Mixed-effects Multinomial Logistic Regression Research Articles

Impact of bedaquiline resistance probability on treatment decision for rifampicin-resistant TB

<sec><title>BACKGROUND</title>Accurate diagnosis of bedaquiline (BDQ) resistance remains challenging. A Bayesian approach expresses this uncertainty as a probability of BDQ resistance (prBDQR) with a 95% credible interval. We investigated how prBDQR information influences BDQ prescribing decisions.</sec><sec><title>METHOD</title>We performed a discrete choice experiment with 55 international rifampicin-resistant tuberculosis physicians. We employed mixed-effects multinomial logistic regression to quantify the effect of prBDQR, patient attributes, and contextual factors on the decision to continue BDQ or not when sequencing results become available.</sec><sec><title>RESULTS</title>PrBDQR was the most influential factor for BDQ decision-making, three times greater than treatment response. Each percentage point increase in prBDQR resulted in 8.2% lower odds (OR 0.92, 95% CI 0.90–0.93) of continuing BDQ as a fully effective drug and 5.0% lower odds (OR 0.95, 95% CI 0.94–0.96) of continuing it but not counting it as an effective drug. The most favourable patient profile for prescribing BDQ as a fully effective drug was a patient receiving the BPaLM regimen (BDQ, pretomanid, linezolid and moxifloxacin) with low prBDQR, good 1-month treatment response, fluoroquinolone-susceptible TB, and no prior BDQ treatment. Physicians with higher discomfort with uncertainty and more years of experience with BDQ were more inclined to stop BDQ.</sec><sec><title>CONCLUSION</title>Given the uncertainty of genotype-phenotype associations, physicians valued prBDQR for BDQ decision-making in rifampicin-resistant TB treatment.</sec>

Continuity of Health Insurance Coverage and Choice of Contraception Method.

Objective: To evaluate whether part-year or year-round uninsurance is associated with reduced likelihood of using prescription contraception methods rather than using nonprescription methods or using no contraceptive methods. Methods: We identified nonpregnant and sexually active female respondents participating in the National Longitudinal Survey of Youth, 1997 cohort between 2007 and 2019. At each interview, we classified the contraceptive method used most frequently as prescription, nonprescription, or none, and used mixed-effects multinomial logistic regression to predict contraceptive method based on health insurance coverage over the past year (classified as continuous private, continuous public, part-year uninsured, or year-round uninsured). Results: Our sample included 3,738 respondents and 18,678 observations (person-years). In the most recent interview, 35% of respondents used prescription contraception, 16% used nonprescription methods only, and 49% used no method. On multivariable analysis using all available years of data, respondents with part-year uninsurance were 20% less likely to use prescription rather than nonprescription methods, as compared to respondents with continuous private insurance (95% confidence interval: -31%, -6%; p = 0.007), but did not differ on the likelihood of using prescription methods rather than no method. Conclusions: Part-year uninsurance was associated with lower use of prescription contraceptive methods rather than nonprescription methods when compared with continuous private insurance coverage. Use of prescription contraceptives was lowest among people with year-round uninsurance. Policy efforts ensuring continuous insurance coverage with greater flexibility of eligibility and enrollment periods may promote greater access to prescription contraceptives.

Women empowerment domains and unmet need for contraception among married and cohabiting fecund women in Sub-Saharan Africa: A multilevel analysis based on gender role framework.

Low women empowerment, is a known contributing factor to unmet needs for contraception by limiting access to health services through negative cultural beliefs and practices. However, little is known about the association between unmet needs and domains of women empowerment in Sub-Saharan African (SSA) countries. Hence, this study aimed at assessing the influence of women empowerment domains on the unmet need for contraception in the region using the most recent Demographic and Health Survey (DHS) data (2016-2021). The data for the study was derived from the appended women's (IR) file of eighteen SSA countries. A weighted sample of 128,939 married women was analyzed by STATA version 16. The Harvard Institute's Gender Roles Framework, which comprised of influencer, resource, and decision-making domains was employed to identify and categorize the covariates across three levels. The effects of each predictor on the unmet need for spacing and limiting were examined using a multivariable multilevel mixed-effect multinomial logistic regression analysis. Adjusted relative risk ratio (aRRR) with its corresponding 95% confidence interval was used to declare the statistical significance of the independent variables. The pooled prevalence of unmet needs for contraception was 26.36% (95% CI: 24.83-30.40) in the region, with unmet needs for spacing and limiting being 16.74% (95% CI: 16.55, 17.02) and 9.62% (95% CI: 9.45, 12.78), respectively. Among variables in the influencer domain, educational level, family size of more than five, parity, number of children, attitude towards wife beating, and media exposure were substantially linked with an unmet need for spacing and limiting. Being in the poorest wealth quintile and enrollment in health insurance schemes, on the other hand, were the two variables in the resource domain that had a significant influence on unmet needs. The overall decision-making capacity of women was found to be the sole significant predictor of unmet needs among the covariates in the decision-making domain. Unmet needs for contraception in SSA countries were found to be high. Reproductive health program planners and contraceptive service providers should place due emphasis on women who lack formal education, are from low-income families, and have large family sizes. Governments should collaborate with insurance providers to increase health insurance coverage alongside incorporating family planning within the service package to minimize out-of-pocket costs. NGOs, government bodies, and program planners should collaborate across sectors to pool resources, advocate for policies, share best practices, and coordinate initiatives to maximize the capacity of women's decision-making autonomy.

202-LB: Association between Diabetes and the Vaginal Microbiota

Type 2 diabetes is a known risk factor for urinary tract infection and vulvovaginal candidiasis, and this risk is modulated in part by vaginal microbiota. We sought to assess the associations between diabetes and the vaginal microbiota, specifically whether diabetes may reduce vaginal Lactobacillus spp., which produce lactic acid and protect the urogenital tract. This study included 811 participants (2,301 person-visits) aged 35-60 recruited to a 2-year observational cohort with mid-vaginal sampling every 6 months. Diabetes was defined as self-reported diabetes medications, history, or serum glycated albumin (GA) levels >16.5% (HbA1c not available). Vaginal microbiota was characterized by 16S rRNA gene sequencing and clustered into community state types (CSTs) by VALENCIA, a nearest centroid algorithm based on a reference dataset. Mixed-effects multinomial logistic regression with bootstrapping was used to assess the association between case status and CSTs. Among diabetes cases (n=45, 130 person-visits), 45% had L. crispatus, L. jensenii, L. gasseri-dominated (CST I/II/V), 12% had L. iners-dominated (CST III) and 43% had low-Lactobacillus (CST IV) microbiota. Diabetes cases had reduced odds of L. iners-dominated CST III (aOR: 0.54, 95% CI: 0.31-0.81) than controls (12% vs 24%) and no association with low-Lactobacillus CST IV (aOR=01.05, 95% CI: 0.76-1.44) compared to Lactobacillus-dominated CST I/II/V in a model adjusted for menopause status and race. Estimates were similar when only GA was used for a diabetes case definition on a subset (1,001 person-visits). The results suggest diabetes is associated with reduced L. iners dominance. Diabetes cases had even frequencies of low-Lactobacillus or the optimal Lactobacillus-dominated CSTs. L. iners is metabolically distinct from other lactobacilli, producing only L-lactic acid (not the more protective D-lactic acid), and is associated with increased reproductive tract infection risk. This data informs how diabetes may affect the vaginal microenvironment. Disclosure S. J. Robbins: None. R. M. Brotman: None. R. S. Miller: None. A. L. Beitelshees: None. M. Shardell: None. J. Ravel: None. S. I. Taylor: Consultant; Ionis Pharmaceuticals. Funding National Institute of Allergy and Infectious Diseases (F31-AI164859)

Mental health profiles of Finnish adolescents before and after the peak of the COVID-19 pandemic

BackgroundThe COVID-19 pandemic has had implications for adolescents’ interpersonal relationships, communication patterns, education, recreational activities and well-being. An understanding of the impact of the pandemic on their mental health is crucial in measures to promote the post-pandemic recovery. Using a person-centered approach, the current study aimed to identify mental health profiles in two cross-sectional samples of Finnish adolescents before and after the peak of the pandemic, and to examine how socio-demographic and psychosocial factors, academic expectations, health literacy, and self-rated health are associated with the emerging profiles.Methods and findingsSurvey data from the Health Behaviour in School-aged Children (HBSC) study conducted in Finland in 2018 (N = 3498, age M = 13.44) and 2022 (N = 3838, age M = 13.21) were analyzed. A four-profile model using cluster analysis was selected for both samples. In Sample 1, the identified profiles were (1) “Good mental health”, (2) “Mixed psychosocial health”, (3) “Somatically challenged”, and (4) “Poor mental health”. In Sample 2, the identified profiles were (1) “Good mental health”, (2) “Mixed psychosomatic health”, (3) “Poor mental health and low loneliness”, and (4) “Poor mental health and high loneliness”. The results of the mixed effect multinomial logistic regression analysis showed that in both samples, being a girl and reporting lower maternal monitoring; lower family, peer, and teacher support; higher intensity of online communication; a less positive home atmosphere and school climate; and poor self-rated health were most strongly linked to belonging to a poorer mental health profile. In addition, in Sample 2, low subjective health literacy was a key factor associated with poorer mental health profiles, and teacher support was more important than before COVID.ConclusionsThe current study stresses the importance of identifying those vulnerable to developing poor mental health. To maximize post-pandemic recovery, the role of schools, especially teacher support and health literacy, along with the factors that remained important over time should be taken into account in public health and health promotion interventions.

Leptospira Infection in Rural Areas of Urabá Region, Colombia: A Prospective Study.

The objective of this study was to analyze the eco-epidemiological aspects of Leptospira seroprevalence and seroincidence and its associated factors in two municipalities of northwest Colombia. A prospective study was performed in rural areas of Urabá, Antioquia, Colombia. The study enrolled 597 people between November 2015 and January 2016, of which 274 were followed up 1 year later. Serologic testing was performed by a microscopic agglutination. The outcomes were seroprevalent and seroincident cases, and the primary exposure was an outdoor occupation. A binary and mixed-effect multinomial logistic regression model was used to estimate factors associated with seroprevalent or seroincident cases of Leptospira infection. The overall Leptospira seroprevalence was 27.81% (95% confidence interval: 23.62-32.49), and the overall cumulative seroincidence for Leptospira was 14.60% (95% confidence interval: 10.33-20.23). Multivariable analysis showed that factors associated with L. interrogans serogroups seropositivity were outdoor occupation, male gender, older age, the presence of dirt floor in the household, and the presence of piglets and opossums. It also showed that factors associated with other Leptospira species serogroups were the presence of pit latrines and of turkeys. In addition, the multivariable model of seroincident cases of L. interrogans serogroups evidenced outdoor occupations, the presence of rats, and corn cultivation as risk factors. Likewise, the multivariable model for seroincident cases of other Leptospira species showed that the presence of hunting canines and cassava cultivation were risk factors. We found specific factors associated with the transmission of Leptospira serogroups contribute to the understanding of the epidemiology of Leptospira infection in rural areas of Urabá, Colombia.

Obesity is associated with myelin oligodendrocyte glycoprotein antibody-associated disease in acute optic neuritis

Optic neuritis (ON) is a frequent presentation at onset of multiple sclerosis (MS), neuromyelitis optica spectrum disorder (NMOSD), and myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD). The pathophysiology underlying these diseases, especially MOGAD, is still being elucidated. While obesity has been reported to potentially be a risk factor for MS, this has not been explored in NMOSD or MOGAD. We aimed to investigate a possible association between obesity (body mass index [BMI] > 30 kg/m2) in patients with MOGAD, aquaporin 4-IgG positive NMOSD (AQP4-IgG+ NMOSD) or MS. In this multicenter non-interventional retrospective study, data was collected from patients with a first ever demyelinating attack of ON subsequently diagnosed with MOGAD (n = 44), AQP4-IgG+ NMOSD (n = 49) or MS (n = 90) between 2005 and 2020. The following data was collected: age, sex, ethnicity, BMI (documented before corticosteroid treatment), and the ON etiology after diagnostic work-up. A mixed model analysis was performed to assess the potential of obesity or BMI to predict MOGAD-ON, and to distinguish MOGAD-ON from AQP4-IgG+ NMOSD-ON and MS-ON. Main outcome measures included BMI in patients with acute ON and subsequent diagnosis of MOGAD, AQP4-IgG+ NMOSD or MS. A higher BMI was significantly associated with a diagnosis of MOGAD-ON (p < 0.001); in MOGAD patients the mean BMI was 31.6 kg/m2 (standard deviation (SD) 7.2), while the mean BMI was 24.7 kg/m2 (SD 5.3) in AQP4-IgG+ NMOSD patients, and 26.9 kg/m2 (SD 6.2) in MS patients. Mixed-effects multinomial logistic regression, adjusted for age and sex, with obesity as a binary variable, revealed that obesity was associated with a higher odds ratio (OR) of a subsequent MOGAD diagnosis (OR 5.466, 95% CI [2.039, 14.650], p = 0.001) in contradistinction with AQP4-IgG+ NMOSD. This study suggests an association between obesity and MOGAD. Our findings require further exploration, but could have significant pathophysiologic implications if confirmed in larger prospective studies.

Sexual Concurrency Among Adolescent Women With Multiple Partners: A Daily Diary Study

PurposeSexual partnerships that overlap in some period of time (i.e., concurrent) are a key factor in sexually transmitted infection (STI) transmission. Research examining concurrency among adolescents typically uses person-level, cross-sectional, and/or retrospective data, obscuring factors that manifest on any given occasion of concurrency. We used sexual diaries to examine: (1) daily prevalence of vaginal sex concurrency when two partners are reported by adolescent women; and (2) individual, relational, and behavioral attributes that impact the likelihood of sex with neither partner, with one partner or the other, or with both partners on any given day. MethodsDaily diaries were drawn from a large longitudinal cohort study examining sexual relationships, behaviors, and STIs among adolescent women. Participants contributed 280,844 partner-associated diary entries, of which (27.9%; n = 78,356) indicated multiple partners. All two-partner diary entries were retained for analysis (N = 323; Mage = 17.55; 90% African American). Random intercept mixed-effects multinomial logistic regression was used to evaluate the impact of predictor variables on odds of having sex with one or the other, both, or neither partner. ResultsMost two-partner days (93.5%) involved no sex with either partner. Few reports (.2%) indicated same-day sexual concurrency. Older age, greater partner support, and higher sexual interest increased the likelihood of sex with both partners on the same day. ConclusionsIndividual, relational, and behavioral factors predict concurrency in ways that challenge assumptions that secondary partnerships constitute negligible relationships. Programs targeting partner concurrency in adolescents may find success tailoring messaging around these findings.

Fidelity evaluation of the dialogue around respiratory illness treatment (DART) program communication training

ObjectiveTo evaluate receipt fidelity of communication training content included in a multifaceted intervention known to reduce antibiotic over-prescribing for pediatric acute respiratory tract infections (ARTIs), by examining the degree to which clinicians implemented the intended communication behavior changes. MethodsParents were surveyed regarding clinician communication behaviors immediately after attending 1026 visits by children 6 months to < 11 years old diagnosed with ARTIs by 53 clinicians in 18 pediatric practices. Communication outcomes analyzed were whether clinicians: (A) provided both a combined (negative + positive) treatment recommendation and a contingency plan (full implementation); (B) provided either a combined treatment recommendation or a contingency plan (partial implementation); or (C) provided neither (no implementation). We used mixed effects multinomial logistic regression to determine whether these 3 communication outcomes changed between baseline and the time periods following each of 3 training modules. ResultsAfter completing the communication training, the adjusted probability of clinicians fully implementing the intended communication behavior changes increased by an absolute 8.1% compared to baseline (95% Confidence Interval [CI]: 2.4%, 13.8%, p = .005). ConclusionsOur findings support the receipt fidelity of the intervention’s communication training content. Practical implicationsClinicians can be trained to implement communication behaviors that may aid in reducing antibiotic over-prescribing for ARTIs.

The beneficial effect of providing kidney transplantation information on transplantation status differs between for-profit and nonprofit dialysis centers.

Informing end-stage kidney disease patients about kidney transplantation options increases the likelihood of kidney transplant waiting list (WL) enrollment and live donor kidney transplant (LDKT) receipt. Patients in for-profit dialysis centers have lower rates of WL enrollment and LDKT receipt. This study examined if the ownership status of dialysis centers modified the association between informing patients about transplantation options and patients' transplantation status. Multilevel analysis using mixed-effect multinomial logistic regression was performed using the United States Renal Data System (USRDS) data (January 2005 to December 2017). The study showed that informing patients improved the odds of WL enrollment and LDKT receipt. However, the effect of informing patients on transplantation status was less pronounced at for-profit as compared with nonprofit centers (Nonprofit: WL enrollment OR: 2.23 [95% CI: 2.07-2.40], and LDKT receipt OR: 3.35 [95% CI: 2.65-4.25]. For-profit: WL enrollment OR: 1.73 [95% CI: 1.66-1.79], and LDKT receipt OR: 2.35 [95% CI: 2.08-2.66]), although the odds of informing patients was higher for for-profit centers, and type of patients informed were similar across both types of centers. Information provided by for-profit centers was potentially less effective than those provided by nonprofit centers. Standardized guidelines for transplantation information provision are needed in order to ensure similar informational quality across centers.

Leveraging the temporal dynamics of anticipatory vowel-to-vowel coarticulation in linguistic prediction: A statistical modeling approach

Previous research has shown that coarticulatory information in the signal orients listeners in spoken word recognition, and that articulatory and perceptual dynamics closely parallel one another. The current study uses statistical classification to test the power of time-varying anticipatory coarticulatory information present in the acoustic signal for predicting upcoming sounds in the speech stream. Bayesian mixed-effects multinomial logistic regression models were trained on several different representations of spectral variation present in V1 in order to predict the identity of V2 in naturally coarticulated transconsonantal V1…V2 sequences. Models trained on simple measures of spectral variation (e.g. formant measures taken at V1 midpoint) were compared with models trained on more sophisticated time-varying representations (e.g. the estimated coefficients of polynomial curves fit to whole formant trajectories of V1). Accuracy in predicting V2 was greater when models were trained on dynamic representations of spectral variation in V1, and those trained on quadratic and cubic polynomial representations achieved the greatest accuracy, with more than 15 percentage points in correct classification over using midpoint formant frequencies alone. The results demonstrate that spectral representations with high temporal resolution capture more disambiguating anticipatory information available in the signal than representations with lower temporal resolution.

Race, Education, and Gender Disparities in Transplantation of Kidneys From Hepatitis C Viremic Donors.

Transplantation of kidneys from hepatitis C virus (HCV)-viremic donors into HCV-negative patients followed by direct-acting antiviral therapy was an important breakthrough to increase the number of life-saving kidney transplants. Data suggest that these transplants offer several benefits; however, it is unknown whether adoption of this practice has been shared equitably, especially among disadvantaged groups. We evaluated United Network for Organ Sharing data on HCV-seronegative adult deceased-donor kidney transplant recipients from January 1, 2017, to June 12, 2020. We compared recipients of a kidney from an HCV antibody- (Ab-)/nucleic acid test- (NAT-), HCV Ab+/NAT-, and HCV NAT+ donor. The primary covariates were as follows: (1) race/ethnicity; (2) female sex; and (3) highest level of education. Models included variables associated with being offered an HCV NAT+ kidney. We fit mixed-effects multinomial logistic regression models with the center as a random effect to account for patient clustering. Of 48 255 adult kidney-alone deceased-donor kidney transplant HCV-seronegative recipients, 1641 (3.4%) donors were HCV NAT+-, increasing from 0.3% (January 2017-June 2017) to 6.9% (January 2020-June 2020). In multivariable models, racial/ethnic minorities, women, and those with less education were significantly less likely to receive a kidney from an HCV NAT+ donor relative to an HCV Ab-/NAT- and HCV Ab+/NAT- donor. The disparities were most pronounced among Hispanic and Asian patients with less educational attainment (grade school, high school, or some college/tech school). Despite an increase in transplants from HCV NAT+ donors, we found substantial racial/ethnic disparities in transplantation of these kidneys. These data highlight how the benefits of a scientific breakthrough are often made less available to disadvantaged patients.

Relationship between childhood obesity and socio-economic status among primary school children in Costa Rica.

This article analyses the relationship between socio-economic status and the prevalence of overweight and obesity in the primary school population in Costa Rica. A National School Weight/Height Census was disseminated across Costa Rica in 2016. The percentage of children who were overweight or obese was calculated by sex, age and socio-economic indicators (type of institution: private, public, mix; type of geographic location: rural, urban and the level of development of the district of residence: quartiles). A mixed-effects multinomial logistic regression model and mixed-effects logistic regression model were used to analyse the association between the prevalence of being overweight or obese and district socio-economic status. The survey was carried out in public and private primary schools across Costa Rica in 2016. In total, 347 366 students from 6 to 12 years were enrolled in public and private primary schools. The prevalence of overweight and obesity among children was 34·0 %. Children in private schools were more likely to be overweight or obese than students in public schools (OR = 1·10 [1·07, 1·13]). Additionally, children were less likely to be overweight or obese if attending a school in a district of the lowest socio-economic quartile compared with the highest socio-economic quartile (OR = 0·79 [0·75, 0·83]) and in a rural area compared with the urban area (OR = 0·92 [0·87, 0·97]). Childhood obesity in Costa Rica continues to be a public health problem. Prevalence of overweight and obesity in children was associated with indicators of higher socio-economic status.

Spatial statistical analysis of pre-existing mortalities of 20 diseases with COVID-19 mortalities in the continental United States

BackgroundAlthough the United States is among the countries with the highest mortalities of COVID-19, inadequate geospatial studies have analyzed the disease mortalities across the nation. MethodsIn this county-level study, we investigated age-adjusted co-mortalities of 20 diseases, including cardiovascular, cancer, drug and alcohol disorder, respiratory and infectious diseases with COVID-19 over the first ten months of epidemic. One-way analysis of variance was applied to the Local Moran's I classes (High-High and Low-Low clusters, and non-significant counties of COVID-19) to examine whether the mean mortality measures of covariates that fall into the classes are significantly different. Moreover, a mixed-effects multinomial logistic regression model was employed to estimate the effects of mortalities on COVID-19 classes. ResultsResults showed that the distribution of COVID-19 case fatality ratio (CFR) and mortality rate co-occurrence of High-High clusters were mainly concentrated in Louisiana, Connecticut, and New Jersey. Also, positive associations were observed between High-High cluster of COVID-19 CFR and Asthma (OR = 4.584, 95 % Confidence Interval (CI): 2.583–8.137), Hepatitis (OR = 5.602, CI: 1.265–24.814) and Leukemia (OR = 2.172, CI: 1.518–3.106) mortality rates compared to the non-significant counties, respectively. ConclusionsOur results indicated that counties with higher mortality of some cancers and respiratory diseases are more vulnerable to fall into clusters of HH COVID-19 CFR. Future vaccine allocation and more medical professionals and treatment equipment should be a priority to those High-High clusters.

Trends in Mineral Metabolism Treatment Strategies in Patients Receiving Hemodialysis in the United States.

With multiple medications indicated for mineral metabolism, dialysis providers can apply various strategies to achieve target phosphate and parathyroid hormone (PTH) levels. We describe common prescribing patterns and practice variation in mineral metabolism treatment strategies over the last decade. In a cohort of adults initiating hemodialysis at Dialysis Clinic, Inc. facilities, we assessed prescriptions of vitamin D sterols, phosphate binders, and cinacalcet longitudinally. To identify the influence of secular trends in clinical practice, we stratified the cohort by dialysis initiation year (2006-2008, 2009-2011, and 2012-2015). To measure practice variation, we estimated the median odds ratio for prescribing different mineral metabolism treatment strategies at 12 months post-dialysis initiation across facilities using mixed effects multinomial logistic regression. Sensitivity analyses evaluated strategies used after detection of first elevated PTH. Among 23,549 incident patients on hemodialysis, there was a decline in vitamin D sterol-based strategies and a corresponding increase in strategies without PTH-modifying agents (i.e., phosphate binders alone or no mineral metabolism medications) and cinacalcet-containing treatment strategies between 2006 and 2015. The proportion with active vitamin D sterol-based strategies at dialysis initiation decreased across cohorts: 15% (2006-2008) to 5% (2012-2015). The proportion with active vitamin D sterol-based strategies after 18 months of dialysis decreased across cohorts: 52% (2006-2008) to 34% (2012-2015). The odds of using individual strategies compared with reference (active vitamin D sterol with phosphate binder) varied from 1.5- to two-fold across facilities in 2006-2008 and 2009-2011 cohorts, and increased to two- to three-fold in the 2012-2015 cohort. Findings were similar in sensitivity analyses starting from first elevated PTH measurement. Over time, mineral metabolism management involved less use of vitamin D sterol-based strategies, greater use of both more conservative and cinacalcet-containing strategies, and increased practice variation, suggesting growing equipoise.

Keeping kids safe for active travel to school: A mixed method examination of school policies and practices and children’s school travel behaviour

Active school travel contributes to children’s physical, mental and social wellbeing. The prevalence of children’s active school travel, however, has been declining in many developed countries. Gaining insights into school culture and environments in relation to school travel behaviour is crucial to inform interventions. Using a multiphase mixed methods approach, this study aimed to provide a comprehensive understanding of how school policies and practices supported or inhibited school travel behaviour in Auckland, New Zealand. Data were drawn from Neighbourhoods for Active Kids, a cross-sectional study of 1085 children aged 8–13 years between February 2015 and December 2016. School representatives were interviewed regarding their policies and practices related to school travel behaviour and traffic around school, and the data were analysed thematically. An overarching theme, sub-themes and categories were contextualised for quantitative modelling using objectively measured school variables (school socioeconomic status, active school travel programme, built environments around school). Mixed effects multinomial logistic regression models were employed to determine associations between school travel mode and objectively measured child (sociodemographic characteristics, traffic safety perceptions) and school variables. Safety was the core concept of school travel policies, procedures and programmes. Significant differences in child variables, school socioeconomic status, and cycle lanes and traffic lights around school were found between children who actively travelled or used public transport to school and those driven to school. Overall, this study demonstrated the important role of school policy and procedures and the potential application of an intersectoral approach for interventions to support changes in school travel behaviour.

Illness, Social Disadvantage, and Sexual Risk Behavior in Adolescence and the Transition to Adulthood.

This study investigated the influence of illness on sexual risk behavior in adolescence and the transition to adulthood, both directly and through moderation of the impact of social disadvantage. We hypothesized positive effects for social disadvantages and illness on sexual risk behavior, consistent with the development of faster life history strategies among young people facing greater life adversity. Using the first two waves of the National Longitudinal Study of Adolescent to Adult Health, we developed a mixed-effects multinomial logistic regression model predicting sexual risk behavior in three comparisons: risky nonmonogamous sex versus safer nonmonogamous sex, versus monogamous sex, and versus being sexually inactive, by social characteristics, illness, interactions thereof, and control covariates. Multiple imputation was used to address a modest amount of missing data. Subjects reporting higher levels of illness had lower odds of having safer nonmonogamous sex (OR = 0.84, p < .001), monogamous sex (OR = 0.82, p < .001), and being sexually inactive (OR = 0.74, p < .001) versus risky nonmonogamous sex, relative to subjects in better health. Illness significantly moderated the sex (OR = 0.88, p < .01), race/ethnicity (e.g., OR = 1.21, p < .001), and childhood SES (OR = 0.94; p < .01) effects for the sexually inactive versus risky nonmonogamous sex comparison. Substantive findings were generally robust across waves and in sensitivity analyses. These findings offer general support for the predictions of life history theory. Illness and various social disadvantages are associated with increased sexual risk behavior in adolescence and the transition to adulthood. Further, analyses indicate that the buffering effects of several protective social statuses against sexual risk-taking are substantially eroded by illness.

Exploring the association between food insecurity and food skills among school-aged children.

To examine the relationship between household food insecurity (FI) and children's involvement in family meal choices and food preparation, used as proxies for children's food skills, and to explore gender differences within these associations. Households were classified as food-secure or food-insecure using the six-item, short-form Household Food Security Survey Module. Children's involvement in family meal choices and food preparation were treated as proxies for children's food skills. Mixed-effects multinomial logistic regression models were used. Public schools in Nova Scotia, Canada. 5244 children in the fifth grade (10-11 years old) participating in the Children's Lifestyle and School Performance Study (CLASS). Most children reported being involved in family meal choices or food preparation at least weekly (74 and 68 %). The likelihood of helping choose family meals once a week was 33 % lower among girls from food-insecure households compared to girls from food-secure households. No differences in boys' involvement in family meal choices were observed according to household FI status. Boys from food-insecure households were 65 % more likely than boys from food-secure households to assist with food preparation/cooking four times per week. No differences in girls' involvement in food preparation were observed according to household FI status. Findings support that household FI is not due to a lack of food skills but most likely due to inadequate access to resources. This supports the call for upstream policies targeting the structural issues underpinning household FI such as low income.

A Markov Mixed-Effect Multinomial Logistic Regression Model for Nominal Repeated Measures with an Application to Syntactic Self-Priming Effects

Syntactic priming effects have been investigated for several decades in psycholinguistics and the cognitive sciences to understand the cognitive mechanisms that support language production and comprehension. The question of whether speakers prime themselves is central to adjudicating between two theories of syntactic priming, activation-based theories and expectation-based theories. However, there is a lack of a statistical model to investigate the two different theories when nominal repeated measures are obtained from multiple participants and items. This paper presents a Markov mixed-effect multinomial logistic regression model in which there are fixed and random effects for own-category lags and cross-category lags in a multivariate structure and there are category-specific crossed random effects (random person and item effects). The model is illustrated with experimental data that investigates the average and participant-specific deviations in syntactic self-priming effects. Results of the model suggest that evidence of self-priming is consistent with the predictions of activation-based theories. Accuracy of parameter estimates and precision is evaluated via a simulation study using Bayesian analysis.

Nutritional paradox in Ethiopian women: Multilevel multinomial analysis

Nutrition-related diseases have emerged at a faster rate in lower economic conditions before the battle against poverty and undernutrition has been won. In Ethiopia, the double burden of malnutrition seems to come sooner. The present study aims to determine the prevalence and identify the determinants of malnutrition among reproductive-age women using the 2016 Ethiopian Demographic and Health Survey dataset. Data from Ethiopian Demographic and Health Survey 2016 were used to identify individual and community-level associated factors of malnutrition among reproductive-age women. Nonpregnant, non-puerperal women aged 15-49 were included. Six hundred forty-two communities and 11,513 women were involved from this nationally representative data. The analysis was done using multi-level mixed-effects multinomial logistic regression to determine fixed effects of individual and community-level factors and random intercept of between characteristics. Thirty-seven percent of the women had malnutrition generally. One-quarter of Ethiopian women were underweight and 11.8% were overweight/obese. Household wealth, women's age and contraceptive use were individual factors negatively associated with undernutrition. Region, place of residence and community-level wealth were cluster characteristics associated with undernutrition. On the other hand, the educational status of women, household wealth index, women's age, contraceptive use, breastfeeding, region and place of residence were the factors associated with overweight. Wealthier women had lowest of the odds for undernutrition OR=0.84 (95% CI, 0.71, 0.98: P-value<0.05). Middle-aged women had lower odds of being underweight, whereas reproductive-age extremities on both ends were prone to malnutrition. Modern contraceptive use had decreased the odds of being underweight OR=0.71 (95% CI 0.63, 0.81P-value <0.05). Women residing in rural areas had higher odds for undernutrition OR=1.31 (95% CI 1.10, 1.56: p-value <0.001). In the same scenario, wealthier cluster had lower odds for undernutrition as rivalled with poorer clusters OR=0.75 (95% CI 0.65, 0.87: P-value<0.0001). Nutritional paradox in the form of overweight/obesity and underweight exists in Ethiopian women. Underweight has a high prevalence and it is in a serious situation, whereas, the burden of overweight/obesity is rising. Both individual and community-level characteristics were significant predictors of malnutrition in Ethiopian women. Besides, the individual-level factors, interventions should also consider community-level associates in tackling malnutrition.