Meta-analysis prediction intervals are under reported in sport and exercise medicine.

AIM: Prediction intervals are a useful measure of uncertainty for meta-analyses that capture the likely effect size of a new (similar) study based on the included studies. In comparison, confidence intervals reflect the uncertainty around the point estimate but provide an incomplete summary of the underlying heterogeneity in the meta-analysis. This study aimed to estimate (i) the proportion of meta-analysis studies that report a prediction interval in sports medicine; and (ii) the proportion of studies with a discrepancy between the reported confidence interval and a calculated prediction interval. METHODS: We screened, at random, 1500 meta-analysis studies published between 2012 and 2022 in highly ranked sports medicine and medical journals. Articles that used a random effect meta-analysis model were included in the study. We randomly selected one meta-analysis from each article to extract data from, which included the number of estimates, the pooled effect, and the confidence and prediction interval. RESULTS: Of the 1500 articles screened, 866 (514 from sports medicine) used a random effect model. The probability of a prediction interval being reported in sports medicine was 1.7% (95% CI = 0.9%, 3.3%). In medicine the probability was 3.9% (95% CI = 2.4%, 6.6%). A prediction interval was able to be calculated for 220 sports medicine studies. For 60% of these studies, there was a discrepancy in study findings between the reported confidence interval and the calculated prediction interval. Prediction intervals were 3.4 times wider than confidence intervals. CONCLUSION: Very few meta-analyses report prediction intervals and hence are prone to missing the impact of between-study heterogeneity on the overall conclusions. The widespread misinterpretation of random effect meta-analyses could mean that potentially harmful treatments, or those lacking a sufficient evidence base, are being used in practice. Authors, reviewers, and editors should be aware of the importance of prediction intervals.

Process evaluation of a tailored nudge intervention to promote appropriate care and treatment of older patients at the end-of-life.

BACKGROUND: Non-beneficial treatment affects a considerable proportion of older people in hospital, and some will choose to decline invasive treatments when they are approaching the end of their life. The Intervention for Appropriate Care and Treatment (InterACT) intervention was a 12-month stepped wedge randomised controlled trial with an embedded process evaluation in three hospitals in Brisbane, Australia. The aim was to increase appropriate care and treatment decisions for older people at the end-of-life, through implementing a nudge intervention in the form of a prospective feedback loop. However, the trial results indicated that the expected practice change did not occur. The process evaluation aimed to assess implementation using the Consolidated Framework for Implementation Research, identify barriers and enablers to implementation and provide insights into the lack of effect of the InterACT intervention. METHODS: Qualitative data collection involved 38 semi-structured interviews with participating clinicians, members of the executive advisory groups overseeing the intervention at a site level, clinical auditors, and project leads. Online interviews were conducted at two times: implementation onset and completion. Data were coded to the Consolidated Framework for Implementation Research and deductively analysed. RESULTS: Overall, clinicians felt the premise and clinical reasoning behind InterACT were strong and could improve patient management. However, several prominent barriers affected implementation. These related to the potency of the nudge intervention and its integration into routine clinical practice, clinician beliefs and perceived self-efficacy, and wider contextual factors at the health system level. CONCLUSIONS: An intervention designed to change clinical practice for patients at or near to end-of-life did not have the intended effect. Future interventions targeting this area of care should consider using multi-component strategies that address the identified barriers to implementation and clinician change of practice. TRIAL REGISTRATION: Australia New Zealand Clinical Trial Registry (ANZCTR), ACTRN12619000675123p (approved 06/05/2019).

Gestational diabetes mellitus screening and diagnosis criteria before and during the COVID-19 pandemic: a retrospective pre-post study.

OBJECTIVE: To determine whether perinatal outcomes after excluding gestational diabetes mellitus (GDM) on the basis of fasting venous plasma glucose (FVPG) assessment during the coronavirus disease 2019 (COVID-19) pandemic in 2020 were similar to those during the preceding year after excluding GDM using the standard oral glucose tolerance test (OGTT) procedure. DESIGN: Retrospective pre-post study. SETTING, PARTICIPANTS: All women who gave birth in Queensland during 1 July - 31 December 2019 and 1 July - 31 December 2020. MAIN OUTCOME MEASURES: Perinatal (maternal and neonatal) outcomes for pregnant women assessed for GDM, by assessment method (2019: OGTT/glycated haemoglobin [HbA1c ] assessment; 2020: GDM could be excluded by an FVPG value below 4.7 mmol/L). RESULTS: 3968 of 29 113 pregnant women in Queensland during 1 July - 31 December 2019 (13.6%) were diagnosed with GDM, and 4029 of 28 778 during 1 July - 31 December 2020 (14.0%). In 2020, FVPG assessments established GDM in 216 women (1.1%) and excluded it in 1660 (5.8%). The frequencies of most perinatal outcomes were similar for women without GDM in 2019 and those for whom it was excluded in 2020 on the basis of FVPG values; the exception was caesarean delivery, for which the estimated probability increase in 2020 was 3.9 percentage points (95% credibility interval, 2.2-5.6 percentage points), corresponding to an extra 6.5 caesarean deliveries per 1000 births. The probabilities of several outcomes - respiratory distress, neonatal intensive care or special nursery admission, large for gestational age babies - were about one percentage point higher for women without GDM in 2020 (excluding those diagnosed on the basis of FVPG assessment alone) than for women without GDM in 2019. CONCLUSIONS: Identifying women at low absolute risk of gestational diabetes-related pregnancy complications on the basis of FVPG assessment as an initial step in GDM screening could reduce the burden for pregnant women and save the health system substantial costs.

Is requiring Research Integrity Advisors a useful policy for improving research integrity? A census of advisors in Australia.

Research Integrity Advisors are used in Australia to provide impartial guidance to researchers who have questions about any aspect of responsible research practice. Every Australian institution conducting research must provide access to trained advisors. This national policy could be an important part of creating a safe environment for discussing research integrity issues and thus resolving issues. We conducted the first formal study of advisors, using a census of every Australian advisor to discover their workload and attitudes to their role. We estimated there are 739 advisors nationally. We received responses to our questions from 192. Most advisors had a very light workload, with an median of just 0.5 days per month. Thirteen percent of advisors had not received any training, and some advisors only discovered they were an advisor after our approach. Most advisors were positive about their ability to help colleagues deal with integrity issues. The main desired changes were for greater advertising of their role and a desire to promote good practice rather than just supporting potential issues. Advisors might be a useful policy for supporting research integrity, but some advisors need better institutional support in terms of training and raising awareness.

Vital signs-based deterioration prediction model assumptions can lead to losses in prediction performance.

OBJECTIVE: Vital signs-based models are complicated by repeated measures per patient and frequently missing data. This paper investigated the impacts of common vital signs modeling assumptions during clinical deterioration prediction model development. STUDY DESIGN AND SETTING: Electronic medical record (EMR) data from five Australian hospitals (1 January 2019-31 December 2020) were used. Summary statistics for each observation's prior vital signs were created. Missing data patterns were investigated using boosted decision trees, then imputed with common methods. Two example models predicting in-hospital mortality were developed, as follows: logistic regression and eXtreme Gradient Boosting. Model discrimination and calibration were assessed using the C-statistic and nonparametric calibration plots. RESULTS: The data contained 5,620,641 observations from 342,149 admissions. Missing vitals were associated with observation frequency, vital sign variability, and patient consciousness. Summary statistics improved discrimination slightly for logistic regression and markedly for eXtreme Gradient Boosting. Imputation method led to notable differences in model discrimination and calibration. Model calibration was generally poor. CONCLUSION: Summary statistics and imputation methods can improve model discrimination and reduce bias during model development, but it is questionable whether these differences are clinically significant. Researchers should consider why data are missing during model development and how this may impact clinical utility.

The bias for statistical significance in sport and exercise medicine.

OBJECTIVES: We aimed to examine the bias for statistical significance using published confidence intervals in sport and exercise medicine research. DESIGN: Observational study. METHODS: The abstracts of 48,390 articles, published in 18 sports and exercise medicine journals between 2002 and 2022, were searched using a validated text-mining algorithm that identified and extracted ratio confidence intervals (odds, hazard, and risk ratios). The algorithm identified 1744 abstracts that included ratio confidence intervals, from which 4484 intervals were extracted. After excluding ineligible intervals, the analysis used 3819 intervals, reported as 95 % confidence intervals, from 1599 articles. The cumulative distributions of lower and upper confidence limits were plotted to identify any abnormal patterns, particularly around a ratio of 1 (the null hypothesis). The distributions were compared to those from unbiased reference data, which was not subjected to p-hacking or publication bias. A bias for statistical significance was further investigated using a histogram plot of z-values calculated from the extracted 95 % confidence intervals. RESULTS: There was a marked change in the cumulative distribution of lower and upper bound intervals just over and just under a ratio of 1. The bias for statistical significance was also clear in a stark under-representation of z-values between -1.96 and +1.96, corresponding to p-values above 0.05. CONCLUSIONS: There was an excess of published research with statistically significant results just below the standard significance threshold of 0.05, which is indicative of publication bias. Transparent research practices, including the use of registered reports, are needed to reduce the bias in published research.

Morbid obesity's impact on COVID-19 patients requiring venovenous extracorporeal membrane oxygenation: The covid-19 critical care consortium database review.

Introduction: Obesity is associated with a worse prognosis in COVID-19 patients with acute respiratory distress syndrome (ARDS). Veno-venous (V-V) Extracorporeal Membrane Oxygenation (ECMO) can be a rescue option, however, the direct impact of morbid obesity in this select group of patients remains unclear.Methods: This is an observational study of critically ill adults with COVID-19 and ARDS supported by V-V ECMO. Data are from 82 institutions participating in the COVID-19 Critical Care Consortium international registry. Patients were admitted between 12 January 2020 to 27 April 2021. They were stratified based on Body Mass Index (BMI) at 40 kg/m2. The endpoint was survival to hospital discharge.Results: Complete data available on 354 of 401 patients supported on V-V ECMO. The characteristics of the high BMI (>40 kg/m2) and lower BMI (≤40 kg/m2) groups were statistically similar. However, the 'high BMI' group were comparatively younger and had a lower APACHE II score. Using survival analysis, older age (Hazard Ratio, HR 1.49 per-10-years, CI 1.25-1.79) and higher BMI (HR 1.15 per-5 kg/m2 increase, CI 1.03-1.28) were associated with a decreased patient survival. A safe BMI threshold above which V-V ECMO would be prohibitive was not apparent and instead, the risk of an adverse outcome increased linearly with BMI.Conclusion: In COVID-19 patients with severe ARDS who require V-V ECMO, there is an increased risk of death associated with age and BMI. The risk is linear and there is no BMI threshold beyond which the risk for death greatly increases.

Impact of a prospective feedback loop on care review activities in older patients at the end of life. A stepped-wedge randomised trial.

BACKGROUND: Hospitalisation rates for older people are increasing, with end-of-life care becoming a more medicalised experience. Innovative approaches are warranted to support early identification of the end-of-life phase, communicate prognosis, provide care consistent with people's preferences, and improve the use of healthcare resources. The Intervention for Appropriate Care and Treatment (InterACT) trial aimed to increase appropriate care and treatment decisions for older people at the end of life, through implementation of a prospective feedback loop. This paper reports on the care review outcomes. METHODS: A stepped-wedge randomised controlled trial was conducted in three large acute hospitals in Queensland, Australia between May 2020 and June 2021. The trial identified older people nearing the end of life using two validated tools for detecting deterioration and short-term death. Admitting clinical teams were provided with details of patients identified as at-risk with the goal of increasing awareness that end of life was approaching to facilitate appropriate patient centred care and avoid non-beneficial treatment. We examined the time between when the patient was identified as 'at-risk' and three outcomes: clinician-led care review discussions, review of care directive measures and palliative care referrals. These were considered useful indicators of appropriate care at the end of life. RESULTS: In two hospitals there was a reduction in the review of care directive measures during the intervention compared with usual care at 21 days (reduced probability of - 0.08; 95% CI: - 0.12 to - 0.04 and - 0.14; 95% CI: - 0.21 to - 0.06). In one hospital there was a large reduction in clinician-led care review discussions at 21 days during the intervention (reduced probability of - 0.20; 95% CI: - 0.28 to - 0.13). There was little change in palliative care referrals in any hospital, with average probability differences at 21 days of - 0.01, 0.02 and 0.04. DISCUSSION: The results are disappointing as an intervention designed to improve care of hospitalised older people appeared to have the opposite effect on care review outcomes. The reasons for this may be a combination of the intervention design and health system challenges due to the pandemic that highlight the complexity of providing more appropriate care at the end of life. TRIAL REGISTRATION: Australia New Zealand Clinical Trial Registry, ACTRN12619000675123 (registered 6 May 2019).

The comparative mortality of an elite group in the long run of history: an observational analysis of politicians from 11 countries.

This study aims to compare the mortality rate and life expectancy of politicians with those of the age and gender-matched general populations. This was an observational analysis of mortality rates of politicians (i.e. members of national parliaments with available data on dates of birth, death and election, gender, and life tables) in 11 developed countries. Politicians were followed from date of first election until either death or the last available year with life table data. Relative mortality differences were estimated using standardised mortality ratios (SMRs). Absolute inequalities were quantified as the difference in survival by deducting a population's remaining life expectancy from politicians' remaining life expectancy at age 45, estimated using Gompertz parametric proportional hazards models. We included 57,561 politicians (with follow-up ranging from 1816-2016 for France to 1949-2017 for Germany). In almost all countries politicians had similar rates of mortality to the general population in the early part of the twentieth century. Relative mortality and survival differences (favouring politicians) increased considerably over the course of the twentieth century, with recent SMRs ranging from 0.45 (95%CI 0.41-0.50) in Italy to 0.82 (95%CI 0.69-0.95) in New Zealand. The peak life expectancy gaps ranged from 4.4 (95% CI, 3.5-5.4) years in the Netherlands to 7.8 (95% CI, 7.2-8.4) years in the US. Our results show large relative and absolute inequalities favouring politicians in every country. In some countries, such as the US, relative inequalities are at the greatest level in over 150 years.

Venovenous extracorporeal membrane oxygenation in patients with acute covid-19 associated respiratory failure: comparative effectiveness study.

OBJECTIVE: To estimate the effect of extracorporeal membrane oxygenation (ECMO) compared with conventional mechanical ventilation on outcomes of patients with covid-19 associated respiratory failure. DESIGN: Observational study. SETTING: 30 countries across five continents, 3 January 2020 to 29 August 2021. PARTICIPANTS: 7345 adults admitted to the intensive care unit with clinically suspected or laboratory confirmed SARS-CoV-2 infection. INTERVENTIONS: ECMO in patients with a partial pressure of arterial oxygen to fraction of inspired oxygen (PaO2/FiO2) ratio <80 mm Hg compared with conventional mechanical ventilation without ECMO. MAIN OUTCOME MEASURE: The primary outcome was hospital mortality within 60 days of admission to the intensive care unit. Adherence adjusted estimates were calculated using marginal structural models with inverse probability weighting, accounting for competing events and for baseline and time varying confounding. RESULTS: 844 of 7345 eligible patients (11.5%) received ECMO at any time point during follow-up. Adherence adjusted mortality was 26.0% (95% confidence interval 24.5% to 27.5%) for a treatment strategy that included ECMO if the PaO2/FiO2 ratio decreased <80 mm Hg compared with 33.2% (31.8% to 34.6%) had patients received conventional treatment without ECMO (risk difference -7.1%, 95% confidence interval -8.2% to -6.1%; risk ratio 0.78, 95% confidence interval 0.75 to 0.82). In secondary analyses, ECMO was most effective in patients aged <65 years and with a PaO2/FiO2 <80 mm Hg or with driving pressures >15 cmH2O during the first 10 days of mechanical ventilation. CONCLUSIONS: ECMO was associated with a reduction in mortality in selected adults with covid-19 associated respiratory failure. Age, severity of hypoxaemia, and duration and intensity of mechanical ventilation were found to be modifiers of treatment effectiveness and should be considered when deciding to initiate ECMO in patients with covid-19.

An observational analysis of the trope "A p-value of < 0.05 was considered statistically significant" and other cut-and-paste statistical methods.

Appropriate descriptions of statistical methods are essential for evaluating research quality and reproducibility. Despite continued efforts to improve reporting in publications, inadequate descriptions of statistical methods persist. At times, reading statistical methods sections can conjure feelings of dèjá vu, with content resembling cut-and-pasted or "boilerplate text" from already published work. Instances of boilerplate text suggest a mechanistic approach to statistical analysis, where the same default methods are being used and described using standardized text. To investigate the extent of this practice, we analyzed text extracted from published statistical methods sections from PLOS ONE and the Australian and New Zealand Clinical Trials Registry (ANZCTR). Topic modeling was applied to analyze data from 111,731 papers published in PLOS ONE and 9,523 studies registered with the ANZCTR. PLOS ONE topics emphasized definitions of statistical significance, software and descriptive statistics. One in three PLOS ONE papers contained at least 1 sentence that was a direct copy from another paper. 12,675 papers (11%) closely matched to the sentence "a p-value < 0.05 was considered statistically significant". Common topics across ANZCTR studies differentiated between study designs and analysis methods, with matching text found in approximately 3% of sections. Our findings quantify a serious problem affecting the reporting of statistical methods and shed light on perceptions about the communication of statistics as part of the scientific process. Results further emphasize the importance of rigorous statistical review to ensure that adequate descriptions of methods are prioritized over relatively minor details such as p-values and software when reporting research outcomes.

"No Country Bureaucratised its way to Excellence": A Content Analysis of Comments on a Petition to Streamline Australian Research Ethics and Governance Processes.

We created a petition for a national inquiry into the Australian system of research ethics and governance, to inform the politicians about the problems with the existing system. We analyzed the reasons that signatories offered for why signing the petition was important to them. A total of 409 comments (by 805 signatories) focused on five major themes: (1) views on previous changes to the system of research ethics and governance; (2) drawbacks of the existing system; (3) suggested changes to the system; (4) anticipated impacts of changing the system; and (5) miscellaneous/other comments. Comments ranged from several words to over 400 words in length, and most often focused on the procedural aspects, and commented on theme 2: drawbacks of the existing system.

Prenatal exposure to mixtures of persistent environmental chemicals and fetal growth outcomes in Western Australia.

BACKGROUND: Environmental chemicals have been implicated in the etiology of impaired fetal growth. However, few studies have assessed the effects of chemical mixtures or considered the possibility of non-monotonic exposure-response relationships for chemicals that act through the endocrine system. METHODS: We assessed exposure to polybrominated diphenyl ethers, organochlorine pesticides, metals, and perfluorinated alkyl substances in blood and urine samples collected approximately two weeks prior to delivery in 166 non-smoking pregnant women, and subsequent birth weight, length, and head circumference of neonates who were part of the Australian Maternal Exposures to Toxic Substances (AMETS) study. We used Bayesian structured additive regression models with spike-slab priors to estimate mixture effects, identify important exposures, and model non-linearity in exposure-response relationships. RESULTS: Mixtures of polybrominated diphenyl ethers, organochlorine pesticides, metals, and perfluorinated alkyl substances were not associated with fetal growth outcomes. Estimated change in fetal growth outcomes for an increase in exposure from the 25th to 75th percentile suggested no meaningful associations; the strongest evidence was for a small inverse association between birth weight and cesium exposure measured in whole blood (-124 g, 90% credible interval: -240 to -3 g). We identified several chemicals that may be associated with fetal growth non-linearly; however, 90% credible intervals contained small values consistent with no meaningful association. CONCLUSIONS: Using a Bayesian penalized regression method, we assessed the shapes of exposure-response relationships, controlled for confounding by co-exposure, and estimated the single and combined effects of a large mixture of correlated environmental chemicals on fetal growth. Our findings, based on a small sample of mother-neonate pairs, suggest that mixtures of persistent chemicals are not associated with birth weight, length, and head circumference. The potential for non-monotonic relationships between environmental chemicals and fetal growth outcomes warrants further study.

Beneficial Effect of Prone Positioning During Venovenous Extracorporeal Membrane Oxygenation for Coronavirus Disease 2019.

OBJECTIVES: The study investigated the impact of prone positioning during venovenous extracorporeal membrane oxygenation support for coronavirus disease 2019 acute respiratory failure on the patient outcome. DESIGN: An observational study of venovenous extracorporeal membrane oxygenation patients. We used a multistate survival model to compare the outcomes of patients treated with or without prone positioning during extracorporeal membrane oxygenation, which incorporates the dynamic nature of prone positioning and adjusts for potential confounders. SETTING: Seventy-two international institutions participating in the Coronavirus Disease 2019 Critical Care Consortium international registry. PATIENTS: Coronavirus disease 2019 patients who were supported by venovenous extracorporeal membrane oxygenation during the study period. INTERVENTION: None. MEASUREMENTS AND MAIN RESULTS: There were 232 coronavirus disease 2019 patients at 72 participating institutions who were supported with venovenous extracorporeal membrane oxygenation during the study period from February 16, 2020, to October 31, 2020. Proning was used in 176 patients (76%) before initiation of extracorporeal membrane oxygenation and in 67 patients (29%) during extracorporeal membrane oxygenation. Survival to hospital discharge was 33% in the extracorporeal membrane oxygenation prone group versus 22% in the extracorporeal membrane oxygenation supine group. Prone positioning during extracorporeal membrane oxygenation support was associated with reduced mortality (hazard ratio, 0.31; 95% CI, 0.14-0.68). CONCLUSIONS: Our study highlights that prone positioning during venovenous extracorporeal membrane oxygenation support for refractory coronavirus disease 2019-related acute respiratory distress syndrome is associated with reduced mortality. Given the observational nature of the study, a randomized controlled trial of prone positioning on venovenous extracorporeal membrane oxygenation is needed to confirm these findings.

Cost-effectiveness analysis of percutaneous coronary intervention for single-vessel coronary artery disease: an economic evaluation of the ORBITA trial.

OBJECTIVE: To evaluate the cost-effectiveness of percutaneous coronary intervention (PCI) compared with placebo in patients with single-vessel coronary artery disease and angina despite anti-anginal therapy. DESIGN: A cost-effectiveness analysis comparing PCI with placebo. A Markov model was used to measure incremental cost-effectiveness, in cost per quality-adjusted life-years (QALYs) gained, over 12 months. Health utility weights were estimated using responses to the EuroQol 5-level questionnaire, from the Objective Randomised Blinded Investigation with optimal medical Therapy of Angioplasty in stable angina trial and UK preference weights. Costs of procedures and follow-up consultations were derived from Healthcare Resource Group reference costs and drug costs from the National Health Service (NHS) drug tariff. Probabilistic sensitivity analysis was undertaken to test the robustness of results to parameter uncertainty. Scenario analyses were performed to test the effect on results of reduced pharmaceutical costs in patients undergoing PCI, and the effect of patients crossing over from placebo to PCI due to refractory angina within 12 months. SETTING: Five UK NHS hospitals. PARTICIPANTS: 200 adult patients with stable angina and angiographically severe single-vessel coronary artery disease on anti-anginal therapy. INTERVENTIONS: At recruitment, patients received 6 weeks of optimisation of medical therapy for angina after which they were randomised to PCI or a placebo procedure. OUTCOME MEASURES: Incremental cost-effectiveness ratio (ICER) expressed as cost (in £) per QALY gained for PCI compared with placebo. RESULTS: The estimated ICER is £90 218/QALY gained when using PCI compared with placebo in patients receiving medical treatment for angina due to single-vessel coronary artery disease. Results were robust under sensitivity analyses. CONCLUSIONS: The ICER for PCI compared with placebo, in patients with single-vessel coronary artery disease and angina on anti-anginal medication, exceeds the threshold of £30 000 used by the National Institute of Health and Care Excellence when undertaking health technology assessment for the NHS context.Trial registration: The ORBITA study is registered with ClinicalTrials.gov, number NCT02062593.

Performance of variable and function selection methods for estimating the nonlinear health effects of correlated chemical mixtures: A simulation study.

Statistical methods for identifying harmful chemicals in a correlated mixture often assume linearity in exposure-response relationships. Nonmonotonic relationships are increasingly recognized (eg, for endocrine-disrupting chemicals); however, the impact of nonmonotonicity on exposure selection has not been evaluated. In a simulation study, we assessed the performance of Bayesian kernel machine regression (BKMR), Bayesian additive regression trees (BART), Bayesian structured additive regression with spike-slab priors (BSTARSS), generalized additive models with double penalty (GAMDP) and thin plate shrinkage smoothers (GAMTS), multivariate adaptive regression splines (MARS), and lasso penalized regression. We simulated realistic exposure data based on pregnancy exposure to 17 phthalates and phenols in the US National Health and Nutrition Examination Survey using a multivariate copula. We simulated data sets of size N = 250 and compared methods across 32 scenarios, varying by model size and sparsity, signal-to-noise ratio, correlation structure, and exposure-response relationship shapes. We compared methods in terms of their sensitivity, specificity, and estimation accuracy. In most scenarios, BKMR, BSTARSS, GAMDP, and GAMTS achieved moderate to high sensitivity (0.52-0.98) and specificity (0.21-0.99). BART and MARS achieved high specificity (≥0.90), but low sensitivity in low signal-to-noise ratio scenarios (0.20-0.51). Lasso was highly sensitive (0.71-0.99), except for quadratic relationships (≤0.27). Penalized regression methods that assume linearity, such as lasso, may not be suitable for studies of environmental chemicals hypothesized to have nonmonotonic relationships with outcomes. Instead, BKMR, BSTARSS, GAMDP, and GAMTS are attractive methods for flexibly estimating the shapes of exposure-response relationships and selecting among correlated exposures.