A method for fitting regression splines with varying polynomial order in the linear mixed model

Is High Body Mass Index Protective or Detrimental in Surgical Fixation of Hip Fractures?: A Spline Regression Analysis of 22,388 Patients

OBJECTIVE

We aimed to characterize the association between BMI as a continuous variable and 30-day postoperative outcomes following hip fracture surgery through (1) 30-day readmission and reoperation; (2) local wound-related; and (3) systemic complications.

METHODS

The National Surgical Quality Improvement Program database (January 2016-December 2019) was queried for patients undergoing hip fracture open reduction and internal fixation. Baseline patient demographics, comorbidities, and patient outcomes were recorded. Multivariable regression models accounted for baseline demographics, comorbidities, and fracture patterns. Significant associations were analyzed using spline regression models to evaluate the continuous association between BMI and the aforementioned outcomes.

RESULTS

Spline models demonstrated a U-shaped curve for the odds of 30-day readmission and 30-day reoperation with nadirs at the BMI of 27.5 and 22.0 kg/m 2 . The odd ratios of superficial infection, deep infection, any wound complication, and inability to weight bear on POD 1 rose progressively starting at a BMI of 25.6, 35.5, 25.6, and 32.7 kg/m 2 respectively. Odds of 30-day mortality, transfusion, pneumonia, and delirium were greatest at the lowest recorded BMI (11.9 kg/m 2 ).

CONCLUSION

BMI has a U-shaped association with 30-day readmission and reoperation. Conversely, the highest risk of mortality and systemic complications (transfusion, pneumonia, and delirium) were within the lower BMI range, with diminishing risk as BMI increased. Local wound complications and systemic sepsis exhibited a third unique pattern with progressive rise in odds as BMI increased. The odds of any complications demonstrated a U-shaped pattern with a nadir in the overweight to obese I categories, suggesting that patients may be at lowest risk within this range.

LEVEL OF EVIDENCE

Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.

Incidence and development of cholestasis in surgical neonates receiving an intravenous mixed-oil lipid emulsion

BACKGROUND

Intestinal failure-associated liver disease (IFALD), initially manifesting as cholestasis, is a complication in neonates receiving parenteral nutrition (PN). Soybean oil lipid emulsion (SOLE), though implicated in IFALD, was the only US Food and Drug Administration (FDA)-approved initial intravenous lipid emulsion (ILE) for infants and children in the United States. A mixed-oil lipid emulsion (MOLE) gained popularity in patients at risk for IFALD and was recently FDA approved as an initial ILE in children. Given the presence of soybean oil in MOLE, we hypothesized that MOLE would not be effective at preventing cholestasis in surgical neonates.

METHODS

Neonates with gastrointestinal surgical conditions necessitating PN for ≥14 days and receiving MOLE (SMOFlipid) from July 2016 to July 2019 were analyzed retrospectively. Unpaired and pair-matched historical surgical neonates treated with SOLE (Intralipid) served as controls. The primary outcome measure was development of cholestasis (direct bilirubin ≥2 mg/dl).

RESULTS

Overall, 63% (10 of 16) of MOLE patients and 22% (30 of 136) of SOLE patients developed cholestasis after ≥14 days of therapy (P = 0.005). The latency to developing cholestasis was significantly shorter in MOLE patients compared with SOLE patients.

CONCLUSION

In surgical neonates, MOLE may not prevent cholestasis and should not be considered hepatoprotective. Regardless of ILE source, all surgical neonates should be closely monitored for development of IFALD. To date, there is still no ILE able to prevent IFALD.

Risk assessment for health insurance using equation modeling and machine learning

Due to the advancement of medical sensor technologies new vectors can be added to the health insurance packages. Such medical sensors can help the health as well as the insurance sector to construct mathematical risk equation models with parameters that can map the real-life risk conditions. In this paper parameter analysis in terms of medical relevancy as well in terms of correlation has been done. Considering it as ‘inverse problem’ the mathematical relationship has been found and are tested against the ground truth between the risk indicators. The pairwise correlation analysis gives a stable mathematical equation model can be used for health risk analysis. The equation gives coefficient values from which classification regarding health insurance risk can be derived and quantified. The Logistic Regression equation model gives the maximum accuracy (86.32%) among the Ridge Bayesian and Ordinary Least Square algorithms. Machine learning algorithm based risk analysis approach was formulated and the series of experiments show that K-Nearest Neighbor classifier has the highest accuracy of 93.21% to do risk classification.

Turnover events of animal caretakers and its impact on productivity in swine farms

Attracting and retaining quality animal caretaking personnel is one of most pressing issues the US swine industry currently faces. On-farm employee turnover can be costly and have an impact on productivity. The primary objectives of this study were to describe the amount of animal caretaker turnover events that occurred in a single year in eleven swine farms, and to investigate associations between employee turnover events and two subsequent production parameters of interest: number of pigs weaned per sow (PWS) and pre-weaning mortality (PWM). A retrospective cohort study was conducted with eleven commercial farrow-to-wean swine farms belonging to two vertically integrated multi-site production systems within two swine production companies. Human resources and production data for the year of 2019 were obtained monthly from each farm. The primary predictor of interest was the occurrence of an employee turnover event, defined as 'voluntary' (employee decided to leave or quit) or 'involuntary' (employee was terminated by company decision). The primary outcomes of interest included the monthly average PWS and PWM. These associations were assessed with 1-, 2-, 3-, and 6-months between the turnover events and the outcome. Linear mixed effects models were fit in STATA 15, with system and farm included as random effects. To account for temporal and seasonal trends of production, season and the monthly production were included in the models. There were a total of 152 turnover events, with 4 and 148 turnover events in systems 1 and 2, respectively. The average turnover, calculated as a percent of total turnover among full time employee positions, was 92 % (SD = 62 %; Range = 8-217 %). Improved production efficiency in both monthly PWS (p = 0.01) and PWM (p = 0.02) was observed with the occurrence of an involuntary turnover event two months prior, after controlling for season, previous month production, farm, and system. For the PWS outcome, there was a significant interaction between an involuntary turnover event two months prior and monthly county-level unemployment rate (p = 0.02), indicative of the improved performance being most profound at the lowest levels of unemployment rate and diminishing at the highest levels. Turnover of animal caretaking personnel in farrow-to-wean farms was confirmed to be highly variable and high for the majority of farms in this study. Furthermore, animal caretaker turnover was associated with subsequent trends of production efficiency, warranting closer consideration of prioritizing managerial efforts in worker recruitment, training and retention.

A recipe for accurate estimation of lifespan brain trajectories, distinguishing longitudinal and cohort effects

We address the problem of estimating how different parts of the brain develop and change throughout the lifespan, and how these trajectories are affected by genetic and environmental factors. Estimation of these lifespan trajectories is statistically challenging, since their shapes are typically highly nonlinear, and although true change can only be quantified by longitudinal examinations, as follow-up intervals in neuroimaging studies typically cover less than 10% of the lifespan, use of cross-sectional information is necessary. Linear mixed models (LMMs) and structural equation models (SEMs) commonly used in longitudinal analysis rely on assumptions which are typically not met with lifespan data, in particular when the data consist of observations combined from multiple studies. While LMMs require a priori specification of a polynomial functional form, SEMs do not easily handle data with unstructured time intervals between measurements. Generalized additive mixed models (GAMMs) offer an attractive alternative, and in this paper we propose various ways of formulating GAMMs for estimation of lifespan trajectories of 12 brain regions, using a large longitudinal dataset and realistic simulation experiments. We show that GAMMs are able to more accurately fit lifespan trajectories, distinguish longitudinal and cross-sectional effects, and estimate effects of genetic and environmental exposures. Finally, we discuss and contrast questions related to lifespan research which strictly require repeated measures data and questions which can be answered with a single measurement per participant, and in the latter case, which simplifying assumptions that need to be made. The examples are accompanied with R code, providing a tutorial for researchers interested in using GAMMs.

Surface electromyographic evaluation of the neuromuscular activation of the inspiratory muscles during progressively increased inspiratory flow under inspiratory-resistive loading

This study aimed to evaluate neuromuscular activation in the scalene and sternocleidomastoid muscles using surface electromyography (EMG) during progressively increased inspiratory flow, produced by increasing the respiratory rate under inspiratory-resistive loading using a mask ventilator. Moreover, we attempted to identify the EMG inflection point (EMG^IP) on the graph, at which the root mean square (RMS) of the EMG signal values of the inspiratory muscles against the inspiratory flow velocity acceleration abruptly increases, similarly to the EMG anaerobic threshold (EMG^AT) reported during incremental-resistive loading in other skeletal muscles. We measured neuromuscular activation of healthy male subjects and found that the inspiratory flow velocity increased by approximately 1.6-fold. We successfully observed an increase in RMS that corresponded to inspiratory flow acceleration with ρ ≥ 0.7 (Spearman's rank correlation) in 17 of 27 subjects who completed the experimental protocol. To identify EMG^IP, we analyzed the fitting to either a straight or non-straight line related to the increasing inspiratory flow and RMS using piecewise linear spline functions. As a result, EMG^IP was identified in the scalene and sternocleidomastoid muscles of 17 subjects. We believe that the identification of EMG^IP in this study infers the existence of EMG^AT in inspiratory muscles. Application of surface EMG, followed by identification of EMG^IP, for evaluating the neuromuscular activation of respiratory muscles may be allowed to estimate the signs of the respiratory failure, including labored respiration, objectively and non-invasively accompanied using accessory muscles in clinical respiratory care.

Evolution of respiratory function in Duchenne muscular dystrophy from childhood to adulthood

In Duchenne muscular dystrophy (DMD), it is still to be determined if specific timepoints can be identified during the natural evolution of respiratory dysfunction from childhood to adulthood and if scoliosis, steroid therapy and nocturnal noninvasive mechanical ventilation (NIMV) have any effect on it.In a 7-year retrospective study performed on 115 DMD patients (6-24 years), evaluated once or twice per year, with 574 visits in total, evolution mean curves of spirometry, lung volumes, spontaneous breathing and thoraco-abdominal pattern (measured by optoelectronic plethysmography) parameters were obtained by nonlinear regression model analysis.While predicted values of forced vital capacity, forced expiratory volume in 1 s, and peak expiratory flow decline continuously since childhood, during spontaneous breathing the following parameters become significantly different than normal in sequence: abdominal contribution to tidal volume (lower after 14.8 years), tidal volume (lower after 17.2 years), minute ventilation (lower after 18.1 years) and respiratory rate (higher after 22.1 years). Restrictive lung pattern and diaphragmatic impairment are exacerbated by scoliosis severity, slowed by steroids treatment and significantly affected by NIMV.Spirometry, lung volumes, breathing pattern and thoraco-abdominal contributions show different evolution curves over time. Specific timepoints of respiratory impairment are identified during disease progression. These should be considered when defining outcome measures in clinical trials and treatment strategies in DMD.

Blood kinetics of Ebola virus in survivors and nonsurvivors

BACKGROUND

Infection with Ebola virus (EBOV) results in a life-threatening disease, with reported mortality rates between 50%-70%. The factors that determine patient survival are poorly understood; however, clinical observations indicate that EBOV viremia may be associated with fatal outcome. We conducted a study of the kinetics of Zaire EBOV viremia in patients with EBOV disease (EVD) who were managed at an Ebola Treatment Centre in Sierra Leone during the recent West African outbreak.

METHODS

Data from 84 EVD patients (38 survivors, 46 nonsurvivors) were analyzed, and EBOV viremia was quantified between 2 and 13 days after symptom onset. Time since symptom onset and clinical outcome were used as independent variables to compare EBOV viral kinetics in survivors and nonsurvivors.

RESULTS

In all patients, EBOV viremia kinetics was a quadratic function of time; however, EBOV viremia was 0.94 logarithm (log) copies per ml (cp/ml) (P = 0.011) higher in nonsurvivors than in survivors from day 2 after the onset of symptoms. Survivors reached peak viremia levels at an earlier time after symptom onset than nonsurvivors (day 5 versus day 7) and had lower mean peak viremia levels compared with nonsurvivors (7.46 log cp/ml; 95% CI, 7.17-7.76 vs. 8.60 log cp/ml; 95% CI, 8.27-8.93). Before reaching peak values, EBOV viremia similarly increased both in survivors and nonsurvivors; however, the decay of viremia after the peak was much stronger in survivors than in nonsurvivors.

CONCLUSION

Our results demonstrate that plasma concentrations of EBOV are markedly different between survivors and nonsurvivors at very early time points after symptom onset and may be predicative of outcome. Further studies focused on the early phase of the disease will be required to identify the causal and prognostic factors that determine patient outcome.

FUNDING

Italian Ministry of Health; Italian Ministry of Foreign Affairs; EMERGENCY's private donations; and Royal Engineers for DFID-UK.

Variation in maternal urinary cortisol profiles across the peri-conceptional period: a longitudinal description and evaluation of potential functions

STUDY QUESTION

How do women's first morning urinary cortisol levels, a marker of stress axis activity, vary during the peri-conceptional period (the 12 weeks around conception)?

SUMMARY ANSWER

First morning urinary cortisol follows an overall increasing trajectory across the peri-conceptional period, interrupted by 2 week-long decreases during the week preceding conception and the fifth week following conception.

WHAT IS KNOWN ALREADY

Later gestational stages (i.e. second and third trimesters) are characterized by increasing levels of circulating cortisol. This increase is hypothesized to constitute a response to the energy demands imposed by fetal growth, and the development of energy reserves in preparation for nursing and performing regular activities while carrying pregnancy's extra weight and volume.

STUDY DESIGN, SIZE, DURATION

This study is based on a data set collected as part of a longitudinal, naturalistic investigation into the interactions between the stress (hypothalamic-pituitary-adrenal axis (HPAA)) and reproductive (hypothalamic-pituitary-gonadal axis (HPGA)) axes. Biomarkers of HPAA and HPGA function were quantified in first morning urinary specimens collected every other day from 22 healthy women who conceived a pregnancy during the study. We analyzed the longitudinal within- and between-individual variation in first morning urinary cortisol levels across the 12-week peri-conceptional period.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Participants were recruited from two rural, aboriginal, neighboring communities in Guatemala. Cortisol, estradiol and progesterone metabolites (estrone-3-glucuronide and pregnanediol glucuronide, respectively) and hCG levels were quantified in first morning urinary specimens using immunoassays to determine time of conception and confirm pregnancy maintenance. Linear mixed-effects models with regression splines were used to evaluate the magnitude and significance of changes in cortisol trajectories.

MAIN RESULTS AND THE ROLE OF CHANCE

Overall, maternal first morning urinary cortisol increased from 6 weeks prior to conception (geometric mean ± SD = 58.14 ± 36.00 ng/ml) to 6 weeks post-conception (89.29 ± 46.76 ng/ml). The magnitude of the increase between the pre- and post-conception periods varied significantly between women (likelihood ratio test statistic = 8.0017, P = 0.005). The peri-conceptional period is characterized by an increasing cortisol trajectory (+1.36% per day; P = 0.007) interrupted by a week-long decline immediately prior to conception (-4.02% per day; P = 0.0013). After conception cortisol increased again (+1.73% per day; P = 0.0008) for 4 weeks, fell in the fifth week (-6.60% per day; P = 0.0002) and increased again in post-conceptional week 6 (+8.86% per day; P = 0.002). Maternal urinary cortisol levels varied with sex of the gestating embryo. During gestational week 2, mothers carrying female embryos (N = 10) had higher mean cortisol levels than those carrying male embryos (N = 9) (t(17) = 2.28, P = 0.04).

LIMITATIONS, REASONS FOR CAUTION

Our results are based on a relatively small sample (n = 22) of women. However, our repeated-measures design with an average of 27 ± 8 (mean ± SD) data points per woman strengthens the precision of estimates resulting in high statistical power. Additionally, our study population's high degree of ethnic and cultural homogeneity reduces the effects of confounders compared with those found in industrialized populations. This higher level of homogeneity also increases our statistical power. However, since there may be small differences in absolute cortisol values among ethnic groups, the social and biological background of our sample may affect the generalizability of our results. General patterns of HPAA activity, however, are expected to be universal across women. Finally, as there is, to the best of our knowledge, no evidence to the contrary, we assumed that urinary cortisol levels reflect HPAA activity and that changes in gonadal steroids across the menstrual cycle do not affect the levels of free cortisol measured in urine.

WIDER IMPLICATIONS OF THE FINDINGS

To our knowledge, this is the first longitudinal profile of basal maternal HPAA activity across the peri-conceptional period. A basic understanding of the normative (basal as opposed to stress-induced) changes in HPAA activity across this period is needed to accurately assess women's stress at this juncture. Importantly, changes in HPAA activity are likely to play a critical role in ovulation, fertilization, implantation, placentation and embryonic programing. Thus, this novel information should aid in the development of interventions aimed at preventing or moderating undesired effects of maternal physiological stress during the peri-conceptional period on reproductive outcomes as well as embryonic development.

STUDY FUNDING/COMPETING INTERESTS

This research was funded by a CIHR IGH Open Operating grant (CIHR 106705) to P.A.N. and L.Z.; a Simon Fraser University (SFU) President's Start-up grant, a Community Trust Endowment Fund grant through SFU's Human Evolutionary Studies Program and a Michael Smith Foundation for Health Research Career Investigator Scholar Award to P.A.N.; an NSERC Discovery grant to L.Z.; a CIHR Post-Doctoral Fellowship to C.K.B. and an NSERC Undergraduate Student Research Award to H.M. and J.C.B. The funding agencies had no role in the design, analysis, interpretation or reporting of the findings. There are no competing interests.

TRIAL REGISTRATION NUMBER

Not applicable.

An analysis of 24-h ambulatory blood pressure monitoring data using orthonormal polynomials in the linear mixed model

BACKGROUND

The use of 24-h ambulatory blood pressure monitoring (ABPM) in clinical practice and observational epidemiological studies has grown considerably in the past 25 years. ABPM is a very effective technique for assessing biological, environmental, and drug effects on blood pressure.

OBJECTIVES

In order to enhance the effectiveness of ABPM for clinical and observational research studies using analytical and graphical results, developing alternative data analysis approaches using modern statistical techniques are important.

METHODS

The linear mixed model for the analysis of longitudinal data is particularly well suited for the estimation of, inference about, and interpretation of both population (mean) and subject-specific trajectories for ABPM data. We propose using a linear mixed model with orthonormal polynomials across time in both the fixed and random effects to analyze ABPM data.

RESULTS

We demonstrate the proposed analysis technique using data from the Dietary Approaches to Stop Hypertension (DASH) study, a multicenter, randomized, parallel arm feeding study that tested the effects of dietary patterns on blood pressure.

CONCLUSION

The linear mixed model is relatively easy to implement (given the complexity of the technique) using available software, allows for straightforward testing of multiple hypotheses, and the results can be presented to research clinicians using both graphical and tabular displays. Using orthonormal polynomials provides the ability to model the nonlinear trajectories of each subject with the same complexity as the mean model (fixed effects).

A circular LEAR correlation structure for cyclical longitudinal data

Circular covariance patterns arise naturally from many important biological and physical processes. Modelling these patterns can be immensely important for proper analyses. In this article, we propose a circular linear exponent autoregressive (LEAR) correlation structure for cyclical longitudinal data. Special cases of this parsimonious correlation model include the equal correlation and first-order moving average (MA(1)) correlation structures and a circular analogue of the continuous-time AR(1) model. We discuss properties and estimation of the circular LEAR model in the context of cyclical longitudinal data concerning diet and hypertension (the DASH study). Analysis of these data exemplifies the benefits of the circular LEAR correlation structure.

Model choice can obscure results in longitudinal studies

BACKGROUND

This article examines how different parameterizations of age and time in modeling observational longitudinal data can affect results.

METHODS

When individuals of different ages at study entry are considered, it becomes necessary to distinguish between longitudinal and cross-sectional differences to overcome possible selection biases.

RESULTS

Various models were fitted using data from longitudinal studies with participants with different ages and different follow-up lengths. Decomposing age into two components-age at entry into the study (first age) and the longitudinal follow-up (time) compared with considering age alone-leads to different conclusions.

CONCLUSIONS

In general, models using both first age and time terms performed better, and these terms are usually necessary to correctly analyze longitudinal data.

Identifying patterns in treatment response profiles in acute bipolar mania: a cluster analysis approach

Patients with acute mania respond differentially to treatment and, in many cases, fail to obtain or sustain symptom remission. The objective of this exploratory analysis was to characterize response in bipolar disorder by identifying groups of patients with similar manic symptom response profiles.

METHODS

Patients (n = 222) were selected from a randomized, double-blind study of treatment with olanzapine or divalproex in bipolar I disorder, manic or mixed episode, with or without psychotic features. Hierarchical clustering based on Ward's distance was used to identify groups of patients based on Young-Mania Rating Scale (YMRS) total scores at each of 5 assessments over 7 weeks. Logistic regression was used to identify baseline predictors for clusters of interest.

RESULTS

Four distinct clusters of patients were identified: Cluster 1 (n = 64): patients did not maintain a response (YMRS total scores < or = 12); Cluster 2 (n = 92): patients responded rapidly (within less than a week) and response was maintained; Cluster 3 (n = 36): patients responded rapidly but relapsed soon afterwards (YMRS > or = 15); Cluster 4 (n = 30): patients responded slowly (> or = 2 weeks) and response was maintained. Predictive models using baseline variables found YMRS Item 10 (Appearance), and psychosis to be significant predictors for Clusters 1 and 4 vs. Clusters 2 and 3, but none of the baseline characteristics allowed discriminating between Clusters 1 vs. 4. Experiencing a mixed episode at baseline predicted membership in Clusters 2 and 3 vs. Clusters 1 and 4. Treatment with divalproex, larger number of previous manic episodes, lack of disruptive-aggressive behavior, and more prominent depressive symptoms at baseline were predictors for Cluster 3 vs. 2.

CONCLUSION

Distinct treatment response profiles can be predicted by clinical features at baseline. The presence of these features as potential risk factors for relapse in patients who have responded to treatment should be considered prior to discharge.

TRIAL REGISTRATION

The clinical trial cited in this report has not been registered because it was conducted and completed prior to the inception of clinical trial registries.

Amplified transmission of HIV-1: comparison of HIV-1 concentrations in semen and blood during acute and chronic infection

OBJECTIVES

This study was conducted to compare viral dynamics in blood and semen between subjects with antibody negative, acute HIV-1 infection and other subjects with later stages of infection.

DESIGN

A prospective cohort study was embedded within a cross-sectional study of HIV screening in a Lilongwe, Malawi STD clinic.

METHODS

Blood samples from HIV antibody negative or indeterminate volunteers were used to detect HIV RNA in plasma using a pooling strategy. Blood and seminal plasma HIV-1 RNA concentrations were measured over 16 weeks.

RESULTS

Sixteen men with acute HIV infection and 25 men with chronic HIV infection were studied. Blood viral load in subjects with acute HIV infection was highest about 17 days after infection (mean +/- SE, 6.9 +/- 0.5 log10 copies/ml), while semen viral load peaked about 30 days after infection (4.5 +/- 0.4 log10 copies/ml). Semen viral load declined by 1.7 log10 to a nadir by week 10 of HIV infection. Semen and blood viral loads were more stable in chronically infected subjects over 16 weeks. Higher semen levels of HIV RNA were noted in subjects with low CD4 cell counts.

CONCLUSIONS

These results provide a biological explanation for reported increases in HIV transmission during the very early (acute) and late stages of infection. Recognizing temporal differences in HIV shedding in the genital tract is important in the development of effective HIV prevention strategies.