There was a mistake in Table 1 as published. The dietary caffeine intake is right-skewed, the median (interquartile range) is now reported instead of the mean ± SD: 73.00 (4.00, 192.00), 112.00 (26.00, 246.02), and 146.77 (49.70, 289.00) in the low-, middle-, and high-income groups, respectively. In addition, two values in the “Marital status” rows of the high-income group were transcribed incorrectly: “Separated” should read 40 (1.52%) rather than 296 (11.22%), and “Living with partner” should read 100 (3.79%) rather than 277 (5.80%).
The corrected column sums to the stated group total of n = 2,637.
Reviewing the original statistical results of this study, the PIR groups were as follows: PIR < 1; 1 ≤ PIR < 4; PIR ≥ 4. All statistical processes were conducted under these groups. Therefore, the group description in Table 1 is incorrect.
The corrected Table 1 appears below:
In the published article, there was an error in the Methods section, “2.3 Statistical analysis”. The sentence describing the tests used to compare participant characteristics across the three PIR groups does not describe the analysis that was actually performed. All baseline comparisons reported in Table 1 were generated in EmpowerStats (version 2.0) and R (version 4.1.3): continuous variables were compared using the Kruskal–Wallis rank-sum test and categorical variables using the chi-square test.
No comparison in Table 1 was performed using a t-test, and none of the p-values reported in Table 1 have changed. Additionally, there is an error in the PIR grouping description.
A correction has been made to Methods, section 2.3, paragraph 1. This sentence previously stated: “T-test and chi-square test were used to characterize participants' demographic information after PIR grouping (PIR < 1, PIR1-4, PIR > 4) in the study.”
The corrected sentence appears below:
“Continuous variables are presented as mean ± SD and were compared across the three PIR groups (PIR < 1, 1 ≤ P IR < 4, PIR ≥ 4) using the Kruskal–Wallis rank-sum test; categorical variables are presented as n (%) and were compared using the chi-square test.”
According to the original statistical results of this study, the PIR groups were as follows: PIR < 1, 1 ≤ PIR < 4, PIR ≥ 4; All statistical processes were conducted within the above groups. Therefore, there are still multiple errors in the PIR group descriptions in the original article, as follows:
In the “Abstract; Results” section:
The corrected sentence appears below: The variable PIR was categorized as the low-income (PIR < 1), the middle-income (1 ≤ PIR < 4), and the high-income (PIR ≥ 4). Notably, there was a significant difference in trend for the high-income group (PIR ≥ 4) compared to the control low-income group (PIR < 1) (all P for interaction < 0.05).
In the “2.2 Study variables” section:
The corrected sentence appears below: We then categorized these into three classes: the low-income (PIR < 1), the middle-income (1 ≤ PIR < 4), and the high-income (PIR ≥ 4).
In the “3.1 Baseline characteristics of participants” section:
The corrected sentence appears below: There were 1,385 (15.74%) participants with low-income (PIR < 1), 4,778 (54.30%) with middle-income (1 ≤ PIR < 4), and 2,637 (29.96%) with high-income (PIR ≥ 4). Notably, the low-income individuals (PIR < 1) were younger, more likely to have less than a high school education, had lower coffee intake, and had a higher prevalence of severe headaches/migraines compared to the high-income individuals (PIR ≥ 4).
In the “3.3 Correlation of PIR with severe headache/migraine” section:
The corrected sentence appears below: After triple categorizing the variable PIR (PIR < 1, 1 ≤ PIR < 4, PIR ≥ 4), there was a significant difference in trend for the high-income group (PIR ≥ 4) compared to the low-income group (PIR < 1).
In the “Table 3” section
The corrected Table 3 is as follows:
The original version of this article has been updated.