Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

I think this paper presents first evidence of the association between monthly average temperature and stroke and myocardial infarction mortality in tropical climate. Although residents in tropical climate may have been acclimatized to high temperature year round, small increase in average temperatures can still lead to increases in mortality, especially among vulnerable population. The findings in this paper provides good evidence for policy makers to consider in anticipation of higher temperatures in this Century. Overall, the study was thoughtfully designed and the manuscript was clearly written. I only have a few minor suggestions as follows:

Line 49-50: "Although global cardiovascular 49
mortality has been rising, earlier evidence up to 2013 did not reveal an association 50
between temperature and cardiovascular mortality in the region" - needs a citation here. 

2.1 Study area: I suggest using a map to show the actual study are, as most readers are not familiar with the geography in Brazil. 

Line 104: What did the "mean temperature" refer to here? Mean daily, monthly or yearly? Please clarify. 

I am not familiar with the statistical model employed for data analysis in this study, so I defer to other reviewers who may be able to provide a more informed assessment.

Author Response

Comment 1: I think this paper presents first evidence of the association between monthly average temperature and stroke and myocardial infarction mortality in tropical climate. Although residents in tropical climate may have been acclimatized to high temperature year round, small increase in average temperatures can still lead to increases in mortality, especially among vulnerable population. The findings in this paper provides good evidence for policy makers to consider in anticipation of higher temperatures in this Century. Overall, the study was thoughtfully designed and the manuscript was clearly written. I only have a few minor suggestions as follows:

Response 1: We thank the reviewer for the positive assessment of our study and for highlighting its relevance to understanding temperature-related cardiovascular mortality in tropical climates.

 

Comment 2: Line 49-50: "Although global cardiovascular 49 mortality has been rising, earlier evidence up to 2013 did not reveal an association 50 between temperature and cardiovascular mortality in the region" - needs a citation here.

Response 2: Thank you for this observation. We have added an appropriate citation to support this statement, which reports the absence of an association between temperature and cardiovascular mortality in northern Brazil during the period up to 2013. See line 51.

 

Comment 3: 2.1 Study area: I suggest using a map to show the actual study are, as most readers are not familiar with the geography in Brazil.

Response 3: We agree with the reviewer. A map of the Brazilian Amazon indicating the study region has been added to Section 2.1 (now Figure 1). This improves geographic contextualization for readers unfamiliar with the region.

 

Comment 4: Line 104: What did the "mean temperature" refer to here? Mean daily, monthly or yearly? Please clarify.

Response 4: We have clarified that “mean temperature” refers specifically to annual mean temperature. We included this information in the text (see line 120).

 

Comment 5: I am not familiar with the statistical model employed for data analysis in this study, so I defer to other reviewers who may be able to provide a more informed assessment.

Response 5: We appreciate the reviewer’s comment and acknowledge that additional reviewers may offer further insight on the statistical modeling. The methods were described in detail to ensure transparency and reproducibility.

Reviewer 2 Report

Comments and Suggestions for Authors

dear  authors

 I accept your manuscript as it is, with minor considerations.

line 65 Please clarify which cities the study where made, or States. And give the readers the total population of those specific areas.

Line 104, add 'annual' before 'mean'

figure 4 the legend must be self contained. 90% percentil means how much 29C for stroke and 30C for myocardial? All the values are monthly means?

why April of the year 2020 is so high? 

The overall result is a newty for tropical/equatorial regions. The authors addressed them correctly.

 

 

 

Comments for author File: Comments.pdf

Author Response

Comment 1: I accept your manuscript as it is, with minor considerations.

Response 1: We thank the reviewer for the positive evaluation and for considering the manuscript acceptable pending minor revisions.

 

Comment 2: Line 65 Please clarify which cities the study where made, or States. And give the readers the total population of those specific areas.

Response 2: We have clarified that the analysis covers all municipalities in the Brazilian Amazon region, distributed across the states of Acre, Amapá, Amazonas, Pará, Rondônia, Roraima, Tocantins, and the northern portion of Mato Grosso. We also added the total population of the study area based on the most recent IBGE estimates. See lines 62 to 65.

 

Comment 3: Line 104, add 'annual' before 'mean'

Response 3: The text has been revised to specify “annual mean,” improving clarity. See line 120.

 

Comment 4: Figure 4 the legend must be self contained. 90% percentil means how much 29C for stroke and 30C for myocardial? All the values are monthly means?

Response 4: We revised the legend so it is fully self-contained. It now specifies the exact temperature values corresponding to the 90th percentile (29°C for stroke; 30°C for myocardial infarction) and states explicitly that all values represent monthly mean temperatures. See legends for Figures 5 and 6.

 

Comment 5: Why April of the year 2020 is so high?

Response 5: We added a paragraph to the Discussion clarifying that periods of elevated mortality, exemplified by April 2020, may be associated with the early overlap between the COVID-19 pandemic and limited healthcare capacity in the region. The revised text discusses how healthcare disruption, delayed emergency care, and other concurrent stressors may have contributed to increased deaths from acute cardiovascular events, while emphasizing that temperature patterns remained strongly associated with mortality risk. See lines 194 to 205.

 

Comment 6: The overall result is a newty for tropical/equatorial regions. The authors addressed them correctly.

Response 6: We thank the reviewer for the positive feedback on the novelty and interpretation of the results.

Reviewer 3 Report

Comments and Suggestions for Authors

This article examines the association between temperature and mortality due to stroke and myocardial infarction in the Amazonian population of Brazil. The subject is original due to the unique living conditions and climate of this region.

However, the results are not convincing. The following major issues have been identified:

• The observed increase in mortality for both causes over the study period is substantial, even before the onset of months with extreme heat. This suggests either a major and growing health problem independent of temperature since 2000, or a significant measurement bias related to a possible improvement in the mortality registration system. The latter hypothesis appears to be favored by the authors (line 154, page 5, in the discussion). It is therefore highly likely that the temperature-mortality association is overestimated due to this concurrent improvement in registration, which coincides with the rise in extreme temperature events. The analysis should be restricted to a period when the registration system is considered reliable. This period cannot be identified based solely on the results presented. If data completeness is better, I recommend analyzing all-cause mortality rather than specific causes.
• Another approach is to assume that the improvement in the registration system is gradual and not related to intra-annual variations. In this case, it is recommended to adjust for a long-term temporal trend in the regression model, which does not appear to have been done here. Figure 2 should also adjust the estimates for this temporal trend.
• It is unusual to study the effects of heat at the monthly scale, as effects are generally observed within 2–3 days following temperature increases. Can the authors obtain more precise data?
• Figure 5 should be enhanced with confidence intervals to identify the statistical significance of the observed increases in mortality.

In conclusion, the article requires substantial revision before it can be reconsidered.

Comments for author File: Comments.pdf

Author Response

Comment 1: This article examines the association between temperature and mortality due to stroke and myocardial infarction in the Amazonian population of Brazil. The subject is original due to the unique living conditions and climate of this region.

Response 1: We thank the reviewer for recognizing the originality of the study and the relevance of examining temperature-related cardiovascular mortality in the Amazonian population, given the region’s unique climatic conditions and living environments.

 

Comment 2: However, the results are not convincing. The following major issues have been identified: The observed increase in mortality for both causes over the study period is substantial, even before the onset of months with extreme heat. This suggests either a major and growing health problem independent of temperature since 2000, or a significant measurement bias related to a possible improvement in the mortality registration system. The latter hypothesis appears to be favored by the authors (line 154, page 5, in the discussion). It is therefore highly likely that the temperature-mortality association is overestimated due to this concurrent improvement in registration, which coincides with the rise in extreme temperature events. The analysis should be restricted to a period when the registration system is considered reliable. This period cannot be identified based solely on the results presented. If data completeness is better, I recommend analyzing all-cause mortality rather than specific causes.

Response 2: We thank the reviewer for this important concern. We acknowledge that long-term increases in stroke and myocardial infarction mortality reflect non-thermal factors, including demographic aging and gradual improvements in mortality registration, and we clarified that these trends should not be attributed to temperature alone (see lines 178 to 184). Our analyses focus on intra-annual and interannual variability and adjust for long-term temporal trends and seasonality (see lines 90 to 94 and 98 to 99). To further address potential bias related to data completeness, we conducted sensitivity analyses excluding extreme years and simulating possible under- and overreporting (see lines 110 to 114 and 155 to 158). Although all-cause mortality and period restriction are valid alternatives, gradual improvements in the registration system make it difficult to define a precise cutoff, and cause-specific outcomes remain relevant for assessing cardiovascular sensitivity to heat.

 

Comment 3: Another approach is to assume that the improvement in the registration system is gradual and not related to intra-annual variations. In this case, it is recommended to adjust for a long-term temporal trend in the regression model, which does not appear to have been done here. Figure 2 should also adjust the estimates for this temporal trend.

Response 3: We appreciate this methodological suggestion. All models were adjusted for long-term temporal trends and seasonality to separate gradual changes in mortality from short-term and interannual temperature effects (see lines 90 to 94).

 

Comment 4: It is unusual to study the effects of heat at the monthly scale, as effects are generally observed within 2–3 days following temperature increases. Can the authors obtain more precise data?

Response 4: We thank the reviewer for raising this point. Although heat effects are often assessed at daily scales, monthly analyses are commonly used when working with population-based mortality data in settings where finer temporal resolution is not consistently available. In Brazil, mortality data are officially disseminated at monthly resolution, which constrains temporal granularity. Monthly mean temperature reflects sustained thermal exposure and cumulative heat stress, which are particularly relevant in tropical regions characterized by prolonged high temperatures.

 

Comment 5: Figure 5 should be enhanced with confidence intervals to identify the statistical significance of the observed increases in mortality.

Response 5: We thank the reviewer for this suggestion. Figures 5 and 6 were revised to include 95% confidence intervals, allowing clearer identification of the statistical significance and uncertainty associated with the observed increases in mortality.

 

Comment 6: In conclusion, the article requires substantial revision before it can be reconsidered.

Response 6: We appreciate the reviewer’s positive assessment. We agree that the findings provide novel evidence for tropical regions, and we have further refined the manuscript to strengthen the interpretation and clarity of the results in line with the reviewer’s comments.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for revising the manuscript in response to my comments.

However, I do not see any changes in the Results section. Does this mean that my suggestion to adjust for long-term variations was already implemented in the previous version of the results? If so, I have the following questions and recommendations:

1. Model specification: The model used needs to be described more clearly. In particular, how was the long-term trend introduced?
2. Figure 2: This figure should be adjusted to reflect the long-term trend.
3. Sensitivity analyses: It remains unclear how the simulations accounted for a possible 10% variation in the reporting rate. Over what period was this correction applied?

Author Response

Comment 1: Thank you for revising the manuscript in response to my comments. However, I do not see any changes in the Results section. Does this mean that my suggestion to adjust for long-term variations was already implemented in the previous version of the results? If so, I have the following questions and recommendations:

Response 1: Thank you for this clarification request. Adjustment for long-term temporal variation was already incorporated in the modeling framework and therefore did not require changes to the Results. All models explicitly control for long-term trends and seasonality, ensuring that the reported associations capture intra-annual and interannual temperature variability rather than mortality trends. To improve clarity, we have further clarified this point in the Methods (lines 90 to 95 and 100 to 103) and Results (lines 129 to 134) sections.

 

Comment 2: Model specification: The model used needs to be described more clearly. In particular, how was the long-term trend introduced?

Response 2: We thank the reviewer for this comment. Long-term trends were explicitly controlled in the models. In the ARIMAX framework, trends and autocorrelation were captured by the autoregressive and moving-average components selected via auto.arima. In the DLNM analyses, long-term trends and seasonality were adjusted using a smooth function of time (natural splines). This specification allowed us to isolate temperature-related effects from underlying long-term mortality changes. The Methods section has been revised to clarify these points. See lines 90 to 95 and 100 to 103.

 

Comment 3: Figure 2: This figure should be adjusted to reflect the long-term trend.

Response 3: We thank the reviewer for this comment. Figure 2 presents observed mortality trends and age-specific patterns for descriptive purposes. The long-term temporal trend was explicitly accounted for in the statistical models, as described in the Methods (lines 90 to 95) and reflected in the Results (lines 128 and 131). We also revised the figure legend (lines 140 to 145) to clarify that temperature-mortality associations were estimated from models adjusted for long-term trends and seasonality.

 

Comment 4: Sensitivity analyses: It remains unclear how the simulations accounted for a possible 10% variation in the reporting rate. Over what period was this correction applied?

Response 4: We thank the reviewer for this question. The ±10% variation was applied uniformly across the entire study period as a global sensitivity test, rather than as a time-specific correction. Mortality counts were systematically increased and decreased by 10% for all months to simulate potential under- and overreporting in the registration system. This approach was intended to assess whether the temperature-mortality associations were robust to plausible levels of reporting uncertainty, independent of any specific year or period. The consistency of the results under these scenarios indicates that the observed associations are not driven by moderate variations in reporting rates. See lines 117 to 121.