Creating a nationally representative sample of patients from trauma centers.

BACKGROUND: The National Trauma Data Bank (NTDB) was developed as a convenience sample of registry data from contributing trauma centers (TCs), thus, inferences about trauma patients may not be valid at the national level. The NTDB National Sample was created to obtain nationally representative estimates of trauma patients treated in the US level I and II TCs. METHODS: Level I and II TCs in the Trauma Information Exchange Program were identified and a random stratified sample of 100 TCs was selected. The probability-proportional-to-size method was used to select TCs and sample weights were calculated. National Sample Program estimates from 2003 to 2006 were compared with raw NTDB data, and to a subset of TCs in the Healthcare Cost and Utilization Project Nationwide Inpatient Sample, a population-based dataset drawn from community hospitals. RESULTS: Weighted estimates from the NTDB National Sample range from 484,000 (2004) to 608,000 (2006) trauma incidents. Crude NTDB data over-represented the proportion of younger patients (0 years-14 years) compared with the NTDB National Sample, which does not include children's hospitals. Few TCs in Trauma Information Exchange Program are included in Healthcare Cost and Utilization Project Nationwide Inpatient Sample, but estimates based on this subset indicate a higher percentage of older patients (age 65 year or older, 23.98% versus 17.85%), lower percentage male patients, and a lower percentage of motor vehicle accidents compared with NTDB National Sample. CONCLUSION: Although nationally representative data regarding trauma patients are available in other population-based samples, they do not represent TCs patients and lack the specificity of National Sample Program data, which contains detailed information on injury mechanisms, diagnoses, and hospital treatment.

Regression modeling of combined data from multiple sample surveys.

Combined data from multiple sample surveys are often used in population-based epidemiologic studies. Combining data can be beneficial in that sampling errors are reduced and coverage biases are corrected. Also, it is often necessary in order to use information lacking in one survey but available in another. We propose an estimation equations method for generalized linear models from the combined data. The estimation procedures for logistic regression models and Poisson regression models are developed. An example of estimating the relative risk of death by smoking status is used as an illustration and a simulation study is performed to examine the performance of the method.

Exposure to beryllium and occurrence of lung cancer: findings from a cox proportional hazards analysis of data from a retrospective cohort mortality study.

OBJECTIVE: The objective was to assess highly confounded patterns in a standardized mortality ratio (SMR) analysis of lung cancer in beryllium worker cohorts. METHODS: We used Cox proportional hazards single- and multi-variate models to assess confounding and the SMR patterns. RESULTS: We confirmed the lack of association of lung cancer with time worked. We could not confirm the original study's finding of lung cancer highly associated with earlier plants and or with workers hired in the 1940s compared to the 1950s. The pattern of higher rates of lung cancer with increasing latency was attenuated when covariates were added to the model. We could not exclude that the lower SMR and hazard ratios for workers hired in the 1960s might be related to assumed lower beryllium exposures. CONCLUSION: The patterns observed provide little support for an association of lung cancer with beryllium work factors. This result is likely due to the absence in the original study of a significant overall excess of lung cancer after smoking adjustment.

Dietary supplement use among infants, children, and adolescents in the United States, 1999-2002.

OBJECTIVE: To describe dietary supplement use among US children. DESIGN: Analysis of nationally representative data from the 1999-2002 National Health and Nutrition Examination Survey (NHANES). SETTING: Home interviews and a mobile examination center. PARTICIPANTS: Children from birth through 18 years who participated in NHANES (N=10,136). MAIN EXPOSURE: Frequency of use of any dietary supplement product. OUTCOME MEASURE: Prevalence of use and intake of key nutrients from supplements among children. RESULTS: In 1999-2002, 31.8% of children used dietary supplements, with the lowest use reported among infants younger than 1 year (11.9%) and teenagers 14 to 18 years old (25.7%) and highest use among 4- to 8-year-old children (48.5%). Use was highest among non-Hispanic white (38.1%) and Mexican American (22.4%) participants, lowest among non-Hispanic black participants (18.8%), and was not found to differ by sex. The type of supplement most commonly used was multivitamins and multiminerals (18.3%). Ascorbic acid (28.6%), retinol (25.8%), vitamin D (25.6%), calcium (21.1%), and iron (19.3%) were the primary supplemental nutrients consumed. Supplement use was associated with families with higher incomes; a smoke-free environment; not being certified by the US Department of Agriculture Special Supplemental Nutrition Program for Women, Infants and Children in the last 12 months; lower child body mass index; and less daily recreational screen time (television, video games, computers, etc) (P<.005). The highest prevalence of supplement use (P<.005) was in children who were underweight or at risk for underweight (P<.005). CONCLUSIONS: More than 30% of children in the United States take dietary supplements regularly, most often multivitamins and multiminerals. Given such extensive use, nutrient intakes from dietary supplements must be included to obtain accurate estimates of overall nutrient intake in children.

Hepatitis A 2004 vaccination in children: methods and findings of a survey in two states.

BACKGROUND: Hepatitis A vaccine coverage estimates needed for surveillance and vaccine policy decisions are not readily available for children older than 35 months or for adolescents. This article reports methodology developed for obtaining such estimates by telephone survey with and without provider record verification. METHODS: A random-digit-dial telephone survey with provider verification was conducted in Arizona and Oregon in 2004-2005 to obtain coverage estimates for children aged 2.5 to 15 years based on parental reports from telephone survey data alone, and from multiple logistic regressions using both telephone survey and provider data. Analysis was performed during 2006. RESULTS: Vaccination information was collected from parents of 1266 children, and provider verification from 488. Telephone survey and provider record-based hepatitis A vaccine coverage (one or more doses) was 60% and 65%, respectively, in Arizona, and 39% and 26%, respectively, in Oregon. Children who were younger, lived in metropolitan areas, or were Hispanic or nonwhite had significantly higher coverage; parents with immunization records provided more-accurate information. While a logistic model-based estimator developed using both parent and provider data performed slightly better than the estimator based on parent data alone, they differed mostly in the subgroups that had small sample sizes. CONCLUSIONS: These are the first statewide provider-verified hepatitis A vaccine coverage estimates for children older than 35 months and indicate that telephone survey estimates as developed using this methodology could prove useful for immunization surveillance activities if interpreted cautiously.

Empirical evaluation of complex epidemiologic study designs: workplace exposure and cancer.

OBJECTIVE: To test whether a frequently used cohort-nested case-control study design exaggerated exposure-response relationships because of unrecognized study design bias. Our aim was to evaluate empirically the performance of this complex study design. METHODS: We applied the design from one such study to a closely related cohort using randomly selected probands as cases. Values for average exposures were assigned to probands equal to, greater than, and less than those assigned to controls (matches). RESULTS: Under certain lag scenarios, the nested study design produced higher average exposure in probands compared with their matches, even when this was clearly not the case. CONCLUSIONS: Empirical evaluation demonstrated that the study design produced a biased case-control lagged exposure difference under the null hypothesis and could not distinguish qualitatively between null and alternate hypotheses. Empirical evaluation provided a useful check on results generated from a complex study design. It gave useful insight into the behavior of the index study design that was not otherwise readily deducible.

Predictors of smallpox vaccination among healthcare workers and other first responders.

BACKGROUND: The goal of the National Smallpox Vaccination Program was to vaccinate a cadre of healthcare workers and first responders who could care for smallpox patients in the event of an attack. METHODS: Using a convenience sample of 36 health departments and 34 hospitals in California, we conducted a telephone interview between July 2003 and April 2004 of healthcare workers and first responders to determine predictors of smallpox vaccination. FINDINGS: The response rate was 54.1%. Of 477 respondents with no contraindications to vaccination, 106 were vaccinated and 371 were unvaccinated. Among the vaccinated, the leading reason for vaccination was wanting to be part of a smallpox response team (74%). Among the unvaccinated, leading reasons for not being vaccinated included thinking the risk of smallpox was not high enough (25%) and concern about side effects (19%). Factors independently associated with vaccination include previous smallpox vaccination (adjusted odds ratio [AOR]=4.1, 95% confidence interval [CI]=1.2-14.1), having little or no concern about vaccine adverse events (AOR=3.0, CI=1.3-7.0, compared with somewhat/very), reporting their employer had a policy to reimburse for travel or other out of pocket costs (AOR=2.5, CI=1.1-5.7, compared with no policy), very high to high chance of compensation if adverse events occurred (AOR=2.9, CI=1.2-6.3, compared with low chance), and answering in the negative to questions about concerns about potential costs. Blacks were less likely than whites to be vaccinated (AOR=0.04, CI=0.03-0.6). CONCLUSIONS: Clearly communicating the risks and benefits of vaccination and addressing issues of cost, convenience, and compensation may be important for any program where vaccination is provided in the national interest and when the direct benefits of vaccination are unknown.

Exposure to beryllium and occurrence of lung cancer: a reexamination of findings from a nested case-control study.

OBJECTIVE: Our aim was to reanalyze a nested case-control study of beryllium and lung cancer because we identified analysis and study design issues that could have led to the elevated odds ratios obtained in the study. METHODS: We reanalyzed the data using nontransformed exposure metrics instead of log-transformed metrics used in the publication. We identified and examined effects on estimated odds ratios of imbalances between cases and controls caused by the control selection method. RESULTS: This reanalysis found no elevated odds ratios for any exposure variable. CONCLUSION: : Our conclusions differ from the authors' interpretation that the findings are due to a causal relationship between beryllium exposure and lung cancer. Our alternative explanation is that they may be due to methodological problems that could have been controlled by closer matching of controls to cases. CLINICAL SIGNIFICANCE: This study challenges conclusions made from a large case-control study concerning beryllium-lung cancer associations. Occupational medicine practitioners may want to integrate findings from this study into advice they give beryllium-exposed workers concerned about lung cancer.

Monitoring county-level vaccination coverage during the 2004-2005 influenza season.

BACKGROUND: During the 2004-2005 influenza season, the United States faced a sudden shortage of influenza vaccine. In response, the Centers for Disease Control and Prevention (CDC) and the Advisory Committee on Immunization Practices (ACIP) recommended prioritizing vaccination for persons aged 65 and older and others at high risk. To monitor subsequent vaccination coverage, several questions about influenza vaccination were added to the ongoing Behavioral Risk Factor Surveillance System (BRFSS). This study provided real-time county-level estimates of influenza vaccination coverage from the BRFSS each month from October 2004 through January 2005. METHOD: The methods used a variation of small area estimation procedures suitable for situations in which most small areas have few or no survey respondents, and rapid assessment is essential. Both model-based methods and nonparametric spatial-smoothing methods were used in a three-step procedure. RESULTS: The highest vaccination rates during the 2004-2005 influenza season were seen in the upper Midwest and the Southeast. Areas with the lowest vaccination rates were the intermountain West, southern California, portions of Washington and Oregon, and various areas across the Eastern United States, often coinciding with urban areas. Intrastate variations were especially pronounced in the Eastern United States, particularly in Georgia, Florida, Tennessee, Kentucky, North Carolina, Virginia, and New York. These states all had areas with low immunization rates as well as areas with high rates. CONCLUSIONS: The results showed that vaccination coverage varied significantly across states and substate regions. Our findings show that this methodology can provide estimates with reasonable reliability for planning during public health emergencies.

Healthcare workers who elected not to receive smallpox vaccination.

BACKGROUND: The goal of the National Smallpox Vaccination Program was to vaccinate a cadre of healthcare workers and first responders who could care for smallpox patients in the event of an attack. METHODS: Using a convenience sample of health departments (n=49) and hospitals (n=60) in five states, we conducted a telephone interview between July 2003 and April 2004 of healthcare workers and first responders who chose not to receive smallpox vaccination. (Data were analyzed in 2004 and 2005.) RESULTS: The response rate was 63%. Of 1895 respondents, 723 (38.2%) reported having a contraindication, 280 (14.8%) reported being contraindicated because of a household member's condition, and 892 (47.0%) reported having no contraindication to smallpox vaccination. Among respondents with no contraindication, the leading reasons for nonvaccination were concerns about side effects (20.6%) and not feeling that the risk of outbreak was high enough (19.5%). More than half (54.8%) were somewhat or very concerned about having an adverse reaction to the vaccine; Hispanics, blacks, and Asians were significantly more likely than whites to be somewhat or very concerned about side effects. Less than one fifth (17.9%) reported that there was a policy to financially compensate employees who developed side effects from vaccination, and 40.7% reported that there was a policy to provide liability coverage to employees who transmitted vaccinia to a patient. CONCLUSIONS: Many people who chose not to receive smallpox vaccine perceived their personal risk-benefit balance as not favoring vaccination. The success of future smallpox vaccination efforts or vaccination against other bioterrorist health threats depends on addressing potential barriers to participation including compensation and liability issues, in addition to clearly communicating risks and benefits.

Adaptation of multiple logistic regression to a multiple inverse sampling design: application to the Isfahan healthy heart program.

In observational and experimental studies in the health sciences involving human populations, it is sometimes considered desirable to recruit subjects according to designs that specify a predetermined number of subjects in each of several mutually exclusive classes (generally but not necessarily demographic in nature). This type of adaptive sampling design, now generally referred to as multiple inverse sampling (MIS), has received recent attention, and estimation methods are now available for several sequential MIS sampling designs. In this class of designs, subjects are sampled randomly and sequentially, usually one at a time, until all classes have the pre-specified number of subjects. In this paper, we extend MIS for finite population sampling to estimation of the parameters in multiple logistic regression under MIS. Using estimated logistic regression parameters and cost components obtained from the Isfahan Healthy Heart Program (IHHP), we report findings from a simulation experiment in which it appears that, at fixed cost, MIS at the last stage of sampling compares favourably to simple random sampling. The IHHP is a large community intervention study for prevention of cardiovascular disease being conducted in Isfahan, Iran and two other cities in Iran. The IHHP identified subjects through a multistage sample survey in which MIS was used at the final stage of sampling. MIS is one of several methods of adaptive sampling that are generating considerable interest and show promise of being useful in a wide variety of applications.

Beryllium and lung cancer: a reanalysis of a NIOSH cohort mortality study.

This analysis is motivated by recent reviews on the carcinogenicity of beryllium by the International Agency for Research on Cancer, the U.S. Environmental Protection Agency, and the American Conference of Governmental Industrial Hygienists, and reconsideration by the National Toxicology Program on its classification of the carcinogenicity of beryllium. It reanalyzes data from a 1992 publication of a cohort mortality study conducted by the National Institute of Occupational Safety and Health (NIOSH) of workers employed in seven plants producing beryllium in the United States (Ward et al., 1992). That publication reported an increased risk of lung cancer in these workers and concluded that it is most likely due to occupational exposure to beryllium compounds. This present report uses: (1) an adjustment for smoking based on more germane estimates of the association between smoking and mortality from lung cancer; (2) computations of expected lung cancer rates based on alternative comparison populations; and (3) an overall combined estimate of the findings from the individual plants based on meta-analysis. Our findings indicate lower and generally not statistically significant standard mortality ratios that are not compatible with the interpretation of a likely causal association.