Statistical inference for ordinal predictors in generalized additive models with application to Bronchopulmonary Dysplasia.

OBJECTIVE: Discrete but ordered covariates are quite common in applied statistics, and some regularized fitting procedures have been proposed for proper handling of ordinal predictors in statistical models. Motivated by a study from neonatal medicine on Bronchopulmonary Dysplasia (BPD), we show how quadratic penalties on adjacent dummy coefficients of ordinal factors proposed in the literature can be incorporated in the framework of generalized additive models, making tools for statistical inference developed there available for ordinal predictors as well. RESULTS: The approach presented allows to exploit the scale level of ordinally scaled factors in a sound statistical framework. Furthermore, several ordinal factors can be considered jointly without the need to collapse levels even if the number of observations per level is small. By doing so, results obtained earlier on the BPD data analyzed could be confirmed.

Protein intake and outcome of critically ill patients: analysis of a large international database using piece-wise exponential additive mixed models.

BACKGROUND: Proteins are an essential part of medical nutrition therapy in critically ill patients. Guidelines almost universally recommend a high protein intake without robust evidence supporting its use. METHODS: Using a large international database, we modelled associations between the hazard rate of in-hospital death and live hospital discharge (competing risks) and three categories of protein intake (low: < 0.8 g/kg per day, standard: 0.8-1.2 g/kg per day, high: > 1.2 g/kg per day) during the first 11 days after ICU admission (acute phase). Time-varying cause-specific hazard ratios (HR) were calculated from piece-wise exponential additive mixed models. We used the estimated model to compare five different hypothetical protein diets (an exclusively low protein diet, a standard protein diet administered early (day 1 to 4) or late (day 5 to 11) after ICU admission, and an early or late high protein diet). RESULTS: Of 21,100 critically ill patients in the database, 16,489 fulfilled inclusion criteria for the analysis. By day 60, 11,360 (68.9%) patients had been discharged from hospital, 4,192 patients (25.4%) had died in hospital, and 937 patients (5.7%) were still hospitalized. Median daily low protein intake was 0.49 g/kg [IQR 0.27-0.66], standard intake 0.99 g/kg [IQR 0.89- 1.09], and high intake 1.41 g/kg [IQR 1.29-1.60]. In comparison with an exclusively low protein diet, a late standard protein diet was associated with a lower hazard of in-hospital death: minimum 0.75 (95% CI 0.64, 0.87), and a higher hazard of live hospital discharge: maximum HR 1.98 (95% CI 1.72, 2.28). Results on hospital discharge, however, were qualitatively changed by a sensitivity analysis. There was no evidence that an early standard or a high protein intake during the acute phase was associated with a further improvement of outcome. CONCLUSIONS: Provision of a standard protein intake during the late acute phase may improve outcome compared to an exclusively low protein diet. In unselected critically ill patients, clinical outcome may not be improved by a high protein intake during the acute phase. Study registration ID number ISRCTN17829198.

Reported and Recorded Sleepiness in Obesity and Depression.

BACKGROUND: Obesity and depression are both associated with changes in sleep/wake regulation, with potential implications for individualized treatment especially in comorbid individuals suffering from both. However, the associations between obesity, depression, and subjective, questionnaire-based and objective, EEG-based measurements of sleepiness used to assess disturbed sleep/wake regulation in clinical practice are not well known. OBJECTIVES: The study investigates associations between sleep/wake regulation measures based on self-reported subjective questionnaires and EEG-derived measurements of sleep/wake regulation patterns with depression and obesity and how/whether depression and/or obesity affect associations between such self-reported subjective questionnaires and EEG-derived measurements. METHODS: Healthy controls (HC, NHC = 66), normal-weighted depressed (DEP, NDEP = 16), non-depressed obese (OB, NOB = 68), and obese depressed patients (OBDEP, NOBDEP = 43) were included from the OBDEP (Obesity and Depression, University Leipzig, Germany) study. All subjects completed standardized questionnaires related to daytime sleepiness (ESS), sleep quality and sleep duration once as well as questionnaires related to situational sleepiness (KSS, SSS, VAS) before and after a 20 min resting state EEG in eyes-closed condition. EEG-based measurements of objective sleepiness were extracted by the VIGALL algorithm. Associations of subjective sleepiness with objective sleepiness and moderating effects of obesity, depression, and additional confounders were investigated by correlation analyses and regression analyses. RESULTS: Depressed and non-depressed subgroups differed significantly in most subjective sleepiness measures, while obese and non-obese subgroups only differed significantly in few. Objective sleepiness measures did not differ significantly between the subgroups. Moderating effects of obesity and/or depression on the associations between subjective and objective measures of sleepiness were rarely significant, but associations between subjective and objective measures of sleepiness in the depressed subgroup were systematically weaker when patients comorbidly suffered from obesity than when they did not. CONCLUSION: This study provides some evidence that both depression and obesity can affect the association between objective and subjective sleepiness. If confirmed, this insight may have implications for individualized diagnosis and treatment approaches in comorbid depression and obesity.

Calorie intake and short-term survival of critically ill patients.

BACKGROUND & AIMS: The association between calorie supply and outcome of critically ill patients is unclear. Results from observational studies contradict findings of randomized studies, and have been questioned because of unrecognized confounding by indication. The present study wanted to re-examine the associations between the daily amount of calorie intake and short-term survival of critically ill patients using several novel statistical approaches. METHODS: 9661 critically ill patients from 451 ICUs were extracted from an international database. We examined associations between survival time and three pragmatic nutritional categories (I: <30% of target, II: 30-70%, III: >70%) reflecting different amounts of total daily calorie intake. We compared hazard ratios for the 30-day risk of dying estimated for different hypothetical nutrition support plans (different categories of daily calorie intake during the first 11 days after ICU admission). To minimize indication bias, we used a lag time between nutrition and outcome, we particularly considered daily amounts of calorie intake, and we adjusted results to the route of calorie supply (enteral, parenteral, oral). RESULTS: 1974 patients (20.4%) died in hospital before day 30. Median of daily artificial calorie intake was 1.0 kcal/kg [IQR 0.0-4.1] in category I, 12.3 kcal/kg [9.4-15.4] in category II, and 23.5 kcal/kg [19.5-27.8] in category III. When compared to a plan providing daily minimal amounts of calories (category I), the adjusted minimal hazard ratios for a delayed (from day 5-11) or an early (from day 1-11) mildly hypocaloric nutrition (category II) were 0.71 (95% confidence interval [CI], 0.54 to 0.94) and 0.56 (95% CI, 0.38 to 0.82), respectively. No substantial hazard change could be detected, when a delayed or an early, near target calorie intake (category III) was compared to an early, mildly hypocaloric nutrition. CONCLUSIONS: Compared to a severely hypocaloric nutrition, a mildly hypocaloric nutrition is associated with a decreased risk of death. In unselected critically ill patients, this risk cannot be reduced further by providing amounts of calories close to the calculated target. STUDY REGISTRATION: ID number ISRCTN17829198, website http://www.isrctn.org.

Penalized estimation of complex, non-linear exposure-lag-response associations.

We propose a novel approach for the flexible modeling of complex exposure-lag-response associations in time-to-event data, where multiple past exposures within a defined time window are cumulatively associated with the hazard. Our method allows for the estimation of a wide variety of effects, including potentially smooth and smoothly time-varying effects as well as cumulative effects with leads and lags, taking advantage of the inference methods that have recently been developed for generalized additive mixed models. We apply our method to data from a large observational study of intensive care patients in order to analyze the association of both the timing and the amount of artificial nutrition with the short term survival of critically ill patients. We evaluate the properties of the proposed method by performing extensive simulation studies and provide a systematic comparison with related approaches.

Why some obese people become depressed whilst others do not: exploring links between cognitive reactivity, depression and obesity.

Obesity and depression are two major public health concerns, particularly when they co-occur. To date, mechanisms underlying this association have not been established and it is unclear why some obese people become depressed whilst others do not. However, considering the strong association between depression and cognitive reactivity (CR), the present study explores possible associations between obesity, depression and CR in light of the scarce and conflicting nature of past literature. 254 participants were included for measures of depression, CR and obesity. Multivariate analyses of covariance examined the effects of depression and obesity as well as interaction effects of depression x obesity controlling for age and gender. Directions of effects were analysed by means of regression analyses and group contrasts. Linear analyses revealed (1) a significant effect of obesity on the rumination (RUM) and control/perfectionism subscales of CR, (2) a significant effect of depression on CR and all of its subscales, and (3) a significant interaction effect between obesity x depression on RUM. Results may support the 'Jolly Fat Hypothesis' and the existence of a psychologically protected subgroup of obese patients characterised by a lower ruminative thinking style and fewer depressive symptoms. Thus, incorporating anti-rumination therapy into treatment for obese individuals may be beneficial to prevent the development of comorbid depression.

An experimental test of the habitat-amount hypothesis for saproxylic beetles in a forested region.

The habitat-amount hypothesis challenges traditional concepts that explain species richness within habitats, such as the habitat-patch hypothesis, where species number is a function of patch size and patch isolation. It posits that effects of patch size and patch isolation are driven by effects of sample area, and thus that the number of species at a site is basically a function of the total habitat amount surrounding this site. We tested the habitat-amount hypothesis for saproxylic beetles and their habitat of dead wood by using an experiment comprising 190 plots with manipulated patch sizes situated in a forested region with a high variation in habitat amount (i.e., density of dead trees in the surrounding landscape). Although dead wood is a spatio-temporally dynamic habitat, saproxylic insects have life cycles shorter than the time needed for habitat turnover and they closely track their resource. Patch size was manipulated by adding various amounts of downed dead wood to the plots (~800 m³ in total); dead trees in the surrounding landscape (~240 km2 ) were identified using airborne laser scanning (light detection and ranging). Over 3 yr, 477 saproxylic species (101,416 individuals) were recorded. Considering 20-1,000 m radii around the patches, local landscapes were identified as having a radius of 40-120 m. Both patch size and habitat amount in the local landscapes independently affected species numbers without a significant interaction effect, hence refuting the island effect. Species accumulation curves relative to cumulative patch size were not consistent with either the habitat-patch hypothesis or the habitat-amount hypothesis: several small dead-wood patches held more species than a single large patch with an amount of dead wood equal to the sum of that of the small patches. Our results indicate that conservation of saproxylic beetles in forested regions should primarily focus on increasing the overall amount of dead wood without considering its spatial arrangement. This means dead wood should be added wherever possible including in local landscapes with low or high dead-wood amounts. For species that have disappeared from most forests owing to anthropogenic habitat degradation, this should, however, be complemented by specific conservation measures pursued within their extant distributional ranges.

A penalized framework for distributed lag non-linear models.

Distributed lag non-linear models (DLNMs) are a modelling tool for describing potentially non-linear and delayed dependencies. Here, we illustrate an extension of the DLNM framework through the use of penalized splines within generalized additive models (GAM). This extension offers built-in model selection procedures and the possibility of accommodating assumptions on the shape of the lag structure through specific penalties. In addition, this framework includes, as special cases, simpler models previously proposed for linear relationships (DLMs). Alternative versions of penalized DLNMs are compared with each other and with the standard unpenalized version in a simulation study. Results show that this penalized extension to the DLNM class provides greater flexibility and improved inferential properties. The framework exploits recent theoretical developments of GAMs and is implemented using efficient routines within freely available software. Real-data applications are illustrated through two reproducible examples in time series and survival analysis.

Thrombus Histology Suggests Cardioembolic Cause in Cryptogenic Stroke.

BACKGROUND AND PURPOSE: Ischemic stroke of undetermined cause is a major health issue because of its high frequency and clinical relevance. Histopathologic analysis of human thrombi, retrieved from stroke patients with large-vessel occlusion during mechanical thrombectomy, may provide information about underlying pathologies. This study examines the relationship between stroke causes and histological clot composition to identify specific patterns that might help to distinguish causes of cryptogenic stroke. METHODS: Thrombi of 145 consecutive stroke patients with large-vessel occlusion were collected during intracranial mechanical recanalization. The hematoxylin and eosin-stained specimens were quantitatively analyzed in terms of the relative fractions of the main constituents (red and white blood cells and fibrin/platelets). These data, along with additional clinical and interventional parameters, were compared for different stroke subtypes, as defined by the international Trial of Org 10172 in Acute Stroke Treatment criteria. RESULTS: The composition of thrombi from cardioembolic and noncardioembolic stroke patients differed significantly for all main thrombus components. Cardioembolic thrombi had higher proportions of fibrin/platelets (P=0.009), less erythrocytes (P=0.003), and more leucocytes (P=0.035) than noncardioembolic thrombi. Cryptogenic strokes showed strong overlap with cardioembolic strokes but not with noncardioembolic strokes, in terms of both thrombus histology and interventional and clinical outcome parameters. CONCLUSIONS: Quantitative evaluation of thrombus composition may help to distinguish between different stroke causes. Our findings support the notion that the majority of cryptogenic strokes are cardioembolic.

Functional Additive Mixed Models.

We propose an extensive framework for additive regression models for correlated functional responses, allowing for multiple partially nested or crossed functional random effects with flexible correlation structures for, e.g., spatial, temporal, or longitudinal functional data. Additionally, our framework includes linear and nonlinear effects of functional and scalar covariates that may vary smoothly over the index of the functional response. It accommodates densely or sparsely observed functional responses and predictors which may be observed with additional error and includes both spline-based and functional principal component-based terms. Estimation and inference in this framework is based on standard additive mixed models, allowing us to take advantage of established methods and robust, flexible algorithms. We provide easy-to-use open source software in the pffr() function for the R-package refund. Simulations show that the proposed method recovers relevant effects reliably, handles small sample sizes well and also scales to larger data sets. Applications with spatially and longitudinally observed functional data demonstrate the flexibility in modeling and interpretability of results of our approach.

Functional Generalized Additive Models.

We introduce the functional generalized additive model (FGAM), a novel regression model for association studies between a scalar response and a functional predictor. We model the link-transformed mean response as the integral with respect to t of F{X(t), t} where F(·,·) is an unknown regression function and X(t) is a functional covariate. Rather than having an additive model in a finite number of principal components as in Müller and Yao (2008), our model incorporates the functional predictor directly and thus our model can be viewed as the natural functional extension of generalized additive models. We estimate F(·,·) using tensor-product B-splines with roughness penalties. A pointwise quantile transformation of the functional predictor is also considered to ensure each tensor-product B-spline has observed data on its support. The methods are evaluated using simulated data and their predictive performance is compared with other competing scalar-on-function regression alternatives. We illustrate the usefulness of our approach through an application to brain tractography, where X(t) is a signal from diffusion tensor imaging at position, t, along a tract in the brain. In one example, the response is disease-status (case or control) and in a second example, it is the score on a cognitive test. R code for performing the simulations and fitting the FGAM can be found in supplemental materials available online.

Clinical effectiveness of hymenoptera venom immunotherapy: a prospective observational multicenter study of the European academy of allergology and clinical immunology interest group on insect venom hypersensitivity.

BACKGROUND: Treatment failure during venom immunotherapy (VIT) may be associated with a variety of risk factors. OBJECTIVE: Our aim was to evaluate the association of baseline serum tryptase concentration (BTC) and of other parameters with the frequency of VIT failure during the maintenance phase. METHODS: In this observational prospective multicenter study, we followed 357 patients with established honey bee or vespid venom allergy after the maintenance dose of VIT had been reached. In all patients, VIT effectiveness was either verified by sting challenge (n = 154) or patient self-reporting of the outcome of a field sting (n = 203). Data were collected on BTC, age, gender, preventive use of anti-allergic drugs (oral antihistamines and/or corticosteroids) right after a field sting, venom dose, antihypertensive medication, type of venom, side effects during VIT, severity of index sting reaction preceding VIT, and duration of VIT. Relative rates were calculated with generalized additive models. RESULTS: 22 patients (6.2%) developed generalized symptoms during sting challenge or after a field sting. A strong association between the frequency of VIT failure and BTC could be excluded. Due to wide confidence bands, however, weaker effects (odds ratios <3) of BTC were still possible, and were also suggested by a selective analysis of patients who had a sting challenge. The most important factor associated with VIT failure was a honey bee venom allergy. Preventive use of anti-allergic drugs may be associated with a higher protection rate. INTERPRETATION: It is unlikely that an elevated BTC has a strong negative effect on the rate of treatment failures. The magnitude of the latter, however, may depend on the method of effectiveness assessment. Failure rate is higher in patients suffering from bee venom allergy.

Quantification of leukocyte genomic 5-methylcytosine levels reveals epigenetic plasticity in healthy adult cloned cattle.

Successful somatic cell nuclear transfer (SCNT) requires epigenetic reprogramming of a differentiated donor cell nucleus. Incorrect reprogramming of epigenetic markings such as DNA methylation is associated with compromised prenatal development and postnatal abnormalities. Clones that survive into adulthood, in contrast, are assumed to possess a normalized epigenome corresponding to their normal phenotype. To address this point, we used capillary electrophoresis to measure 5-methylcytosine (5mC) levels in leukocyte DNA of 38 healthy female bovine clones that represented five genotypes from the Simmental breed and four genotypes from the Holstein breed. The estimated variance in 5mC level within clone genotypes of both breeds [0.104, 95% confidence interval (CI): 0.070-0.168] was higher than between clone genotypes (0, CI: 0-0.047). We quantified the contribution of SCNT to this unexpected variability by comparing the 19 Simmental clones with 12 female Simmental monozygotic twin pairs of similar age. In Simmental clones, the estimated variability within genotype (0.0636, CI: 0.0358-0.127) was clearly higher than in twin pairs (0.0091, CI: 0.0047-0.0229). In clones, variability within genotype (0.0636) was again higher than between genotypes (0, CI: 0-0.077). Twins, in contrast, showed lower variability within genotypes (0.0091) than between genotypes (0.0136, CI: 0.00250-0.0428). Importantly, the absolute deviations of 5mC values of individual SCNT clones from their genotype means were fivefold increased in comparison to twins. Further comparisons with noncloned controls revealed DNA hypermethylation in most of the clones. The clone-specific variability in DNA methylation and DNA hypermethylation clearly show that healthy adult SCNT clones must be considered as epigenome variants.

Predictors of side effects during the buildup phase of venom immunotherapy for Hymenoptera venom allergy: the importance of baseline serum tryptase.

BACKGROUND: Severe side effects during venom immunotherapy (VIT) are associated with a variety of risk factors. OBJECTIVE: Our aim was to evaluate the association of baseline serum tryptase concentration (BTC) and of other parameters, which are routinely recorded during patient evaluation, with the frequency of severe reactions requiring an emergency intervention during the buildup phase of VIT. METHODS: In this observational prospective multicenter study, we enrolled 680 patients with established honeybee or vespid venom allergy who underwent VIT. Data were collected on tryptase concentration, age, sex, culprit insect, cardiovascular medication, degree of preceding sting reaction, preventive antiallergic medication before therapy, time between last preceding sting reaction and VIT, venom specific IgE concentration, and type of buildup procedure. Relative rates were calculated with generalized additive models. RESULTS: Fifty-seven patients (8.4%) required an emergency intervention during buildup because of a severe systemic reaction. The frequency of interventions increased significantly with higher BTC (log-linear association; adjusted odds ratio, 1.56; 95% CI, 1.15-2.11; P < .005). The predictive power of BTC was markedly greater when VIT was performed for vespid venom allergy than for bee venom (for bee VIT, no significant association; for vespid VIT, log-linear association; adjusted odds ratio, 2.33; 95% CI, 1.28-4.26; P = .005). The most important other factor significantly associated with severe reactions during the buildup phase of VIT was bee venom allergy. CONCLUSION: Before vespid VIT, measurement of baseline serum tryptase concentration should be used to identify patients with a high risk for side effects. Patients with bee venom allergy require a particularly high degree of surveillance during VIT.

Predictors of severe systemic anaphylactic reactions in patients with Hymenoptera venom allergy: importance of baseline serum tryptase-a study of the European Academy of Allergology and Clinical Immunology Interest Group on Insect Venom Hypersensitivity.

BACKGROUND: Severe anaphylaxis to honeybee or vespid stings is associated with a variety of risk factors, which are poorly defined. OBJECTIVE: Our aim was to evaluate the association of baseline serum tryptase concentrations and other variables routinely recorded during patient evaluation with the frequency of past severe anaphylaxis after a field sting. METHODS: In this observational multicenter study, we enrolled 962 patients with established bee or vespid venom allergy who had a systemic reaction after a field sting. Data were collected on tryptase concentration, age, sex, culprit insect, cardiovascular medication, and the number of preceding minor systemic reactions before the index field sting. A severe reaction was defined as anaphylactic shock, loss of consciousness, or cardiopulmonary arrest. The index sting was defined as the hitherto first, most severe systemic field-sting reaction. Relative rates were calculated with generalized additive models. RESULTS: Two hundred six (21.4%) patients had a severe anaphylactic reaction after a field sting. The frequency of this event increased significantly with higher tryptase concentrations (nonlinear association). Other factors significantly associated with severe reactions after a field sting were vespid venom allergy, older age, male sex, angiotensin-converting enzyme inhibitor medication, and 1 or more preceding field stings with a less severe systemic reaction. CONCLUSION: In patients with honeybee or vespid venom allergy, baseline serum tryptase concentrations are associated with the risk for severe anaphylactic reactions. Preventive measures should include substitution of angiotensin-converting enzyme inhibitors.

Fear-related behaviour of dogs in veterinary practice.

To investigate further the occurrence of fear-related behaviour in dogs in veterinary practice and to evaluate associated factors, 135 dogs were observed under practice conditions within the framework of a standardised test examination and the owners interviewed using a questionnaire. Most dogs exhibited fear reactions, particularly on the examination table, with 78.5% (106/135) categorised as 'fearful' based on their behaviour. Unlike weight and castration, age, gender and previous experience were significantly (P<0.05) associated with fearful behaviour. Male dogs were significantly less 'fearful' than females and animals under <2 years were significantly less 'fearful' compared with older dogs. Those with only positive previous experiences in veterinary surgeries were significantly less 'fearful' than dogs that had a previous negative experience. Fear-related behaviour in veterinary practice is an issue of importance.