Health inequity on access to services in the ethnic minority regions of Northeastern Myanmar: a cross-sectional study

Objectives To evaluate health inequity on access to services in the ethnic regions of Northeastern Myanmar from three points of analysis: geographic barrier, gender-based disparity and financial burden of health services. Setting A multistage-stratified random cluster survey was conducted in Shan State Special Region 2 and Eastern Shan State Special Region 4 of Northeastern Myanmar in 2016, including a total number of 774 households. Participants A total number of 4235 participants were recruited during the survey. Primary and secondary outcome measures Geographic distance, gender, household income and inpatient/outpatient service utilisation. Results The study results showed that residents living within 5 km of any form of healthcare facilities paid more outpatient visits (90.06 visits per thousand population) in the past 2 weeks, compared with those living 5–20 km and over 20 km (54.84 and 54.02 per thousand population, respectively) from healthcare facilities. A similar trend with little significant differences was found for inpatient service use. Regarding household income, adults with an annual household income of above US$720 were more likely to seek outpatient service (OR=1.43, 95% CI 0.98 to 2.10) compared with those with an annual income of <US$360. After adjusting for other covariates, female adults were less likely to seek inpatient treatment (OR=0.55, 95% CI 0.35 to 0.84) and outpatient services (OR=0.86, 95% CI 0.64 to 1.15) than male adults. Conclusions Geographic barrier, gender-based disparity and financial burdens were identified as key causes that significantly restrict ethnic people’s access to healthcare facilities. The study concludes that tackling health inequity in Northeastern Myanmar ethnic regions requires an improved primary healthcare system, proper financial protection mechanisms and a special focus on women.


GENERAL COMMENTS
This is a very good and straightforward study that is presented in a very clear and transparent manner. The discussion is appropriate. Given the setting, the topic is obviously very timely and highly relevant. I am not aware of similar studies that have been performed in the specific geo-cultural context. The discussion is coherent.
As the main body of text is written in clear language, I have not ticked "minor revision" with a request for an extra round of language editing. However I see a few very minor glitches in the abstract and suggest that the language in the abstract be polished, as this forms the figurehead of your work.

GENERAL COMMENTS
This is a very interesting and relevant paper which evaluates with an appropriate statistical analysis the inequalities in the access to healthcare in Northeastern Myanmar. A journal such as BMJ Open can be the right place for this paper, having this manuscript relevance in particular for policymakers and epidemiologist.
I suggest only some minor revisions: 1. The title suggests a study on health inequities instead of a study on inequities on the access to healthcare.
3. Line 133. I assume 360 USD (and 720) is per year, but I do not see it specified. Moreover considered that almost 50% of the population in the We region is below this threashold (while in the SR$ is 8%), I suggest, if possible, to conduct a sensitivity analysis, or a complementary analysis, with lower threasholds. The relation incomeaccess to healthcare (outpatient and inpatient) should be non-linear, so if you use a threshould of 150 or 200 for example you should find apossibly by farhigher OR. This is relevant because still probably represents 10-30% of the population (I do not know the distribution). 4. I would suggest substituting the term "nutritional status" with something as food security; it seems more adequate considering the measure you use to evaluate it. 5. As fara as I can see Wa region and SR4 region have very different socioeconomic profiles, and you correctly have stratified the analyses per region. But when you study the exposure factors you use all the sample, so somehow you mix two population which could have different population-specific effects worthy to be shown. In particular the Wa region, could show some more extreme exposure effects that what is shown when you analyze them together. If you have done this in your data analysis, could be interesting to mention it as sensitivity or complementary analysis.

VERSION 1 -AUTHOR RESPONSE
Reviewer: 1 Reviewer Name: Dr Michael Thiede Institution and Country: Scenarium Group GmbH, Germany Please state any competing interests: None declared Comment: This is a very good and straightforward study that is presented in a very clear and transparent manner. The discussion is appropriate. Given the setting, the topic is obviously very timely and highly relevant. I am not aware of similar studies that have been performed in the specific geo-cultural context. The discussion is coherent.
As the main body of text is written in clear language, I have not ticked "minor revision" with a request for an extra round of language editing. However I see a few very minor glitches in the abstract and suggest that the language in the abstract be polished, as this forms the figurehead of your work.
Response: Thank you for your comments and encouragement, we have polished our language by a native English speaker as you kindly requested. Response: Thank you for the kind reminder, we have added in page 4 line 38.

Reviewer
-Moreover, considered that almost 50% of the population in the We region is below this threashold (while in the SR$ is 8%), I suggest, if possible, to conduct a sensitivity analysis, or a complementary analysis, with lower threasholds. The relation incomeaccess to healthcare (outpatient and inpatient) should be non-linear, so if you use a threshould of 150 or 200 for example you should find apossibly by farhigher OR. This is relevant because still probably represents 10-30% of the population (I do not know the distribution).
Response: We have done an additional analysis specifically for Wa region residents aged above 14, and set a separate group for household income below 150 USD, the result was similar as shown below. Also considering that there are few people with annual income below 150 USD in SR4, we have therefore chosen not to present the result in the manuscript to keep categories consistent across the two regions. Response: Thank you for pointing out the issue. To provide a better estimate of the odds ratios, we have added "region" as a confounding variable into the regression model, and presented the regionadjusted results in Table 3. This will minimize the influence of region to the association. Besides, the geopolitical condition of Wa region and SR4 region are similar, both were governed by ethnic minority authorities and health services were provided in parallel by Myanmar's Ministry of Health and Sports as well as local ethnic health authorities, thus we believe practically, it is reasonable to pool the data of two regions together while adjusting for the "cluster" in the regression models.