PURPOSE:We examined whether increasing severe maternal morbidity (SMM) rates are driven by changes in maternal age distribution or age-specific SMM rates. METHODS:In a retrospective cohort study across three states, we analyzed data from two time points (2008-2009 to 2019-2020). We used Kitagawa decomposition analysis to determine the contributions of changes in maternal age distribution and age-specific SMM rates to SMM rates both with and without transfusion during birth and up to 1 year postpartum, stratified by race/ethnicity. We examined the following racial/ethnic groups: non-Hispanic white, non-Hispanic Black, Hispanic, and non-Hispanic Asian/Pacific Islander. RESULTS:Between 2008-2009 and 2019-2020, SMM and non-transfusion SMM rates increased from 213.6 to 260.5 and from 109.6 to 154.8 per 10,000 births, respectively. Across all racial and ethnic groups, the proportion of younger birthing individuals (<25 years) decreased and the proportion of older individuals (≥30 years) increased. The decomposition analysis showed that increases in SMM and non-transfusion SMM were primarily due to increases in age-specific SMM rates (100.0% and 94.6%, respectively), particularly among younger birthing individuals. Changes in maternal age distribution had a minimal overall contribution. However, when stratified by race/ethnicity, changes in maternal age distribution had a greater contribution to SMM (28.9%) and non-transfusion SMM (22.7%) rates among non-Hispanic Black individuals, with little to no contribution observed in the other groups. CONCLUSION:The increasing rates of birth-related and postpartum SMM rates stem from increasing rates of SMM in every age group rather than shifts in maternal age distribution.

Objective: To examine the correlations between pairs of maternal, infant, and maternal-infant dyad quality measures to provide a comprehensive assessment of perinatal care. Study Design: In a retrospective cohort study using birth and fetal death certificates linked to hospital discharge data from Michigan, Oregon, Pennsylvania, and South Carolina (2016-2018), we examined correlations between pairs of maternal, infant, and maternal-infant dyad quality measures. Maternal quality measures included nulliparous term singleton vertex (NTSV) cesarean birth, non-transfusion severe maternal morbidity (SMM), and a composite maternal outcome. Infant quality was assessed with a composite outcome measure, while the dyad measure combined maternal and infant outcomes. Results: Among 955,904 dyads across 266 hospitals, 25.9% had NTSV, 0.7% had non-transfusion SMM, 12.3% had the composite infant measure, and 19.3% had the dyad measure. The correlation between non-transfusion SMM and the dyad measure was 0.12 while the correlation between the composite infant measure and the dyad measure was 0.86 which was higher than the correlation between the composite maternal measure and the dyad measure (0.47). Conclusion: We observed minimal correlations among these perinatal quality measures, especially when aggregated beyond individual outcomes.

Importance:Few studies have investigated the association of composite measures of neighborhood social determinants of health with severe maternal morbidity (SMM), and no research has examined this association for indices tailored to maternal health. Objective:To examine the association of scores in the Maternal Vulnerability Index (MVI), a tool developed to measure maternal risk of adverse health outcomes, with SMM. Design, Setting, and Participants:This retrospective, population-based cohort study was conducted in 5 states (2008-2020 for Michigan, Oregon, and South Carolina; 2008-2018 for Pennsylvania; and 2008-2012 for California) among individuals delivering a fetal death or a live birth between 22 and 44 weeks. Analysis was conducted between August and October 2024. Exposure:The MVI, a composite measure of 43 area-level indicators, was categorized into 6 themes encompassing physical, social, and health care environments. MVI score and themes were examined in quartiles (quartile 1 = lowest risk to quartile 4 = highest risk) based on residential zip code tabulation area. Main Outcomes and Measures:SMM during delivery hospitalization and after discharge within 42 days after delivery. Results:Among 6 543 255 birthing individuals (3 568 631 ages 25-34 years [54.5%]; 472 145 Asian or Pacific Islander [7.2%], 824 239 Black [12.6%], 1 673 917 Hispanic [25.6%], and 3 346 807 White [51.2%]), there were 1 087 936 individuals in MVI quartile 1 (16.6%) and 1 376 658 individuals in MVI quartile 4 (21.0%). A total of 45 051 individuals (0.7%) had SMM during delivery hospitalization, while 13 534 individuals (0.2%) had SMM after discharge within 42 days after delivery. In adjusted analyses, there were no associations between MVI score or themes and SMM during delivery hospitalization. However, a dose-response association was observed between MVI score and SMM within 42 days after delivery (second MVI quartile: adjusted relative risk [aRR], 1.03; 95% CI, 0.95-1.11; third MVI quartile: aRR, 1.12; 95% CI, 1.03-1.23; fourth MVI quartile: aRR, 1.27; 95% CI, 1.14-1.41). The highest MVI quartile in themes of general health care (aRR, 1.27; 95% CI, 1.14-1.43), physical environment (aRR, 1.33; 95% CI, 1.22-1.46), physical health (aRR, 1.23; 95% CI, 1.12-1.35), reproductive health care (aRR, 1.30; 95% CI, 1.15-1.47), and socioeconomic determinants (aRR, 1.19; 95% CI, 1.02-1.39) was associated with SMM within 42 days after delivery. A dose-response association was observed between all MVI themes and SMM within 42 days after delivery (eg, physical environment MVI theme second quartile: aRR, 1.04; 95% CI, 0.96-1.13; third quartile: aRR, 1.14; 95% CI, 1.05-1.25; fourth quartile: aRR, 1.33; 95% CI, 1.22-1.46), except for the mental health and general health care themes. Conclusions and Relevance:In this study, MVI score was not associated with SMM during delivery but was associated with postpartum SMM, suggesting that MVI may capture long-term risks more effectively than acute conditions during delivery hospitalization.

Importance:As a result of consolidation in the health care delivery system, most very preterm infants in the US are born and receive care in multihospital health systems. The extent of variation in patient outcomes and length of stay for this vulnerable population across health systems and across hospitals within systems is not known. Objective:To evaluate the extent of variation in mortality and length of stay within and across health systems for infants born very preterm (gestational age 24-29 weeks). Design, Setting, and Participants:This cross-sectional study examined data contributed by Vermont Oxford Network US member hospitals in 224 health systems that delivered care to very preterm infants born between January 1, 2021, and December 31, 2022. Exposure:Receipt of neonatal intensive care unit (NICU) care in a horizontally integrated multihospital health system. Main Outcomes and Measures:Mortality rates and length of stay among surviving infants were estimated using multilevel logistic and linear models. Results:The sample included 38 501 infants (median [IQR] gestational age, 27 [26-28] weeks; 52.8% boys). The median (IQR) number of infants receiving care at a hospital system during the 2-year period was 108 (59-198); 91.0% were born at the reporting hospital, and 95.4% were born in the reporting system. The mean adjusted mortality rate in the highest performing quartile of systems was 7.8% (95% credible interval [CrI], 7.3%-8.3%) compared with 9.8% (95% CrI, 9.1%-10.7%) for the lowest performing quartile. The mean adjusted length of stay for surviving infants ranged from 78 days (95% CrI, 77-79 days) to 90 days (95% CrI, 88-91 days) between the highest and lowest performing quartiles of systems, respectively. Conclusions and Relevance:In this cross-sectional study of very preterm infants, there was a 2-percentage point difference in mortality between systems in the highest and lowest performing quartiles and a 12-day difference in mean length of stay among surviving infants, which are potentially clinically meaningful. Opportunities exist for health systems to improve quality at the health system level to decrease mortality among infants born very preterm and reduce resources used in patient care.

Importance:Minoritized racial and ethnic groups, such as American Indian and Black individuals, often receive lower quality health care compared with White individuals. There is limited understanding of how these disparities extend to obstetric care, particularly when comparing the quality of care at the actual delivery hospital vs the nearest obstetric hospital based on the birthing individual's residence. Objective:To examine inequality in care based on the actual delivery hospital and the closest delivery hospital to the birthing individual's residential zip code centroid. Design, Setting, and Participants:This population-based retrospective cohort study used data from 5 states (2008 to 2020 for Michigan, Oregon, and South Carolina; 2008 to 2018 for Pennsylvania; and 2008 to 2012 for California). Individuals delivering a fetal death or a live birth with gestational age between 22 to 44 weeks were included. Analysis was conducted between February and August 2024. Exposure:Race and ethnicity. Main Outcomes and Measures:The obstetric inequality index was calculated using Gini coefficients from Lorenz curves for American Indian, Asian, Black, and Hispanic birthing individuals compared with White individuals, with hospitals ranked by their standardized morbidity ratio for nontransfusion severe maternal morbidity. Results:There were 6 418 635 birthing individuals across 549 hospitals (23 050 American Indian individuals [0.4%], 463 342 Asian individuals [7.2%], 807 738 Black individuals [12.6%], 1 645 922 Hispanic individuals [25.6%], and 3 279 315 White individuals [51.1%]). Compared with White individuals, American Indian and Black individuals delivered at lower-quality hospitals, while there was no significant difference for Asian and Hispanic individuals (delivery hospital inequality index: American Indian, 0.07 [95% CI, 0.03 to 0.11]; Asian, -0.02 [95% CI, -0.08 to 0.04]; Black, 0.15 [95% CI, 0.12 to 0.19]; Hispanic -0.04 [95% CI, -0.09 to 0.01]). Black individuals lived closer to lower-quality hospitals than White individuals (closest hospital inequality index for Black individuals: 0.11 [95% CI, 0.07 to 0.14]). Asian and Hispanic individuals had similar closest hospital inequality indices to White individuals. The inequality index for Black individuals would have been lower if individuals had delivered at their nearest hospital. Conclusions and Relevance:This cohort study found that American Indian and Black individuals delivered at lower-quality hospitals than White individuals. The disparity in care between Black and White birthing individuals would have been reduced if individuals had delivered at their nearest hospital.

Severe maternal morbidity (SMM) is a significant complication associated with preterm delivery. However, most studies have focused solely on SMM during the delivery hospitalization, without differentiating by preterm birth subtype or examining postpartum SMM and readmissions. To examine the association between preterm birth subtypes and SMM during delivery hospitalization, up to 1-year postpartum, and postpartum readmissions within 365 days. We conducted a retrospective cohort study using linked birth and fetal death certificates and maternal hospital discharge data from Michigan, Oregon, and South Carolina (2008–2020). Modified Poisson regression models were used to estimate adjusted relative risks (aRR) and 95

PURPOSE:We examined the association between iron deficiency anemia (IDA) and severe maternal morbidity (SMM) during delivery and up to 1-year postpartum. METHODS:In a retrospective cohort study across 3 states, we computed adjusted relative risks (aRR) for SMM comparing individuals with IDA versus those without, using modified Poisson regression models. RESULTS:Among 2459,106 individuals, 10.3 % (n = 252,240) had IDA. Individuals with IDA experienced higher rates of blood transfusion and non-transfusion SMM (329 and 122 per 10,000 deliveries, respectively) than those without IDA (33 and 46 per 10,000 deliveries, respectively). The risk of blood transfusion (aRR: 8.2; 95 % CI 7.9-8.5) and non-transfusion SMM (aRR: 1.9; 95 % CI: 1.8-2.0) were higher among individuals with IDA. The attributable risk per 10,000 deliveries due to IDA for blood transfusion and non-transfusion SMM during delivery were 29.5 (95 % CI: 28.9-30.0) and 5.7 (95 % CI: 5.3-6.2), respectively. Within 1-year postpartum, the relative risk of non-transfusion SMM (aRR:1.3; 95 % CI: 1.2-1.3) was 30 % higher among individuals with IDA. CONCLUSION:IDA is associated with increased SMM risk. Addressing IDA in pregnant individuals may reduce SMM rates.

BACKGROUND:Few recent studies have examined the rate of severe maternal morbidity occurring during the antenatal and/or postpartum period to 42 days after delivery. However, little is known about the rate of severe maternal morbidity occurring beyond 42 days after delivery. OBJECTIVE:This study aimed to examine the distribution of severe maternal morbidity and its indicators during antenatal, delivery, and postpartum hospitalizations to 365 days after delivery and to estimate the increase in severe maternal morbidity rate and its indicators after accounting for antenatal and postpartum severe maternal morbidity to 365 days after delivery. STUDY DESIGN:This was a retrospective cohort study using birth and fetal death certificate data linked to hospital discharge records from Michigan, Oregon, and South Carolina from 2008 to 2020. This study examined the distribution of severe maternal morbidity, nontransfusion severe maternal morbidity, and severe maternal morbidity indicators during antenatal, delivery, and postpartum hospitalizations to 365 days after delivery. Subsequently, this study examined "severe maternal morbidity cases added," which represent cases among unique individuals that are included by considering the antenatal and postpartum periods but that would be missed if only the delivery hospitalization cases were included. RESULTS:A total of 64,661 (2.5%) individuals experienced severe maternal morbidity, whereas 37,112 (1.4%) individuals experienced nontransfusion severe maternal morbidity during antenatal, delivery, and/or postpartum hospitalization. A total of 31% of severe maternal morbidity cases were added after accounting for severe maternal morbidity occurring during the antenatal or postpartum hospitalization to 365 days after delivery, whereas 49% of nontransfusion severe maternal morbidity cases were added after accounting for nontransfusion severe maternal morbidity occurring during the antenatal or postpartum periods. Severe maternal morbidity occurring between 43 and 365 days after delivery contributed to 12% of all severe maternal morbidity cases, whereas nontransfusion severe maternal morbidity occurring between 43 and 365 days after delivery contributed to 19% of all nontransfusion severe maternal morbidity cases. CONCLUSION:Our study showed that a total of 31% of severe maternal morbidity and 49% of nontransfusion severe maternal morbidity cases were added after accounting for severe maternal morbidity occurring during the antenatal or postpartum hospitalization to 365 days after delivery. Our findings highlight the importance of expanding the severe maternal morbidity definition beyond the delivery hospitalization to better capture the full period of increased risk, identify contributing factors, and design strategies to mitigate this risk. Only then can we improve outcomes for mothers and subsequently the quality of life of their infants.

Importance Little is known about the association between sickle cell disease (SCD) and severe maternal morbidity (SMM).Objective To examine the association of SCD with racial disparities in SMM and with SMM among Black individuals.Design, Setting, and Participants This cohort study was a retrospective population-based investigation of individuals with and without SCD in 5 states (California [2008-2018], Michigan [2008-2020], Missouri [2008-2014], Pennsylvania [2008-2014], and South Carolina [2008-2020]) delivering a fetal death or live birth. Data were analyzed between July and December 2022.Exposure Sickle cell disease identified during the delivery admission by using International Classification of Diseases, Ninth Revision and Tenth Revision codes.Main Outcomes and Measures The primary outcomes were SMM including and excluding blood transfusions during the delivery hospitalization. Modified Poisson regression was used to estimate risk ratios (RRs) adjusted for birth year, state, insurance type, education, maternal age, Adequacy of Prenatal Care Utilization Index, and obstetric comorbidity index.Results From a sample of 8 693 616 patients (mean [SD] age, 28.5 [6.1] years), 956 951 were Black individuals (11.0%), of whom 3586 (0.37%) had SCD. Black individuals with SCD vs Black individuals without SCD were more likely to have Medicaid insurance (70.2% vs 64.6%), to have a cesarean delivery (44.6% vs 34.0%), and to reside in South Carolina (25.2% vs 21.5%). Sickle cell disease accounted for 8.9% and for 14.3% of the Black-White disparity in SMM and nontransfusion SMM, respectively. Among Black individuals, SCD complicated 0.37% of the pregnancies but contributed to 4.3% of the SMM cases and to 6.9% of the nontransfusion SMM cases. Among Black individuals with SCD compared with those without, the crude RRs of SMM and nontransfusion SMM during the delivery hospitalization were 11.9 (95% CI, 11.3-12.5) and 19.8 (95% CI, 18.5-21.2), respectively, while the adjusted RRs were 3.8 (95% CI, 3.3-4.5) and 6.5 (95% CI, 5.3-8.0), respectively. The SMM indicators that incurred the highest adjusted RRs included air and thrombotic embolism (4.8; 95% CI, 2.9-7.8), puerperal cerebrovascular disorders (4.7; 95% CI, 3.0-7.4), and blood transfusion (3.7; 95% CI, 3.2-4.3).Conclusions and Relevance In this retrospective cohort study, SCD was found to be an important contributor to racial disparities in SMM and was associated with an elevated risk of SMM among Black individuals. Efforts from the research community, policy makers, and funding agencies are needed to advance care among individuals with SCD.

BACKGROUND:Severe maternal morbidity has been increasing in the past few decades. Few studies have examined the risk of severe maternal morbidity among individuals with stillbirths vs individuals with live-birth deliveries. OBJECTIVE:This study aimed to examine the prevalence and risk of severe maternal morbidity among individuals with stillbirths vs individuals with live-birth deliveries during delivery hospitalization as a primary outcome and during the postpartum period as a secondary outcome. STUDY DESIGN:This was a retrospective cohort study using birth and fetal death certificate data linked to hospital discharge records from California (2008-2018), Michigan (2008-2020), Missouri (2008-2014), Pennsylvania (2008-2014), and South Carolina (2008-2020). Relative risk regression analysis was used to examine the crude and adjusted relative risks of severe maternal morbidity along with 95% confidence intervals among individuals with stillbirths vs individuals with live-birth deliveries, adjusting for birth year, state of residence, maternal sociodemographic characteristics, and the obstetric comorbidity index. RESULTS:Of the 8,694,912 deliveries, 35,012 (0.40%) were stillbirths. Compared with individuals with live-birth deliveries, those with stillbirths were more likely to be non-Hispanic Black (10.8% vs 20.5%); have Medicaid (46.5% vs 52.0%); have pregnancy complications, including preexisting diabetes mellitus (1.1% vs 4.3%), preexisting hypertension (2.3% vs 6.2%), and preeclampsia (4.4% vs 8.4%); have multiple pregnancies (1.6% vs 6.2%); and reside in South Carolina (7.4% vs 11.6%). During delivery hospitalization, the prevalence rates of severe maternal morbidity were 791 cases per 10,000 deliveries for stillbirths and 154 cases per 10,000 deliveries for live-birth deliveries, whereas the prevalence rates for nontransfusion severe maternal morbidity were 502 cases per 10,000 deliveries for stillbirths and 68 cases per 10,000 deliveries for live-birth deliveries. The crude relative risk for severe maternal morbidity was 5.1 (95% confidence interval, 4.9-5.3), whereas the adjusted relative risk was 1.6 (95% confidence interval, 1.5-1.8). For nontransfusion severe maternal morbidity among stillbirths vs live-birth deliveries, the crude relative risk was 7.4 (95% confidence interval, 7.0-7.7), whereas the adjusted relative risk was 2.0 (95% confidence interval, 1.8-2.3). This risk was not only elevated among individuals with stillbirth during the delivery hospitalization but also through 1 year after delivery (severe maternal morbidity adjusted relative risk, 1.3; 95% confidence interval, 1.1-1.4; nontransfusion severe maternal morbidity adjusted relative risk, 1.2; 95% confidence interval, 1.1-1.3). CONCLUSION:Stillbirth was found to be an important contributor to severe maternal morbidity.

PurposeThis paper examines how U.S. consumer intentions to adopt hemp vary across product types using the theory of planned behavior (TPB).Design/methodology/approachData were collected via an online survey of U.S. residents in 2022 (n = 1,948). Two-step structural equation modeling is used to examine how TPB constructs and background factors influence intent to use five different hemp-based products: cannabidiol (CBD), clothing, food, personal care products, and pet products. Data are analyzed using R.FindingsPositive attitudes towards all categories of hemp-based products increase the probability of adoption, while subjective norm and perceived behavioral control have limited and varied significant influence across product models. Age has a consistent significant and negative influence on adoption.Research limitations/implicationsFindings highlight consumer segmentation and marketing opportunities, inform hemp stakeholder decision-making, and provide directions for future research. Given the absence of explanatory power of SN and PBC on most product models and the diversity of products and nuanced U.S. hemp policy, future research could investigate expanded iterations of TPB. Using revealed behavior could also highlight potential intention-behavior gaps and offer more robust insights for hemp stakeholders.Originality/valueFindings contribute to a limited body of information on markets and consumer demand for hemp in the U.S.

BACKGROUND Mortality and morbidity for very preterm infants in the United States decreased for years. The current study describes recent changes to assess whether the pace of improvement has changed. METHODS Vermont Oxford Network members contributed data on infants born at 24 to 28 weeks’ gestation from 1997 to 2021. We modeled mortality, late-onset sepsis, necrotizing enterocolitis, chronic lung disease, severe intraventricular hemorrhage, severe retinopathy of prematurity, and death or morbidity by year of birth using segmented relative risk regression, reporting risk-adjusted annual percentage changes with 95% confidence intervals overall and by gestational age week. RESULTS Analyses of data for 447 396 infants at 888 hospitals identified 3 time point segments for mortality, late onset sepsis, chronic lung disease, severe intraventricular hemorrhage, severe retinopathy of prematurity, and death or morbidity, and 4 for necrotizing enterocolitis. Mortality decreased from 2005 to 2021, but more slowly since 2012. Late-onset sepsis decreased from 1997 to 2021, but more slowly since 2012. Severe retinopathy of prematurity decreased from 2002 to 2021, but more slowly since 2011. Necrotizing enterocolitis, severe intraventricular hemorrhage, and death or morbidity were stable since 2015. Chronic lung disease has increased since 2012. Trends by gestational age generally mirror those for the overall cohort. CONCLUSIONS Improvements in mortality and morbidity have slowed, stalled, or reversed in recent years. We propose a 3-part strategy to regain the pace of improvement: research; quality improvement; and follow through, practicing social as well as technical medicine to improve the health and well-being of infants and families.

This paper explores connections between margin-based loss functions and consistency in binary classification and regression applications. It is shown that a large class of margin-based loss functions for binary classification/regression result in estimating scores equivalent to log-likelihood scores weighted by an even function. A simple characterization for conformable (consistent) loss functions is given, which allows for straightforward comparison of different losses, including exponential loss, logistic loss, and others. The characterization is used to construct a new Huber-type loss function for the logistic model. A simple relation between the margin and standardized logistic regression residuals is derived, demonstrating that all margin-based loss can be viewed as loss functions of squared standardized logistic regression residuals. The relation provides new, straightforward interpretations for exponential and logistic loss, and aids in understanding why exponential loss is sensitive to outliers. In particular, it is shown that minimizing empirical exponential loss is equivalent to minimizing the sum of squared standardized logistic regression residuals. The relation also provides new insight into the AdaBoost algorithm.

The US National Bioengineered Food Disclosure Standard mandates disclosures on foods containing genetically modified (GM) ingredients, while allowing voluntary non-GM claims to coexist in the marketplace. Using cross-sectional survey data, accounting for error term correlation and controlling for demographic variability, we describe the types of GM food labels seen by respondents and quantify the association of other label reading behaviors and demographic characteristics on seeing GM and/or non-GM labels. The percent of respondents seeing both contains and does not contain GM labels increased between 2018 and 2021, with more respondents seeing non-GM compared with contains GM labels each year. Estimated conditional probabilities reveal that time, other label reading behaviors, and attitudes are heterogeneously associated with seeing contains GM and non-GM labels.

We introduce simulated packing and cracking as a technique for evaluating partisan-gerrymandering measures. We apply it to historical congressional and legislative elections to evaluate four measures: partisan bias, declination, efficiency gap, and mean-median difference. While the efficiency gap recognizes simulated packing and cracking in a completely predictable manner (a fact that follows immediately from the efficiency gap's definition) and the declination does a very good job of recording simulated packing and cracking, we conclude that both of the other two measures record it poorly. This deficiency is especially notable given the frequent use of such measures in outlier analyses.

Because benthic foraminifera exhibit spatial heterogeneity, a number of replicates or multiple biological samples are necessary to estimate population densities. In this study, we empirically examine the efficacy of taking four or fewer replicates to differentiate among mean densities in location and time using p-values as a metric for strength of evidence against the null hypothesis of no difference in taxon density. For spatial analyses, four stations along a traverse with four replicates per station were compared with ANOVA within Mission Bay, Texas, using the four most abundant taxa. The p-values for comparing mean densities among stations increased markedly for all taxa, as the number of samples per station decreased from four to two. Using a test level of 0.05, four replicates per station resulted, on average, in significant differences for three of four taxa, three replicates distinguished two of four taxa, and two replicates detected only one difference. For temporal analyses, a single station was sampled in the Indian River Lagoon, Florida, seasonally over four years. Again, p-values increased markedly as the number of samples per station decreased. Using a test level of 0.05, both four- and three-replicate groups were found to separate mean densities among the four years for three of four taxa, two replicates distinguished one taxon, and use of only one replicate could not detect any difference in mean densities among the four years. Based on these and previous field results, we recommend at least four replicates per station for environmental monitoring. However, when examining mean densities within larger ecological entities such as biofacies, just one sample at each station along a single traverse containing four stations in each bay could delineate Mission, Copano, and Mesquite bays in Texas.

This chapter reviews features of survival models and survivial data in the absence of measurement error. Models for covariate measurement error are then considered, followed by a discussion of the effects of measurement error on parameter estimation. Several approaches to correcting for the effects of covariate measurement error are then described, including regression calibration, SIMEX (Simulation-extrapolation), likelihood methods and Bayesian methods. The chapter closes with a discussion and conclusions.

In this chapter, we take a brief tour of the development of measurement error strategies so as to appreciate the challenges facing analysts hoping to address measurement error in applications. To conduct sensible analysis for real world data which are commonly error-corrupted, it is critical to understand the impact of measurement error effects and develop correction adjustments for measurement error effects accordingly. In the literature, research on measurement error models may be categorized into three areas: (1) measurement error in covariates, (2) measurement error in the response variable, and (3) measurement error in both covariate and response variables. The discussion in this chapter will focus on the first category. In particular, we are interested in addressing the following questions: (1) How has research in measurement error evolved over time, and what are some essential findings? (2) What has been the impact of the research addressing covariate measurement error in applications? (3) How can the impact be amplified?