Disparities in Geographical Access to Hospitals in Portugal
Round 1
Reviewer 1 Report
This a well written article addressing an important topic: the differences in the geographical access to hospitals in Portugal using a population driving-time indicator over two decades. Nevertheless, I have some doubts about the relevance of this paper regarding the methodological novelty although it presents new information for the study area. I make here some remarks which will hopefully help the authors to improve their manuscript before being considered for publication after a major review:
Title
- The concept of “disparities/inequalities” is more commonly used when the study of social and economic characteristics is involved. However, this is not the case. I confess I was expecting to find the analysis of these concepts in the paper.
- Shouldn’t hospital be in plural?
Introduction
- I understand that this is an important aspect to study in any country/region and new information is presented for Portugal. However, these methods have been applied in the past as recognised and referenced by the authors. In lines 11-15 there are some sentences starting with the word “still”. I am not sure if these are the gaps being addressed in this paper. It should be clearly stated what is the original contribution of the paper when compared to previous studies. The contribution should not be deducted by the reader but stated by the authors. This is a very important aspect of the paper which needs clarification.
Methods
- Line 154 Tracts instead of tracks, there are other cases in the document.
- Population driving time should have a reference.
Results
- Why did the census tracts go from 91,615 in 1991 to 232,625 in 2011? I do not understand these numbers.
- In figure 3, hospitals are hardly visible.
Discussion
- Is figure 7 necessary? I think it does not add much to the discussion.
- 5 is about the relevance of the indicator used for the specific case of Portugal according to its characteristics. Would this approach be recommended for other cases? Is it generalisable? This should be connected to the relevance of the paper in the introduction which I commented previously.
- One limitation that could be referred in 4.6 is the MAUP. This analysis relies on census tracts, however, it would be more interesting to use dasymetric mapping to better consider the location of populations as done in previous studies for other locations. This is particularly important in Alentejo.
Conclusions
- In the last sentence of the conclusions is stated that this method is a step forward in terms of spatial health care accessibility. However, measuring accessibility using time has been done many times before so for me this statement is not clear. Also the term “accuracy” is mentioned in line 525. How does this method increase accuracy? These are statements that are not supported on the results.
Author Response
Dear Reviewer,
We would like to express our sincere thanks to you for the constructive and positive comments. The manuscript has been fully revised accordingly and changes were highlighted in blue. The following is a point-to-point response to your comments.
--------------------------------------------------------------------------------------------
Comments and Suggestions for Authors
This a well written article addressing an important topic: the differences in the geographical access to hospitals in Portugal using a population driving-time indicator over two decades. Nevertheless, I have some doubts about the relevance of this paper regarding the methodological novelty although it presents new information for the study area. I make here some remarks which will hopefully help the authors to improve their manuscript before being considered for publication after a major review:
Title
- The concept of “disparities/inequalities” is more commonly used when the study of social and economic characteristics is involved. However, this is not the case. I confess I was expecting to find the analysis of these concepts in the paper.
Response:
The concept of health disparities that shapes this paper was published by Paula Braveman in 2006: “A health disparity/inequality is a particular type of difference in health or in the most important influences on health that could potentially be shaped by policies” (https://www.annualreviews.org/doi/pdf/10.1146/ annurev.publhealth.27.021405.102103).
On the manuscript, we clarified this concept on the manuscript (line 103-104), the discussion is based on the increasing vulnerability of the population associated with weak to bad geographical accessibility to reach the hospital (line 339). The analysis of the social characteristics of those living on the municipalities was strengthened, focusing on the association between population-weighted driving time, share of older population, population density and premature mortality (lines 377-390).
- Shouldn’t hospital be in plural?
Response:
Thank you for the suggestion. We made the change accordingly.
Introduction
- I understand that this is an important aspect to study in any country/region and new information is presented for Portugal. However, these methods have been applied in the past as recognised and referenced by the authors. In lines 11-15 there are some sentences starting with the word “still”. I am not sure if these are the gaps being addressed in this paper. It should be clearly stated what is the original contribution of the paper when compared to previous studies. The contribution should not be deducted by the reader but stated by the authors. This is a very important aspect of the paper which needs clarification.
Response:
Thank you for the suggestion. We included information on lines 114-132 to explain why this study is important for Portugal and other peripheric countries and to highlight the original contribution when compared to previous ones. Regarding the abstract, we reformulate it in order to clarify where is important to act in order to reduce the gaps on geographical accessibility (lines 22 – 26).
Methods
- Line 154 Tracts instead of tracks, there are other cases in the document.
Response:
Thank you for the suggestion. We corrected as suggested.
- Population driving time should have a reference.
Response:
Thank you for the suggestion. The references are on line 192.
Results
- Why did the census tracts go from 91,615 in 1991 to 232,625 in 2011? I do not understand these numbers.
Response:
Statistics Portugal (Portuguese National Statistics Institute) changed the census tracts, as well as their delimitation, in order to make them more coherent, in each census. Therefore, the number of census tracts increased over the census years and, simultaneously, the average area of the census tracts decreased. We include more information on table 1 to make this clearer (lines 303 - 304).
- In figure 3, hospitals are hardly visible.
Response:
Thank you for the suggestion. We changed the symbol of the hospital in order to became more clear.
Discussion
- Is figure 7 necessary? I think it does not add much to the discussion.
Response:
Thank you for the suggestion. We removed the figure. We also decided to remove figure 6 in order to exclude all the figures from the discussion.
- 5 is about the relevance of the indicator used for the specific case of Portugal according to its characteristics. Would this approach be recommended for other cases? Is it generalisable? This should be connected to the relevance of the paper in the introduction which I commented previously.
Response:
Thank you for the suggestion. We rewrite the conclusion to include the relevance of this paper and why this method can be recommended to other studies (lines 658 - 672).
- One limitation that could be referred in 4.6 is the MAUP. This analysis relies on census tracts, however, it would be more interesting to use dasymetric mapping to better consider the location of populations as done in previous studies for other locations. This is particularly important in Alentejo.
Response:
Thank you for the suggestion. We agree that it would be more interesting to use dasymetric data to better consider the location of the population. However, it was not possible to have data with quality for 1991 and 2001. We include a reference to this on the limitations section (lines 639 - 640). Nonetheless, considering the size of the census tracts (average: 0.3km2), it is our belief that this level already provides enough detail to apply the methodology and produce adequate evidence. We include this on the manuscript (line 300 and 303 - 304).
Conclusions
- In the last sentence of the conclusions is stated that this method is a step forward in terms of spatial health care accessibility. However, measuring accessibility using time has been done many times before so for me this statement is not clear. Also the term “accuracy” is mentioned in line 525. How does this method increase accuracy? These are statements that are not supported on the results.
Response:
Thank you very much for noticing this. We remove the term “accuracy”. It is our belief that this work is a step forward on how to communicate geographic accessibility and on the extreme importance to critically analyse the area of the administrative area being evaluated before considering the centroid as representative of the geographical accessibility. We clarified this on the manuscript (lines 663 – 672).
Reviewer 2 Report
The aim of the paper is to assess differences in hospital’s accessibility for residential population in Portugal using a population-weighted driving time measurement. The supplementary web application (saudemunicipio.uc.pt) seems to be useful for local/regional authorities. The methodology is illustratively explained and extended discussion about health care services and plans is provided but the methodology is really simple. The behaviour of the indicator should be more tested or compared with alternatives/previous approaches of hospital accessibility evaluation. The discussion of geographical factors is insufficient and evaluation of different impacts to local conditions is absent. It is necessary to improve it to better fit into the journal profile. Also the structure of the paper should be improved – some parts are more appropriate at the beginning of the paper than at the end. More detailed comments follow.
Comments:
1991-2011 = 20, and not 30 years.
The transport mode has to be explained in the abstract and the introduction.
Specify what kind of hospitals were used in the analysis (see 287, there are 3 levels of hospitals).
66-115 A review provided in the introduction has to be extended and deepened. E.g. for classification of accessibility measures following sources are recommended:
- Geurs, K.T.; van Wee, B. Accessibility evaluation of land-use and transport strategies: review and research directions. J. Transp. Geogr. 2004, 12, 127–140.
- El-Geneidy, M., A., Levinson, D., M., 2006. Access to Destinations: Development of Accessibility Measures. Report #1 in the series Access to destinations study. Minnesota Department of Transportation.
- Bhat, Ch., Handy, S., Kockelman, K., Mahmassani, H., Chen, Q., Weston, L., 2000. Urban Accessibility Index: Literature Review. Center for Transportation Research, the University of Texas at Austin, 92p.
- Geuers, K., T., Van Eck, JR., R., 2001. Accessibility Measures: Review and Applications. RIMV – Research for Man and Environment; Report 408 505 006, Utrecht: National Institute of Public Health and the Environment. Utrecht University, 2001.
104 A fifth parameter was not mentioned at 70 (only 4 parameters)
143, 177 „an influence area“ is an unusual term; consider to use „catchment“
157 „time from the centroid of each small-area level (census track) to the hospital’s influence area of the small-area;“ – better to „reference hospital“
162 mistakes in indexes within Eq.1 – the sum is from i=1 and what about j?
165 „revealing the time that more than half of the population from a municipality would need to reach the hospital“ – this it not true. The calculation is based on weighted averages and not on quantiles. Try to calculate for the following testing data: 20 min for 1000, 30 min for 8000 and 150 min for 1000.
171 fig. 1 instead of 3
187 instead of a maximum speed the average speed has to be used for road network analysis
201-204 confusing interpretation of fig.2 which portrays internal variability of access time in 2011 for municipalities and not a temporal development
229 Hospital instead of Hospitals in the legend for the individual symbol. And the symbol is almost invisible in the maps.
233 an interpretation is missing (or confused with fig.2?)
247-255 the analysis and interpretation of age influence is not sufficient.
274-310 influencing factors are not satisfactory explained and, mainly, evaluation of their impact is missing. It is anticipated authors will interpret geographical differences based on analyses of each factor – e.g. what municipalities improved its access due to improvement of road network (and can you quantify this influence?), what municipalities take advantages of 2 new hospitals, where a depopulation significantly improves the situation (and quantify this impact). Also, according to l.295, some changes of catchments in the referenced time period occur – where and what is the impact?
334-336 not clear, time seems to be the only indicator
342-370 majority of the text should be moved to the introduction/state-of-the-art. It is necessary to explain at the beginning what is e.g. LHU.
379-409 such extended discussion about organisation of transboundary health services does not fit into a geographically oriented journal. It is anticipated authors provide modelling of accessibility including selected Spain hospitals, evaluate (geographically) differences and recognize (quantify) a potential for hospital accessibility improvement.
410-425 discussion about ageing and elder population should be deepened and related to territorial distribution
427-436 discussion about telehealth solutions seems to be out of focus
475 „sensitive to locally low areas of accessibility where populations live“ – the problem is in averaging - it is still equalizing differences and hiding problematic places. Alternatively, some threshold (e.g. 30 min) can be applied and sum population in these bad accessible census tracks per municipality.
492-3 „Most studies published to date consider the closest hospital to evaluate geographical accessibility in Portugal [24,53,54,73].“ - Why results of the new method are not compared with results of the previous methodology based on the closest hospitals?
504-513 and what about hospital’s capacity? What about temporary (not residential) population?
506 what about projection of population from the census’s year to the current state?
Conclusion – clear recommendations are missing. What are priorities (and where): build roads/improve transport services, build new hospitals, establish cross-border health services, change LHUs (catchments)? Or just wait for further depopulation of peripheral and rural settlements?
Author Response
Dear Reviewer,
We would like to express our sincere thanks to you for the constructive and positive comments. The manuscript has been fully revised accordingly and changes were highlighted in blue. The following is a point-to-point response to your comments.
--------------------------------------------------------------------------------------------
Comments and Suggestions for Authors
The aim of the paper is to assess differences in hospital’s accessibility for residential population in Portugal using a population-weighted driving time measurement. The supplementary web application (saudemunicipio.uc.pt) seems to be useful for local/regional authorities. The methodology is illustratively explained and extended discussion about health care services and plans is provided but the methodology is really simple. The behaviour of the indicator should be more tested or compared with alternatives/previous approaches of hospital accessibility evaluation. The discussion of geographical factors is insufficient and evaluation of different impacts to local conditions is absent. It is necessary to improve it to better fit into the journal profile. Also the structure of the paper should be improved – some parts are more appropriate at the beginning of the paper than at the end. More detailed comments follow.
Comments:
1991-2011 = 20, and not 30 years.
Response:
Thank you for the suggestion. We made the change throughout the manuscript (e.g., line 18).
The transport mode has to be explained in the abstract and the introduction.
Response:
Thank you for the suggestion. We include references to the transport mode in the abstract and methods section (e.g., line 232).
Specify what kind of hospitals were used in the analysis (see 287, there are 3 levels of hospitals).
Response:
Thank you for the suggestion. We clarified the kind of data being used (lines 281 – 292).
66-115 A review provided in the introduction has to be extended and deepened. E.g. for classification of accessibility measures following sources are recommended:
- Geurs, K.T.; van Wee, B. Accessibility evaluation of land-use and transport strategies: review and research directions. Transp. Geogr.2004, 12, 127–140.
- El-Geneidy, M., A., Levinson, D., M., 2006. Access to Destinations: Development of Accessibility Measures. Report #1 in the series Access to destinations study. Minnesota Department of Transportation.
- Bhat, Ch., Handy, S., Kockelman, K., Mahmassani, H., Chen, Q., Weston, L., 2000. Urban Accessibility Index: Literature Review. Center for Transportation Research, the University of Texas at Austin, 92p.
- Geuers, K., T., Van Eck, JR., R., 2001. Accessibility Measures: Review and Applications. RIMV – Research for Man and Environment; Report 408 505 006, Utrecht: National Institute of Public Health and the Environment. Utrecht University, 2001.
Response:
Thank you very much for the suggestions. They were very useful to extend and deepen the introduction. We include them all on the manuscript (lines 31 – 73).
104 A fifth parameter was not mentioned at 70 (only 4 parameters)
Response:
Most studies only refer the four parameters mentioned at line 70. However, we believe that a fifth parameter should be added: the outcome of the measurement. We change the sentence in order to make it more clear (line 173).
143, 177 „an influence area“ is an unusual term; consider to use „catchment“
Response:
Thank you for the suggestion. We made the change throughout the manuscript.
157 „time from the centroid of each small-area level (census track) to the hospital’s influence area of the small-area;“ – better to „reference hospital“
Response:
Thank you for the suggestion. We made the change accordingly.
162 mistakes in indexes within Eq.1 – the sum is from i=1 and what about j?
Response:
Thank you very much for the note. The formula has been corrected (line 247).
165 „revealing the time that more than half of the population from a municipality would need to reach the hospital“ – this it not true. The calculation is based on weighted averages and not on quantiles. Try to calculate for the following testing data: 20 min for 1000, 30 min for 8000 and 150 min for 1000.
Response:
Thank you very much for your question. We have rewritten this sentence (lines 240 – 241).
171 fig. 1 instead of 3
Response:
Thank you for the suggestion. We made the change accordingly.
187 instead of a maximum speed the average speed has to be used for road network analysis
Response:
We agree that using average speed would bring higher accuracy to the study. However, it is not possible to access that information from all the roads from 1991 and 2001. We included this as a limitation (line 648).
201-204 confusing interpretation of fig.2 which portrays internal variability of access time in 2011 for municipalities and not a temporal development
Response:
In order to clarify figure 2, we decided to improve the text regarding this figure and to include the same analysis for 1991 in order to support the temporal analysis (lines 310 - 318).
229 Hospital instead of Hospitals in the legend for the individual symbol. And the symbol is almost invisible in the maps.
Response:
Thank you for the suggestion. We made changes on the legend and symbol (line 331).
233 an interpretation is missing (or confused with fig.2?)
Response:
The interpretation of figure 3 is on lines 319-329. We put the figure closer to the text to facilitate the interpretation.
247-255 the analysis and interpretation of age influence is not sufficient.
Response:
Thank you for the suggestion. We decided to strength the analysis of the association between older population and population-weighted driving time to the reference hospital. Moreover, we also included other characteristics as population density and premature mortality (lines 377-389).
274-310 influencing factors are not satisfactory explained and, mainly, evaluation of their impact is missing. It is anticipated authors will interpret geographical differences based on analyses of each factor – e.g. what municipalities improved its access due to improvement of road network (and can you quantify this influence?), what municipalities take advantages of 2 new hospitals, where a depopulation significantly improves the situation (and quantify this impact). Also, according to l.295, some changes of catchments in the referenced time period occur – where and what is the impact?
Response:
Thank you for the suggestion. We decided to include a new analysis to the text to reveal the municipalities affected by changes on hospital distribution, extension of the road network and/or changes on population distribution (lines 345 – 363). Moreover, we improved the discussion by including an interpretation of the geographical differences of each influencing factor (lines 425 – 428, 433 – 434 and 440).
Regarding the quantification of the influence, it is very interesting It is something that we are already working and is planned to be part of another paper.
334-336 not clear, time seems to be the only indicator
Response:
Thank you for the suggestion. We clarified the sentence to provide evidence that conclusions are based on the results from the Population weighted driving time indicator (line 459).
342-370 majority of the text should be moved to the introduction/state-of-the-art. It is necessary to explain at the beginning what is e.g. LHU.
Response:
Thank you for the suggestion. We moved part of the text to the introduction (lines 92 – 99, 117 – 123, and 127 - 133) and we explain the LHU (line 477).
379-409 such extended discussion about organisation of transboundary health services does not fit into a geographically oriented journal. It is anticipated authors provide modelling of accessibility including selected Spain hospitals, evaluate (geographically) differences and recognize (quantify) a potential for hospital accessibility improvement.
Response:
Thank you for the suggestion. Besides producing an outcome that can influence policy and, therefore, decrease health disparities, the aim of this paper was also to describe the method and analyse it considering the criteria to better evaluate geographical accessibility. Therefore, it would only make sense to publish this on a geographically oriented publication and on a special issue linking Geographical Information Systems to Health.
We agree that perhaps the presentation of possible interventions to be implemented on the municipalities is to big. However, we wanted to give the reader information to better understand the type of intervention that we recommend.
Regarding the modelling alternatives, recently we start to work with a research team from Spain in order to replicate this method and calculate the population-weighted driving time from the Iberian Peninsula. On that paper we want to strength the analysis of the outcome and not so much on the method. Moreover, that paper will have co-authors from both teams so it would not make sense to strength the method on that paper.
410-425 discussion about ageing and elder population should be deepened and related to territorial distribution
Response:
Thank you for the suggestion. We increased the analysis on the association between time to the hospital and elderly population rate (lines 377 – 389).
427-436 discussion about telehealth solutions seems to be out of focus
Response:
Telehealth is included as a suggestion of intervention that can help vulnerable populations to have access to healthcare. We change the sentence to make it clearer (line 544-545).
475 „sensitive to locally low areas of accessibility where populations live“ – the problem is in averaging - it is still equalizing differences and hiding problematic places. Alternatively, some threshold (e.g. 30 min) can be applied and sum population in these bad accessible census tracks per municipality.
Response:
Thank you for the suggestion. The limitations of our method are referred to in the manuscript. In fact, it still has limitations, but it increases the accuracy in the calculation of accessibility because it uses the population registered in each tract census as a power factor. This way, it is very sensitive to make a comparison with other methods that do not have the same calculation option.
492-3 „Most studies published to date consider the closest hospital to evaluate geographical accessibility in Portugal [24,53,54,73].“ - Why results of the new method are not compared with results of the previous methodology based on the closest hospitals?
Response:
Thank you for the suggestion. We included a paragraph on the manuscript where we compare the results (lines 409 – 416) and we also added why most studies decided to use the closest hospital instead of the reference one (line 625 - 627).
504-513 and what about hospital’s capacity? What about temporary (not residential) population?
Response:
Thank you for the suggestion. We agree that in further studies we should consider both data regarding the hospital capacity and the weight of temporary population, so we include both ideas on the manuscript as future developments of this study (lines 644 and 647).
506 what about projection of population from the census’s year to the current state?
Response:
Statistics Portugal only calculate projections for Municipality and there is no other data source of projected population for census tracts. We agree that this would be important, so we added it as a limitation (line 639).
Conclusion – clear recommendations are missing. What are priorities (and where): build roads/improve transport services, build new hospitals, establish cross-border health services, change LHUs (catchments)? Or just wait for further depopulation of peripheral and rural settlements?
Response:
Thank you for the suggestion. On section 4.2 we highlight the municipalities where action should be taken and include recommendations on how municipalities can act to overtake the inadequate geographical accessibility. For instance, we refer the importance of strengthening the Local Health Units and the communication between the units within to support the municipalities that have a population-weighted driving time to the hospital higher than 30 minutes; changes to the cross-border healthcare directive and implementation of health interventions within the Euroregions to support the municipalities closer to the border with Spain; and adequate telehealth units on the municipalities with higher rates of elderly population (lines 463 – 554).
Reviewer 3 Report
This manuscript is addressing the temporal changes of the spatial disproportion of hospitals and their drivers. The issue and results are interesting and promising. However, it requires some improvements with more evidence.
[Major]
- The relationship between the aging ratio and hospital accessibility is impressive. The author(s) should also confirm if the variation of population density or hospitals' expansion or closure affected the accessibility.
- Can you compare your results with two other papers below?
- Doetsch, J., Pilot, E., Santana, P. et al. Potential barriers in healthcare access of the elderly population influenced by the economic crisis and the troika agreement: a qualitative case study in Lisbon, Portugal. Int J Equity Health 16, 184 (2017). https://doi.org/10.1186/s12939-017-0679-7
- Baeten, R., Spasova, S., Vanhercke, B., & Coster, S. (2018). Inequalities in access to healthcare. A study of national policies 2018. https://doi.org/10.2767/371408
[Minor]
- The expression "the last 30 years": It may give confusion to readers. It should be clear that the period of investigating data is 20 years from 1991 to 2011 (exactly, 1991, 2001, and 2011).
- NUTS 2 or 3: Briefly explain the NUTS I, II, and III boundaries in the manuscript for Non-EU readers.
Author Response
Dear Reviewer,
We would like to express our sincere thanks to you for the constructive and positive comments. The manuscript has been fully revised accordingly and changes were highlighted in blue. The following is a point-to-point response to your comments.
--------------------------------------------------------------------------------------------
Comments and Suggestions for Authors
This manuscript is addressing the temporal changes of the spatial disproportion of hospitals and their drivers. The issue and results are interesting and promising. However, it requires some improvements with more evidence.
[Major]
- The relationship between the aging ratio and hospital accessibility is impressive. The author(s) should also confirm if the variation of population density or hospitals' expansion or closure affected the accessibility.
Response:
Thank you for your acknowledgment. We included population density to the analysis to confirm that it affects accessibility since hospitals are located on the municipalities with higher population density and this haven´t changed with hospital’s expansion because the new hospitals are located on the metropolitan areas (lines 377 – 389).
- Can you compare your results with two other papers below?
- Doetsch, J., Pilot, E., Santana, P. et al.Potential barriers in healthcare access of the elderly population influenced by the economic crisis and the troika agreement: a qualitative case study in Lisbon, Portugal. Int J Equity Health 16, 184 (2017). https://doi.org/10.1186/s12939-017-0679-7
- Baeten, R., Spasova, S., Vanhercke, B., & Coster, S. (2018). Inequalities in access to healthcare. A study of national policies 2018. https://doi.org/10.2767/371408
Response:
Both papers are regarding access to healthcare on a broader sense and our focus is on geographical accessibility. Although we cannot compare results, both papers are very interesting, and we included them on the manuscript (lines 88, 115, 285, 483 and 536).
[Minor]
- The expression "the last 30 years": It may give confusion to readers. It should be clear that the period of investigating data is 20 years from 1991 to 2011 (exactly, 1991, 2001, and 2011).
Response:
Thank you for the suggestion. Throughout the manuscript we have made corrections, as suggested, to clarify that it is a 20 years period (e.g., line 395).
- NUTS 2 or 3: Briefly explain the NUTS I, II, and III boundaries in the manuscript for Non-EU readers.
Response:
Thank you for the suggestion. Since there was only one reference to the NUTS in the text, we decided to change and refer region instead (line 323).
Reviewer 4 Report
The manuscript is interesting and well-organized. I just have a few suggestions for authors to consider.
- Section 1: Whys suddenly Portugal is mentioned here? What is the uniqueness of Portugal? Please provide more context of the study and explain the importance and implications of doing this research for Portugal?
- Section 2.2: Please discuss the quality of these data. Providing some metadata would be helpful.
- Figure 2 and Figure 4: What was the spatial reference used? Where is the reference hospital?
- Section 4: Please discuss your analysis and results based on the four parameters (a framework) that you mentioned in the introduction section.
Author Response
Dear Reviewer,
We would like to express our sincere thanks to you for the constructive and positive comments. The manuscript has been fully revised accordingly and changes were highlighted in blue. The following is a point-to-point response to your comments.
--------------------------------------------------------------------------------------------
Comments and Suggestions for Authors
The manuscript is interesting and well-organized. I just have a few suggestions for authors to consider.
- Section 1: Whys suddenly Portugal is mentioned here? What is the uniqueness of Portugal? Please provide more context of the study and explain the importance and implications of doing this research for Portugal?
Response:
Thank you for the suggestion. We remove some text that was on discussion to the introduction. This way we were able to explain the importance of this study and why we focused on Portuguese Municipalities (lines 114 – 132).
- Section 2.2: Please discuss the quality of these data. Providing some metadata would be helpful.
Response:
Thank you for the suggestion. We included information regarding the quality of the data being used on section 2.3 – Geographical data (lines 252 – 295).
- Figure 2 and Figure 4: What was the spatial reference used? Where is the reference hospital?
Response:
Thank you for the suggestion. We included the location of the reference hospitals on all maps included on the manuscript. Moreover, we clarify the legend of figures 2 and 4 to clarify that the spatial reference is the municipality (lines 343 - 344).
- Section 4: Please discuss your analysis and results based on the four parameters (a framework) that you mentioned in the introduction section.
Response:
Thank you for the suggestion. We included a sentence where we refer the choices, we made to build the population-weighted driving time, taking into account the parameters referred on introduction (lines 561 – 565).
Round 2
Reviewer 1 Report
Dear Authors
Thanks for swiftly addressing my recommendations.
Good luck.
Best regards
Author Response
Dear Reviewer,
Again, we would like to express our sincere thanks to you for the constructive support to improve our manuscript.
Reviewer 2 Report
Dear authors,
Thank you for changes which improved the paper (both the content and the structure). Nevertheless some sensitive comments are still not solved well. There are also several new comments:
171 fig. 1 instead of 3
Response:
Thank you for the suggestion. We made the change accordingly.
RE: 245 No change, still fig. 3
201-204 confusing interpretation of fig.2 which portrays internal variability of access time in 2011 for municipalities and not a temporal development
Response:
In order to clarify figure 2, we decided to improve the text regarding this figure and to include the same analysis for 1991 in order to support the temporal analysis (lines 310 - 318).
RE: 310-314 the text is still confusing. Parts such as “gap .. longer than 30 minutes” lead to misinterpretation because fig.2 represents an internal variability and NOT a travel time to a hospital. Consider to use “range of values” instead of the “gap”. E.g. “municipalities with a travel time gap to the reference hospital longer than 30 minutes“ -> „municipalities with a range of travel times to the reference hospital larger than 30 minutes“.
BTW, this type of analysis is strange and readers will welcome to directly evaluate how many people (or a share) lives in census tracts (inside municipalities) with a bad access time.
317 fig.2: use ≥46 instead of ≤46
318 in 1991 and 2011
247-255 the analysis and interpretation of age influence is not sufficient.
Response:
Thank you for the suggestion. We decided to strength the analysis of the association between older population and population-weighted driving time to the reference hospital. Moreover, we also included other characteristics as population density and premature mortality (lines 377-389).
RE: 377-388 I do not see any substantial change in the text. You have added a new graph and a table but the text (=interpretation) contains only 2 sentences about this phenomena which is still insufficient. Similarly, new factors (density, mortality) are interpreted by only 1 sentence.
387 table 3: What coefficient (of association?, slope of a regression line?) is used? What is the interpretation of this coefficient?
Add R2 as a new column to the table.
Correct following typos: 0-25, 0-84, 0.17-0.09
379 why is the association in 1991 lower than in 2011?
274-310 influencing factors are not satisfactory explained and, mainly, evaluation of their impact is missing. It is anticipated authors will interpret geographical differences based on analyses of each factor – e.g. what municipalities improved its access due to improvement of road network (and can you quantify this influence?), what municipalities take advantages of 2 new hospitals, where a depopulation significantly improves the situation (and quantify this impact). Also, according to l.295, some changes of catchments in the referenced time period occur – where and what is the impact?
Response:
Thank you for the suggestion. We decided to include a new analysis to the text to reveal the municipalities affected by changes on hospital distribution, extension of the road network and/or changes on population distribution (lines 345 – 363). Moreover, we improved the discussion by including an interpretation of the geographical differences of each influencing factor (lines 425 – 428, 433 – 434 and 440).
Regarding the quantification of the influence, it is very interesting It is something that we are already working and is planned to be part of another paper.
RE: Not enough. Now we know what municipalities are influenced by what factors but not on the level of census tracts and mainly - majority of these “influenced” municipalities are under the combination and we still do not know what factor is important. What is the dominant influence in the given municipality where all 3 reasons occur. Is the accessibility here better mainly due to highway construction or due to better administrative catchments?
379-409 such extended discussion about organisation of transboundary health services does not fit into a geographically oriented journal. It is anticipated authors provide modelling of accessibility including selected Spain hospitals, evaluate (geographically) differences and recognize (quantify) a potential for hospital accessibility improvement.
Response:
Thank you for the suggestion. Besides producing an outcome that can influence policy and, therefore, decrease health disparities, the aim of this paper was also to describe the method and analyse it considering the criteria to better evaluate geographical accessibility. Therefore, it would only make sense to publish this on a geographically oriented publication and on a special issue linking Geographical Information Systems to Health.
We agree that perhaps the presentation of possible interventions to be implemented on the municipalities is to big. However, we wanted to give the reader information to better understand the type of intervention that we recommend.
Regarding the modelling alternatives, recently we start to work with a research team from Spain in order to replicate this method and calculate the population-weighted driving time from the Iberian Peninsula. On that paper we want to strength the analysis of the outcome and not so much on the method. Moreover, that paper will have co-authors from both teams so it would not make sense to strength the method on that paper.
RE: This discussion is still too long and out of geographical focus. You write about possible intervention but we do not know if it is relevant from a geographical point of view – we do not know where Spain hospitals are located and where the accessibility can be improved. The topic of your paper is “Disparities in geographical access..” thus these question should be addressed.
492-3 „Most studies published to date consider the closest hospital to evaluate geographical accessibility in Portugal [24,53,54,73].“ - Why results of the new method are not compared with results of the previous methodology based on the closest hospitals?
Response:
Thank you for the suggestion. We included a paragraph on the manuscript where we compare the results (lines 409 – 416) and we also added why most studies decided to use the closest hospital instead of the reference one (line 625 - 627).
RE: Better to show differences in maps.
Conclusion – clear recommendations are missing. What are priorities (and where): build roads/improve transport services, build new hospitals, establish cross-border health services, change LHUs (catchments)? Or just wait for further depopulation of peripheral and rural settlements?
Response:
Thank you for the suggestion. On section 4.2 we highlight the municipalities where action should be taken and include recommendations on how municipalities can act to overtake the inadequate geographical accessibility. For instance, we refer the importance of strengthening the Local Health Units and the communication between the units within to support the municipalities that have a population-weighted driving time to the hospital higher than 30 minutes; changes to the cross-border healthcare directive and implementation of health interventions within the Euroregions to support the municipalities closer to the border with Spain; and adequate telehealth units on the municipalities with higher rates of elderly population (lines 463 – 554).
RE: I consider better and more local specific recommendations could be provided as I mentioned.
Author Response
Dear Reviewer,
We would like to express our sincere thanks to you for the constructive and positive comments. The manuscript has been fully revised according to your suggestions. Changes are highlighted in yellow. The following is a point-to-point response (A1 – round 1; A2 – round 2) to your comments (R1 – round 1; R2 – round 2), and responses are in green (see document in attach).
Author Response File: 
Author Response.pdf
Reviewer 3 Report
The author(s) have addressed all my comments properly.
Author Response
Dear Reviewer,
Again, we would like to express our sincere thanks to you for the constructive support to improve our manuscript.
