Group‐based trajectory modeling of body mass index and body size over the life course: A scoping review

Empirical sample-specific approaches to define HPV16 and HPV18 seropositivity in unvaccinated, young, sexually active women

Given low seroconversion rates following human papillomavirus (HPV) infection, fixed external cutoffs may lead to errors in estimating HPV seroprevalence. We evaluated finite mixture modeling (FMM) and group-based trajectory modeling (GBTM) among unvaccinated, sexually active, HPV-exposed women to determine study-specific HPV16 and HPV18 seropositivity thresholds. We included 399 women (aged 18-24 years) enrolled in the HPV Infection and Transmission Among Couples Through Heterosexual Activity (HITCH) cohort study between 2005 and 2011 in Montreal, Canada. Participants' blood samples from up to six visits spanning 2 years were tested by multiplex serology for antibodies [median fluorescence intensity (MFI)] specific to bacterially expressed HPV16 and HPV18 L1 glutathione S-transferase fusion proteins. We applied FMM and GBTM to baseline and longitudinal antibody titer measurements, respectively, to define HPV type-specific seronegative and seropositive distributions. Study-specific thresholds were generated as five standard deviations above the mean seronegative antibody titers, mimicking cutoffs (HPV16: 422 MFI; HPV18: 394 MFI) derived from an external population of sexually inactive, HPV DNA-negative Korean women (aged 15-29 years). Agreement (kappa) of study-specific thresholds was evaluated against external cutoffs. Seroprevalence estimates using FMM (HPV16: 27.5%-43.2%; HPV18: 21.7%-49.5%) and GBTM (HPV16: 11.8%-11.8%; HPV18: 9.9%-13.4%) thresholds exceeded those of external cutoffs (HPV: 10.2%; HPV18: 9.7%). FMM thresholds showed slight-to-moderate agreement with external cutoffs (HPV16: 0.26%-0.46%; HPV18: 0.20%-0.56%), while GBTM thresholds exhibited high agreement (HPV16: 0.92%-0.92%; HPV18: 0.82%-0.99%). Kappa values suggest that GBTM, used for longitudinal serological data, and otherwise FMM, for cross-sectional data, are robust methods for determining the HPV serostatus without prior classification rules.IMPORTANCEWhile human papillomavirus (HPV) seropositivity has been employed as an epidemiologic determinant of the natural history of genital HPV infections, only a fraction of women incidentally infected with HPV respond by developing significant antibody levels. HPV seropositivity is often determined by a dichotomous fixed cutoff based on the seroreactivity of an external population of women presumed as seronegative, given the lack of evidence of HPV exposure. However, considering the variable nature of seroreactivity upon HPV infection, which arguably varies across populations, such externally defined cutoffs may lack specificity to the population of interest, causing inappropriate assessment of HPV seroprevalence and related epidemiologic uses of that information. This study demonstrates that finite mixture modeling (FMM) and group-based trajectory modeling (GBTM) can be used to independently estimate seroprevalence or serve as the basis for defining study-specific seropositivity thresholds without requiring prior subjective assumptions, consequently providing a more apt internally valid discrimination of seropositive from seronegative individuals.

Risk assessment of obesity-related noncommunicable diseases through body mass index trajectories in adulthood: NHANES 2007-2018

AIM

To assess the impact of adult body mass index (BMI) trajectories on the risk of obesity-related noncommunicable diseases (NCDs) in the U.S. adults after adjustment for sociodemographic and lifestyle factors.

METHODS

Data were extracted from the National Health and Nutrition Examination Survey conducted from 2007 to 2018, including male and female participants aged 29-59 years. Rao-Scott adjusted chi-square was employed to detect associations between categorical variables in descriptive analyses. Cox proportional hazards models estimated hazard ratios (HR) and 95% confidence intervals (CI) for NCDs and BMI trajectories, adjusted for sociodemographic and lifestyle factors. Kaplan-Meier curves illustrated the cumulative incidence over time.

RESULTS

Analyses were carried out on 15 721 participants and revealing significant differences among BMI trajectories in terms of demographic, lifestyle, and health characteristics. The overall prevalence of NCDs was 28.0% (95%CI:26.6-28.9). The cumulative incidence over time was higher in the high increase, moderate increase, and mixed trajectory groups, with a correspondingly higher cumulative risk (p < 0.001). Non-overweight trajectory was considered reference category in Cox models. The BMI trajectories were independently associated with an increased risk of NCDs, even after adjusting for potential confounders (HR: 1.7; 95%CI: 1.4-1.9 for moderate increase; HR: 3.6; 95%CI: 3.2-4.1 for high increase; and HR: 2.4; 95%CI: 2.1-2.7, for mixed). Furthermore, differences between males and females were also observed.

CONCLUSION

The transition to and persistence of obesity into adulthood increases the risk of NCDs. The implementation of targeted interventions with long-term monitoring of BMI may be beneficial in the prevention of future obesity-related NCDs.

Resilience, Accelerated Aging and Persistently Poor Health: Diverse Trajectories of Health in Malawi

Individuals age at vastly different rates resulting in significant within-population heterogeneity in health and aging outcomes. This diversity in health and aging trajectories has rarely been investigated among low-income aging populations that have experienced substantial hardships throughout their lifecourses. Utilizing 2006-2018 data from the Malawi Longitudinal Study of Families and Health (MLSFH) and estimating group-based trajectory models (GBTM), our analyses identified three distinct lifecourse health trajectories: (1) comparatively good initial mental and physical health that persisted throughout the lifecourse ("resilient aging"); (2) relatively good initial mental and physical health that started to deteriorate during mid-adulthood ("accelerated aging"); and (3) poor initial mental and physical health that further declined over the lifecourse ("aging with persistently poor health"). For both physical and mental health, men were more likely to enjoy resilient aging than women. Predictors other than gender of trajectory membership sometimes confirmed, and sometimes contradicted, hypotheses derived from high-income country studies. Our analyses highlight the long arm of early life conditions and gender in determining aging trajectories and show that a non-trivial sub-population is characterized by aging with persistently poor health. The study uncovers widening gaps in health outcomes between those who age with resilience and those who experience accelerated aging.

Clustering and trajectories of key noncommunicable disease risk factors in Norway: the NCDNOR project

Noncommunicable diseases (NCDs) are a leading cause of premature death globally and have common preventable risk factors. In Norway, the NCDNOR-project aims at establishing new knowledge in the prevention of NCDs by combining information from national registries with data from population-based health studies. In the present study, we aimed to harmonize data on key NCD risk factors from the health studies, describe clustering of risk factors using intersection diagrams and latent class analysis, and identify long-term risk factor trajectories using latent class mixed models. The harmonized study sample consisted of 808,732 individuals (1,197,158 participations). Two-thirds were exposed to ≥ 1 NCD risk factor (daily smoking, physical inactivity, obesity, hypertension, hypercholesterolaemia or hypertriglyceridaemia). In individuals exposed to ≥ 2 risk factors (24%), we identified five distinct clusters, all characterized by fewer years of education and lower income compared to individuals exposed to < 2 risk factors. We identified distinct long-term trajectories of smoking intensity, leisure-time physical activity, body mass index, blood pressure, and blood lipids. Individuals in the trajectories tended to differ across sex, education, and body mass index. This provides important insights into the mechanisms by which NCD risk factors can occur and may help the development of interventions aimed at preventing NCDs.

Gene-Environment Interplay in the Development of Overweight

PURPOSE

Overweight in youth is influenced by genes and environment. Gene-environment interaction (G×E) has been demonstrated in twin studies and recent developments in genetics allow for studying G×E using individual genetic predispositions for overweight. We examine genetic influence on trajectories of overweight during adolescence and early adulthood and determine whether genetic predisposition is attenuated by higher socioeconomic status and having physically active parents.

METHODS

Latent class growth models of overweight were fitted using data from the TRacking Adolescents' Individual Lives Survey (n = 2720). A polygenic score for body mass index (BMI) was derived using summary statistics from a genome-wide association study of adult BMI (N = ∼700,000) and tested as predictor of developmental pathways of overweight. Multinomial logistic regression models were used to examine effects of interactions of genetic predisposition with socioeconomic status and parental physical activity (n = 1675).

RESULTS

A three-class model of developmental pathways of overweight fitted the data best ("non-overweight", "adolescent-onset overweight", and "persistent overweight"). The polygenic score for BMI and socioeconomic status distinguished the persistent overweight and adolescent-onset overweight trajectories from the non-overweight trajectory. Only genetic predisposition differentiated the adolescent-onset from the persistent overweight trajectory. There was no evidence for G×E.

DISCUSSION

Higher genetic predisposition increased the risk of developing overweight during adolescence and young adulthood and was associated with an earlier age at onset. We did not find that genetic predisposition was offset by higher socioeconomic status or having physically active parents. Instead, lower socioeconomic status and higher genetic predisposition acted as additive risk factors for developing overweight.

Body mass index, waist circumference and pre-frailty/frailty: the Tromsø study 1994-2016

OBJECTIVE

This study investigated the association between obesity, assessed using body mass index (BMI) and waist circumference (WC), and pre-frailty/frailty among older adults over 21 years of follow-up.

DESIGN

Prospective cohort study.

SETTING

Population-based study among community-dwelling adults in Tromsø municipality, Norway.

PARTICIPANTS

2340 women and 2169 men aged ≥45 years attending the Tromsø study in 1994-1995 (Tromsø4) and 2015-2016 (Tromsø7), with additional BMI and WC measurements in 2001 (Tromsø5) and 2007-2008 (Tromsø6).

PRIMARY OUTCOME MEASURE

Physical frailty was defined as the presence of three or more and pre-frailty as the presence of one to two of the five frailty components suggested by Fried et al: low grip strength, slow walking speed, exhaustion, unintentional weight loss and low physical activity.

RESULTS

Participants with baseline obesity (adjusted OR 2.41, 95% CI 1.93 to 3.02), assessed by BMI, were more likely to be pre-frail/frail than those with normal BMI. Participants with high (OR 2.14, 95% CI 1.59 to 2.87) or moderately high (OR 1.57, 95% CI 1.21 to 2.03) baseline WC were more likely to be pre-frail/frail than those with normal WC. Those at baseline with normal BMI but moderately high/high WC or overweight with normal WC had no significantly increased odds for pre-frailty/frailty. However, those with both obesity and moderately high/high WC had increased odds of pre-frailty/frailty. Higher odds of pre-frailty/frailty were observed among those in 'overweight to obesity' or 'increasing obesity' trajectories than those with stable normal BMI. Compared with participants in a stable normal WC trajectory, those with high WC throughout follow-up were more likely to be pre-frail/frail.

CONCLUSION

Both general and abdominal obesity, especially over time during adulthood, is associated with an increased risk of pre-frailty/frailty in later years. Thus maintaining normal BMI and WC throughout adult life is important.

Kidney-Function Trajectories From Young Adulthood to Midlife: Identifying Risk Strata and Opportunities for Intervention

Introduction

Understanding normative patterns of change in kidney function over the life course may allow targeting of early interventions to slow or prevent the onset of kidney disease, but knowledge about kidney functional change before middle age is limited. This study used prospective longitudinal data from a representative birth cohort to examine common patterns of change from young to midadulthood and to identify risk factors and outcomes associated with poorer trajectories.

Methods

We used group-based trajectory modeling in the Dunedin study birth cohort (n = 857) to identify the following: (i) common kidney function trajectories between the ages 32 and 45 years, (ii) early-life factors associated with those trajectories, (iii) modifiable physical and psychosocial factors across adulthood associated with differences in trajectory slope, and (iv) links between trajectories and kidney-related outcomes at age 45 years.

Results

Three trajectory groups were identified and could be differentiated by age 32 years as follows: normal (58% of participants), low-normal (36%), and high-risk (6%) groups. Those from low socioeconomic backgrounds had higher odds of following a high-risk (vs. normal) trajectory. Modifiable factors (blood pressure, body mass index, inflammation, glycated hemoglobin, smoking, and socioeconomic status) across adulthood were associated with steeper age-related declines in kidney function, particularly among those in the low-normal and high-risk groups. Those in the low-normal and high-risk groups also had more adverse kidney-related outcomes at age 45 years.

Conclusion

The current findings could be used to inform the development of early interventions and point to socioeconomic conditions across the life course and health-related risk factors and behaviors in adulthood as kidney health promotion targets.

Association between distinct body mass index trajectories according to the group-based trajectory modeling and the risk of incident diabetes: A systematic review

We aimed to determine the association between distinct body mass index (BMI) trajectories, using group-based trajectory modeling, and the subsequent risk of incident diabetes. Five databases were systematically searched. Fourteen population-based cohort studies that summarized the association between different BMI trajectories and subsequent diabetes, with the four most common BMI trajectories including the "stable," "increasing," "decreasing," and "turning" groups, were included. The rapid increase and stable high-level BMI groups showed the strongest association with the subsequent risk of diabetes compared with the stable normal BMI group. Increased baseline BMI levels resulted in a steeper slope and greater risk of subsequent diabetes. In the decreasing BMI group, one study reported that those aged &gt;50 years showed the highest incidence of subsequent diabetes, whereas the other two studies reported no association between these two variables. In the turning group, an increase followed by a decrease in BMI levels from adolescence to late adulthood could reduce the risk of developing diabetes, although the residual risk remained. By contrast, the incidence of subsequent diabetes remained high in the middle-aged BMI-turning group. This study can provide further insights for identifying populations at high risk of diabetes and for developing targeted prevention strategies.

Inter-subject variability of pleasant pain relief using a data-driven approach in healthy volunteers

Background

The offset of a painful and unpleasant sensation can elicit pleasure. This phenomenon, namely pleasant pain relief (PPR), is attracting growing interest in research. While the cold pressor test (CPT) has been frequently used to study the inhibition of pain by the administration of another painful stimulation (inhibitory conditioned pain modulation; ICPM), a preliminary study from our research team has shown that CPT can also elicit a robust and long-lasting PPR. However, its effects on pain relief and inhibition vary greatly between subjects. Although substantial research has been carried out on inter-individual variability in the case of ICPM, the same cannot be said of PPR. Therefore, the current study sought to identify clusters of healthy volunteers with similar dynamic pain responses during the CPT, using a data-driven approach, and to investigate the inter-subject variability for PPR and ICPM.

Methods

One hundred and twenty-two healthy volunteers were recruited. A sequential ICPM paradigm was carried out with CPT (water at 10°C) and a Peltier Thermode to evaluate pain intensity and unpleasantness. Moreover, PPR was measured for four minutes at CPT offset. Statistical analyses were performed using group-based trajectory modelling.

Results

Four trajectories (groups) were identified for CPT pain intensity and unpleasantness ratings with varying levels of tonic pain and pain sensitization (e.g., temporal summation). PPR scores were correlated with both pain ratings trajectories (p &lt; 0.001). On the other hand, no differences were found between groups regarding ICPM efficacy (percentage pain inhibition).

Discussion

This study has provided a first step into the investigation of PPR and ICPM interindividual variability. Using a data-driven approach, it was shown that PPR at CPT offset differs between clusters of participants identified based on dynamic pain intensity and unpleasantness responses from CPT. Thus, it was brought to light that both the levels of tonic pain and pain sensitization underlie individual differences in PPR. The lack of correlation between CPT pain trajectories and ICPM efficacy may be explained by the hypotheses that eliciting ICPM requires only a certain threshold of stimulation which doesn’t need to be noxious. In the future, studies on the inter-subject variability of PPR in large samples of chronic pain patients are warranted.

Caregiver perceptions of neighbourhood green space quality, heavy traffic conditions, and asthma symptoms: Group-based trajectory modelling and multilevel longitudinal analysis of 9,589 Australian children

BACKGROUND

This study assessed the associations between changes in exposure to green space quality, heavy traffic conditions, and asthma symptoms among children.

METHODS

10-year cohort data of 9589 children, retrieved from the Longitudinal Study of Australian Children, were analysed. Caregiver-reported neighbourhood green space quality, heavy traffic conditions, and asthma symptoms were measured biennially. Group-based trajectory mixture models were used to develop trajectory groups, denoting different patterns of, or changes in, exposure to green space quality, heavy traffic conditions, and asthma symptoms across childhood. Multilevel multinomial logistic regression was used to identify factors associated with trajectory group membership and examine the confounders-adjusted associations between trajectory groups of green space quality, heavy traffic conditions, and asthma symptoms.

RESULTS

Four trajectory groups for each green space quality and heavy traffic conditions, and five trajectory groups for asthma symptoms were developed. Children in less disadvantaged areas were more likely to be in trajectory groups with exposure to quality green space, but less likely to be exposed to heavy neighbourhood traffic. Living in more remote areas was associated with the decreased likelihood to be in groups with exposure to both quality green space and heavy traffic conditions over time. Accumulation of exposure to quality green space across childhood was not found to be protective against asthma symptoms. However, children whose caregiver perceptions of heavy traffic conditions trended from low to moderate levels; or were consistently in high levels across childhood had a higher likelihood to be in trajectory groups with a higher risk of asthma symptoms.

CONCLUSION

Exposure to quality green space was not associated with the reduced risk of asthma symptoms. The accumulation of exposure to heavy traffic conditions increased the likelihood of asthma symptoms among children. Reducing the presence of heavy traffic in neighbourhoods might reduce the risk of childhood asthma.

Does group-based trajectory modeling estimate spurious trajectories?

Background

Group-based trajectory modelling (GBTM) is increasingly used to identify subgroups of individuals with similar patterns. In this paper, we use simulated and real-life data to illustrate that GBTM is susceptible to generating spurious findings in some circumstances.

Methods

Six plausible scenarios, two of which mimicked published analyses, were simulated. Models with 1 to 10 trajectory subgroups were estimated and the model that minimized the Bayes criterion was selected. For each scenario, we assessed whether the method identified the correct number of trajectories, the correct shapes of the trajectories, and the mean number of participants of each trajectory subgroup. The performance of the average posterior probabilities, relative entropy and mismatch criteria to assess classification adequacy were compared.

Results

Among the six scenarios, the correct number of trajectories was identified in two, the correct shapes in four and the mean number of participants of each trajectory subgroup in only one. Relative entropy and mismatch outperformed the average posterior probability in detecting spurious trajectories.

Conclusion

Researchers should be aware that GBTM can generate spurious findings, especially when the average posterior probability is used as the sole criterion to evaluate model fit. Several model adequacy criteria should be used to assess classification adequacy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-022-01622-9.

Perceived green space quality, child biomarkers and health-related outcomes: A longitudinal study

Accumulating exposure to quality green space over time is posited to influence child health, yet longitudinal studies are scarce. This study aimed to examine the associations between trajectories of perceived green space quality and child health-related outcomes. We used data from 1874 childrenin the B-cohort of the Longitudinal Study of Australian Children who participated in the Child Health Checkpoint module at 11-12 years. Data on caregiver perceived green space quality measured biennially was assessed using discrete trajectory mixture models to group children by contrasting distributions in green space quality over time. Examination of associations between trajectory groups of perceived green space quality and child biomarkers (i.e., albumin-to-creatinine ratio, total, cholesterol, total triglycerides, and glucose), physical health and behavioural assessments (i.e., anthropometric measurements, blood pressure, sedentary behaviour, physical activity, sleep, aerobic work capacity, and general wellbeing), and health care use were assessed using multilevel models, adjusted for sociodemographic variables. Four perceived green space quality trajectories were identified: "decreasing quality from high to moderate"; "increasing quality from low to high"; "consistently high quality"; "consistently low quality". Compared with consistently low levels of quality green space, adjusted models indicated consistently high-quality green space was associated with lower total triglycerides (β -0.13; 95%CI -0.25, -0.01). Lower odds of hospital admission was observed among children who accumulated quality green space over time (OR 0.45; 95%CI 0.23, 0.87). These associations were observed in boys only in sex-stratified analyses. Moreover, boys accumulating quality green space through time tended to have lower diastolic blood pressure (β -2.76; 95%CI -5.17, -0.35) and girls who experienced loss in quality green space tended to have a higher percentage of body fat (β 2.81; 95%CI 0.43, 5.20). Accumulating quality green space over time is important for various aspects of child health, with contrasting benefits by sex.

Association between built environments and weight status: evidence from longitudinal data of 9589 Australian children

Background

No studies appear to examine potential associations between changes in built environments across childhood and the developmental trajectories of child weight status.

Objective

Examine the developmental trajectories of child weight status with respect to changes in childhood exposure to the built environments.

Methods

This study used data of 9589 children with biennial follow-up (2004–2016), retrieved from the Longitudinal Study of Australian Children. Changes in objectively-measured child weight status (i.e., body mass index-BMI, waist circumference) were investigated in relation to changes in seven built environments (i.e., neighbourhood safety; green space quality; footpaths and street conditions; public transport; shopping facilities; basic services; and local traffic) subjectively reported by caregivers. Group-based discrete trajectory mixture models were used to classify children according to their developmental trajectories of built environments and weight status. Multilevel multinomial logistic regression was employed to examine associations between built environments and child weight status adjusted for confounding.

Results

Two, four, and six trajectory groups were developed for built environment variables. Three groups namely “moderate”, “high”, and “extreme increase” were generated for each BMI and waist circumference. Findings from multilevel analyses indicated that growing up in neighbourhoods that are considered highly safe, with better quality of green space nearby, and in areas with low local traffic over time are protective against unhealthy weight increase in childhood. Meanwhile, living with better access to shopping facilities and basic services was associated with an unhealthy increase in BMI and/or waist circumference. No clear associations appeared between the quality of footpath and street conditions, access to public transport, and child weight status.

Conclusions

Built environments might act either as a risk or protective factor of an unhealthy increase in child weight status. Enabling health-promoting neighbourhoods (i.e., highly safe, quality green space nearby, low local traffic) is important to support a healthy weight trajectory across childhood.

Life-course trajectories of weight and their impact on the incidence of type 2 diabetes

Although exposure to overweight and obesity at different ages is associated to a higher risk of type 2 diabetes, the effect of different patterns of exposure through life remains unclear. We aimed to characterize life-course trajectories of weight categories and estimate their impact on the incidence of type 2 diabetes. We categorized the weight of 7203 participants as lean, normal or overweight at five time-points from ages 7–55 using retrospective data. Participants were followed for an average of 19 years for the development of type 2 diabetes. We used latent class analysis to describe distinctive trajectories and estimated the risk ratio, absolute risk difference and population attributable fraction (PAF) associated to different trajectories using Poisson regression. We found five distinctive life-course trajectories. Using the stable-normal weight trajectory as reference, the stable overweight, lean increasing weight, overweight from early adulthood and overweight from late adulthood trajectories were associated to higher risk of type 2 diabetes. The estimated risk ratios and absolute risk differences were statistically significant for all trajectories, except for the risk ratio of the lean increasing trajectory group among men. Of the 981 incident cases of type 2 diabetes, 47.4% among women and 42.9% among men were attributable to exposure to any life-course trajectory different from stable normal weight. Most of the risk was attributable to trajectories including overweight or obesity at any point of life (36.8% of the cases among women and 36.7% among men). The overweight from early adulthood trajectory had the highest impact (PAF: 23.2% for woman and 28.5% for men). We described five distinctive life-course trajectories of weight that were associated to increased risk of type 2 diabetes over 19 years of follow-up. The variability of the effect of exposure to overweight and obesity on the risk of developing type 2 diabetes was largely explained by exposure to the different life-course trajectories of weight.