| . | Step 5: perform the initial analysis and report . | Step 6: analyse biomarker data . | Step 7: consider covariates . | Step 8: interpret and present results . |
|---|---|---|---|---|
| Example 1 Butts et al. (2023)2 | Descriptive statistics were analysed for the study variables, and the data were reviewed for normality assumptions and outliers | ANCOVA was used to examine between-group differences at 3 months. Paired t tests were used to examine within-group differences over time. Pearson correlations were used to examine the relationship between telomere length and cytokines. Effect sizes were calculated using Hedges’ g for paired t tests and η2 for analysis of covariance models | Adjusted for baseline values with ANCOVA. As this was a pilot study with a small sample (n = 32), additional covariates were not included | Total telomere length increased and plasma IL-1β levels decreased in the exercise group from baseline to 3 months. Total telomere length was negatively associated with IL-1β at baseline |
| Example 2 Butts et al. (2023)3 | Descriptive statistics were analysed for the study variables, and the data were reviewed for normality assumptions and outliers | Zero-truncated Poisson regression models were constructed to estimate the effect of a predictor variable | Society of Thoracic Surgeons-Predicted Risk of Morbidity pre-operative risk score was included in multi-variate models | Multivariable Poisson regression models with 4-h PCF chymase activity plus risk score as predictor variables demonstrate best model fitting for prediction of ICU and total hospital length of stay |
| Example 3 Denfeld et al. (2021)4 | sST2 was interpolated from a standard curve. Descriptive statistics for all variables, and biomarker data were natural log-transformed | Latent growth curve modelling was used to estimate change in NT-proBNP and sST2 from pre- to 1, 3, and 6 months post-LVAD implantation. Quantified the effect of gender on intercepts and each shape or phase of change | There were very few significant differences between women and men pre-LVAD, and thus, did not adjust for covariates | NT-proBNP: women and men had similar values of NT-proBNP at baseline and through 1 month post-implantation, but NT-proBNP decreased at a steeper rate from 1 to 6 months post-implantation compared with women. sST2: women and men had similar values of sST2 at baseline, but the trajectories of change thereafter were significantly different as sST2 increased and then decreased for women, whereas for men, it just decreased |
| Example 4 Butts et al. (2019)5 | Descriptive statistics were calculated for all study variables and data were reviewed for normality assumptions and outliers in preparation for analysis | Multiple linear regression was used to examine linear relationships. Student’s t-tests were used to compare groups (normotensive vs. resistant hypertension) | Adjusted for age, sex, and body mass index | Xanthine oxidase activity was increased two-fold in resistant hypertension vs. normal and was positively associated with left ventricular mass, left ventricular diastolic function, and 24-h urinary sodium |
| Worked example | Performed descriptive analyses; transformed data to general normal distributions; calculated HOMA-IR | Using generalized linear modelling, examined associations between biomarkers and HRQOL | Adjusted for severity of heart failure (Seattle Heart Failure Model Score) and comorbidities (Charlson Comorbidity Index) | Worse HRQOL may be associated with insulin resistance or related pathways |
| . | Step 5: perform the initial analysis and report . | Step 6: analyse biomarker data . | Step 7: consider covariates . | Step 8: interpret and present results . |
|---|---|---|---|---|
| Example 1 Butts et al. (2023)2 | Descriptive statistics were analysed for the study variables, and the data were reviewed for normality assumptions and outliers | ANCOVA was used to examine between-group differences at 3 months. Paired t tests were used to examine within-group differences over time. Pearson correlations were used to examine the relationship between telomere length and cytokines. Effect sizes were calculated using Hedges’ g for paired t tests and η2 for analysis of covariance models | Adjusted for baseline values with ANCOVA. As this was a pilot study with a small sample (n = 32), additional covariates were not included | Total telomere length increased and plasma IL-1β levels decreased in the exercise group from baseline to 3 months. Total telomere length was negatively associated with IL-1β at baseline |
| Example 2 Butts et al. (2023)3 | Descriptive statistics were analysed for the study variables, and the data were reviewed for normality assumptions and outliers | Zero-truncated Poisson regression models were constructed to estimate the effect of a predictor variable | Society of Thoracic Surgeons-Predicted Risk of Morbidity pre-operative risk score was included in multi-variate models | Multivariable Poisson regression models with 4-h PCF chymase activity plus risk score as predictor variables demonstrate best model fitting for prediction of ICU and total hospital length of stay |
| Example 3 Denfeld et al. (2021)4 | sST2 was interpolated from a standard curve. Descriptive statistics for all variables, and biomarker data were natural log-transformed | Latent growth curve modelling was used to estimate change in NT-proBNP and sST2 from pre- to 1, 3, and 6 months post-LVAD implantation. Quantified the effect of gender on intercepts and each shape or phase of change | There were very few significant differences between women and men pre-LVAD, and thus, did not adjust for covariates | NT-proBNP: women and men had similar values of NT-proBNP at baseline and through 1 month post-implantation, but NT-proBNP decreased at a steeper rate from 1 to 6 months post-implantation compared with women. sST2: women and men had similar values of sST2 at baseline, but the trajectories of change thereafter were significantly different as sST2 increased and then decreased for women, whereas for men, it just decreased |
| Example 4 Butts et al. (2019)5 | Descriptive statistics were calculated for all study variables and data were reviewed for normality assumptions and outliers in preparation for analysis | Multiple linear regression was used to examine linear relationships. Student’s t-tests were used to compare groups (normotensive vs. resistant hypertension) | Adjusted for age, sex, and body mass index | Xanthine oxidase activity was increased two-fold in resistant hypertension vs. normal and was positively associated with left ventricular mass, left ventricular diastolic function, and 24-h urinary sodium |
| Worked example | Performed descriptive analyses; transformed data to general normal distributions; calculated HOMA-IR | Using generalized linear modelling, examined associations between biomarkers and HRQOL | Adjusted for severity of heart failure (Seattle Heart Failure Model Score) and comorbidities (Charlson Comorbidity Index) | Worse HRQOL may be associated with insulin resistance or related pathways |
ANCOVA, analysis of covariance; HF, heart failure; HOMA-IR, homeostatic model assessment for insulin resistance (HOMA-IR); HRQOL, health-related quality of life; ICU, intensive care unit; IL-1β, interleukin 1 beta; LVAD, left ventricular assist device; NT-proBNP, N-terminal B-type natriuretic peptide; sST2, soluble suppressor of tumorgenicity.
aSteps 1–4 are presented in Part 1.
| . | Step 5: perform the initial analysis and report . | Step 6: analyse biomarker data . | Step 7: consider covariates . | Step 8: interpret and present results . |
|---|---|---|---|---|
| Example 1 Butts et al. (2023)2 | Descriptive statistics were analysed for the study variables, and the data were reviewed for normality assumptions and outliers | ANCOVA was used to examine between-group differences at 3 months. Paired t tests were used to examine within-group differences over time. Pearson correlations were used to examine the relationship between telomere length and cytokines. Effect sizes were calculated using Hedges’ g for paired t tests and η2 for analysis of covariance models | Adjusted for baseline values with ANCOVA. As this was a pilot study with a small sample (n = 32), additional covariates were not included | Total telomere length increased and plasma IL-1β levels decreased in the exercise group from baseline to 3 months. Total telomere length was negatively associated with IL-1β at baseline |
| Example 2 Butts et al. (2023)3 | Descriptive statistics were analysed for the study variables, and the data were reviewed for normality assumptions and outliers | Zero-truncated Poisson regression models were constructed to estimate the effect of a predictor variable | Society of Thoracic Surgeons-Predicted Risk of Morbidity pre-operative risk score was included in multi-variate models | Multivariable Poisson regression models with 4-h PCF chymase activity plus risk score as predictor variables demonstrate best model fitting for prediction of ICU and total hospital length of stay |
| Example 3 Denfeld et al. (2021)4 | sST2 was interpolated from a standard curve. Descriptive statistics for all variables, and biomarker data were natural log-transformed | Latent growth curve modelling was used to estimate change in NT-proBNP and sST2 from pre- to 1, 3, and 6 months post-LVAD implantation. Quantified the effect of gender on intercepts and each shape or phase of change | There were very few significant differences between women and men pre-LVAD, and thus, did not adjust for covariates | NT-proBNP: women and men had similar values of NT-proBNP at baseline and through 1 month post-implantation, but NT-proBNP decreased at a steeper rate from 1 to 6 months post-implantation compared with women. sST2: women and men had similar values of sST2 at baseline, but the trajectories of change thereafter were significantly different as sST2 increased and then decreased for women, whereas for men, it just decreased |
| Example 4 Butts et al. (2019)5 | Descriptive statistics were calculated for all study variables and data were reviewed for normality assumptions and outliers in preparation for analysis | Multiple linear regression was used to examine linear relationships. Student’s t-tests were used to compare groups (normotensive vs. resistant hypertension) | Adjusted for age, sex, and body mass index | Xanthine oxidase activity was increased two-fold in resistant hypertension vs. normal and was positively associated with left ventricular mass, left ventricular diastolic function, and 24-h urinary sodium |
| Worked example | Performed descriptive analyses; transformed data to general normal distributions; calculated HOMA-IR | Using generalized linear modelling, examined associations between biomarkers and HRQOL | Adjusted for severity of heart failure (Seattle Heart Failure Model Score) and comorbidities (Charlson Comorbidity Index) | Worse HRQOL may be associated with insulin resistance or related pathways |
| . | Step 5: perform the initial analysis and report . | Step 6: analyse biomarker data . | Step 7: consider covariates . | Step 8: interpret and present results . |
|---|---|---|---|---|
| Example 1 Butts et al. (2023)2 | Descriptive statistics were analysed for the study variables, and the data were reviewed for normality assumptions and outliers | ANCOVA was used to examine between-group differences at 3 months. Paired t tests were used to examine within-group differences over time. Pearson correlations were used to examine the relationship between telomere length and cytokines. Effect sizes were calculated using Hedges’ g for paired t tests and η2 for analysis of covariance models | Adjusted for baseline values with ANCOVA. As this was a pilot study with a small sample (n = 32), additional covariates were not included | Total telomere length increased and plasma IL-1β levels decreased in the exercise group from baseline to 3 months. Total telomere length was negatively associated with IL-1β at baseline |
| Example 2 Butts et al. (2023)3 | Descriptive statistics were analysed for the study variables, and the data were reviewed for normality assumptions and outliers | Zero-truncated Poisson regression models were constructed to estimate the effect of a predictor variable | Society of Thoracic Surgeons-Predicted Risk of Morbidity pre-operative risk score was included in multi-variate models | Multivariable Poisson regression models with 4-h PCF chymase activity plus risk score as predictor variables demonstrate best model fitting for prediction of ICU and total hospital length of stay |
| Example 3 Denfeld et al. (2021)4 | sST2 was interpolated from a standard curve. Descriptive statistics for all variables, and biomarker data were natural log-transformed | Latent growth curve modelling was used to estimate change in NT-proBNP and sST2 from pre- to 1, 3, and 6 months post-LVAD implantation. Quantified the effect of gender on intercepts and each shape or phase of change | There were very few significant differences between women and men pre-LVAD, and thus, did not adjust for covariates | NT-proBNP: women and men had similar values of NT-proBNP at baseline and through 1 month post-implantation, but NT-proBNP decreased at a steeper rate from 1 to 6 months post-implantation compared with women. sST2: women and men had similar values of sST2 at baseline, but the trajectories of change thereafter were significantly different as sST2 increased and then decreased for women, whereas for men, it just decreased |
| Example 4 Butts et al. (2019)5 | Descriptive statistics were calculated for all study variables and data were reviewed for normality assumptions and outliers in preparation for analysis | Multiple linear regression was used to examine linear relationships. Student’s t-tests were used to compare groups (normotensive vs. resistant hypertension) | Adjusted for age, sex, and body mass index | Xanthine oxidase activity was increased two-fold in resistant hypertension vs. normal and was positively associated with left ventricular mass, left ventricular diastolic function, and 24-h urinary sodium |
| Worked example | Performed descriptive analyses; transformed data to general normal distributions; calculated HOMA-IR | Using generalized linear modelling, examined associations between biomarkers and HRQOL | Adjusted for severity of heart failure (Seattle Heart Failure Model Score) and comorbidities (Charlson Comorbidity Index) | Worse HRQOL may be associated with insulin resistance or related pathways |
ANCOVA, analysis of covariance; HF, heart failure; HOMA-IR, homeostatic model assessment for insulin resistance (HOMA-IR); HRQOL, health-related quality of life; ICU, intensive care unit; IL-1β, interleukin 1 beta; LVAD, left ventricular assist device; NT-proBNP, N-terminal B-type natriuretic peptide; sST2, soluble suppressor of tumorgenicity.
aSteps 1–4 are presented in Part 1.
This PDF is available to Subscribers Only
View Article Abstract & Purchase OptionsFor full access to this pdf, sign in to an existing account, or purchase an annual subscription.
