Genetic association analysis under complex survey sampling: the Hispanic Community Health Study/Study of Latinos.

The cohort design allows investigators to explore the genetic basis of a variety of diseases and traits in a single study while avoiding major weaknesses of the case-control design. Most cohort studies employ multistage cluster sampling with unequal probabilities to conveniently select participants with desired characteristics, and participants from different clusters might be genetically related. Analysis that ignores the complex sampling design can yield biased estimation of the genetic association and inflation of the type I error. Herein, we develop weighted estimators that reflect unequal selection probabilities and differential nonresponse rates, and we derive variance estimators that properly account for the sampling design and the potential relatedness of participants in different sampling units. We compare, both analytically and numerically, the performance of the proposed weighted estimators with unweighted estimators that disregard the sampling design. We demonstrate the usefulness of the proposed methods through analysis of MetaboChip data in the Hispanic Community Health Study/Study of Latinos, which is the largest health study of the Hispanic/Latino population in the United States aimed at identifying risk factors for various diseases and determining the role of genes and environment in the occurrence of diseases. We provide guidelines on the use of weighted and unweighted estimators, as well as the relevant software.

Local Dependence in Latent Class Analysis of Rare and Sensitive Events.

For survey methodologists, latent class analysis (LCA) is a powerful tool for assessing the measurement error in survey questions, evaluating survey methods, and estimating the bias in estimates of population prevalence. LCA can be used when gold standard measurements are not available and applied to essentially any set of indicators that meet certain criteria for identifiability. LCA offers quality inference, provided the key threat to model validity-namely, local dependence-can be appropriately addressed either in the study design or in the model-building process. Three potential causes threaten local independence: bivocality, behaviorally correlated error, and latent heterogeneity. In this article, these threats are examined separately to obtain insights regarding (a) questionnaire designs that reduce local dependence, (b) the effects of local dependence on parameter estimation, and (c) modeling strategies to mitigate these effects in statistical inference. The article focuses primarily on the analysis of rare and sensitivity outcomes and proposes a practical approach for diagnosing and mitigating model failures. The proposed approach is empirically tested using real data from a national survey of inmate sexual abuse where measurement errors are a serious concern. Our findings suggest that the proposed modeling strategy was successful in reducing local dependence bias in the estimates, but its success varied by the quality of the indicators available for analysis. With only three indicators, the biasing effects of local dependence can usually be reduced but not always to acceptable levels.

Research priorities for FCTC Articles 20, 21, and 22: surveillance/evaluation and information exchange.

INTRODUCTION: Framework Convention on Tobacco Control (FCTC) Articles 20, 21, and 22 call for strong monitoring and reporting of tobacco use and factors influencing use and disease (Articles 20 and 21) and for collaboration among the Parties and relevant organizations to share resources, knowledge, and expertise on all relevant tobacco control strategies (Article 22). METHODS: This paper provides background information and discusses research strategies that would strengthen these efforts and better inform the parties. By necessity, Articles 20 and 21 are discussed separately from Article 22, although 1 example that relates to both 20/21 and 22 is discussed at the end. RESULTS: Twelve important research opportunities on surveillance and evaluation are recognized, along with 4 on collaboration. The authors believe that the 6 most important areas for research would study (a) possible underreporting of tobacco use among certain demographic groups in some countries, (b) measures of industry activities, (c) optimal sampling strategies, (d) sentinel surveillance, (e) networks of tobacco companies and their partners as they promote tobacco use and interfere with implementation of the FCTC, and (f) network/relationship factors that impact diffusion of knowledge and decision making on the implementation of the FCTC. In addition, we call for a review process of existing surveillance and evaluation strategies to coordinate activities to make optimal use of existing resources. This activity would involve networking as prescribed in Article 22. CONCLUSIONS: Studies and activities such as these would facilitate control of the tobacco epidemic.

Sample design and cohort selection in the Hispanic Community Health Study/Study of Latinos.

PURPOSE: The Hispanic Community Health Study (HCHS)/Study of Latinos (SOL) is a multicenter, community-based cohort study of Hispanic/Latino adults in the United States. A diverse participant sample is required that is both representative of the target population and likely to remain engaged throughout follow-up. The choice of sample design, its rationale, and benefits and challenges of design decisions are described in this study. METHODS: The study design calls for recruitment and follow-up of a cohort of 16,000 Hispanics/Latinos 18-74 years of age, with 62.5% (10,000) over 44 years of age and adequate subgroup sample sizes to support inference by Hispanic/Latino background. Participants are recruited in community areas surrounding four field centers in the Bronx, Chicago, Miami, and San Diego. A two-stage area probability sample of households is selected with stratification and oversampling incorporated at each stage to provide a broadly diverse sample, offer efficiencies in field operations, and ensure that the target age distribution is obtained. CONCLUSIONS: Embedding probability sampling within this traditional, multisite cohort study design enables competing research objectives to be met. However, the use of probability sampling requires developing solutions to some unique challenges in both sample selection and recruitment, as described here.

Prevalence of hip symptoms and radiographic and symptomatic hip osteoarthritis in African Americans and Caucasians: the Johnston County Osteoarthritis Project.

OBJECTIVE: To report contemporary estimates of the prevalence of hip-related osteoarthritis (OA) outcomes in African Americans and Caucasians aged>or=45 years. METHODS: Weighted prevalence estimates and their corresponding 95% confidence intervals for hip symptoms, radiographic hip OA, symptomatic hip OA, and severe radiographic hip OA were calculated using SUDAAN for age, race, and sex subgroups among 3068 participants (33% African Americans, 38% men) in the baseline examination (1991-97) of The Johnston County Osteoarthritis Project, a population-based study of OA in North Carolina. Radiographic hip OA was defined as Kellgren-Lawrence radiographic grade>or=2, moderate/severe radiographic hip OA as grades 3 and 4, and symptomatic hip OA as hip symptoms in a hip with radiographic OA. RESULTS: Hip symptoms were present in 36%; 28% had radiographic hip OA; nearly 10% had symptomatic hip OA; and 2.5% had moderate/severe radiographic hip OA. Prevalence of all 4 outcomes was higher in older individuals; most outcomes were higher for women and African Americans. CONCLUSION: African Americans in this population do not have a lower prevalence of hip-related OA outcomes as previous studies suggested. Increasing public and health system awareness of the relatively high prevalence of these outcomes, which can be disabling, may help to decrease their effects and ultimately prevent them.

The rising prevalence of chronic low back pain.

BACKGROUND: National or state-level estimates on trends in the prevalence of chronic low back pain (LBP) are lacking. The objective of this study was to determine whether the prevalence of chronic LBP and the demographic, health-related, and health care-seeking characteristics of individuals with the condition have changed over the last 14 years. METHODS: A cross-sectional, telephone survey of a representative sample of North Carolina households was conducted in 1992 and repeated in 2006. A total of 4437 households were contacted in 1992 and 5357 households in 2006 to identify noninstitutionalized adults 21 years or older with chronic (>3 months), impairing LBP or neck pain that limits daily activities. These individuals were interviewed in more detail about their health and health care seeking. RESULTS: The prevalence of chronic, impairing LBP rose significantly over the 14-year interval, from 3.9% (95% confidence interval [CI], 3.4%-4.4%) in 1992 to 10.2% (95% CI, 9.3%-11.0%) in 2006. Increases were seen for all adult age strata, in men and women, and in white and black races. Symptom severity and general health were similar for both years. The proportion of individuals who sought care from a health care provider in the past year increased from 73.1% (95% CI, 65.2%-79.8%) to 84.0% (95% CI, 80.8%-86.8%), while the mean number of visits to all health care providers were similar (19.5 [1992] vs 19.4 [2006]). CONCLUSIONS: The prevalence of chronic, impairing LBP has risen significantly in North Carolina, with continuing high levels of disability and health care use. A substantial portion of the rise in LBP care costs over the past 2 decades may be related to this rising prevalence.

Lifetime risk of symptomatic knee osteoarthritis.

OBJECTIVE: To estimate the lifetime risk of symptomatic knee osteoarthritis (OA), overall and stratified by sex, race, education, history of knee injury, and body mass index (BMI). METHODS: The lifetime risk of symptomatic OA in at least 1 knee was estimated from logistic regression models with generalized estimating equations among 3,068 participants of the Johnston County Osteoarthritis Project, a longitudinal study of black and white women and men age >or=45 years living in rural North Carolina. Radiographic, sociodemographic, and symptomatic knee data measured at baseline (1990-1997) and first followup (1999-2003) were analyzed. RESULTS: The lifetime risk of symptomatic knee OA was 44.7% (95% confidence interval [95% CI] 40.0-49.3%). Cohort members with history of a knee injury had a lifetime risk of 56.8% (95% CI 48.4-65.2%). Lifetime risk rose with increasing BMI, with a risk of 2 in 3 among those who were obese. CONCLUSION: Nearly half of the adults in Johnston County will develop symptomatic knee OA by age 85 years, with lifetime risk highest among obese persons. These current high risks in Johnston County may suggest similar risks in the general US population, especially given the increase in 2 major risk factors for knee OA, aging, and obesity. This underscores the immediate need for greater use of clinical and public health interventions, especially those that address weight loss and self-management, to reduce the impact of having knee OA.

Prevalence of low sexual desire and hypoactive sexual desire disorder in a nationally representative sample of US women.

BACKGROUND: We sought to estimate the prevalence of low sexual desire and hypoactive sexual desire disorder (HSDD) in US women, focusing on their menopausal status. METHODS: We performed a cross-sectional study. From a probability sample of households, 2207 US women aged 30 to 70 years and in stable relationships (>or=3 months) were interviewed by telephone. The analysis focused on 755 premenopausal women and 552 naturally and 637 surgically menopausal women. Low sexual desire was defined using the Profile of Female Sexual Function desire domain, and HSDD was defined using the Profile of Female Sexual Function and the Personal Distress Scale. RESULTS: Prevalence of low sexual desire ranged from 26.7% among premenopausal women to 52.4% among naturally menopausal women. The prevalence of HSDD was highest among surgically menopausal women (12.5%). Compared with premenopausal women and adjusting for age, race/ethnicity, educational level, and smoking status, the prevalence ratios for HSDD were 2.3 (95% confidence interval, 1.2-4.5) for surgically menopausal women and 1.2 (0.5-2.8) for naturally menopausal women; the prevalence ratios for low sexual desire were 1.3 (0.9-1.9) and 1.5 (1.0-2.2) for surgically and naturally menopausal women, respectively. CONCLUSIONS: Prevalence of low sexual desire is elevated among surgically and naturally menopausal women vs premenopausal women. Distress about low desire (HSDD) appears to be more than twice as prevalent among surgically menopausal women vs premenopausal women, although the estimate is fairly imprecise.

Disproportionate sampling for population subgroups in telephone surveys.

Population studies often seek to examine phenomena in important population subgroups or to compare results among these and other subgroups. When subgroups of interest comprise a relatively small percentage of the population and acceptable subgroup member lists are not available to serve as sampling frames, it may be prohibitively expensive even by telephone to screen through a sample of the entire population. This paper considers some statistical effects of estimation from a class of two-stratum telephone sample designs where part of the frame with a higher subgroup concentration is disproportionately sampled compared to the rest of the frame. Using proportionate sampling as a reference, the relative impact of this disproportionate design is determined for nominal and effective sample sizes, where the latter are tied to the effect of variation in sample weights that occurs in disproportionately allocated samples. Findings are illustrated using two recent telephone surveys. Whereas nominal subgroup sample sizes may be improved by disproportionate sampling, we conclude that both the survey designer and analyst should use this type of design cautiously in telephone surveys.

Prevalence of knee symptoms and radiographic and symptomatic knee osteoarthritis in African Americans and Caucasians: the Johnston County Osteoarthritis Project.

OBJECTIVE: To report contemporary estimates of the prevalence of knee-related osteoarthritis (OA) outcomes in African Americans and Caucasians aged > or = 45 years. METHODS: Weighted prevalence estimates for knee symptoms, radiographic knee OA, symptomatic knee OA, and severe radiographic knee OA were calculated for age, ethnic, and sex subgroups, in 3018 participants (33% African Americans, 38% men) in the baseline examination (1991-97) of The Johnston County Osteoarthritis Project, a population-based study of OA in North Carolina. Radiographic knee OA was defined as Kellgren-Lawrence radiographic grade > or = 2, severe radiographic knee OA as grades 3 and 4, and symptomatic knee OA as knee symptoms in a knee with radiographic OA. RESULTS: Knee symptoms were present in 43%, 28% had radiographic knee OA, 16% had symptomatic knee OA, and 8% had severe radiographic knee OA. Prevalence was higher in older individuals and women. African Americans had slightly higher prevalence of knee symptoms, radiographic knee OA, and symptomatic knee OA, but significantly higher prevalence of severe radiographic knee OA compared to Caucasians. CONCLUSION: Policy should be directed to increasing education of the public and the medical community about the high prevalence of these conditions, especially in these subgroups, to decrease their impact and ultimately prevent them.

Incidence and risk factors for concussion in high school athletes, North Carolina, 1996-1999.

A prospective cohort study was used to quantify risk factors for sports concussions. Analysis was based on a stratified cluster sample of North Carolina high school athletes followed during 1996-1999. Clustering was by school and sport, and the sample included 15,802 athletes with 1-8 seasons of follow-up per athlete. Concussion rates were estimated for 12 sports, and risk factors were quantified using generalized Poisson regression. Concussion rates ranged from 9.36 (95% confidence interval: 1.93, 16.80) per 100,000 athlete-exposures in cheerleading to 33.09 (95% confidence interval: 24.74, 41.44) per 100,000 athlete-exposures in football, where "athlete-exposure" is one athlete participating in one practice or game. The overall rate of concussion was 17.15 (95% confidence interval: 13.30, 21.00) per 100,000 athlete-exposures. Cheerleading was the only sport for which the practice rate was greater than the game rate. Almost two thirds of cheerleading concussions involved two-level pyramids. Concussion rates were elevated for athletes with a history of concussion, and they increased with the increasing level of body contact permitted in the sport. After adjustment for sport, body mass index, and year in school, history of concussion(s) remained a moderately strong risk factor for concussion (rate ratio = 2.28, 95% confidence interval: 1.24, 4.19). The fact that concussion history is an important predictor of concussion incidence, even in this young population, underscores the importance of primary prevention efforts, timely identification, and careful clinical management of these injuries.

Oral cancer knowledge and examination experiences among North Carolina adults.

OBJECTIVE: This study assesses knowledge of oral cancer risk factors, clinical signs, and oral cancer examination experience among North Carolina adults. METHODS: A statewide random digit dial, computer-assisted telephone interview was conducted in 2002. Data from 1,096 respondents, with a response rate of 62 percent, were poststratified to 2000 census data by sex, race, and age group to produce population-based estimates. Knowledge of one sign of oral cancer, four or more risk factors for oral cancer, and having ever had an oral cancer examination were compared in logistic regression models using normalized weights. RESULTS: Fourteen (95% confidence interval [CI] +/-2) percent of adults had never heard of oral or mouth cancer. Risk factor knowledge was high for 56 percent (95% CI+/-3) and associated in a logistic regression model with younger age, feeling personal factors cause cancer, and nonuse of snuff. One sign of oral cancer (sore/lesion, red or white patch in mouth, and bleeding in the mouth) was correctly identified by 53 percent (95% CI+/-3) with significantly more correct responses from younger people, nonsmokers, and some college education. Only 29 percent (95% CI+/-3) reported ever having had an oral cancer examination when this procedure was described. Most respondents reported exams performed by dentists. In a weighted logistic regression model, older age, being dentate, nonsmokers, alcohol users, and those with some college education were significantly more likely to report having ever had an oral cancer examination. CONCLUSIONS: Although there is moderate knowledge of signs and risk factors for oral cancer among North Carolina adults, knowledge deficits remain. Oral cancer examinations need to be increased, particularly among tobacco smokers.

Development and testing of interview questions to determine last menstrual period in Mexican immigrant populations.

The last menstrual period is used to estimate gestational age. This paper examines sources of measurement error related to the recall of last menstrual period among Mexican immigrant women living in the United States. Qualitative analyses (focus groups and cognitive interviews) suggest that last menstrual period recall does not seem to be a large source of measurement error in the calculation of gestational age and the impact of this type of error on the misclassification of preterm births appears to be minimal. Questions for querying about last menstrual period in this population are offered.

Sampling minority groups in health surveys.

Sampling racial and ethnic minority groups in population-based health studies presents the statistical designer with a variety of notable challenges, many of which can be traced to the manner in which the sampling frame maps into the set of group members. A brief review of sampling approaches aimed at race/ethnic groups is followed by findings on two particularly important statistical issues one may face when sampling minorities: the implications of the geographic diffusion of minorities within sampled area clusters, and the relative utility of different forms of space and time sampling to deal with sometimes dramatic change over time in the linkage between entries on a sampling frame and members of relatively mobile minority groups. Simulated findings from practical illustrations involving the sampling of Hispanics are presented to address each issue.