Michael Rotondi

Sample Size Requirements for Interval Estimation of the Kappa Statistic for Interobserver Agreement Studies with a Binary Outcome and Multiple Raters

Sample size requirements that achieve a prespecified expected lower limit for a confidence interval about the intraclass kappa statistic are supplied for the case of multiple raters and a binary outcome variable. The expected lower confidence limit achievable for a given number of subjects and raters is also presented. These results should be useful in the planning stages of an interobserver agreement study in which the focus is on interval estimation rather than hypothesis-testing.

Urinary Biomarkers of Polycyclic Aromatic Hydrocarbons Are Associated with Cardiometabolic Health Risk

Background Polycyclic aromatic hydrocarbons (PAH) are both man-made and naturally occurring environmental pollutants that may be related to cardiometabolic health risk. Objective To determine whether PAH is associated with obesity in the adult population and to examine whether urinary concentrations of PAH metabolites are associated with differences in how obesity relates to 3 or more risk factors for the metabolic syndrome (3RFMetS), type 2 diabetes (T2D), hypertension, and dyslipidemia. Methods A total of 4765 adult participants from the 2001–2008 National Health and Nutrition Examination Survey were examined. The association between 8 urinary hydroxylated PAH metabolites, obesity, and health were examined using weighted logistic regressions adjusting for age, sex, ethnicity, PIR, smoking status, and urinary creatinine. Results There was a positive dose-dependent association between obesity and 2-phenanthrene quintiles (P trend <0.0001). Contrarily, higher quintiles of 1-naphthalene were associated with lower risk of obesity (P trend = 0.0004). For a given BMI, those in the highest quintile of 2-naphthalene, 2-fluorene, 3-fluorene and 2-phenanthrene had a 66–80% greater likelihood of 3RFMetS (P≤0.05) compared to low levels. Higher quintiles of 1-naphthalene, 2-naphthalene, 2-phenanthrene and 1-pyrene were associated with a 78–124% greater likelihood of T2D (P≤0.05) compared to low levels while high 1-naphthalene, 2-naphthalene, 2-fluorene, 3-fluorene and 2-phenanthrene were associated with a 38–68% greater likelihood of dyslipidemia (P≤0.05) compared to lower levels. Finally, 2-naphthalene and 2-phenanthrene were positively associated with hypertension (P trend = 0.008 and P trend = 0.02 respectively). Conclusions PAH is related to obesity and the expression of a number of obesity-related cardiometabolic health risk factors. Future research is needed to bring to light the mechanistic pathways related to these findings.

Sample Size Estimation in Cluster Randomized Educational Trials: An Empirical Bayes Approach

The educational field has now accumulated an extensive literature reporting on values of the intraclass correlation coefficient, a parameter essential to determining the required size of a planned cluster randomized trial. We propose here a simple simulation-based approach including all relevant information that can facilitate this task. An example and corresponding computer code is attached.

Effect of tai chi on musculoskeletal health-related fitness and self-reported physical health changes in low income, multiple ethnicity mid to older adults.

BackgroundTai Chi (TC) has proven to be effective at improving musculoskeletal fitness by increasing upper and lower body strength, low back flexibility and overall physical health. The objectives of this study were to examine changes in musculoskeletal health-related fitness and self-reported physical health after a 16 week TC program in a low income multiple ethnicity mid to older adult population.MethodsTwo hundred and nine ethnically diverse mid to older community dwelling Canadian adults residing in low income neighbourhoods were enrolled in a 16 week Yang style TC program. Body Mass Index and select musculoskeletal fitness measures including upper and lower body strength, low back flexibility and self-reported physical health measured by SF 36 were collected pre and post the TC program. Determinants of health such as age, sex, marital status, education, income, ethnicity of origin, multi-morbidity conditions, weekly physical activity, previous TC experience as well as program adherence were examined as possible musculoskeletal health-related fitness change predictors.ResultsUsing paired sample t-tests significant improvements were found in both upper and lower body strength, low back flexibility, and the SF 36 physical health scores (p < 0.05). Based on multiple linear regression analyses, no common health determinants explained a significant portion of the variation in percent changes of the musculoskeletal fitness and SF 36 measures.ConclusionsThese results reveal that TC has the potential of having a beneficial influence on musculoskeletal health-related fitness and self-reported physical health in a mid to older low socioeconomic, ethnically diverse sample.

Sample size estimation in cluster randomized trials: An evidence-based perspective

The evidence-based perspective to sample size estimation determines appropriate trial size by examining its potential impact on the literature. This approach is extended to determine the appropriate size of a planned cluster randomized trial by considering the role of the planned trial on a future meta-analysis (including current literature and the proposed study). A simulation-based algorithm allows consideration of variable cluster sizes and intracluster correlation coefficient values in conjunction with three approaches to sample size estimation, namely the power-based, variance reduction and non-inferiority perspectives. Two examples employing the sample size estimation techniques described are discussed in detail, while appropriate code is provided in the accompanying R package CRTSize.

Formulation of evidence-based messages to promote the use of physical activity to prevent and manage Alzheimer’s disease

BackgroundThe impending public health impact of Alzheimer’s disease is tremendous. Physical activity is a promising intervention for preventing and managing Alzheimer’s disease. However, there is a lack of evidence-based public health messaging to support this position. This paper describes the application of the Appraisal of Guidelines Research and Evaluation II (AGREE-II) principles to formulate an evidence-based message to promote physical activity for the purposes of preventing and managing Alzheimer’s disease.MethodsA messaging statement was developed using the AGREE-II instrument as guidance. Methods included (a) conducting a systematic review of reviews summarizing research on physical activity to prevent and manage Alzheimer’s disease, and (b) engaging stakeholders to deliberate the evidence and formulate the messaging statement.ResultsThe evidence base consisted of seven systematic reviews focused on Alzheimer’s disease prevention and 20 reviews focused on symptom management. Virtually all of the reviews of symptom management conflated patients with Alzheimer’s disease and patients with other dementias, and this limitation was reflected in the second part of the messaging statement. After deliberating the evidence base, an expert panel achieved consensus on the following statement: “Regular participation in physical activity is associated with a reduced risk of developing Alzheimer’s disease. Among older adults with Alzheimer’s disease and other dementias, regular physical activity can improve performance of activities of daily living and mobility, and may improve general cognition and balance.” The statement was rated favourably by a sample of older adults and physicians who treat Alzheimer’s disease patients in terms of its appropriateness, utility, and clarity.ConclusionPublic health and other organizations that promote physical activity, health and well-being to older adults are encouraged to use the evidence-based statement in their programs and resources. Researchers, clinicians, people with Alzheimer’s disease and caregivers are encouraged to adopt the messaging statement and the recommendations in the companion informational resource.

Application of the McKenzie system of Mechanical Diagnosis and Therapy (MDT) in patients with shoulder pain; a prospective longitudinal study

Abstract Objectives: The primary objective was to determine if the pain and function response to the McKenzie system of Mechanical Diagnosis and Therapy (MDT) differs by MDT classification category at two and four weeks following the start of MDT treatment for shoulder complaints. The secondary objective was to describe the frequency of discharge over time by MDT classification. Methods: International, MDT-trained study collaborators recruited 93 patients attending physiotherapy for rehabilitation of a shoulder problem. The Numeric Pain Rating Scale (NPRS) and the Upper Extremity Functional Index (UEFI) were collected at the initial assessment and two and four weeks after treatment commenced. A two-way mixed model analysis of variance with planned pairwise comparisons was performed to identify where the differences between MDT classification groups actually existed. Results: The Derangement and Spinal classifications had significantly lower NPRS scores than the Dysfunction group at week 2 and week 4 (p < 0.05). The Derangement and Spinal classifications had significantly higher UEFI scores than the Dysfunction group at week 2 and week 4 (p < 0.05). The frequency of discharge at week 2 was 37% for both Derangement and Spinal classifications, with no discharges for the Dysfunction classification at this time point. The frequency of discharge at week 4 was 83, 82 and 15% for the Derangement, Spinal and Dysfunction classifications, respectively. Discussion: Classifying patients with shoulder pain using the MDT system can impact treatment outcomes and the frequency of discharge. When MDT-trained clinicians are allowed to match the intervention to a specific MDT classification, the outcome is aligned with the response expectation of the classification. Level of Evidence: 2b

The associations of resting metabolic rate with chronic conditions and weight loss

The aim of this study was to examine the associations between baseline and changes in resting metabolic rate (RMR) with chronic condition(s) and weight loss (WL). Sex stratified analysis was undertaken on 393 adults from the Wharton Weight Management Clinics. The association between baseline RMR and WL was examined adjusting for age, BMI, ethnicity and treatment time. The association between changes in RMR (ΔRMR) and WL was also examined adjusting for baseline RMR and above covariates. Models were further adjusted for high glucose, triglycerides, blood pressure, low‐density lipoprotein (LDL) and low high‐density lipoprotein (HDL). While men (6.0 ± 8.6 kg) and women (5.6 ± 8.3 kg) had significant WL throughout the intervention, their measured decreases in RMR (−48 ± 322 kcal and −5 ± 322 kcal, respectively) were non‐significant (P > 0.05). Individuals with a high blood pressure had a higher baseline RMR and women with a high LDL had a lower baseline RMR than those without the chronic condition (P < 0.05). Regardless of sex, WL was not significantly associated with baseline RMR or ΔRMR (P > 0.05) in both models. Participants with a low baseline RMR do not appear to be at a disadvantage for WL. Further, WL can occur without decreases in RMR in populations with high levels of obesity and obesity‐related comorbidities.

The Influence of Urinary Concentrations of Organophosphate Metabolites on the Relationship between BMI and Cardiometabolic Health Risk

The objective was to determine whether detectable levels of OP metabolites influence the relationship between BMI and cardiometabolic health. This cross-sectional study was conducted using 2227 adults from the 1999–2008 NHANES datasets. Urinary concentrations of six dialkyl phosphate metabolites were dichotomized to above and below the detection limit. Weighted multiple regression analysis was performed adjusting for confounding variables. Independent of BMI, individuals with detectable metabolites had higher diastolic blood pressure (for dimethylphosphate, diethylphosphate, and diethyldithiophosphate; P < 0.05), lower HDL (for diethyldithiophosphate; P = 0.02), and higher triglyceride (for dimethyldithiophosphate; P = 0.05) than those below detection. Contrarily, those with detectable dimethylthiophosphate had better LDL, HDL, and total cholesterol, independent of BMI. Individuals at a higher BMI range who had detectable diethylphosphate (interaction: P = 0.03) and diethylthiophosphate (interaction: P = 0.02) exhibited lower HDL, while little difference existed between OP metabolite detection statuses at lower BMIs. Similarly, individuals with high BMIs and detectable diethylphosphate had higher triglyceride than those without detectable levels, while minimal differences between diethylphosphate detection statuses were observed at lower BMIs (interaction: P = 0.02). Thus, cardiometabolic health outcome differs depending on the specific OP metabolite being examined, with higher BMIs amplifying health risk.

Individuals with obesity but no other metabolic risk factors are not at significantly elevated all-cause mortality risk in men and women: Metabolic healthy obesity and mortality risk

Studies have examined mortality risk for metabolically healthy obesity, defined as zero or one metabolic risk factors but not as zero risk factors. Thus, we sought to determine the independent mortality risk associated with obesity or elevated glucose, blood pressure or lipids in isolation or clustered together. The sample included 54 089 men and women from five cohort studies (follow‐up = 12.8 ± 7.2 years and 4864 [9.0%] deaths). Individuals were categorized as having obesity or elevated glucose, blood pressure or lipids alone or clustered with obesity or another metabolic factor. In our study sample, 6% of individuals presented with obesity but no other metabolic abnormalities. General obesity (hazard ratios [HR], 95% CI = 1.10, 0.8–1.6) and abdominal obesity (HR = 1.24, 0.9–1.7) in the absence of metabolic risk factors were not associated with mortality risk compared to lean individuals. Conversely, diabetes, hypertension and dyslipidaemia in isolation were significantly associated with mortality risk (HR range = 1.17–1.94, P < 0.05). However, when using traditional approaches, obesity (HR = 1.12, 1.02–1.23) is independently associated with mortality risk after statistical adjustment for the other metabolic risk factors. Similarly, metabolically healthy obesity, when defined as zero or one risk factor, is also associated with increased mortality risk (HR = 1.15, 1.01–1.32) as compared to lean healthy individuals. Obesity in the absence of metabolic abnormalities may not be associated with higher risk for all‐cause mortality compared to lean healthy individuals. Conversely, elevation of even a single metabolic risk factor is associated with increased mortality risk.