OBJECTIVE: To evaluate whether the use of salmeterol is associated with an increased risk of an asthma-related hospitalization or emergency care among asthma patients. METHODS: The data for this study were extracted from the Medstat MarketScan® claims databases occurring between 01/01/00 and 12/31/01. A nested case-control study design was employed. A cohort representing asthma patients was identified in 2000. Among the study cohort, hospitalized cases were identified as those who had the firsttime asthma-related hospitalization in 2001, and were matched to select controls by age (± 5 years), sex, and the number of ambulatory visits for asthma (5:1 control to case ratio). A similar process was applied to evaluating an asthma-related emergency department (ED) visit. The odds of prior salmeterol exposure among cases compared to controls were estimated using conditional multiple logistic regressions. The covariates of interest comprised age, prior hospital admission or ED visit for asthma, number of canisters of inhaled short-acting β₂ agonists and use of other asthma medications. RESULTS: A total of 35,312 subjects were eligible to be the study cohort. In addition, 285 and 640 subjects were identified as hospitalized and ED cases, respectively. The non-significant association was found when the prior salmeterol exposure was treated as a dichotomized variable. However, it was found that one unit increase in the number of canisters of salmeterol was associated with a seven percent decrease in the risk of a hospital admission for asthma (p <0.001). Additionally, current use of salmeterol was associated with a 48 percent decrease in the risk of an asthma-related hospitalization (OR = 0.52; p <0.001). The protective effect of salmeterol did not exist for those with recent or past use of salmeterol. Similar findings were observed for the ED visit outcome. CONCLUSIONS: The use of salmeterol was not found to be associated with an increased risk of an asthma-related hospital admission or ED visit. Conversely, one unit increase in the number of canisters of salmeterol and current use of salmeterol, respectively, were found to decrease the risk in an asthma-related hospitalization or ED visit among asthma patients.

Cadmium and lead accumulate in humans and can have toxic effects. Exposure to cadmium is well known to cause kidney damage. Cadmium binds to metallothioneins, proteins that play a role in cadmium transport. Lead exposure’s main effect is on the central nervous system, but associations with kidney disease have also been found, although it is unknown if the latter is a causal association. The main source of both metals within the non-smoking population is from the diet. This thesis aims to 1) compare the biomarkers lead and cadmium concentration in whole-blood, plasma and urine with regard to their ability to discriminate between individuals with different mean concentrations, and to describe the effect of urinary dilution, 2) estimate the association between end-stage renal disease and blood concentrations of cadmium, lead and mercury, using prospectively collected samples for exposure evaluation, 3) use longitudinal data on kidney function makers to evaluate kidney recovery after a substantial decrease in cadmium exposure, and 4) assess the influence of metallothionein polymorphisms (MT1A rs11076161, MT2A rs10636 and MT2A rs28366003) on cadmium-associated kidney toxicity and recovery due to a reduction in Cd exposure. Repeated sampling of whole-blood, plasma and urine was conducted on 48 occupationally lead-exposed men and 20 individuals under normal environmental lead exposure, for estimation of the day-to-day and between individual-variation. Prospective samples were obtained for 118 cases that later in life developed end-stage renal disease, and 378 matched controls. Erythrocyte cadmium, lead, and mercury concentrations were determined and the risk of developing end-stage renal disease associated with metal concentrations was estimated. For evaluation of kidney recovery after a reduction in cadmium exposure and to test for gene-environment interactions, follow-up data on N-acetyl-β‑d-glucosaminidase, β2‑microglobulin, albumin, and gene polymorphisms were obtained for 412 individuals within the Chinese population and the relation to blood and urinary cadmium was assessed. The concentration of lead in blood was found to be the biomarker with the largest fraction of the total variance attributable to between-individual variation, and was therefore the biomarker with the best ability to discriminate between individuals with different mean concentrations, both for individuals under occupational and normal environmental exposure (91 and 95%, respectively). Adjusting for urinary dilution had a great effect on the fraction of the total variance attributable to between-individual variation among individuals with normal lead exposure but only a minor effect among those who were occupationally exposed. Variance analysis showed that blood concentrations were also the best discriminating biomarker for cadmium. Erythrocyte lead was, in a univariate model, associated with an increased risk of developing end-stage renal disease [odds ratio (OR) = 1.54 for an interquartile range increase, with a 95% confidence interval (CI) = 1.18-2.00], while erythrocyte mercury was negatively associated (OR = 0.75 for an interquartile range increase, with a 95% CI = 0.56-0.99). For erythrocyte cadmium, the OR was 1.15 with a 95% CI of 0.99-1.34. Associations with lead and cadmium were only seen among men. In the study on kidney recovery, the proportion of individuals with albumin level above the 95th percentile decreased between baseline and follow up, but no decrease was found for the tubular markers N-acetyl-β‑d-glucosaminidase and β2-microglobulin. Metallothionein polymorphisms modified cadmium-associated effects on N-acetyl-β‑d-glucosaminidase and β2-microglobulin levels but did not modify cadmium-associated change in any of the kidney function markers between baseline and follow up after a substantial decrease in exposure. Blood concentrations of lead and cadmium are the biomarkers with the best ability to discriminate between individuals with different mean concentrations. Adjustment for urinary dilution has great influence on the fraction of the total variance attributed to between individual variation among urine samples with low lead concentrations, but only a small influence on samples with high lead concentrations. This suggests a difference in excretion. The association between end-stage renal disease and low-level lead exposure, as assessed through prospective erythrocyte samples, gives reason for concern, although further studies are needed to determine causality. A cadmium-associated increase in albumin is reversible after a substantial reduction in exposure, but this is not the case for the observed tubular effects. The tubular kidney effects of cadmium might be modified by the MT1A rs11076161 polymorphism. / För att bedöma exponering för kadmium och bly mäts ofta deras koncentrationer i blod eller urin. Dessa studerades i longitudinella data för 48 blyarbetare och 20 individer med normal miljömässig exponering. Blod- och urinprover togs var annan till var tredje månad. Kadmium- och blykoncentrationer mättes sedan i helblod, plasma och urin. Koncentrationer av bly i blod var den biomarkör som hade den största andelen av den totala variationen som kunde förklaras av skillnader mellan individer, och var därför den biomarkör med den bästa förmågan att särskilja på individer med olika medelkoncentration, både för individer med yrkesexponering och normal miljömässig exponering (91 respektive 95% av variansen berodde på vilken individ blodprovet kom ifrån). Justering för urinens utspädning av bly i urin förbättrar oftast urins användbarhet som biomarkör. För bly stämde detta bara hos dem som inte var blyarbetare. Blodkoncentrationer var också den biomarkör med störst andel av den totala variation som kunde förklaras med skillnader mellan individer för kadmium. Kadmium och bly ackumuleras i njure respektive ben och kan ha toxikologiska effekter. Det är välkänt att höga exponeringsnivåer av kadmium orsakar njurskada och även vid lägre exponeringsnivåer har studier funnit samband med markörer för njurfunktion. Exponering för bly påverkar i första hand det centrala nervsystemet. Studier har dock funnit samband mellan koncentrationer av bly i blod och njurens glomerulära filtrationshastighet. Det är oklart både om dessa associationer, vid låga exponeringsnivåer, är viktiga för hälsan och om de verkligen beror på att kadmium och bly orsakar njurskada. För att studera end-stage renal disease användes prospektiva kohorter där personer lämnat blodprov för forskning: Västerbottens interventionsprogram med prover som tagits vid Västerbottens hälsoundersökningar, MONICA-undersökningar i Norr- och Västerbotten, mammografiundersökningarna i Västerbotten och Malmö kost cancer. Sammanlagt ingick över ett hundra tusen individer i dessa kohorter. Med hjälp av det Svenska njurregistret identifierades sedan 118 personer som senare i livet fått end-stage renal disease. Dessa jämfördes med 378 kontroller. För dessa 496 personer tinades blodprovet (närmare bestämt röda blodkroppar) upp och analyserades för kadmium och bly. För att undersöka njurens förmåga till återhämtning studerades tre områden i Kina varav ett tidigare varit kraftigt kadmiumexponerat. Erytrocytkoncentrationer av bly var, utan att ta hänsyn till några andra variabler, associerat med en ökad risk för att utveckla end-stage renal disease (med oddskvoten 1.54 för en interquartile range ökning av erytrocytbly, med ett 95% konfidensintervall 1.18-2.00). Sambanden kvarstod också efter att ha tagit hänsyn till övriga variabler. För erytrocytkadmium var oddskvoten 1.15 med 95% konfidensintervall 0.99-1.34, och sambandet försvagades när hänsyn togs till andra variabler. Associationerna sågs bland män men inte bland kvinnor. Eftersom kadmium vid höga nivåer orsakar njurskada är det också av intresse att studera om påverkan på njuren går över om exponeringen minskas. Totalt följdes 412 individer upp med mätningar av markörer för njurfunktion och kadmiumkoncentrationer i blod och urin. Första undersökningen gjordes 1998, då man just hade slutat äta kadmiumförorenat ris. En andra undersökning gjordes 2006. Andelen individer med avvikande albuminvärde i urin var lägre vid uppföljningen jämfört med vid baslinjen, men ingen minskning sågs för markörer för tubulär förmåga att återta proteiner. Åttioprocent av kadmium i celler är bundet till proteinet metallotheonin, vilket skyddar mot cellskada, men har också en roll i transporten av kadmium från levern till njurarna. En tidigare studie har visat att njurens känslighet för kadmiumexponering var associerad med genetiska skillnader i detta protein. För att studera genetiska associationer studerades de 412 personerna i den kinesiska studien [då också individernas genotyper av metallotheonin-polymorfierna MT1A rs11076161 (G/A), MT2A rs10636 (G/C) och MT2A rs28366003 (A/G) bestämdes]. Genetiken spelade roll för sambandet mellan förmåga att återta proteiner och kadmium men påverkade inte förändring av njurfunktion efter att man slutat äta kadmiumförorenat ris. Kadmium- och blykoncentrationer i blod är de biomarkörer, av koncentrationer i blod, plasma och urin, med den bästa förmågan att skilja på individer med olika medelkoncentrationer. Justering för urinutspädning påverkade andelen av den totala variationen som kunde förklaras av skillnader mellan individer i stor utsträckning för individer med normal miljömässig exponering men inte bland yrkesexponerade, vilket tyder på en skillnad i hur utsöndringen går till. Associationen mellan end-stage renal disease och låg exponering för bly, uppmätta i prospektiva erytrocytprover, ger orsak till oro, men ytterligare studier behövs för att kunna utvärdera om detta är ett kausalt samband. En kadmiumrelaterad skada av den glomerulära filtrationen är reversibel efter en kraftig reducering i exponering, men detta är inte fallet för tubulär skada. De tubulära njureffekterna av kadmiumexponering kan påverkas av metallotheonin-polymorfier.

In public health, biomedical, epidemiological, and other applications, data collected are often measured with error. When mismeasured data is used in a regression analysis, not accounting for the measurement error can lead to incorrect inference about the relationships between the covariates and the response. We investigate measurement error in the covariates of two types of regression models.  For each we propose a fully Bayesian approach that treats the variable measured with error as a latent variable to be integrated over, and a semi-Bayesian approach which uses a first order Laplace approximation to marginalize the variable measured with error out of the likelihood. The first model is the matched case-control study for analyzing clustered binary outcomes. We develop low-rank thin plate splines for the case where a variable measured with error has an unknown, nonlinear relationship with the response. In addition to the semi- and fully Bayesian approaches, we propose another using expectation-maximization to detect both parametric and nonparametric relationships between the covariates and the binary outcome. We assess the performance of each method via simulation terms of mean squared error and mean bias. We illustrate each method on a perturbed example of 1--4 matched case-control study. The second regression model is the generalized linear model (GLM) with unknown link function. Usually, the link function is chosen by the user based on the distribution of the response variable, often to be the canonical link. However, when covariates are measured with error, incorrect inference as a result of the error can be compounded by incorrect choice of link function. We assess performance via simulation of the semi- and fully Bayesian methods in terms of mean squared error. We illustrate each method on the Framingham Heart Study dataset. The simulation results for both regression models support that the fully Bayesian approach is at least as good as the semi-Bayesian approach for adjusting for measurement error, particularly when the distribution of the variable of measure with error and the distribution of the measurement error are misspecified. / Ph. D.

This dissertation consists of three projects in matched case-control studies. In the first project, we employ a general bias preventive approach developed by Firth (1993) to handle the bias of an estimator of the log-odds ratio parameter in conditional logistic regression by solving a modified score equation. The resultant estimator not only reduces bias but also can prevent producing infinite value. Furthermore, we propose a method to calculate the standard error of the resultant estimator. A closed form expression for the estimator of the log-odds ratio parameter is derived in the case of a dichotomous exposure variable. Finite sample properties of the estimator are investigated via a simulation study. Finally, we apply the method to analyze a matched case-control data from a low-birth-weight study. In the second project of this dissertation, we propose a score typed test for checking adequacy of a functional form of a covariate of interest in matched case-control studies by using penalized regression splines to approximate an unknown function. The asymptotic distribution of the test statistics under the null model is a linear combination of several chi-square random variables. We also derive the asymptotic distribution of the test statistic when the alternative model holds. Through a simulation study we assess and compare the finite sample properties of the proposed test with that of Arbogast and Lin (2004). To illustrate the usefulness of the method, we apply the proposed test to a matched case-control data constructed from the breast cancer data of the SEER study. Usually a logistic model is needed to associate the risk of the disease with the covariates of interests. However, this logistic model may not be appropriate in some instances. In the last project , we adopt idea to matched case-control studies and derive an information matrix based test for testing overall model adequacy and investigate the properties against the cumulative residual based test in Arbogast and Lin (2004) via a simulation study. The proposed method is less time consuming and has comparative power for small parameters. It is suitable to explore the overall model fitting.

This project discusses some of the methodologies developed over the years to estimate attributable risk among exposed persons and the attributable risk in the entire population (also called Etiologic Fraction). It provides a general framework for estimating attributable risk among the exposed (denoted lambda_e). By making use of the recent observation that the two measures of attributable risk can be linked through the prevalence of the risk factor among the cases (denoted V_x), an estimate of population attributable risk (denoted lambda) for matched case-control studies is determined. Using the methodology developed recently by Kuritz and Landis (1987), this project provides explicit formulas for estimating the attributable risk among the exposed and the population attributable risk, and their large sample variances. This has been done both in situations where exactly R controls have been matched to a case and for a variable number of controls per case. The methodologies are illustrated with data from some case-control studies reported in the literature. Asymptotic relative efficiencies of different matching designs computed in terms of the costs of gathering cases and controls, are presented, together with some recommendations on what design is considered optimal. / Thesis / Master of Science (MSc)

Semiparametric modeling is a combination of the parametric and nonparametric models in which some functions follow a known form and some others follow an unknown form. In this dissertation we made contributions to semiparametric modeling for matched case-crossover data. In matched case-crossover studies, it is generally accepted that the covariates on which a case and associated controls are matched cannot exert a confounding effect on independent predictors included in the conditional logistic regression model. Any stratum effect is removed by the conditioning on the fixed number of sets of the case and controls in the stratum. However, some matching covariates such as time, and/or spatial location often play an important role as an effect modification. Failure to include them makes incorrect statistical estimation, prediction and inference. Hence in this dissertation, we propose several approaches that will allow the inclusion of time and spatial location as well as other effect modifications such as heterogeneous subpopulations among the data. To address modification due to time, three methods are developed: the first is a parametric approach, the second is a semiparametric penalized approach and the third is a semiparametric Bayesian approach. We demonstrate the advantage of the one stage semiparametric approaches using both a simulation study and an epidemiological example of a 1-4 bi-directional case-crossover study of childhood aseptic meningitis with drinking water turbidity. To address modifications due to time and spatial location, two methods are developed: the first one is a semiparametric spatial-temporal varying coefficient model for a small number of locations. The second method is a semiparametric spatial-temporal varying coefficient model, and is appropriate when the number of locations among the subjects is medium to large. We demonstrate the accuracy of these approaches by using simulation studies, and when appropriate, an epidemiological example of a 1-4 bi-directional case-crossover study. Finally, to explore further effect modifications by heterogeneous subpopulations among strata we propose a nonparametric Bayesian approach constructed with Dirichlet process priors, which clusters subpopulations and assesses heterogeneity. We demonstrate the accuracy of our approach using a simulation study, as well a an example of a 1-4 bi-directional case-crossover study. / Ph. D.

Both spinal pain and temporomandibular disorders (TMD) commonly occur in the general population. Previous studies demonstrate neurophysiologic and biomechanical couplings between the trigeminal and cervical regions. This investigation tested the null hypothesis of no relationship between spinal pain (neck, shoulder and/or low back) and TMD, by using questionnaires and clinical examinations of the jaw function. In an age- and sex-matched case-control study, the specific aim was to compare the prevalence of signs and symptoms of TMD among cases with long-term spinal pain and controls without spinal pain. The results showed that subjects with spinal pain had signs and symptoms of TMD significantly more often than did controls. The associations remained after excluding all participants with jaw pain. Furthermore, the comorbidity pattern was similar, regardless of location of spinal pain. In a cross-sectional study, the specific aim was to test whether there is a reciprocal cross-sectional dose-response-like relationship between spinal pain and TMD. Two different designs were used, one with frequency/severity of spinal pain as independent variable, and the other, with frequency/severity of TMD symptoms as independent variable. The analysis showed increasing odds for presence of TMD symptoms with increasing frequency/severity of spinal pain, and increasing odds for presence of spinal pain with increasing frequency/severity of TMD symptoms. In a case-control study within a 2-year prospective cohort, the specific aim was to test whether there is a reciprocal temporal relationship between signs and symptoms in trigeminally, and symptoms in spinally, innervated areas. Incidence of symptoms in these areas was analyzed in relation to presence of spinal pain, headaches, and signs and symptoms of TMD at baseline. The main findings were that presence of signs of TMD at baseline increased the onset of spinal pain and symptoms in the trigeminal area, and that spinal pain increased the onset of symptoms in the trigeminal area. An augmentation effect between the significant baseline variables was observed for the incidence of headaches and jaw pain. In conclusion, the investigation demonstrated a cross-sectional and temporal relationship between spinal pain and TMD; thus, the null hypothesis was rejected. The results indicate common pathophysiological mechanisms in the development of spinal pain and TMD. The comorbidity and reciprocal influence that were found call for an integrated and multidimensional approach in the management of individuals with long-term spinal pain and TMD.

A variety of complex data has broadened in many research fields such as epidemiology, genomics, and analytical chemistry with the development of science, technologies, and design scheme over the past few decades. For example, in epidemiology, the matched case-crossover study design is used to investigate the association between the clustered binary outcomes of disease and a measurement error in covariate within a certain period by stratifying subjects' conditions. In genomics, high-correlated and high-dimensional(HCHD) data are required to identify important genes and their interaction effect over diseases. In analytical chemistry, multiple time series data are generated to recognize the complex patterns among multiple classes. Due to the great diversity, we encounter three problems in analyzing those complex data in this dissertation. We have then provided several contributions to semiparametric and nonparametric methods for dealing with the following problems: the first is to propose a method for testing the significance of a functional association under the matched study; the second is to develop a method to simultaneously identify important variables and build a network in HDHC data; the third is to propose a multi-class dynamic model for recognizing a pattern in the time-trend analysis. For the first topic, we propose a semiparametric omnibus test for testing the significance of a functional association between the clustered binary outcomes and covariates with measurement error by taking into account the effect modification of matching covariates. We develop a flexible omnibus test for testing purposes without a specific alternative form of a hypothesis. The advantages of our omnibus test are demonstrated through simulation studies and 1-4 bidirectional matched data analyses from an epidemiology study. For the second topic, we propose a joint semiparametric kernel machine network approach to provide a connection between variable selection and network estimation. Our approach is a unified and integrated method that can simultaneously identify important variables and build a network among them. We develop our approach under a semiparametric kernel machine regression framework, which can allow for the possibility that each variable might be nonlinear and is likely to interact with each other in a complicated way. We demonstrate our approach using simulation studies and real application on genetic pathway analysis. Lastly, for the third project, we propose a Bayesian focal-area detection method for a multi-class dynamic model under a Bayesian hierarchical framework. Two-step Bayesian sequential procedures are developed to estimate patterns and detect focal intervals, which can be used for gas chromatography. We demonstrate the performance of our proposed method using a simulation study and real application on gas chromatography on Fast Odor Chromatographic Sniffer (FOX) system. / Doctor of Philosophy / A variety of complex data has broadened in many research fields such as epidemiology, genomics, and analytical chemistry with the development of science, technologies, and design scheme over the past few decades. For example, in epidemiology, the matched case-crossover study design is used to investigate the association between the clustered binary outcomes of disease and a measurement error in covariate within a certain period by stratifying subjects' conditions. In genomics, high-correlated and high-dimensional(HCHD) data are required to identify important genes and their interaction effect over diseases. In analytical chemistry, multiple time series data are generated to recognize the complex patterns among multiple classes. Due to the great diversity, we encounter three problems in analyzing the following three types of data: (1) matched case-crossover data, (2) HCHD data, and (3) Time-series data. We contribute to the development of statistical methods to deal with such complex data. First, under the matched study, we discuss an idea about hypothesis testing to effectively determine the association between observed factors and risk of interested disease. Because, in practice, we do not know the specific form of the association, it might be challenging to set a specific alternative hypothesis. By reflecting the reality, we consider the possibility that some observations are measured with errors. By considering these measurement errors, we develop a testing procedure under the matched case-crossover framework. This testing procedure has the flexibility to make inferences on various hypothesis settings. Second, we consider the data where the number of variables is very large compared to the sample size, and the variables are correlated to each other. In this case, our goal is to identify important variables for outcome among a large amount of the variables and build their network. For example, identifying few genes among whole genomics associated with diabetes can be used to develop biomarkers. By our proposed approach in the second project, we can identify differentially expressed and important genes and their network structure with consideration for the outcome. Lastly, we consider the scenario of changing patterns of interest over time with application to gas chromatography. We propose an efficient detection method to effectively distinguish the patterns of multi-level subjects in time-trend analysis. We suggest that our proposed method can give precious information on efficient search for the distinguishable patterns so as to reduce the burden of examining all observations in the data.

Background: Non-disclosure of HIV positive status to a partner threatens to reverse gains made in prevention of mother-to-child transmission (PMTCT) in resource limited settings. Determining the association between non-disclosure and infant HIV acquisition is important to justify focussing on disclosure as a strategy in PMTCT programmes. Objective: To determine the association between non-disclosure of HIV positive status to a partner and mother-to-child transmission (MTCT). Methods: Using a matched case-control design, we compared 34 HIV positive infants to 146 HIV negative infants and evaluated whether the mothers had disclosed their HIV status to their partner. Results: Non-disclosure was more frequent among cases (overall, 16.7%; cases, 52.8%; controls 7.6%), p<0.001 and significantly associated with MTCT (aOR 8.9 (3.0-26.3); p<0.0001), with male partner involvement partially mediating the effect of non-disclosure on MTCT. Conclusions: There is a need for PMTCT programs to focus on strategies to improve male partner involvement and partner disclosure without compromising the woman’s safety. / Health Studies / M. (Public Health)

Background: Non-disclosure of HIV positive status to a partner threatens to reverse gains made in prevention of mother-to-child transmission (PMTCT) in resource limited settings. Determining the association between non-disclosure and infant HIV acquisition is important to justify focussing on disclosure as a strategy in PMTCT programmes. Objective: To determine the association between non-disclosure of HIV positive status to a partner and mother-to-child transmission (MTCT). Methods: Using a matched case-control design, we compared 34 HIV positive infants to 146 HIV negative infants and evaluated whether the mothers had disclosed their HIV status to their partner. Results: Non-disclosure was more frequent among cases (overall, 16.7%; cases, 52.8%; controls 7.6%), p<0.001 and significantly associated with MTCT (aOR 8.9 (3.0-26.3); p<0.0001), with male partner involvement partially mediating the effect of non-disclosure on MTCT. Conclusions: There is a need for PMTCT programs to focus on strategies to improve male partner involvement and partner disclosure without compromising the woman’s safety. / Health Studies / M. (Public Health)