Importance of responsible interpretation of observational studies with an impact on public health policy

doi:10.21203/rs.3.rs-1431992/v1

Download PDF

Article

Importance of responsible interpretation of observational studies with an impact on public health policy

https://doi.org/10.21203/rs.3.rs-1431992/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Observational studies are often referenced when informing policy decisions in public health. However, there limitations associated with the nature of observational studies may lead to erroneous interpretations and recommendations. A peer-reviewed publication published in 2018 reported that communities living near hog Concentrated Animal Feeding Operations (CAFO) have increased negative health outcomes and mortalities, in the U.S. state of North Carolina (NC). While authors stated that associations does not imply causation, speculative interpretation of their results by media caused detrimental effects on the swine industry. The objective here was to repeat their study using updated data and to evaluate the strength of conclusions and appropriateness of methods used with the ultimate goal of alerting on the impact that study limitations may have when used as evidence for decision-making.

The referenced manuscript reported results from an ecological study design that used available data from 2007–2013. They analyzed exposure to CAFOs using a logistic regression model conducted at the individual-level, while presumably correcting for six confounding factors. We used same data sources and analytical methods, with additional years of data (2007–2018). Mortality, hospital admissions, and emergency department visits related eight diseases were analyzed (anemia, kidney disease, infectious diseases, tuberculosis, low birth weight, HIV, and diabetes). The study population and their exposure was defined by categorizing NC zip codes into three by pig density; where, > 1 hogs/km² (G1), > 232hogs/km2 (G2), and no hogs (Control).

Our repeated analysis recognized that the early study approach was inadequate to assess the intended hypothesis and had shortcomings including presence of ecological fallacy, residual confounding, inconsistency of associations, and over estimation of exposure. Diseases that are not causally relatable to hog CAFOs, such as HIV and diabetes, were also prominent in the communities neighboring CAFOs (G1 and G2). These associations, in part, reflects an alternative explanation of systemic health disparities of the communities near hog CAFOs.

An ecological study would not confirm or deny exposure over time nor indicate causation. Hence, we emphasize the importance of responsible interpretation of ecological studies that concerns public health, agriculture, and economy, for science to provide appropriate and accurate support to the ultimate objective of targeting policy to enhance the quality of life and health of the communities.

animal agriculture

public health

CAFO

epidemiology

risk factors

Concentrated animal feeding operations (CAFOs) are the result of a response to the necessity to reduce land use while maintaining the demand for increased food production. CAFOs are defined by the US Environmental Protection Agency (EPA) as “an animal feeding operation with more than 1000 animal units confined on site for more than 45 days during the year” (USDA, 2011; USDA NRCS, 2021), in which the animal unit of pig CAFO is defined as 2500 swine weighing more than 45 days during the year. EPA further defines that, an animal feeding operation that discharges manure or wastewater into a natural or manmade ditch, stream or other waterway is defined as a CAFO, regardless of size.

United States is the third largest swine producing country of the world (USDA, 2019). North Carolina alongside Midwestern states of Iowa and Minnesota are the three major swine producing states in the U.S. (USDA NASS 2017). The swine (i.e. hog and pig) industry is an economic powerhouse in North Carolina that supports 44,000 total jobs and provides over USD 10-billion economic output for the state yearly (National Hog Farmer, 2019). Environmental activist groups in North Carolina have claimed that CAFOs are a source of air and water pollution (Wing et al., 2000; Wing, 2002; Wing et al., 2002; Wing et al., 2008; Nicole, 2013), and several public health issues (Smit et al., 2014; Kravchenko et al., 2018; Barnett, 2018). While securing public health should be a primary focus, observational studies that concerns public health, agriculture, and economy need interpretation with caution.

A relatively recent peer-reviewed publication reported that communities living near hog CAFOs have increased negative health outcomes and mortalities, in the U.S. state of North Carolina (Kravchenko et al., 2018). While authors stated that associations do not imply causation, media and environmental activist groups interpreted the results out-of-context, adding fuel to ongoing debates on hog farming and it is impact on public and environmental health. Consequently, swine producers associations have reported that the published work was used as evidence to file multiple lawsuits causing detrimental effects and financial losses (https://deq.nc.gov/; Freese 2020). In the 2018 publication, the authors analyzed the association of all-cause mortality, hospitalizations, and emergency department visits with the CAFO density in a zip code, by conducting a series of independent logistic regression analyses. The 2018 study chose six disease conditions (anemia, kidney disease, infant mortality, low birth weight, septicemia, and tuberculosis) that were proposed to have associations with CAFO exposure, although, unfortunately, the rationale for selecting specific health outcomes while leaving others out was unclear. For example, it would have been beneficial to include negative health outcomes that one would not expect to be associated with location of swine operations to serve as a control or reference points.

Conducting epidemiological analysis and modeling of hazardous environmental exposures in relation to negative health outcomes is challenging. These approaches require that the exposure, disease outcomes, and confounding factors are accurately measured and their effects on the exposure or outcome are modeled correctly. It is often difficult to derive the causal relationship of residing in a high CAFO dense zip code leading to negative health outcomes from an ecological study design, mainly because the duration and extent of exposure cannot be measured. Moreover, insufficient accounting for confounding factors that are indicative of health and social disparities can further lead to misleading results.

The objective of the study here was to re-analyze the previous study using updated data for repeatability, and identify whether their study was influenced by preconceived notion, data quality, and choice of analytical methods. Then, if needed, to propose solutions to the issues identified, conduct the analysis accordingly, interpret, and convey the output to the interested parties including swine producers and the North Carolina communities. The null hypothesis was that the associations observed in the previous study were consistent through time, and thus, repeatable. In addition to the six disease conditions selected in the 2018 publication, diabetes and HIV were chosen for this study as “controls”, in which these diseases have no established or proposed associations with CAFOs. The control diseases were chosen because observing compatible associations between the residence in zip codes with CAFOs and HIV or diabetes, it is reasonable to theorize that the underlying health disparities and social determinants of health in these communities are the key reason, and not due to health risks directly imparted by the presence or proximity of hog CAFOs.

The overarching objective here was to identify technical soundness of the analytical methods of the previous study rather than proving or denial of the associations, because we encourage responsible and cautious interpretation of ecological studies that concerns public health, agriculture, and economy. We identify the importance of transparency when communicating scientific information derived from an observational study to the hog industry and the residents of North Carolina (Jepsen et al., 2004). Therefore, our study outcomes would be shared with hog industry and the public and would be used to support the communications between the stakeholders when taking actions to improve the animal, environmental, and public health in North Carolina. This work intends to encourage the private and public sectors in North Carolina to direct efforts and resources towards factors and forces truly affecting community health, with the ultimate goal of improve the quality of life of the communities.

Zip codes classified by exposure

The three zip codes divisions of North Carolina (n= 808) based on hog density were Group 1 (≥1 hogs/square kilometer; n= 211), Group 2 (>234 hogs/sqkm; n=53; Group 2 is a subset of Group 1), and controls (n= 597) (Fig 01). The matched control zip codes of Group 2 were randomly distributed compared to the Group 2 area that was clustered in the southeastern part of the state (Supplementary figure 01).

The negative health outcomes

When the primary diagnosis listed in H-CUP data were summarized by year (Fig 02), the all-cause mortality, hospital admissions, and emergency department visits did not exhibit a trend over the years. The disease specific morbidities (both hospital and emergency department visits collectively) related to anemia and septicemia exhibited an increase in the rates in both Group 1 and Control zip codes. Kidney diseases rates were less in the years 2011 – 2014 compared to the other years.

Directed Acyclic Graph (DAG) on variable associations

A simple and concise DAG was used to define the variable associations (Fig 03).

Individual-level analysis

Results of the individual-level logistic regressions, that included all six confounding factors, are summarized in Tables 02 (‘07-‘14 data) and Table 03 (‘15 – ‘18 data). In the ‘07-‘14 analysis (Table 02); 43/76 (56.6%) of the associations were statistically significant and had odds ratios (OR) >1.

For the ease of comparison to the subject of this re-analysis HIV and diabetes can be excluded as they were novel diseases analyzed only in this study. When compared the 52 associations that were common in both the studies (Kravchenko et al., 2018) and our reanalysis, previous study showed 41/52 (79%) statistically significant associations with odd ratios >1, compared to the 26/52 (50%) in our re-analysis of ‘07-‘14 data. With regard to mortalities, previous study indicated anemia, kidney disease, tuberculosis, and septicemia were statistically significant with OR >1. Specifically, they highlighted “In Group 2 zip codes, mortality ORs were 1.50 for anemia (P < 0.0001), 1.31 for kidney disease (P < 0.0001), 2.30 for septicemia (P < 0.0001), and 2.22 for tuberculosis (P = 0.0061) (Kravchenko et al., 2018).” In the reanalysis (Table 02), Group 2 mortality OR for primary and secondary diagnoses were 1.197 for anemia (p = 0.29; ns i.e. non-significant association), 1.136 for kidney disease (p=0.002), 0.870 for tuberculosis (p=0.88; ns), 1.073 for septicemia (p=0. 038), 0.797 for HIV (p=0.309; ns), and 0.863 for diabetes (p=0.19; ns). The tuberculosis logistic regression models for Group 2 associations indicated wide confidence intervals with majority of associations with no statistical significance (Tables 02 and 03). This is attributable to the low case numbers. In Group 2, for ‘07-‘14 period there were 10 cases with primary diagnosis and 27 cases with primary+secondary diagnosis. In the Group 2 matched control zip codes had 15 primary cases and 30 primary+secondary cases with tuberculosis. According to the Hosmer-Lemeshow test, all the tuberculosis models for Group 2 were not good model fits and therefore need to be interpreted with caution.

The associations and trends observed in study period of ‘07-‘14 were not held over time. For the ‘15-‘18 data (Table 03), 38/74 (51.3%) were statistically significant associations with OR >1. Of which, only 18/74 (24.3%) were matched with an OR within 10% compared to that of ‘07-‘14 associations (Table 03). From both time periods (i.e. ‘07-‘14 and ‘15-‘18), the general pattern of mortalities in Group 2 residents with primary+secondary diagnosis only resulted in statistically significant OR for septicemia (1.073 and 1.139 in Tables 02 and 03 respectively).

Testing for confounding

The odds ratios in logistic regression models for kidney disease or septicemia hospitalizations did not vary more than 10%, when the simple model was compared to the model with confounding factors (Supplementary Table 03).

Sensitivity analysis

The sensitivity analysis performed with and without the urban zip codes indicated no difference in all-cause mortalities between the groups (Supplementary table 04).

A deep dive into a previous publication on mortalities and negative health outcomes in North Carolina residents living near hog CAFOs unveiled that the study approach was insufficient to assess the intended hypothesis of associations or support the causative associations implied in media reports. The technical soundness of the previous study data and analysis had several shortcomings including ecological fallacy, residual confounding, inconsistency of associations, and over estimation of exposure. The repeatability and confidence of the associations between exposure associations (odds ratios of the logistic regression) were weak and varied between the years (Tables 02 and 03). Discounted for the fact that corrections for confounding was not successful, the six diseases selected by the previous study exhibited varying levels of associations, diseases that are not causally relatable to hog CAFOs–HIV and diabetes–were also prominent in the communities neighboring CAFOs. The associations in fact reflects systemic health disparities of the communities near hog CAFOs, which were not caused by the exposure.

An ecological study would not confirm or deny exposure over time nor indicate causation. While it is understandable that collecting individual-level data for all the confounding factors in a retrospective study is nearly impossible, that does not justify for not correctly adjusting for the key social determinants of health in the individual-level logistic regression models. Except for age, all other confounding factors used in the individual-level logistic regressions were deducted from aggregated data, which indicates a classic case of ecological fallacy. The odds ratios in regression models did not vary more than 10%, when the simple model was compared to the model with confounding factors (Supplementary Table 03), which leads to the interpretation that none of the six factors proved to be true confounding factors. This suggests the associations are not adjusted for confounding and the presence of residual confounding.

Aside the unclear rationale behind the choices, the previous study selected five disease hypothesized to have associations with hog farm exposure—anemia, kidney disease, septicemia, tuberculosis and low birth weight. However, these known associations were not supported with evidence of causation or systematic review. Especially, human tuberculosis, which is commonly caused by Mycobacterium tuberculosis and the zoonotic origins are commonly related to M. bovis; a Mycobacterium species relatable to cattle. Human tuberculosis relatable to the origin of swine is extremely rare. Tuberculosis is often reported to affect marginalized populations, including racial minorities, migrant and seasonal farmworkers (Ciesielski et al., 1994; Stout et al., 2006). Without the ability to trace back and investigate the potential exposure to tuberculosis, the claim of tuberculosis cases being high in the zip code with CAFOs is not plausible. In our repeated analysis, we included HIV and diabetes, two chronic disease conditions that are not causally relatable to hog CAFOs. Results showed that across the board, HIV and diabetes rates are higher in these North Carolina communities (Tables 02 and 03), indicating profound and systemic health disparity in the communities in general.

The variability of associations across diseases and time periods in terms of the direction of association (ORs > 1 or < 1) and P-value based statistical significance were inconsistent to make the claims by the previous study. For example when compared the mortalities between the previous study and the re-analysis of ’07-’14 data (Table 02: Group 2 primary + secondary diagnosis), only kidney disease showed a consistent statistical significant result wit OR > 1. Within our reanalysis itself, the associations for mortalities agreed between the ’07-’14 and then ’15-’18 for septicemia only. This exemplifies the danger of implying exposure associations based on a single ecological assessment of data, especially when the associations were not corrected for confounding. Across all the associations, ’07-‘14 results of kidney disease had 9 of 12 odds ratios > 1 (related to mortality, hospitalizations, and ED visits) with statistical significance, but only 5 of 12 anemia odds ratios showed these results. Five of 12 septicemia odds ratios actually showed statistically significant < 1 relationships. In fact, the results we found for HIV and diabetes were among the most consistent and significant results; all odds ratio results for hospital admissions and emergency department visits in the ’07-’14 re-analysis were > 1 and statistically significant i.e. zip codes with hog CAFOs had greater odds of having HIV and diabetes than the control group zip codes.

The key question to answer using an epidemiological approach is the counterfactual scenario whether these communities hypothetically were living in an area where there were no hog CAFOs would they still have the same negative health outcomes. A study by Son et al., (2021), conducted a comparable study to Kravchenko et al., (2018) using a spatial analysis and concluded that the communities near CAFOs are disproportionally affected. However, both the studies neglected the alternative explanation that zip codes with hog CAFOs have specific demographic characteristics with greater proportions of African American and American Indian residents, lower median household incomes, fewer residents with a bachelor’s degree education level, and fewer primary care providers. These four aspects of race, income, education, and health care demographics have all been well established as social determinants of health (Sources: health.gov links, Braveman 2011 and 2014, Catalyst 2017). North Carolina, as a state, faces multiple demographic, economic, and environmental challenges including extreme racial, gender, and income inequities (Mitchell, 2015; Wilson et al., 2002; County Health Rankings, 2021). Racially uneven development of the neighborhoods in North Carolina may have disproportionately exposed rural communities to agricultural and industrial toxicity (Bullard et al. 2007; Wilson et al. 2011; Wing et al. 2000). This emphasize the need for focusing solving health disparities in these communities, instead of falsely blaming the agriculture. Previous studies also suggested that while minorities are not directly targeted for exposure to hog farm locations, but are disproportionately exposed and that this disparity may relate to poverty and being a rural population (Yeboah et. al., 2009). Therefore, instead of ecological studies or survey-based opinion pieces that can be detrimental to the economy, we emphasize the need for conducting better epidemiological studies that are spatiotemporally explicit, and involve systematic reviews to evaluate hog CAFO exposure-over-time and evaluating the cause-effect.

Through the exercise of re-analysis, we identified a few ways to improve the assessment. First, in the absence of individual-level confounding variables, the analysis should have been conducted at zip code or county level, ideally using a spatial regression model. This is because the neighboring administrative units pose an effect on each other and accounting for such spatial dependence is vital when establishing risk factor association. Secondly, there are multiple other social determinants of health need to be considered as confounding factors and having an index that can represent overall health inequality is the best instead of collecting data from several unrelated sources (Berkowitz et al., 2015). The CDC/ATSDR Social Vulnerability Index (CDC/ATSDR, 2018) would be an example of such an indicator where socioeconomic status, household composition, minority status, housing type, transportation were all considered when the SVI index was assigned to each county. The key differences in our re-analysis and previous study are that we did not calculate the age-adjusted rate comparison tables, did not conduct ad-hoc comparisons of mortalities with other selected U.S. states, and did not repeat the DiSC analysis-based logistic regression because we cannot compare that to the geostatistical methods we proposed to better estimate the distance-based exposure.

Hazard identification and risk quantification in environmental epidemiology is challenging (Saracci, 2017). Potential environmental contamination including soil, ground water, and air contamination relatable to CAFOs is a broad topic that cannot be captured using the ecological study used. Hog farms are managed by farming systems and each farm is different in their management practices, building structure, lagoon structure, and maintenance. Therefore, the use of hog numbers is not an accurate representation of the potential exposure. If the exposure to be truly enumerated several factors are to be considered and the exposure should be assigned based on farm characteristics, length and routes of potential exposure, and the distance to the residence. This need to be further supported by air, soil, and water quality indices. Similarly potential environmental exposure through land application need to be measured differently in terms of manure consistency, hydrology, runoff, soil types, and watershed characteristics. Number of hogs would not offer a proxy for the exposure. An assessment that incorporates duration of exposure and the outcome, after correctly adjusting for the major confounding factors are needed to establish such association. Further confirmation of associations are needed to be done by comparing several similar studies (i.e. systematic revision). Compared to the Midwest where manure digestion is done in earthen deep pits, swine waste management in North Carolina is predominantly done via uncovered lagoons, where animal waste undergoes anaerobic digestion in a man-made ponds (USDA: Agricultural Waste Management Field Book, 2011). However, the swine waste disposal systems and land application of manure is highly regulated by North Carolina Department of Environmental Quality and the state legislature (https://deq.nc.gov/; Swine Farm Siting Act, 1995; General Permit, 2019). Expanded CAFOs are required to use environmentally superior technologies (ESTs) to substantially reduce emissions and prevent waste discharges into surface and ground waters (XXX Legislature XX; Nicole, 2013). Regardless of the fact that associations implied here are not well supported, several farming communities are working towards sustainable environmental goals where the health of people, animals, and the environment is being at the center of their focus (Han et al., 2001). Hog farming as an industry faces challenges of animal diseases, biosecurity, and securing the health of farm workers while making profits. Changes in farming culture is costly, labor and technology intensive. Proposing a true One Health approach where human, animal, and environmental health would be a solution for the issue where stakeholders representing the relevant government, swine producers, community members discuss and plan together multiple solutions and plans for follow-up and follow-through.

A peer-reviewed publication concluded NC communities near hog CAFOs are associated with increased negative health outcomes. The same study was repeated using updated data and concluded that the study approach was inadequate to assess the intended hypothesis of establishing an impact of CAFOs on the human mortalities in the neighborhood. The shortcomings included preconceived associations, data quality, and choice of analytics. The unfortunate reality is that based on the presence of key social determinants alone, one would already expect these communities in North Carolina who lives in close proximity to hog CAFOs to have higher rates of negative health outcomes. An ecological study would not confirm or deny exposure over time nor indicate causation. Out-of-context interpretation of their results by media arguably caused detrimental effects on the swine industry. Hence, we emphasize the importance of responsible interpretation of ecological studies that concerns public health, agriculture, and economy to support the ultimate goal of directing efforts and resources toward factors and determinants truly associated with disease.

North Carolina is a southeastern state of the United States (Latitude: 35^oN, Longitude: 79^o W). The state has wide range of elevations and landscapes including the Appalachian Mountains and the Atlantic coastal plain. The state inhabits over 10 million people, with 12.9% of persons in poverty (i.e. 11^th lowest household income rank among the 50 U.S. states) (U.S. Census Bureau, 2021). Raleigh and Charlotte, located in the central and western part of the state, are the largest metropolitan areas. Rural parts of North Carolina, especially in the mountains and in the east, are affected by persistent inter-generational poverty (Mitchell, 2015; Wilson et al., 2002; County Health Rankings, 2021).

Swine operations are largely concentrated in the southeastern region of the state where over 60% of the state’s 8.9 million hogs (NASS 2017) reside within the top five producing counties (Duplin, Sampson, Bladen, Wayne, and Greene). The swine feeding operations in North Carolina are required to obtain a permit from North Carolina Department of Environmental Quality confirming the swine waste management system comply with the state requirements (https://deq.nc.gov/; USDA: Agricultural Waste Management Field Book, 2011).

Data

As outlined in Fig 04, data on negative health outcomes, hog CAFOs, and confounding variables were collected following the data sources described by previous study (Kravchenko et al., 2018). The data sources and relevant descriptions follows.

GIS files: Zip codes

The GIS shapefile of North Carolina state and the zip code tabulation areas (n= 808 zip codes) was obtained from NC OneMap web site (NC OneMap, 2017; https://www.nconemap.gov/). It is important to note that previous (Kravchenko et al., 2018) study used a shapefile with 822 zip codes, whereas, our study used a shapefile with 808 zipcodes. This is because zip codes are divisions assigned by the U.S. postal services and are subjected to change over time.

Hog CAFOs

Hog CAFO data were extracted from the North Carolina Department of Environmental Quality web site (https://deq.nc.gov/; Animal Facility Map, 2020). The data included CAFO permits (n total=2,614) for cattle (n=250), wet and dry poultry (n=20), horses (n=5), manure hauler (n=88), swine (n=2,248), and other animals (n=2). The data contained farm location, allowable count of animals, number of lagoons, and details related to permit issuance and expiration. Among the registered swine CAFOS, 175 were duplicated. The records with larger head count and/or number of lagoons were included in the analysis. 2,072 hog CAFO locations were included in mapping process. However, among these, 59 CAFO locations did not contain allowable counts and therefore were not used in the hog density calculation. The hog operations in North Carolina were defined as ‘any agricultural feedlot activity involving 250 or more swine with a waste management system, or any agricultural feedlot activity with a liquid animal waste management system that discharge to the surface waters of the state’.

Human disease data

Negative health outcome data on mortality, hospitalizations, and emergency department visits were obtained at person/individual-level from the Healthcare Cost and Utilization Project’s (H-CUP) database (https://www.H-CUP-us.ahrq.gov/). The H-CUP data included State Emergency Department Database (SEDD) (SEDD, 2021) and State Inpatient Database (SID) (SID, 2021) data from 2007 – 2018. The SID and SEDD data were checked for any duplicated records and were merged into one database to be used in the regression analysis. The international classification of diseases (ICD) codes used when choosing the selected disease conditions are listed in Supplementary Table 02. H-CUP data were cleaned to remove any duplicated or erroneous records. It is important to note the difference between ICD-9 and ICD-10 code changes and it may affect the disease diagnostics extracted from the H-CUP database.

Previous study used data from CDC Wonder (Kravchenko et al., 2018; http://wonder.cdc.gov/) to obtain all-cause mortalities, compare mortalities with other states, and to identify major causes of mortalities (Table 2 of Kravchenko et al., 2018). Given that we will only repeat the study focusing on the North Carolina state health data, and not comparing across the data sources, we will use all-cause mortality data obtained solely from H-CUP database.

Confounding variables

Data on the following six confounding factors were gathered: age, education, median household income, health insurance coverage, number of primary care providers, and smoking prevalence. Except for age, which was available from H-CUP database, the data for other confounding variables were not available at the individual-level. In the absence of these social determinants of health, previous study assumed a person’s median household income, education level, insurance coverage, smoking prevalence, and the number of primary care physicians is same as everyone else in the zip code they live. Therefore, following the same methods described by previous study (Kravchenko et al., 2018), available data were gathered at different granularities (Table 01), and the inference about the individual’s nature was deduced based on the aggregated data at zip code or the county the individual resides. While this assumption represents the epidemiological phenomenon known as ‘ecological fallacy’, given the primary objective was to repeat the previous study (Kravchenko et al., 2018) the same approach was used. Ecological fallacy occurs when conclusions are made about individuals using averaged combined data from larger geographical areas (zip codes, counties, etc.) and assumption about an individual was made based on average data obtained for a group to which the individual belongs (Rothman, 1986; Portnov et al., 2007). Ecological fallacy can lead to misleading conclusions about associations of exposures and outcomes. Zip code-level data on median household income and education, defined as the percentage of people aged 25+ who attained an education level higher than a bachelor’s degree were obtained from the 2011 – 2016 American community survey (ACS, 2020).

County-level data on percentage of those who are under 65 years of age without health insurance (F1475110 - F1547418) and number of primary care providers, specifically physicians (F1467510 - F1467518), were obtained from the Area Health Resources Files (AHRF) for the years 2010 – 2018, values were averaged across the years (AHRF, 2021). The data were used to calculate the average of primary care providers per 100,000 people in each county. Given the definition for ‘primary care providers’ used in the previous study (Kravchenko et al., 2018) was not mentioned, in this re-analysis, only physicians were used as per the definition provided by Healthcare.gov was used (Specific variable name: Phys, Primary Care, Patient Care; Healthcare.gov).

While no further details were described, (Kravchenko et al., 2018) claims to have obtain current smoking prevalence at county-level using Behavioral Risk Factor Surveillance System (BRFSS) data (BRFSS CDC, 2020). BRFSS surveys are a random digit–dial telephone (cellular and landline) survey of noninstitutionalized adults aged 18 years or older, conducted through collaborations between the Centers for Disease Control and Prevention (CDC) and all 50 US states. As per the published literature, unless extraneous steps such as those by Song et al., (2016), which uses a Bayesian geostatistical approach to estimate the smoking prevalence using BRFSS survey data, it is not straightforward to calculate smoking prevalence using BRFSS data. Thus, in this re-analysis study, we used the model-based county-level estimates of ‘percentage of adults who are current smokers (age adjusted)’ available from County Health Rankings database (www.countyhealthrankings.org; PLACES project; Adult smoking). Specifically, the county-level smoking prevalence estimates for 2014 (which covered the BRFSS survey data from 2006 – 2012) and 2017 were downloaded.

Classifying zip code areas for comparison

The hog density per square kilometer was calculated by dividing the number of hogs by the area of the zip code. Comparable to the previous study (Kravchenko et al., 2018), the zip codes were divided into three study groups based on the hog density. The zipcodes with upper quartile of hog density ( with > 234 hogs per sqkm) was defined as Group 2; zip codes with ≥ 1 hogs per sqkm defined as Group 1; and the zip codes with zero hog density were considered as the control group. Notice that, Group 2 was included in Group 1. All zip codes without CAFOs were the control group for Group 1. Previous study justified their reason for separating out Group 2 was to highlight the magnitude of the effects. The control zip codes for Group 2 were matched at 1:2 ratio based on propensity score using demographic and socioeconomic characteristics (Supplementary Fig. 01). For the step of matching control zip codes for Group 2, a generalized linear model was fitted to obtain the propensity scores. The greedy matching algorithm was used in the matching process, similar to the approach by previous study (Kravchenko et al., 2018).

Five matching variables that are compatible to the previous study (Kravchenko et al., 2018) were used. The factors used when matching the Group 2 control zip codes by the previous study (Kravchenko et al., 2018) included percent of non-Hispanic or Latino black or African-Americans, percent of children (aged 0–19) and people aged 65 + among the residents, median household income, and percent of people with bachelor or higher degree in people aged 25+. All these were continuous variables and they were categorized in to five based on natural breaks (Jenks; Jenks, 1967) criteria and the zipcodes of Group 2 (n = 53) were matched with two control zip codes (n = 106). The stratification of the continuous variable is compatible to stratification of propensity score approach, which is commonly used in observational epidemiological studies (Cochran, 1968; Huppler Hullsiek and Louis, 2002; Austin, 2011).

However, the reasoning for selecting these factors as matching variables was unclear in the publication by (Kravchenko et al., 2018). To obtain unbiased and precise estimation, it is essential to identify matching variables correctly and determine whether the adjustment is needed (Li at el., 2016). Previous study summarized the age-adjusted rates (per 100,000) for all the negative health outcomes comparing the numbers for Group 1, Group 2, and the controls, here, instead of tabulation of these numbers, we summarized the disease data by year for the whole state.

Directed Acyclic Graph (DAG) on variable associations

A simple and concise DAG was used to clarify the exposure, outcome, and confounding variables. Confounding factors considered could either increase or decrease bias in the presence or absence of a true association between exposure and outcome. As done in previous study (Kravchenko et al., 2018), the exposure was the presence of CAFOs in the resident’s zip code. Under the outcome of negative health outcomes mortality, hospitalization, and emergency department visits related to all causes, infant mortality, anemia, tuberculosis, HIV, and diabetes were listed.

Estimation of the risk of exposure based on distance

In the ecological study design followed by previous study, the exposure to CAFOs was measured using two methods: 1) presence of CAFOs in the resident’s zip code, and 2) assigning the exposure risk as a function of distance from the CAFOs while accounting for hog number and the human population. Previous study named this distance based risk calculation ‘Distance from the Source of potential Contamination (DiSC)’ analysis (Kravchenko et al., 2018). In DiSC analysis, researchers hypothesized that the risk of negative health outcome is proportional to the number of hogs and calculated the risk of exposure using the number of hogs allowed per farm and the distance from the farm to the center of the census block codes with in each zip code. The calculation also assumed all people live at the center of the census block and normalized the calculation using the human population counts. Inclusion of human population in the calculation makes an artificial weight in the highly populated zipcodes and overestimate the exposure. Instead, exposure to an environmental contaminant is commonly quantified using distance-decay functions and there are well-established geostatistical techniques to perform such analysis.

We compared DiSC analysis with two well-established geostatistical methods; 1) point density estimation weighted by the number of hogs in each farm and 2) geostatistical interpolation technique called co-Kriging. Point density analysis generates raster cells in which the neighborhood values are calculated using the points (farms in this case) falls within the defined cell size. Kriging can be understood as a two-step process, where, step 1 is fitting the spatial variogram or likelihood for the data observed at the sampled points; and step 2 involves the interpolation of values for unsampled points using the weights derived from this covariance structure (Isaaks and Srivastava, 1989; Kanankege et al., 2020). In co-Kriging, we used both number of hogs and the number of lagoons to calculate the distance based exposure.

In the context of this study, the co-Kriging estimates can be defined as a spatially continuous variable that represent the risk of exposure to CAFOs as a function of distance from all the CAFOs, in which the hog numbers and number of lagooons at close proximity would contribute to the exposure estimate more than those that are distant. There was no mapped output of DiSC analysis in the previous publication to compare with the geostatstical techniques we used. Considering the differences between DiSC analysis and Kriging, in our analysis, we focused only on the exposure assigned based on the presence of CAFOs in the individual’s residential zip code, i.e. the main predictor variable of the regression models. The resulting maps of the two geostatistical methods of point density and co-Kriging are presented in supplementary documents (Supplementary Figs. 03 and 04).

Individual-level logistic regression

A series of individual-level multiple logistic regressions were conducted, similar to previous study (Kravchenko et al., 2018). The main predictor variable was the presence of CAFOs in the zip code, defined as Group 1, Group 2, or controls. The control group was used as the reference to calculate odds ratios (OR). The outcome encompassed several negative health outcomes including hospitalizations, emergency department visits, and mortalities related to all causes, infant mortality, anemia, kidney diseases, septicemia, tuberculosis, low birthweight of infants, HIV, and diabetes. Negative health outcomes were available at individual level from the H-CUP database. Analyses were performed for the underlying cause of death (i.e. primary diagnosis) and for underlying-plus-secondary causes of death (primary-plus-secondary) diagnoses listed in the H-CUP database. The exposure and confounding factor variables were available at zipcode or county level (see Table 01). A case - control matched conditional logistic regression was used to compare Group 2 zip codes with the selected and matched control zip codes. The Proc logistic procedure statement of SAS 9.4 software (SAS Institute, Carry NC) was used to conduct the regression analysis and estimate OR, 95% confidence intervals (CI), and frequentist statistical p-values (See Supplementary documents: SAS codes). Previous publication did not explicitly discussed their approach on addressing low case numbers. In the re-analysis, when encountered with low case numbers (n < 30), specifically for tuberculosis, the Proclogit models were evaluated using Hosmer-Lemmeshow goodness-of-fit test.

One of the goals of the re-analysis was to add data from additional years and determine if the associations and trends observed by previous study (data from 2007–2013) would hold over time. Therefore, H-CUP data of 2007–2018 was split into two periods: 1.) 2007–2014 and 2) 2015–2018. This split year was chosen for ease of comparison with previous study and that ICD codes underwent a formatting change in 2015: ICD-9 to ICD-10.

Testing for confounding

The logistic regression model with hospitalizations due to kidney disease and septicemia were introduced with the six confounding variables described in the 2018 publication, first one by one and then all six variables together to test for changes in odds ratio with and without them. The resulting OR were compared to the base model, and a change of ≥ 10% of the OR were considered as a potential confounder (Westreich and Greenland, 2013). By forcing the six variables that were considered as potential confounders into the logistic regression, a conditional logistic regression was fitted to obtain conditional rather than causal estimation of odds ratio (Breslow and Day, 1980).

Sensitivity analysis

A sensitivity analysis was performed by removing the zip codes in cities and urban areas and performing the regression analyses, then the OR were compared to that of the complete dataset. The zip codes in cities (Raleigh and Charlotte) and urbanized areas where the population was ≥ 50,000 were excluded from the individual-level and zip code-level analysis (n = 70 zip codes were removed). Previous study reasoned that this exclusion was because CAFOs are predominantly located in rural North Carolina and the access to medical case likely is different compared to cities and urban areas.

Acknowledgement

This project was funded in part by a grant provided by the National Pork Board (NPB; grant ID: CON000000088038). The authors would also like to acknowledge the openness of authors form the previous study, Drs. Kravchenko and Akushevich, to share the analytical details of their study.

Conflict of interest

Authors of the 2018 publication titled “Mortality and Health Outcomes in North Carolina Communities Located in Close Proximity to Hog Concentrated Animal Feeding Operations” of the Duke University including Drs Kravchenko J, Rhew SH, Akushevich I, Agarwal P, Lyerly HK may have potential conflict of interest on this re-evaluation.

Ethics statement

The Institutional Review Board of the University of Minnesota determined that the research activity of re-evaluation of Kravchenko et al., (2018) does not involve human subjects as defined by DHHS and FDA regulations (IRB ID: STUDY00010813). De-identified health data were obtained from the H-CUP database and were stored and analyzed in accordance with the data security measures determined by the University of Minnesota.

Author contributions

KK contributed to the research design, and conducted the analysis, compiled the manuscript. IT conducted the data cleaning and analysis on SAS and edited the manuscript. AP proposed the research design, and contributed to the results interpretation and discussion.

Data availability statement

The hog CAFO data are open accessible via North Carolina Department of Environmental Quality web site (https://deq.nc.gov/). The data on confounding factors are all available open access from relevant data sources listed in Table 01. The health data are available for purchase via the Healthcare Cost and Utilization Project’s (H-CUP) database (https://www.H-CUP-us.ahrq.gov/). The exact open-source data that were used in the study can be provided on request.

USDA Natural Resources Conservation Services. 2021. Animal Feeding Operations (AFO) and Concentrated Animal Feeding Operations (CAFO). Online access: https://www.nrcs.usda.gov/wps/portal/nrcs/main/national/plantsanimals/livestock/afo/.
United States Department of Agriculture (USDA). Agricultural Waste Management Field Book. Chapter 9. Part 651. Natural Resources Conservation Service. 9-18 (2011). Online access: https://directives.sc.egov.usda.gov/viewerfs.aspx?hid=21430.
USDA: Hogs and Pork. Economic Research Service. Online access: https://www.ers.usda.gov/topics/animal-products/hogs-pork/. Last access: July, 20^th 2020.
USDA National Agricultural Statistics Service. Hogs and Pigs – Inventory. Census of Agriculture. (2017) Online access: www.nass.usda.gov/AgCensus
American Community Survey. 2020. Median HH income and education. Online access: https://www.census.gov/programs-surveys/acs/guidance/comparing-acs-data/2019.html
Rothman, K.J. (1986) Modern Epidemiology. 5. Measures of Disease Frequency, 23-34. Little, Brown and Company. Boston/Toronto.
Wing, S., Cole, D. & Grant, G. Environmental injustice in North Carolina's hog industry. Environmental Health Perspectives 108, 225-231, doi:10.2307/3454438 (2000).
Wing, S., Freedman, S. & Band, L. The potential impact of flooding on confined animal feeding operations in eastern North Carolina. Environmental Health Perspectives 110, 387-391, doi:10.1289/ehp.02110387 (2002).
Wing, S. Social responsibility and research ethics in community-driven studies of industrialized hog production. Environmental Health Perspectives 110, 437-444, doi:10.1289/ehp.02110437 (2002).
Wing, S. et al. Air pollution and odor in communities near industrial swine operations. Environmental Health Perspectives 116, 1362-1368, doi:10.1289/ehp.11250 (2008).
Kravchenko, J., Rhew, S.H., Akushevich, I., Agarwal, P. & Lyerly, H.K. Mortality and health outcomes in North Carolina communities located in close proximity to hog concentrated animal feeding operations. North Carolina Medical Journal. 79(5), 278-288. doi:10.18043/ncm.79.5.278 (2018).
Smit, L. A. M. et al. Air pollution from livestock farms, and asthma, allergic rhinitis and COPD among neighbouring residents. Occupational and Environmental Medicine 71, 134-140, doi:10.1136/oemed-2013-101485 (2014).
Nicole, W. CAFOs and Environmental Justice The Case of North Carolina. Environmental Health Perspectives 121, A182-A189, doi:10.1289/ehp.121-a182 (2013).
Jepsen, P., Johnsen, S. P., Gillman, M. W. & Sorensen, H. T. Interpretation of observational studies. Heart 90, 956-960, doi:10.1136/hrt.2003.017269 (2004).
Ciesielski, S., Esposito, D., Protiva, J. & Piehl, M. The incidence of tuberculosis among North Carolina migrant farmworkers, 1991. American Journal of Public Health. 84, 1836-1838, doi:10.2105/ajph.84.11.1836 (1994).
Stout, J. E., Saharia, K. K., Nageswaran, S., Ahmed, A. & Hamilton, C. D. Racial and ethnic disparities in pediatric tuberculosis in North Carolina. Archives of Pediatrics & Adolescent Medicine. 160, 631-637, doi:10.1001/archpedi.160.6.631 (2006).
Braveman, P. & Gottlieb, L. The social determinants of health: It's time to consider the causes of the causes. Public Health Reports 129, 19-31, doi:10.1177/00333549141291s206 (2014).
Catalyst, N. E. J. M. (2017). Social determinants of health (SDOH). NEJM Catalyst, 3(6). Online access: https://catalyst.nejm.org/doi/full/10.1056/CAT.17.0312?targetBtn=articleToolSaveBtn
Mitchell T. North Carolina’s Greatest Challenge: Widespread struggles remain grave threat to economic growth and us all. 2015. North Carolina Justice Center. https://www.nccourts.gov/assets/inline-files/civil-justice-BTC-Report_North-Carolinas-Greatest-Challenge.pdf? (2015).
Anonymous. Social determinants of health (SDOH). NEJM Catalyst, 3(6). Online access: https://catalyst.nejm.org/doi/full/10.1056/CAT.17.0312?targetBtn=articleToolSaveBtn
Mitchell T. North Carolina’s Greatest Challenge: Widespread struggles remain grave threat to economic growth and us all. 2015. North Carolina Justice Center. https://www.nccourts.gov/assets/inline-files/civil-justice-BTC-Report_North-Carolinas-Greatest-Challenge.pdf? (2015).
Wilson, S. M., Howell, F., Wing, S. & Sobsey, M. Environmental injustice and the Mississippi hog industry. Environmental Health Perspectives 110, 195-201, doi:10.1289/ehp.02110s2195 (2002).
NC County Health Rankings. Premature death. Year of potential life lost before age 75 per 100,000 population (age adjusted). The 2021 County Health Rankings used data from 2017 – 2019 for this measure. https://www.countyhealthrankings.org/app/north-carolina/2021/measure/outcomes/1/map (2021).
Centers for Disease Control and Prevention/ Agency for Toxic Substances and Disease Registry/ Geospatial Research, Analysis, and Services Program. CDC/ATSDR Social Vulnerability Index. 2018. Database: North Carolina. Online access: https://www.atsdr.cdc.gov/placeandhealth/svi/data_documentation_download.html. Accessed on February 25, 2022.
Centers for Disease Control and Prevention (CDC). Behavioral Risk Factor Surveillance System Survey Data. Atlanta, Georgia: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention (2020).
Yeboah, O., Thomas, T., Foster, T. & Fosu, E. Pollution, Environmental Justice, and the North Carolina Pork Industry. Proceedings of the 2007 National Conference on Environmental Science and Technology, 127-137, doi:10.1007/978-0-387-88483-7_18 (2009).
Kanankege, K. S. T., Alvarez, J., Zhang, L. & Perez, A. M. An introductory framework for choosing spatiotemporal analytical tools in population-level eco-epidemiological research. Frontiers in Veterinary Science. 7, doi:10.3389/fvets.2020.00339 (2020).
Saracci, R. The hazards of hazard identification in environmental epidemiology. Environmental Health. 16, doi:10.1186/s12940-017-0296-3 (2017).
Animal Facility Map. List of permitted animal facilities - 4-1-2020. Last accessed: December 25, 2020. https://deq.nc.gov/about/divisions/water-resources/water-resources-permits/wastewater-branch/animal-feeding-operation-permits/animal-facility-map (2020).
Area Health Resources Files. (AHRF) 2019-2020. US Department of Health and Human Services, Health Resources and Services Administration, Bureau of Health Workforce, Rockville, MD. Last accessed: February 19, 2021. https://data.hrsa.gov/topics/health-workforce/ahrf (2021).
Austin, P. C. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behavioral Research 46, 399-424, doi:10.1080/00273171.2011.568786 (2011).
Berkowitz SA, Traore CY, Singer DE, Atlas SJ. Evaluating area-based socioeconomic status indicators for monitoring disparities within healthcare systes: results from a primary care network. Health Serv Res. 50(2): 398–417. doi: 10.1111/1475-6773.12229 (2015).
Braveman, P., Egerter, S. & Williams, D. R. The social determinants of health: coming of age. Annual Review of Public Health, Vol 32, 381-398, doi:10.1146/annurev-publhealth-031210-101218 (2011).
Breslow N.E. & Day NE. Conditional logistic regression for matched sets. Statistical Methods in Cancer Research. 1, 248–79 (1980).
Cochran, W. G. The effectiveness of adjustment by subclassification in removing bias in observational studies. Biometrics 24, 295-313, doi:10.2307/2528036 (1968).
Westreich, D. & Greenland, S. The Table 2 Fallacy: Presenting and Interpreting Confounder and Modifier Coefficients. American Journal of Epidemiology 177, 292-298, doi:10.1093/aje/kws412 (2013).
Definitions. North Carolina General Statute 143-215.10B. Available online: https://www.ncleg.net/enactedlegislation/statutes/html/bysection/chapter_143/gs_143-215.10b.html. Last accessed: December 25, 2020.
Department of Environmental Quality (DEQ). Swine waste management system: General permit. 2019. North Carolina environmental management commission. North Carolina Administrative Code 15A. Available online: https://deq.nc.gov/media/15424/download. Last updated: April 12, 2019. Last accessed January 17, 2022.
Freese B. 2020. What farmers can learn from the hog nuisance suits in North Carolina. Successful Farming. Online: https://www.agriculture.com/livestock/pork-powerhouses/what-farmers-can-learn-from-the-hog-nuisance-suits-in-north-carolina. [Accessed: June 15, 2020]
Bullard, R. D. Equity, unnatural man-made disasters, and race: Why environmental justice matters. Equity and the Environment 15, 51-85, doi:10.1016/s0196-1152(07)15002-x (2007).
Han, I. K., Lee, J. H., Piao, X. S. & Li, D. F. Feeding and management system to reduce environmental pollution in swine production - Review. Asian-Australasian Journal of Animal Sciences 14, 432-444, doi:10.5713/ajas.2001.432 (2001).
Portnov, B. A., Dubnov, J. & Barchana, M. On ecological fallacy, assessment errors stemming from misguided variable selection, and the effect of aggregation on the outcome of epidemiological study. Journal of Exposure Science and Environmental Epidemiology 17, 106-121, doi:10.1038/sj.jes.7500533 (2007).
Hullsiek, K. H. & Louis, T. A. Propensity score modeling strategies for the causal analysis of observational data. Biostatistics 3, 179-193, doi:10.1093/biostatistics/3.2.179 (2002).
Isaaks, E.H. & Srivastava, R.M. Applied Geostatistics. New York, NY: Oxford University Press (1989).
NC OneMap. ZIP code Tabulation Areas. Last updated June 21, 2018. Last accessed: December 25, 2020. https://services.nconemap.gov/secure/rest/services/NC1Map_Boundaries/MapServer/1 (2017)
Jenks GF. The data model concept in statistical mapping. International Yearbook of Cartography. 7;186–90. (1967).
Overview of the State Emergency Department Databases (SEDD). Rockville, MD: Healthcare Cost and Utilization Project (H-CUP); 2007-2018; https://www.H-CUP-us.ahrq.gov/seddoverview.jsp. Accessed: Jan 15th, 2021.
Overview of the State Inpatient Databases (SID). Rockville, MD: Healthcare Cost and Utilization Project (H-CUP); 2007-2018; https://www.H-CUP-us.ahrq.gov/sidoverview.jsp. Accessed Jan 15th, 2021.
PLACES Project. Centers for Disease Control and Prevention. Accessed March 9, 2021. https://www.cdc.gov/places.
54. SAS Institute Inc. SAS/ACCESS® 9.4 Interface to ADABAS: Reference. Cary, NC: SAS Institute Inc. (2020).
Son, J.Y., Muenich, R.L., Schaffer-Smith D., Miranda, M.L. & Bell, M.L. Distribution of environmental justice metrics for exposure to CAFOs in North Carolina, USA. Environmental Research. 198, 110862. https://doi.org/10.1016/j.envres.2021.110862 (2021).
Song, L., Mercer, L., Wakefield, J., Laurent, A. & Solet, D. Using small-area estimation to calculate the prevalence of smoking by subcounty geographic areas in King County, Washington, behavioral risk factor surveillance system, 2009-2013. Preventing Chronic Disease 13, doi:10.5888/pcd13.150536 (2016).
Wing, S. Social responsibility and research ethics in community-driven studies of industrialized hog production. Environmental Health Perspectives 110, 437-444, doi:10.1289/ehp.02110437 (2002).
United States Census Bureau. Population Census: North Carolina. Population estimates as of July 1st 2021. https://www.census.gov/quickfacts/NC (2021).
Swine Farm Siting Act. 1995, c. 420, s. 1; 1995 (Reg. Sess., 1996), c. 626, s. 7(a); 1997-458, s. 4.1. https://files.nc.gov/ncdeq/Water%20Quality/Aquifer%20Protection/AFO/Statutes%20and%20Rules/Swine%20Farm%20siting-2011update.pdf (1995).
Li, H. K. et al. A simulation study on matched case-control designs in the perspective of causal diagrams. Bmc Medical Research Methodology 16, doi:10.1186/s12874-016-0206-3 (2016).
Han, I. K., Lee, J. H., Piao, X. S. & Li, D. F. Feeding and management system to reduce environmental pollution in swine production - Review. Asian-Australasian Journal of Animal Sciences 14, 432-444, doi:10.5713/ajas.2001.432 (2001).
Hullsiek, K. H. & Louis, T. A. Propensity score modeling strategies for the causal analysis of observational data. Biostatistics 3, 179-193, doi:10.1093/biostatistics/3.2.179 (2002).
NC General Statute 143-215.10B. available on line at https://www.ncleg.gov/EnactedLegislation/Statutes/PDF/BySection/Chapter_143/GS_143-215.10C.pdf

Table 01

Summary of variables and data sources.

Variable	Time	Granularity	Source	Attributes/ Comments
CAFOs	2020	Farm-level	NC Dept. of Environmental quality (DEQ)	Farm location, number of animals, number of manure lagoons
Negative health outcomes	2007 - 2018	Individual-level	The Healthcare Cost and Utilization Project (HCUP) - Emergency dept: SEDD - SID: Hospitalizations	Age, gender, date, diagnosis, resolution
Confounding factors
- Age	2007 – 2018	Individual-level	HCUP
- Household income	2011 - 2016	Zipcode-level	American Community Survey
- Education	2011 - 2016	Zipcode-level	American Community Survey
- Health insurance	2010, 2015 - 2018	County-level	Area Health Res. Files (AHRF)
- Primary-care	2010 - 2018	County-level	Area Health Res. Files (AHRF)
- Smoking prevalence	2005 – 2012, 2017	County-level	Behavioral Risk Factor Surveillance System (BRFSS); County Health Rankings	Model-based estimates of adult smoking prevalence (https://www.countyhealthrankings.org/

Table 02

Results of the logistic regression models. Odds ratios and 95% confidence intervals of mortality, hospital admissions, and emergency department visits for selected disease conditions from 2007-2014 in North Carolina communities are summarized. Primary and secondary diagnosis listed in the H-CUP database (https://www.H-CUP-us.ahrq.gov/) were analyzed. Study Group 1) represents North Carolina communities in zip codes with >1 hogs/km², Study Group 2) > 234 hogs/km², and zip codes without hog CAFOs were the controls (reference).

#=Compatible result (<10% change) with Kravchenko et al. 2018.

Table 03

Results of the logistic regression models. Odds ratios and 95% confidence intervals of mortality, hospital admissions, and emergency department visits for selected disease conditions from 2015 - 2018 in North Carolina communities are summarized. Primary and secondary diagnosis listed in the H-CUP database (https://www.H-CUP-us.ahrq.gov/) were analyzed. Study Group 1) represents North Carolina communities in zip codes with >1 hogs/km², Study Group 2) > 234 hogs/km², and zip codes without hog CAFOs were the controls (reference).

#=Compatible result (<10% change) with 2007-2014 results (Table 2)

No competing interests reported.

02272022NCpublichealthstudyD1sup.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Importance of responsible interpretation of observational studies with an impact on public health policy

Status:

Version 1

Abstract

Figures

Introduction

Results

Zip codes classified by exposure

The negative health outcomes

Directed Acyclic Graph (DAG) on variable associations

Individual-level analysis

Testing for confounding

Sensitivity analysis

Discussion

Conclusion

The Study Area

Data

GIS files: Zip codes

Hog CAFOs

Human disease data

Confounding variables

Methods

Classifying zip code areas for comparison

Directed Acyclic Graph (DAG) on variable associations

Estimation of the risk of exposure based on distance

Individual-level logistic regression

Testing for confounding

Sensitivity analysis

Declarations

Acknowledgement

Conflict of interest

Ethics statement

Author contributions

Data availability statement

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1