Alan Hubbard

To GEE or not to GEE: comparing population average and mixed models for estimating the associations between neighborhood risk factors and health.

Two modeling approaches are commonly used to estimate the associations between neighborhood characteristics and individual-level health outcomes in multilevel studies (subjects within neighborhoods). Random effects models (or mixed models) use maximum likelihood estimation. Population average models typically use a generalized estimating equation (GEE) approach. These methods are used in place of basic regression approaches because the health of residents in the same neighborhood may be correlated, thus violating independence assumptions made by traditional regression procedures. This violation is particularly relevant to estimates of the variability of estimates. Though the literature appears to favor the mixed-model approach, little theoretical guidance has been offered to justify this choice. In this paper, we review the assumptions behind the estimates and inference provided by these 2 approaches. We propose a perspective that treats regression models for what they are in most circumstances: reasonable approximations of some true underlying relationship. We argue in general that mixed models involve unverifiable assumptions on the data-generating distribution, which lead to potentially misleading estimates and biased inference. We conclude that the estimation-equation approach of population average models provides a more useful approximation of the truth.

Effect of reduction in household air pollution on childhood pneumonia in Guatemala (RESPIRE): a randomised controlled trial

BACKGROUND Pneumonia causes more child deaths than does any other disease. Observational studies have indicated that smoke from household solid fuel is a significant risk factor that affects about half the worlds children. We investigated whether an intervention to lower indoor wood smoke emissions would reduce pneumonia in children. METHODS We undertook a parallel randomised controlled trial in highland Guatemala, in a population using open indoor wood fires for cooking. We randomly assigned 534 households with a pregnant woman or young infant to receive a woodstove with chimney (n=269) or to remain as controls using open woodfires (n=265), by concealed permuted blocks of ten homes. Fieldworkers visited homes every week until children were aged 18 months to record the childs health status. Sick children with cough and fast breathing, or signs of severe illness were referred to study physicians, masked to intervention status, for clinical examination. The primary outcome was physician-diagnosed pneumonia, without use of a chest radiograph. Analysis was by intention to treat (ITT). Infant 48-h carbon monoxide measurements were used for exposure-response analysis after adjustment for covariates. This trial is registered, number ISRCTN29007941. FINDINGS During 29,125 child-weeks of surveillance of 265 intervention and 253 control children, there were 124 physician-diagnosed pneumonia cases in intervention households and 139 in control households (rate ratio [RR] 0·84, 95% CI 0·63-1·13; p=0·257). After multiple imputation, there were 149 cases in intervention households and 180 in controls (0·78, 0·59-1·06, p=0·095; reduction 22%, 95% CI -6% to 41%). ITT analysis was undertaken for secondary outcomes: all and severe fieldworker-assessed pneumonia; severe (hypoxaemic) physician-diagnosed pneumonia; and radiologically confirmed, RSV-negative, and RSV-positive pneumonia, both total and severe. We recorded significant reductions in the intervention group for three severe outcomes-fieldworker-assessed, physician-diagnosed, and RSV-negative pneumonia--but not for others. We identified no adverse effects from the intervention. The chimney stove reduced exposure by 50% on average (from 2·2 to 1·1 ppm carbon monoxide), but exposure distributions for the two groups overlapped substantially. In exposure-response analysis, a 50% exposure reduction was significantly associated with physician-diagnosed pneumonia (RR 0·82, 0·70-0·98), the greater precision resulting from less exposure misclassification compared with use of stove type alone in ITT analysis. INTERPRETATION In a population heavily exposed to wood smoke from cooking, a reduction in exposure achieved with chimney stoves did not significantly reduce physician-diagnosed pneumonia for children younger than 18 months. The significant reduction of a third in severe pneumonia, however, if confirmed, could have important implications for reduction of child mortality. The significant exposure-response associations contribute to causal inference and suggest that stove or fuel interventions producing lower average exposures than these chimney stoves might be needed to substantially reduce pneumonia in populations heavily exposed to biomass fuel air pollution. FUNDING US National Institute of Environmental Health Sciences and WHO.

Toward Understanding the Risk of Secondary Airborne Infection: Emission of Respirable Pathogens

Certain respiratory tract infections are transmitted through air. Coughing and sneezing by an infected person can emit pathogen-containing particles with diameters less than 10 μ m that can reach the alveolar region. Based on our analysis of the sparse literature on respiratory aerosols, we estimated that emitted particles quickly decrease in diameter due to water loss to one-half the initial values, and that in one cough the volume in particles with initial diameters less than 20 μ m is 6 × 10− 8 mL. The pathogen emission rate from a source case depends on the frequency of expiratory events, the respirable particle volume, and the pathogen concentration in respiratory fluid. Viable airborne pathogens are removed by exhaust ventilation, particle settling, die-off, and air disinfection methods; each removal mechanism can be assigned a first-order rate constant. The pathogen concentration in well-mixed room air depends on the emission rate, the size distribution of respirable particles carrying pathogens, and the removal rate constants. The particle settling rate and the alveolar deposition fraction depend on particle size. Given these inputs plus a susceptible persons breathing rate and exposure duration to room air, an expected alveolar dose μ is estimated. If the infectious dose is one organism, as appears to be true for tuberculosis, infection risk is estimated by the expression: R = 1 − exp(−μ). Using published tuberculosis data concerning cough frequency, bacilli concentration in respiratory fluid, and die-off rate, we illustrate the model via a plausible scenario for a person visiting the room of a pulmonary tuberculosis case. We suggest that patients termed “superspreaders” or “dangerous disseminators” are those infrequently encountered persons with high values of cough and/or sneeze frequency, elevated pathogen concentration in respiratory fluid, and/or increased respirable aerosol volume per expiratory event such that their pathogen emission rate is much higher than average.

Diagnostic Accuracy of Canine Scent Detection in Early- and Late-Stage Lung and Breast Cancers

Background: Lung and breast cancers are leading causes of cancer death worldwide. Prior exploratory work has shown that patterns of biochemical markers have been found in the exhaled breath of patients with lung and breast cancers that are distinguishable from those of controls. However, chemical analysis of exhaled breath has not shown suitability for individual clinical diagnosis. Methods: The authors used a food reward-based method of training 5 ordinary household dogs to distinguish, by scent alone, exhaled breath samples of 55 lung and 31 breast cancer patients from those of 83 healthy controls. A correct indication of cancer samples by the dogs was sitting/lying in front of the sample. A correct response to control samples was to ignore the sample. The authors first trained the dogs in a 3-phase sequential process with gradually increasing levels of challenge. Once trained, the dogs’ ability to distinguish cancer patients from controls was then tested using breath samples from subjects not previously encountered by the dogs. The researchers blinded both dog handlers and experimental observers to the identity of breath samples. The diagnostic accuracy data reported were obtained solely from the dogs’ sniffing, in double-blinded conditions, of these breath samples obtained from subjects not previously encountered by the dogs during the training period. Results: Among lung cancer patients and controls, overall sensitivity of canine scent detection compared to biopsy-confirmed conventional diagnosis was 0.99 (95% confidence interval [CI], 0.99, 1.00) and overall specificity 0.99 (95% CI, 0.96, 1.00). Among breast cancer patients and controls, sensitivity was 0.88 (95% CI, 0.75, 1.00) and specificity 0.98 (95% CI, 0.90, 0.99). Sensitivity and specificity were remarkably similar across all 4 stages of both diseases. Conclusion: Training was efficient and cancer identification was accurate; in a matter of weeks, ordinary household dogs with only basic behavioral “puppy training” were trained to accurately distinguish breath samples of lung and breast cancer patients from those of controls. This pilot work using canine scent detection demonstrates the validity of using a biological system to examine exhaled breath in the diagnostic identification of lung and breast cancers. Future work should closely examine the chemistry of exhaled breath to identify which chemical compounds can most accurately identify the presence of cancer.

Mitochondrial oxidative stress causes hyperphosphorylation of tau.

Age-related neurodegenerative disease has been mechanistically linked with mitochondrial dysfunction via damage from reactive oxygen species produced within the cell. We determined whether increased mitochondrial oxidative stress could modulate or regulate two of the key neurochemical hallmarks of Alzheimers disease (AD): tau phosphorylation, and ß-amyloid deposition. Mice lacking superoxide dismutase 2 (SOD2) die within the first week of life, and develop a complex heterogeneous phenotype arising from mitochondrial dysfunction and oxidative stress. Treatment of these mice with catalytic antioxidants increases their lifespan and rescues the peripheral phenotypes, while uncovering central nervous system pathology. We examined sod2 null mice differentially treated with high and low doses of a catalytic antioxidant and observed striking elevations in the levels of tau phosphorylation (at Ser-396 and other phospho-epitopes of tau) in the low-dose antioxidant treated mice at AD-associated residues. This hyperphosphorylation of tau was prevented with an increased dose of the antioxidant, previously reported to be sufficient to prevent neuropathology. We then genetically combined a well-characterized mouse model of AD (Tg2576) with heterozygous sod2 knockout mice to study the interactions between mitochondrial oxidative stress and cerebral Aß load. We found that mitochondrial SOD2 deficiency exacerbates amyloid burden and significantly reduces metal levels in the brain, while increasing levels of Ser-396 phosphorylated tau. These findings mechanistically link mitochondrial oxidative stress with the pathological features of AD.

Seasonality of rotavirus disease in the tropics: a systematic review and meta-analysis

BACKGROUND To date little conclusive evidence exists on the seasonality of rotavirus incidence in the tropics. We present a systematic review and meta-analysis on the seasonal epidemiology of rotavirus in the tropics, including 26 studies reporting continuous monthly rotavirus incidence for which corresponding climatological data was available. METHODS Using linear regression models that account for serial correlation between months, monthly rotavirus incidence was significantly negatively correlated with temperature, rainfall and relative humidity in 65%, 55% and 60% of studies, respectively. We carried out pooled analyses using a generalized estimating equation (GEE) that accounts for correlation from between-study variation and serial correlation between months within a given study. RESULTS For every 1 degrees C (1.8 degrees F) increase in mean temperature, 1 cm (0.39 in.) increase in mean monthly rainfall, and 1% increase in relative humidity (22%) this analysis showed reductions in rotavirus incidence of 10% (95% CI: 6-13%), 1% (95% CI: 0-1%), and 3% (95% CI:0-5%), respectively. CONCLUSIONS On the basis of the evidence, we conclude that rotavirus responds to changes in climate in the tropics, with the highest number of infections found at the colder and drier times of the year.

Do changes in spousal employment status lead to domestic violence? Insights from a prospective study in Bangalore, India

The prevalence of physical domestic violence--violence against women perpetrated by husbands--is staggeringly high across the Indian subcontinent. Although gender-based power dynamics are thought to underlie womens vulnerability, relatively little is known about risk and protective factors. This prospective study in southern India examined the association between key economic aspects of gender-based power, namely spousal employment status, and physical domestic violence. In 2005-2006, 744 married women, aged 16-25, residing in low-income communities in Bangalore, India were enrolled in the study. Data were collected at enrollment, 12 and 24 months. Multivariable logistic regression models were used to examine the prospective association between womens employment status, their perceptions of their husbands employment stability, and domestic violence. Women who were unemployed at one visit and began employment by the next visit had an 80% higher odds of violence, as compared to women who maintained their unemployed status. Similarly, women whose husbands had stable employment at one visit and newly had difficulty with employment had 1.7 times the odds of violence, as compared to women whose husbands maintained their stable employment. To our knowledge, this study is the first from a developing country to confirm that changes in spousal employment status are associated with subsequent changes in violence risk. It points to the complex challenges of violence prevention, including the need for interventions among men and gender-transformative approaches to promote gender-equitable attitudes, practices and norms among men and women.

An application of Random Forests to a genome-wide association dataset: Methodological considerations & new findings

BackgroundAs computational power improves, the application of more advanced machine learning techniques to the analysis of large genome-wide association (GWA) datasets becomes possible. While most traditional statistical methods can only elucidate main effects of genetic variants on risk for disease, certain machine learning approaches are particularly suited to discover higher order and non-linear effects. One such approach is the Random Forests (RF) algorithm. The use of RF for SNP discovery related to human disease has grown in recent years; however, most work has focused on small datasets or simulation studies which are limited.ResultsUsing a multiple sclerosis (MS) case-control dataset comprised of 300 K SNP genotypes across the genome, we outline an approach and some considerations for optimally tuning the RF algorithm based on the empirical dataset. Importantly, results show that typical default parameter values are not appropriate for large GWA datasets. Furthermore, gains can be made by sub-sampling the data, pruning based on linkage disequilibrium (LD), and removing strong effects from RF analyses. The new RF results are compared to findings from the original MS GWA study and demonstrate overlap. In addition, four new interesting candidate MS genes are identified, MPHOSPH9, CTNNA3, PHACTR2 and IL7, by RF analysis and warrant further follow-up in independent studies.ConclusionsThis study presents one of the first illustrations of successfully analyzing GWA data with a machine learning algorithm. It is shown that RF is computationally feasible for GWA data and the results obtained make biologic sense based on previous studies. More importantly, new genes were identified as potentially being associated with MS, suggesting new avenues of investigation for this complex disease.

Massage Therapy Attenuates Inflammatory Signaling After Exercise-Induced Muscle Damage

Beneficial effects of massage on tired muscles work through anti-inflammatory and mitochondrial biogenesis pathways. The Mechanism of Massage Massage is a popular treatment—both for its putative healing properties and because it feels good—but is not among the usual tools of physicians. To validate its usefulness and understand how massage affects muscles in biomedical terms, Crane and his colleagues have documented the biological changes that massage evokes in the leg muscles of 11 young men who had just pushed themselves to exhaustion with heavy exercise. The exercise itself caused massive changes in gene expression, but after 10 min of massage, signaling pathways responsive to mechanical stresses were activated. Massage reduced signs of inflammation, and massaged muscle cells were better able to make new mitochondria—promoting faster recovery from exercise-induced muscle damage. Massage stretches and pulls muscles and, as one might expect, the authors found that mechanosensory sensors focal adhesion kinase–1 and its downstream effectors extracellular signaling kinases 1 and 2 were activated, as revealed by their increased phosphorylation. Several hours after massage, another downstream target of this pathway, PGC-1α, shifted into the nucleus, where it in turn activated transcription of its own targets COX7B and ND1. This set of responses indicated that additional mitochondria were forming, presumably accelerating healing of the muscle. Massage also altered the behavior of NFκB, causing less of this key inflammatory mediator to accumulate in the nucleus. Consequently, the NFκB-regulated heat shock proteins and immune cytokines interleukin-6 and tumor necrosis factor–α were less active, a sign of less cellular stress and inflammation. But one oft-repeated idea turned out not to be true. As others have shown, massage did not help clear lactic acid from tired muscles. And glycogen levels were also unchanged. Now that we know something about how massage heals, perhaps we will soon get some idea of why it is so enjoyable. Massage therapy is commonly used during physical rehabilitation of skeletal muscle to ameliorate pain and promote recovery from injury. Although there is evidence that massage may relieve pain in injured muscle, how massage affects cellular function remains unknown. To assess the effects of massage, we administered either massage therapy or no treatment to separate quadriceps of 11 young male participants after exercise-induced muscle damage. Muscle biopsies were acquired from the quadriceps (vastus lateralis) at baseline, immediately after 10 min of massage treatment, and after a 2.5-hour period of recovery. We found that massage activated the mechanotransduction signaling pathways focal adhesion kinase (FAK) and extracellular signal–regulated kinase 1/2 (ERK1/2), potentiated mitochondrial biogenesis signaling [nuclear peroxisome proliferator–activated receptor γ coactivator 1α (PGC-1α)], and mitigated the rise in nuclear factor κB (NFκB) (p65) nuclear accumulation caused by exercise-induced muscle trauma. Moreover, despite having no effect on muscle metabolites (glycogen, lactate), massage attenuated the production of the inflammatory cytokines tumor necrosis factor–α (TNF-α) and interleukin-6 (IL-6) and reduced heat shock protein 27 (HSP27) phosphorylation, thereby mitigating cellular stress resulting from myofiber injury. In summary, when administered to skeletal muscle that has been acutely damaged through exercise, massage therapy appears to be clinically beneficial by reducing inflammation and promoting mitochondrial biogenesis.

Occupational exposure to formaldehyde, hematotoxicity, and leukemia-specific chromosome changes in cultured myeloid progenitor cells.

There are concerns about the health effects of formaldehyde exposure, including carcinogenicity, in light of elevated indoor air levels in new homes and occupational exposures experienced by workers in health care, embalming, manufacturing, and other industries. Epidemiologic studies suggest that formaldehyde exposure is associated with an increased risk of leukemia. However, the biological plausibility of these findings has been questioned because limited information is available on the ability of formaldehyde to disrupt hematopoietic function. Our objective was to determine if formaldehyde exposure disrupts hematopoietic function and produces leukemia-related chromosome changes in exposed humans. We examined the ability of formaldehyde to disrupt hematopoiesis in a study of 94 workers in China (43 exposed to formaldehyde and 51 frequency-matched controls) by measuring complete blood counts and peripheral stem/progenitor cell colony formation. Further, myeloid progenitor cells, the target for leukemogenesis, were cultured from the workers to quantify the level of leukemia-specific chromosome changes, including monosomy 7 and trisomy 8, in metaphase spreads of these cells. Among exposed workers, peripheral blood cell counts were significantly lowered in a manner consistent with toxic effects on the bone marrow and leukemia-specific chromosome changes were significantly elevated in myeloid blood progenitor cells. These findings suggest that formaldehyde exposure can have an adverse effect on the hematopoietic system and that leukemia induction by formaldehyde is biologically plausible, which heightens concerns about its leukemogenic potential from occupational and environmental exposures. Cancer Epidemiol Biomarkers Prev; 19(1); 80–8.