A generalized Hosmer-Lemeshow goodness-of-fit test for a family of generalized linear models.

Generalized linear models (GLMs) are very widely used, but formal goodness-of-fit (GOF) tests for the overall fit of the model seem to be in wide use only for certain classes of GLMs. We develop and apply a new goodness-of-fit test, similar to the well-known and commonly used Hosmer-Lemeshow (HL) test, that can be used with a wide variety of GLMs. The test statistic is a variant of the HL statistic, but we rigorously derive an asymptotically correct sampling distribution using methods of Stute and Zhu (Scand J Stat 29(3):535-545, 2002) and demonstrate its consistency. We compare the performance of our new test with other GOF tests for GLMs, including a naive direct application of the HL test to the Poisson problem. Our test provides competitive or comparable power in various simulation settings and we identify a situation where a naive version of the test fails to hold its size. Our generalized HL test is straightforward to implement and interpret and an R package is publicly available. Supplementary Information: The online version contains supplementary material available at 10.1007/s11749-023-00912-8.

Causal pathways to health-related quality of life in children with juvenile idiopathic arthritis: results from the ReACCh-Out cohort.

OBJECTIVE: Structural equation modelling was applied to data from the Research in Arthritis in Canadian Children emphasizing Outcomes (ReACCh-Out) cohort to help elucidate causal pathways to decreased health-related quality of life (HRQoL) in children with JIA. METHODS: Based on published literature and clinical plausibility, a priori models were constructed with explicit root causes (disease activity, treatment intensity) and mediators (pain, disease symptoms, functional impairments) leading to HRQoL [measured by the Quality of my Life (QoML) scale and the Juvenile Arthritis Quality of Life Questionnaire (JAQQ)] at five disease stages: (i) diagnosis, (ii) 3-9 months after diagnosis, (iii) flare, (iv) remission on medications, (v) remission off medications. Following structural equation modelling, a posteriori models were selected based on data fit and clinical plausibility. RESULTS: We included 561, 887, 137, 186 and 182 patients at each stage, respectively. In a posteriori models for active disease stages, paths from disease activity led through pain, functional impairments, and disease symptoms, directly or through restrictions in participation, to decreased QoML scores. Treatment intensity had detrimental effects through psychosocial domains; while treatment side effects had a lesser role. Pathways were similar for QoML and JAQQ, but JAQQ models provided greater specificity. Models for remission stages were not supported by the data. CONCLUSION: Our findings support disease activity and treatment intensity as being root causes of decreased HRQoL in children with JIA, with pain, functional impairments, and participation restrictions being mediators for disease activity; they support psychosocial effects and side effects as being mediators for treatment intensity.

Evaluation of cardiovascular disease risk in individuals with chronic spinal cord injury.

STUDY DESIGN: Multicentre, cross-sectional study. OBJECTIVES: To identify which markers of obesity, injury characteristics and autonomic function variables are related to cardiovascular disease (CVD) risk after spinal cord injury (SCI), and establish cut-points for detection and risk management. SETTING: Eight SCI rehabilitation centres in the Netherlands. METHODS: Individuals (n = 257) with a traumatic, chronic (≥10 years) SCI, with age at injury between 18 and 35 years, completed a self-report questionnaire and a one-day visit to a rehabilitation centre for testing. Three anthropometric measures were tested: body mass index (BMI); waist circumference (WC); and waist-to-height ratio (WHtR). Injury characteristics included: American Spinal Injury Association impairment scale (AIS); duration of injury (DOI); and neurological level of injury (LOI). Cardiovascular autonomic function was assessed from peak heart rate during maximal exercise (HRpeak). Systolic arterial pressure (SAP) and aerobic capacity (VO2peak) were also determined. CVD risk was calculated using the Framingham risk score (FRS). RESULTS: All anthropometric variables were associated with FRS, with WC showing the strongest correlation (r = 0.41, p < 0.001) and greatest area under the curve (0.73) for 10-year CVD risk (%). WC, DOI, SAP, HRpeak, LOI, and VO2peak (variable importance: 0.81, 1.0, 0.98, 0.98, 0.66, 0.68, respectively) were important predictive variables for 10-year CVD risk in individuals with SCI. CONCLUSIONS: We confirm that WC is a simple, practical measure of CVD risk, and along with DOI and markers of cardiovascular autonomic function, plays a role in the increased CVD risk following SCI.

Cardiovascular responses to orthostasis and their association with falls in older adults.

BACKGROUND: Orthostatic hypotension (OH) refers to a marked decline in blood pressure when upright. OH has a high incidence and prevalence in older adults and represents a potential intrinsic risk factor for falls in these individuals. Previous studies have not included more recent definitions for blood pressure responses to orthostasis, including initial, delayed, and recovery blood pressure responses. Furthermore, there is little research examining the relationships between cerebrovascular functioning and falling risk. Therefore, we aimed to: (i) test the association between different blood pressure responses to orthostatic stress and retrospective falling history and; (ii) test the association between cerebrovascular responses to orthostatic stress and falling history. METHODS: We tested 59 elderly residents in long term care facilities who underwent a passive seated orthostatic stress test. Beat-to-beat blood pressure and cerebral blood flow velocity (CBFV) responses were assessed throughout testing. Risk factors for falls and falling history were collected from facility records. Cardiovascular responses to orthostasis were compared between retrospective fallers (≥1 fall in the previous year) and non-fallers. RESULTS: Retrospective fallers had larger delayed declines in systolic arterial pressure (SAP) compared to non-fallers (p = 0.015). Fallers also showed poorer early (2 min) and late (15 min) recovery of SAP. Fallers had a greater decline in systolic CBFV. CONCLUSIONS: Older adults with a positive falling history have impaired orthostatic control of blood pressure and CBFV. With better identification and understanding of orthostatic blood pressure impairments earlier intervention and management can be implemented, potentially reducing the associated risk of morbidity and mortality. Future studies should utilize the updated OH definitions using beat-to-beat technology, rather than conventional methods that may offer less accurate detection.

Improving the Hosmer-Lemeshow goodness-of-fit test in large models with replicated Bernoulli trials.

The Hosmer-Lemeshow (HL) test is a commonly used global goodness-of-fit (GOF) test that assesses the quality of the overall fit of a logistic regression model. In this paper, we give results from simulations showing that the type I error rate (and hence power) of the HL test decreases as model complexity grows, provided that the sample size remains fixed and binary replicates (multiple Bernoulli trials) are present in the data. We demonstrate that a generalized version of the HL test (GHL) presented in previous work can offer some protection against this power loss. These results are also supported by application of both the HL and GHL test to a real-life data set. We conclude with a brief discussion explaining the behavior of the HL test, along with some guidance on how to choose between the two tests. In particular, we suggest the GHL test to be used when there are binary replicates or clusters in the covariate space, provided that the sample size is sufficiently large.

Random forests for survival data: which methods work best and under what conditions?

Few systematic comparisons of methods for constructing survival trees and forests exist in the literature. Importantly, when the goal is to predict a survival time or estimate a survival function, the optimal choice of method is unclear. We use an extensive simulation study to systematically investigate various factors that influence survival forest performance - forest construction method, censoring, sample size, distribution of the response, structure of the linear predictor, and presence of correlated or noisy covariates. In particular, we study 11 methods that have recently been proposed in the literature and identify 6 top performers. We find that all the factors that we investigate have significant impact on the methods' relative accuracy of point predictions of survival times and survival function estimates. We use our results to make recommendations for which methods to use in a given context and offer explanations for the observed differences in relative performance.

Geographic variation in rodent-flea relationships in the presence of black-tailed prairie dog colonies.

We characterized the relationship between fleas and their rodent hosts in the presence of prairie dog colonies and compared them to adjacent assemblages away from colonies. We evaluated the rodent-flea relationship by quantifying prevalence, probability of infestation, flea load, and intensity of fleas on rodents. As prairie dog burrows provide refugia for fleas, we hypothesized that prevalence, flea load, and intensity would be higher for rodents that are associated with black-tailed prairie dog colonies. Rodents were trapped at off- and on-colony grids, resulting in the collection of 4,509 fleas from 1,430 rodents in six study areas. The rodent community composition varied between these study areas. Flea species richness was not different between prairie dog colonies and the surrounding grasslands (p = 0.883) but was positively correlated with rodent species richness (p = 0.055). Prairie dog colonies did not increase the prevalence of fleas (p > 0.10). Flea loads on rodents did not vary between off- and on-colony grids at three of the study areas (p > 0.10). Based on the prevalence, infestation rates, and flea loads, we identified Peromyscus maniculatus, Onychomys leucogaster, and two Neotoma species as important rodent hosts for fleas and Aetheca wagneri, Orchopeus leucopus, Peromyscopsylla hesperomys, Pleochaetis exilis, and Thrassisfotus as the most important fleas associated with these rodents. Prairie dog colonies did not seem to facilitate transmission of fleas between rodent hosts, and the few rodent-flea associations exhibited significant differences between off- and on-colony grids.

Determinants of staff commitment to hip protectors in long-term care: A cross-sectional survey.

BACKGROUND: If worn, certain models of hip protectors are highly effective at preventing hip fractures from falls in residents of long-term care, but modest acceptance and adherence have limited the effectiveness of hip protectors. Residents of long-term care are more likely to accept the initial offer of hip protectors and to adhere to recommendations concerning the use of hip protectors when staff are committed to supporting the application of hip protectors. Yet, we know very little about the nature of and factors associated with staff commitment to hip protectors in long-term care. OBJECTIVE: To identify factors associated with staff commitment to hip protectors in long-term care. DESIGN: A cross-sectional survey. SETTING: Thirteen long-term care homes (total beds = 1816) from a single regional health district in British Columbia, Canada. PARTICIPANTS: A convenience sample of 535 paid staff who worked most of their time (>50% of work hours) at a participating long-term care home, for at least one month, and for at least 8 h per week. We excluded six (1.1%) respondents who were unaware of hip protectors. Of the remaining 529 respondents, 90% were female and 55% were health care assistants. METHODS: Respondents completed the Commitment to Hip Protectors Index to indicate their commitment to hip protectors. We used Bayesian Model Averaging logistic regression to model staff commitment as a function of personal variables, experiences with hip protectors, intraorganizational communication and influence, and organizational context. RESULTS: Staff commitment was negatively related to organizational tenure >20 years (posterior probability = 97%; logistic regression coefficient = -0.28; 95% confidence interval = -0.48, -0.08), and awareness of a padded hip fracture (100%; -0.57; -0.69, -0.44). Staff commitment was positively related to the existence of a champion of hip protectors within the home (100%; 0.24; 0.17, 0.31), perceived quality of intraorganizational communication (100%; 0.04; 0.02, 0.05), extent of mutual respect between residents and staff and perceived contribution to quality of life of the residents they serve (100%; 0.10; 0.05, 0.15), and frequency of transformational leadership practices by respondents' primary supervisors (100%; 0.01; 0.01, 0.02). CONCLUSIONS: We provide novel insight into the factors governing staff commitment to hip protectors in long-term care. Targeting of these factors could improve acceptance and adherence with hip protectors, thereby contributing to enhanced effectiveness of hip protectors to prevent hip fractures in long-term care.

Efficacy of Bauveria bassiana for red flour beetle when applied with plant essential oils or in mineral oil and organosilicone carriers.

The carriers mineral oil and Silwet L-77 and the botanical insecticides Neemix 4.5 and Hexacide were evaluated for their impacts on the efficacy of Beauveria bassiana (Balsamo) Vuillemin conidia against red flour beetle, Tribolium castaneum (Herbst), larvae. The dosages of liquid treatments were quantified by both conidia concentration in the spray volume and conidia deposition on the target surface. The latter approach allowed comparison with dry, unformulated conidia. The median lethal concentrations of B. bassiana in 0.05% Silwet L-77 solution or without a carrier were approximately double that for conidia in mineral oil. Carriers had highly significant effects on the efficacy of B. bassiana. The lower efficacy of conidia in aqueous Silwet L-77 may have been the result of conidia loss from the larval surface because of the siloxane's spreading properties. Neemix 4.5 (4.5% azadirachtin) delayed pupation and did not reduce the germination rate of B. bassiana conidia, but it significantly reduced T. castaneum mortality at two of four tested fungus doses. Hexacide (5% rosemary oil) caused significant mortality when applied without B. bassiana, but it did not affect pupation, the germination rate of conidia, or T. castaneum mortality when used in combination with the fungus.

Physical activity minimizes the association of body fatness with abdominal obesity in white, premenopausal women: results from the Third National Health and Nutrition Examination Survey.

This study explored the association of physical activity with body fatness and abdominal obesity in 1,004 premenopausal white women who participated in the Third National Health and Nutrition Examination Survey. Physical activity was classified into four levels based on fitness criteria from the American College of Sports Medicine. Standard protocols were used for measuring body mass index (BMI), percent body fat, and waist-to-hip ratio (WHR). Means for BMI, percent body fat, and WHR were calculated for each physical activity level. Analysis of variance procedures were performed, followed by pair-wise comparisons, to determine differences among means and for trends. Mean BMI, percent body fat, and WHR were significantly less ( P < .001) at each higher physical activity level except between Levels 0 and 1. The ability of increased daily physical activity to minimize age-related increases in abdominal obesity could be a strong incentive for women approaching menopause to become more physically active.

The effect of orthostatic stress type on cardiovascular control.

OBJECTIVES: We aimed to compare the cardiovascular responses of a novel orthostatic stress test, the passive seated orthostatic stress test (PSOST), with those during passive head-up tilt testing (HUTT). We hypothesized that cardiovascular responses during PSOST would be similar to those during HUTT (the 'gold standard'). METHODS: We tested 15 healthy volunteers, who underwent both PSOST and HUTT during one session in a random order. We measured beat-to-beat blood pressure, heart rate, peripheral resistance, stroke volume, cardiac output, and middle cerebral artery blood flow velocity during each test. RESULTS: Blood pressure responses were not significantly different between PSOST and HUTT, except for a significantly lower delayed nadir and 15-min recovery value in systolic arterial pressure during HUTT. HUTT elicited a significantly larger increase in heart rate during all test intervals in comparison with PSOST, as well as a larger decline in stroke volume during almost all test intervals. Responses for the other hemodynamic variables were not significantly different between the tests at any test interval. Repeated HUTT has large inherent variability, which was also evident from the variability in the mean differences on comparing PSOST and HUTT. There was a significant bias for larger heart rate increases (P<0.01) and a greater delayed systolic arterial pressure decline during HUTT. CONCLUSION: We have shown that PSOST and HUTT elicit similar blood pressure and cerebrovascular responses in the early stages of the upright phase. We believe that PSOST is a reasonable surrogate for HUTT in assessing orthostatic hypotension in population groups that are unable to stand for prolonged periods of time.

Assessing pitcher and catcher influences on base stealing in Major League Baseball.

A formal statistical analysis is performed to determine the extent to which pitchers and catchers can influence stolen-base attempts and successes. Two response proportions, attempt/opportunity and success/attempt, are modelled separately using mixed-effects logistic regression models applied to situations with a runner on first and other bases empty. Data include the first innings of all Major League Baseball games played between 1978 and 1990, which encompasses over 48,000 opportunities and 9000 attempts. Pitchers and catchers are entered as random effects and various other factors thought to influence stolen-base attempts and successes are entered as fixed effects. Variance components are estimated and hypotheses tests indicate that the population variance components for both pitchers and catchers are significant for both response proportions. The presence of variation among players at the respective positions is interpreted as evidence that stolen-base defence is a real skill exhibited to varying extents by different players. Furthermore, the variance component for pitchers is greater than that for catchers for both response proportions, indicating that pitchers have greater potential to affect stolen-base attempts and successes. Under usual conditions, it is estimated that 95% of pitchers have first-inning stolen-base success/attempt probabilities between 0.50 and 0.84, while 95% of catchers have probabilities between 0.59 and 0.79.

Consumer sensory analysis of organically and conventionally grown vegetables.

Consumers of organically grown fruits and vegetables often believe that these products taste better than conventional produce. However, comparison of produce from supermarket shelves does not permit adequate assessment of this consumer perception, given potentially confounding cultivar and environmental effects. We used replicated side-by-side plots to produce organic and conventional vegetables for consumer sensory studies. In one test, red loose leaf lettuce, spinach, arugula, and mustard greens, grown organically and conventionally, were evaluated for overall liking as well as for intensity of flavor and bitterness. Another consumer test was conducted comparing organically and conventionally grown tomatoes, cucumbers, and onions. Overall, organically and conventionally grown vegetables did not show significant differences in consumer liking or consumer-perceived sensory quality. The only exception was in tomatoes where the conventionally produced tomato was rated as having significantly stronger flavor than the organically produced tomato. However, overall liking was the same for both organic and conventional samples. As conventional tomatoes also were scored marginally significantly higher in ripeness and a positive correlation was found between ratings of flavor intensity and ripeness, the flavor difference observed could not be simply ascribed to the contrasting growing conditions. Consumer panelists in both tests considered organic produce to be healthier (72%) and more environmentally friendly (51%) than conventional produce, while 28% considered organic produce to have better taste. Covariance analysis indicated that consumer demographics affected sensory comparisons of organic and conventional lettuce and cucumbers. Future study is needed to substantiate the influence of segmentation of consumers on their preference for organic food.

Testing for conditional multiple marginal independence.

Survey respondents are often prompted to pick any number of responses from a set of possible responses. Categorical variables that summarize this kind of data are called pick any/c variables. Counts from surveys that contain a pick any/c variable along with a group variable (r levels) and stratification variable (q levels) can be marginally summarized into an r x c x q contingency table. A question that may naturally arise from this setup is to determine if the group and pick any/c variable are marginally independent given the stratification variable. A test for conditional multiple marginal independence (CMMI) can be used to answer this question. Since subjects may pick any number out of c possible responses, the Cochran (1954, Biometrics 10, 417-451) and Mantel and Haenszel (1959, Journal of the National Cancer Institute 22, 719-748) tests cannot be used directly because they assume that units in the contingency table are independent of each other. Therefore, new testing methods are developed. Cochran's test statistic is extended to r x 2 x q tables, and a modified version of this statistic is proposed to test CMMI. Its sampling distribution can be approximated through bootstrapping. Other CMMI testing methods discussed are bootstrap p-value combination methods and Bonferroni adjustments. Simulation findings suggest that the proposed bootstrap procedures and the Bonferroni adjustments consistently hold the correct size and provide power against various alternatives.

Testing for marginal independence between two categorical variables with multiple responses.

Questions that ask respondents to "choose all that apply" from a set of items occur frequently in surveys. Categorical variables that summarize this type of survey data are called both pick any/c variables and multiple-response categorical variables. It is often of interest to test for independence between two categorical variables. When both categorical variables can have multiple responses, traditional Pearson chi-square tests for independence should not be used because of the within-subject dependence among responses. An intuitively constructed version of the Pearson statistic is proposed to perform the test using bootstrap procedures to approximate its sampling distribution. First- and second-order adjustments to the proposed statistic are given in order to use a chi-square distribution approximation. A Bonferroni adjustment is proposed to perform the test when the joint set of responses for individual subjects is unavailable. Simulations show that the bootstrap procedures hold the correct size more consistently than the other procedures.