American Journal of Epidemiology

L.J. REED, H. MUENCH; A SIMPLE METHOD OF ESTIMATING FIFTY PER CENT ENDPOINTS12, American Journal of Epidemiology, Volume 27, Issue 3, 1 May 1938, Pages 493

Relative risk is usually the parameter of interest in epidemiologic and medical studies. In this paper, the author proposes a modified Poisson regression approach (i.e., Poisson regression with a robust error variance) to estimate this effect measure directly. A simple 2-by-2 table is used to justify the validity of this approach. Results from a limited simulation study indicate that this approach is very reliable even with total sample sizes as small as 100. The method is illustrated with two data sets.

With advances in the effectiveness of treatment and disease management, the contribution of chronic comorbid diseases (comorbidities) found within the Charlson comorbidity index to mortality is likely to have changed since development of the index in 1984. The authors reevaluated the Charlson index and reassigned weights to each condition by identifying and following patients to observe mortality within 1 year after hospital discharge. They applied the updated index and weights to hospital discharge data from 6 countries and tested for their ability to predict in-hospital mortality. Compared with the original Charlson weights, weights generated from the Calgary, Alberta, Canada, data (2004) were 0 for 5 comorbidities, decreased for 3 comorbidities, increased for 4 comorbidities, and did not change for 5 comorbidities. The C statistics for discriminating in-hospital mortality between the new score generated from the 12 comorbidities and the Charlson score were 0.825 (new) and 0.808 (old), respectively, in Australian data (2008), 0.828 and 0.825 in Canadian data (2008), 0.878 and 0.882 in French data (2004), 0.727 and 0.723 in Japanese data (2008), 0.831 and 0.836 in New Zealand data (2008), and 0.869 and 0.876 in Swiss data (2008). The updated index of 12 comorbidities showed good-to-excellent discrimination in predicting in-hospital mortality in data from 6 countries and may be more appropriate for use with more recent administrative data.

The relationship between social and community ties and mortality was assessed using the 1965 Human Population Laboratory survey of a random sample of 6928 adults in Alameda County, California and a subsequent nine-year mortality follow-up. The findings show that people who lacked social and community ties were more likely to die in the follow-up period than those with more extensive contacts. The age-adjusted relative risks for those most isolated when compared to those with the most social contacts were 2.3 for men and 2.8 for women. The association between social ties and mortality was found to be independent of self-reported physical health status at the time of the 1965 survey, year of death, socioeconomic status, and health practices such as smoking, alcoholic beverage consumption, obesity, physical activity, and utilization of preventive health services as well as a cumulative index of health practices.

Sleep-disordered breathing is a common disorder with a range of harmful sequelae. Obesity is a strong causal factor for sleep-disordered breathing, and because of the ongoing obesity epidemic, previous estimates of sleep-disordered breathing prevalence require updating. We estimated the prevalence of sleep-disordered breathing in the United States for the periods of 1988-1994 and 2007-2010 using data from the Wisconsin Sleep Cohort Study, an ongoing community-based study that was established in 1988 with participants randomly selected from an employed population of Wisconsin adults. A total of 1,520 participants who were 30-70 years of age had baseline polysomnography studies to assess the presence of sleep-disordered breathing. Participants were invited for repeat studies at 4-year intervals. The prevalence of sleep-disordered breathing was modeled as a function of age, sex, and body mass index, and estimates were extrapolated to US body mass index distributions estimated using data from the National Health and Nutrition Examination Survey. The current prevalence estimates of moderate to severe sleep-disordered breathing (apnea-hypopnea index, measured as events/hour, ≥15) are 10% (95% confidence interval (CI): 7, 12) among 30-49-year-old men; 17% (95% CI: 15, 21) among 50-70-year-old men; 3% (95% CI: 2, 4) among 30-49-year-old women; and 9% (95% CI: 7, 11) among 50-70 year-old women. These estimated prevalence rates represent substantial increases over the last 2 decades (relative increases of between 14% and 55% depending on the subgroup).

The rule of thumb that logistic and Cox models should be used with a minimum of 10 outcome events per predictor variable (EPV), based on two simulation studies, may be too conservative. The authors conducted a large simulation study of other influences on confidence interval coverage, type I error, relative bias, and other model performance measures. They found a range of circumstances in which coverage and bias were within acceptable levels despite less than 10 EPV, as well as other factors that were as influential as or more influential than EPV. They conclude that this rule can be relaxed, in particular for sensitivity analyses undertaken to demonstrate adequate control of confounding.

The aim of this study was to evaluate the reproducibility and validity of a 61-item semiquantitative food frequency questionnaire used in a large prospective study among women. This form was administered twice to 173 participants at an interval of approximately one year (1980-1981), and four one-week diet records for each subject were collected during that period. Intraclass correlation coefficients for nutrient intakes estimated by the one-week diet records (range = 0.41 for total vitamin A without supplements to 0.79 for vitamin B6 with supplements) were similar to those computed from the questionnaire (range = 0.49 for total vitamin A without supplements to 0.71 for sucrose), indicating that these methods were generally comparable with respect to reproducibility. With the exception of sucrose and total carbohydrate, nutrient intakes from the diet records tended to correlate more strongly with those computed from the questionnaire after adjustment for total caloric intake. Correlation coefficients between the mean calorie-adjusted intakes from the four one-week diet records and those from the questionnaire completed after the diet records ranged from 0.36 for vitamin A without supplements to 0.75 for vitamin C with supplements. Overall, 48% of subjects in the lowest quintile of calorie-adjusted intake computed from the diet records were also in the lowest questionnaire quintile, and 74% were in the lowest one of two questionnaire quintiles. Similarly, 49% of those in the highest diet record quintile were also in the highest questionnaire quintile, and 77% were in the highest one or two questionnaire quintiles. These data indicate that a simple self-administered dietary questionnaire can provide useful information about individual nutrient intakes over a one-year period.

Muscle mass decreases with age, leading to "sarcopenia," or low relative muscle mass, in elderly people. Sarcopenia is believed to be associated with metabolic, physiologic, and functional impairments and disability. Methods of estimating the prevalence of sarcopenia and its associated risks in elderly populations are lacking. Data from a population-based survey of 883 elderly Hispanic and non-Hispanic white men and women living in New Mexico (the New Mexico Elder Health Survey, 1993-1995) were analyzed to develop a method for estimating the prevalence of sarcopenia. An anthropometric equation for predicting appendicular skeletal muscle mass was developed from a random subsample (n = 199) of participants and was extended to the total sample. Sarcopenia was defined as appendicular skeletal muscle mass (kg)/height2 (m2) being less than two standard deviations below the mean of a young reference group. Prevalences increased from 13-24% in persons under 70 years of age to >50% in persons over 80 years of age, and were slightly greater in Hispanics than in non-Hispanic whites. Sarcopenia was significantly associated with self-reported physical disability in both men and women, independent of ethnicity, age, morbidity, obesity, income, and health behaviors. This study provides some of the first estimates of the extent of the public health problem posed by sarcopenia.

The UK Biobank cohort is a population-based cohort of 500,000 participants recruited in the United Kingdom (UK) between 2006 and 2010. Approximately 9.2 million individuals aged 40-69 years who lived within 25 miles (40 km) of one of 22 assessment centers in England, Wales, and Scotland were invited to enter the cohort, and 5.5% participated in the baseline assessment. The representativeness of the UK Biobank cohort was investigated by comparing demographic characteristics between nonresponders and responders. Sociodemographic, physical, lifestyle, and health-related characteristics of the cohort were compared with nationally representative data sources. UK Biobank participants were more likely to be older, to be female, and to live in less socioeconomically deprived areas than nonparticipants. Compared with the general population, participants were less likely to be obese, to smoke, and to drink alcohol on a daily basis and had fewer self-reported health conditions. At age 70-74 years, rates of all-cause mortality and total cancer incidence were 46.2% and 11.8% lower, respectively, in men and 55.5% and 18.1% lower, respectively, in women than in the general population of the same age. UK Biobank is not representative of the sampling population; there is evidence of a "healthy volunteer" selection bias. Nonetheless, valid assessment of exposure-disease relationships may be widely generalizable and does not require participants to be representative of the population at large.

The Multi-Ethnic Study of Atherosclerosis was initiated in July 2000 to investigate the prevalence, correlates, and progression of subclinical cardiovascular disease (CVD) in a population-based sample of 6,500 men and women aged 45-84 years. The cohort will be selected from six US field centers. Approximately 38% of the cohort will be White, 28% African-American, 23% Hispanic, and 11% Asian (of Chinese descent). Baseline measurements will include measurement of coronary calcium using computed tomography; measurement of ventricular mass and function using cardiac magnetic resonance imaging; measurement of flow-mediated brachial artery endothelial vasodilation, carotid intimal-medial wall thickness, and distensibility of the carotid arteries using ultrasonography; measurement of peripheral vascular disease using ankle and brachial blood pressures; electrocardiography; and assessments of microalbuminuria, standard CVD risk factors, sociodemographic factors, life habits, and psychosocial factors. Blood samples will be assayed for putative biochemical risk factors and stored for use in nested case-control studies. DNA will be extracted and lymphocytes will be immortalized for genetic studies. Measurement of selected subclinical disease indicators and risk factors will be repeated for the study of progression over 7 years. Participants will be followed through 2008 for identification and characterization of CVD events, including acute myocardial infarction and other coronary heart disease, stroke, peripheral vascular disease, and congestive heart failure; therapeutic interventions for CVD; and mortality.

Abstract Although social isolation is known to impact those with cardiovascular disease, its relationship with incident hypertension is unclear. In this retrospective cohort study, we used data from the All of Us research program. Participants were enrolled between May 2017-April 2023, aged 50+ years, and had at least 60-days of EHR data pre-index date. We identified incident hypertension during follow-up. Those classified as hypertensive at or prior to the index date were excluded. We defined low social support and loneliness as binary variables. Using causal forests, we estimated treatment effects as unweighted and overlap-weighted risk differences (RD). A total of 24,720 participants with a median follow-up of 2.6 years were included. While loneliness and social support were not significantly associated with incident hypertension in the overall population, causal forests revealed meaningful effect heterogeneity, with socially isolated male (RD: 0.038, 95% CI 0.002 to 0.075; overlap-weighted RD: 0.027, 95% CI -0.005 to 0.059) and Gay or Lesbian participants (RD: 0.108, 95% CI 0.011 to 0.205; overlap-weighted RD: 0.11, 95% CI 0.008 to 0.211) experiencing a significantly elevated risk. This illustrates how machine learning, when embedded within epidemiologic frameworks, can detect subgroup vulnerability that other approaches may obscure, enabling targeting of limited resources.

Understanding treatment effect heterogeneity (TEH) is essential to improve care for patients living with chronic diseases like type 2 diabetes. Individual responses to treatment can vary substantially, and identifying the factors that drive TEH may enable increased personalization of treatment and follow-up. In this study, we conduct an exploratory analysis of TEH for glucagon-like peptide-1 receptor agonists used to treat type 2 diabetes. Endpoints are defined using hemoglobin A1c measurements, body mass index, and blood pressure. We use causal forests to conduct global tests for TEH, estimate variable importances to rank candidate effect modifiers, and estimate the magnitude of heterogeneity. Data from three clinical trials are analyzed: SUSTAIN 6 (subcutaneous semaglutide versus placebo, n=3297), PIONEER 6 (oral semaglutide versus placebo, n=3183), and LEADER (subcutaneous liraglutide versus placebo, n=9340), pooled to 13,670 patients after exclusions. Baseline hemoglobin A1c ranks highest among candidate effect modifiers across methods, and kidney function ranks next in the relative risk causal forest; we discuss their clinical implications. The magnitude estimates, however, show that the variation these covariates jointly explain is small. As such, large covariate-driven heterogeneity can be ruled out in these trials, and the average treatment effect provides a reasonable summary for most patients.

One aspect of the Hispanic Health Paradox is the longstanding observation that Latin American immigrants in the United States often experience equal or better health outcomes than U.S.-born counterparts despite greater socioeconomic disadvantage. This advantage extends to their infants and is most pronounced among mothers with the least education. Whether such patterns reflect selective migration remains unclear. This study evaluates the role of educational selectivity in explaining nativity disparities in infant health. Using U.S. natality data among immigrants from Mexico, Cuba, El Salvador, Guatemala, the Dominican Republic, Colombia, Honduras, and Venezuela (2012-2023), linked with internationally harmonized educational data, we assess preterm birth disparities using two measures of maternal education: absolute and relative attainment. Immigrants at the bottom of the U.S. education distribution often rank much higher in their origin countries, consistent with positive selection. For most births to Latin American mothers (except Colombian and Venezuelan), measuring education relative to origin-country distributions reverses immigrant advantages. At equivalent relative education ranks, immigrant mothers face comparable or higher risks of preterm birth than U.S.-born Hispanic mothers. These findings offer a new perspective on the Hispanic Health Paradox, highlighting the importance of relative education in understanding the association between maternal education and infant health.

Yating Xie, Ting Ye; Live-birth selection and the interpretation of hyperemesis gravidarum and adverse pregnancy outcomes, American Journal of Epidemiology

Abstract Poor sleep may negatively impact reproductive outcomes, yet its role in uterine leiomyomata (UL) development has not been studied. We estimated prospectively the association of self-reported sleep health with ultrasound-detected UL incidence and growth in the Study of Environment, Lifestyle, and Fibroids, a prospective cohort of Black individuals aged 23-35 years in Detroit, Michigan. At each of five clinic visits (2010-2020), participants reported sleep duration, frequency of trouble sleeping, and sleep quality, and underwent transvaginal ultrasound to detect incident UL. We used Cox proportional hazards regression models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) and applied inverse probability weights to adjust for time-dependent confounding. Among 1,137 participants, 351 developed UL (10-year cumulative incidence: 30.9%). Short (<6 hours/day, HR=0.30 95% CI: 0.17, 0.55) sleep duration and consistently poorer sleep quality (<4 days/week feeling well-rested, HR=0.78, 95% CI: 0.56, 1.08) were associated with lower UL incidence. Increased frequency of trouble sleeping was positively associated with UL incidence (≥15 days/month, HR=1.96, 95% CI: 0.76, 5.09). Sleep health was not associated with UL growth. In this prospective ultrasound study, sleep measures were inconsistently associated with UL incidence and not associated with UL growth.

Abstract Among couples with non-smoking wives, the relationship between husband smoking and fecundability remains uncertain. Evidence is limited regarding whether smoking cessation before conception is associated with improved fecundability. We included a total of 4,569,530 reproductive-aged couples with non-smoking wives from the National Free Pre-conception Check-up Project in China from 2015 to 2017. Cox proportional hazards models were used to estimate fecundability hazard ratios (HRs) and 95% confidence intervals (CIs). Among 4,569,530 eligible couples, 1,228,222 (26.88%) husbands reported smoking at baseline. Within 1 year, 3,046,804 couples (66.68%) conceived, with a median TTP of 2.76 cycles. Husband smoking was associated with lower fecundability (HR=0.97, 95% CI, 0.97-0.97), demonstrating a dose-response relationship. Among 781,481 couples in which the husband smoked and conception occurred within 1 year, 507,530 (64.94%) reported smoking cessation. Smoking cessation was associated with higher fecundability (HR=1.23, 95% CI, 1.23-1.24), with consistent benefits across prior smoking levels, although the magnitude of improvement was smaller among husbands who smoked more heavily. Associations varied by age and were more pronounced in older couples. These findings support the inclusion of husband smoking cessation in preconception care and highlight its potential role in promoting smoke-free households and broader tobacco control efforts.

Abstract Service members experience unique occupational hazards that may increase their risk for melanoma. This longitudinal study aimed to determine associations between deployment status, environmental exposures, and solar exposures and incident melanoma among active duty service members. Millennium Cohort Study survey data (2001-2021) were linked with deployment and military health care records (2000-2021). Adjusted Cox proportional hazards models estimated effects of exposures and covariates on incident melanoma. Risk of melanoma diagnosis by age (early vs late onset) and anatomical site (sun-protected vs sun-exposed skin) were also estimated. Among 115,512 service members, 184 were diagnosed with melanoma during the study period (61% were <40 years old and 77% diagnosed on sun-protected skin). Melanoma risk by demographics was consistent with civilian studies. Among this military population, those who deployed (adjusted hazard ratio [AHR] 0.55 [95% CI 0.40, 0.75]) had reduced melanoma risk. Conversely, those who self-reported being exposed to occupational hazards (AHR 1.43 [95% CI 1.05, 1.95]) had elevated melanoma risk. Findings were consistent with the “healthy deployer effect,” while self-reported occupational hazards were associated with an increased risk of melanoma. Results may be informative for understanding melanoma etiology, risk counseling, and screening.