Mixed Models For Longitudinal Data Research Articles

The effects of cholecalciferol and afamelanotide on vitamin D levels in erythropoietic protoporphyria; a multicentre cohort study.

Erythropoietic protoporphyria (EPP) patients experience lifelong painful photosensitivity resulting in a lack of sunlight exposure. Previous studies have shown that 47-63% of EPP patients suffer from vitamin D deficiency and a high prevalence of osteoporosis. As of 2016 an effective treatment for EPP is available: the alpha-MSH analogue afamelanotide. So far studies on vitamin D levels in EPP have only investigated patients who were not treated with afamelanotide. To investigate the effects of afamelanotide treatment on vitamin D levels in EPP. A multi-centre observational cohort study, in adult patients with EPP from the Erasmus Medical Centre, the Netherlands and the University Hospital Düsseldorf, Germany. Routinely-collected vitamin D levels between 2005 and 2021 were used for analysis. Patient exposure to cholecalciferol or afamelanotide was categorized into four treatment groups; untreated, cholecalciferol, afamelanotide, and combined treatment. A linear mixed model for longitudinal data was applied to measure the effect of the treatment groups, compared to the untreated, on vitamin D levels. A total of 230 patients and 1774 vitamin D measurements were included. Prevalence of vitamin D deficiency remained high despite afamelanotide treatment: <50 nmol/l in 71.8% of patients, and severe deficiency <30 nmol/l in 48.1%. Afamelanotide treatment alone did not lead to a significant average increase in vitamin D levels (β:0.5, 95% Confidence Interval [CI]: -3.2 - 4.2). In contrast, cholecalciferol and combined therapy with afamelanotide, led to a significant increase in vitamin D levels (β:11.6, CI: 7.2-15.9 and β:15.2, CI: 12.3-18.1). Cholecalciferol remains essential for treatment of vitamin D deficiency in EPP, irrespective of new treatment options like afamelanotide. Afamelanotide treatment did not affect vitamin D levels. We suggest that future guidelines include continuous monitoring of vitamin D and prescription of cholecalciferol in all patients with EPP, including those treated with afamelanotide.

Novel cross-linked polysaccharide-polyelectrolyte hemostatic foam improves survival compared to CombatGauze in swine femoral artery hemorrhage model.

Uncontrolled hemorrhage is the leading cause of preventable death in combat and civilian trauma. Efficacious hemostatic agents in junctional hemorrhage can quell blood loss and improve survival. We hypothesized that a novel hemostatic foam of starch and chitosan would improve hemostasis, and thereby increase survival in a swine femoral artery hemorrhage model when compared to CombatGauze. A novel hemostatic foam of starch and chitosan was created and modified over the study period. 30 pigs (4 excluded) were assigned to treatment using either foam version 1 (FV1, n = 9) or 2 (FV2, n = 8), or CombatGauze (CG, n = 9) in a standard swine femoral artery hemorrhage model. Animals were observed for 150 minutes. Outcomes assessed included hemostasis, survival, post-treatment blood loss, IV fluid volume, and hemodynamic and laboratory trends. Hemostasis prior to 150 minutes was similar with 44.4%, 77.8%, and 50% of swine treated with CG, FV1 and FV2, respectively (Kaplan-Meyer and log-rank test [KM-LR] p > 0.05). Survival to 150 minutes was improved in swine treated with FV1 (100%) compared to CG (55.6%) (KM-LR p = 0.02). Survival was similar between FV1 and FV2 (75%) (KM-LR p > 0.05), and between CG and FV2 (KM-LR p > 0.05). Using mixed model for longitudinal data, MAP decreased significantly in CG- and FV2-treated swine, while there was no significant change in MAP in FV1-treated swine. Trends in lactic acid, hematocrit, platelets, INR, and TEG were more favorable for FV1 compared to CG. In this preclinical study of junctional hemorrhage, survival was improved in swine treated with version 1 of a novel chitosan/starch foam compared to CombatGauze. Trends in hemodynamics and laboratory data were also more favorable in the FV1-treated swine. This novel hemostatic foam may be an effective alternative to current hemostatic agents. not applicableStudy type: Original research: animal research.

Incomplete Antiretroviral Therapy Adherence Is Associated with Lower CD4-CD8 Ratio in Virally Suppressed Patients with HIV Infection in Mexico.

Suboptimal adherence to antiretroviral therapy (ART) in people with HIV, even during sustained viral suppression, is associated with persistent inflammation, immune activation, and coagulopathy. Persistently low CD4-CD8 Ratio has been also associated with residual inflammation, is a good predictor of increased risk of death and more widely available than inflammatory biomarkers. We tested the hypothesis that the CD4-CD8 Ratio is associated with ART adherence during periods of complete viral suppression. We used the Medication Possession Ratio based in pharmacy registries as measure of adherence and time-varying, routine care CD4 and CD8 measurements as outcome. We used a linear mixed model for longitudinal data, including fixed effects for sex, age, education, date of ART initiation, AIDS-related conditions, and baseline CD4 to model the outcome. In 988 adults with a median follow-up of 4.13 years, higher ART adherence was independently associated with a modest increase in CD4-CD8. For each increasing percentage point in adherence, the CD4-CD8 Ratio increased 0.000857 (95% confidence interval [CI] -0.000494 to 0.002209, p = .213731) in the first year after achieving viral suppression; 0.001057 (95% CI 0.000262-0.001853, p = .009160) in years 1 to 3; 0.000323 (95% CI -0.000448 to 0.001095, p = .411441) in years 3 to 5; and 0.000850 (95% CI 0.000272-0.001429, p = .003946) 5-10 years after achieving viral suppression. The magnitude of the effect of adherence over CD4-CD8 Ratios varied over time and by baseline CD4 count, with increasing adherence having a larger effect early after ART initiation in people with higher baseline CD4 (>500 cells/μL) and in later years in people with lower baseline CD4 count (≥200 cells/μL). Our findings expand on previous evidence suggesting that the benefits of optimal adherence to modern ART regimens goes beyond maintaining viral suppression. These results highlight the importance of including objective measurements of adherence as part of routine care, even in patients with complete HIV suppression over long-term follow-up.

Bayesian variable selection and estimation in multivariate skew-normal generalized partial linear mixed models for longitudinal data

This paper proposes a novel generalized partial linear mixed models (GPLMMs) with the random effects distribution distributed as a multivariate skew-normal distribution for discrete longitudinal data over time. Under the Bayesian framework, we extended the Bayesian version of adaptive lasso (least absolute shrinkage and selection operator) for linear regression to the Bayesian adaptive lasso with the least squares approximation (LSA) method for generalized partial linear mixed models. A hybrid algorithm combining the Gibbs sampler and the Metropolis-Hastings algorithm in conjunction with the Bayesian adaptive lasso with LSA is used to simultaneously obtain the Bayesian estimates of unknown parameters, smoothing function and random effects, as well as select important covariates in GPLMMs. Several simulation studies and a real example are presented to illustrate the proposed methodologies.

Extending multivariate Student's- semiparametric mixed models for longitudinal data with censored responses and heavy tails.

This article extends the semiparametric mixed model for longitudinal censored data with Gaussian errors by considering the Student's -distribution. This model allows us to consider a flexible, functional dependence of an outcome variable over the covariates using nonparametric regression. Moreover, the proposed model takes into account the correlation between observations by using random effects. Penalized likelihood equations are applied to derive the maximum likelihood estimates that appear to be robust against outlying observations with respect to the Mahalanobis distance. We estimate nonparametric functions using smoothing splines under an EM-type algorithm framework. Finally, the proposed approach's performance is evaluated through extensive simulation studies and an application to two datasets from acquired immunodeficiency syndrome clinical trials.

Estimation in multivariate linear mixed models for longitudinal data with multiple outputs: Application to PBCseq data analysis

In many biomedical studies or clinical trials, we have data with more than one response variable on the same subject repeatedly measured over time. In analyzing such data, we adopt a multivariate linear mixed-effects longitudinal model. On the other hand, in longitudinal data, we often find features that do not impact modeling the response variable and are eliminated from the study. In this paper, we consider the problem of simultaneous variable selection and estimation in a multivariate t linear mixed-effects model (MtLMM) for analyzing longitudinally measured multioutcome data. This work's motivation comes from a cohort study of patients with primary biliary cirrhosis. The interest is eliminating insignificant variables using the smoothly clipped and absolute deviation penalty function in the MtLMM. The proposed penalized model offers robustness and flexibility to accommodate fat tails. An expectation conditional maximization algorithm is employed for the computation of maximum likelihood estimates of parameters. The calculation of standard errors is affected by an information-based method. The methodology is illustrated by analyzing Mayo Clinic Primary Biliary Cirrhosis sequential (PBCseq) data and a simulation study. We found drugs and sex can be eliminated from the PBCseq analysis, and over time the diseaseprogresses.

Successful Treatment Of Type 2 Diabetes In Older Black And Hispanic Community Members

PURPOSE: As the seventh leading cause of death in the United States, diabetes continues to have a major impact on morbidity and mortality. Diabetes is significantly more prevalent in Black (18.7%) and Hispanic Americans (11.8%) compared to non-Hispanic white populations (7.1%). Specific daily physical activities (PA) guidelines for people with chronic diseases, particularly for ethnic minorities, is lacking. Our hypothesis is that lower levels (moderate or low, rather than high intensity) of physical activities can reduce HbA1c in older Black and Hispanic Americans with type 2 diabetes. METHODS: Seventeen participants were conveniently recruited from UConn Health outpatient clinics and surrounding communities for JUMP in 2019-2020. Participants completed four face-to-face interviews, established SMART goals, and wore an activity tracker to measure their daily PA for three months. The activity tracker classifies PA into high, moderate and low intensity levels. HbA1c was measured through a fingerstick at each visit. Change scores for minutes engaged in each PA level and HbA1c values were created. A linear mixed model for longitudinal data was used to analyze these relationships over time. The study protocol was approved by the university IRB. RESULTS: Eleven Black and six Hispanic individuals (mean age 56) completed the study. At baseline, mean HbA1c was 8.58 mmol/mol, mean BMI was 34.8, and mean systolic/diastolic blood pressure was 135/81. Over three months, HbA1c linearly declined from 8.58 at baseline to 8.21 at Month 1, 8.19 at Month 2, and 7.96 at Month 3 (p < 0.05). Average daily minutes of moderate intensity PA increased from 15.8 (Month 1) to 20.3 (Month 2) (p = 0.045) to 20.8 (Month 3) (p = 0.022). These changes were significantly associated with the reduction of HbA1c. Lastly, average daily minutes of high intensity PA increased from 12.1 (Month 1) to 13.6 (Month 2) (p = 0.075) to 15.2 (Month 3) (p = 0.051). CONCLUSION: Over time, increases of 4 to 5 minutes to moderate daily PA were associated with decreased HbA1c in Black and Hispanic Americans with type 2 diabetes. Increasing moderate daily physical activity may be a manageable guideline for glycemic control.

Hand grip strength before and after SARS-CoV-2 infection in community-dwelling older adults.

ObjectiveTo assess the association between SARS‐CoV‐2 infection and decreased hand grip strength (HGS).DesignLongitudinal population‐based study.SettingCommunity‐dwelling older adults (aged ≥60 years) living in a rural Ecuadorian village struck by the SARS‐CoV‐2 pandemic.ParticipantsOf 282 enrolled individuals, 254 (90%) finished the study.MeasurementsHGS was measured 3 months before (January 2020) and 9 months after the introduction of the virus into the population (January 2021). SARS‐CoV‐2 antibody testing was performed in two rounds: in May–June (early) and September–November (late), 2020. An independent association between SARS‐CoV‐2 infection and HGS decline was assessed by fitting linear mixed models for longitudinal data. Changes in HGS scores in SARS‐CoV‐2 seropositive subjects, according to the time elapsed since seroconversion, were compared with those who remained seronegative.ResultsOverall, 149 (59%) individuals became seropositive for SARS‐CoV‐2. The mean HGS (in kg) was 25.3 ± 8.3 at baseline and 23.7 ± 8.1 at follow‐up (p = 0.028), with 140 individuals having >5% HGS decline between both measurements. The follow‐up HGS measurement decreased by 1.72 kg in seropositive individuals, and by 0.57 kg in their seronegative counterparts (p < 0.001). SARS‐CoV‐2 seropositive individuals were 2.27 times more likely (95% CI: 1.33–3.87) to have a lower HGS measurement at the time of follow‐up than those who remained seronegative. When compared with seronegative subjects, seropositive patients with early seroconversion were 3.41 times (95% CI: 1.73–6.74) more likely to have >5% HGS decline at the time of the follow‐up than those with later, i.e., more recent, infections.ConclusionsThis study shows an independent deleterious impact of SARS‐CoV‐2 on HGS that is more marked among individuals with infections that occurred more than 8 months before follow‐up HGS. Results suggest the possibility of chronic damage to skeletal muscles by SARS‐CoV‐2.

Sublesional Bevacizumab Injection for Recurrent Respiratory Papillomatosis: Evaluation of Utility in a Typical Clinical Practice

The goal of this study was to evaluate the benefit of sublesional bevacizumab injection for recurrent respiratory papillomatosis (RRP) as used in a typical clinical practice. A retrospective review of patients with RRP treated between 2011 and 2016 was undertaken. All patients were treated with in-office potassium titanyl phosphate (KTP) laser photoablation. Sublesional bevacizumab injection was used based on joint patient-physician decision making. Papilloma burden was objectively measured on prior recordings by 2 blinded reviewers and described as vocal fold segments affected (VFSA) by papilloma. Each patient served as their own control by comparing times when bevacizumab was or was not used. Mixed model for longitudinal data was used to determine if the previous use of bevacizumab decreased the disease burden. A total of 19 patients met inclusion criteria and all prior laryngoscopic exams were reviewed for VFSA as described above. The mean (SD) number of VFSA when bevacizumab was not used at the prior procedure was 15 (14) as compared to 8 (11) when bevacizumab was used. When adjusted for age, time from initial presentation and baseline disease burden, mixed model showed a decrease of 11 VFSA (95% CI 6.5, 15.5, P < .0001) when bevacizumab was used as compared to when bevacizumab was not used at the prior procedure. In mixed model analysis, there was not a significant association between bevacizumab dose used and VFSA at the subsequent visit (P = .8). Using sublesional bevacizumab intermittently based on clinical findings appears to be effective in improving disease control. Papilloma burden is significantly decreased at the subsequent clinical follow-up following injection of sublesional bevacizumab. The beneficial effect of sublesional bevacizumab may be not be dose-dependent.

Cognitive decline among individuals with history of mild symptomatic SARS-CoV-2 infection: A longitudinal prospective study nested to a population cohort.

Background and purposeNeurological complications of SARS‐CoV‐2 infection are noticed among critically ill patients soon after disease onset. Information on delayed neurological sequelae of SARS‐CoV‐2 infection is nil. Following a longitudinal study design, the occurrence of cognitive decline among individuals with a history of mild symptomatic SARS‐CoV‐2 infection was assessed.MethodsStroke‐ and seizure‐free Atahualpa residents aged ≥40 years, who had pre‐pandemic cognitive assessments as well as normal brain magnetic resonance imaging and electroencephalogram recordings, underwent repeated evaluations 6 months after a SARS‐CoV‐2 outbreak infection in Atahualpa. Patients requiring oxygen therapy, hospitalization, and those who had initial neurological manifestations were excluded. Cognitive decline was defined as a reduction in the Montreal Cognitive Assessment (MoCA) score between the post‐pandemic and pre‐pandemic assessments that was ≥4 points greater than the reduction observed between two pre‐pandemic MoCAs. The relationship between SARS‐CoV‐2 infection and cognitive decline was assessed by fitting logistic mixed models for longitudinal data as well as exposure‐effect models.ResultsOf 93 included individuals (mean age 62.6 ± 11 years), 52 (56%) had a history of mild symptomatic SARS‐CoV‐2 infection. Post‐pandemic MoCA decay was worse in seropositive individuals. Cognitive decline was recognized in 11/52 (21%) seropositive and 1/41 (2%) seronegative individuals. In multivariate analyses, the odds for developing cognitive decline were 18.1 times higher among SARS‐CoV‐2 seropositive individuals (95% confidence interval 1.75–188; p = 0.015). Exposure‐effect models confirmed this association (β = 0.24; 95% confidence interval 0.07–0.41; p = 0.006).ConclusionsThis study provides evidence of cognitive decline among individuals with mild symptomatic SARS‐CoV‐2 infection. The pathogenesis of this complication remains unknown.

D-optimal population designs in linear mixed effects models for multiple longitudinal data

The main purpose of this paper is to investigate D-optimal population designs in multi-response linear mixed models for longitudinal data. Observations of each response variable within subjects are assumed to have a first-order autoregressive structure, possibly with observation error. The equivalence theorems are provided to characterise the D-optimal population designs for the estimation of fixed effects in the model. The semi-Bayesian D-optimal design which is robust against the serial correlation coefficient is also considered. Simulation studies show that the correlation between multi-response variables has tiny effects on the optimal design, while the experimental costs are important factors in the optimal designs.

Body Composition in Community-Dwelling Older Adults Before and After SARS-CoV-2 Infection: A Longitudinal Prospective Study in a Rural Village Struck by the Pandemic.

Background:Information on the body composition of inhabitants of remote communities during the SARS-CoV-2 pandemic is limited. Using a longitudinal population-based study design, we assessed the association between SARS-CoV-2 infection and changes in body composition.Methods:Community-dwelling older adults living in a rural Ecuadorian village received body composition determinations before and 1 year after the pandemic as well as serological tests for detection of SARS-CoV-2 antibodies. The independent association between SARS-CoV-2 infection and abnormalities in body composition at follow-up was assessed by fitting linear mixed models for longitudinal data.Results:Of 327 enrolled individuals, 277 (85%) received baseline and follow-up body composition determinations, and 175 (63%) of them became SARS-CoV-2 seropositive. Overall, diet and physical activity deteriorated during the follow-up. Multivariate random-effects generalized least squares regression models that included the impact of time and seropositivity on follow-up body composition, showed that neither variable contributed to a worsening in body composition. Multivariate logistic regression models disclosed that the serological status at follow-up cannot be predicted by differences in body composition and other baseline covariates.Conclusions:Study results suggest no increased susceptibility to SARS-CoV-2 infection among older adults with abnormal body composition and no significant changes as a result of worse physical activity and dietary habits or seropositivity during the length of the study. Together with a previous study in the same population that showed decrease in hand-grip strength after SARS-CoV-2, results confirm that dynapenia (and not sarcopenia) is associated with SARS-CoV-2 infection in older adults.

Bayesian variable selection in linear quantile mixed models for longitudinal data with application to macular degeneration.

This paper presents a Bayesian analysis of linear mixed models for quantile regression based on a Cholesky decomposition for the covariance matrix of random effects. We develop a Bayesian shrinkage approach to quantile mixed regression models using a Bayesian adaptive lasso and an extended Bayesian adaptive group lasso. We also consider variable selection procedures for both fixed and random effects in a linear quantile mixed model via the Bayesian adaptive lasso and extended Bayesian adaptive group lasso with spike and slab priors. To improve mixing of the Markov chains, a simple and efficient partially collapsed Gibbs sampling algorithm is developed for posterior inference. Simulation experiments and an application to the Age-Related Macular Degeneration Trial data to demonstrate the proposed methods.

Social Inequalities and Loneliness as Predictors of Ageing Well: A Trend Analysis Using Mixed Models.

Background: This study examines if education, income, and loneliness are associated with physical functioning and optimism in an ageing population in Germany. Furthermore, time trends of physical functioning and optimism as well as of associations with social inequality and loneliness are analyzed. Methods: The German Ageing Survey (DEAS), a longitudinal population-based survey of individuals aged 40 years and older, was used (four waves between 2008 and 2017, total sample size N = 23,572). Physical functioning and optimism were introduced as indicators of ageing well. Educational level, net equivalent income, and loneliness were used as predictors in linear mixed models for longitudinal data. Results: Time trends show that physical functioning decreases over time, while optimism slightly increases. Education and income are positively associated with physical functioning, while higher loneliness correlates with lower physical functioning. Higher optimism was associated with higher income and particularly with lower loneliness. Income and notable educational inequalities in physical functioning increase over time. Time trends of the associations with optimism show decreasing income inequalities and increasing disparities in loneliness. Conclusions: Increasing educational inequalities in physical functioning and a strong association of loneliness with optimism provide information for further interventions. Targeted health promotion among the aged and addressing maladaptive social cognition are options to tackle these issues. Key areas for action on healthy ageing include, for instance, the alignment of health systems to the needs of older populations or the creation of age-friendly environments.

Worse Quality of Life, Function, and Pain in Children With Enthesitis, Irrespective of Their Juvenile Arthritis Category.

To estimate the impact of enthesitis on patient-reported outcomes in children with juvenile idiopathic arthritis (JIA), irrespective of JIA category. Children enrolled in the Research in Arthritis in Canadian Children Emphasizing Outcomes cohort were studied. Entheseal tenderness by physician examination in 33 defined locations, Juvenile Arthritis Quality of Life Questionnaire (JAQQ), Quality of My Life (QoML) Questionnaire, Childhood Health Assessment Questionnaire (C-HAQ), and a pain visual analog scale were completed at enrollment, every 6 months for 2 years, and then yearly up to 5 years. Analyses consisted of descriptive statistics, linear mixed models for longitudinal data, and analysis of covariance. Among 1,371 patients followed for a median of 35.3 months (interquartile range 22.1, 49.2), 214 (16%) had enthesitis, of whom 137 (64%) were classified as having enthesitis-related arthritis. After adjusting for JIA category and covariates, children with enthesitis reported higher JAQQ (mean raw score 2.71 versus 2.16, adjusted difference 0.41 points; 95% confidence interval [95% CI] 0.22, 0.59), higher C-HAQ (0.47 versus 0.31, adjusted difference 0.14 points; 95% CI 0.07, 0.22), higher pain (3.01 versus 1.68, adjusted difference 0.94 points; 95% CI 0.64, 1.25), and lower QoML (7.02 versus 8.23, adjusted difference -0.80 points; 95% CI -1.09, -0.51) scores than children without enthesitis. These differences persisted up to 5 years. Children with enthesitis, regardless of JIA category, report worse patient-reported outcomes than those without enthesitis. Thus, enthesitis should be assessed in all children with JIA.

Abstract PD3-09: Sulindac reduces breast density and alters collagen alignment in patients on aromatase inhibitors

Abstract Background. Preclinical and observational evidence supports cancer prevention activity of non-steroidal anti-inflammatory drugs (NSAIDs) in the breast via suppression of prostaglandin E2 (PGE2) synthesis by cyclooxygenase-2 (COX2). Evidence includes linking of PGE2 to aromatase activity and estrogen synthesis in breast adipose tissue, as well as effects on collagen and breast density (BD). Methods. In an open-label trial, we evaluated the effect of the non-selective NSAID sulindac at 150 mg bid for 12 months on BD in 52 postmenopausal women taking adjuvant aromatase inhibitors (AIs) for breast cancer. BD was measured using a fat-water decomposition MRI based BD measure (MRD) previously shown to be more quantitative than mammographic density. A non-randomized observation cohort of 46 postmenopausal women on AI without NSAID use was conducted in parallel to assess the effect of AI on MRD over 12 months. Eligible participants were recruited at two study sites and included women with an intact, unaffected contralateral breast and BI-RADS ≥ 2. Each subject’s MRI_BD measures at baseline and at 6 and 12 months were included in linear mixed models for longitudinal data. Log-transformation was applied to the outcome of BD. Covariates included log-transformed baseline BD, time on AI, and baseline body mass index (BMI) and change in BMI. Breast tissue collagen fiber alignment for 30 women with paired breast biopsies, before and after 6 months on sulindac, was examined using Second-Harmonic Generation (SHG) microscopy and analysis of the distribution (histogram) of ‘straight’ fibers in three randomly selected areas of breast tissue. Straightness of individual fibers was calculated as the linear length of a fiber divided by the distance along the fiber. Results. Participants on sulindac intervention had a significant change in BD relative to baseline BD at 6 (p=0.05) and 12 months (p&amp;lt;0.001). After adjusting for time on AI, baseline BMI and BMI change, the relative change in BD at 12 months from baseline BD was 10% lower (95% CI; -14.4, -5.4%) for women on sulindac. For the AI only group, no significant change from baseline BD was observed at 6 (p=0.50) or 12 months (p=0.17). The relative change in BD for the AI only group at 12 months was -3.5% (95% CI; -8.3, +1.6%). For women with baseline MRD_BD of ≥ 25%, the relative decrease in BD from baseline in the sulindac group was greater, at -16.3% (95% CI, -25.4, -6.2%). In contrast, there was no evidence for baseline BD effect on change in BD at 12 months in the AI only group whose baseline BD was ≥25% [-3.3% change at 12 mo (95% CI; -12.7, +7.1%)]. Exploratory analyses of collagen fiber alignment in the sulindac group at 6 months showed a significant decrease in straight collagen fibers (p=0.01). The decrease in collagen straightness correlated with a decrease in BD (rho=0.66, p=0.01). Conclusions. Twice daily 150 mg sulindac for 12 months in postmenopausal women on AIs was associated with a significant decrease in BD using MRD that was slightly higher for women with higher baseline BD. At the tissue level, the decrease in BD was associated with change in collagen straightness. No decrease in BD was observed in a non-randomized population of postmenopausal women on AI only therapy. These results suggest that sulindac decreases BD and may do so through effects on collagen alignment. Further investigation of sulindac for effects on BD, including synergy with AIs, for breast cancer prevention is indicated. Citation Format: Patricia Thompson, Chuan Huang, Betsy Wertheim, Christina Preece, Jie Yang, Jessica Martinez, Denise Roe, Pavani Chalasani, Alison Stopeck. Sulindac reduces breast density and alters collagen alignment in patients on aromatase inhibitors [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr PD3-09.

Empirical evaluation of the implementation of the EMA guideline on missing data in confirmatory clinical trials: Specification of mixed models for longitudinal data in study protocols.

SummaryIn confirmatory clinical trials, the prespecification of the primary analysis model is a universally accepted scientific principle to allow strict control of the type I error. Consequently, both the ICH E9 guideline and the European Medicines Agency (EMA) guideline on missing data in confirmatory clinical trials require that the primary analysis model is defined unambiguously. This requirement applies to mixed models for longitudinal data handling missing data implicitly. To evaluate the compliance with the EMA guideline, we evaluated the model specifications in those clinical study protocols from development phases II and III submitted between 2015 and 2018 to the Ethics Committee at Hannover Medical School under the German Medicinal Products Act, which planned to use a mixed model for longitudinal data in the confirmatory testing strategy. Overall, 39 trials from different types of sponsors and a wide range of therapeutic areas were evaluated. While nearly all protocols specify the fixed and random effects of the analysis model (95%), only 77% give the structure of the covariance matrix used for modeling the repeated measurements. Moreover, the testing method (36%), the estimation method (28%), the computation method (3%), and the fallback strategy (18%) are given by less than half the study protocols. Subgroup analyses indicate that these findings are universal and not specific to clinical trial phases or size of company. Altogether, our results show that guideline compliance is to various degrees poor and consequently, strict type I error rate control at the intended level is not guaranteed.

Jason Lezak again and again - linear mixed modelling analysis of change–over times in relay swimming races

ABSTRACTThe aim of this study was to examine the importance of the change-over time in swimming relay races. Top-class international 4 x 100 m freestyle races were analysed across a 10-year period including three Olympic Games and five European and World Championships. A total of 220 swimmers (116 female, 104 male) were included in this study with an average participation of 1.7 ± 1.2 races. To consider such repeated measurements and other factors (e.g., ranking in the relay race, position in the relay team) linear mixed models for longitudinal data were used for the statistical evaluation. Our results showed significantly longer change-over times for male medallists (0.23 ± 0.08 s) than non-medallists (0.20 ± 0.09 s) which reflects a very likely effect (94.2%). Furthermore, there were significant differences in change-over times between female and male swimmers depending on the current race positions. In total, the influence of change-over time on the final performance in 4 x 100-m freestyle relay appears to be overrated in previous studies.

Pseudo-Bayesian D-optimal designs for longitudinal Poisson mixed models with correlated errors

This paper is concerned with the problem of pseudo-Bayesian D-optimal designs for the first-order Poisson mixed model for longitudinal data with time-dependent correlated errors. A standard approximate covariance matrix of the parameter estimation is obtained based on the quasi-likelihood method. Furthermore, to overcome the dependence of pseudo-Bayesian D-optimal designs on the choice of the prior mean, a hierarchical pseudo-Bayesian D-optimal designs based on the hierarchical prior distribution of unknown parameters is proposed. The results show that the optimal number of time points depends on both the interclass autoregressive coefficients and different cost constraints. The relative efficiency of equidistant designs compared with the hierarchical pseudo-Bayesian D-optimal designs is also discussed.

Shrinkage estimation in linear mixed models for longitudinal data

This paper is concerned with the selection and estimation of fixed effects in linear mixed models while the random effects are treated as nuisance parameters. We propose the non-penalty James–Stein shrinkage and pretest estimation methods based on linear mixed models for longitudinal data when some of the fixed effect parameters are under a linear restriction. We establish the asymptotic distributional biases and risks of the proposed estimators, and investigate their relative performance with respect to the unrestricted maximum likelihood estimator (UE). Furthermore, we investigate the penalty (LASSO and adaptive LASSO) estimation methods and compare their relative performance with the non-penalty pretest and shrinkage estimators. A simulation study for various combinations of the inactive covariates shows that the shrinkage estimators perform better than the penalty estimators in certain parts of the parameter space. This particularly happens when there are many inactive covariates in the model. It also shows that the pretest, shrinkage, and penalty estimators all outperform the UE. We further illustrate the proposed procedures via a real data example.