Multilevel modeling of county-level excessive alcohol use, rurality, and COVID-19 case fatality rates in the US.

OBJECTIVE: Reports of disparities in COVID-19 mortality rates are emerging in the public health literature as the pandemic continues to unfold. Alcohol misuse varies across the US and is related to poorer health and comorbidities that likely affect the severity of COVID-19 infection. High levels of pre-pandemic alcohol misuse in some counties may have set the stage for worse COVID-19 outcomes. Furthermore, this relationship may depend on how rural a county is, as access to healthcare in rural communities has lagged behind more urban areas. The objective of this study was to test for associations between county-level COVID-19 mortality, pre-pandemic county-level excessive drinking, and county rurality.
METHOD: We used national COVID-19 data from the New York Times to calculate county-level case fatality rates (n = 3,039 counties and county equivalents; October 1 -December 31, 2020) and other external county-level data sources for indicators of rurality and health. We used beta regression to model case fatality rates, adjusted for several county-level population characteristics. We included a multilevel component to our model and defined state as a random intercept. Our focal predictor was a single variable representing nine possible combinations of low/mid/high alcohol misuse and low/mid/high rurality.
RESULTS: The median county-level COVID-19 case fatality rate was 1.57%. Compared to counties with low alcohol misuse and low rurality (referent), counties with high levels of alcohol and mid (β = -0.17, p = 0.008) or high levels of rurality (β = -0.24, p<0.001) demonstrated significantly lower case fatality rates.
CONCLUSIONS: Our findings highlight the intersecting roles of county-level alcohol consumption, rurality, and COVID-19 mortality.

Introduction

COVID-19, the disease caused by the novel coronavirus SARS-CoV-2, is an ongoing global pandemic. At this writing (January, 2021), the US has seen more cases than any other country, with more than 20 million reported cases and surpassing 350,000 deaths [1]. Co-occurring and pre-existing health conditions exacerbate susceptibility to and the severity of COVID-19 infections [2]. While co-occurring health conditions have received considerable attention, particularly in relation to racial/ethnic disparities [3, 4], there has been less attention paid to behavioral and geographic risk factors associated with COVID-19 outcomes, such as alcohol use and rurality.

Levels of alcohol use and related problems vary across the US. The majority of Americans (73%) report drinking alcohol in the past year, and 13% meet the clinical criteria for a past-year alcohol use disorder [5]. However, alcohol use is not distributed evenly. Previous research has identified distinct regional differences in alcohol consumption [6], and ongoing public health surveillance finds varying levels of hazardous drinking by state [7].

Importantly, moderate to chronic heavy drinking suppresses immune responses and can increase susceptibility to and the seriousness of infectious and respiratory diseases, such as pneumonia, tuberculosis, and acute respiratory stress syndromes [8-10]. This same alcohol-induced immunosuppression may also increase the severity of COVID-19 infections at a population level. Given the effects of alcohol on infectious disease progression, high levels of county-level alcohol use preceding the COVID-19 pandemic are a particularly concerning population health problem.

Many burdens of heavy alcohol use can be exacerbated by other factors. For example, residence in more rural communities is an important social and environmental determinant of health and is associated with lower life expectancy [11], higher rates of all-cause morbidity and mortality [12], and multiple intersecting barriers in access to healthcare [13]. Roughly 60 million people, or nearly one in five Americans, live in a rural area [14].

Rural counties and towns have faced considerable challenges in dealing with the COVID-19 pandemic, partly attributable to weakened health care infrastructure, health care provider shortages, lower socio-economic status, and higher proportions of aging residents [15-17]. In the summer of 2020, the incidence rate of COVID-19 infections began to increase faster in rural areas, following the initial outbreaks in larger and more metropolitan urban centers [18]. Case fatality rates have also been shown to be higher in rural counties [19], rural states [20], and in rural counties with larger proportions of Black and Hispanic populations [21].

Furthermore, alcohol use and misuse vary by rurality. Rural populations are more likely than their urban counterparts to abstain from alcohol, but among non-abstainers, rates of alcohol use disorders (AUD) are higher in more rural areas [22]. Better understanding the association between county-level alcohol use and rurality will inform efforts to mitigate health inequities across regions.

To our knowledge, no studies have investigated the intersection between alcohol misuse, rurality, and COVID-19. This study was designed to measure disparities in COVID-19 infection severity at the county population level, which we operationalized as case fatality rates. We considered two important determinants of population health–excessive alcohol consumption and rurality–and their relationship to case fatality rates at the county level. Although largely an exploratory study, we expected to see higher burdens and more disparities in rural areas.

Materials and methods

Data source and sample

We used the New York Times’ COVID-19 data repository (https://github.com/nytimes/covid-19-data) to calculate case fatality rates in US counties with at least one COVID-19 death reported between October 1, 2020 and December 31, 2020. We restricted our sample to the last quarter of 2020 because this was the most recent data available at the time of analysis, and to highlight relationships during a time of particularly high case and death rates across the US. Case fatality rates were calculated by dividing the total number of deaths by the total number of confirmed cases, indicating the percentage of individuals with COVID-19 who died. Case fatality rates help to illustrate disproportionate mortality burdens between populations or geographies where the case rates may be similar but underlying social, economic, and health conditions differ. Furthermore, case fatality rates allow for more appropriate comparison between counties with substantial variation in population size.

Variables

Our focal predictor was a single combined variable representing levels of pre-pandemic county-level excessive alcohol consumption and rurality. The alcohol component was sourced from the Robert Wood Johnson Foundation’s 2018 County Health Rankings [23] and merged to the analytic dataset by county. County-level excessive alcohol consumption was defined as the percentage of adults who reported exceeding recommended drinking limits, which included: 1) four or more alcoholic drinks consumed on a single occasion in the past month for women and five or more alcoholic drinks on a single occasion in the past month for men (i.e., binge drinking), and/or 2) more than one drink on average per day for women, and more than two drinks on average per day for men (i.e., exceeding recommended daily limits). We used Waldorf and Kim’s Index of Relative Rurality (IRR) to define county-level rurality [24]. The IRR is a threshold-free measurement of rurality at the county level, where greater scores indicate greater rurality. The IRR takes into account population size, population density, remoteness, and built-up infrastructure as a percentage of total land area. Our final analytic variable included nine possible combinations of low/mid/high alcohol consumption and low/mid/high rurality. The low/mid/high cutoffs for both variables were based on tertiles, resulting in nine categorical levels of 1) low alcohol use and low rurality, 2) low alcohol use and mid rurality, 3) low alcohol use and high rurality, 4) mid alcohol use and low rurality, 5) mid alcohol use and mid rurality, 6) mid alcohol use and high rurality, 7) high alcohol use and low rurality, 8) high alcohol use and mid rurality, and 9) high alcohol use and high rurality.

We included several county-level variables in our model that likely confound the relationships between excessive alcohol consumption, rurality, and COVID-19 case fatality rates. All covariates (prevalence of smoking, obesity, diabetes, older age, race/ethnicity, and unemployment) were sourced from the County Health Rankings data. Smoking was defined as the percent of a county’s adults who reported currently smoking tobacco every day or on most days, and who have smoked at least 100 cigarettes in their lifetime. Older age was defined as the percent of a county’s population aged 65 years and older. Obesity was defined as the percentage of a county’s adults, aged 20 years and older, who report a body mass index of 30 kg/m2 or greater. Diabetes was defined as the percentage of a county’s adults, aged 20 years and older, who have been diagnosed with diabetes. Race/ethnicity was represented by the percentage of a county made up of self-identified nonwhite residents. Finally, unemployment was defined as the percentage of a county’s civilian labor force, aged 16 and older, who were unemployed but seeking work during the past week.

Our final analytic sample included 3,039 counties and county equivalents in the US that had at least one death recorded and for which we had complete data for all study variables, (i.e., matching FIPS codes between the New York Times and County Health Rankings datasets), representing 96% of all 3,143 US counties and county equivalents.

Analysis

We used SAS software for all analyses (Version 9.4) [25]. We described the percentage of counties that fall into each level of our categorical predictor, and the median, minimum, and maximum values for all continuous study variables. We reported the IRR scores and case fatality rates for the ten counties with the lowest percentage of excessive drinking and for the ten counties with the highest percentage of excessive drinking. We also developed an original, county-level US map to illustrate the geographic distribution of excessive alcohol consumption and rurality using Mathematica software.

We used beta regression to model case fatality rates. Beta regression is a more appropriate framework than linear regression to model outcomes that are proportions or rates, such as case fatality rates [26]. We considered three separate structures for our model and selected the final model based on how well the model fit the data, as indicated by the lowest Akaike Information Criterion (AIC) value. All candidate models were adjusted for county-level indicators of smoking, obesity, diabetes, age, race/ethnicity, and unemployment. Candidate model #1 used a non-structural, fixed effects design (AIC = -19337). Candidate model #2 included state as a random effect, under the assumption that counties clustered within one state are likely more similar to each other than to counties in other states (AIC = -19845) [27]. Finally, candidate model #3 included state defined as a random effect and accounted for the structural spatial autocorrelation between counties. This model incorporated a Gaussian correlation structure using geographic county centroids (AIC = -19843) [28]. Candidate model #2 was selected as the final analytic model because it demonstrated the lowest AIC value and the best fit to the data. As a multilevel model, we reported the Intraclass Correlation Coefficient (ICC), which is an indicator of how much of the total variation in county case fatality rates is accounted for by the state. Finally, to model our nine-level categorical focal predictor, we included eight dummy variables and used low alcohol use/low rurality as the referent group.

We interpreted study results using a significance threshold of α = 0.05. We also identified no problematic multicollinearity between the study variables, defined as a Pearson correlation coefficient greater than 0.80.

Results

County population characteristics are reported in Table 1. The median county-level COVID-19 case fatality rate was 1.57% (range 0.01% - 16.08%) (Table 2). The high upper limit of this range was occupied by Kenedy County, Texas (16.08%; 184 deaths out of 1,144 reported cases), and the second highest was Franklin County, Massachusetts (11.45%; 7,003 deaths out of 61,156 reported cases). Among all counties, 7.15% (n = 217) were at levels of low alcohol and low rurality, and 10.46% (n = 318) were at levels of high alcohol and high rurality. Eight out of the ten counties with the lowest percentages of excessive drinking were in the US South (Table 3). The ten counties with the lowest alcohol levels had a combined average IRR score of 0.55 (more rural) and an average case fatality rate of 2.31%. Conversely, all ten counties with the highest levels of excessive alcohol use were in Wisconsin, with a combined average IRR score of 0.45 (less rural) and an average case fatality rate of 0.65%. Counties with both high alcohol and high rurality tended to be clustered in the upper Midwest, parts of the mountain West, and in the states of Nevada and Alaska (Fig 1).

Table 1

County characteristics (n = 3,039 counties).

Variables	Median	Min, Max
Percentage of a county’s adult population who reported smoking every day or most days	17.03	5.91, 41.49
Percentage of a county’s adult population with a BMI of 30 or higher	33.20	12.40, 57.70
Percentage of a county’s adult population with diagnosed diabetes	11.70	1.90, 34.10
Percentage of a county’s population aged 65 years and older	18.83	4.83, 57.58
Percentage of a county’s population that is nonwhite	16.76	2.11, 97.31
Percentage of a county’s civilian labor force, aged 16 and older, that is unemployed but seeking work	3.88	1.30, 18.09

Table 2

COVID-19 case fatality rates by county-level alcohol use and rurality (n = 3,039 counties).

Variables	n	%	Case fatality rate
			median (min, max)
Focal predictor
Low alcohol use counties
Low rurality	217	7.15	1.84 (0.17, 8.76)
Mid rurality	420	13.82	1.77 (0.14, 8.99)
High rurality	368	12.11	1.72 (0.04, 16.08)
Mid alcohol use counties
Low rurality	352	11.58	1.68 (0.01, 7.24)
Mid rurality	301	9.90	1.71 (0.01, 7.60)
High rurality	352	11.58	1.65 (0.02, 8.30)
High alcohol use counties
Low rurality	433	14.25	1.48 (0.07, 11.45)
Mid rurality	278	9.15	1.25 (0.01, 5.62)
High rurality	318	10.46	1.20 (0.01, 9.30)

Table 3

Characteristics of top 10 and lowest 10 counties by county-level excessive drinking.

County	State	Excessive drinkinga (%)	IRR score	Case fatality rate (%)
Lowest ten counties
Utah	UT	7.81	0.39	0.29
Clay	GA	9.32	0.59	2.09
Jefferson	MS	9.49	0.57	3.34
Holmes	MS	9.50	0.54	4.79
Piete	UT	9.55	0.65	2.25
Greene	AL	9.57	0.57	4.10
Perry	AL	9.68	0.56	1.03
McDowell	WV	9.71	0.53	0.30
Humphreys	MS	9.80	0.55	3.53
Quitman	MS	9.81	0.56	1.36
Top ten counties
Portage	WI	27.32	0.48	0.80
La Crosse	WI	27.33	0.42	0.44
Calumet	WI	27.41	0.46	0.59
Outagamie	WI	27.54	0.41	0.81
Dunn	WI	27.61	0.50	0.39
Dodge	WI	27.92	0.47	0.91
St. Croix	WI	27.96	0.47	0.48
Dane	WI	28.22	0.38	0.37
Washington	WI	28.25	0.41	0.83
Pierce	WI	28.62	0.50	0.84

a Excessive drinking was defined as past-month binge drinking (≥4 drinks on a single occasion for women; ≥5 drinks on a single occasion for men) or exceeding recommended daily limits (>1 drink per day for women; >2 drinks per day for men).

Fig 1

US county map of levels of excessive alcohol consumption and rurality.

Original map was created by the study team with Mathematica software.

In our fitted model, compared to counties with low alcohol use/low rurality (referent), counties with mid levels of alcohol use and high levels of rurality (β = -0.14, p = 0.019), as well as counties with high levels of alcohol use and mid (β = -0.17, p = 0.008) and high levels of rurality (β = -0.24, p<0.001), demonstrated significantly lower case fatality rates (Table 4). County prevalence of smoking, obesity, and unemployment was not associated with case fatality rate; however, county prevalence of diabetes, older adults, and non-white residents were each positively associated with greater case fatality rates. The ICC value demonstrated that roughly 4% of the variability of case fatality rates between states was attributed to county clustering within states (p<0.001).

Table 4

Multivariate beta regression modeling case fatality rate (n = 3,039 counties).

Variables	β	SE	p
Focal predictor
Low alcohol use counties
Low rurality	Ref.	Ref.	Ref.
Mid rurality	-0.01	0.04	0.982
High rurality	-0.07	0.05	0.138
Mid alcohol use counties
Low rurality	-0.06	0.05	0.197
Mid rurality	-0.07	0.06	0.207
High rurality	-0.14	0.06	0.019
High alcohol use counties
Low rurality	-0.09	0.06	0.110
Mid rurality	-0.17	0.07	0.008
High rurality	-0.24	0.07	<0.001
Covariates
Percentage of a county’s adult population who reported smoking every day or most days	-0.01	0.01	0.671
Percentage of a county’s adult population with a BMI of 30 or higher	0.01	0.01	0.259
Percentage of a county’s adult population with diagnosed diabetes	0.01	0.01	0.032
Percentage of a county’s population aged 65 years and older	0.02	0.01	<0.0001
Percentage of a county’s population that is nonwhite	0.01	0.01	<0.0001
Percentage of a county’s civilian labor force, aged 16 and older, that is unemployed but seeking work	0.01	0.01	0.552

Intraclass correlation coefficient = 0.037, p<0.001.

Discussion

To our knowledge, this is the first study to examine county-level relationships between excessive alcohol consumption before the pandemic, rurality, and COVID-19 mortality. Notably, results indicated a joint contribution of alcohol use patterns and level of rurality. Case fatality rates were generally lowest in more rural counties; compared to low alcohol use/low rurality counties, rates were lower in mid alcohol use/high rurality, high alcohol use/mid rurality and high alcohol use/high rurality counties in particular. Our findings highlight the importance of considering both behavioral and environmental determinants of COVID-19 outcomes.

Our findings did not support our hypothesis of higher burdens and more disparities in rural areas. Instead, we found lower fatality burdens in some mid and high rural counties. In rural settings, other factors besides alcohol use may be more important drivers of COVID-19 fatality, such as a county’s age distribution, racial/ethnic distribution, and unemployment levels. Some rural areas may offer protection against the effect of alcohol on mortality. Transmission may be less efficient than in larger and more concentrated urban populations, and drinking norms in rural areas may include smaller and less frequent group meetings in bars or indoor environments, or higher levels of drinking in isolation. Future studies that incorporate longitudinal changes in alcohol use behaviors could improve our understanding of rural and urban differences in health behavior and substance misuse at multiple time points before, during, and following the pandemic.

Potential limitations

These results should be considered in light of several potential limitations. County-level excessive drinking may have been misestimated as individual survey responses about drinking habits may be subject to social desirability biases [29]. Such underreporting would decrease our ability to detect an association. Second, as a secondary data analysis, some variables were not available, such as asymptomatic COVID-19 cases, COVID-19 testing capacity, relevant county-level comorbidities (i.e., chronic lung disease or asthma, liver disease, serious heart conditions, and other diseases affecting immune response), and county healthcare characteristics (i.e., available ventilators and provider shortage areas). Finally, as an ecological study using county-level data, no individual-level information was available. As such, there is no way to identify the demographic characteristics of the population that made up the cases and deaths. Future studies that evaluate relationships between individual-level alcohol use and COVID-19 mortality outcomes would contribute substantially to the literature around behavioral health and infectious disease outbreaks. As this study was undertaken in the midst of the COVID-19 pandemic, our pattern of results should be confirmed as more data accumulate. Nevertheless, our findings address a gap in knowledge about the association of excessive alcohol consumption, rurality, and COVID-19 outcomes.

Conclusion

Our findings demonstrate lower COVID-19 fatality burdens in some counties, in particular counties with mid/high excessive alcohol use and counties with mid/high levels of rurality. These findings were contrary to our hypothesis and highlight the intersecting roles that excessive alcohol consumption and geography play in baseline risk for COVID-19 outcomes at the community and population levels.

29 Mar 2021

PONE-D-21-01356

Multilevel modeling of county-level excessive alcohol use, rurality, and COVID-19 case fatality rates in the US

PLOS ONE

Dear Dr. Pro,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by May 13 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Antonio Palazón-Bru, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please refer to the specific statistical analyses performed as well as any post-hoc corrections to correct for multiple comparisons. If these were not performed please justify the reasons. Please refer to our statistical reporting guidelines for assistance (https://journals.plos.org/plosone/s/submission-guidelines.#loc-statistical-reporting). Additionally, please ensure you have thoroughly discussed any potential limitations of this study within the Discussion.

3. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. Please see the following video for instructions on linking an ORCID iD to your Editorial Manager account: https://www.youtube.com/watch?v=_xcclfuvtxQ

4. We note that Figure 1 in your submission contain map images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

4.1.    You may seek permission from the original copyright holder of Figure 1 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

4.2.    If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: No

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Summary:

This study wanted to determine if county COVID-19 case fatality rate could be predicted by the 9-level grouping of alcohol use (low, medium, high) and rurality (low, medium, high) levels, while controlling for state, and other health factors (such as smoking, age, diabetes, etc.). They found that, contrary to their exploratory hypothesis, there were lower case fatality rates in higher alcohol-using and more rural counties than the baseline (low alcohol, low rural).

This manuscript is well-written and covers a relevant topic. Furthermore, the model design is well-founded and had appropriate controls and justification. Finally, the results were very interesting, as they were the opposite of what is expected and had a nice bevy of detail. The major weakness of the manuscript was an incorrect assertion in the discussion/conclusion about the burden of COVID-19 which was not supported by any results. A related weakness is the call-to-action statement, which was in no way justified by the results.

-----------------------------------------------------------------------------------------------------------------------------

Specific Areas of Improvement:

Abstract:

Major:

Page 2, Line 23-27: Your conclusion has a call-to-action that includes alcohol prevention/treatment in the pandemic response plan. While mitigating alcohol abuse is a good in of itself, increasing efforts based on the pandemic is not supported by your results. Counties with higher alcohol levels had lower case fatality rates. You shouldn’t include such a call-to-action unless you can convincingly demonstrate why, despite the contrary results, that such a response is warranted, and I don’t see how you could.

Minor:

Page 2, Lines 15-17: consider adding ‘while controlling for confounding factors’ to the model description.

Page 2, Line 21: p-values should have consistent significant digits (I typically see 3-4 as the preferred number in other manuscripts).

Introduction:

Major:

none

Minor:

none

Methods:

Major:

none

Minor:

Page 5, Line 5: For the COVID-19 deaths, you used only the most recent 3 months. Is there a reason why? That time period, while likely a reasonable approach, should be justified.

Results:

Major:

none

Minor:

Page 12, Lines 2-4, 5: As noted in the abstract, the p-values should have consistent significant digits.

Figures and Tables:

Major:

none

Minor:

Table 2: The header for the binge drinking column is much longer than the other headers. Perhaps shorten the header for binge drinking and give the additional information into the table description [example for revised header: Excess drinking (%)].

Table 3: For the second column header, change the B to β to match the format from the text portion of the results. For the p-value column, again set the p-values to consistent significant digits.

Discussion:

Major:

Page 14, Lines 4-7: You write that the burden of higher case fatality rates lay predominately in mid alcohol use/mid rural and high alcohol use/high rural counties. This is not what the results show. High alcohol use/high rural counties had the lowest case fatality rates, while mid alcohol use/mid rural had lower rates (though not significant) to the referent (low alcohol/low rural). These two sentences need to be revised to reflect what the results found.

Page 14, Lines 19-23; Page 15, Lines 1-7: You write that alcohol consumption during self-isolation poses a substantial public health challenge. However, none of your results provide any evidence for this claim. The three categories of counties with high alcohol had the lowest case fatality rates (1.48%, 1.25%, and 1.20%, from Table 1). Furthermore, even looking at the top 10 highest and lowest alcohol abuse counties, the average case fatality rates are far lower in in the highest (0.646%) compared to the lowest (2.308%) (values calculated by averaging the case-fatality-rate values in Table 2). As noted in the abstract, while mitigating alcohol abuse is a good in of itself, increasing efforts based on the pandemic is not supported by your results. This whole section needs to be dropped or revised to justify such action despite contrary results (which I do not see how you could).

Page 15, Lines 21-22; Page 16, Line 1: This sentence needs to be corrected in the same way as the Page 14, Lines 4-7.

Page 16, Lines 3-5: This call-to-action sentence, like the section of Page 14, Lines 19-23; Page 15, Lines 1-7, has no support from your results. It should not be included.

Minor:

Page 14, Line 5: the word ‘lied’ should be ‘lay’

Page 15, Line 8: The biggest potential limitation is that this is an ecological study (i.e., the data was aggregated at the county level, so individual metrics were not available). You cannot know if the population that made up the cases and deaths have the same demographics as the other variables (rurality, alcohol use, etc.). While this may well be an obvious statement, it is good to mention it since you’ve included a section on potential limitations.

Page 15, Line 20: For your conclusion, because of the following the facts: 1) the results were contrary to expectations, 2) it was an ecological study, and 3) you’re trying to give alcohol prevention recommendations, it may be a good idea to outline a future experiment that would test the link between mortality and alcohol use on an individual level.

-----------------------------------------------------------------------------------------------------------------------------

Specific Areas of Achievement:

Abstract:

Objective, methods, and results sections were clear and concise, presenting the reader with an understandable summary of your project.

Introduction:

Page 4, Lines 20-23: The statement of the project and the hypothesis are good additions.

Methods:

Page 7, Lines 10-23; Page 8, Lines 1-3: the description of the model building and selection was very clear and well justified. I thoroughly enjoyed reading this section.

Results:

Short and to the point. It conveyed the major results while letting the details reside in the appropriate figures and graphs.

Figures and Tables:

Figure 1 was particularly well made. Tables were clear and informative.

Discussion:

While the call-to-action was no appropriate for the results found, I do like the attempt to translate results into action.

Reviewer #2: This paper explored the relationship between county-level excessive alcohol consumption, rurality and COVID-19 case-fatality rate using multilevel beta regression analyses. The authors make a compelling argument for the pathway through which a higher prevalence of excessive alcohol consumption may affect COVID-19 fatality. They also make a compelling argument in the discussion section about potential reasons for the reversed associations between the outcome and focal predictor. The paper is methodologically sound. However, as indicated in the feedback below, some of the discussion and conclusion points are in contrast to the results shown in the tables. Minor revisions are suggested.

Tables.

Table 1. For ease of readability, the authors should consider providing summary statistics in table 1 for only the covariates (second part of table 1) and having the first part showing case-fatality rates for values of the focal predictor as a separate table.

Table 3. The low alcohol mid rurality coefficient is reported as 0.00, the coefficient should be rounded to a non-zero value. Same applies to “percentage of a county’s adult population with a BMI of 30 or higher” covariate and several standard errors.

Discussion.

Page 14 line 4-6 “We found that the burden of higher case fatality rates lied predominately in mid alcohol use/mid rural and high alcohol use/high rural counties.” However, table 3 shows that the mid alcohol /mid rural was not statistically significant and also did not have the largest effect size (the mid alcohol/high rural and high alcohol/mid rural were significant).

Page 15 line 2 “Given indications that alcohol use is associated with higher COVID-19 mortality in some areas,…” this is in contrast to the index study finding and the authors do not provide any citation to support the statement.

Page 15 line 21-page 16 line 1 “our findings demonstrate disparate COVID-19 fatality burdens in some counties, in particular counties with mid/high excessive alcohol use and counties with mid/high levels of rurality.” This again is not supported by the study findings. Table 1 and table 2 indicate that the case fatality burden is more for the low alcohol/low rural counties.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Mark R Williamson

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: Comments_PLOSone_03.24.2021.pdf

Click here for additional data file.

7 May 2021

Thank you to the academic editor and our two reviewers for the feedback and constructive criticism of our manuscript. The changes we made in response to the reviews have strengthened the paper overall, with a focus on keeping our discussion in line with our reported findings and refraining from making recommendations that do not stem directly from our results. Below we have summarized some of the reviews and combined others that were noted by both reviewers. We have provided one revised version with tracked changes, and another clean version with changes accepted.

General feedback from the academic editor

1) We note that Figure 1 in your submission contains map images which may be copyrighted. We cannot publish previously copyrighted maps or satellite images created using proprietary data. We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission.

We have communicated with our PLOS publishing editor, Theodore Peng, about our map. This original map was created with Mathematica software by the study team using the same data from the study analyses. The map image is not copyrighted.

We have included additional information about the creation of the map in the methods section and added a caption to the figure specifying that the map was created by the study team using Mathematica.

Abstract

2) Page 2, Line 23-27: Your conclusion has a call-to-action that includes alcohol prevention/treatment in the pandemic response plan. While mitigating alcohol abuse is a good in of itself, increasing efforts based on the pandemic is not supported by your results. Counties with higher alcohol levels had lower case fatality rates. You shouldn’t include such a call-to-action unless you can convincingly demonstrate why, despite the contrary results, that such a response is warranted, and I don’t see how you could. (Reviewer 1)

We agree that this call-to-action is beyond the scope of the study findings. We have removed this statement from the abstract as well as throughout the discussion section.

3) Page 2, Lines 15-17: consider adding ‘while controlling for confounding factors’ to the model description. (Reviewer 1)

We have included a statement in the abstract clarifying that our model was adjusted for relevant county-level population characteristics.

4) Page 2, Line 21: p-values should have consistent significant digits (I typically see 3-4 as the preferred number in other manuscripts). (Reviewer 1)

We have reviewed the PLOS reporting guidelines for p-values, and have reported all p-values to three decimal places throughout the manuscript.

Methods

5) Page 5, Line 5: For the COVID-19 deaths, you used only the most recent 3 months. Is there a reason why? That time period, while it is likely a reasonable approach, should be justified. (Reviewer 1)

We have clarified our justification for using this date range, including that it was a time of particularly high case and death rates across the US, and was also the most recent data available at the time of analysis.

Results

6) Table 1. For ease of readability, the authors should consider providing summary statistics in table 1 for only the covariates (second part of table 1) and having the first part showing case-fatality rates for values of the focal predictor as a separate table. (Reviewer 2)

Thank you for helping us improve the readability of our tables. We have revised our tables to present only summary statistics for county population characteristics in Table 1, then moved the case fatality rate information to a new Table 2.

7) Table 2: The header for the binge drinking column is much longer than the other headers. Perhaps shorten the header for binge drinking and give the additional information into the table description [example for revised header: Excess drinking (%)]. (Reviewer 1)

We shortened the column header to the suggested “Excessive drinking (%)”, which helped clean up this table and keep column headers small and similarly sized.

8) Table 3: For the second column header, change the B to β to match the format from the text portion of the results. (Reviewer 1)

This has been updated in Table 3.

9) Table 3. The low alcohol mid rurality coefficient is reported as 0.00, the coefficient should be rounded to a non-zero value. Same applies to “percentage of a county’s adult population with a BMI of 30 or higher” covariate and several standard errors. (Reviewer 2)

We have rounded the small coefficient and standard error values to the nearest non-zero value using two decimal places.

Discussion

10) Page 14, Lines 4-7: You write that the burden of higher case fatality rates lay predominately in mid alcohol use/mid rural and high alcohol use/high rural counties. This is not what the results show. High alcohol use/high rural counties had the lowest case fatality rates, while mid alcohol use/mid rural had lower rates (though not significant) to the referent (low alcohol/low rural). These two sentences need to be revised to reflect what the results found. (Reviewer 1)

Page 14 line 4-6 “We found that the burden of higher case fatality rates lied predominately in mid alcohol use/mid rural and high alcohol use/high rural counties.” However, table 3 shows that the mid alcohol /mid rural was not statistically significant and also did not have the largest effect size (the mid alcohol/high rural and high alcohol/mid rural were significant). (Reviewer 2)

We appreciate that these mistakes were pointed out. We have clarified our findings and the direction of significant associations in this first discussion paragraph.

11) Page 14, Lines 19-23; Page 15, Lines 1-7: You write that alcohol consumption during self-isolation poses a substantial public health challenge. However, none of your results provide any evidence for this claim. This whole section needs to be dropped or revised to justify such action despite contrary results (which I do not see how you could). (Reviewer 1)

We agree that this paragraph expands the discussion beyond the study findings. In order to keep our discussion aligned with the reported findings – and not venture into a call-for-action or recommendations for treatment interventions – we have removed this paragraph.

Limitations

12) Page 15, Line 8: The biggest potential limitation is that this is an ecological study. You cannot know if the population that made up the cases and deaths have the same demographics as the other variables. While this may well be an obvious statement, it is good to mention it since you’ve included a section on potential limitations. (Reviewer 1)

We have added a short paragraph outlining the limitations of ecological study designs, as well as a recommendation for future observational studies identifying pathways between behavioral health and COVID-19 outcomes at the individual level.

Conclusion

13) Page 15, Line 20: For your conclusion, because of the following the facts: 1) the results were contrary to expectations, 2) it was an ecological study, and 3) you’re trying to give alcohol prevention recommendations, it may be a good idea to outline a future experiment that would test the link between mortality and alcohol use on an individual level. (Reviewer 1)

Please see our response to Comment #12, above. We have drawn attention to the limitations of interpreting results from ecological studies, while also making a recommendation for future observational studies that are well suited to identify associations between individual-level characteristics like health behavior and disease outcomes.

14) Page 15 line 21-page 16 line 1 “our findings demonstrate disparate COVID-19 fatality burdens in some counties, in particular counties with mid/high excessive alcohol use and counties with mid/high levels of rurality.” This again is not supported by the study findings. Table 1 and table 2 indicate that the case fatality burden is more for the low alcohol/low rural counties. (Reviewer 2)

We have made efforts throughout the manuscript to correctly report our findings and clarify the direction of associations, including in the conclusion section. We have also removed our recommendations for strengthened alcohol screening and treatment as they are not within the scope of our reported findings.

7 Jun 2021

Multilevel modeling of county-level excessive alcohol use, rurality, and COVID-19 case fatality rates in the US

PONE-D-21-01356R1

Dear Dr. Pro,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Antonio Palazón-Bru, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Revised version looks good. A strength of this manuscript is the clarity of statistical analysis and the use of a single variable to capture both alcohol usage and rurality into categories.

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Mark R Williamson

Reviewer #2: No

9 Jun 2021

PONE-D-21-01356R1

Multilevel modeling of county-level excessive alcohol use, rurality, and COVID-19 case fatality rates in the US

Dear Dr. Pro:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Antonio Palazón-Bru

Academic Editor

PLOS ONE