The outcome of the Shapiro-Wilk test show that the FAS scores cannot be assumed to be normally distributed. This is true for both the 1st (p = 0.09, due to the high kurtosis: −1.14) and the 2nd (p = 0.002) questionnaire scores. As we have collected data in two steps, we will discuss at first the outcomes for each step (i.e 1st questionnaire, 2nd questionnaire) and later compare the two questionnaires outcomes to investigate the evolution of fatigue.
3.1. 1st Questionnaire
An overview of the total FAS scores is reported in
Figure 1 (see also
Table S2 in SI). A significant difference exists in FAS score between females and males (
p < 0.001), with higher score for females. The presence of a significant gender difference in the FAS score drove us to perform a more detailed analysis of FAS answers per gender, aimed to better elucidate the issue. The comparison between genders of the scores for each answer is reported in
Figure 2 (see also
Table S3 in SI). Mann-Whitney U test shows that there are significant differences for items 1 (
p = 0.010), 2 (
p = 0.003), 4 (
p = 0.006), 5 (
p = 0.006), 6 (
p = 0.005), 7 (
p = 0.005), 9 (
p = 0.002) while no significant difference was found for items 3 (
p = 0.132,
d = 0.56, power = 0.34), 8 (
p = 0.034;
d = 1.02, power = 0.78), and 10 (
p = 0.031;
d = 0.89, power = 0.60). In all cases females evidence more fatigue. The outcome (female have significant higher scores at the p < 0.01 level) holds also for the mental (
p = 0.002) and physical fatigue (
p < 0.001) subsets.
A non significant difference exists between the total scores of hospitalized and home confined patients (p = 0.0475, d = 0.81, power = 0.56) although home confined ones have nevertheless higher scores. A non significant difference exists for mental (p = 0.054, d = 0.76, power = 0.52) and physical (p = 0.034, d = 0.77, power = 0.51) fatigue too. It has to be noted that, given the p values close to the threshold and the large effect size, larger samples will be needed to confirm or deny this result. To remain on the conservative side, we assumed that a false negative cannot be ruled out and included place of care as an impact parameter for Bonferroni correction calculation.
3.2. 2nd Questionnaire
An overview of the total FAS scores is reported in
Figure 3 (see also
Table S5 in SI). A significant difference exists in FAS score between females and males (
p = 0.012), with higher score for females. The comparison between genders of the scores for each answer is reported in
Figure 4 (see also
Table S6 in SI). Mann-Whitney U test shows that there are significant differences for items 1 (
p = 0.005), 2 (
p = 0.017), 4 (
p = 0.019), and 5 (
p = 0.003) while no significant differences were found for items 3 (
p = 0.154,
d = 0.68, power = 0.50), 6 (
p = 0.057;
d = 0.77, power = 0.53), 7 (
p = 0.119;
d = 0.78, power = 0.59), 8 (
p = 0.026,
d = 1.01, power = 0.79), 9 (
p = 0.038,
d = 1.01, power = 0.75) and 10 (
p = 0.047;
d = 0.82, power = 0.60). For most of them a larger sample will surely provide a more reliable outcome (see comment in the previous paragraph). In all cases females evidence more fatigue. The outcome (female have significant higher scores) holds also for the physical fatigue (
p = 0.007) while mental fatigue (
p = 0.029,
d = 1.00, power = 0.73) does not show a significant difference. No significant difference exists between hospitalized and home confined patients for the total FAS score (
p = 0.206,
d = 0.47, power = 0.27), the mental (
p = 0.149,
d = 0.42, power = 0.24) and the physical (
p = 0.174,
d = 0.49, power = 0.28) fatigue.
3.3. Comparison of the Two Questionnaires
A significant improvement is observed for the whole set of patients for:
total FAS score (p < 0.001), as well as mental (p = 0.004) and physical (p < 0.001) subsets;
items 1 (p < 0.001), 2 (p < 0.001), 3 (p = 0.007), 4 (p = 0.014), 5 (p = 0.009), 8 (p = 0.007), 9 (p = 0.004)
No significant improvement was detected for items 6 (p = 0.028, d = 0.50, power = 0.49), 7 (p = 0.134, d = 0.25, power = 0.25) and 10 (p = 0.113, d = 0.34, power = 0.30)
All males reduced their FAS scores. The significant improvements are the following:
total FAS score (p < 0.001), as well as mental (p = 0.009) and physical (p < 0.001) subsets;
items 1 (p < 0.001), 2 (p = 0.001), 3 (p = 0.012), 4 (p = 0.021), 5 (p = 0.018), 8 (p = 0.007), 9 (p = 0.017)
Items 6 (p = 0.119, d = 0.41, power = 0.26), 7 (p = 0.352, d = 0.21, power = 0.13), 10 (p = 0.184, d = 0.36, power = 0.22) have no significant improvement. These items are related to mental fatigue.
For females the only significant improvement was found for item 3 (p = 0.019) of the questionnaire. No significant improvement was found for items 1 (p = 0.032, d = 1.01, power = 0.68), 2 (p = 0.029, d = 1.15, power = 0.78), 4 (p = 0.154, d = 0.77, power = 0.49), 5 (p = 0.081, d = 0.77, power = 0.49), 6 (p = 0.056, d = 0.69, power = 0.43), 7 (p = 0.192, d = 0.34, power = 0.17), 8 (p = 0.106, d = 0.67, power = 0.41), 9 (p = 0.048, d = 0.97, power = 0.65), 10 (p = 0.224, d = 0.37, power = 0.19). Non significant are also the improvements in FAS score (p = 0.061, d = 0.87, power = 0.57), and physical (p = 0.052, d = 0.94, power = 0.63) and mental (p = 0.061, d = 0.76, power = 0.48) fatigue. These results are related to the fact that 4 of the 10 females worsened their FAS scores while 6 reduced them. However, as the average variation in the FAS score for the 4 females that increased it is 4, while the average decrease for the remaining 6 is −20.3, it can be readily understood why most of the p values are close to the significance threshold. As for those p values the effect size is usually large (>0.8), an increase in the sample size in further studies can reduce the risk of a false negative conclusion.
The difference in the ability to reduce the FAS score between males and females (
Table 4) is significant, despite the small size of the sample, as witnessed by the outcome of Fisher’s Exact test:
p = 0.024.
3.4. Additional Outcomes
The smoking habitus (previous smokers vs no smokers) has no significant impact on the FAS in our group of patients. In fact for the 1st questionnaire p = 0.440, d = 0.01, power = 0.05 while for the 2nd questionnaire p = 0.364, d = 0.01, power = 0.08.
The Pearson’s correlation coefficient between FAS score and BMI is r(18) = 0.25, p = 0.288 for the 1st questionnaire and r(18) = 0.28, p = 0.23 for the 2nd one. This shows that BMI has at most a marginal impact on the FAS score. The number of degree of freedom is limited to 18 because the BMI values of 3 patients were not available.
The Pearson’s correlation coefficient between FAS score and age is r(21) = −0.09, p = 0.683 for the 1st questionnaire and r(18) = 0.13, p = 0.55 for the 2nd one. This shows that age has no impact on the FAS score in our case.