 97% (95% CI: 93% to 99%) at C&T centers. Receipt of the pregnancy (20% of 1518, to 64% of 1405) and infant (16%-31%) packages improved post decentralization.

CONCLUSIONS

Services were delivered less efficiently at ANC/L&D sites than C&T centers. Although access improved with decentralization, its potential cannot be realized without sufficient and sustained support.

Temporal changes in the outcomes of HIV-exposed infants in Kinshasa, Democratic Republic of Congo during a period of rapidly evolving guidelines for care (2007-2013)

OBJECTIVE

Guidelines for prevention of mother-to-child transmission of HIV have developed rapidly, yet little is known about how outcomes of HIV-exposed infants have changed over time. We describe HIV-exposed infant outcomes in Kinshasa, Democratic Republic of Congo, between 2007 and 2013.

DESIGN

Cohort study of mother-infant pairs enrolled in family-centered comprehensive HIV care.

METHODS

Accounting for competing risks, we estimated the cumulative incidences of early infant diagnosis, HIV transmission, death, loss to follow-up, and combination antiretroviral therapy (cART) initiation for infants enrolled in three periods (2007-2008, 2009-2010, and 2011-2012).

RESULTS

1707 HIV-exposed infants enrolled at a median age of 2.6 weeks. Among infants whose mothers had recently enrolled into HIV care (N = 1411), access to EID by age two months increased from 28% (95% confidence limits [CL]: 24,34%) among infants enrolled in 2007-2008 to 63% (95% CL: 59,68%) among infants enrolled in 2011-2012 (Gray's p-value <0.01). The 18-month cumulative incidence of HIV declined from 16% (95% CL: 11,22%) for infants enrolled in 2007-2008 to 11% (95% CL: 8,16%) for infants enrolled in 2011-2012 (Gray's p-value = 0.19). The 18-month cumulative incidence of death also declined, from 8% (95% CL: 5,12%) to 3% (95% CL: 2,5%) (Gray's p-value = 0.02). LTFU did not improve, with 18-month cumulative incidences of 19% (95% CL: 15,23%) for infants enrolled in 2007-2008 and 22% (95% CL: 18,26%) for infants enrolled in 2011-2012 (Gray's p-value = 0.06). Among HIV-infected infants, the 24-month cumulative incidence of cART increased from 61% (95% CL: 43,75%) to 97% (95% CL: 82,100%) (Gray's p-value <0.01); the median age at cART decreased from 17.9 to 9.3 months. Outcomes were better for infants whose mothers enrolled before pregnancy.

CONCLUSIONS

We observed encouraging improvements, but continued efforts are needed.

Antenatal and delivery services in Kinshasa, Democratic Republic of Congo: care-seeking and experiences reported by women in a household-based survey

OBJECTIVES

Increasing coverage of quality reproductive health services, including prevention of mother-to-child transmission services, requires understanding where and how these services are provided. To inform scale-up, we conducted a population-based survey in Kinshasa, Democratic Republic of Congo.

METHODS

Stratified two-stage cluster sampling was used to select women ≥18 years old who had been pregnant within the prior three years. Participants were interviewed about their reproductive healthcare utilization and impressions of services received.

RESULTS

We interviewed 1221 women, 98% of whom sought antenatal care (ANC). 78% of women began ANC after the first trimester and 22% reported <4 visits. Reasons for choosing an ANC facility included reputation (51%), friendly/accessible staff (39%), availability of comprehensive services (29%), medication access (26%), location (26%), and cost (21%). Most women reported satisfactory treatment by staff, but 47% reported that the ANC provider ignored their complaints, 23% had difficulty understanding responses to their questions, 22% wanted more time with the provider, 21% wanted more privacy, and 12% felt uncomfortable asking questions. Only 56% reported someone talked to them about HIV/AIDS. Strongest predictors of seeking inadequate ANC included low participant and partner education and lack of certain assets. Only 32% of women sought postnatal care. Some results varied by health zone.

CONCLUSIONS

Scaling-up interventions to improve reproductive health services should include broad-based health systems strengthening and promote equitable access to quality ANC, delivery, and postnatal services. Personal and structural-level barriers to seeking ANC need to be addressed, with consideration given to local contexts.

Reducing transplant toxicity

Conventional myeloablative allogeneic hematopoietic cell transplantation produces considerable morbidity and mortality. These generally limit this treatment to patients in good medical condition who are younger than 55 years of age. T-cell-mediated graft-versus-tumor effects play a key role in the elimination of malignancy after allografting. Several investigators have sought to reduce regimen-related toxicities while optimizing graft-versus-tumor effects. Strategies can be broadly classified as (1) reduced-intensity regimens that retain some toxicity, and (2) minimally myelosuppressive regimens that rely on immunosuppression for allogeneic engraftment and resultant graft-versus-tumor effects. Although follow-up has been short, preliminary results are encouraging. Current challenges include defining a regimen that will facilitate full donor engraftment while minimizing toxicities and graft-versus-host disease. If long-term efficacy is demonstrated, such strategies will expand the options for patients who would not qualify for conventional allogeneic transplants.

Nonmyeloablative hematopoietic cell transplantation. Replacing high-dose cytotoxic therapy by the graft-versus-tumor effect

Conventional allografting produces considerable regimen-related toxicities that generally limit this treatment to patients younger than 55 years and in otherwise good medical condition. T cell-mediated graft-versus-tumor (GVT) effects are known to play an important role in the elimination of malignant disease after allotransplants. A minimally myelosuppressive regimen that relies on immunosuppression for allogeneic engraftment was developed to reduce toxicities while optimizing GVT effects. Pre-transplant total-body irradiation (200 cGy) followed by post-transplant immunosuppression with cyclosporine (CSP) and mycophenolate mofetil (MMF) permitted human leukocyte antigen (HLA)-matched sibling donor hematopoietic cell engraftment in 82% of patients (n = 55) without prior high-dose therapy. The addition of fludarabine (90 mg/m2) facilitated engraftment in all 28 subsequent patients. Overall, fatal progression of underlying disease occurred in 20% of patients after transplant. Non-relapse mortality occurred in 11% of patients. Toxicities were low. Grade 2-4 acute graft-versus-host disease (GVHD) associated with primary engraftment developed in 47% of patients, and was readily controlled in all but two patients. Donor lymphocyte infusions (DLI) were not very effective at converting a low degree of mixed donor/host chimerism to full donor chimerism; however, the addition of fludarabine reduced the need for DLI. With a median follow-up of 244 days, 68% of patients were alive, with 42% of patients in complete remission, including molecular remissions. Remissions occurred gradually over periods of weeks to a year. If long-term efficacy is demonstrated, such a strategy would expand treatment options for patients who would otherwise be excluded from conventional allografting.

Nonmyeloablative hematopoietic cell transplantation

The myeloablative doses of chemotherapy and radiation used with conventional allogeneic hematopoietic cell transplantation produce considerable morbidity and mortality that generally limit this treatment to patients younger than 55 years of age and in good general medical condition. It has become clear that T-cell-mediated graft-versus-tumor effects play an important role in the elimination of malignant disease after allotransplants. Several investigators have sought to reduce regimen-related toxicities while optimizing graft-versus-tumor effects. Strategies can be broadly categorized as reduced-intensity regimens that retain some toxicities and require hospitalization, and minimally myelosuppressive regimens that rely on immunosuppression for allogeneic engraftment and resultant graft-versus-tumor effects. The latter approach can be performed in the ambulatory care setting. Preliminary results are encouraging. If long-term efficacy is demonstrated, such strategies would expand treatment options for patients who would otherwise be excluded from receiving conventional allografts.

Predicting the local dynamics of epizootic rabies among raccoons in the United States

Mathematical models have been developed to explore the population dynamics of viral diseases among wildlife. However, assessing the predictions stemming from these models with wildlife databases adequate in size and temporal duration is uncommon. An epizootic of raccoon rabies that began in the mid-Atlantic region of the United States in the late 1970s has developed into one of the largest and most extensive in the history of wildlife rabies. We analyzed the dynamics of local epizootics at the county level by examining a database spanning more than 20 years and including 35,387 rabid raccoons. The size, number, and periodicity of rabies epizootics among raccoons were compared with predictions derived from a susceptible, exposed, infectious, and recovered model of raccoon rabies [Coyne, J., Smith, G. & McAllister, F. E. (1989) Am. J. Vet. Res. 50, 2148-2154]. After our methods for defining epizootics were applied to solutions of the model, the time series revealed recurrent epizootics in some counties, with a median first epizootic period of 48 months. Successive epizootics declined in size and the epizootic period progressively decreased. Our reanalysis of the model predicted the initial-epizootic period of 4-5 years, with a progressive dampening of epizootic size and progressive decrease in epizootic period. The best quantitative agreement between data and model assumed low levels of immunity (1-5%) within raccoon populations, suggesting that raccoons develop little or no rabies immune class. These results encourage the use of data obtained through wildlife surveillance in assessing and refining epidemic models for wildlife diseases.

Will your academic department survive managed care?

The current form of academic department is likely to vanish from many institutions. Changes occurring in health care are part of the evolution other industries have experienced, following the product life cycle. Physicians are becoming "deprofessionalized" and as such are beginning to resemble technical workers seen in other industries. The rearrangements in health care are bringing together organizations with different missions, priorities, culture and even language. An academic department may not be considered as an asset to the larger organization or network, representing but one option for product differentiation in the market place. There are strategies for maintaining the viability of the academic component of an organization that necessitate congruence with the overall strategy for the greater organization.

X-Ray-Induced Bulk Photovoltaic Effect in Insulators

When an insulator, sandwiched between metals of dissimilar work functions, is irradiated with x-radiation, a voltage related to the contact potential difference of the metals is observed. This phenomenon, known as the bulk photovoltaic effect, has been demonstrated in a variety of metal-insulator-metal combinations. Evidence that the voltage is not generated by junctions, barrier layers, or simillar other phenomena is presented.