Paper 25: Impact of negative tuberculin skin test on growth among disadvantaged Bangladeshi children

Author

Lee Jones - Senior Biostatistician - Statistical Review

Published

March 15, 2026

Reference

Gaffar SMA, Chisti MJ, Mahfuz M, Ahmed T (2019) Impact of negative tuberculin skin test on growth among disadvantaged Bangladeshi children. PLoS ONE 14(11): e0224752. https://doi.org/10.1371/journal.pone.0224752

Disclosure

This reproducibility project was conducted to the best of our ability, with careful attention to statistical methods and assumptions. The research team comprises four senior biostatisticians (three of whom are accredited), with 20 to 30 years of experience in statistical modelling and analysis of healthcare data. While statistical assumptions play a crucial role in analysis, their evaluation is inherently subjective, and contextual knowledge can influence judgements about the importance of assumption violations. Differences in interpretation may arise among statisticians and researchers, leading to reasonable disagreements about methodological choices.

Our approach aimed to reproduce published analyses as faithfully as possible, using the details provided in the original papers. We acknowledge that other statisticians may have differing success in reproducing results due to variations in data handling and implicit methodological choices not fully described in publications. However, we maintain that research articles should contain sufficient detail for any qualified statistician to reproduce the analyses independently.

Methods used in our reproducibility analyses

There were two parts to our study. First, 100 articles published in PLOS ONE were randomly selected from the health domain and sent for post-publication peer review by statisticians. Of these, 95 included linear regression analyses and were therefore assessed for reporting quality. The statisticians evaluated what was reported, including regression coefficients, 95% confidence intervals, and p-values, as well as whether model assumptions were described and how those assumptions were evaluated. This report provides a brief summary of the initial statistical review.

The second part of the study involved reproducing linear regression analyses for papers with available data to assess both computational and inferential reproducibility. All papers were initially assessed for data availability and the statistical software used. From those with accessible data, the first 20 papers (from the original random sample) were evaluated for computational reproducibility. Within each paper, individual linear regression models were identified and assigned a unique number. A maximum of three models per paper were selected for assessment. When more than three models were reported, priority was given to the final model or the primary models of interest as identified by the authors; any remaining models were selected at random.

To assess computational reproducibility, differences between the original and reproduced results were evaluated using absolute discrepancies and rounding error thresholds, tailored to the number of decimal places reported in each paper. Results for each reported statistic, e.g., regression coefficient, were categorised as Reproduced, Incorrect Rounding, or Not Reproduced, depending on how closely they matched the original values. Each paper was then classified as Reproduced, Mostly Reproduced, Partially Reproduced, or Not Reproduced. The mostly reproduced category included cases with minor rounding or typographical errors, whereas partially reproduced indicated substantial errors were observed, but some results were reproduced.

For models deemed at least partially computationally reproducible, inferential reproducibility was further assessed by examining whether statistical assumptions were met and by conducting sensitivity analyses, including bootstrapping where appropriate. We examined changes in standardized regression coefficients, which reflect the change in the outcome (in standard deviation units) for a one standard deviation increase in the predictor. Meaningful differences were defined as a relative change of 10% or more, or absolute differences of 0.1 (moderate) and 0.2 (substantial). When non-linear relationships were identified, inferential reproducibility was assessed by comparing model fit measures, including R², Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC). When the Gaussian distribution was not appropriate for the dependent variable, alternative distributions were considered, and model fit was evaluated using AIC and BIC.

Results from the reproduction of the Gaffar et al. (2019) paper are presented below. An overall summary of results is presented first, followed by model-specific results organised within tab panels. Within each panel, the Original results tab displays the linear regression outputs extracted from the published paper. The Reproduced results tab presents estimates derived from the authors’ shared data, along with a comprehensive assessment of linear regression assumptions. The Differences tab compares the original and reproduced models to assess computational reproducibility. Finally, the Sensitivity analysis tab evaluates inferential reproducibility by examining whether identified assumption violations meaningfully affected the results.

Summary from statistical review

This paper examines growth differences between Tuberculin skin test and negative undernourished children growth among disadvantaged Bangladeshi children. Statistical methods were described well and presented clearly in tables identifying the test used. There was the overuse of acronyms, but generally easy to read. No assumptions were discussed. The author’s used pre-post change scores for linear regression but did not explicitly mention independence. They have not interpreted the coefficients except in terms of significance.

Data availability and software used

The data was available in the supporting information as a wide CSV file, with no data dictionary provided. Linear regression results were analysed using Stata.

Regression sample

Twelve linear regression models identified, Three main outcomes with multivariable linear regression reported, therefore, randomisation was not required. The three outcomes considered were change in length for age Z score LAZ, Change in weight for age Z score WAZ, and change in weight for length Z score WLZ, the main independent variable was whether Tuberculin skin test was positive or negative with adjustment for sex (male or female) and post-intervention age.

Computational reproducibility results

The models assessed were mostly computationally reproducible, with only minor rounding differences. P-values reported in the paper showed minor rounding discrepancies, consistent with truncation to two decimal places rather than rounding, but these differences did not affect their interpretation. Regression coefficients retained their direction. Some data manipulation was required to create change scores, and age measured in days required converting into months. Overall, the study was manageable to reproduce.

Inferential reproducibility results

Two of the three models were inferentially reproducible, meaning that overall the paper was not inferentially reproducible. For the LAZ model, the Breusch–Pagan test was not statistically significant. However, the residuals versus fitted plot showed a funnel-shaped pattern, suggesting some heteroscedasticity and wild bootstrapping was used as a comparison. The Q–Q plot and Kolmogorov–Smirnov test indicated that the residuals were approximately normal, although the Shapiro–Wilk test suggested minor deviations from normality. The covariance ratio flagged potential outliers that could inflate coefficient variances and widen confidence intervals. The study sampling method was unclear and independence could not be tested as there were no identifiers in the dataset. The WAZ and WLZ models exhibited similar assumption issues but did not show evidence of heteroscedasticity and BCa bootsrapping was used as a comparison for these models. The direction and statistical significance of effects remained consistent between the reproduced and bootstrapped models for all outcomes. However, the WLZ model was not inferentially reproducible, with standardized coefficient range differing by 0.11 (14%). In contrast, the LAZ and WAZ models differences in standardized coefficients and CI ranges were <0.1, indicating inferential reproducibility.

Recommended Changes

Derived variables, such as change scores, should be included in the dataset to facilitate reproducibility.
A unique participant-level identifier should be provided to allow linkage of observations and verification of analyses.
The original study design should be described in sufficient detail to clarify whether participants were clustered (e.g., within neighbourhoods or families) and whether adjustment for clustering was required.
Evaluate the assumptions of the linear regression models by examining residuals, identifying influential outliers, and assessing multicollinearity among predictors. If any assumptions are violated, address them using appropriate methods.
Include a data dictionary.
Place greater emphasis on the magnitude and practical relevance of regression coefficients, including their confidence intervals, rather than focusing predominantly on statistical significance.

Model 1

Model results for LAZ change

Term	B	Lower	Upper	p-value
Intercept
TST:
Positive – Negative	−0.04	−0.15	0.08	0.51
Sex:
Female – Male	−0.02	−0.10	0.05	0.52
Age	0.003	−0.01	0.02	0.74
SE = Standard error; Lower = lower confidence interval; Upper = upper confidence interval.

Fit statistics for LAZ change

R	R2	R2Adj	AIC	RMSE	F	DF1	DF2	p-value

R2 Adj = Adjusted R2; AIC = Akaike Information Criterion; RMSE = The Root Mean Squared Error; DF1 = Degrees of freedom for the model; DF2 = Degrees of freedom for the residuals.

ANOVA table for LAZ change

Term	SS	DF	MS	F	p-value
TST
Sex
Age
Residuals
SS = Sum of Squares; DF = Degrees of freedom; MS = Mean Square.

Model results for LAZ change

Term	B	SE	Lower	Upper	t	p-value
Intercept	0.004	0.171	−0.333	0.340	0.021	0.9831
TST:
Positive – Negative	−0.039	0.059	−0.155	0.077	−0.657	0.5116
Sex:
Female – Male	−0.025	0.038	−0.100	0.051	−0.642	0.5213
Age	0.003	0.009	−0.015	0.021	0.321	0.7487
SE = Standard error; Lower = lower confidence interval; Upper = upper confidence interval.

Fit statistics for LAZ change

R	R2	R2Adj	AIC	RMSE	F	DF1	DF2	p-value
0.063	0.004	−0.009	105.179	0.294	0.314	3	239	0.8155
R2 Adj = Adjusted R2; AIC = Akaike Information Criterion; RMSE = The Root Mean Squared Error; DF1 = Degrees of freedom for the model; DF2 = Degrees of freedom for the residuals.

ANOVA table for LAZ change

Term	SS	DF	MS	F	p-value
TST	0.038	1	0.038	0.432	0.5116
Sex	0.036	1	0.036	0.413	0.5213
Age	0.009	1	0.009	0.103	0.7487
Residuals	21.049	239	0.088
SS = Sum of Squares; DF = Degrees of freedom; MS = Mean Square; Calculated using type III SS.

Visualisation of regression model

The blue line shows the best line of fit with shading representing 95% confidence intervals, while holding all other covariates constant. The dots show partial residuals, which reflect the observed data adjusted for all other predictors except the one being plotted.

Checking residuals plots for patterns

Blue line showing quadratic fit for residuals

Testing residuals for non linear relationships

Term	Statistic	p-value	Results
TST
Sex
Age	−1.576	0.1163	No linearity violation
Tukey test	−0.098	0.9222	No linearity violation
Specification test for predictors using quadratic tests, for fitted values curvature is tested through Tukey's one-degree-of-freedom test for nonadditivity.

Checking univariate relationships with the dependent variable using scatterplots

Blue line shows linear relationship, red line indicates relationship inferred by GAM modelling

Linearity results

No linearity violation was observed in either plots or tests.

Testing for homoscedasticity

Statistic	p-value	Parameter	Method
5.341	0.1484	3	studentized Breusch-Pagan test

Homoscedasticity results

The studentized Breusch-Pagan test supports homoscedasticity.
Some heteroscedasticity is present in plots, and a sensitivity analysis using weighted or robust regression or wild bootstrapping is recommended.

Model descriptives including cook’s distance and leverage to understand outliers

Term	N	Mean	SD	Median	Min	Max	Skewness	Kurtosis
LAZ change	243	0.040	0.296	0.000	−0.810	1.020	0.358	0.329
TST*	243	1.119	0.325	1.000	1.000	2.000	2.334	3.461
Sex*	243	1.473	0.500	1.000	1.000	2.000	0.106	−1.997
Age	243	18.632	2.127	18.533	14.867	23.067	0.122	−1.174
.fitted	243	0.040	0.019	0.039	−0.015	0.069	−0.748	0.648
.resid	243	0.000	0.295	−0.030	−0.842	0.994	0.370	0.337
.leverage	243	0.016	0.011	0.013	0.008	0.054	1.967	2.780
.sigma	243	0.297	0.001	0.297	0.290	0.297	−3.483	15.964
.cooksd	243	0.004	0.006	0.002	0.000	0.042	3.296	12.789
.std.resid	243	−0.000	1.002	−0.102	−2.850	3.372	0.371	0.325
dfb.1_	243	0.000	0.061	0.000	−0.175	0.301	0.362	2.623
dfb.TSTP	243	−0.000	0.064	0.002	−0.308	0.362	0.938	11.083
dfb.SxMl	243	0.000	0.065	0.003	−0.205	0.228	0.160	0.448
dfb.Age	243	−0.000	0.059	−0.001	−0.255	0.174	−0.175	1.666
dffit	243	−0.002	0.126	−0.012	−0.350	0.413	0.491	0.684
cov.r	243	1.017	0.028	1.023	0.848	1.066	−2.475	10.174
* categorical variable

Cooks threshold

Cook’s distance measures the overall change in fit, if the ith observation is removed. Potential influential observations are identified by \(\text{Cook's Distance}_i > \frac{4}{n}\), where n is the number of observations. In practice a threshold of 0.5 to 1 is often used to identify influential observations.

DFFIT threshold

DFFIT measures how many standard deviations the fitted values will change when the ith observation is removed. Potential influential observations \(\left| \text{DFFITS}_i \right| > \frac{2\sqrt{p}}{\sqrt{n}}\) where p is the number of predictors (including the intercept) and n is the number of observations. In practice this can result in a large number of points identified, a practical cut-off of 1 was used to flag observations with meaningful impact.

DFBETA threshold

DFBETAS quantify the influence of the ith observation on the jth regression coefficient as the change in that coefficient when the observation is omitted, expressed in units of the coefficient’s estimated standard error. There is a DFBETA for each parameter in the model. Potential influential observations \(|\text{DFBETA}_{ij}| > \frac{2}{\sqrt{n}}\), where n is the number of observations. In larger datasets this threshold can flag a high number of observations with only minor influence on the model. A practical cut-off of 1 was used to flag observations with meaningful impact.

Influence plot

Observations with high leverage (horizontal) and large residuals (vertical, typically at ±2 or ±3 studentized residuals) are concerning, as they may disproportionately influence the model. This combination is reflected by large bubbles with high Cook’s distance indicated by darker shadings of blue.

COVRATIO plot

COVRATIO measures the overall change in the precision (covariance matrix) of the estimated regression coefficients if the ith observation is removed. Values close to 1 indicate little influence on the model’s precision. Values below 1 indicate inflated variances and reduced precision (wider confidence intervals), whereas values above 1 indicate deflated variances and increased precision (narrower confidence intervals). A commonly cited guideline is \(\left|\mathrm{COVRATIO}_i - 1\right| > \frac{3p}{n}\), where p is the number of parameters and n is the number of observations. A practical cut-off between 0.9 to 1.1 was used to flag observations with meaningful impact on precision, although there is no agreed universal alternative cut-off.

Observations of interest identified by the influence plot

ID	StudRes	Leverage	CookD	dfb.1_	dfb.TSTP	dfb.SxMl	dfb.Age	dffit	cov.r
50	−1.064	0.050	0.015	0.125	−0.196	−0.062	−0.120	−0.244	1.050
19	3.268	0.011	0.028	0.119	−0.100	0.228	−0.117	0.340	0.862
233	1.409	0.054	0.028	−0.161	0.272	−0.104	0.174	0.337	1.040
89	3.448	0.014	0.039	0.301	−0.091	−0.205	−0.255	0.405	0.848
49	2.085	0.038	0.042	−0.015	0.362	0.123	0.002	0.413	0.983
StudRes = studentized residuals; CookD = Cook's Distance a combined measure of leverage and influence. DFBETAS (dfb.*) measures how much a specific regression coefficient changes (in standard errors) when an observation is removed; DFFITS measures how much the fitted (predicted) value for an observation changes (in standard deviations) when that observation is removed; cov.r = coefficient covariance ratio which measures how much the overall variance (precision) of the coefficients changes when that observation is removed.

Results for outliers and influential points

Two observations had studentized residuals greater than 3, both had low leverage and Cook’s distance values, with cook’s distance, DFBETAs, and DFFITS within reasonable ranges, but the covariance ratio values were substantially lower than one. Suggesting minimal impact on the coefficient estimates but potential inflation of standard errors and consequent widening of confidence intervals.

Checking for normality of the residuals using a Q–Q plot

Normality of residuals using Shapiro-Wilk and Kolmogorov-Smirnov tests

Statistic	p-value	Method
0.065	0.2613	Asymptotic one-sample Kolmogorov-Smirnov test

Statistic	p-value	Method
0.988	0.0415	Shapiro-Wilk normality test

Normality results

The Kolmogorov-Smirnov supports residuals being normally distributed.
The Shapiro-Wilk normality test indicates residuals may not be normally distributed.
QQ-plot looks roughly normal.

Assessing collinearity with VIF

Term	VIF	Tolerance
TST	1.010	0.990
Sex	1.003	0.997
Age	1.007	0.993
VIF = Variance Inflation Factor.

Collinearity results

All VIF values are under three, indicating no collinearity issues.
Overall, when taking into account VIF and SE, the model does not have collinearity issues.

Assessing independence with the Durbin–Watson test for autocorrelation

AutoCorrelation	Statistic	p-value
0.256	1.483	<0.001

Independence results

The Durbin–Watson test suggests there are auto-correlation issues.
Autocorrelation of residuals was assessed using the Durbin–Watson test, which evaluates first-order serial correlation. Evidence of autocorrelation may arise from unmodelled clustering or repeated measurements, but can also reflect artefacts of data ordering (e.g. sorting by time or subject) rather than true temporal dependence.
It was unclear if the study design was independent as recruitment of children was not explained.

Assumption conclusions

The model met the assumption of linearity. The residuals versus fitted plot showed a funnel-shaped pattern, suggesting mild heteroscedasticity; however, the Breusch–Pagan test was not statistically significant. A sensitivity analysis using weighted regression or wild bootstrapping is recommended. The Q–Q plot and Kolmogorov–Smirnov test indicated that the residuals were approximately normal, although the Shapiro–Wilk test suggested some deviation from normality. The covariance ratio identified potential outliers that could inflate coefficient variances and widen confidence intervals. The study design and recruitment of children were not clearly described, making it unclear whether observations were independent.

Forest plot showing original and reproduced coefficients and 95% confidence intervals for LAZ change

Change in regression coefficients

term	O_B	R_B	Change.B	reproduce.B
Intercept		0.0036
TST:
Positive – Negative	−0.040	−0.0388	0.0012	Reproduced
Sex:
Female – Male	−0.020	−0.0245	−0.0045	Reproduced
Age	0.003	0.0029	−0.0001	Reproduced
O_B = original B; R_B = reproduced B; Change.B = change in R_B - O_B; Reproduce.B = B reproduced.

Change in lower 95% confidence intervals for coefficients

term	O_lower	R_lower	Change.lci	Reproduce.lower
Intercept		−0.3329
TST:
Positive – Negative	−0.15	−0.1551	−0.0051	Incorrect Rounding
Sex:
Female – Male	−0.10	−0.0998	0.0002	Reproduced
Age	−0.01	−0.0148	−0.0048	Reproduced
O_lower = original lower confidence interval; R_lower = reproduced lower confidence interval; change.lci = change in R_lower - O_lower; Reproduce.lower = lower confidence interval reproduced.

Change in upper 95% confidence intervals for coefficients

term	O_upper	R_upper	Change.uci	Reproduce.upper
Intercept		0.3402
TST:
Positive – Negative	0.08	0.0775	−0.0025	Reproduced
Sex:
Female – Male	0.05	0.0507	0.0007	Reproduced
Age	0.02	0.0206	0.0006	Reproduced
O_upper = original upper confidence interval; R_upper = reproduced upper confidence interval; change.uci = change in R_upper - O_upper; Reproduce.upper = upper confidence interval reproduced.

Change in p-values

Term	O_p	R_p	Change.p	Reproduce.p	SigChangeDirection
Intercept		0.9831
TST:
Positive – Negative	0.51	0.5116	0.0016	Reproduced	Remains non-sig, B same direction
Sex:
Female – Male	0.52	0.5213	0.0013	Reproduced	Remains non-sig, B same direction
Age	0.74	0.7487	0.0087	Incorrect Rounding	Remains non-sig, B same direction
O_p = original p-value; R_p = reproduced p-value; Changep = change in p-value R_p - O_p; Reproduce.p = p-values reproduced. SigChangeDirection = statistical significance and B change between original and reproduced models. Note, p-values that were <0.001 were set to 0.00099 for the purposes of comparison.

Bland Altman plot showing differences between original and reproduced p-values for LAZ change

Results for p-values

P-values for this model were mostly reproduced. One p-value differed due to a rounding error, but this did not alter its statistical significance.

Conclusion computational reproducibility

This model was mostly computationally reproducible, with minor rounding errors. P-values were reproduced and had the same interpretation, and regression coefficients did not change direction.

Methods

The model was successfully reproduced; however, the residuals showed mild heteroscedasticity. To further verify the findings, bootstrapped standardized regression coefficients and their 95% confidence intervals were examined. Percentage and absolute changes in estimates and confidence-interval bounds relative to the linear model were summarised using thresholds of 10% change and standardized coefficient differences of <0.10 and <0.20. Coefficient direction and statistical significance were assessed for consistency. Wild bootstrapping was used to account for potential heteroscedasticity.

Results

Bootstrapped results

Wild bootstrapping was performed with 10,000 iterations.

Change in regression coefficients

Term	B	boot.B	B_diff	%_Diff	Diff_10%	Diff_0.1	Diff_0.2
TSTPositive	−0.1313	−0.1287	−0.0026	−1.9700	No	No	No
SexMale	0.0830	0.0826	0.0004	0.4500	No	No	No
z_Age	0.0208	0.0207	0.0001	0.5100	No	No	No
B = standardized regression coefficient reproduced B; boot.B = boostrapped standardized reproduced B; B_diff = change in B - boot.B; %_Diff = percentage difference; Diff_10% = difference ≥10% ; Diff_0.1 and Diff_0.2 = absolute difference ≥0.1 and ≥0.2, respectively.

Change in lower 95% confidence interval

Term	Lower	boot.Lower	Lower_diff	%_Diff	Diff_10%	Diff_0.1	Diff_0.2
TSTPositive	−0.5248	−0.5063	−0.0185	−3.5200	No	No	No
SexMale	−0.1716	−0.1712	−0.0004	−0.2500	No	No	No
z_Age	−0.1068	−0.0933	−0.0135	−12.6600	Yes	No	No
Lower = standardized reproduced lower CI; boot.Lower = boostrapped standardized reproduced lower CI; Lower_diff = change in Lower - boot.Lower; %_change = percentage difference; Diff_10% = difference ≥10% ; Diff_0.1 and Diff_0.2 = absolute difference ≥0.1 and ≥0.2, respectively.

Change in upper 95% confidence interval

Term	Upper	boot.Upper	Upper_diff	%_Diff	Diff_10%	Diff_0.1	Diff_0.2
TSTPositive	0.2622	0.2427	0.0195	7.4300	No	No	No
SexMale	0.3376	0.3371	0.0005	0.1500	No	No	No
z_Age	0.1484	0.1327	0.0157	10.5800	Yes	No	No
Upper = standardized reproduced upper CI; boot.Upper = boostrapped standardized reproduced upper CI; Upper_diff = change in Upper - boot.Upper; %_change = percentage difference; Diff_10% = difference ≥10% ; Diff_0.1 and Diff_0.2 = absolute difference ≥0.1 and ≥0.2, respectively.

Change in Range of 95% confidence interval

Term	Range	boot.Range	Range_Diff	%_Diff	Diff_10%	Diff_0.1	Diff_0.2
TSTPositive	0.7870	0.7491	−0.0379	−4.8200	No	No	No
SexMale	0.5092	0.5083	−0.0010	−0.1900	No	No	No
z_Age	0.2552	0.2260	−0.0292	−11.4500	Yes	No	No
Range = standardized reproduced CI range; boot.B = boostrapped standardized reproduced CI range; Range_diff = change in CI Range ; %_change = percentage difference, Diff_10% = difference ≥10% ; Diff_0.1 and Diff_0.2 = absolute difference ≥0.1 and ≥0.2, respectively.

Change in p-value significance and regression coefficient direction

Term	p-value	boot.p-value	changep	SigChangeDirection
TSTPositive	0.5116	0.5006	0.0110	Remains non-sig, B same direction
SexMale	0.5213	0.5200	0.0013	Remains non-sig, B same direction
z_Age	0.7487	0.7193	0.0294	Remains non-sig, B same direction
p-value = standardized reproduced p-value; boot.p-value = boostrapped standardized reproduced p-value; changep = change in p-value - boot.p-value; SigChangeDirection = statistical significance and B change between reproduced and bootstrapped model.

Check distribution of bootstrap estimates

The bootstrap distribution of each coefficient appeared approximately normal and centered near the original estimate (red dashed line), suggesting that the estimates are relatively stable. No strong skewness or multimodality was observed.

Conclusions based on bootstrapped model

This model was inferentially reproducible. While some statistics changed by 10% or more these differences were not meaningful with change in standardized regression coefficients and confidence intervals less than 0.1. The direction of effects and statistical significance remained consistent between the reproduced and bootstrapped models.

Model 2

Model results for WAZ change

Term	B	Lower	Upper	p-value
Intercept
TST:
Positive – Negative	0.04	−0.11	0.18	0.62
Sex:
Female – Male	0.01	−0.08	0.11	0.80
Age	0.01	−0.01	0.04	0.23
SE = Standard error; Lower = lower confidence interval; Upper = upper confidence interval.

Fit statistics for WAZ change

R	R2	R2Adj	AIC	RMSE	F	DF1	DF2	p-value

R2 Adj = Adjusted R2; AIC = Akaike Information Criterion; RMSE = The Root Mean Squared Error; DF1 = Degrees of freedom for the model; DF2 = Degrees of freedom for the residuals.

ANOVA table for WAZ change

Term	SS	DF	MS	F	p-value
TST
Sex
Age
Residuals
SS = Sum of Squares; DF = Degrees of freedom; MS = Mean Square.

Model results WAZ change

Term	B	SE	Lower	Upper	t	p-value
Intercept	−0.198	0.217	−0.626	0.230	−0.911	0.3634
TST:
Positive – Negative	0.037	0.075	−0.111	0.185	0.491	0.6236
Sex:
Female – Male	0.012	0.049	−0.084	0.108	0.246	0.8059
Age	0.014	0.011	−0.009	0.036	1.180	0.2391
SE = Standard error; Lower = lower confidence interval; Upper = upper confidence interval.

Model fit for WAZ change

R	R2	R2Adj	AIC	RMSE	F	DF1	DF2	p-value
0.081	0.007	−0.006	222.309	0.375	0.532	3	239	0.6609
R2 Adj = Adjusted R2; AIC = Akaike Information Criterion; RMSE = The Root Mean Squared Error; DF1 = Degrees of freedom for the model; DF2 = Degrees of freedom for the residuals.

ANOVA table for WAZ change

Term	SS	DF	MS	F	p-value
TST	0.034	1	0.034	0.241	0.6236
Sex	0.009	1	0.009	0.061	0.8059
Age	0.199	1	0.199	1.393	0.2391
Residuals	34.086	239	0.143
SS = Sum of Squares; DF = Degrees of freedom; MS = Mean Square; Calculated using type III SS.