Statistical analysis

1. Continuous data were expressed as mean (standard deviation) or median

(interquartile range, IQR), and categorical variables were presented as
absolute numbers or percentages.
2. Significance of associated risk factors was tested with the chi-squared test or
Fisher's exact test, as appropriate, for categorical variables.
3. Qualitative variables were described as frequency and percentage and
analyzed by Chi-square or Fisher’s exact tests.
4. Normally distributed quantitative variables were described as mean (standard
deviation) and analyzed by analysis of variance (ANOVA).
5. With respect to the comparison of data between sexes and children from
different BMI groups, the independent t-test was used for normally distributed
variables, while the Mann–Whitney U test was used for non-normally
distributed variables.
6. With respect to the comparison of data between sexes and children from
different BMI groups, the one-way analysis of variance (ANOVA) was used
for normally distributed variables, while the Kruskal-Wallis test was used for
non-normally distributed variables.
7. Comparison of aerobic fitness in relation to age was assessed using one-way
analysis of variance (ANOVA) or Kruskal-Wallis test, as appropriate, and
a post-hoc tests were calculated by using Bonferroni multiple comparison test.
8. Correlations of BMI, FMI, and aerobic fitness (AT MET, peak MET,
peak V˙O2, and peak PD) of all subjects were examined using Pearson's
correlation analysis for normally distributed variables and Spearman's
correlation analysis for non-normally distributed variables.
9. We analysed the association between SNPs and risk of gout using binary
logistic regression analysis.
10. OR with 95% confidence intervals (CI) was adjusted for age, gender, body
mass index, hypertension, alcohol use, hyperuricaemia, total cholesterol, log
(triglycerides) and creatinine using a multiple logistic regression model.
11. The performance of the binary logistic regression model was assessed by the
area under the Receiver operating characteristic (ROC) curve (AUC).
12. Multinomial logistic regression (often just called 'multinomial regression') is
used to predict a nominal dependent variable given one or more independent
13. Adjusted OR was calculated after adjustments of age group, cancer type,
stage, ECOG, and current disease condition using a multinomial logistic
regression model.
14. The Kaplan–Meier method was used to estimate the survival curves for each
group, and the log-rank test was used to test for homogeneity among the
survival curves.
15. The hazard ratios(HRs) and 95% confidence intervals(CIs) for benign
prostatic hyperplasia were calculated using the Cox proportional hazards
model.
16. Potential risk factors, such as comorbidities, were incorporated into the Cox
proportional hazards model(Adjusted HR).
17. Interactions between the values of age and gout were tested using the Cox
proportional hazards model with an added interaction term(age by gout) and
covariates.
18. A P-value ≤ 0.05 was considered significant.
19. All statistical analyses were performed using SAS statistical software(version
9.4, SAS Institute, Cary, NC, USA).
20. SPSS for Windows version 19.0 (IBM Corp., Armonk, NY) was used for all
analyses.
21. Each variable with a significant association (P<0.05) and additional variables
that were not significant but had potential clinical importance were introduced
into a forward, stepwise, logistic-regression model.
22. The adjusted OR and 95% CI were estimated by a stepwise logistic regression
method, the significant variables were entered to this model (P<0.05),
otherwise indicated by ‘–’.
23. All variables with a P value of less than 0.05 in the univariate model were
further entered the multivariate analysis.
24. Standard deviation measures the spread of a data distribution. The more spread
out a data distribution is, the greater its standard deviation.
25. Cox proportional hazards regression models were fitted with age as the time
scale and adjusted for known psoriasis risk factors.
26. Collinearity Diagnostics: A collinearity problem occurs when a component
associated with a high condition index contributes strongly (variance
proportion greater than about 0.5) to the variance of two or more variables.
https://www.sfu.ca/sasdoc/sashtml/stat/chap55/sect36.htm
27. A major assumption of the Cox proportional hazards model is that the effect of
a given covariate does not change over time. If this assumption is violated, the
simple Cox model is invalid, and more sophisticated analyses are required.
https://pubmed.ncbi.nlm.nih.gov/7481205/
28. Pre–post changes within groups were estimated via the standardised response
mean, with mean differences between T1 and T0 divided by the standard
deviation of the difference scores.
Paired t test: Standardised Response Mean (SRM)= (𝑀)/ standard deviation,
𝑆𝐷, Eur Respir J. 2018 Jan 25;51(1). PMID: 29371382

29. ANCOVA with T1 values as outcomes and treatment group (intervention

group/control group) as predictor was used to estimate treatment effects.
Adjusted mean differences (AMDs) with 95% confidence intervals and
Cohen's d were calculated to quantify the between-group effects. We
computed two models for each outcome. The unadjusted model only included
corresponding T0 values as covariates. The adjusted model also included sex,
GOLD stage, smoking status (active versus ex-/never-smoker) and baseline
PImax as covariates. The residuals of all models were checked for distortions
of model assumptions (e.g. normality and linearity)
Eur Respir J. 2018 Jan 25;51(1). PMID: 29371382
30. The Mann-Whitney U test (also called the Wilcoxon-Mann-Whitney test) is a
rank-based nonparametric test that can be used to determine if there are
differences between two groups on a continuous or ordinal dependent variable.
https://statistics.laerd.com/premium/spss/mwut/mann-whitney-test-in-spss.php
31. The Kruskal-Wallis H test (sometimes also called the "one-way ANOVA on
ranks") is a rank-based nonparametric test that can be used to determine if
there are statistically significant differences between two or more groups of an
independent variable on a continuous or ordinal dependent variable.
https://statistics.laerd.com/premium/spss/kwht/kruskal-wallis-test-in-spss.php
32. The χ²tests: The distribution of a categorical variable in a sample often needs
to be compared with the distribution of a categorical variable in another
sample.
https://www.bmj.com/about-bmj/resources-readers/publications/statistics-
square-one/8-chi-squared-tests
33. Multinomial logistic regression is used to model nominal outcome variables,
in which the log odds of the outcomes are modeled as a linear combination of
the predictor variables.
https://stats.oarc.ucla.edu/stata/dae/multinomiallogistic-regression/
34. We assessed the proportional hazards assumption in all models by visually
examining plots of the survival function versus survival time, as well as
log(−log(survival)) versus log(-survival time). The assumption was not
violated in any model. https://gut.bmj.com/content/gutjnl/71/7/1332.full.pdf?
with-ds=yes Gut. 2022 Jul;71(7):1332-1339.
35. We explored non-linear relationships between pregnancy weight gain,
including rate of early weight gain and total weight gain, and CRC in offspring
using restricted cubic splines, with three knots at the 10th, 50th and 90th
percentiles. The relationship did not deviate from linearity, and model fit was
similar when pregnancy weight gain was modelled as a continuous measure
versus restricted cubic spline (assessed via Akaike information criterion). Gut.
2022 Jul;71(7):1332-1339.
36. We examined effect measure modification using single-referent models to
estimate the relative excess risk due to interaction (RERI). Gut. 2022
Jul;71(7):1332-1339.
37. The collected material was subjected to statistical analysis, in order to examine
the relationship between the variables, non-parametric Chi-square test and V-
square tests of independence were used. The strength of the relationship
(correlation) was checked using the V-Cramer coefficient. A significance level
of α=0.05 was used for the analyses, and the statistically significant results
were defined by the p-value < 0.05. Statistical analysis of the 179 collected
material was performed using the Statistica program (Software TIBCO
Software 180 Inc. - 2017, version 13). Reproducibility was assessed based on
the magnitude of the percentage of concordant responses to the same questions
in both rounds and by calculating the kappa statistic. In order to assess the
reproducibility of the results obtained with the used questionnaire, the value of
the χ (kappa Cohen’a) parameter was calculated for individual questions in the
questionnaire (results obtained in the baseline and retest). 78.3% of the
question ot-methods had a very good agreement >0.75. Microplastic in food
and water. Current knowledge and aware-2 ness of consumers.
38. Initially, all data was checked for normal or abnormal distribution. For
quantitative data analysis the Mann–Whitney U test and Pearson’s or
Spearman’s correlations was used. Assessment of qualitative results was 168
evaluated with the use of the chi-squared test. Data were expressed as
minimum, maximum, mean ± SD with p < 0.05 being considered statistically
significant.
39. Statistical analysis was carried out with the SPSS 18 programme at a 95%
confidence level, using Levene´s Test to contrast variances, Welch’s Test and
ANOVA to compare means, and Games-Howell´s and Bonferroni´s Test to
determine differences between groups.(
 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9328485/)
40. The statistical analyses were performed using SPSS 20 for Windows (SPSS
Inc., Chicago, IL, USA) software. Descriptive statistics including the mean,
median, mode, standard deviation, variance, range, skewness and
minimum/maximum values were calculated for each of the four groups tested.
Normality tests were done to analyze the distribution of data collected for each
group using the Kolmogorov-Smirnov and Shapiro-Wilk tests. Homogeneity
of variance was tested with the Levene test.( https://www.ijdr.in/article.asp?
issn=0970-
9290;year=2021;volume=32;issue=1;spage=98;epage=103;aulast=Pulgaonkar)
41. The significance of mean difference between the groups was done by Tukey's
post hoc test after ascertaining normality by Shapiro-Wilk (W) test and
homogeneity of variance by Levene's test. A two-tailed P value less than 0.05
(P < 0.05) was considered statistically significant. All analyses were
performed on SPSS software (Windows version 17.0).
https://www.ijdr.in/article.asp?issn=0970-
9290;year=2019;volume=30;issue=4;spage=590;epage=594;aulast=Gangwar
42. Initially, all data was checked for normal or abnormal distribution.
43. Did the authors perform the Levene’s test before the Student’s t test. Levene’s
test was used to assess the equality of variances, and an independent sample t-
test (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9537748/)
44. Did the authors consider either a two-tailed or a one-tailed P-value. A two-
tailed P value less than 0.05 (P < 0.05) was considered statistically significant.
45. The authors should specify how was the normality distribution of the variables
assessed. The normal distribution of variables was evaluated with the Shapiro-
Wilk test. Normality tests were done to analyze the distribution of data
collected for each group using the Kolmogorov-Smirnov and Shapiro-Wilk
tests.
46. The Shapiro–Wilk test is more appropriate method for small sample sizes (<50
samples) although it can also be handling on larger sample size while
Kolmogorov–Smirnov test is used for n ≥50.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6350423/
47. Comparison of aerobic fitness in relation to age was assessed using oneway
analysis of variance (ANOVA), and a post-hoc test was performed by Scheffe
method if there was homogeneity of variance or the Dunnett T3 test if there
was heterogeneity of variance.
48. The beta coefficient is the degree of change in the outcome variable for every
1-unit of change in the predictor variable.
49. All data were analyzed using SPSS Statistics 23.0 software.
50. The measurement data are expressed as median (25% quantile, 75% quantile)
(M(P25,P75)), and the results of qualitative variables are expressed as
frequency and corresponding percentage.
51. Spearman's correlation analysis was performed to assess the correlation
between the variables. The groups were compared by Mann-Whitney test.
52. Binary logistic regression was used to determine the independent predictors,
the regression uses a forward stepwise selection with a stopping rule based on
P<0.05.
53. A P-value of <0.05 means that the difference is statistically significant.
54. The probability of new-onset (incident) type 2 diabetes during follow-up as a
 function of baseline lipoprotein(a) as a spline effect of degree 3 (piecewise
cubic curve) was estimated for each treatment group by logistic regression
with a logit link function and the logarithm of follow-up time as an offset
variable in the models.
55. In a sensitivity analysis, the outcome of death from neurodegenerative disease
was subjected to a competing risks regression analysis to ascertain whether the
estimated hazard ratio was sensitive to the competing risks of death from
ischemic heart disease and death from any cancer. N Engl J Med
2019;381:1801-8
56. The probability of new-onset (incident) type 2 diabetes during follow-up as a
function of baseline lipoprotein(a) as a spline effect of degree 3 (piecewise
cubic curve) was estimated for each treatment group by logistic regression
with a logit link function and the logarithm of follow-up time as an offset
variable in the models. Diabetes Care. 2021 May;44(5):1219-1227.
57. Because prescribing outcomes were available only from 2009 onward, we
applied a nested case–control design in which conditional logistic regression
was used to analyze the whole cohort (matched data)30 and standard logistic
regression was used to analyze the subgroups of former players defined
according to player position. N Engl J Med 2019;381:1801-8
58. We tested the assumption of proportional subhazards (hazard ratios adjusted
for competing risks of death from ischemic heart disease and death from any
cancer) by including an interaction term between time of analysis and a
dummy variable for former soccer player status and assessing whether the
interaction was significant (P<0.05). N Engl J Med 2019;381:1801-8
59. we also performed a competing-risk analysis in which death due to a cause
other than a cardiovascular cause was considered as a competing risk, and
hazard ratios with 95% confidence intervals were calculated with the use of
the method of Fine and Gray. January 9, 2020N Engl J Med 2020; 382:111-
119
60. Participant characteristics are reported as median and IQR or percentages.
Correlation analyses were performed using Spearman rank correlation. For
prospective analyses, we z-standardized all glycans (mean 0, SD 1) to allow
for comparison of the effect estimates. Diabetes Care. 2022 Nov
1;45(11):2729-2736.
61. We selected IgG-GPs significantly associated with each end point in mutually
adjusted models with multiple fractional polynomials (MFPs), applying false
discovery rate (FDR) correction. MFP-selected glycans were subsequently
tested in confounder-adjusted Prentice-weighted Cox proportional hazards
models. Diabetes Care. 2022 Nov 1;45(11):2729-2736.
62. In the interaction analyses, a two-sided P < 0.05 denoted statistical
significance. Diabetes Care. 2022 Nov 1;45(11):2729-2736.
63. The median follow-up time was 6.5 (interquartile range [IQR] 6.0–8.6) years
for diabetes and 8.3 (IQR 7.5–9.2) years for CVD. Diabetes Care. 2022 Nov
1;45(11):2729-2736.
64. For prospective analyses, we z-standardized all glycans (mean 0, SD 1) to
allow for comparison of the effect estimates. (proc stdize?? Per SD??)
Diabetes Care. 2022 Nov 1;45(11):2729-2736.
65. Means and SDs for continuous variables and numbers with proportions for
baseline characteristics are presented for the subcohort and for each event
 outside the subcohort. A descriptive analysis was done using an ordinary least-
squares regression with plasma ACE2 concentration as the outcome in a cross-
sectional analysis at baseline Lancet. 2020 Oct 3;396(10256):968-976.
66. The association measure was presented as a hazard ratio (HR) per 1 SD unit
increase in the marker, adjusted for the following: age, sex, smoking, BMI,
systolic blood pressure, non- HDL cholesterol, and geographic ancestry.
Lancet. 2020 Oct 3;396(10256):968-976
67. In addition, we used a regression model with restricted cubic splines (three
knots) to assess the shape of the association of the composite biomarker score
with fruit and vegetable intake in the subcohort, adjusted for the same
covariates as above.
68. The difference among groups were calculated by One-way Anova, followed
by Bonferoni multiple comparision test, using IBM SPSS Statistics software
(IBM, Armonk, NY, USA). P<0.05 value was considered as statistically
significant.
69. The Kolmogorov Smirnov test and histogram were used to evaluate data
normality. Descriptive analysis was conducted according to data type,
including constituent ratios, mean ± standard deviation, and median
(interquartile range).
70. Baseline information for both groups was compared using Chi-square (χ2) test
(for categorical variables), Student’s t-test (for normal continuous variables),
and Mann-Whitney U test (for non-normal continuous variables).
71. A goodness-of-fit χ2-test was used to determine Hardy-Weinberg equilibrium.
72. Logistic regression analysis based on four genetic models (co-dominant
model, dominant model, recessive model and additive model) was performed
to construct odds ratio (OR) and 95% confidence interval (CI) to investigate
the correlations of SNPs with susceptibility to NAFLD, with age, sex,
triglyceride, and glucose adjusted for potential confounding effects.
73. To account for the impact of multiple comparisons, the false discovery rate
(FDR) is adjusted.
74. Cochran-Armitage trend was tested to analyze the combined effects of
statistically significant SNPs.
75. We further performed subgroup analysis to explore the influence of
confounding factors, and heterogeneity between subgroups was assessed by Q
test.
76. To identify NAFLD predictors and proceed to construct a predictive model,
multivariate stepwise logistic regression and receiver-operating characteristic
curve (ROC) were used.
77. The discrimination and calibration of the prediction model were assessed by
the area under the receiver operating characteristic curve (AUROC) and the
Hosmer-Lemeshow test, respectively.
78. In all analyses, a P-value<0.05 in the two-tailed test was considered
statistically significant.
79. f. Cox & Snell R Square and Nagelkerke R Square – These are pseudo R-
squares. Logistic regression does not have an equivalent to the R-squared that
is found in OLS regression; however, many people have tried to come up with
one. There are a wide variety of pseudo-R-square statistics (these are only two
of them). Because this statistic does not mean what R-squared means in OLS
regression (the proportion of variance explained by the predictors), we suggest
 interpreting this statistic with great caution.
https://stats.oarc.ucla.edu/spss/output/logistic-regression/
80. All values are depicted as mean ± standard error of the 473 mean (SEM). One-
way ANOVAs followed by a Dunnet post hoc test, or paired and 474 unpaired
t-tests (two-tailed) were used to calculate significances between groups.
81. Crude and adjusted odds ratio with their 95% CI were calculated to determine
the strength and presence of association
82. The Hosmer-Lemeshow test analysis was used to check the appropriateness of
the model.
83. Both bivariable and multivariable logistic regression analyses were done to
identify associations between dependent and independent variables.
84. Descriptive statistics like frequency distributions, mean, median and standard
deviation were computed.
85. The dataset was scaled and outliers that had SD > 4 from the mean metabolite
and particle concentrations were removed.
86. The multiple-testing threshold was set according to p < 0.05 / 19 PCs; i.e. p <
0.0029.
87. Because of the multiple testing issue, what is the evidence for the significant
threshold definition. " We used a Bonferroni correction to adjust for multiple
testing of three traits with each other (p<0.05/9=0.0056).
88. In experimental research a scientific conclusion is always drawn from the
statistical testing of hypothesis, in which an acceptable cutoff of
probability(significance level α), such as 0.05 or 0.01, is used for decision-
making. However, the probability of committing false statistical inferences
would considerably increase when more than one hypothesis is simultaneously
tested (namely the multiple comparisons), which therefore requires proper
adjustment. For example, it is common in clinical trials to simultaneously
compare the therapeutic effects of more than one dose levels of a new drug in
comparison with standard treatment. Therefore, it is obvious that the proper
adjustment of statistical inference is required for multiple comparisons.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5506159/
89. Using the direct standardization method, age-standardized mortality rates
(ASMR) were calculated by multiplying the age-specific mortality rates by the
number of persons in each age group of the standard population.
90. The age structure was divided into every ten years from 25 to 85+ based on
the 2000 U.S. Census Standard Population.
91. Plasma insulin, HOMA-IR, AST, ALT, GGT and triglycerides levels were log-
transformed to normality prior being used in statistical models.
92. The Youden index (J) is a measure for evaluating the biomarker effectiveness.
This measure was first introduced to the medical literature by Youden [5]. J is
a function of Se(c) and Sp(c), such that J( c ) = { Se( c ) + Sp( c ) − 1 }
= { Se( c ) − ( 1 − Sp( c ) ) }
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5470053/
93. We used Cox proportional-hazards regression to model time to death and
Schoenfeld residuals to test the assumption of proportional hazards. N Engl J
Med 2019;381:1801-8.
94. All the tests of effects were conducted at a two-sided alpha level of 0.05. The
primary outcome, the change from baseline to 80 weeks in the ADAS-cog14
score, was analyzed with the use of a mixed-model repeated-measures
 analysis, with the change from baseline in the ADAS-cog14 score at each
scheduled visit at weeks 12, 28, 40, 52, 64, and 80 after baseline as the
dependent variable. Panel A shows the results for the primary outcome, the
least-squares mean change from baseline (dashed line) in the score on the 14-
item cognitive subscale of the Alzheimer’s Disease Assessment. The estimated
difference is the least-squares mean (±SE) change from baseline between the
two trial groups at 80 weeks. N Engl J Med 2018;378:321-30
95. Categorical variables are shown as frequencies and percentages, and
continuous variables as means and standard deviations (SD) or median and
interquartile range(IQR).
96. Linear regression analysis is used to predict the value of No radiographic
lucency of fracture site based on the value of Standing (3-mon).
97. Spearman's rank correlation measures the strength and direction of association
between No radiographic lucency of fracture site and Standing (3-mon).
98. The primary outcome, the change from baseline to Post-op day 1 in the Pain
(VAS) or Majeed score, was analyzed with the use of a generalized linear
model repeated-measures analysis, with the change from baseline in the Pain
(VAS) or Majeed score at each scheduled visit at Day 7, 1-month, and 3-
months. The estimated difference is the least-squares mean (±standard error,
SE) change from baseline.
99. All the tests of effects were conducted at a two-sided alpha level of 0.05
100. To explore whether there existed potential effect modifications, we
conducted four pre-defined subgroup analyses by sex (males and females), age
(< 65 and ≥ 65 years), diabetes mellitus (yes or no), and MH status (yes or no).
Cardiovasc Diabetol. 2022 Jul 15;21(1):133.
101. Baseline variables were reported as means (standard deviations [SDs]) for
continuous variables and counts (percentages) for categorical variables,
respectively. Cardiovasc Diabetol. 2022 Jul 15;21(1):133.
102. Restricted cubic splines, as commonly used to model non-linear
associations in regression models, were transformation of an independent
continuous variable and could be used in various regression models. The range
of values of the independent variable was first split up, with “knots” defining
the end of one segment and the start of the next. Subsequently, separate curves
were fitted to each segment so that the resulting overall fitted curve was
smooth and continuous [18]. We used the restricted cubic spline based on
multivariable Cox proportional hazards model to assess the potential non-
linear effect of HbA1c on CVD risk, where the HbA1c level was treated by
using a restricted cubic spline with four knots laid at the 5th, 35th, 65th, and
95th percentiles.
103. The Bonferroni method is used in the adjustment of multiple comparisons.
5-point Likert scale was used to measure the respondents’ perception of
benefits. J Hum Genet. 2022 Dec 12. doi: 10.1038/s10038-022-01100-6.
104. The proportional hazard assumption is supported by a non-significant
relationship between residuals and time, and refuted by a significant
relationship. PH assumption is supported by a non-significant relationship
between residuals and time, and refuted by a significant relationship. Regress
Schoenfeld residuals against time to test for independence between residuals
and time(Statistics 262: Intermediate Biostatistics)
105. Continuous variables were expressed as median and Inter-Quartile Range
 (IQR) and com-136 pared using the Mann whitney-U test. Categorical
variables were expressed in number and percentage and compared using chi-
square tests. Bilateral P values less than 0.05 were considered statistically
significant.
106. ‡Denotes % difference between predicted and observed values, which was
calculated as100%×(observed mortality – predicted mortality)/predicted
mortality
107. A two-way mixed model ANOVA was used to test the effects of the within-
subject factor (time: baseline and after 24 months) and the between-subject
factors (groups: using insulin or not; private or public health service) and the
interaction between these factors in the observed variance of the body weight
and clinical analysis parameters: glycemia, glycated hemoglobin (HbA1c),
HDL-c, LDL-c, and TAGs, among both adults and elderly individuals
(separately).
108. The results of Box’s test, Mauchly’s sphericity test and Levene’s test for
equality of variances were evaluated as assumptions; when necessary, the
Huynh-Feldt correction was applied.
109. Outliers were considered when the studentized residual was greater than ±3
SDs.
110. The Mann Kendall Trend Test (sometimes called the M-K test) is used to
analyze data collected over time for consistently increasing or decreasing
trends (monotonic) in Y values. It is a non-parametric test, which means it
works for all distributions (i.e. your data doesn’t have to meet the assumption
of normality), but your data should have no serial correlation. If your data
does follow a normal distribution, you can run simple linear
regression instead. https://www.statisticshowto.com/mann-kendall-trend-test/
111. Statistical power: the likelihood that a test will detect an effect of a certain
size if there is one, usually set at 80% or higher.
https://www.scribbr.com/statistics/statistical-power/
112. Variables with a probability result of P < 0.05 in the univariate analyses
were introduced into the multivariate logistic regression model. Those
variables were assigned values for conducting the multivariate logistic
regression analysis, and P =0.05 was set as the test value.
113. P values were calculated with the use of a Cox proportional-hazards
regression model.
114. The variance inflation factor (VIF) was used to detect multicollinearity in
the regression model. Explanatory variables with multicollinearity of ≥5
indicates a multicollinearity problem.
115. What is multiple linear regression? Multiple linear regression is a regression
model that estimates the relationship between a quantitative dependent
variable and two or more independent variables using a straight line.
116. The coefficient of determination or R squared method is the proportion of
the variance in the dependent variable that is predicted from the independent
variable. It indicates the level of variation in the given data set. The coefficient
of determination is the square of the correlation(r), thus it ranges from 0 to 1.
117. Stepwise regression is the step-by-step iterative construction of a regression
model that involves the selection of independent variables to be used in a final
model. It involves adding or removing potential explanatory variables in
succession and testing for statistical significance after each iteration.
118. The Shapiro–Wilk test is more appropriate method for small sample sizes
(<50 samples) although it can also be handling on larger sample size while
Kolmogorov–Smirnov test is used for n ≥50. For both of the above tests, null
hypothesis states that data are taken from normal distributed population. Ann
Card Anaesth. 2019 Jan-Mar; 22(1): 67–72.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6350423/
119. Too many variables were included in multivariable analyses, resulting in
overfitting of models. The authors may want to restrict the number of
variables according to the general rule of one variable to 10 outcome events.
120. Generalized linear model repeated measure is a statistical technique that
takes a dependent, or criterion variable, measured as correlated, non-
independent data. Commonly used when measuring the effect of a treatment at
different time points.
121. The Log-rank test is a statistical method for comparing the distribution
of time until the occurrence of an event of interest in independent groups.
122. Survival curves are estimated for each group, considered separately, using
the Kaplan-Meier method and compared statistically using the log rank test.
https://sphweb.bumc.bu.edu/otlt/mph-modules/bs/bs704_survival/BS704_Surv
ival5.html
123. The Cox proportional-hazards regression model for time-to-event data may
be used with covariates, independent variables, or predictor variables that vary
over time. These are called time-dependent covariates. Their use is much more
complicated in practice than the fixed (time-independent) covariates.
https://www.annualreviews.org/doi/10.1146/annurev.publhealth.20.1.145
124. Here is the strategy we use at Statistics Solutions: select the correct test, and
then select the correct sample size for that test.
125. It is in the author’s interest to provide the reader with all the necessary
information to judge whether the statistical tools used in the paper are the most
suitable to answer the scientific question and are suited to the data structure.
https://link.springer.com/article/10.1007/s43440-020-00110-5
126. The normal distribution of variables was ascertained via the Shapiro–Wilk
test. Continuous variables are expressed as the mean ± standard deviation (SD)
or absolute numbers, percentage. Regression analysis was conducted to test
the association between protein intake and cardiometabolic markers.
127. Stratification is defined as the act of sorting data, people, and objects into
distinct groups or layers. It is a technique used in combination with other data
analysis tools. When data from a variety of sources or categories have been
lumped together, the meaning of the data can be difficult to see.
https://asq.org/quality-resources/stratification
128. We used cox proportional hazards models to investigate the relations
between sleep factors and the risk of new-onset gout.
129. Possible effect modifications of the association between healthy sleep
scores and the risk of new-onset gout were assessed by stratified analyses,
according to gout GRS, age, sex, BMI, smoking status, drinking status,
diuretics medications use, eGFR and uric acid levels. Interactions between
healthy sleep score and subgroups were evaluated by likelihood ratio testing.
130. Continuous variables are presented as mean ± SD, category variables are
presented as n (%).
131. A Bonferronicorrected p value of <0.004 (0.05/13) was used to determine
 significant associations in the replication analysis. Ann Rheum Dis . 2021
Sep;80(9):1220-1226.
132. Participants were divided into low, intermediate, or high genetic risk for
gout according to the tertiles of gout GRS
133. The multivariable model was adjusted for all the risk factor variables
included in the full model, the matching variables (age, sex, ethnicity,
deprivation (NZDep13), and region), and an additional deprivation measure
(the caregiver’s individual deprivation score, NZiDep). These risk factor
variables were identified a priori in the study protocol6 and refined by
stepwise regression, with collinear or non-significant variables removed. The
Lancet Regional Health - Western Pacific 2022;26: 100508
134. A within-subjects factor is sometimes referred to as a repeated-measures
factor since repeated measurements are taken on each
subject. https://onlinestatbook.com/2/analysis_of_variance/within-
subjects.html
135. Between-subject variables are independent variables or factors in which a
different group of subjects is used for each level of the variable.
https://onlinestatbook.com/glossary/between-subjects.html
136. The correlation coefficient is measured on a scale that varies from + 1
through 0 to – 1. Complete correlation between two variables is expressed by
either + 1 or -1. When one variable increases as the other increases the
correlation is positive; when one decreases as the other increases it is negative.
https://www.bmj.com/about-bmj/resources-readers/publications/statistics-
square-one/11-correlation-and-regression
137. The FDR is the ratio of the number of false positive results to the number of
total positive test results: a p-value of 0.05 implies that 5% of all tests will
result in false positives. An FDR-adjusted p-value (also called q-value) of 0.05
indicates that 5% of significant tests will result in false positives.
138. P values of comparing categorical variables between groups are from two-
sided χ2 test; P values of comparing continuous variable between two groups
are from two-sided Wilcoxon’s test. https://www.nature.com/articles/s41467-
021-21458-z
139. Characteristics of the study participants were presented as means (standard
error) for continuous variables or percent (standard error) for categorical
variables. Generalized linear models were used to compare differences in
continuous variables, andχ2 tests were used for categorical variables.
Multivariable Cox proportional hazards regression was used to estimate
hazard ratios and 95% CIs of mortality in relation to frequency of eating meals
prepared away from home. Stratified analyses were performed according to
age (20-44 years vs 45 years or older), sex (male vs female), race (White vs
non-White), ratio of family income to poverty (<2 vs ≥2), and BMI (<30 vs
≥30). Interaction tests were conducted by including multiplicative interaction
terms in the multivariable models. J Acad Nutr Diet. 2021 Sep;121(9):1741-
1749.e1
140. Missing values can be imputed with a provided constant value, or using the
statistics (mean, median or most frequent) of each column in which the
missing values are located. This class also allows for different missing values
encodings. https://scikit-learn.org/stable/modules/impute.html
141. Historically, a common approach to imputation was “mean-value
 imputation,” in which subjects for whom a given variable is missing have the
missing value replaced with the mean value of that variable among all subjects
for whom the variable is present. A specific implementation of this strategy in
which every variable is imputed conditional on all other variables is now
known as the multivariate imputation. The imputation process described above
uses linear regression and takes the imputed values as random draws from a
normal distribution.
https://www.sciencedirect.com/science/article/pii/S0828282X20311119
142. Plasma n-3 was analyzed as a 84 continuous variable after log-
normalization and standardization.
143. The treatment interventions were compared with the use of a repeated-
measures linear mixed model with adjustment for period as a fixed effect and
site as a random effect and with accounting for the baseline value as a separate
period. A 95% confidence interval was reported for the difference between
interventions on the basis of the linear mixed model. For highly skewed data, a
ranked normal score transformation was used. Missing data were not imputed
for the primary analysis. https://www.nejm.org/doi/10.1056/NEJMoa2111673
144. Relative rates are described with hazard ratios and 95% confidence
intervals. The 95% confidence intervals have not been adjusted for
multiplicity, and therefore inferences drawn from these intervals should not be
used for hypothesis testing. For the principal analyses, missing data were not
replaced, and complete case data are presented. In a sensitivity analysis,
multiple imputation was used to account for missing follow-up data. All the
statistical analyses were performed with the use of SAS software, version 9.4
(SAS Institute). https://www.nejm.org/doi/full/10.1056/NEJMoa2300213?
query=featured_home
145. A hierarchical approach was prespecified to evaluate sequentially the
primary end point and each of the six key secondary efficacy end points;
statistical significance at each step was required to test the next hypothesis,
thereby preserving the study-wise type I error rate at 5%. All efficacy analyses
were based on the intention-to-treat principle. All efficacy end points were
analyzed with the use of a Cox proportional-hazards model that included the
trial-group assignment as a factor to generate the hazard ratio and 95%
confidence interval. P values were obtained with the use of a two-sided log-
rank test. https://www.nejm.org/doi/full/10.1056/NEJMoa2215024?
query=featured_home ??
146. All the analyses were performed on an intention-to-treat basis. The primary
end-point analysis included an estimation of the cumulative-incidence function
of target-vessel failure and a comparison of the randomized groups with the
use of the method of Fine and Gray to adjust for the potential competing risk
of death from noncardiac causes.13 As a sensitivity analysis, Kaplan–Meier
analyses were performed for time-to-event end points with treatment effects
estimated by means of Cox proportional-hazards regression models, and
results are presented as hazard ratios with 95% confidence intervals. The
proportional-hazards assumption was evaluated with a two-sided score test of
the scaled Schoenfeld residuals over time at the 0.05 level. Subgroup analysis
of the primary end point was performed according to the type of intravascular
imaging device and clinical factors. No imputation methods were used to infer
missing values of baseline variables. All the models were adjusted for the
 patient’s clinical presentation and for participating centers.
https://www.nejm.org/doi/full/10.1056/NEJMoa2216607?
query=featured_home
147. The general linear model (GLM) refers to conventional linear regression
models for a continuous response variable given continuous and/or categorical
predictors. It includes multiple linear regression, as well as analysis of
variance (ANOVA) and analysis of covariance(ANCOVA, with fixed effects
only). https://online.stat.psu.edu/stat504/lesson/6/6.1
148. The primary analysis of vaccine efficacy was performed in the according-
to-protocol efficacy cohort with the use of Cox proportional-hazards
regression models (vaccine efficacy=1−hazard ratio), with 90% confidence
intervals and P values for Wald tests. N Engl J Med 2019; 381:2429-2439
149. We used Cox proportional-hazards regression after adjustment for trial
center, patient’s age, Gleason score, and baseline PSA (log-transformed) to
compare prostate cancer–specific mortality at 15 years in the three groups on
an intention-to-treat basis. Interaction terms were added to this model to
investigate differential treatment effects across the eight prespecified
subgroups. March 11, 2023 DOI: 10.1056/NEJMoa2214122.
150. Several potential confounders were included in the adjusted logistic
regression model: age (expressed in years), sex, COVID-19 history, district
and epidemiological week of the SOD. The same covariables taken into
account for the logistic regression model were considered as predictors. The
AIC was used to define the functional form of the age variable included in the
final model. The assumption of proportional risks was evaluated by estimating
the correlation of Schoenfeld residuals of the vaccination status variable with
the linear, quadratic and cubic forms of the days until death. The results of
these models are expressed as Hazard Ratios, taking individuals who did not
receive any vaccine as the reference category. The survival function was
estimated fitting Kapplan-Maier curves for each immunization status for each
vaccine analysis. https://ars.els-cdn.com/content/image/1-s2.0-
S0140673622000113-mmc1.pdf
151. For the prespecified composite secondary outcome of invasive mechanical
ventilation or death within 28 days (among patients who were not receiving
invasive mechanical ventilation at randomization), the precise date of invasive
mechanical ventilation was not available, so a log-binomial regression model
was used to estimate the risk ratio. February 25, 2021N Engl J Med 2021;
384:693-704
152. An interaction plot was used to plot the estimated effect of physical activity
on MetS and their corresponding 95% CI as a function of individual pollutant
concentrations using Interplot method. All models adjusted for age, gender,
education level, marital status, average monthly income, smoking and drinking
status, high-fat diet, fruit, and vegetable intake as well as family histories of
T2DM and hypertension. Environ Int. 2020 Mar;136:105459.
153. The categorical variables, normal and non-normal distribution of continuous
variables were expressed as the number (percentage), mean (standard
deviation, SD) and median (interquartile range), respectively. Environ Int.
2020 Mar;136:105459.
154. We assessed the heterogeneity of the treatment effect on the primary
outcome in eight prespecified subgroups by adding an interaction term in an
 unadjusted logistic-regression model. June 20, 2013 N Engl J Med 2013;
368:2355-2365
155. The relative change in hematoma volume was log-transformed to remove
skewness after the addition of the value 1.1 to eliminate negative values. The
nominal level of significance for all analyses was P<0.048, since two interim
analyses were performed in which the Haybittle–Peto efficacy stopping rule
was used. June 20, 2013 N Engl J Med 2013; 368:2355-2365
156. We confirmed the veracity of the proportional-hazards assumption in the
model both by inspecting plots of the Martingale residuals and by testing for
an interaction between a constructed time-dependent covariable and the study
group. We confirmed the robustness of the treatment effect in a model that
included 22 additional preoperative covariables, selected from 48 potential
covariables in a manner similar to that used for building the propensity-score
model. February 21, 2008 N Engl J Med 2008; 358:784-793
157. We assessed study-group differences in growth-velocity trends for
participants between the ages of 5 and 15 years separately for each sex by
means of a repeated-measures multiple linear regression model,24 including
an indicator variable for treatment, splines for each year of age, and interaction
terms for treatment according to age. September 6, 2012 N Engl J Med 2012;
367:904-912
158. Missing covariate values (which occurred in <1% of the participants) were
handled with the use of missing indicators. January 20, 2022 N Engl J Med
2022; 386:252-263
159. The proportional-hazard assumptions were tested and verified by the
inclusion of an interaction term with time in the model. January 20, 2022 N
Engl J Med 2022; 386:252-263
160. Our primary analyses were based on a priori hypotheses, but in order to
account for the hypothesis tests for three exposures, we present P values that
were corrected for multiple comparisons with the use of the Holm–Bonferroni
procedure. January 20, 2022 N Engl J Med 2022; 386:252-263
161. The spline analysis with pooled data further supported a linear association
over the range of sodium excretion within this population (P<0.001 for
linearity) January 20, 2022 N Engl J Med 2022; 386:252-263
162. The Cox proportional hazards model is the most popular model for the
analysis of survival data. The use of cubic spline functions allows
investigation of non-linear effects of continuous covariates and flexible
assessment of time-by-covariate interactions.
163. Proportional-hazards models were used to compare the risk of death among
patients with HPV-positive cancer and those with HPV-negative cancer. July 1,
2010 N Engl J Med 2010; 363:24-35
164. The continuous variables were presented as the mean (standard deviation
[SD]) or the median (interquartile range), and the categorical variables were
presented as n (%). The Kolmogorov-Smirnov statistical test was used to
determine if continuous variables were normally distributed. To normalize
their distribution, the blood metals were log-transformed. Spearman’s
correlation was utilized to determine the exposure correlation coefficients for
all metals. To investigate the dose-response curves of exposure variables and
the prevalence of hyperuricemia and gout, a restricted cubic spline regression
model with knots at the 5th, 35th, 65th, and 95th percentile of each exposure
 variable was utilized. Front Endocrinol (Lausanne). 2022 Dec 2;13:1052784.
165. trend test was performed as the SUA levels increased and logistic regression
was performed to calculate the ORs with per SD increase in UA. We
performed restricted cubic spline analysis with 3 knots of the SUA levels to
characterize the dose–response relationship between SUA and diabetic
complications in logistic regression Model 2. A two-sided p < 0.05 was
considered statistically significant. J Clin Med. 2023 Jan 16;12(2):725.
166. Residual analysis confirmed that the assumption of normal error was
satisfied. The net intervention effect (the mean change in the experimental
group minus the mean change in the control group) was calculated from the
parameters of the fitted model.
https://www.nejm.org/doi/full/10.1056/nejmoa1203388October 11, 2012 N
Engl J Med 2012; 367:1407-1416
167. The logistic regression assumptions are similar to the linear regression
assumptions. However, linearity and additivity are checked with respect to the
logit of the outcome variable. In addition, homoscedasticity and normality of
residuals are not assumptions of binary logistic regression. https://carpentries-
incubator.github.io/logistic-regression-public-health/05-assumpLogistic/
index.html
168. The statistical power of a hypothesis test is the probability of detecting an
effect, if there is a true effect present to detect.
https://machinelearningmastery.com/statistical-power-and-power-analysis-in-
python/
169. Categorical data are presented as percentages, while continuous data are
presented as means with their corresponding standard deviations. We used the
standardized difference to compare the distribution of general attributes
between the cohorts
170. In the overlap-weighting scheme, CKD participants were weighted
according to the propensity score's probability, whereas control participants
were weighted according to200 the probability of 1 minus the propensity
score. We used this overlap weighting, ranging between 0 and 1, to achieve an
ideal balance and enhance precision in the analyses[34-36].
171. This study utilized standardized differences to compare the general
characteristics between the CKD and control groups, both before and after
weighting. Additionally, the effectiveness of the matching process was
assessed by comparing the absolute standardized differences of the covariates
before and after matching. We regarded an absolute standardized difference of
less than 0.20 as an indication of satisfactory balance [37].
172. Furthermore, the crude incidence rates and the differences in incidence rates
were computed by dividing the number of participants who experienced a
particular event by the total person-years of observation, and this was
expressed as instances per 1,000 person-years. We used the Kaplan-Meier
method and the log-rank test to compare the cumulative incidence of PD in the
CKD group with that in the control group. 202306050612nutrients-2448331
173. When the assumption of homogeneity of variance was violated according to
Levene’s test, Welch’s corrected F was used.
174. Continuous data were expressed as mean (standard deviation, SD) or
median (interquartile range, IQR), and the categorical variables were
presented as n (%, percentages may not sum to 100 because of rounding).
175. Most statistical analyses and software packages assume that all variables in
the model are measured. The default procedure normally deletes cases with
missing data on the variables of interest, which known as listwise
deletion(Definition: Listwise deletion (also known as casewise deletion or
complete case analysis) removes all observations from your data, which have a
missing value in one or more variables https://statisticsglobe.com/listwise-
deletion-missing-data/). The major disadvantage of this method is that it
regularly removes a large proportion of the sample, leading to a severe loss of
statistical power.
http://www.jtle.net/uploadfile/2016/1108/20161108041850649.pdf
Listwise or case deletion
By far the most common approach to the missing data is to simply omit those
cases with the missing data and analyze the remaining data. This approach is
known as the complete case (or available case) analysis or listwise deletion.
Korean J Anesthesiol. 2013 May;64(5):402-6.
176. The chi-squared test and Fisher's exact test can assess for independence
between two variables when the comparing groups are independent and not
correlated. The chi-squared test applies an approximation assuming the sample
is large, while the Fisher's exact test runs an exact procedure especially for
small-sized samples. If the chi-squared test concludes that there is significant
association, we may want to know if there is any significant difference in three
compared pairs, between control and experiment 1, between control and
experiment 2, and between experiment 1 and experiment 2. We can reduce the
table into multiple 2 × 2 contingency tables and perform the chi-squared test
with applying the Bonferroni corrected alpha level (corrected α = 0.05/3
compared pairs = 0.017). Restor Dent Endod. 2017 May; 42(2): 152–155.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5426219/
177. Data with a weight of 0 or missing data (N=7,885) were also excluded. In
addition, participants with a daily vitamin intake above average (+3 SD) or 80
below average (-3 SD) (N=3,048) were also excluded. In the analysis, the
vitamin 126 intake was divided into quartiles, and the previous 25% was used
as reference values for 127 comparative analysis. To examine the relationship
between trends in individual vitamin 128 intake and the risk of kidney stones,
we performed dose‒response analysis using a re-129 stricted cubic spline
function curve. Finally, according to the results of clustering, multivariate
logistic regression analysis 132 was carried out again.
178. The Shapiro–Wilk test was used to assess the normality of distribution of
the continuous variables. In the presence of symmetry of the distributions, the
variables were represented with mean and Standard Deviation (SD)錯誤 or, in
the case of non-normal distribution, with the median value and interquartile
range [1Q 3Q]; categorical data were expressed as total numbers and
percentages (%).In case of normal distribution of data, continuous variables
were presented with mean and Standard Deviation (SD), while in case of non-
normal distribution, with median value and interquartile range [1Q 3Q].
179. We expressed quantitative variables as means (±SD) and used t-tests to
compare them when the sample size in each group was 30 or more (in
accordance with the central limit theorem) and the Wilcoxon rank-sum test
when the sample size in one group was less than 30. We used chi-square tests
or Fisher’s exact test to compare qualitative variables, as appropriate. N Engl J
Med 2014;370:1583-93.