Publications by authors named "Kristin Sainani"

82 Publications

Multinomial and Ordinal Logistic Regression.

PM R 2021 Apr 27. Epub 2021 Apr 27.

Department of Epidemiology and Population Health, Stanford University.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1002/pmrj.12622DOI Listing
April 2021

Assessing Diagnostic and Severity Grading Accuracy of Ultrasound Measurements for Carpal Tunnel Syndrome Compared to Electrodiagnostics.

PM R 2020 Dec 11. Epub 2020 Dec 11.

Division of Physical Medicine & Rehabilitation, Department of Orthopaedics, Stanford University, Stanford, CA, USA.

Background: The combined sensory index (CSI) is the most sensitive electrodiagnostic criteria for carpal tunnel syndrome (CTS), and the CSI and Bland criteria have been shown to predict surgical treatment outcomes. The proposed ultrasound measurements have not been assessed against the CSI for diagnostic accuracy and grading of CTS severity.

Objective: To investigate the use of ultrasound evaluations for both diagnosis and assessment of severity grading of CTS in comparison to electrodiagnostic assessment.

Design: All patients underwent an electrodiagnostic evaluation using the CSI and Bland severity grading. Each patient underwent an ultrasound evaluation including cross-sectional area (CSA), the change in CSA from the forearm to the tunnel (∆CSA), and the wrist-forearm ratio (WFR). These measurements were assessed for diagnostic and severity grading accuracy using the CSI as the gold standard.

Setting: Tertiary academic center.

Participants: All patients referred for electrodiagnostic evaluation for CTS were eligible for the study. Only those with idiopathic CTS were included and those with prior CTS treatment were also excluded. Ninety-five patients were included in the study.

Interventions: Not applicable.

Main Outcome Measures: The primary study outcome measure was concordance between CSI diagnosis and severity categories and the ultrasound measurements. Both outcomes were also assessed using Bland criteria.

Results: Optimal cut-points for diagnosis of CTS were found to be CSA ≥12 mm , ∆CSA ≥4 mm , WFR ≥1.4. Using these cut-points, C-statistics comparing diagnosis of CTS using ultrasound measurements versus using the CSI ranged from 0.893-0.966. When looking at CSI severity grading compared to ∆CSA, however, the C-statistics were 0.640-0.661 with substantial overlap between severity groups.

Conclusions: Although ultrasound measurements had high diagnostic accuracy for CTS based on the CSI criteria, ultrasound measurements were unable to adequately distinguish between CSI severity groups among patients with CTS.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1002/pmrj.12533DOI Listing
December 2020

Lower Trabecular Bone Score and Spine Bone Mineral Density Are Associated With Bone Stress Injuries and Triad Risk Factors in Collegiate Athletes.

PM R 2020 Oct 10. Epub 2020 Oct 10.

Boswell Human Performance Laboratory, Department of Orthopaedic Surgery, Stanford, CA.

Introduction: Determinants of bone health and injury are important to identify in athletes. Bone mineral density (BMD) is commonly measured in athletes with Female Athlete Triad (Triad) risk factors; the trabecular bone score (TBS) has been proposed to predict fracture risk independent of BMD. Evaluation of TBS and spine BMD in relation bone stress injury (BSI) risk has not been studied in female collegiate athletes.

Objective: We hypothesized that spine BMD and TBS would each independently predict BSI and that the combined measures would improve injury prediction in female collegiate athletes. We also hypothesized that each measure would be correlated with Triad risk factors.

Design: Retrospective cohort.

Setting: Academic Institution.

Methods: Dual energy x-ray absorptiometry (DXA) of the lumbar spine was used to calculate BMD and TBS values. Chart review was used to identify BSI that occurred after the DXA measurement and to obtain Triad risk factors. We used logistic regression to examine the ability of TBS and BMD alone or in combination to predict prospective BSI.

Results: Within 321 athletes, 29 (9.0%) sustained a BSI after DXA. BMD and TBS were highly correlated (Pearson correlation r = 0.62, P < .0001). Spine BMD and TBS had similar ability to predict BSI; the C-statistic and 95% confidence intervals were 0.69 (0.58 to 0.81) for spine BMD versus 0.68 (0.57 to 0.79) for TBS. No improvement in discrimination was observed with combined BMD + TBS (C-statistic 0.70, 0.59 to 0.81). Both TBS and BMD predicted trabecular-rich BSI (defined as pelvis, femoral neck, and calcaneus) better than cortical-rich BSI. Both measures had similar correlations with Triad risk factors.

Conclusion: Lower BMD and TBS values are associated with elevated risk for BSI and similar correlation to Triad risk factors. TBS does not improve prediction of BSI. Collectively, our findings suggest that BMD may be a sufficient measure of skeletal integrity from DXA in female collegiate athletes.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1002/pmrj.12510DOI Listing
October 2020

Virological Failure and Acquired Genotypic Resistance Associated With Contemporary Antiretroviral Treatment Regimens.

Open Forum Infect Dis 2020 Sep 6;7(9):ofaa316. Epub 2020 Aug 6.

Division of Infectious Diseases, Department of Medicine, Stanford University, Stanford, California, USA.

Background: There are few descriptions of virologic failure (VF) and acquired drug resistance (HIVDR) in large cohorts initiating contemporary antiretroviral therapy (ART).

Methods: We studied all persons with HIV (PWH) in a California clinic population initiating ART between 2010 and 2017. VF was defined as not attaining virologic suppression, discontinuing ART, or virologic rebound prompting change in ART.

Results: During the study, 2315 PWH began ART. Six companion drugs were used in 93.3% of regimens: efavirenz, elvitegravir/c, dolutegravir, b-darunavir, rilpivirine, and raltegravir. During a median follow-up of 36 months, 214 (9.2%) PWH experienced VF (2.8 per 100 person-years) and 62 (2.7%) experienced HIVDR (0.8 per 100 person-years). In multivariable analyses, younger age, lower CD4 count, higher virus load, and b-atazanavir were associated with increased VF risk; lower CD4 count, higher virus load, and nevirapine were associated with increased HIVDR risk. Compared with efavirenz, dolutegravir, raltegravir, and b-darunavir were associated with reduced HIVDR risk. Risks of VF and HIVDR were not significantly associated with ART initiation year. Of the 62 PWH with HIVDR, 42 received an non-nucleoside RT inhibitor (NNRTI), 15 an integrase-strand transfer inhibitor (INSTI), and 5 a protease inhibitor (PI). Among those with HIVDR on an NNRTI or first-generation INSTI, 59% acquired dual class resistance and 29% developed tenofovir resistance; those receiving a PI or dolutegravir developed just M184V.

Conclusions: Despite the frequent use of contemporary ART regimens, VF and HIVDR continue to occur. Further efforts are required to improve long-term ART virological responses to prevent the consequences of ongoing HIV-1 replication including virus transmission and HIVDR.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1093/ofid/ofaa316DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7462367PMC
September 2020

Systematic review of the use of "magnitude-based inference" in sports science and medicine.

PLoS One 2020 26;15(6):e0235318. Epub 2020 Jun 26.

Department of Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer Center, New York, New York, United States of America.

Magnitude-based inference (MBI) is a controversial statistical method that has been used in hundreds of papers in sports science despite criticism from statisticians. To better understand how this method has been applied in practice, we systematically reviewed 232 papers that used MBI. We extracted data on study design, sample size, and choice of MBI settings and parameters. Median sample size was 10 per group (interquartile range, IQR: 8-15) for multi-group studies and 14 (IQR: 10-24) for single-group studies; few studies reported a priori sample size calculations (15%). Authors predominantly applied MBI's default settings and chose "mechanistic/non-clinical" rather than "clinical" MBI even when testing clinical interventions (only 16 studies out of 232 used clinical MBI). Using these data, we can estimate the Type I error rates for the typical MBI study. Authors frequently made dichotomous claims about effects based on the MBI criterion of a "likely" effect and sometimes based on the MBI criterion of a "possible" effect. When the sample size is n = 8 to 15 per group, these inferences have Type I error rates of 12%-22% and 22%-45%, respectively. High Type I error rates were compounded by multiple testing: Authors reported results from a median of 30 tests related to outcomes; and few studies specified a primary outcome (14%). We conclude that MBI has promoted small studies, promulgated a "black box" approach to statistics, and led to numerous papers where the conclusions are not supported by the data. Amidst debates over the role of p-values and significance testing in science, MBI also provides an important natural experiment: we find no evidence that moving researchers away from p-values or null hypothesis significance testing makes them less prone to dichotomization or over-interpretation of findings.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0235318PLOS
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7319293PMC
September 2020

Ten Common Statistical Errors from All Phases of Research, and Their Fixes.

PM R 2020 06 23;12(6):610-614. Epub 2020 May 23.

Department of Epidemiology and Population Health, Stanford University, Stanford, CA, USA.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1002/pmrj.12395DOI Listing
June 2020

The 2016 California policy to eliminate nonmedical vaccine exemptions and changes in vaccine coverage: An empirical policy analysis.

PLoS Med 2019 12 23;16(12):e1002994. Epub 2019 Dec 23.

Department of Medicine, University of California, San Francisco, San Francisco, California, United States of America.

Background: Vaccine hesitancy, the reluctance or refusal to receive vaccination, is a growing public health problem in the United States and globally. State policies that eliminate nonmedical ("personal belief") exemptions to childhood vaccination requirements are controversial, and their effectiveness to improve vaccination coverage remains unclear given limited rigorous policy analysis. In 2016, a California policy (Senate Bill 277) eliminated nonmedical exemptions from school entry requirements. The objective of this study was to estimate the association between California's 2016 policy and changes in vaccine coverage.

Methods And Findings: We used a quasi-experimental state-level synthetic control analysis and a county-level difference-in-differences analysis to estimate the impact of the 2016 California policy on vaccination coverage and prevalence of exemptions to vaccine requirements (nonmedical and medical). We used publicly available state-level data from the US Centers for Disease Control and Prevention on coverage of measles, mumps, and rubella (MMR) vaccination, nonmedical exemption, and medical exemption in children entering kindergarten. We used county-level data individually requested from state departments of public health on overall vaccine coverage and exemptions. Based on data availability, we included state-level data for 45 states, including California, from 2011 to 2017 and county-level data for 17 states from 2010 to 2017. The prespecified primary study outcome was MMR vaccination in the state analysis and overall vaccine coverage in the county analysis. In the state-level synthetic control analysis, MMR coverage in California increased by 3.3% relative to its synthetic control in the postpolicy period (top 2 of 43 states evaluated in the placebo tests, top 5%), nonmedical exemptions decreased by 2.4% (top 2 of 43 states evaluated in the placebo tests, top 5%), and medical exemptions increased by 0.4% (top 1 of 44 states evaluated in the placebo tests, top 2%). In the county-level analysis, overall vaccination coverage increased by 4.3% (95% confidence interval [CI] 2.9%-5.8%, p < 0.001), nonmedical exemptions decreased by 3.9% (95% CI 2.4%-5.4%, p < 0.001), and medical exemptions increased by 2.4% (95% CI 2.0%-2.9%, p < 0.001). Changes in vaccination coverage across counties after the policy implementation from 2015 to 2017 ranged from -6% to 26%, with larger increases in coverage in counties with lower prepolicy vaccine coverage. Results were robust to alternative model specifications. The limitations of the study were the exclusion of a subset of US states from the analysis and the use of only 2 years of postpolicy data based on data availability.

Conclusions: In this study, implementation of the California policy that eliminated nonmedical childhood vaccine exemptions was associated with an estimated increase in vaccination coverage and a reduction in nonmedical exemptions at state and county levels. The observed increase in medical exemptions was offset by the larger reduction in nonmedical exemptions. The largest increases in vaccine coverage were observed in the most "high-risk" counties, meaning those with the lowest prepolicy vaccine coverage. Our findings suggest that government policies removing nonmedical exemptions can be effective at increasing vaccination coverage.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1371/journal.pmed.1002994DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC6927583PMC
December 2019

How to Be a Statistical Detective.

PM R 2020 02 18;12(2):211-215. Epub 2020 Jan 18.

Department of Epidemiology and Population Health, Stanford University, Stanford, CA.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1002/pmrj.12305DOI Listing
February 2020

Magnitude-based Inference is not Bayesian and is not a valid method of inference.

Scand J Med Sci Sports 2019 09;29(9):1428-1436

Department of Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer Center, New York, New York.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1111/sms.13491DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC6684445PMC
September 2019

Case Studies in Statistics.

PM R 2019 06 21;11(6):654-656. Epub 2019 May 21.

Department of Health Research and Policy, Stanford University, Stanford, CA.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1002/pmrj.12178DOI Listing
June 2019

Lack of Diagnostic Utility of "Amino Acid Dysregulation Metabotypes".

Biol Psychiatry 2019 04 27;85(7):e41-e42. Epub 2018 Dec 27.

Division of Epidemiology, Department of Health Research and Policy, Stanford University, Stanford, California. Electronic address:

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.biopsych.2018.11.012DOI Listing
April 2019

Bone stress injuries in male distance runners: higher modified Female Athlete Triad Cumulative Risk Assessment scores predict increased rates of injury.

Br J Sports Med 2019 Feb 22;53(4):237-242. Epub 2018 Dec 22.

Stanford University, Stanford, California, USA.

Objectives: Bone stress injuries (BSI) are common in runners of both sexes. The purpose of this study was to determine if a modified Female Athlete Triad Cumulative Risk Assessment tool would predict BSI in male distance runners.

Methods: 156 male runners at two collegiate programmes were studied using mixed retrospective and prospective design for a total of 7 years. Point values were assigned using risk assessment categories including low energy availability, low body mass index (BMI), low bone mineral density (BMD) and prior BSI. The outcome was subsequent development of BSI. Statistical models used a mixed effects Poisson regression model with p<0.05 as threshold for significance. Two regression analyses were performed: (1) baseline risk factors as the independent variable; and (2) annual change in risk factors (longitudinal data) as the independent variable.

Results: 42/156 runners (27%) sustained 61 BSIs over an average 1.9 years of follow-up. In the baseline risk factor model, each 1 point increase in prior BSI score was associated with a 57% increased risk for prospective BSI (p=0.0042) and each 1 point increase in cumulative risk score was associated with a 37% increase in prospective BSI risk (p=0.0079). In the longitudinal model, each 1 point increase in cumulative risk score was associated with a 27% increase in prospective BSI risk (p=0.05). BMI (rate ratio (RR)=1.91, p=0.11) and BMD (RR=1.58, p=0.19) risk scores were not associated with BSI.

Conclusion: A modified cumulative risk assessment tool may help identify male runners at elevated risk for BSI. Identifying risk factors may guide treatment and prevention strategies.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1136/bjsports-2018-099861DOI Listing
February 2019

The Burden of Caring for a Child or Adolescent With Pediatric Acute-Onset Neuropsychiatric Syndrome (PANS): An Observational Longitudinal Study.

J Clin Psychiatry 2018 12 11;80(1). Epub 2018 Dec 11.

Pediatric Divisions of Child & Adolescent Psychiatry, Stanford University School of Medicine, Palo Alto, California, USA.

Objective: To describe the longitudinal association between disease severity, time established in clinical treatment, and caregiver burden in a community-based patient population diagnosed with pediatric acute-onset neuropsychiatric syndrome (PANS).

Methods: The study included an observational longitudinal cohort design, with Caregiver Burden Inventories (CBIs) collected between April 2013 and November 2016 at the Stanford PANS multidisciplinary clinic. Inclusion criteria for this study were as follows: pediatric patients meeting strict PANS/pediatric autoimmune neuropsychiatric disorders associated with streptococcal infections (PANDAS) diagnostic criteria (n = 187), having a caregiver fill out at least 1 complete CBI during a disease flare (n = 114); and having family who lives locally (n = 97). For longitudinal analyses, only patients whose caregiver had filled out 2 or more CBIs (n = 94 with 892 CBIs) were included. In the study sample, most primary caregivers were mothers (69 [71.1%] of 97), the majority of PANS patients were male (58 [59.8%] of 97), and mean age at PANS onset was 8.8 years.

Results: In a patient's first flare tracked by the clinic, 50% of caregivers exceeded the caregiver burden score threshold used to determine respite need in care receiver adult populations. Longitudinally, flares, compared with quiescence, predicted increases in mean CBI score (6.6 points; 95% CI, 5.1 to 8.0). Each year established in clinic predicted decreased CBI score (-3.5 points per year; 95% CI, -2.3 to -4.6). Also, shorter time between PANS onset and entry into the multidisciplinary clinic predicted greater improvement in mean CBI score over time (0.7 points per year squared; 95% CI, 0.1 to 1.3). Time between PANS onset and treatment with antibiotics or immunomodulation did not moderate the relationship between CBI score and time in clinic.

Conclusions: PANS caregivers suffer high caregiver burden. Neuropsychiatric disease severity predicts increased caregiver burden. Caregiver burden tends to decrease over time in a group of patients undergoing clinical treatment at a specialty PANS clinic. This decrease could be independent of clinical treatment.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.4088/JCP.17m12091DOI Listing
December 2018

Dealing With Binary Repeated Measures Data.

PM R 2018 12 22;10(12):1412-1416. Epub 2018 Nov 22.

Division of Epidemiology, Department of Health Research and Policy, Stanford University, HRP Redwood Building, Stanford, CA 94305(‡). Electronic address:

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.pmrj.2018.11.002DOI Listing
December 2018

Response.

Med Sci Sports Exerc 2018 12;50(12):2611

Department of Health Research and Policy Division of Epidemiology Stanford University Stanford, CA.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1249/MSS.0000000000001737DOI Listing
December 2018

Response.

Med Sci Sports Exerc 2019 03;51(3):600

Division of Epidemiology Department of Health Research and Policy Stanford University Stanford, CA.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1249/MSS.0000000000001824DOI Listing
March 2019

A Checklist for Analyzing Data.

PM R 2018 09;10(9):963-965

Department of Health Research and Policy, Division of Epidemiology, Stanford University, HRP Redwood Building, Stanford, CA 94305(†). Electronic address:

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.pmrj.2018.07.015DOI Listing
September 2018

Breastfeeding mitigates the effects of maternal HIV on infant infectious morbidity in the Option B+ era.

AIDS 2018 10;32(16):2383-2391

Division of Immunology, Institute of Infections Disease and Molecular Medicine, University of Cape Town, South Africa.

Objective: The effects of in-utero HIV-exposure on infectious morbidity and mortality in settings with universal maternal treatment and high breastfeeding rates are unclear. Further, the benefits of exclusive feeding options have not been assessed in the Option B+ era. We investigated these in two African settings with high breastfeeding uptake and good HIV treatment infrastructure during the first year of life.

Methods: Cox regression with time-changing variables in a birth cohort of 749 HIV-exposed uninfected and HIV-unexposed uninfected infants from Cape Town, South Africa and Jos, Nigeria.

Results: There was no difference in infectious morbidity incidence between HIV-exposed uninfected and HIV-unexposed uninfected infants (hazard ratio 1.01; 95% CI 0.78-1.32) after adjusting for confounding variables. Formula-fed infants had significantly higher infectious morbidity incidence when compared with exclusively breastfed infants (hazard ratio 1.64; 95% CI 1.03-2.63) and mixed-breastfed infants (hazard ratio 1.42; 95% CI 1.00-2.02) after adjusting for potential confounding variables. There was no significant difference in mortality among HIV-exposed infants and HIV-unexposed infants during the first year of life in this cohort (2.04 versus 0.94%, P = 0.38). Notably, exclusive breastfeeding for only 4 months had protective effects on morbidity up to 1 year.

Conclusion: In settings with universal antiretroviral coverage and high breastfeeding rates, breastfeeding mitigates the effects of in-utero HIV exposure among infants during the first year of life. These findings support previous recommendations for exclusive breastfeeding among HIV-infected women and highlight the role that breastfeeding plays on the health of infants in settings where exclusive breastfeeding is not always feasible or where replacement feeding is recommended.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1097/QAD.0000000000001974DOI Listing
October 2018

Sport and Triad Risk Factors Influence Bone Mineral Density in Collegiate Athletes.

Med Sci Sports Exerc 2018 12;50(12):2536-2543

Department of Orthopaedic Surgery, Boswell Human Performance Laboratory, Stanford, CA.

Purpose: Athletes in weight-bearing sports may benefit from higher bone mineral density (BMD). However, some athletes are at risk for impaired BMD with female athlete triad (Triad). The purpose of this study is to understand the influence of sports participation and Triad on BMD. We hypothesize that athletes in high-impact and multidirectional loading sports will have highest BMD, whereas nonimpact and low-impact sports will have lowest BMD. Triad risk factors are expected to reduce BMD values independent of sports participation.

Methods: Two hundred thirty-nine female athletes participating in 16 collegiate sports completed dual-energy x-ray absorptiometry (DXA) scans to measure BMD z-scores of the lumbar spine (LS) and total body (TB). Height and weight were measured to calculate body mass index (BMI). Triad risk assessment variables were obtained from preparticipation examination. Mean BMD z-scores were compared between sports and by sport category (high-impact, multidirectional, low-impact, and nonimpact). Multivariable regression analyses were performed to identify differences of BMD z-scores accounting for Triad and body size/composition.

Results: Athlete populations with lowest average BMD z-scores included synchronized swimming (LS, -0.34; TB, 0.21) swimming/diving (LS, 0.34; TB, -0.06), crew/rowing (LS, 0.27; TB, 0.62), and cross-country (LS, 0.29; TB, 0.91). Highest values were in gymnastics (LS, 1.96; TB, 1.37), volleyball (LS, 1.90; TB, 1.74), basketball (LS, 1.73; TB, 1.99), and softball (LS, 1.68; TB, 1.78). All Triad risk factors were associated with lower BMD z-scores in univariable analyses; only low BMI and oligomenorrhea/amenorrhea were associated in multivariable analyses (all P < 0.05). Accounting for Triad risk factors and body size/composition, high-impact sports were associated with higher LS and TB BMD z-scores and nonimpact sports with lower LS and TB BMD z-scores compared to low-impact sport (all P < 0.05).

Conclusions: Both sport type and Triad risk factors influence BMD. Athletes in low-impact and nonimpact sports and athletes with low BMI and oligomenorrhea/amenorrhea are at highest risk for reduced BMD.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1249/MSS.0000000000001711DOI Listing
December 2018

Reply.

PM R 2018 05;10(5):563

Department of Health Research and Policy, Division of Epidemiology, Stanford University, Stanford, CA.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.pmrj.2017.12.013DOI Listing
May 2018

The Problem with "Magnitude-based Inference".

Med Sci Sports Exerc 2018 Oct;50(10):2166-2176

Division of Epidemiology, Department of Health Research and Policy, Stanford University, Stanford, CA.

Purpose: A statistical method called "magnitude-based inference" (MBI) has gained a following in the sports science literature, despite concerns voiced by statisticians. Its proponents have claimed that MBI exhibits superior type I and type II error rates compared with standard null hypothesis testing for most cases. I have performed a reanalysis to evaluate this claim.

Methods: Using simulation code provided by MBI's proponents, I estimated type I and type II error rates for clinical and nonclinical MBI for a range of effect sizes, sample sizes, and smallest important effects. I plotted these results in a way that makes transparent the empirical behavior of MBI. I also reran the simulations after correcting mistakes in the definitions of type I and type II error provided by MBI's proponents. Finally, I confirmed the findings mathematically; and I provide general equations for calculating MBI's error rates without the need for simulation.

Results: Contrary to what MBI's proponents have claimed, MBI does not exhibit "superior" type I and type II error rates to standard null hypothesis testing. As expected, there is a tradeoff between type I and type II error. At precisely the small-to-moderate sample sizes that MBI's proponents deem "optimal," MBI reduces the type II error rate at the cost of greatly inflating the type I error rate-to two to six times that of standard hypothesis testing.

Conclusions: Magnitude-based inference exhibits worrisome empirical behavior. In contrast to standard null hypothesis testing, which has predictable type I error rates, the type I error rates for MBI vary widely depending on the sample size and choice of smallest important effect, and are often unacceptably high. Magnitude-based inference should not be used.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1249/MSS.0000000000001645DOI Listing
October 2018

Instrumental Variables: Uses and Limitations.

PM R 2018 03;10(3):303-308

Division of Epidemiology, Department of Health Research and Policy, Stanford University, HRP Redwood Building, Stanford, CA 94305.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.pmrj.2018.02.002DOI Listing
March 2018

Semantic Memory in the Clinical Progression of Alzheimer Disease.

Cogn Behav Neurol 2017 09;30(3):81-89

Departments of *Health Research & Policy (Epidemiology) and †Neurology & Neurological Sciences, Stanford University, Stanford, California ‡Department of Clinical Epidemiology, Aarhus University, Aarhus, Denmark.

Background And Objective: Semantic memory measures may be useful in tracking and predicting progression of Alzheimer disease. We investigated relationships among semantic memory tasks and their 1-year predictive value in women with Alzheimer disease.

Methods: We conducted secondary analyses of a randomized clinical trial of raloxifene in 42 women with late-onset mild-to-moderate Alzheimer disease. We assessed semantic memory with tests of oral confrontation naming, category fluency, semantic recognition and semantic naming, and semantic density in written narrative discourse. We measured global cognition (Alzheimer Disease Assessment Scale, cognitive subscale), dementia severity (Clinical Dementia Rating sum of boxes), and daily function (Activities of Daily Living Inventory) at baseline and 1 year.

Results: At baseline and 1 year, most semantic memory scores correlated highly or moderately with each other and with global cognition, dementia severity, and daily function. Semantic memory task performance at 1 year had worsened one-third to one-half standard deviation. Factor analysis of baseline test scores distinguished processes in semantic and lexical retrieval (semantic recognition, semantic naming, confrontation naming) from processes in lexical search (semantic density, category fluency). The semantic-lexical retrieval factor predicted global cognition at 1 year. Considered separately, baseline confrontation naming and category fluency predicted dementia severity, while semantic recognition and a composite of semantic recognition and semantic naming predicted global cognition. No individual semantic memory test predicted daily function.

Conclusions: Semantic-lexical retrieval and lexical search may represent distinct aspects of semantic memory. Semantic memory processes are sensitive to cognitive decline and dementia severity in Alzheimer disease.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1097/WNN.0000000000000131DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC5617354PMC
September 2017

Getting the Right Answer: Four Statistical Principles.

PM R 2017 09;9(9):933-937

Department of Health Research and Policy, Stanford University, Stanford, CA(∗). Electronic address:

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.pmrj.2017.06.015DOI Listing
September 2017

Reliability Statistics.

PM R 2017 06;9(6):622-628

Stanford University, HRP Redwood Building, Welch Road, Stanford, CA 94305(∗). Electronic address:

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.pmrj.2017.05.001DOI Listing
June 2017

Evaluation of Evidence of Statistical Support and Corroboration of Subgroup Claims in Randomized Clinical Trials.

JAMA Intern Med 2017 04;177(4):554-560

Department of Health Research and Policy, Stanford University School of Medicine, Stanford, California2Meta-Research Innovation Center at Stanford (METRICS), Stanford University School of Medicine, Stanford, California3Department of Medicine, Stanford University School of Medicine, Stanford, California4Stanford Prevention Research Center, Department of Medicine, Stanford University School of Medicine, Stanford, California6Department of Statistics, Stanford University School of Humanities and Sciences, Stanford, California.

Importance: Many published randomized clinical trials (RCTs) make claims for subgroup differences.

Objective: To evaluate how often subgroup claims reported in the abstracts of RCTs are actually supported by statistical evidence (P < .05 from an interaction test) and corroborated by subsequent RCTs and meta-analyses.

Data Sources: This meta-epidemiological survey examines data sets of trials with at least 1 subgroup claim, including Subgroup Analysis of Trials Is Rarely Easy (SATIRE) articles and Discontinuation of Randomized Trials (DISCO) articles. We used Scopus (updated July 2016) to search for English-language articles citing each of the eligible index articles with at least 1 subgroup finding in the abstract.

Study Selection: Articles with a subgroup claim in the abstract with or without evidence of statistical heterogeneity (P < .05 from an interaction test) in the text and articles attempting to corroborate the subgroup findings.

Data Extraction And Synthesis: Study characteristics of trials with at least 1 subgroup claim in the abstract were recorded. Two reviewers extracted the data necessary to calculate subgroup-level effect sizes, standard errors, and the P values for interaction. For individual RCTs and meta-analyses that attempted to corroborate the subgroup findings from the index articles, trial characteristics were extracted. Cochran Q test was used to reevaluate heterogeneity with the data from all available trials.

Main Outcomes And Measures: The number of subgroup claims in the abstracts of RCTs, the number of subgroup claims in the abstracts of RCTs with statistical support (subgroup findings), and the number of subgroup findings corroborated by subsequent RCTs and meta-analyses.

Results: Sixty-four eligible RCTs made a total of 117 subgroup claims in their abstracts. Of these 117 claims, only 46 (39.3%) in 33 articles had evidence of statistically significant heterogeneity from a test for interaction. In addition, out of these 46 subgroup findings, only 16 (34.8%) ensured balance between randomization groups within the subgroups (eg, through stratified randomization), 13 (28.3%) entailed a prespecified subgroup analysis, and 1 (2.2%) was adjusted for multiple testing. Only 5 (10.9%) of the 46 subgroup findings had at least 1 subsequent pure corroboration attempt by a meta-analysis or an RCT. In all 5 cases, the corroboration attempts found no evidence of a statistically significant subgroup effect. In addition, all effect sizes from meta-analyses were attenuated toward the null.

Conclusions And Relevance: A minority of subgroup claims made in the abstracts of RCTs are supported by their own data (ie, a significant interaction effect). For those that have statistical support (P < .05 from an interaction test), most fail to meet other best practices for subgroup tests, including prespecification, stratified randomization, and adjustment for multiple testing. Attempts to corroborate statistically significant subgroup differences are rare; when done, the initially observed subgroup differences are not reproduced.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1001/jamainternmed.2016.9125DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC6657347PMC
April 2017

Association of the Female Athlete Triad Risk Assessment Stratification to the Development of Bone Stress Injuries in Collegiate Athletes.

Am J Sports Med 2017 Feb 30;45(2):302-310. Epub 2016 Dec 30.

Division of Physical Medicine and Rehabilitation, Department of Orthopaedic Surgery, Stanford University, Stanford, California, USA.

Background: The female athlete triad (referred to as the triad) contributes to adverse health outcomes, including bone stress injuries (BSIs), in female athletes. Guidelines were published in 2014 for clinical management of athletes affected by the triad.

Purpose: This study aimed to (1) classify athletes from a collegiate population of 16 sports into low-, moderate-, and high-risk categories using the Female Athlete Triad Cumulative Risk Assessment score and (2) evaluate the predictive value of the risk categories for subsequent BSIs.

Study Design: Cohort study; Level of evidence, 3.

Methods: A total of 323 athletes completed both electronic preparticipation physical examination and dual-energy x-ray absorptiometry scans. Of these, 239 athletes with known oligomenorrhea/amenorrhea status were assigned to a low-, moderate-, or high-risk category. Chart review was used to identify athletes who sustained a subsequent BSI during collegiate sports participation; the injury required a physician diagnosis and imaging confirmation.

Results: Of 239 athletes, 61 (25.5%) were classified into moderate-risk and 9 (3.8%) into high-risk categories. Sports with the highest proportion of athletes assigned to the moderate- and high-risk categories included gymnastics (56.3%), lacrosse (50%), cross-country (48.9%), swimming/diving (42.9%), sailing (33%), and volleyball (33%). Twenty-five athletes (10.5%) assigned to risk categories sustained ≥1 BSI. Cross-country runners contributed the majority of BSIs (16; 64%). After adjusting for age and participation in cross-country, we found that moderate-risk athletes were twice as likely as low-risk athletes to sustain a BSI (risk ratio [RR], 2.6; 95% confidence interval [95% CI], 1.3-5.5) and high-risk athletes were nearly 4 times as likely (RR, 3.8; 95% CI, 1.8-8.0). When examining the 6 individual components of the triad risk assessment score, both the oligomenorrhea/amenorrhea score ( P = .0069) and the prior stress fracture/reaction score ( P = .0315) were identified as independent predictors for subsequent BSIs (after adjusting for cross-country participation and age).

Conclusion: Using published guidelines, 29% of female collegiate athletes in this study were classified into moderate- or high-risk categories using the Female Athlete Triad Cumulative Risk Assessment Score. Moderate- and high-risk athletes were more likely to subsequently sustain a BSI; most BSIs were sustained by cross-country runners.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1177/0363546516676262DOI Listing
February 2017

The Value of Scatter Plots.

PM R 2016 12 2;8(12):1213-1217. Epub 2016 Nov 2.

Department of Health Research and Policy, Stanford University, Stanford, CA 94305(∗). Electronic address:

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.pmrj.2016.10.018DOI Listing
December 2016