Modelling correlated data: Multilevel models and generalized estimating equations and their use with data from research in developmental disabilities

Cardiovascular risk burden and disability: findings from the International Mobility in Aging Study (IMIAS)

BACKGROUND

The association of cardiovascular risk burden with disability is unclear. We examined the association between trajectories of the Framingham general cardiovascular disease risk score (FGCRS) with the trajectories of limitations of physical function in older adults.

METHODS

A total of 1219 participants with no disabilities from the International Mobility in Aging Study (IMIAS) study who had up to three repeated measures of FGCRS between 2012-2016 and without a history of stroke or coronary heart disease at baseline and follow-up were included. FGCRS at baseline was assessed and categorized into tertiles. Physical function was evaluated with the Short Physical Performance Battery (SPPB). The data were analyzed using linear mixed-effects models.

RESULTS

At baseline, FGCRS ranged between 3-94 (mean score: 24 ± 15.8), participants were 32 (2.6%), 502 (41.2%) and 685 (56.2%) in lowest, middle, and highest tertiles, respectively. In the trajectories of limitations of physical function, the lowest FGCRS had no differences, while the middle and highest had a decrease in physical performance between 2012-2014 (P = 0.0001). Age, being female, living in Andes Mountains, having middle and highest FGCRS, higher alcohol consumption, being obese, lack of exercise and cognitive impairment increase the probability of disability (P < 0.05). Alternatively, living in more developed regions and having a higher educational level reduced the probability of disability during the follow-up time (P < 0.05).

CONCLUSIONS

Higher cardiovascular risk burden is associated with decreased physical performance, especially in gait. Results suggest SPPB may provide a measure of cardiovascular health in older adults.

Locomotor-cognitive dual-tasking in children with developmental coordination disorder

Introduction

Children with Developmental Coordination Disorder (DCD) demonstrate deficits in predictive motor control and aspects of cognitive control compared with their typically developing (TD) peers. Adjustment to dynamic environments depends on both aspects of control and the deficits for children with DCD may constrain their ability to perform daily actions that involve dual-tasking. Under the assumption that motor-cognitive integration is compromised in children with DCD, we examined proportional dual-task costs using a novel locomotor-cognitive dual-task paradigm that enlisted augmented reality. We expect proportional dual-task performance costs to be greater for children with DCD compared to their TD peers.

Methods

Participants were 34 children aged 6-12 years (16 TD, 18 DCD) who walked along a straight 12 m path under single- and dual-task conditions, the cognitive task being visual discrimination under simple or complex stimulus conditions presented via augmented reality. Dual-task performance was measured in two ways: first, proportional dual-task costs (pDTC) were computed for cognitive and gait outcomes and, second, within-trial costs (p-WTC) were measured as the difference on gait outcomes between pre- and post-stimulus presentation.

Results

On measures of pDTC, TD children increased their double-limb support time when walking in response to a dual-task, while the children with DCD increased their locomotor velocity. On p-WTC, both groups increased their gait variability (step length and step width) when walking in response to a dual-task, of which the TD group had a larger proportional change than the DCD group. Greater pDTCs on motor rather than cognitive outcomes were consistent across groups and method of dual-task performance measurement.

Discussion

Contrary to predictions, our results failed to support dramatic differences in locomotor-cognitive dual-task performance between children with DCD and TD, with both groups tending to priorities the cognitive over the motor task. Inclusion of a within-trial calculation of dual-task interference revealed an expectancy effect for both groups in relation to an impending visual stimulus. It is recommended that dual-task paradigms in the future continue to use augmented reality to present the cognitive task and consider motor tasks of sufficient complexity to probe the limits of performance in children with DCD.

Use of Light Protection Equipment at Night Reduces Time Until Discharge From the Neonatal Intensive Care Unit: A Randomized Interventional Study

Newborn infants' circadian systems are not completely developed and rely on external temporal cues for synchronizing their biological rhythms to the environment. In neonatal intensive care units (NICUs), lighting is usually continuous or irregular and infants are exposed to artificial light at night, which can have negative health consequences. Therefore, the aim of this study was to evaluate the impact of the use of individual light protection equipment at night on the development and growth of preterm neonates. Infants born at less than 37 gestational weeks who no longer needed constant intensive care were admitted into a newborn nursery and randomized to either use eye masks at night (intervention, n = 21) or not (control, n = 20). Infants who used eye protection at night were discharged earlier than those in the control group (8 [5] vs 12 [3.75] days; p < 0.05). A greater variation within the day in heart rate was observed in the intervention group, with lower values of beats per minute at 1400 and 2000 h. There was no significant difference in weight gain between groups. In view of our results and of previous findings present in the literature, we suggest that combining a darkened environment at night with individual light protection devices creates better conditions for the development of preterm infants in the NICU. In addition, eye masks are an affordable and simple-to-use tool that can reduce hospitalization costs by decreasing the number of days spent in the NICU.

Learning about crash causation from countermeasure evaluation: The example of the Queensland minimum passing distance rule

Close passes by motor vehicles endanger both the safety and comfort of bicycle riders. Governments in many countries have introduced laws requiring drivers to maintain at least a minimum distance between their vehicle and the cyclist they are passing, despite relatively poor understanding of the causes of bicycle overtaking crashes at the time. Queensland was the first state in Australia to introduce such a law, with a two-year trial commencing in April 2014. The data collected during the evaluation of the trial were later analysed to answer two main questions: "Under what circumstances do close passes occur?" and "Why do drivers pass too close?". The first question was largely approached by analysing the video observations of more than 18,000 riders (including 2,000 passing events) at 15 locations on Queensland roads and examining the infrastructure, traffic and road user characteristics that influenced passing distances. The second question was addressed in experimental studies which used the video observations as stimuli. This paper demonstrates how the political need for evaluation of a countermeasure can act as a stimulus for research funding that then allows data collection, analysis and better understanding of crash causation. Logically, introduction of a countermeasure should be based on a rigorous understanding of crash causation. But when this does not occur, evaluation may provide data that can be used to answer questions about crash causation - or at least pose new questions.

Tuberculosis in Prisons: Importance of Considering the Clustering in the Analysis of Cross-Sectional Studies

The level of clustering and the adjustment by cluster-robust standard errors have yet to be widely considered and reported in cross-sectional studies of tuberculosis (TB) in prisons. In two cross-sectional studies of people deprived of liberty (PDL) in Medellin, we evaluated the impact of adjustment versus failure to adjust by clustering on prevalence ratio (PR) and 95% confidence interval (CI). We used log-binomial regression, Poisson regression, generalized estimating equations (GEE), and mixed-effects regression models. We used cluster-robust standard errors and bias-corrected standard errors. The odds ratio (OR) was 20% higher than the PR when the TB prevalence was >10% in at least one of the exposure factors. When there are three levels of clusters (city, prison, and courtyard), the cluster that had the strongest effect was the courtyard, and the 95% CI estimated with GEE and mixed-effect models were narrower than those estimated with Poisson and binomial models. Exposure factors lost their significance when we used bias-corrected standard errors due to the smaller number of clusters. Tuberculosis transmission dynamics in prisons dictate a strong cluster effect that needs to be considered and adjusted for. The omission of cluster structure and bias-corrected by the small number of clusters can lead to wrong inferences.

Effectiveness of a Transtheoretical Model-Based Intervention to Improve Blood Pressure Control of Hypertensive Patients in China: A Clustered Randomized Controlled Trial

Background

Hypertension can result in great cardiovascular risk, while medication adherence and blood pressure control of patients were suboptimal. Therefore, we conducted a 12-month clustered randomized controlled trial to evaluate the effectiveness of transtheoretical model (TTM)-based health intervention mode on blood pressure of Chinese newly-diagnosed hypertensive patients.

Methods

This trial was conducted in six primary healthcare centers, Shunyi District, Beijing, China from September 2016 to September 2017. A total of 400 patients were included and randomized, 194 patients in the control group and 206 patients in the intervention group. Patients in the intervention group received TTM-based health intervention and those in the control group received usual care. Multilevel modeling was used to adjust for clustering effect and repeated measurements.

Results

Systolic blood pressure of patient was reduced by 4.534 mm Hg at 3-month follow-up [95% CI (−3.054, −1.403), p = 0.005], 3.982 mm Hg at 6-month follow-up [95% CI (−7.387, −0.577), p = 0.022], and 5.803 mm Hg at 12-month follow-up [95% (−9.716, −1.891), p = 0.004]. Diastolic blood pressure of patient was reduced by 3.383 mm Hg at 3-month follow-up [95% CI (−5.724, −1.042), p = 0.005], 0.330 mm Hg at 6-month follow-up [95% CI (−2.870, 2.210), p = 0.799], and 3.129 mm Hg at 12-month follow-up [95% CI (−6.048, −0.21), p = 0.036]. Medication adherence of patients was improved at all the three time points. For stages of change for taking medication, patients in the intervention group were 8.401-fold more likely to be in a higher stage at 3-month follow-up [95% CI (4.186, 16.862), p < 0.001]. The odds ratio of being in a higher stage was 8.454 at 6-month follow-up [95% CI (3.943, 18.123), p < 0.001] and 19.263 at 12-month follow-up [95% CI (7.979, 46.505), p < 0.001].

Conclusion

Transtheoretical model-based health intervention might be a promising strategy to improve medication adherence of newly-diagnosed hypertensive patients in community.

An Evaluation of the Implementation of a "No Force First" Informed Organisational Guide to Reduce Physical Restraint in Mental Health and Learning Disability Inpatient Settings in the UK

BACKGROUND

The use of physical restraint on vulnerable people with learning disabilities and mental health problems is one of the most controversial and criticised forms of restrictive practice. This paper reports on the implementation of an organisational approach called "No Force First" within a large mental health organisation in England, UK. The aim was to investigate changes in violence/aggression, harm, and physical restraint following implementation.

METHODS

The study used a pretest-posttest quasi-experimental design. Recorded incidents of violence/aggression from 44 inpatient mental health and learning disabilities (including forensic) wards were included (n = 13,599). Two study groups were created for comparison: the "intervention" group comprising all incidents on these wards during the 24 months post-implementation (2018-2019) (n = 6,551) and the "control" group comprising all incidents in the 24 months preceding implementation (2015-2016) (n = 7,048). Incidents recorded during implementation (i.e., 2017) were excluded (n = 3,705). Incidence rate ratios (IRR) were calculated with 95% confidence intervals (95% CI). Multivariate regression models using generalised estimating equations were performed to estimate unadjusted and adjusted prevalence ratios (aPR) of physical restraint and harm, using type of wards, incident, and violence/aggression as key covariates.

RESULTS

A significant 17% reduction in incidence of physical restraint was observed [IRR = 0.83, 95% CI 0.77-0.88, p < 0.0001]. Significant reductions in rates of harm sustained and aggression/violence were also observed, but not concerning the use of medication during restraint. The prevalence of physical restraint was significantly higher in inpatients on forensic learning disability wards than those on forensic mental health wards both pre- (aPR = 4.26, 95% CI 2.91-6.23) and post-intervention (aPR = 9.09, 95% CI 5.09-16.23), when controlling for type of incident and type of violence/aggression. Physical assault was a significantly more prevalent risk factor of restraint use than other forms of violence/aggression, especially that directed to staff (not to other patients).

CONCLUSIONS

This is a key study reporting the positive impact that organisational models and guides such as "No Force First" can have on equipping staff to focus more on primary and secondary prevention as opposed to tertiary coercive practices such as restraint in mental health and learning disabilities settings.

The association between statewide vaping prevalence and COVID-19

Existing literature indicated electronic cigarette users (vapers) have impaired immune response that might increase vulnerability to coronavirus disease 2019 (COVID-19) infection and death. However, whether vapers are more susceptible to COVID-19 is unknown. Using integrated data in each US state from the 2018 Behavioral Risk Factor Surveillance System (BRFSS), United States Census Bureau and the 1Point3Acres.com website, generalized estimating equation (GEE) models with negative binomial distribution assumption and log link functions were used to examine the association of statewide e-cigarette use prevalence with number of COVID-19 cases and deaths in the US on a state level from January 21, 2020 to April 25, 2020. The weighted proportion of vapers who used e-cigarettes every day or some days ranged from 2.86% to 6.42% for US states. Statistically significant associations were observed between the weighted proportion of vapers and number of COVID-19 cases as well as COVID-19 deaths in the US after adjusting for the weighted proportion of smokers and other significant covariates in the GEE models. With every one percent increase in weighted proportion of vapers in each state, the number of COVID-19 cases increase by 0.3139 (95% CI: 0.0554-0.5723) and the number of COVID-19 deaths increase by 0.3730 (95% CI: 0.0815-0.6646) in log scale in each US state. The positive associations between the proportion of vapers and the number of COVID-19 cases and deaths in each US state in this ecological study suggest an increased susceptibility of vapers to COVID-19 on a state level and warrants further investigation.

Are Vapers More Susceptible to COVID-19 Infection?

Background COVID-19, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), was declared a global pandemic in March 2020. Electronic cigarette use (vaping) rapidly gained popularity in the US in recent years. Whether electronic cigarette users (vapers) are more susceptible to COVID-19 infection is unknown. Methods Using integrated data in each US state from the 2018 Behavioral Risk Factor Surveillance System (BRFSS), United States Census Bureau and the 1Point3Acres.com website, generalized estimating equation (GEE) models with negative binomial distribution assumption and log link functions were used to examine the association of weighted proportions of vapers with number of COVID-19 infections and deaths in the US. Results The weighted proportion of vapers who used e-cigarettes every day or some days ranged from 2.86% to 6.42% for US states. Statistically significant associations were observed between the weighted proportion of vapers and number of COVID-19 infected cases as well as COVID-19 deaths in the US after adjusting for the weighted proportion of smokers and other significant covariates in the GEE models. With every one percent increase in weighted proportion of vapers in each state, the number of COVID-19 infected cases increase by 0.3139 (95% CI: 0.0554 - 0.5723) and the number of COVID-19 deaths increase by 0.3705 (95% CI: 0.0623 - 0.6786) in log scale in each US state. Conclusions The positive associations between the proportion of vapers and the number of COVID-19 infected cases and deaths in each US state suggest an increased susceptibility of vapers to COVID-19 infections and deaths.

Understanding health management and safety decisions using signal processing and machine learning

Background

Small group research in healthcare is important because it deals with interaction and decision-making processes that can help to identify and improve safer patient treatment and care. However, the number of studies is limited due to time- and resource-intensive data processing. The aim of this study was to examine the feasibility of using signal processing and machine learning techniques to understand teamwork and behaviour related to healthcare management and patient safety, and to contribute to literature and research of teamwork in healthcare.

Methods

Clinical and non-clinical healthcare professionals organised into 28 teams took part in a video- and audio-recorded role-play exercise that represented a fictional healthcare system, and included the opportunity to discuss and improve healthcare management and patient safety. Group interactions were analysed using the recurrence quantification analysis (RQA; Knight et al., 2016), a signal processing method that examines stability, determinism, and complexity of group interactions. Data were benchmarked against self-reported quality of team participation and social support. Transcripts of group conversations were explored using the topic modelling approach (Blei et al., 2003), a machine learning method that helps to identify emerging themes within large corpora of qualitative data.

Results

Groups exhibited stable group interactions that were positively correlated with perceived social support, and negatively correlated with predictive behaviour. Data processing of the qualitative data revealed conversations focused on: (1) the management of patient incidents; (2) the responsibilities among team members; (3) the importance of a good internal team environment; and (4) the hospital culture.

Conclusions

This study has shed new light on small group research using signal processing and machine learning methods. Future studies are encouraged to use these methods in the healthcare context, and to conduct further research on how the nature of group interaction and communication processes contribute to the quality of team and task decision-making.

Electronic supplementary material

The online version of this article (10.1186/s12874-019-0756-2) contains supplementary material, which is available to authorized users.

Linear Mixed-Model Analysis to Examine Longitudinal Trajectories in Vocabulary Depth and Breadth in Children Who Are Hard of Hearing

Purpose Children who are hard of hearing (CHH) tend to have reduced vocabularies compared to children with normal hearing (CNH). Prior research on vocabulary skills in children with hearing loss has focused primarily on their breadth of knowledge (how many words are known). Depth of vocabulary knowledge (how well words are known) is not well documented for CHH. The current study used linear mixed models (LMMs) to investigate growth trajectories of vocabulary depth and breadth in CHH relative to age-matched CNH. Method Participants for this study included 155 children (93 CHH, 62 CNH) enrolled in a longitudinal study. Examiners administered a standardized measure of vocabulary knowledge at ages 7, 8, and 9 years. We constructed multiple LMMs with fixed effects for group and age. The models included various combinations of random intercepts for subject and item and random slope for age. Results For depth, CHH showed significant and stable deficits compared to CNH over time. For breadth, CNH showed greater vocabulary breadth, but the group differences diminished with age. For CHH, higher aided audibility, age, and maternal educational level were associated with greater vocabulary breadth and depth. Age at hearing aid fitting was not. Conclusions A major advantage of using LMM is that it allowed us to cope with missing data points while still accounting for variability within and across participants. Assessment of both vocabulary breadth and depth may be useful in identifying school-age CHH who are at risk of delays in language outcomes.