To dichotomize or not to dichotomize?

Social determinants of deaths from pneumonia and tuberculosis in children in Brazil: an ecological study

OBJECTIVE

To identify the risk areas of deaths due to unspecified pneumonia and tuberculosis (TB) in children, and to identify if there is a relationship between these events with higher TB incidence and social determinants.

METHODS

Ecological study carried out in Brazil. All cases of TB or unspecified pneumonia deaths in children under 5 years of age reported between 2006 and 2016 were included and collected through Department of Informatics of the Unified Health System (Brazil's electronic database). The Spatial Scan Statistics was used to identify areas at higher risk of dying from this event. The spatial association was verified through the Getis-Ord techniques. The Bivariate Moran Global Index was used to verify the spatial autocorrelation between the two events. To identify the association of TB and pneumonia deaths with endemic areas of pulmonary TB and social determinants, four explanatory statistical models were identified.

RESULTS

A total of 21 391 cases of pneumonia and 238 cases of TB were identified. Spatial scanning analysis enabled the detection of four clusters of risk for TB (relative risk, RR, between 3.30 and 18.18) and 22 clusters for pneumonia (RR between 1.38 and 5.24). The spatial association of the events was confirmed (z-score 3.74 and 64.34) and spatial autocorrelation between events (Moran Index:0.031 (p=0.001)). The zero-inflated negative binomial distribution was chosen, and an association for both events was identified with the TB incidence rate (OR 5.3, 95% CI 2.85 to 9.84; OR 6.63, 95% CI 5.62 to 7.81), with the Gini Index (OR 1.78, 95% CI 1.12 to 2.82; OR 4.22, 95% CI 3.63 to4.92). Primary care coverage showed an inverse association for both events (OR 0.10, 95% CI 0.67 to 0.17; OR 0.18, 95% CI 0.15 to 0.21) for pneumonia). Finally, a family that benefited from the Bolsa Família Programme had an inverse association for deaths from pneumonia (OR 0.81, 95% CI 0.52 to 1.25).

CONCLUSIONS

The results do not just contribute to reduce mortality in children, but mainly contribute to prevent premature deaths through identification of critical areas in Brazil, which is crucial to qualify health surveillance services.

Improving IoT Predictions through the Identification of Graphical Features

IoT sensor networks have an inherent graph structure that can be used to extract graphical features for improving performance in a variety of prediction tasks. We propose a framework that represents IoT sensor network data as a graph, extracts graphical features, and applies feature selection methods to identify the most useful features that are to be used by a classifier for prediction tasks. We show that a set of generic graph-based features can improve performance of sensor network predictions without the need for application-specific and task-specific feature engineering. We apply this approach to three different prediction tasks: activity recognition from motion sensors in a smart home, demographic prediction from GPS sensor data in a smart phone, and activity recognition from GPS sensor data in a smart phone. Our approach produced comparable results with most of the state-of-the-art methods, while maintaining the additional advantage of general applicability to IoT sensor networks without using sophisticated and application-specific feature generation techniques or background knowledge. We further investigate the impact of using edge-transition times, categorical features, different sensor window sizes, and normalization in the smart home domain. We also consider deep learning approaches, including the Graph Convolutional Network (GCN), for the elimination of feature engineering in the smart home domain, but our approach provided better performance in most cases. We conclude that the graphical feature-based framework that is based on IoT sensor categorization, nodes and edges as features, and feature selection techniques provides superior results when compared to the non-graph-based features.

Cortisol and behavioral reaction of low and high sensation seekers differ in responding to a sport-specific stressor

BACKGROUND/OBJECTIVES

Sensation seeking is closely related to practicing high-risk sports. This domain offers individuals an opportunity to achieve an increased level of arousal. Moreover, stress reactivity implies interindividual differences in the capacity to respond to a stressor. The purpose of this study was to examine whether high sensation seekers (HSS) compared to low sensation seekers (LSS) experience lower levels of stress on a physiological and psychological basis. Stress was induced in a sport-specific experimental paradigm with a climbing task through a jump into the rope.

METHODS

Two extreme groups (n = 28 male athletes) were examined in a mixed-factorial design (between: HSS vs. LSS; within: repeated measurements of several variables related to stress reactivity - cortisol, heart rate, anxiety).

RESULTS

In response to the stressor HSS compared to LSS released a significantly lower amount of cortisol and needed less time for the climbing task. Heart rate and anxiety were not significant.

CONCLUSIONS

Individuals seem to react physiologically differently to a sport-specific stressor, depending on their level of sensation seeking. HSS seem to be less strained, and seem to perform better. These findings might be relevant not only for the evaluation of (sport-) psychological interventions but also for other contexts (e.g., special forces).

Top ten errors of statistical analysis in observational studies for cancer research

Observational studies using registry data make it possible to compile quality information and can surpass clinical trials in some contexts. However, data heterogeneity, analytical complexity, and the diversity of aspects to be taken into account when interpreting results makes it easy for mistakes to be made and calls for mastery of statistical methodology. Some questionable research practices that include poor analytical data management are responsible for the low reproducibility of some results; yet, there is a paucity of information in the literature regarding specific statistical pitfalls of cancer studies. In addition to proposing how to avoid or solve them, this article seeks to expose ten common problematic situations in the analysis of cancer registries: convenience, dichotomization, stratification, regression to the mean, impact of sample size, competing risks, immortal time and survivor bias, management of missing values, and data dredging.

Statistical review: frequently given comments

From 2006 to 2014, I have carried out approximately 200 statistical reviews of manuscripts for ARD. My most frequent review comments concern the following: 1. Report how missing data were handled. 2. Limit the number of covariates in regression analyses. 3. Do not use stepwise selection of covariates. 4. Use analysis of covariance (ANCOVA) to adjust for baseline values in randomised controlled trials. 5. Do not use ANCOVA to adjust for baseline values in observational studies. 6. Dichotomising a continuous variable: a bad idea. 7. Student's t test is better than non-parametric tests. 8. Do not use Yates' continuity correction. 9. Mean (SD) is also relevant for non-normally distributed data. 10. Report estimate, CI and (possibly) p value-in that order of importance. 11. Post hoc power calculations-do not do it. 12. Do not test for baseline imbalances in a randomised controlled trial. 13. Report actual p values with 2 digits, maximum 3 decimals. 14. Format for reporting CIs.

Psychological symptoms in children of parents with chronic pain-the HUNT study

The aim of the present study was to investigate the associations between parental chronic pain and anxiety, depression, and conduct problems in adolescents. The current study was based on cross-sectional surveys performed during 2006 to 2008 from the Nord Trøndelag Health Study (HUNT 3 and Young-HUNT 3). The sample consisted of 3227 adolescents aged 13 to 18 years for whom information was available on parental chronic pain and health statuses. Separate analyses were conducted for girls and boys. The results indicated that if both parents experienced chronic pain, there was an increased risk of symptoms of anxiety and depression in girls (OR=2.17, CI=1.36-3.45, P=.001) and boys (OR=2.33, CI=1.17-4.63, P=.016) compared with children for whom neither parent had chronic pain. Girls had an increased risk of conduct problems in school if their mothers had chronic pain (OR=1.33, CI=1.02-1.74, P=.034). These results remained after adjusting for the possible effects of confounding factors and parental mental health. The results suggest that the presence of both maternal and paternal chronic pain is a high risk factor for internalizing symptoms in both girls and boys. The present study offers insights that should prove useful in clinical work and further large-scale research.

Determination of moderate-to-severe postoperative pain on the numeric rating scale: a cut-off point analysis applying four different methods

BACKGROUND

Cut-off points (CPs) of the numeric rating scale (NRS 0-10) are regularly used in postoperative pain treatment. However, there is insufficient evidence to identify the optimal CP between mild and moderate pain.

METHODS

A total of 435 patients undergoing general, trauma, or oral and maxillofacial surgery were studied. To determine the optimal CP for pain treatment, four approaches were used: first, patients estimated their tolerable postoperative pain intensity before operation; secondly, 24 h after surgery, they indicated if they would have preferred to receive more analgesics; thirdly, satisfaction with pain treatment was analysed, and fourthly, multivariate analysis was used to calculate the optimal CP for pain intensities in relation to pain-related interference with movement, breathing, sleep, and mood.

RESULTS

The estimated tolerable postoperative pain before operation was median (range) NRS 4.0 (0-10). Patients who would have liked more analgesics reported significantly higher average pain since surgery [median NRS 5.0 (0-9)] compared with those without this request [NRS 3.0 (0-8)]. Patients satisfied with pain treatment reported an average pain intensity of median NRS 3.0 (0-8) compared with less satisfied patients with NRS 5.0 (2-9). Analysis of average postoperative pain in relation to pain-related interference with mood and activity indicated pain categories of NRS 0-2, mild; 3-4, moderate; and 5-10, severe pain.

CONCLUSIONS

Three of the four methods identified a treatment threshold of average pain of NRS≥4. This was considered to identify patients with pain of moderate-to-severe intensity. This cut-off was indentified as the tolerable pain threshold.

Factors associated with internalizing or somatic symptoms in a cross-sectional study of school children in grades 1-10

BACKGROUND

School related factors that may contribute to children's subjective health have not been extensively studied. We assessed whether factors assumed to promote health and factors assumed to have adverse effects were associated with self-reported internalizing or somatic symptoms.

METHODS

In a cross-sectional study, 230 boys and 189 girls in grades 1-10 from five schools responded to the same set of questions. Proportional odds logistic regression was used to assess associations of school related factors with the prevalence of sadness, anxiety, stomach ache, and headache.

RESULTS

In multivariable analyses, perceived loneliness showed strong and positive associations with sadness (odds ratio, 1.94, 95% CI 1.42 to 2.64), anxiety (odds ratio, 1.78, 95% CI 1.31 to 2.42), and headache (odds ratio, 1.47, 95% CI 1.10 to 1.96), with consistently stronger associations for girls than boys. Among assumed health promoting factors, receiving necessary help from teachers was associated with lower prevalence of stomach ache in girls (odds ratio, 0.51, 95% CI 0.30 to 0.87).

CONCLUSIONS

These findings suggest that perceived loneliness may be strongly related to both internalizing and somatic symptoms among school children, and for girls, the associations of loneliness appear to be particularly strong.

School wellbeing among children in grades 1-10

Background

Determinants of children's school wellbeing have not been extensively studied. In this cross-sectional study of school children we assessed how factors assumed to promote wellbeing and factors assumed to adversely influence wellbeing were associated with self-reported wellbeing in school.

Methods

Children from five schools, 230 boys and 189 girls in grades 1-10, responded to the same set of questions. We used proportional odds logistic regression to assess the associations of promoting and restraining factors with school wellbeing.

Results

In a multivariable analysis, degree of school wellbeing in boys was strongly and positively related to enjoying school work (odds ratio, 3.84, 95% CI 2.38 to 6.22) and receiving necessary help (odds ratio, 3.55, 95% CI 2.17 to 5.80) from teachers. In girls, being bothered during lessons was strongly and negatively associated with school wellbeing (odds ratio, 0.43, 95% CI 0.22 to 0.85).

Conclusions

Different factors may determine school wellbeing in boys and girls, but for both genders, factors relevant for lessons may be more important than factors related to recess. Especially in boys, the student-teacher relationship may be of particular importance.