-
Clinical Pharmacology and Therapeutics Jun 2021Null hypothesis significance testing (NHST) with its benchmark P value < 0.05 has long been a stalwart of scientific reporting and such statistically significant... (Review)
Review
Null hypothesis significance testing (NHST) with its benchmark P value < 0.05 has long been a stalwart of scientific reporting and such statistically significant findings have been used to imply scientifically or clinically significant findings. Challenges to this approach have arisen over the past 6 decades, but they have largely been unheeded. There is a growing movement for using Bayesian statistical inference to quantify the probability that a scientific finding is credible. There have been differences of opinion between the frequentist (i.e., NHST) and Bayesian schools of inference, and warnings about the use or misuse of P values have come from both schools of thought spanning many decades. Controversies in this arena have been heightened by the American Statistical Association statement on P values and the further denouncement of the term "statistical significance" by others. My experience has been that many scientists, including many statisticians, do not have a sound conceptual grasp of the fundamental differences in these approaches, thereby creating even greater confusion and acrimony. If we let A represent the observed data, and B represent the hypothesis of interest, then the fundamental distinction between these two approaches can be described as the frequentist approach using the conditional probability pr(A | B) (i.e., the P value), and the Bayesian approach using pr(B | A) (the posterior probability). This paper will further explain the fundamental differences in NHST and Bayesian approaches and demonstrate how they can co-exist harmoniously to guide clinical trial design and inference.
Topics: Algorithms; Bayes Theorem; Data Interpretation, Statistical; Humans; Probability; Research Design
PubMed: 32748400
DOI: 10.1002/cpt.2004 -
Physical Therapy Research 2022Clinical research based on epidemiological study designs requires a good understanding of statistical analysis. This paper discusses the common misconceptions of... (Review)
Review
Clinical research based on epidemiological study designs requires a good understanding of statistical analysis. This paper discusses the common misconceptions of p-values so that researchers and readers of research papers will be able to properly present and understand the results of null hypothesis significance testing (NHST). The p-values calculated by NHST are categorized as three different types: "significant at p <0.05," "significant at p <0.01," or "not significant." If specified, they may be written as p = 0.124. The 95% confidence interval (CI) of the supplementary statistics is presented regardless of the p-value, and the range of the CI is observed and discussed to determine whether the results are clinically valid. The effect size (ES), which is a measure of the magnitude of the effect, is also referenced and discussed. However, the ES should not be overestimated. It is important to examine the actual descriptive statistics and consider them comprehensively as much as possible. A high detection power of 80% or more indicates that NHST with high accuracy was applied. However, even when it falls below 80%, it is important to consider the limitations of the study, because the results are not completely useless.
PubMed: 36118788
DOI: 10.1298/ptr.R0019 -
Veterinary and Comparative Oncology Sep 2022Statements such as 'trend towards' and 'tended to' in regards to 'statistical significance' are ambiguous and reflect the continued focus on proximity of p-values to...
Statements such as 'trend towards' and 'tended to' in regards to 'statistical significance' are ambiguous and reflect the continued focus on proximity of p-values to .05. By themselves, p-values do not provide a measure of effect size or other information needed to judge clinical importance. The goal of this study was to examine original research articles in the journal Veterinary and Comparative Oncology over 2 years, and describe the use of 'trend' statements. Articles were reviewed for 'trend' and 'tended to' statements, the number of statements made per paper, whether a p-value and/or 95% confidence interval was provided to accompany these statements, and what the p-value was (if provided). We noted 15.8% of articles included at least one 'trend' or 'tended to' statement to describe differences between groups. Specifically, 10.5% of articles used these statements to describe differences without providing the associated p-value, or where p > .05. Of the total 36 'trend' statements noted, six were not accompanied by p-values and eight were associated with p-values >.10. Similarly, four of the 16 'tended to' statements were not accompanied by p-values, and one was associated with a p-value >.10. These data reveal a similar emphasis on p-value proximity in veterinary oncology literature compared to that of human oncology literature. Furthermore, these findings highlight the ambiguity of 'trend' statements and support the development of additional guidelines (e.g., the de-emphasis of p-value proximity to an arbitrary threshold, and effect size reporting) for interpreting results in veterinary oncology research.
Topics: Animals; Humans; Medical Oncology
PubMed: 35274802
DOI: 10.1111/vco.12811 -
Foods (Basel, Switzerland) Sep 2017Modeling of microbial inactivation by high hydrostatic pressure (HHP) requires a plot of the log microbial count or survival ratio versus time data under a constant... (Review)
Review
Modeling of microbial inactivation by high hydrostatic pressure (HHP) requires a plot of the log microbial count or survival ratio versus time data under a constant pressure and temperature. However, at low pressure and temperature values, very long holding times are needed to obtain measurable inactivation. Since the time has a significant effect on the cost of HHP processing it may be reasonable to fix the time at an appropriate value and quantify the inactivation with respect to pressure. Such a plot is called dose-response curve and it may be more beneficial than the traditional inactivation modeling since short holding times with different pressure values can be selected and used for the modeling of HHP inactivation. For this purpose, 49 dose-response curves (with at least 4 log reduction and ≥5 data points including the atmospheric pressure value ( = 0.1 MPa), and with holding time ≤10 min) for HHP inactivation of microorganisms obtained from published studies were fitted with four different models, namely the Discrete model, Shoulder model, Fermi equation, and Weibull model, and the pressure value needed for 5 log (₅) inactivation was calculated for all the models above. The Shoulder model and Fermi equation produced exactly the same parameter and ₅ values, while the Discrete model produced similar or sometimes the exact same parameter values as the Fermi equation. The Weibull model produced the worst fit (had the lowest adjusted determination coefficient (R²) and highest mean square error (MSE) values), while the Fermi equation had the best fit (the highest R² and lowest MSE values). Parameters of the models and also ₅ values of each model can be useful for the further experimental design of HHP processing and also for the comparison of the pressure resistance of different microorganisms. Further experiments can be done to verify the ₅ values at given conditions. The procedure given in this study can also be extended for enzyme inactivation by HHP.
PubMed: 28880255
DOI: 10.3390/foods6090079 -
Shanghai Archives of Psychiatry Dec 2015The p-value is the most widely used statistical concept in biomedical research. Recently, there are controversies over its utility and over the possible relationship...
The p-value is the most widely used statistical concept in biomedical research. Recently, there are controversies over its utility and over the possible relationship between p-value misuse and the relatively high proportion of published medical research that cannot be replicated. In this paper, we introduce the p-value in layman's terms and explain its randomness and limitations. However, we also point out that the available alternatives to p-value suffer similar limitations. We conclude that using p values is a valid way to test the null and alternative hypotheses in clinical trials. However, using the p-value from a single statistical test to judge the scientific merit of a research project is a misuse of the p-value; the results of inference tests using p-values need to be integrated with secondary results and other data to arrive at clinically valid conclusions. Understanding the variability and limitations of the p-value is important for the interpretation of statistical results in research studies.
PubMed: 27199532
DOI: 10.11919/j.issn.1002-0829.216027 -
Entropy (Basel, Switzerland) Apr 2023Minimum Bayes factors are commonly used to transform two-sided -values to lower bounds on the posterior probability of the null hypothesis, in particular the bound...
Minimum Bayes factors are commonly used to transform two-sided -values to lower bounds on the posterior probability of the null hypothesis, in particular the bound -e·p·log(p). This bound is easy to compute and explain; however, it does not behave as a Bayes factor. For example, it does not change with the sample size. This is a very serious defect, particularly for moderate to large sample sizes, which is precisely the situation in which -values are the most problematic. In this article, we propose adjusting this minimum Bayes factor with the information to approximate an exact Bayes factor, not only when is a -value but also when is a pseudo--value. Additionally, we develop a version of the adjustment for linear models using the recent refinement of the Prior-Based BIC.
PubMed: 37190406
DOI: 10.3390/e25040618 -
Infectious Diseases and Therapy Sep 2020The ability to predict likely prognosis and infectiousness for patients with COVID-19 would aid patient management decisions. Diagnosis is usually via real-time PCR, and... (Review)
Review
BACKGROUND
The ability to predict likely prognosis and infectiousness for patients with COVID-19 would aid patient management decisions. Diagnosis is usually via real-time PCR, and it is unclear whether the semi-quantitative capability of this method, determining viral load through cycle threshold (Ct) values, can be leveraged.
OBJECTIVES
We aim to review available knowledge on correlations between SARS-COV-2 Ct values and patient- or healthcare-related outcomes to determine whether Ct values provide useful clinical information.
SOURCES
A PubMed search was conducted on 1 June 2020 based on a search strategy of (Ct value OR viral load) AND SARS-CoV-2. Data were extracted from studies reporting on the presence or absence of an association between Ct values, or viral loads determined via Ct value, and clinical outcomes.
CONTENT
Data from 18 studies were relevant for inclusion. One study reported on the correlation between Ct values and mortality and one study reported on the correlation between Ct values and progression to severe disease; both reported a significant association (p < 0.001 and p = 0.008, respectively). Fourteen studies reported on the correlation between Ct value or viral loads determined via Ct value and disease severity, and an association was observed in eight (57%) studies. Studies reporting on the correlation of viral load with biochemical and haematological markers showed an association with at least one marker, including increased lactate dehydrogenase (n = 4), decreased lymphocytes (n = 3) and increased high-sensitivity troponin I (n = 2). Two studies reporting on the correlation with infectivity showed that lower Ct values were associated with higher viral culture positivity.
IMPLICATIONS
Data suggest that lower Ct values may be associated with worse outcomes and that Ct values may be useful in predicting the clinical course and prognosis of patients with COVID-19; however, further studies are warranted to confirm clinical value.
PubMed: 32725536
DOI: 10.1007/s40121-020-00324-3 -
Emergency Medicine International 2022To investigate the changes in thromboelastography (TEG) in patients with dyslipidemia to study its effect on the blood coagulation status.
PURPOSE
To investigate the changes in thromboelastography (TEG) in patients with dyslipidemia to study its effect on the blood coagulation status.
METHODS
131 patients hospitalized in Fujian Provincial Jinshan Hospital from January 2018 to December 2020 were selected, and 64 cases in the hyperlipidemia (HL) group and 67 cases in the non-HL group were set according to whether their blood lipids were abnormal. By measuring the changes of each parameter of TEG in patients, the relevant parameters value, value, angle, and MA value were calculated. And routine blood coagulation (PT, APTT, INR, FIB, and TT) and routine blood (platelet count) tests were performed on all study subjects to analyze the changes of each index of the coagulation function and each parameter of TED in both groups and explore the clinical value of TEG on HL diseases.
RESULTS
Compared with the non-HL group, and K values decreased, and angle and MA values increased in the HL group ( < 0.05). PT, APTT, and INR values decreased, and FIB values increased in the HL group compared with the nonhyperlipidemic group ( < 0.05). The TT levels were similar in the non-HL group and the HL group ( > 0.05). Compared with the non-HL group, PLT values decreased, and PDW and MPV values increased in the HL group ( < 0.05). value was positively correlated with APTT, = 0.373, =0.002. value was negatively correlated with PLT, = -0.399, =0.002. angle and MA values were positively correlated with PLT, = 0.319/0.475, =0.010/ < 0.001. The rest of the indexes did not correlate with each parameter of TEG significant correlation.
CONCLUSION
TEG can predict the hypercoagulability and hypocoagulability of blood by the changes of value, value, angle, and MA to evaluate the effect of hyperlipidemia on the coagulation status, which is important for guiding the adjustment of lipid-lowering, antithrombotic, and anticoagulation programs in patients with atherosclerosis combined with hyperlipidemia or postsurgery combined with hyperlipidemia.
PubMed: 35990371
DOI: 10.1155/2022/1927881 -
Clinical Orthopaedics and Related... Oct 2019Clinical research in orthopaedics typically reports the presence of an association after rejecting a null hypothesis of no association using an alpha threshold of 0.05... (Review)
Review
BACKGROUND
Clinical research in orthopaedics typically reports the presence of an association after rejecting a null hypothesis of no association using an alpha threshold of 0.05 at which to evaluate a calculated p value. This arbitrary value is a factor that results in the current difficulties reproducing research findings. A proposal is gaining attention to lower the alpha threshold to 0.005. However, it is currently unknown how alpha thresholds are used in orthopaedics and the distribution of p values reported.
QUESTIONS/PURPOSES
We sought to describe the use of alpha thresholds in two orthopaedic journals by asking (1) How frequently are alpha threshold values reported? (2) How frequently are power calculations reported? (3) How frequently are p values between 0.005 and 0.05 reported for the main hypothesis? (4) Are p values less than 0.005 associated with study characteristics such as design and reporting power calculations?
METHODS
The 100 most recent original clinical research articles from two leading orthopaedic journals at the time of this proposal were reviewed. For studies without a specified primary hypothesis, a main hypothesis was selected that was most consistent with the title and abstract. The p value for the main hypothesis and lowest p value for each study were recorded. Study characteristics including details of alpha thresholds, beta, and p values were recorded. Associations between study characteristics and p values were described. Of the 200 articles (100 from each journal), 23 were randomized controlled trials, 141 were cohort studies or case series (defined as a study in which authors had access to original data collected for the study purpose), 31 were database studies, and five were classified as other.
RESULTS
An alpha threshold was reported in 166 articles (83%) with all but two reporting a value 0.05. Forty-two articles (21%) reported performing a power calculation. The p value for the main hypothesis was less than 0.005 for 88 articles (44%), between 0.05 and 0.005 for 67 (34%), and greater than 0.05 for 29 (15%). The smallest p value was between 0.05 and 0.005 for 39 articles (20%), less than 0.005 for 143 (72%), and either not provided or greater than 0.05 for 18 (9%). Although 50% (65 of 130) cohort and database papers had a main hypothesis p value less than 0.005, only 26% (6 of 23) randomized controlled trials did. Only 36% (15 of 42) articles reporting a power calculation had a p value less than 0.005 compared with 51% (73 of 142) that did not report one.
CONCLUSIONS
Although a lower alpha threshold may theoretically increase the reproducibility of research findings across orthopaedics, this would preferentially select findings from lower-quality studies or increase the burden on higher quality ones. A more-nuanced approach could be to consider alpha thresholds specific to study characteristics. For example, randomized controlled trials with a prespecified primary hypothesis may still be best evaluated at 0.05 while database studies with an abundance of statistical tests may be best evaluated at a threshold even below 0.005.
CLINICAL RELEVANCE
Surgeons and scientists in orthopaedics should understand that the default alpha threshold of 0.05 represents an arbitrary value that could be lowered to help reduce type-I errors; however, it must also be appreciated that such a change could increase type-II errors, increase resource utilization, and preferentially select findings from lower-quality studies.
Topics: Biomedical Research; Mathematical Concepts; Orthopedic Procedures; Research Design
PubMed: 31283730
DOI: 10.1097/CORR.0000000000000843 -
Ceska a Slovenska Oftalmologie :... 2022The aim of the study was to analyse the values of the anteroposterior corneal optical power ratio (AP ratio), to compare the resulting values with those of theoretical...
AIMS
The aim of the study was to analyse the values of the anteroposterior corneal optical power ratio (AP ratio), to compare the resulting values with those of theoretical models of the eye, and to define the effect of using an individual ratio value on the approximation of the total corneal power.
MATERIAL AND METHODS
A total of 406 eyes were included. Each patient underwent an OCT (RTVue XR) examination, according to which the AP ratio of the cornea was determined, as well as the biometric parameters of the eye (Lenstar LS900). The correlation between the biometric parameters of the eye and the individual AP ratio values was evaluated using Pearsons correlation coefficient. In the analysis, the AP ratio results were compared with selected schematic models of the eye. Using Gaussian equations, a theoretical calculation of the total corneal optical power (KG) was performed, by fitting the AP ratio value and comparing it with the actually measured total corneal power (TCP).
RESULTS
The mean value of the individually determined AP ratio was 1.17 ±0.02. The most frequently represented interval (33.74 %) was 1.17 to 1.18 AP ratio values, with the vast majority of eyes (79.56 %) in the range of 1.15 to 1.20. Individual values of total corneal optical power were statistically significantly different (p < 0.05) from the theoretical values of TCP (except in the Liu-Brennan eye model, where p = 0.06). The lowest mean difference of values was found for the Navarro schematic model. The dependence of the measured AP ratio values and biometric parameters reached a moderate negative correlation (r = -0.50 for p < 0.05) with the parameter corneal posterior surface curvature (Rp), as well as a weak negative correlation with limbal diameter WtW (r = -0.26 for p < 0.05) and a weak positive correlation with central corneal thickness CCT (r = 0.17 for p < 0.05).
CONCLUSION
The assumption of a constant value of the AP ratio according to the selected schematic models of the eye is statistically significantly different from the actual measured values and was defined to have only a negative weak correlation with the size of the limbus diameter. Using the resulting average value of the determined AP ratio (1.17 ±0.02), a lower difference between real and calculated total corneal optical power was achieved.
Topics: Biometry; Cornea; Humans; Reproducibility of Results; Tomography, Optical Coherence
PubMed: 36220362
DOI: 10.31348/2022/23