Frances J. Mather

Statistical methods for linking health, exposure, and hazards

The Environmental Public Health Tracking Network (EPHTN) proposes to link environmental hazards and exposures to health outcomes. Statistical methods used in case–control and cohort studies to link health outcomes to individual exposure estimates are well developed. However, reliable exposure estimates for many contaminants are not available at the individual level. In these cases, exposure/hazard data are often aggregated over a geographic area, and ecologic models are used to relate health outcome and exposure/hazard. Ecologic models are not without limitations in interpretation. EPHTN data are characteristic of much information currently being collected—they are multivariate, with many predictors and response variables, often aggregated over geographic regions (small and large) and correlated in space and/or time. The methods to model trends in space and time, handle correlation structures in the data, estimate effects, test hypotheses, and predict future outcomes are relatively new and without extensive application in environmental public health. In this article we outline a tiered approach to data analysis for EPHTN and review the use of standard methods for relating exposure/hazards, disease mapping and clustering techniques, Bayesian approaches, Markov chain Monte Carlo methods for estimation of posterior parameters, and geostatistical methods. The advantages and limitations of these methods are discussed.

Acral lentiginous melanoma

Between 1958 and 1990, 82 patients with acral lentiginous melanoma were treated by the Tulane Surgical Service with regional perfusion, excision of lesion, and lymph node dissection. The patient group comprised 27 white men, 29 white women, 18 black men, and 8 black women, with an average age of 61 years. More foot lesions than hand lesions were reported, and all the lack men had foot lesions. In stage I patients, overall 5-year survival rates were 65% at 5 years and 44% at 10 years, with differences by race and gender. The black men did poorest, with a 13% 10-year survival rate. Survival rates were worse with increasing disease stage when calculated using univariate analysis. The 5-year survival rate of all patients with stage III and stage IV disease was 26%. A multivariate analysis was performed in 78 of 82 patients in whom all variables of Clarks level, age, race, stage, and sex were known. A strong relationship was observed between decreasing survival time and increasing Clarks level, with stage of marginal significance. In a multivariate analysis of patients with stage I disease, an increasing level of invasion was found to be significant, with a trend for a relationship to thickness. A trend toward decreased survival time was observed in men and blacks.

Long-term survival and prognostic factors in breast cancer patients with localized (no skin, muscle, or chest wall attachment) disease with and without positive lymph nodes†

Uncertainty exists regarding the magnitude of excess mortality from localized breast cancer at long follow‐up times (>15 years) since diagnosis and regarding the effects of race and age as prognostic factors at all follow‐up times. Long‐term survival was determined in 1141 patients (311 white, 830 black) diagnosed as having localized breast cancer with and without positive axillary lymph nodes, but without any signs of complete or incomplete skin, muscle, or chest wall attachment. Survival curves were estimated by means of actuarial methods; prognostic factors were evaluated with the Coxs regression analysis. Survival from all causes was 62%, 43%, 33%, 25%, and 18% at 5, 10, 15, 20, and 30 years, respectively. Breast cancer‐specific survival was 76%, 65%, 63%, 61%, and 59% at 5, 10, 15, 20, and 30 years, respectively. Breast cancer‐specific hazard rates exceeded those expected in the general population by 119 times, 53 times, 12 times, and 6 times at 0 to 5, 5 to 10, 10 to 20, and 20 to 25 years, respectively. Of the 395 patients enrolled after 1968 who had modified radical or radical surgery, 338 had known number of positive nodes and size of tumor. Breast cancer‐specific survival was significantly increased with: (1) a decreasing number of positive lymph nodes, 0, 1 to 3, and 4 or more (P = 0.000); (2) later year of diagnosis (1974 or before versus 1975 or later) (P = 0.000); and (3), possibly, tumor size of 7.0 cm or less (P = 0.09). When these variables were controlled, no significant association of age at diagnosis or race with breast cancer‐specific survival was found. These data suggest that the number of nodes, year of diagnosis and, possibly, tumor size are important prognostic factors for survival, but race and age are not. Also, excess mortality may exist at late intervals; however, it is small in relation to other causes.

Perioperative transfusion, postoperative infection, and recurrence of head and neck cancer

Background Immunologic effects of perioperative transfusion and postoperative infection have been purported to influence cancer recurrence rates. Methods Records of all head and neck cancer patients having surgical extirpation of the primary tumor and/or regional nodes at our institution over a 5-year period were reviewed. Time to recurrence was the outcome measure. All variables were evaluated via univariate analysis using log rank tests, with Cox proportional hazards used for multivariate analyses Results Univariate analysis identified the following as potential prognostic factors associated with recurrence: nodal stage, total lymphocyte count, overall stage, amount transfused, occurrence of a transfusion, and the American Society of Anesthesiologists status. Various backward stepwise multivariate regression models showed that neither transfusion nor postoperative infection independently influenced recurrence. However, transfusion of 3 or more units did surface as an independent contributor to recurrence, and in certain subgroups there was a trend toward improved survival for those who had a postoperative infection. Conclusions In this series, neither perioperative transfusion nor postoperative infection independently influenced recurrence.

Luminal gastric somatostatin-like immunoreactivity in response to various stimuli in man

This study investigates release of somatostatin-like immunoreactivity (SLI) into the gastric lumen of five healthy human subjects in response to pharmacological stimuli (pentagastrin and secretin) and physiological stimuli (sham feeding and intrajejunal perfusion of elemental diet). Basal and poststimulation gastric juice aspirates were collected at 15-min intervals, extracted with acetone, and SLI determined by radioimmunoassay, with these results: (1) A considerable amount of SLI was secreted during the basal period. (2) Pentagastrin stimulated SLI release quickly and was associated with increased acid secretion. (3) Both secretin and sham feeding increased SLI only slightly. (4) During intrajejunal perfusion of the elemental diet, SLI increased significantly, was associated with decreased acid secretion, and rapidly returned to basal level when elemental diet was replaced by saline. Basal levels of gastric luminal SLI thus showed distinct changes in response to each stimulus. Although the physiological action of luminal SLI remains to be studied, its levels may reflect gastric D-cell activities.

Improving malaria control in West Africa: interruption of transmission as a paradigm shift.

With the paradigm shift from the reduction of morbidity and mortality to the interruption of transmission, the focus of malaria control broadens from symptomatic infections in children ≤5 years of age to include asymptomatic infections in older children and adults. In addition, as control efforts intensify and the number of interventions increases, there will be decreases in prevalence, incidence and transmission with additional decreases in morbidity and mortality. Expected secondary consequences of these changes include upward shifts in the peak ages for infection (parasitemia) and disease, increases in the ages for acquisition of antiparasite humoral and cellular immune responses and increases in false-negative blood smears and rapid diagnostic tests. Strategies to monitor these changes must include: (1) studies of the entire population (that are not restricted to children ≤5 or ≤10 years of age), (2) study sites in both cities and rural areas (because of increasing urbanization across sub-Saharan Africa) and (3) innovative strategies for surveillance as the prevalence of infection decreases and the frequency of false-negative smears and rapid diagnostic tests increases.

Does Cigarette Smoking Paradoxically Increase Survival in Idiopathic Dilated Cardiomyopathy

Recent studies have suggested that patients with idiopathic dilated cardiomyopathy (IDCM) who smoke have an improved prognosis as compared with nonsmokers. We examined this paradoxical finding using data from a population-based study in Washington, D.C. (n = 127). Current smokers were more likely to have a left-ventricular ejection fraction (LVEF) of 25% or greater as compared with IDCM patients who were past smokers or lifelong nonsmokers (p < or = 0.02). The cumulative survival among current smokers at 12 and 24 months was 88.1 and 81.4%, respectively, as compared with 77.9 and 71.6% among past smokers and 74.0 and 64.3% among patients who had never smoked. In a univariate analysis using the proportional hazards model, lifelong nonsmokers and former smokers were about twice as likely to die as compared with smokers, although the association was not significant (p > 0.10). In multivariable analysis, older age, LVEF, and ventricular arrhythmias - but not cigarette smoking-were found to be statistically significant independent predictors of survival (p < or = 0.05).

AQ-13, an investigational antimalarial, versus artemether plus lumefantrine for the treatment of uncomplicated Plasmodium falciparum malaria: a randomised, phase 2, non-inferiority clinical trial

Summary Background Chloroquine was used for malaria treatment until resistant Plasmodium falciparum was identified. Because 4-aminoquinolines with modified side chains, such as AQ-13, are active against resistant parasites, we compared AQ-13 against artemether plus lumefantrine for treatment of uncomplicated P falciparum malaria. Methods We did a randomised, non-inferiority trial. We screened men (≥18 years) with uncomplicated malaria in Missira (northeast Mali) and Bamako (capital of Mali) for eligibility (≥2000 asexual P falciparum parasites per μL of blood). Eligible participants were randomly assigned to either the artemether plus lumefantrine group or AQ-13 group by permuting blocks of four with a random number generator. Physicians and others caring for the participants were masked, except for participants who received treatment and the research pharmacist who implemented the randomisation and provided treatment. Participants received either 80 mg of oral artemether and 480 mg of oral lumefantrine twice daily for 3 days or 638·50 mg of AQ-13 base (two oral capsules) on days 1 and 2, and 319·25 mg base (one oral capsule) on day 3. Participants were monitored for parasite clearance (50 μL blood samples twice daily at 12 h intervals until two consecutive negative samples were obtained) and interviewed for adverse events (once every day) as inpatients during week 1. During the 5-week outpatient follow-up, participants were examined for adverse events and recurrent infection twice per week. All participants were included in the intention-to-treat analysis and per-protocol analysis, except for those who dropped out in the per-protocol analysis. The composite primary outcome was clearance of asexual parasites and fever by day 7, and absence of recrudescent infection by parasites with the same molecular markers from days 8 to 42 (defined as cure). Non-inferiority was considered established if the proportion of patients who were cured was higher for artemether plus lumefantrine than for AQ-13 and the upper limit of the 95% CI was less than the non-inferiority margin of 15%. This trial is registered at ClinicalTrials.gov, number NCT01614964. Findings Between Aug 6 and Nov 18, 2013, and between Sept 18 and Nov 20, 2015, 66 Malian men with uncomplicated malaria were enrolled. 33 participants were randomly assigned to each group. There were no serious adverse events (grade 2–4) and asexual parasites were cleared by day 7 in both groups. 453 less-severe adverse events (≤grade 1) were reported: 214 in the combination group and 239 in the AQ-13 group. Two participants withdrew from the AQ-13 group after parasite clearance and three were lost to follow-up. In the artemether plus lumefantrine group, two participants had late treatment failures (same markers as original isolates). On the basis of the per-protocol analysis, the AQ-13 and artemether plus lumefantrine groups had similar proportions cured (28 [100%] of 28 vs 31 [93·9%] of 33; p=0·50) and AQ-13 was not inferior to artemether plus lumefantrine (difference −6·1%, 95% CI −14·7 to 2·4). Proportions cured were also similar between the groups in the intention-to-treat analysis (28 of 33, 84·8% for AQ-13 vs 31 of 33, 93·9% for artemether and lumefantrine; p=0·43) but the upper bound of the 95% CI exceeded the 15% non-inferiority margin (difference 9·1%, 95% CI −5·6 to 23·8). Interpretation The per-protocol analysis suggested non-inferiority of AQ-13 to artemether plus lumefantrine. By contrast, the intention-to-treat analysis, which included two participants who withdrew and three who were lost to follow-up from the AQ-13 group, did not meet the criterion for non-inferiority of AQ-13, although there were no AQ-13 treatment failures. Studies with more participants (and non-immune participants) are needed to decide whether widespread use of modified 4-aminoquinolones should be recommended. Funding US Food and Drug Administration Orphan Product Development, National Institutes of Health, US Centers for Disease Control and Prevention, Burroughs-Wellcome Fund, US State Department, and WHO.

Perceived Discrimination and Heavy Episodic Drinking Among African-American Youth: Differences by Age and Reason for Discrimination.

PURPOSE The purpose of this study was to examine whether associations between perceived discrimination and heavy episodic drinking (HED) vary by age and by discrimination type (e.g., racial, age, physical appearance) among African-American youth. METHODS National data from the Panel Study of Income Dynamics Transition to Adulthood Study were analyzed. Youth participated in up to four interviews (2005, 2007, 2009, 2011; n = 657) between ages 18 and 25 years. Respondents reported past-year engagement in HED (four or more drinks for females, five or more drinks for males) and frequency of discriminatory acts experienced (e.g., receiving poor service, being treated with less courtesy). Categorical latent growth curve models, including perceived discrimination types (racial, age, and physical appearance) as a time-varying predictors of HED, were run. Controls for gender, birth cohort, living arrangement in adolescence, familial wealth, parental alcohol use, and college attendance were explored. RESULTS The average HED trajectory was curvilinear (increasing followed by flattening), whereas perceived discrimination remained flat with age. In models including controls, odds of HED were significantly higher than average around ages 20-21 years with greater frequency of perceived racial discrimination; associations were not significant at other ages. Discrimination attributed to age or physical appearance was not associated with HED at any age. CONCLUSIONS Perceived racial discrimination may be a particularly salient risk factor for HED around the ages of transition to legal access to alcohol among African-American youth. Interventions to reduce discrimination or its impact could be targeted before this transition to ameliorate the negative outcomes associated with HED.

A statistical model for the evaluation of antigen-stimulated blastogenic transformation as measured by scintillation spectrometry

Abstract A quantitative statistical model has been presented to aid the analysis of data from lymphocyte stimulation studies which have been evaluated by scintillation spectrometry. This model is particularly useful when stimulation indices of less than three are encountered. Data are presented which test the model.