Article activity feed

  1. Author Response

    Reviewer #1 (Public Review):

    COVID-19 epidemic conditions are rapidly changing due to behavioral changes, accumulating immunity from prior infections, vaccination roll-outs, and the emergence of new variants. In this analysis, the authors are using a simple mathematical model to reconstruct SARS-CoV-2 transmission dynamics in South Africa through different outbreaks with different prevalent variants. They estimate key characteristics of the epidemic in each of the nine South African provinces while accounting for multiple factors including changing detection rates, seasonality, nonpharmaceutical interventions, and vaccination. The paper is well written and addresses important questions in the field.

    The authors apply a model-inference system to estimate the background population characteristics (e.g., population susceptibility) before the emergence of the new variant, as well as changes in population susceptibility and transmissibility due to the new variant. They come up with projections of cumulative incidence, accumulation, and loss of population immunity over time for different provinces. Inference on the characteristics of different variants is also presented.

    The paper has a couple of key limitations.

    First, simple models come with strong assumptions. The simplicity of the model does not allow to account for several important epidemic drivers including i) heterogeneity in contactness, acquisition risk, and severity (especially with respect to age) which may have a strong impact on the epidemic dynamics; ii) all-or-nothing vaccine which restricts the possible mechanisms of protection to be explored and iii) using the same compartment for vaccinated and recovered from infection which leads to the same duration of immunity and efficacy for these 2 groups. Second, I suspect that the model-inference system has some identifiability issues. It is unclear how it selects between scenarios with low transmissibility but high IFR and scenarios with high transmissibility but low IFR. Some characteristics (including IFR) were estimated independently for each wave and each province. However, correlations across provinces should be expected. The paper will benefit from a more detailed explanation and sensitivity analyses that show how model assumptions influence presented results.

    We thank the reviewer for the comments and suggestions. We agree that model assumptions can affect model estimations. The model used here simulates a single age group, which, when used alone, likely would not be able to capture the heterogeneity in contact rates, acquisition risk, and severity (especially with respect to age). However, a key difference in this study is that the model is used in conjunction with a statistical inference method, i.e. the Ensemble Adjustment Kalman Filter (EAKF), and multiple data streams (i.e., cases, deaths, mobility, vaccination, and weather data). The combined model-inference system (i.e., the model, data, and the filter) enables estimation of short-term dynamics (e.g., changes in IFR due to more infection in older age groups) during each given time step (here, each week).

    Indeed, we have used model-generated synthetic data, for which the true parameters are known, to test a similar model-inference system and shown that it is able accurately estimate the underlying parameters as well as overall variant epidemiological quantities (i.e. immune erosion and change in transmissibility; see details in Yang & Shaman 2021 Nature Communications 12:5573). For this study, we additionally validate the model-inference estimates using three independent data streams (i.e., serology, hospitalization, and excess mortality data) and retrospective predictions (see “Model fit and validation” in Results of the main text). Further, when presenting model results for Gauteng and overall estimates for all nine South African provinces, we compare our model-inference estimates with available estimates in the literature; the consistency provides further support of the study findings (see the remaining Results sections).

    Per the reviewer suggestion, in this revision, we have added more detailed explanation when presenting the estimates (e.g. population susceptibility and variant transmissibility). See e.g., Lines 129 – 138 and 186 – 200 in the main text. We have also added further discussion of the model-inference method (e.g. choice of prior range and diagnosis) in the Appendix 1. In addition, we have added sensitivity analyses, in particular for the infection-detection rate in Gauteng during the Omicron wave, to show how model assumptions influence presented results. We have also plotted and shown the weekly estimates for all parameters included in our model-inference system (Appendix 1-figures 15 -23). Visual inspection of these estimates indicates that posterior estimates for the model parameters are consistent with those reported in the literature, or changed over time and/or across provinces in directions as would be expected. Please see these supporting results in the new Appendix 1.

    Reviewer #2 (Public Review):

    CoVID models have, by necessity, exploded in complexity over the last year. The emergence of new variants with differential spread, the waxing and waning of population immunity, and the constant changes in reporting rates all seem to necessitate the addition of new internal model states and parameters. In the present study, Yang and Shaman have developed a robust methodology that can account for each of these complexities and applied it to reconstruct the first four waves of infections in each province of South Africa. Specifically, they employ an SEIR model with time-varying parameters estimated using a Kalman Filter. Although the model does not explicitly incorporate details such as the waning of immunity, it is present implicitly in the time-varying "population susceptibility" parameter. The authors validate their estimates of infection and CoVID-related death rates over time using seroprevalence, hospitalizations, and excess deaths, which were not used to calibrate the model. Furthermore, they have shown their model's ability to predict the course of waves that have already begun using retrospective predictions of past waves.

    Despite the validity of these methods, it is not clear what conclusions can be drawn. The authors claim that their analysis shows that 1) new waves of infection are still possible, 2) large new waves of deaths can still occur, and 3) any new variant likely requires a loss of pre-existing immunity. Unfortunately, it is not clear how the modeling analysis presented supports these ideas. All three of these conclusions involve the emergence of new variants, something which the model may not be suited for. The transmissibility of new variants has been trending upwards, according to their analysis, suggesting that invasion is a combination of increased transmission and increased loss of immunity. Finally, the Delta wave was not accompanied by a large increase in susceptibility and instead appears to largely have been driven by seasonal fluctuation and increased transmissibility.

    Overall, this work should be of great interest to those modeling CoVID or seeking to understand the history of the epidemic in South Africa.

    We thank the reviewer for the comments. First, regarding waning immunity, the SEIRSV model used here did account for waning immunity, via the term R/Lt in Eqn 1, where R is number the recovered/immune individuals and Lt is the immunity period. This is briefly described in Lines 333 - 337 (grouped under “Virus-specific properties”). To clarify further, we have added a brief note: “Of note, the immunity period Lt and the term R/Lt in Eqn 1 are used to model the waning of immune protection against infection.”

    Second, the main conclusions and findings of this study are the model-inference estimates for the three SARS-CoV-2 variants of concern (i.e. Beta, Delta, and Omicron), as well as the inferenced underlying dynamics. The three general observations we made in the initial submission are related to the SARS-CoV-2 dynamics observed in South Africa, as well as in other places. We have now revised the text to clarify this and provide more direct evidence drawn from specific findings here to support the discussed observations (see Lines 252 - 271).

    Reviewer #3 (Public Review):

    Overall, the authors sought to explain the epidemiological, behavioral, and immunological underpinnings across multiple COVID-19 waves in South Africa using an infectious disease model and statistical framework. In doing so, they hoped to learn about the different emerging variant properties and provide a modeling framework for understanding risk upon future variant emergence.

    Strengths:

    The manuscript uses an epidemiological and statistical modeling framework that has been validated across a number of different diseases, time periods, and regions.

    The researchers have validated their modeling results using multiple separate lines of evidence and data including laboratory results, seroprevalence, forecasting, and other epidemiological studies.

    While not independent from one another, agreement across multiple regions within South Africa enhances the confidence in modeling results

    Weaknesses:

    The model complexity adds some opaqueness to the results due to the presence of many hidden parameters and potential correlations and interactions between them, so I suggest that the authors further validate the convergence of their model fitting and visualize the results of their hidden parameters.

    Conclusions justified:

    Overall I believe the conclusions the authors have provided are justified by their analysis. It appears their analysis is statistically rigorous, and there are multiple independent lines of evidence that agree with and validate their conclusions.

    We thank the reviewer for the comments and suggestions. In response, we have added plots to show estimates for all parameters (see Appendix 1-figures 15 – 23), in addition to those shown in the initial submission. We have also added a brief note in the main text on these results (see Lines 62-65). As the focus of this study is general COVID-19 dynamics and the epidemiological properties of SARS-CoV-2 variants of concern, we present the main estimates (e.g. population susceptibility, transmissibility, infection-detection rate, infection-fatality risk) in the main text, and provide the additional results for the supporting parameters (e.g. latent period) in the Appendix 1.

    Read the original source
    Was this evaluation helpful?
  2. Evaluation Summary:

    This paper presents a modeling framework that can be used to track the complex behavioral and immunological landscape of the COVID-19 pandemic over multiple surges and variants in South Africa, which has been validated previously for other regions and time periods. This work may be useful for infectious disease modelers, epidemiologists, and public health officials as they navigate the next phase of the pandemic or seek to understand the history of the epidemic in South Africa.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

    Read the original source
    Was this evaluation helpful?
  3. Reviewer #1 (Public Review):

    COVID-19 epidemic conditions are rapidly changing due to behavioral changes, accumulating immunity from prior infections, vaccination roll-outs, and the emergence of new variants. In this analysis, the authors are using a simple mathematical model to reconstruct SARS-CoV-2 transmission dynamics in South Africa through different outbreaks with different prevalent variants. They estimate key characteristics of the epidemic in each of the nine South African provinces while accounting for multiple factors including changing detection rates, seasonality, nonpharmaceutical interventions, and vaccination. The paper is well written and addresses important questions in the field.

    The authors apply a model-inference system to estimate the background population characteristics (e.g., population susceptibility) before the emergence of the new variant, as well as changes in population susceptibility and transmissibility due to the new variant. They come up with projections of cumulative incidence, accumulation, and loss of population immunity over time for different provinces. Inference on the characteristics of different variants is also presented.

    The paper has a couple of key limitations.
    First, simple models come with strong assumptions. The simplicity of the model does not allow to account for several important epidemic drivers including i) heterogeneity in contactness, acquisition risk, and severity (especially with respect to age) which may have a strong impact on the epidemic dynamics; ii) all-or-nothing vaccine which restricts the possible mechanisms of protection to be explored and iii) using the same compartment for vaccinated and recovered from infection which leads to the same duration of immunity and efficacy for these 2 groups.
    Second, I suspect that the model-inference system has some identifiability issues. It is unclear how it selects between scenarios with low transmissibility but high IFR and scenarios with high transmissibility but low IFR. Some characteristics (including IFR) were estimated independently for each wave and each province. However, correlations across provinces should be expected.
    The paper will benefit from a more detailed explanation and sensitivity analyses that show how model assumptions influence presented results.

    Read the original source
    Was this evaluation helpful?
  4. Reviewer #2 (Public Review):

    CoVID models have, by necessity, exploded in complexity over the last year. The emergence of new variants with differential spread, the waxing and waning of population immunity, and the constant changes in reporting rates all seem to necessitate the addition of new internal model states and parameters. In the present study, Yang and Shaman have developed a robust methodology that can account for each of these complexities and applied it to reconstruct the first four waves of infections in each province of South Africa. Specifically, they employ an SEIR model with time-varying parameters estimated using a Kalman Filter. Although the model does not explicitly incorporate details such as the waning of immunity, it is present implicitly in the time-varying "population susceptibility" parameter. The authors validate their estimates of infection and CoVID-related death rates over time using seroprevalence, hospitalizations, and excess deaths, which were not used to calibrate the model. Furthermore, they have shown their model's ability to predict the course of waves that have already begun using retrospective predictions of past waves.

    Despite the validity of these methods, it is not clear what conclusions can be drawn. The authors claim that their analysis shows that 1) new waves of infection are still possible, 2) large new waves of deaths can still occur, and 3) any new variant likely requires a loss of pre-existing immunity. Unfortunately, it is not clear how the modeling analysis presented supports these ideas. All three of these conclusions involve the emergence of new variants, something which the model may not be suited for. The transmissibility of new variants has been trending upwards, according to their analysis, suggesting that invasion is a combination of increased transmission and increased loss of immunity. Finally, the Delta wave was not accompanied by a large increase in susceptibility and instead appears to largely have been driven by seasonal fluctuation and increased transmissibility.

    Overall, this work should be of great interest to those modeling CoVID or seeking to understand the history of the epidemic in South Africa.

    Read the original source
    Was this evaluation helpful?
  5. Reviewer #3 (Public Review):

    Overall, the authors sought to explain the epidemiological, behavioral, and immunological underpinnings across multiple COVID-19 waves in South Africa using an infectious disease model and statistical framework. In doing so, they hoped to learn about the different emerging variant properties and provide a modeling framework for understanding risk upon future variant emergence.

    Strengths:
    The manuscript uses an epidemiological and statistical modeling framework that has been validated across a number of different diseases, time periods, and regions.
    The researchers have validated their modeling results using multiple separate lines of evidence and data including laboratory results, seroprevalence, forecasting, and other epidemiological studies.
    While not independent from one another, agreement across multiple regions within South Africa enhances the confidence in modeling results

    Weaknesses:
    The model complexity adds some opaqueness to the results due to the presence of many hidden parameters and potential correlations and interactions between them, so I suggest that the authors further validate the convergence of their model fitting and visualize the results of their hidden parameters.

    Conclusions justified:
    Overall I believe the conclusions the authors have provided are justified by their analysis. It appears their analysis is statistically rigorous, and there are multiple independent lines of evidence that agree with and validate their conclusions.

    Read the original source
    Was this evaluation helpful?
  6. SciScore for 10.1101/2021.12.19.21268073: (What is this?)

    Please note, not all rigor criteria are appropriate for all manuscripts.

    Table 1: Rigor

    NIH rigor criteria are not applicable to paper type.

    Table 2: Resources

    No key resources detected.


    Results from OddPub: Thank you for sharing your code and data.


    Results from LimitationRecognizer: An explicit section about the limitations of the techniques employed in this study was not found. We encourage authors to address study limitations.

    Results from TrialIdentifier: No clinical trial numbers were referenced.


    Results from Barzooka: We did not find any issues relating to the usage of bar graphs.


    Results from JetFighter: We did not find any issues relating to colormaps.


    Results from rtransparent:
    • Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
    • Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
    • No protocol registration statement was detected.

    Results from scite Reference Check: We found no unreliable references.


    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

    Read the original source
    Was this evaluation helpful?