RESUMO
Statistical models incorporating cluster-specific intercepts are commonly used in hierarchical settings, for example, observations clustered within patients or patients clustered within hospitals. Predicted values of these intercepts are often used to identify or "flag" extreme or outlying clusters, such as poorly performing hospitals or patients with rapid declines in their health. We consider a variety of flagging rules, assessing different predictors, and using different accuracy measures. Using theoretical calculations and comprehensive numerical evaluation, we show that previously proposed rules based on the 2 most commonly used predictors, the usual best linear unbiased predictor and fixed effects predictor, perform extremely poorly: the incorrect flagging rates are either unacceptably high (approaching 0.5 in the limit) or overly conservative (eg, much <0.05 for reasonable parameter values, leading to very low correct flagging rates). We develop novel methods for flagging extreme clusters that can control the incorrect flagging rates, including very simple-to-use versions that we call "self-calibrated." The new methods have substantially higher correct flagging rates than previously proposed methods for flagging extreme values, while controlling the incorrect flagging rates. We illustrate their application using data on length of stay in pediatric hospitals for children admitted for asthma diagnoses.
Assuntos
Asma , Modelos Estatísticos , Criança , Humanos , Modelos Lineares , Hospitalização , Asma/diagnósticoRESUMO
BACKGROUND: Individual behaviours are associated with prostate cancer (PC) progression. Behavioural scores, comprised of multiple risk factors, allow assessment of the combined impact of multiple behaviours. METHODS: We examined the association between six a priori scores and risk of PC progression and mortality among 2156 men with PC in the Cancer of the Prostate Strategic Urologic Research Endeavor (CaPSURE) cohort: two scores developed based on the PC survivorship literature ('2021 Score [+ Diet]'); a score developed based on pre-diagnostic PC literature ('2015 Score'); and three scores based on US recommendations for cancer prevention ('WCRF/AICR Score') and survival ('ACS Score [+ Alcohol]'). Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated for progression and PC mortality via parametric survival models (interval censoring) and Cox models, respectively. RESULTS: Over a median (IQR) of 6.4 (1.3, 13.7) years, we observed 192 progression and 73 PC mortality events. Higher (i.e., healthier) 2021 Score + Diet and WCRF/AICR Scores were inversely associated with risk of PC progression (2021 + Diet: HRcontinuous = 0.76, 95% CI: 0.63-0.90. WCRF/AICR: HRcontinuous = 0.83, 95% CI: 0.67-1.02) and mortality (2021 + Diet: HRcontinuous = 0.65, 95% CI: 0.45-0.93. WCRF/AICR: HRcontinuous = 0.71; 95% CI: 0.57-0.89). The ACS Score + Alcohol was only associated with progression (HRcontinuous = 0.89, 95% CI: 0.81-0.98) while the 2021 Score was only associated with PC mortality (HRcontinuous = 0.62, 95% CI: 0.45-0.85). The 2015 was not associated with PC progression or mortality. CONCLUSION: Findings strengthen the evidence that behavioural modifications following a prostate cancer diagnosis may improve clinical outcomes.
Assuntos
Neoplasias da Próstata , Masculino , Humanos , Neoplasias da Próstata/diagnóstico , Comportamentos Relacionados com a Saúde , Dieta , Fatores de RiscoRESUMO
With the advent of electronic health records, information collected in the course of regular health care is increasingly being used for clinical research. The hope is that the wealth of clinical data and the realistic setting (compared with information derived from highly controlled experiments like randomized trials) will aid in the investigation of determinants of disease and understanding of which treatments are effective in regular practice and for which patients. The availability of information in such databases is often driven by how a patient feels and may therefore be associated with the health outcomes being considered. We call this an outcome dependent visit process and recent work has shown that ignoring the outcome dependence can produce significant bias in the regression coefficients when fitting longitudinal data models. It is therefore important to have tools to recognize datasets exhibiting outcome dependence. We develop a score statistic to motivate the form of diagnostic test statistics, suggest a variety of approaches for diagnosing such situations, and evaluate their performance. Simple diagnostic tests achieve high power for diagnosing outcome dependent visit processes. This occurs when generalized estimating equations methods begin to be exhibit bias in estimating regression coefficients and before likelihood based methods are substantially biased.
Assuntos
Assistência Ambulatorial/estatística & dados numéricos , Diagnóstico , Modelos Estatísticos , Visita a Consultório Médico/estatística & dados numéricos , Avaliação de Resultados em Cuidados de Saúde , Bases de Dados Factuais , Conjuntos de Dados como Assunto , Registros Eletrônicos de Saúde , Humanos , Estudos Longitudinais , Doenças do Sistema Nervoso/diagnósticoRESUMO
The timing and frequency of the measurement of longitudinal outcomes in databases may be associated with the value of the outcome. Such visit processes are termed outcome dependent, and previous work showed that conducting standard analyses that ignore outcome-dependent visit times can produce highly biased estimates of the associations of covariates with outcomes. The literature contains several classes of approaches to analyze longitudinal data subject to outcome-dependent visit times, and all of these are based on simplifying assumptions about the visit process. Based on extensive discussions with subject matter investigators, we identified common characteristics of outcome-dependent visit processes that allowed us to evaluate the performance of existing methods in settings with more realistic visit processes than have been previously investigated. This paper uses the analysis of data from a study of kidney function, theory, and simulation studies to examine a range of settings that vary from those where all visits have a low degree of missingness and outcome dependence (which we call "regular" visits) to those where all visits have a high degree of missingness and outcome dependence (which we call "irregular" visits). Our results show that while all the approaches we studied can yield biased estimates of some covariate effects, other covariate effects can be estimated with little bias. In particular, mixed effects models fit by maximum likelihood yielded little bias in estimates of the effects of covariates not associated with the random effects and small bias in estimates of the effects of covariates associated with the random effects. Other approaches produced estimates with greater bias. Our results also show that the presence of some regular visits in the data set protects mixed model analyses from bias but not other methods.
Assuntos
Interpretação Estatística de Dados , Estudos Longitudinais , Resultado do Tratamento , Viés , Taxa de Filtração Glomerular , Humanos , Transplante de Rim/estatística & dados numéricos , Funções Verossimilhança , Modelos Estatísticos , Diálise Renal/estatística & dados numéricos , Insuficiência Renal Crônica/terapia , Fatores de TempoRESUMO
BACKGROUND: With the onset of prevention trials for individuals at high risk for Alzheimer disease, there is increasing need for accurate risk prediction to inform study design and enrollment, but available risk estimates are limited. We developed risk estimates for the incidence of mild cognitive impairment (MCI) or dementia among cognitively unimpaired individuals by APOE-e4 dose for the genetic disclosure process of the Alzheimer's Prevention Initiative Generation Study, a prevention trial in cognitively unimpaired APOE-e4/e4 homozygote individuals. METHODS AND FINDINGS: We included cognitively unimpaired individuals aged 60-75 y, consistent with Generation Study eligibility criteria, from the National Alzheimer's Coordinating Center (NACC) (n = 5,073, 158 APOE-e4/e4), the Rotterdam Study (n = 6,399, 156 APOE-e4/e4), the Framingham Heart Study (n = 4,078, 67 APOE-e4/e4), and the Sacramento Area Latino Study on Aging (SALSA) (n = 1,294, 11 APOE-e4/e4). We computed stratified cumulative incidence curves by age (60-64, 65-69, 70-75 y) and APOE-e4 dose, adjusting for the competing risk of mortality, and determined risk of MCI and/or dementia by genotype and baseline age. We also used subdistribution hazard regression to model relative hazard based on age, APOE genotype, sex, education, family history of dementia, vascular risk, subjective memory concerns, and baseline cognitive performance. The four cohorts varied considerably in age, education, ethnicity/race, and APOE-e4 allele frequency. Overall, cumulative incidence was uniformly higher in NACC than in the population-based cohorts. Among APOE-e4/e4 individuals, 5-y cumulative incidence was as follows: in the 60-64-y age stratum, it ranged from 0% to 5.88% in the three population-based cohorts versus 23.06% in NACC; in the 65-69-y age stratum, from 9.42% to 10.39% versus 34.62%; and in the 70-75-y age stratum, from 18.64% to 33.33% versus 38.34%. Five-year incidence of dementia was negligible except for APOE-e4/e4 individuals and those over 70 y. Lifetime incidence (to age 80-85 y) of MCI or dementia for the APOE-e4/e4 individuals in the long-term Framingham and Rotterdam cohorts was 34.69%-38.45% at age 60-64 y, 30.76%-40.26% at 65-69 y, and 33.3%-35.17% at 70-75 y. Confidence limits for these estimates are often wide, particularly for APOE-e4/e4 individuals and for the dementia outcome at 5 y. In regression models, APOE-e4 dose and age both consistently increased risk, as did lower education, subjective memory concerns, poorer baseline cognitive performance, and family history of dementia. We discuss several limitations of the study, including the small numbers of APOE-e4/e4 individuals, missing data and differential dropout, limited ethnic and racial diversity, and differences in definitions of exposure and outcome variables. CONCLUSIONS: Estimates of the absolute risk of MCI or dementia, particularly over short time intervals, are sensitive to sampling and a variety of methodological factors. Nonetheless, such estimates were fairly consistent across the population-based cohorts, and lower than those from a convenience cohort and those estimated in prior studies-with implications for informed consent and design for clinical trials targeting high-risk individuals.
Assuntos
Apolipoproteínas E/genética , Disfunção Cognitiva/epidemiologia , Demência/epidemiologia , Genótipo , Idoso , Apolipoproteínas E/metabolismo , Disfunção Cognitiva/genética , Estudos de Coortes , Demência/genética , Feminino , Frequência do Gene , Humanos , Masculino , Pessoa de Meia-Idade , Países Baixos/epidemiologia , Fatores de Risco , Estados UnidosRESUMO
The availability of data in longitudinal studies is often driven by features of the characteristics being studied. For example, clinical databases are increasingly being used for research to address longitudinal questions. Because visit times in such data are often driven by patient characteristics that may be related to the outcome being studied, the danger is that this will result in biased estimation compared to designed, prospective studies. We study longitudinal data that follow a generalized linear mixed model and use a log link to relate an informative visit process to random effects in the mixed model. This device allows us to elucidate which parameters are biased under the informative visit process and to what degree. We show that the informative visit process can badly bias estimators of parameters of covariates associated with the random effects, while allowing consistent estimation of other parameters.
Assuntos
Assistência Ambulatorial , Viés , Estudos Longitudinais , Modelos Estatísticos , Bases de Dados Factuais , Humanos , Modelos LinearesRESUMO
OBJECTIVES: We set out to assess the occurrence of new depression and anxiety diagnoses in women 3 years after they sought an abortion. METHODS: We conducted semiannual telephone interviews of 956 women who sought abortions from 30 US facilities. Adjusted multivariable discrete-time logistic survival models examined whether the study group (women who obtained abortions just under a facility's gestational age limit, who were denied abortions and carried to term, who were denied abortions and did not carry to term, and who received first-trimester abortions) predicted depression or anxiety onset during seven 6-month time intervals. RESULTS: The 3-year cumulative probability of professionally diagnosed depression was 9% to 14%; for anxiety it was 10% to 15%, with no study group differences. Women in the first-trimester group and women denied abortions who did not give birth had greater odds of new self-diagnosed anxiety than did women who obtained abortions just under facility gestational limits. CONCLUSIONS: Among women seeking abortions near facility gestational limits, those who obtained abortions were at no greater mental health risk than were women who carried an unwanted pregnancy to term.
Assuntos
Aborto Induzido/psicologia , Ansiedade/epidemiologia , Depressão/epidemiologia , Aborto Induzido/efeitos adversos , Aborto Legal/efeitos adversos , Aborto Legal/psicologia , Adulto , Ansiedade/etiologia , Depressão/etiologia , Feminino , Idade Gestacional , Humanos , Entrevistas como Assunto , Modelos Logísticos , Estudos Longitudinais , Aceitação pelo Paciente de Cuidados de Saúde/psicologia , Aceitação pelo Paciente de Cuidados de Saúde/estatística & dados numéricos , Gravidez , Estudos Prospectivos , Fatores de Risco , Estados Unidos/epidemiologia , Adulto JovemRESUMO
INTRODUCTION: The significance of endothelial injury in children with the acute respiratory distress syndrome (ARDS) has not been well studied. Plasma levels of soluble thrombomodulin (sTM), an endothelial surface protein involved in coagulation, have been associated with endothelial injury. We hypothesized that elevated plasma sTM would correlate with mortality and organ failure in children with ARDS. METHODS: We conducted a multicenter prospective observational study of pediatric patients with ARDS between 2008 and 2014. sTM was measured in plasma collected less than 24 hours from ARDS diagnosis. Outcomes were intensive care unit mortality and organ dysfunction by pediatric logistic organ dysfunction scores. Logistic regression was used to adjust for clinically relevant covariates. RESULTS: Plasma sTM was higher in patients with indirect lung injury compared to direct lung injury (100 ng/mL vs. 86 ng/mL, p = 0.02). Increased sTM levels were correlated with more organ dysfunction in the entire study population (Spearman's rho = 0.37, p < 0.01). Overall mortality was 16%. sTM levels were associated with increased mortality in patients with indirect lung injury (OR 2.7 per log(sTM), p = 0.02). These relationships were independent of age, oxygenation defect, or presence of acute kidney injury. CONCLUSION: Elevated plasma sTM levels are associated with organ dysfunction in children with ARDS and with higher mortality in children with indirect lung injury. These findings highlight the importance of endothelial injury in children with ARDS and may guide the development of future therapies targeted toward endothelial stabilization, repair, or functional replacement in this population.
Assuntos
Lesão Pulmonar/complicações , Pulmão/metabolismo , Insuficiência de Múltiplos Órgãos/etiologia , Síndrome do Desconforto Respiratório/mortalidade , Trombomodulina/sangue , Criança , Pré-Escolar , Estudos de Coortes , Endotélio Vascular/metabolismo , Endotélio Vascular/patologia , Feminino , Humanos , Pulmão/fisiopatologia , Lesão Pulmonar/metabolismo , Lesão Pulmonar/fisiopatologia , Masculino , Insuficiência de Múltiplos Órgãos/mortalidade , Insuficiência de Múltiplos Órgãos/fisiopatologia , Prognóstico , Estudos Prospectivos , Síndrome do Desconforto Respiratório/sangue , Síndrome do Desconforto Respiratório/etiologia , Síndrome do Desconforto Respiratório/fisiopatologia , Trombomodulina/análise , Trombomodulina/metabolismoRESUMO
PURPOSE: Although fear of recurrence (FCR) is common among cancer survivors, it remains unclear what factors predict initial levels (e.g., prior to surgery) or changes in FCR in the post-treatment period. Among women treated for breast cancer, this study evaluated the effects of demographic, clinical, symptom, and psychosocial adjustment characteristics on the initial (preoperative) levels of FCR and trajectories of FCR over 6 months following surgery. METHODS: Prior to and for 6 months following breast cancer surgery, 396 women were assessed for demographic and clinical (disease and treatment) characteristics, symptoms, psychological adjustment characteristics, and quality of life (QOL). FCR was assessed using a four-item subscale from the QOL instrument. Hierarchical linear modeling was used to examine changes in FCR scores and to identify predictors of inter-individual differences in preoperative FCR levels and trajectories over 6 months. RESULTS: From before surgery to 6 months post-operatively, women with breast cancer showed a high degree of inter-individual variability in FCR. Preoperatively, women who lived with someone, experienced greater changes in spiritual life, had higher state anxiety, had more difficulty coping, or experienced more distress due to diagnosis or distress to family members reported higher FCR scores. Patients who reported better overall physical health and higher FCR scores at enrollment demonstrated a steeper decrease in FCR scores over time. CONCLUSIONS: These findings highlight inter-individual heterogeneity in initial levels and changes in FCR over time among women undergoing breast cancer surgery. Further work is needed to identify and provide interventions for women experiencing FCR during and after breast cancer treatment.
Assuntos
Neoplasias da Mama/psicologia , Medo/psicologia , Recidiva Local de Neoplasia/psicologia , Sobreviventes/psicologia , Adaptação Psicológica , Transtornos de Ansiedade/etiologia , Transtornos de Ansiedade/psicologia , Neoplasias da Mama/cirurgia , Feminino , Necessidades e Demandas de Serviços de Saúde , Humanos , Pessoa de Meia-Idade , Período Pós-Operatório , Qualidade de VidaRESUMO
Investigators commonly gather longitudinal data to assess changes in responses over time and to relate these changes to within-subject changes in predictors. With rare or expensive outcomes such as uncommon diseases and costly radiologic measurements, outcome-dependent, and more generally outcome-related, sampling plans can improve estimation efficiency and reduce cost. Longitudinal follow up of subjects gathered in an initial outcome-related sample can then be used to study the trajectories of responses over time and to assess the association of changes in predictors within subjects with change in response. In this article, we develop two likelihood-based approaches for fitting generalized linear mixed models (GLMMs) to longitudinal data from a wide variety of outcome-related sampling designs. The first is an extension of the semi-parametric maximum likelihood approach developed in Neuhaus, Scott and Wild (2002, Biometrika 89, 23-37) and Neuhaus, Scott and Wild (2006, Biometrics 62, 488-494) and applies quite generally. The second approach is an adaptation of standard conditional likelihood methods and is limited to random intercept models with a canonical link. Data from a study of attention deficit hyperactivity disorder in children motivates the work and illustrates the findings.
Assuntos
Interpretação Estatística de Dados , Funções Verossimilhança , Estudos Longitudinais/métodos , Modelos Estatísticos , Resultado do Tratamento , Transtorno do Deficit de Atenção com Hiperatividade/etiologia , Criança , Simulação por Computador , HumanosRESUMO
Generalized linear mixed models with random intercepts and slopes provide useful analyses of clustered and longitudinal data and typically require the specification of the distribution of the random effects. Previous work for models with only random intercepts has shown that misspecifying the shape of this distribution may bias estimates of the intercept, but typically leads to little bias in estimates of covariate effects. Very few papers have examined the effects of misspecifying the joint distribution of random intercepts and slopes. However, simulation results in a recent paper suggest that misspecifying the shape of the random slope distribution can yield severely biased estimates of all model parameters. Using analytic results, simulation studies and fits to example data, this paper examines the bias in parameter estimates due to misspecification of the shape of the joint distribution of random intercepts and slopes. Consistent with results for models with only random intercepts, and contrary to the claims of severe bias in a recent paper, we show that misspecification of the joint distribution typically yields little bias in estimates of covariate effects and is restricted to covariates associated with the misspecified random effects distributions. We also show that misspecification of the distribution of random effects has little effect on confidence interval performance. Coverage rates based on the model-based standard errors from fitted likelihoods were generally quite close to nominal.
Assuntos
Modelos Lineares , Análise de Variância , Viés , Bioestatística , Análise por Conglomerados , Intervalos de Confiança , Interpretação Estatística de Dados , Humanos , Funções Verossimilhança , Estudos LongitudinaisRESUMO
BACKGROUND: Recent endeavors emphasize the importance of understanding early barriers to liver transplantation (LT) by consistently collecting data on patient demographics, socioeconomic factors, and geographic social deprivation indices. METHODS: In this retrospective single-center cohort study of 1657 adults referred for LT evaluation, we assessed the association between community-level vulnerability and individual socioeconomic status measures on the rate of waitlisting and transplantation. Patients' addresses were linked to Social Vulnerability Index (SVI) at the census tract-level to characterize community-level vulnerability. Descriptive statistics were used to describe patient characteristics. Multivariable cause-specific HRs were used to assess the association between community-level vulnerability, individual measures of the socioeconomic status, and LT evaluation outcomes (waitlist and transplantation). RESULTS: Among the 1657 patients referred for LT during the study period, 54% were waitlisted and 26% underwent LT. A 0.1 increase in overall SVI correlated with an 8% lower rate of waitlisting (HR 0.92, 95% CI 0.87-0.96, p < 0.001), with socioeconomic status, household characteristics, housing type and transportation, and racial and ethnic minority status domains contributing significantly to this association. Patients residing in more vulnerable communities experienced a 6% lower rate of transplantation (HR 0.94, 95% CI 0.91- 0.98, p = 0.007), with socioeconomic status and household characteristic domain of SVI significantly contributing to this association. At the individual level, both government insurance and employment status were associated with lower rates of waitlisting and transplantation. There was no association with mortality prior to waitlisting or mortality while on the waitlist. CONCLUSION: Our findings indicate that both individual and community measures of the socioeconomic status (overall SVI) are associated with LT evaluation outcomes. Furthermore, we identified individual measures of neighborhood deprivation associated with both waitlisting and transplantation.
Assuntos
Transplante de Fígado , Adulto , Humanos , Estudos Retrospectivos , Estudos de Coortes , Etnicidade , Vulnerabilidade Social , Grupos Minoritários , Classe Social , Encaminhamento e ConsultaRESUMO
After introducing guidelines for breast cancer screening in 2003, Mexico began to prioritize the implementation of mammography screening nationally. Since then, there have been no studies assessing changes in mammography in Mexico using the two-year prevalence interval that corresponds to national guidelines for screening frequency. The present study analyzes the Mexican Health and Aging Study (MHAS), a national population-based panel study of adults aged 50 and older, to evaluate changes in 2-year mammography prevalence among women aged 50 to 69 across five survey waves from 2001 to 2018 (n = 11,773). We calculated unadjusted and adjusted mammography prevalence by survey year and health insurance type. Overall prevalence increased substantially from 2003 to 2012 and leveled off in the period from 2012 to 2018 (2001: 20.2 % [95 % CI 18.3, 22.1]; 2003: 22.7 % [20.4, 25.0]; 2012: 56.5 % [53.2, 59.7]; 2015: 62.0 % [58.8, 65.2]; 2018: 59.4 % [56.7,62.1]; unadjusted prevalence). Prevalence was higher among respondents with social security insurance, who are more likely to work in the formal economy, than among respondents without social security, who are more likely to work in the informal economy or be unemployed. The overall prevalence estimates observed were higher than previously published estimates of mammography prevalence in Mexico. More research is needed to confirm findings regarding two-year mammography prevalence in Mexico and to better understand the causes of observed disparities.
RESUMO
PURPOSE: To determine whether magnetic resonance (MR) imaging and MR spectroscopic imaging findings can improve predictions made with the Kattan nomogram for radiation therapy. MATERIALS AND METHODS: The institutional review board approved this retrospective HIPAA-compliant study. Ninety-nine men who underwent endorectal MR and MR spectroscopy before external-beam radiation therapy for prostate cancer (January 1998 to June 2007) were included. Linear predictors were calculated with input variables from the study sample and the Kattan original coefficients. The linear predictor is a single weighted value that combines information of all predictor variables in a model, where the weight of each value is its association with the outcome. Two radiologists independently reviewed all MR images to determine extent of disease; a third independent reader resolved discrepancies. Biochemical failure was defined as a serum prostate-specific antigen level of 2 ng/mL (2 µg/L) or more above nadir. Cox proportional hazard models were used to determine the probabilities of treatment failure (biochemical failure) in 5 years. One model included only the Kattan nomogram data; the other also incorporated imaging findings. The discrimination performance of all models was determined with receiver operating characteristics (ROC) curve analyses. These analyses were followed by an assessment of net risk reclassification. RESULTS: The areas under the ROC curve for the Kattan nomogram and the model incorporating MR imaging findings were 61.1% (95% confidence interval: 58.1%, 64.0%) and 78.0% (95% confidence interval: 75.7%, 80.4%), respectively. Comparison of performance showed that the model with imaging findings performed significantly better than did the model with clinical variables alone (P < .001). Overall, the addition of imaging findings led to an improvement in risk classification of about 28%, ranging from approximately a minimum of 16% to a maximum of 39%, depending on the risk change considered important. CONCLUSION: MR imaging data improve the prediction of biochemical failure with the Kattan nomogram after external-beam radiation therapy for prostate cancer. The number needed to image to improve the prediction of biochemical failure in one patient ranged from three to six.
Assuntos
Imageamento por Ressonância Magnética/métodos , Espectroscopia de Ressonância Magnética/métodos , Neoplasias da Próstata/radioterapia , Idoso , Algoritmos , Progressão da Doença , Humanos , Masculino , Pessoa de Meia-Idade , Nomogramas , Valor Preditivo dos Testes , Modelos de Riscos Proporcionais , Antígeno Prostático Específico/sangue , Neoplasias da Próstata/sangue , Neoplasias da Próstata/patologia , Curva ROC , Estudos Retrospectivos , Taxa de Sobrevida , Carga TumoralRESUMO
Statistical models that include random effects are commonly used to analyze longitudinal and correlated data, often with the assumption that the random effects follow a Gaussian distribution. Via theoretical and numerical calculations and simulation, we investigate the impact of misspecification of this distribution on both how well the predicted values recover the true underlying distribution and the accuracy of prediction of the realized values of the random effects. We show that, although the predicted values can vary with the assumed distribution, the prediction accuracy, as measured by mean square error, is little affected for mild-to-moderate violations of the assumptions. Thus, standard approaches, readily available in statistical software, will often suffice. The results are illustrated using data from the Heart and Estrogen/Progestin Replacement Study using models to predict future blood pressure values.
Assuntos
Algoritmos , Artefatos , Biometria/métodos , Interpretação Estatística de Dados , Modelos Lineares , Simulação por Computador , Métodos EpidemiológicosRESUMO
Litière, Alonso, and Molenberghs (2007, Biometrics, 63, 1038-1044) presented the results of simulation studies that they claimed showed that misspecification of the shape of the random effects distribution can produce marked increases in Type II error (decreases in power) of tests based on fits of generalized linear mixed models. However, the article contains a logical fallacy that invalidates this claim. We present logically correct simulation studies that demonstrate little increase in Type II error, consistent with the earlier work that shows little effect due to misspecification.
Assuntos
Algoritmos , Artefatos , Biometria/métodos , Interpretação Estatística de Dados , Modelos Lineares , Modelos BiológicosRESUMO
Genetic epidemiologists often gather outcome-dependent samples of family data to measure within-family associations of genetic factors with disease outcomes. Generalized linear mixed models provide effective methods to estimate within-family associations but typically require parametric specification of the random effects distribution. Although misspecification of the random effects distribution often leads to little bias in estimated regression coefficients in standard, prospective clustered data settings, some recent studies suggest that such misspecification will impact parameter estimates from outcome-dependent cluster sampling designs. Using analytic results, simulation studies and fits to example data, this study examines the effect of misspecification of random effects distributions on parameter estimates in clustered data settings with outcome-dependent sampling. We show that the effects are consistent with results from prospective cluster sampling settings. In particular, ascertainment corrected mixed model methods that assume normally distributed random intercepts and conditional likelihood approaches provide accurate estimates of within-family covariate effects even under a misspecified random effects distribution.
RESUMO
BACKGROUND: Hispanic subjects with cystic fibrosis (CF) have increased morbidity and mortality than non-Hispanic white subjects. The ethnic disparity in mortality varies by region. Factors influencing pulmonary function vary by both ethnicity and region. OBJECTIVE: To determine if the ethnic difference in pulmonary function varies by region. METHODS: This retrospective cohort study compared differences in longitudinal pulmonary function (percent-predicted forced vital capacity [FVC], forced expiratory volume in 1 second [FEV1 ], forced expiratory flow at 25% to 75% [FEF25-75 ], FEV1 /FVC, and FEV1 decline) between Hispanic and non-Hispanic white subjects with CF by Census region of the United States (West, South, Midwest, and Northeast). Subjects were of ages 6 to 25 years and in the CF Foundation Patient Registry from 2008 to 2013. We used linear mixed effects models with subject-specific slopes and intercepts, adjusting for 14 demographic and clinical variables. RESULTS: Of 14 932 subjects, 1433 (9.6%) were Hispanic and 13 499 (90.4%) were non-Hispanic white. Hispanic subjects' FEV1 was 9.0% (8.3%-9.8%) lower than non-Hispanic white subjects in the West, while Hispanic subjects' FEV1 was only 4.0% (3.0%-5.0%) lower in the Midwest, 4.4% (3.1%-5.7%) lower in the Northeast, and 4.4% (3.2%-5.5%) lower in the South. Similarly, FVC and FEF25-75 were lower among Hispanic subjects compared to non-Hispanic white subjects in all US regions, with the biggest differences in the West. Only in the West was FEV1 /FVC significantly lower in Hispanic subjects (-0.019; -0.022 to -0.015). FEV1 decline was not significantly different between ethnicities in any region. CONCLUSIONS: In CF, Hispanic subjects have lower pulmonary function than non-Hispanic white subjects in all geographic regions with the largest difference in occurring in the West.
Assuntos
Hispânico ou Latino , Pulmão/fisiopatologia , Adolescente , Adulto , Criança , Fibrose Cística/etnologia , Fibrose Cística/fisiopatologia , Feminino , Volume Expiratório Forçado , Humanos , Modelos Lineares , Masculino , Testes de Função Respiratória , Estudos Retrospectivos , Estados Unidos , Capacidade Vital , População Branca , Adulto JovemRESUMO
This quality improvement study identifies adverse events for inferior vena cava filters and reports changes in adverse event reporting and estimated insertions between 2016 and 2020 in the US.
Assuntos
Embolia Pulmonar , Filtros de Veia Cava , Humanos , Filtros de Veia Cava/efeitos adversos , Fatores de Risco , Remoção de DispositivoRESUMO
OBJECTIVE: To examine the association between adolescents' perception of clinician communication and adolescents' reported acceptability of the steps involved in chlamydial screening during urgent care visits. DESIGN: Cross-sectional survey of adolescents after urgent care visits. SETTING: Four pediatric clinics in a health maintenance organization. PARTICIPANTS: Three hundred sixty-five adolescents aged 13 to 18 years. OUTCOME MEASURES: Participants' ratings of the acceptability of talking about sexual health and providing a urine sample for chlamydial testing in an urgent care visit. RESULTS: Most adolescents found sexual health discussions and urine collection for chlamydial screening acceptable in the urgent care setting (84% and 80%, respectively). Acceptability of sexual health discussion was significantly associated with adolescents' perception that the clinician explained confidentiality (adjusted odds ratio [AOR], 2.7; 95% confidence interval [CI], 1.3-5.5), knew "how to talk to teens like me" (AOR, 9.0; 95% CI, 3.5-24.2), and "listened carefully as I explained my concerns" (AOR, 14.3; 95% CI, 4.3-54.9). Acceptability of providing a urine sample for chlamydial testing was associated with the adolescents' perception that the clinician knew "how to talk to teens like me" (AOR, 3.7; 95% CI, 1.5-9.3) and "listened carefully as I explained my concerns" (AOR, 3.6; 95% CI, 1.1-11.5). CONCLUSIONS: Sexual history taking and urine collection are 2 key components of chlamydial screening and were reported as acceptable by the great majority of adolescents in the urgent care setting. Aspects of clinician communication appear to be important target areas for pediatric clinician education in supporting expansion of chlamydial screening to adolescents in urgent care visits.