A Statistical Definition of Epidemic Waves

Round 1

Reviewer 1 Report

This study describes a simple way 8 to evaluate whether an epidemic wave is likely to be present based on daily new case count data.

However, I have some comments.

1- The introduction is well-written but too short, it should be extended 

2-According to Raftery [8], the number of data points (n), the coefficient of determina- 68 tion (R2 ), and the number of predictors without the intercept (p) can be used to approxi- 69 mate the Bayesian information criterion (BIC) for linear models as

There are any  Information criteria like AIC and BAIC why you do not use them

3-. If necessary, thresholds for interpretation are available, clas- 82 sifying a Bayes factor between 1 and 3 as weak, between 3 and 20 as positive, between 20 83 and 150 as strong, and above 150 as very strong evidence

Need more references for this statement

4- The figures need some elaborations for the readers 

5- Many typos should be corrected 

5- The conclusion is good, but I suggest making a conclusion section if possible

6- The main contribution of this study must be well defined in the introduction

The English is good

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript presents an interesting approach for detecting epidemic waves that takes into account the timeframe of reference and possible irregularities of the data. this manuscript has the potential to make a valuable contribution to the field of epidemiology, but significant revisions are needed before it can be considered for publication.

 Abstract:

The abstract could benefit from more clarity regarding the significance of the study. While the abstract notes that the proposed measure detects epidemic waves at an early stage, it does not specify how much earlier than traditional methods or what impact this early detection could have on resource allocation and intervention preparation. A clearer explanation of the practical applications and potential implications of the proposed measure could make the study more impactful and relevant to public health decision making.

While the abstract mentions the use of regression analyses to compare exponential and linear models, it does not explain how these models were developed or what variables were included in the analyses. Additionally, the abstract does not discuss the limitations or potential biases of the proposed measure, which could impact the validity of the results.

 

Introduction:

The introduction provides a good overview of the concept of epidemic waves and the challenges associated with defining and identifying them. However, it does not sufficiently explain the significance of the problem or why it is important to accurately identify epidemic waves. The introduction could benefit from a more thorough discussion of the practical implications of misidentifying or failing to detect epidemic waves, such as inadequate allocation of resources or delayed implementation of interventions.

A more thorough explanation of the statistical approach and its advantages over other methods could help readers better understand the novelty and significance of the study. Additionally, the introduction could benefit from a clearer explanation of how the proposed measure relates to the broader goal of identifying and responding to infectious disease outbreaks.

Methodology:

While the proposed methodology seems to have some theoretical foundation, it relies heavily on assumptions, such as the linear approximation of subexponential growth of total case counts and the equivalence of linear associations of logarithmic predictors with exponential associations of untransformed criteria. These assumptions need to be rigorously tested and validated with empirical evidence to ensure the accuracy and reliability of the proposed measure.

The proposed methodology seems to be focused only on the increasing phase of epidemic waves, which may limit its applicability in situations where there are exponential declines in new cases. Additionally, the choice of a time horizon (n days) and the percentage of positive daily new cases to calculate the wave indicator may have a significant impact on the results and interpretation of the Bayes factor. Therefore, the authors should provide a thorough sensitivity analysis to assess the robustness of their findings to changes in these parameters.

 

Results:

The manuscript presents an interesting approach to identifying epidemic waves using a Bayes-factor-based indicator. However, it is important to note that the approach relies solely on daily case counts as the input data, which may not capture the full picture of the epidemic. Other factors, such as changes in testing capacity, reporting methods, and population behavior, may influence the daily case counts and therefore impact the accuracy of the identified epidemic waves.

The manuscript presents the results of applying the proposed indicator to three countries: the United States, the United Kingdom, and Germany. However, it is unclear how generalizable the findings are to other countries or regions. The characteristics of the epidemics, such as the timing and intensity of the waves, may differ across countries due to various factors, including population demographics, healthcare systems, and government policies. Therefore, it would be valuable to test the proposed indicator in other settings to assess its generalizability and robustness.

Discussion:

A limitation of the proposed approach is that it relies on the number of reported cases, which can be subject to inconsistencies due to variation in reporting and testing strategies. The authors acknowledge this limitation, but it deserves further exploration. Future studies may consider alternative data sources and more complex models to account for potential biases in the reported data.

 

The proposed measure has potential applications for predicting epidemic waves and assessing their agreement with similar measures, such as the average of the effective reproduction number R across a defined period of time. However, the manuscript does not provide enough evidence to support its use for prediction or to compare its performance with other measures. Future studies may consider using a larger and more diverse dataset to evaluate the robustness and generalizability of the proposed approach. Additionally, the manuscript could benefit from a more detailed discussion of the assumptions and limitations of the proposed measure.

The manuscript contains several grammatical and syntax errors that need to be addressed. Additionally, some of the sentences are too long and convoluted, making the writing difficult to follow. I would suggest that the authors revise the manuscript for clarity and conciseness.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The manuscript has been in better form.