Latent Class Growth Modeling Research Articles

Eight-year trajectories and predictors of cognitive function in community-dwelling Korean older adults with cardiovascular diseases.

This study aims to identify longitudinal patterns and predictors of cognitive function trajectories among Korean older adults with cardiovascular diseases. This study is a longitudinal panel analysis based on secondary data. Data from the the Korean Longitudinal Study of Ageing (KLoSA) were used for analysis. The KLoSA is a representative panel survey of older Koreans. We analyzed responses from 301 participants aged ≥65 years who completed the same survey more than three times out of five waves between 2012 and 2020. Latent class growth modeling identified two trajectories of cognitive function in older people with cardiovascular diseases: "low and declining" (n = 81, 26.9%) and "high and declining" (n = 220, 73.1%). Participants in "the low and declining trajectory group" were more likely to have a low educational level, weak handgrip strength, depression, and low social participation at baseline than those in "the high and declining trajectory group." Our results indicate a need to develop community-based tailored interventions for improving handgrip strength, mental health, and social participation in delaying cognitive decline in older people with cardiovascular diseases considering their educational level. Healthcare providers should be more concerned about older people with a weaker handgrip, depression, and low social activities as a high-risk group for cognitive decline over time in cardiovascular care. Therefore, it is necessary to evaluate them early with standardized tools and make subsequent strategies for the older population with cardiovascular diseases.

Per- and Polyfluoroalkyl Substances (PFAS) and Lipid Trajectories in Women 45-56 Years of Age: The Study of Women's Health Across the Nation.

Per- and polyfluoroalkyl substances (PFAS) are associated with less favorable blood lipid profiles in epidemiological studies. However, little is known about the potential role of PFAS in longitudinal changes in lipids among midlife women even though women become more susceptible to metabolic alterations during the menopausal transition. To examine associations of serum PFAS concentrations with longitudinal trajectories of blood total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, and triglycerides in midlife women undergoing menopausal transition. The sample included 1,130 women from the Study of Women's Health Across the Nation 45-56 y of age at baseline (1999-2000). We measured serum PFAS concentrations including linear perfluorooctanoic acid (n-PFOA), perfluorononanoic acid (PFNA), linear and branched perfluorooctanesulfonic acid (n-PFOS and Sm-PFOS, respectively), and perfluorohexanesulfonic acid (PFHxS) at baseline. We used k-means clustering to identify subgroups with different patterns of PFAS mixture. Blood lipids were measured annually or biannually through 2016 with an average follow-up of 14.8 y. We identified longitudinal trajectories of each lipid using latent class growth models. We used multinomial log-linear models adjusted for covariates to estimate odds ratios (ORs) and 95% confidence intervals (CIs) of lipid trajectory classes by PFAS and their mixtures. Three distinct trajectories (low, middle, high) of total, LDL, and HDL cholesterol and two distinct trajectories (low and high) of triglycerides were identified. n-PFOS, Sm-PFOS, and PFHxS were positively associated with total and LDL cholesterol trajectories. n-PFOS was inversely associated with triglycerides trajectories. PFAS mixtures (high vs. low) showed positive associations with total and LDL cholesterol trajectories (high vs. low), showing ORs (95% CIs) of 1.69 (95% CI: 1.36, 2.12) and 1.79 (95% CI: 1.44, 2.22), respectively. Concentrations of serum PFAS were positively associated with trajectories of total and LDL cholesterol, providing a line of evidence supporting adverse effects of PFAS on lipid homeostasis. https://doi.org/10.1289/EHP12351.

Digital home monitoring for capturing daily fluctuation of symptoms; a longitudinal repeated measures study: Long Covid Multi-disciplinary Consortium to Optimise Treatments and Services across the NHS (a LOCOMOTION study)

IntroductionA substantial proportion of COVID-19 survivors continue to have symptoms more than 3 months after infection, especially of those who required medical intervention. Lasting symptoms are wide-ranging, and presentation varies...

Developmental trajectories of eating behaviors and cross-lagged associations with weight across infancy

Examining typical developmental trajectories of infant eating behaviors, correlates of those trajectories, and cross-lagged associations between eating behaviors and anthropometry, is important to understand the etiology of these behaviors and their relevance to growth early in the lifespan. Mothers (N = 276) completed the Baby Eating Behavior Questionnaire (BEBQ) and infant anthropometrics were measured at ages 1, 2, 4, 6, and 10 months. Infant and maternal characteristics were collected by maternal report. Trajectories of eating behaviors were identified using latent class growth modeling and bivariate analyses examined associations of infant eating behavior trajectory membership with infant and maternal characteristics. Cross-lagged analyses examined associations between BEBQ subscales and infant weight-for-length z-score. Infant eating behavior trajectories included: Consistently High (62%) and Consistently Moderate (38%) Enjoyment of Food; Consistently High (9%), Moderate & Decreasing (43%), and Low & Decreasing (48%) Food Responsiveness; and Consistently High (62%) and Moderate & Decreasing (38%) General Appetite. Trajectory group membership was not associated with infant sex, gestational age, birthweight, or having been exclusively fed breastmilk at 2 months. Consistently High trajectories for Enjoyment of Food, Food Responsiveness, and General Appetite were associated with maternal demographic markers of psychosocial risk (e.g., lower maternal age and educational attainment). Food Responsiveness and General Appetite tracked strongly across infancy within individuals. Cross-lagged associations of Enjoyment of Food, Food Responsiveness, and General Appetite with weight-for-length z-score across infancy were generally null. Much additional work is needed to understand eating behaviors in infancy, their development, and their etiology. Further understanding of infant eating behaviors will provide the basis for future interventions to improve life course nutrition, growth, and health.

Duchenne Muscular Dystrophy Fatigue Trajectories.

Children with Duchenne muscular dystrophy (DMD) are at risk of experiencing fatigue that negatively impacts their health-related quality of life (HRQoL). This study aimed to assess the association between fatigue and HRQoL, by examining fatigue trajectories over 48 weeks, and assessing factors associated with these fatigue trajectories. The study sample consisted of 173 DMD subjects enrolled in a 48-week-long phase 2 clinical trial (NCT00592553) for a novel therapeutic who were between the ages of 5 and 16 years. The results of regression modeling show baseline fatigue and baseline HRQoL (R 2 = 0. 54 for child self-report and 0.51 for parent proxy report) and change in fatigue and HRQoL over 48 weeks (R 2 = 0.47 for child self-report and 0.36 for parent proxy report) were significantly associated with one another. Three unique fatigue trajectories using Latent Class Growth Models were identified for child and parent proxy reported fatigue. The risk of being in the high fatigue group as compared to the low fatigue group increased by 24% with each year increase in age and also with decreasing walking distance, as reported by children and parent proxy, respectively. This study identified fatigue trajectories and risk factors associated with greater fatigue, helping clinicians and researchers identify the profile of fatigue in DMD children.

A bootstrap approach for evaluating uncertainty in the number of groups identified by latent class growth models.

The use of longitudinal finite mixture models such as group-based trajectory modeling has seen a sharp increase during the last decades in the medical literature. However, these methods have been criticized especially because of the data-driven modelling process which involves statistical decision-making. In this paper, we propose an approach that uses bootstrap to sample observations with replacement from the original data to validate the number of groups identified and to quantify the uncertainty in the number of groups. The method allows investigating the statistical validity and the uncertainty of the groups identified in the original data by checking if the same solution is also found across the bootstrap samples. In a simulation study, we examined whether the bootstrap-estimated variability in the number of groups reflected the replication-wise variability. We evaluated the ability of three commonly used adequacy criteria (average posterior probability, odds of correct classification and relative entropy) to identify uncertainty in the number of groups. Finally, we illustrated the proposed approach using data from the Quebec Integrated Chronic Disease Surveillance System to identify longitudinal medication patterns between 2015 and 2018 in older adults with diabetes.

Men's Smoking Trajectories and Health-Related Quality of Life in the Whole Family: Tehran Lipid and Glucose Study.

We aimed to investigate the latent smoking classes in men and their association with health-related quality of life (HRQoL) of themselves, their wives and offspring. Using Tehran Lipid and Glucose Study (TLGS), 1781 men with marital stability and 8-18-year-old offspring were followed for 15 years (1999-2014). Latent class growth modeling (LCGM) was used to identify men's smoking patterns in 1139 men with at least three non-missing measurements of cigarettes per day (CPD); they had 1908 children (8-18 years at baseline). HRQoL and its physical and mental component summaries (PCS and MCS) was measured at the last follow-up using the SF-12v2 questionnaire. The associations of men's smoking classes with HRQoL of the family members were examined using generalized linear model. LCGM revealed four trajectories for men's smoking patterns; non/rare smokers, decreasing light, persistent moderate smokers, and persistent heavy smokers. Persistent smoking men, regardless of the amount of smoking (light or heavy), had lower scores in general health (ß= -7.80 for moderate and ß= -10.71 for heavy class) but not overall PCS. Al three trajectories of smoker men had poorer overall MCS than non/rare smokers. Living with persistent heavy smoker men was associated with decreased overall MCS in women (ß= -4.20), in particular role emotional (ß= -8.82) and mental health (ß= -9.42). No significant association was detected between fathers' smoking patterns and offspring HRQoL in young adulthood. Our results show men's heavy and persistent smoking worsens their own and their spouses' HRQoL, mainly in mental health dimensions.

Data-Driven Pupil Response Profiles as Transdiagnostic Readouts for the Detection of Neurocognitive Functioning in Affective and Anxiety Disorders

BackgroundNeurocognitive functioning is a relevant transdiagnostic dimension in psychiatry. As pupil size dynamics track cognitive load during a working memory task, we aimed to explore if this parameter allows identification of psychophysiological subtypes in healthy participants and patients with affective and anxiety disorders. MethodsOur sample consisted of 226 participants who completed the n-back task during simultaneous functional magnetic resonance imaging and pupillometry measurements. We used latent class growth modeling to identify clusters based on pupil size in response to cognitive load. In a second step, these clusters were compared on affective and anxiety symptom levels, performance in neurocognitive tests, and functional magnetic resonance imaging activity. ResultsThe clustering analysis resulted in two distinct pupil response profiles: one with a stepwise increasing pupil size with increasing cognitive load (reactive group) and one with a constant pupil size across conditions (nonreactive group). A larger increase in pupil size was significantly associated with better performance in neurocognitive tests in executive functioning and sustained attention. Statistical maps of parametric modulation of pupil size during the n-back task showed the frontoparietal network in the positive contrast and the default mode network in the negative contrast. The pupil response profile of the reactive group was associated with more thalamic activity, likely reflecting better arousal upregulation and less deactivation of the limbic system. ConclusionsPupil measurements have the potential to serve as a highly sensitive psychophysiological readout for detection of neurocognitive deficits in the core domain of executive functioning, adding to the development of valid transdiagnostic constructs in psychiatry.

Problematic use of the internet among adolescents: A four-wave longitudinal study of trajectories, predictors and outcomes.

Problematic use of the internet (PUI) among adolescents has become one of the public problems around the world. Understanding the developmental trajectory of PUI may be beneficial to develop prevention and intervention. The current study aimed to identify the developmental trajectories of PUI among adolescents, considering individual differences over time. And also explored how familial factors contributed to the identified trajectories, and the relationship between PUI changes over time and social, mental health, and academic functioning. A total of 1,149 adolescents (Mage=15.82,SD=0.61; 55.27% girls at Wave 1) participated in assessments at four time points, using 6-month assessment intervals. Based on a latent class growth model, three trajectories of PUI were identified: Low Decreasing, Moderate Increasing, and High Increasing groups. Multivariate logistic regression analyses suggested that inter-parental conflicts and childhood maltreatment served as negative familial predictors for the risk trajectories of PUI (i.e., Moderate Increasing and High Increasing groups). Additionally, adolescents in these two groups displayed more estranged interpersonal relationships, more mental health difficulties, and poorer academic functioning. It is important to consider individual differences in understanding the developmental patterns of PUI among adolescents. Identifying family predictors and the behavioral outcome associated with groups with different developmental trajectories of PUI, which may help to understand better risk factors related to specific developmental patterns of PUI and its adverse correlates. The findings highlight a need to develop more specific effective intervention programs for individuals displaying different problematic developmental trajectories with PUI.

Patient-Concerned Other Dyads' 12-Step Involvement and Patients' Substance Use: A Latent Class Growth Model Analysis.

This study aimed to inform clinical practice by identifying subgroups of patient-concerned other (CO) dyads. Patients with alcohol use disorders (AUDs) were characterized on Alcoholics Anonymous (AA) involvement and substance use together with COs' Al-Anon involvement. Predictors and recovery maintenance outcomes of subgroup membership were examined. Participants were 279 patient-CO dyads. Patients were in residential treatment for AUD. Parallel latent class growth model analysis characterized 12-step involvement and substance use at treatment entry and 3-, 6-, and 12-month follow-ups. Three classes were as follows: 38% Low AA/Low Al-Anon (patients' low AA and COs' low Al-Anon involvement, and patients' high-to-moderate substance use), 10% High AA/High Al-Anon (patients' high AA and COs' high Al-Anon involvement, and patients' moderate-to-low substance use), and 52% High AA/Low Al-Anon (patients' high AA and COs' low Al-Anon involvement, and patients' moderate-to-low substance use). At follow-up, the Low AA/Low Al-Anon classes' patients were less likely to have spirituality as recovery support, confidence about staying abstinent, and satisfaction with recovery progress. The High AA classes' COs had less concern about patients' drinking and scored higher on positive aspects of relationships with patients. Clinicians should encourage patients' and COs' 12-step group involvement (12-step practices' engagement). Among patients treated for AUD, AA involvement was related to better outcomes, and to COs' lessened concern about the patients' drinking. COs' Al-Anon involvement was associated with having a more positive view of their relationship with the patient. That more than one third of dyads had low 12-step group involvement suggests that treatment programs may need to facilitate involvement in non-12-step mutual-help groups.

How final year high school students' depression develop during COVID-19 in China? A latent class growth modeling analysis.

Depression increased sharply during the initial months of COVID-19, but how it developed over time is rarely explored, especially for adolescents. The current study measured depression of 605 final year high school students in China over 11months in 4 waves. The latent growth curve modeling (LGCM) was used to examine overall trends in depression and latent class growth modeling (LCGM) was used to identify potential subgroups of adolescents' depressive trajectories. At the same time, gender, life events, and rumination were included as time-invariant covariates. Overall, the development of depression in the final year of high school students showed a slight downward trend. Meanwhile, the depression trajectories showed heterogeneity, and three categories of depression trajectories were identified, which were low-stable (24.3%), depression-risk (67.9%), and high-stable (7.8%). Neuroticism, rumination, and life events such as punishment and loss were found to significantly predict these trajectories of depression. This study helps to characterize differential depression trajectories among adolescents throughout the COVID-19 pandemic and establish several related predictors of the trajectory of depression.

Me, Myself, and I: Therapy and Self-Concept Change in College Students

ABSTRACT Studies show self-concept highly correlates with mental well-being. Few show self-concept change over time, including the effect of therapeutic interventions and gender differences. Studying how therapy connects with self-concept improvement can lead to improved mental well-being. Over six weeks, 452 college students’ self-concept, self-esteem, satisfaction with life, and psychological well-being was measured every two weeks for a total of four times. Self-concept scores were significantly lower for those in therapy than those not. Compared with females, males had a higher self-concept overall but a lower self-concept when in therapy. Those in and out of therapy had an improving self-concept score. Latent class growth modeling identified six trajectory groups for self-concept, where the lower the starting point, the more likely participants were to be in therapy. Future research may investigate the ceiling effect on improvement and extend the study with less frequent testing to better understand long-term self-concept changes in and out of therapy.

Changes in UK parental mental health symptoms over 10months of the COVID-19 pandemic.

The threats to health, associated restrictions and economic consequences of the COVID-19 pandemic have been linked to increases in mental health difficulties for many. Parents, in particular, have experienced many challenges such as having to combine work with home-schooling their children and other caring responsibilities. Yet, it remains unclear how parental mental health has changed throughout the pandemic or what factors may have mitigated or compounded the impact of the pandemic on parents' mental health. We examined monthly survey data from two linked UK-based longitudinal studies: COVID-19: Supporting Parents, Adolescents and Children during Epidemics' (Co-SPACE) and COVID-19: Supporting Parents and Young Children during Epidemics' (Co-SPYCE). Data from 5576 parents/carers of 2-17-year-old children collected between April 2020 and January 2021 was analysed using mixed-effect modelling and latent class growth (mixture) modelling. Parental stress and depression, but not anxiety, were higher during the periods of restrictions. This pattern was most pronounced for parents with primary-school-aged children, those that worked at home or had other adults in the household. Being younger, reporting secondary or below education, working out of home, having secondary-school-aged children or children with special education needs (SEN)/neurodevelopmental disorders (ND) further moderated whether, how and when parental mental health symptoms changed. Although around three quarters of parents reported consistently low mental health symptoms, a substantial minority reported consistently high or increasing symptoms of anxiety, stress and depression. The latter were more likely to be parents who were younger than average, were a single adult in the household, had a pre-existing mental health diagnosis or had a child with special educational needs or a ND. These findings emphasise how different personal circumstances and pre-existing inequalities shaped how parents were affected by this unprecedented global pandemic and highlight the need for support and consideration to meet the needs of families in the future.

Trajectories of self-rated health among community-dwelling individuals with depressive symptoms: A latent class growth analysis

BackgroundThis study identified differences between individuals with and without depression regarding demographic and socioeconomic variables, health behavior, health status, health care utilization, and self-rated health (SRH) to identify the depressed group's SRH trajectories. MethodsData of individuals with (n = 589) and without (n = 6856) depression aged ≥20 from the 2013–2017 Korean Health Panel were analyzed. A chi-square test and t-tests examined differences in demographic and socioeconomic variables, health behaviors, health status, health care utilization, and the mean of SRH. Latent Growth Curve and Latent Class Growth Modeling identified SRH development trajectories and the most suitable latent classes explaining the trajectories, respectively. Multinomial logistic regression determined the predicting factors that classified latent classes. ResultsThe depressed group had a lower mean SRH than the non-depressed group among most variables. Three latent classes were identified, each showing different SRH trajectories. Body-mass index and pain/discomfort were predicting factors for the “poor” classes compared with the “moderate-stable” class; older age, less national health insurance, less physical activity, more pain/discomfort, and more hospitalization were predictors for the “poor-stable” class. The depressed group's mean SRH was “poor.” LimitationsLatent Class Growth Modeling in individuals with depression was based on experimental data; however, it needed to review other sample data to identify similar types of latent classes to those suggested in the current study. ConclusionsPredictors of the “poor-stable” class that were identified in this study can contribute to the formulation of intervention plans for the health and welfare of individuals with depression.

Long-term trajectories of depressive symptoms and machine learning techniques for fall prediction in older adults: Evidence from the China Health and Retirement Longitudinal Study (CHARLS)

BackgroundFalls are the most common adverse outcome of depression in older adults, yet a accurate risk prediction model for falls stratified by distinct long-term trajectories of depressive symptoms is still lacking. MethodsWe collected the data of 1617 participants from the China Health and Retirement Longitudinal Study register, spanning between 2011 and 2018. The 36 input variables included in the baseline survey were regarded as candidate features. The trajectories of depressive symptoms were classified by the latent class growth model and growth mixture model. Three data balancing technologies and four machine learning algorithms were utilized to develop predictive models for fall classification of depressive prognosis. ResultsDepressive symptom trajectories were divided into four categories, i.e., non-symptoms, new-onset increasing symptoms, slowly decreasing symptoms, and persistent high symptoms. The random forest-TomekLinks model achieved the best performance among the case and incident models with an AUC-ROC of 0.844 and 0.731, respectively. In the chronic model, the gradient boosting decision tree-synthetic minority oversampling technique obtained an AUC-ROC of 0.783. In the three models, the depressive symptom score was the most crucial component. The lung function was a common and significant feature in both the case and the chronic models. ConclusionsThis study suggests that the ideal model has a good chance of identifying older persons with a high risk of falling stratified by long-term trajectories of depressive symptoms. Baseline depressive symptom score, lung function, income, and injury experience are influential factors associated with falls of depression evolution.

A longitudinal study of the effect of individual and socio-cultural factors on students' creativity.

This longitudinal study investigated how characteristics of individual and social relationships affect Korean students' creativity development. Fifth graders (male: 3,623, female: 3,701) from 242 schools in Korea were followed annually from their 5th to 9th grades (indicating from the 5th elementary school grade to the 3rd middle school grade in the Korean school system). Exploratory factor analysis, internal consistency reliability (coefficient alpha), confirmatory factor analysis, and two-level growth model methods were performed. We investigated all nine constructs and their related items by checking metric and scalar invariance assumptions. When the measurement invariance assumptions were satisfied, we used the mean of items that constitute respective factors. We checked growth trajectories of creativity and tapped the possibility of the existence of subgroups based on the growth/change pattern using latent class growth modeling. The results showed that no subgroups existed. Thus, we constructed a two-level growth model to investigate the overall growth pattern of the students. Regarding level 1, we included time-varying variables such as peer attachment, self-regulation habits (self-management), parents' academy-oriented involvement, parent affective support, individualized, interactive teaching methods, teachers' academic pressure, and academic achievement. At level 2, we used gender and parenting style that was obtained at time point 1. The final combined model incorporating level 1 and 2 variables showed that students' self-regulation had the most association with the student's creativity followed by peer attachment, parents' academic support, interaction with parents, interaction with teachers, academic pressure from teachers, and relationships with teachers. Methods for enhancing students' creativity were discussed.

One stream, two channels? A parallel-process latent class growth model of homicide rates and suicide rates in 183 countries, between 2000 and 2019

BackgroundSuicide and homicide have long been viewed in Western culture as moral, ethical and legal equivalents. This view has underpinned many theoretical and empirical explorations into their relationship over the centuries. However, there has been little evaluation of longitudinal heterogeneity. MethodsSuicide and homicide rates in 183 countries between 2000 and 2019 were collected from the World Health Organization Global Health Observatory Repository. Corresponding structural variables (i.e., GDP per capita, unemployment rate, percentage of urban population, percentage of elderly population, and Gini index) were acquired from The World Bank and Standardized World Income Inequality Database. Parallel-process latent class growth modelling was applied to identify different classes within the joint suicide and homicide rate trajectories. Multinomial logistic regression examined relationships between the structural covariates and trajectory classes. ResultsFour trajectory classes were identified, two with inverse relationships between suicide and homicide, and two with parallel relationships: 1) countries with increasing suicide rates and decreasing homicide rates (“suicide up, homicide down”: UD, n = 41) or 2) countries with decreasing suicide rates and increasing homicide rates (“suicide down, homicide up”: DU, n = 17); and 3) countries where suicide and homicide rates both trended up (UU, n = 19), or 4) both trended down (DD, n = 106). A higher average annual growth rate (AAGR) of GDP per capita was related to an increased possibility of being in DD than in DU. Countries with higher AAGR in unemployment rates were more likely to be in UD and UU than in DD, while those with higher AAGR in urbanization were less likely to be in UD than in DD. ConclusionThe over-time relationship between suicide and homicide is heterogenous and complex. It is influenced differently by GDP per capita, urbanization and unemployment in different countries, and it is not well described by a single theory.

Trajectories of maternal D-dimer are associated with the risk of developing adverse maternal and perinatal outcomes: A prospective birth cohort study

ObjectiveTo assess the relationships of maternal D-dimer trajectories with the risk of developing adverse maternal and perinatal outcomes (AMPOs). MethodsA prospective birth cohort study was conducted in China, and 7,095 women who had singleton birth were included. The latent class growth model was used to determine the maternal D-dimer trajectory. ResultsThree maternal D-dimer trajectories were identified: (1) slight increase (43.6%), (2) rapid rise (51.3%), (3) sustained high (5.1%). Compared to pregnant women with a slight increase in D-dimer trajectory, the risk of gestational diabetes mellitus, placenta previa, macrosomia, large for gestational age (LGA), and increased postpartum bleeding was significantly increased in those with a rapid rise trajectory (adjusted OR = 1.22, 2.00, 1.80, and 1.56, adjusted β = 15.92 ∼ 25.1 ml, respectively, P < 0.05), and women with a sustained high trajectory also demonstrated a relatively elevated risk of macrosomia and LGA (adjusted OR = 2.11 and 1.82, respectively, P < 0.05). While the odds of pregnancy-induced hypertension, low birth weight, and small for gestational age in pregnant women with the rapid rise D-dimer trajectory and fetal distress in those with sustained high trajectory exhibited a reduction (adjusted OR = 0.62, 0.38, 0.54, and 0.64, respectively, P < 0.05). ConclusionThis study highlights the influence of inappropriate maternal D-dimer trajectories on the risk of AMPOs.

Distinct hyperuricemia trajectories are associated with different risks of incident diabetes: A prospective cohort study

Background and aimConflicting results suggest a link between serum uric acid and diabetes and previous studies ignored the effect of continuous exposure of serum uric acid on diabetes risk. This study aims to characterize hyperuricemia trajectories in middle-aged adults and to examine its potential impact on diabetes risk, considering the role of obesity, dyslipidemia, and hypertension. Methods and resultsThe cohort included 9192 participants who were free of diabetes before 2013. The hyperuricemia trajectories during 2009–2013 were identified by latent class growth models. Incident diabetes during 2014–2018 was used as the outcome. Modified Poisson regression models were used to assess the association of trajectories with diabetes. Furthermore, marginal structural models were used to estimate the mediating effects of the relationship between hyperuricemia trajectories and diabetes. We identified three discrete hyperuricemia trajectories: high-increasing (n = 5794), moderate-stable (n = 2049), and low-stable (n = 1349). During 5 years of follow-up, we documented 379 incident diabetes cases. Compared with the low-stable pattern, the high-increasing pattern had a higher risk of developing diabetes (RR, 1.42; 95% CI: 1.09–1.84). In addition, the percentages of total effect between the high-increasing hyperuricemia pattern and diabetes mediated by obesity, dyslipidemia, and hypertension were 24.41%, 18.26%, and 6.29%. However, the moderate-stable pattern was not associated with an increased risk of diabetes. ConclusionsThese results indicate that the high-increasing hyperuricemia trajectory is significantly associated with an increased risk of diabetes. Furthermore, obesity, dyslipidemia, and hypertension play mediating roles in the relationship between the high-increasing hyperuricemia pattern and increased diabetes risk.

Association between blood pressure during 12-28 weeks gestation and pre-eclampsia: predictive value of blood pressure trajectories constructed by latent class growth modeling

Objective: To explore the associations between blood pressure trajectories during pregnancy and risk of future pre-eclampsia in a large cohort enrolling pregnant women at gestational age of ~12 weeks from community hospitals in Tianjin. Latent class growth modeling (LCGM) was used to model the blood pressure trajectories. Methods: This was a large prospective cohort study. The study enrolled pregnant women of ~12 weeks of gestation in 19 community hospitals in Tianjin from November 1, 2016 to May 30, 2018. We obtained related information during 5 antepartum examinations before gestational week 28, i.e., week 12, week 16, week 20, week 24 and week 28. LCGM was used to model longitudinal systolic (SBP) and diastolic blood pressure (DBP) trajectories. For the association study, the predictors were set as SBP and DBP trajectory membership (built separately), the outcome was defined as the occurrence of preeclampsia after 28 weeks of gestation. Results: A total of 5 809 cases with known pregnant outcomes were documented. After excluding 249 cases per exclusion criteria, 5 560 cases with singleton pregnancy were included for final analysis. There were 128 cases preeclampsia and 106 cases gestational hypertension in this cohort. Univariate logistic regression and multivariate logistic regression showed the higher baseline SBP level and DBP level were related with increased risk of preeclampsia. Four distinctive SBP trajectories and DBP trajectories from 12 weeks to 28 weeks of gestation were identified by LCGM. After controlling for potential confounders (baseline BMI, being primipara or not, white blood cell counts, hemoglobin level, platelet counts and alanine aminotransferase level), the OR for SBP latent classification trajectory_ 4 was 4.023 (95%CI: 2.368 to 6.835, P<0.001), and the OR for SBP latent classification trajectory_3 was 1.854 (95%CI: 1.223 to 2.811, P=0.004). Logistic regression showed that: using the DBP latent classification trajectory_1 as the reference group, the OR for DBP latent classification trajectory_4 was 4.100 (95%CI: 2.571 to 6.538, P<0.001), and 2.632 (95%CI: 1.570 to 4.414, P<0.001) for DBP latent classification trajectory_2. After controlling for potential confounders (baseline BMI, being primipara or not, white blood cell counts, hemoglobin level, platelet counts and alanine aminotransferase level), the OR for DBP_traj_4 was 2.527 (95%CI: 1.534 to 4.162, P<0.001), and the OR for DBP_traj_3 was 1.297 (95%CI: 0.790 to 2.128, P=0.303), and 2.238 (95%CI: 1.328 to 3.772, P=0.002) for DBP_traj_2. Therefore, BP trajectories from 12 weeks to 28 weeks identified by LCGM served as novel risk factors that independently associated with the occurrence of preeclampsia. Receiver operating characteristic (ROC) curve analysis showed incremental diagnostic performance by combing baseline blood pressure levels with blood pressure trajectories. Conclusion: By applying LCGM, we for the first time identified distinctive BP trajectories from gestational week 12 to 28, which can independently predict the development of preeclampsia after 28 weeks of gestation.