Health Disparities in Staphylococcus aureus Transmission and Carriage in a Border Region of the United States Based on Cultural Differences in Social Relationships: Protocol for a Survey Study

Background Health care–associated Staphylococcus aureus infections are declining but remain common. Conversely, rates of community-associated infections have not decreased because of the inadequacy of public health mechanisms to control transmission in a community setting. Our long-term goal is to use risk-based information from empirical socio-cultural-biological evidence of carriage and transmission to inform intervention strategies that reduce S aureus transmission in the community. Broad differences in social interactions because of cultural affiliation, travel, and residency patterns may impact S aureus carriage and transmission, either as risk or as protective factors. Objective This study aims to (1) characterize S aureus carriage rates and compare circulating pathogen genotypes with those associated with disease isolated from local clinical specimens across resident groups and across Hispanic and non-Hispanic white ethnic groups and (2) evaluate social network relationships and social determinants of health-based risk factors for their impact on carriage and transmission of S aureus. Methods We combine sociocultural survey approaches to population health sampling with S aureus carriage and pathogen genomic analysis to infer transmission patterns. Whole genome sequences of S aureus from community and clinical sampling will be phylogenetically compared to determine if strains that cause disease (clinical samples) are representative of community genotypes. Phylogenetic comparisons of strains collected from participants within social groups can indicate possible transmission within the group. We can therefore combine transmission data with social determinants of health variables (socioeconomic status, health history, etc) and social network variables (both egocentric and relational) to determine the extent to which social relationships are associated with S aureus transmission. Results We conducted a first year pilot test and feasibility test of survey and biological data collection and analytic procedures based on the original funded design for this project (#NIH U54MD012388). That design resulted in survey data collection from 336 groups and 1337 individuals. The protocol, described below, is a revision based on data assessment, new findings for statistical power analyses, and refined data monitoring procedures. Conclusions This study is designed to evaluate ethnic-specific prevalence of S aureus carriage in a US border community. The study will also examine the extent to which kin and nonkin social relationships are concordant with carriage prevalence in social groups. Genetic analysis of S aureus strains will further distinguish putative transmission pathways across social relationship contexts and inform our understanding of the correspondence of S aureus reservoirs across clinical and community settings. Basic community-engaged nonprobabilistic sampling procedures provide a rigorous framework for completion of this 5-year study of the social and cultural parameters of S aureus carriage and transmission.

in reaching "hard to reach" populations in Yuma county, using the type of "targeted sampling" frameworks that are commonly used in these types of studies. Each of these partners has extensive outreach programs and conducts or accommodates sampling of all of the relevant populations (Anglo, Hispanic, resident, snowbird) throughout the county. Our Center for Health Equities Research, and our core staff have created a Yuma community advisory group that has significant exptertise in identifying both locations and timing for sampling the populations in Yuma that will lessen response bias in our data. The first 3 months of the project will include designing and testing the final sampling frame for the research, within the context of meeting all of the scientific goals of the project. Sampling will also be conducted throughout the week to reduce any temporal fluctuations. We recognize that our methods will miss certain people, such as those in short and longterm healthcare facilities, however transmission is relatively well-understood and controlled in these settings and our aim here to focus specifically on community based transmission. Response 3:. Our goal is to use empirical evidence and reported levels of contact to evaluate the likelihood of transmission. We anticipate that the reported relationship type (e.g. friends, brother/sister or mother/son) will correlate with levels of contact, and in the final analysis we may produce a weighted or adjusted level of contact that allows for a more refined test of this hypothesis, but we feel that a priori adjustments based on relationship type might lead to a bias and less accuracy, given the variation in contact levels within specific relationship types. Response 4: We intend to adjust our model for possible confounders such as age, sex, and comorbidities to more clearly determine the effects of such independent variables on the dependent variable (transmission). These will collected by the data collection instrument provided to each participant. Response 5: We will be working closely with infectious disease specialists at both the YRMC and RCBH. In particular, Dr. P. Bhatt MD, (Internal Medicine) who is the Medical Director for RCBH/San Luis Walk In Clinic. We also work closely with Dr. J. Terriquez at the Flagstaff Medical Center on a separate project. We therefore have access to such expertise in Yuma and in Flagstaff. This project is not designed to address transmission in a healthcare setting, however we do have multiple contacts with such expertise that can help guide us as we identify factors that are important for carriage and transmission in the community.

Significance
This project is focused on understanding mechanisms of the disparity in MSRA/MSSA infections among Hispanics and whites living in and around the Yuma Arizona areas. The ultimate objective of this proposal is that enhanced understanding of the disparity may lead to new intervention strategies.

Strengths
Ethnic differences in MRSA infections rates are higher with Native Americans and lower with Hispanics than the general population. The basis of this disparity is unknown and data on whether clinical-based differences (infection rates and pathogen genotype) among ethnic groups extend into their communities in the form of non-symptomatic MSSA/MRSA carriage are lacking. The applicant proposes to identify ethnic differences in MRSA and MSSA carriage rates and the role of social behaviors in transmission. This project is focused on a community setting.

Weaknesses
More information on the impact of community associated MRSA/MSSA infections in the local targeted community would have helped to place the significance of this study in context. Response 6: See Response 1.

Investigator(s) Strengths
Investigators are well suited to perform these studies Weaknesses Need contribution from infectious disease specialist on this project.

Innovation Strengths
Approaches for limiting new and recurrent MRSA infections are proven strategies in a hospital environment but have had limited success in a community environment. This proposal will use genomic based phylogenetic hypotheses overlaid with defined and quantified social contact information to better understand how social interactions impact transmission and ultimately guide intervention models to limit transmission.

Weaknesses
While this research represents the union of bacterial pathogen genome-based epidemiology with social contact analysis, social contact data have been useful in the control of sexually transmitted disease, which limits the innovative potential of this approach. However, it is noted that this approach could prove effective. Response 8: We agree. These individual approaches are highly validated and have been highly effective for controlling different sexually transmitted diseases even though they have not been combined to address other bacterial-based diseases. Part of the innovation here is in the technological (translational) transfer of evidence based techniques to a new disease venue. One note that is relevant, is that these contact tracing network intervention designs have primarily been used in both hard to reach and heavily stigmatized populations (both population sigma and disease stigma). Since MRSA/MSSA are serious public health problems, but are not heavily stigmatized, this may be in important opportunity to determine whether or not the successful model for contact tracing and intervention is, minimally, as effective, more effective, or less effective in relation to non-stigmatized diseases.
 Applicant proposes to recruit and consent over a four months' window each year, individuals from 367 social groups (family/friendship clusters) in public settings and attending public events in Yuma and enroll them into the study for biological sampling and administration of the social instrument.

Weaknesses
Longer sampling periods will improve project Approach. Response 9: Under different circumstances, longer sampling periods would indeed improve the likelihood of capturing the temporal dynamics of S. aureus carriage and transmission. However in this study, we are limited to a four month window because major portions of our subject populations (snowbirds and agricultural workers) are predominantly present in Yuma during the winter months. The majority of both the snowbirds and the migrant farmworker populations exit during the hottest months, and consequently the population in Yuma is cut in half during the summer months.

Environment Strengths
The environment at NAU is well suited to support this study Weaknesses None noted

Significance Strengths
The project investigates the question "Do Pathogen Genotypes, Carriage, and Social Network Differences Lead to Health Disparities in MRSA/MSSA Infections?". The goal is to determine whether national trends in infection rates and asymptomatic carriage are reflected within different ethnic groups in Yuma, Arizona; to determine if clinical strains are representative of community-carriage strains and not due to the emergence of a few, highly fit lineages; and to determine the role that social relationships and interactions have on S. aureus transmission. The project addresses an important public health concern given the mortality associated with MRSA and MSSA infections. These infections are common, with about one third of the healthy US population thought to be asymptomatic carriers of S. aureus and 1.5% being carriers of MRSA. Infection rates are particularly high among American Indian populations, and are lower among Hispanic populations, which relates to health equity objectives of the Collaborative. Differences among communities are not well understood and community-based infections are a growing public health concern. The proposed project meets a health priority area for the community; Yuma County Health District has identified MRSA as one of the top five priorities for infectious disease control in their latest Community Health Improvement Plan.

Weaknesses
The potential for selection bias (resulting from the community-based events/locations of the sampling) and confounding is a concern that limits the inference that can be drawn. Response 10: See response 2 and response 4.

Investigators Strengths
Investigators have appropriate expertise in necessary disciplines including statistics, questionnaire development, microbiology, and bioinformatics.

Weaknesses
Expertise from an infectious disease clinician, hospital/clinical infection control specialist, and infectious disease epidemiologist are lacking from the team.

Innovation Strengths
Social network analysis, coupled with more traditional microbiologic studies, is a novel combination of approaches to address the project aims. Investigation of ethnicity by type of residential status is innovative in assessing community-level disease patterns.

Weaknesses
Alternative approaches regarding transmission dynamics modeling are not addressed.
Response 12: We are not attempting to sample or model an entire outbreak, but rather gather evidence of transmission between individuals in social groups. We are not entirely sure what is meant by this comment. We eventually hope to create models of transmission dynamics that include both phylogentic and social contact variables, and would assume that the data from this study would be highly valuable for those modeling processes. At that point we will address both systems dynamics and agent based modeling parameters, as well as alternative transmission dynamics modeling, but it seems premature to jump from the hypotheses and mechanisms that will be addressed in this study to describing alternative models of transmission dynamics.

Approach Strengths
The general methods are appropriate to meet the project objectives: to gain further insights into important components of S. aureus transmission: community carriage, pathogen genotypes, and the impact of social interactions.

Weaknesses
Methods to investigate sex as biologic variable are not discussed. Response 13: See response 4 One of the key mediating or moderating variables that we will test is differences in both carriage and in levels of social contact based on sex as a biologic characteristics. In addition to sex as a biologic characteristic, gender as a cultural characteristic will be important to our analysis.
The investigators hypothesize that pathogen populations do not differ among groups seen in clinical settings versus community settings. In order to demonstrate that the populations "do not differ", they will need to specify methods that are appropriate for equivalency analyses, based on a pre-specified margin of equivalence. This type of approach is not evident in the application.

Response 14:
We believe that the reviewer is suggesting the use of a non-inferiority or equivalence approach. To appropriately complete this type of analysis, we would need a much larger sample size to maintain the level of statistical power as determined in the sample size estimations. Rather, this must be done in a phylogenetic context that we discuss in section SA#1-7. In short, we will use Bayesian Tip-Significance testing to determine if there are any associations between phylogenetic clades and independent variables such as ethnic group or clinical vs. community isolates.
Alternative approaches regarding transmission dynamics modeling, and infectious disease epidemiology approaches, are not addressed.
Response 15: This appears to be identical to an earlier statement, please see See response 12 Recruitment of study participants will occur at multiple public spaces and events. Methods to minimize selection bias are not specified. It is unclear if these sampled individuals will be representative of the larger community, for example, elderly frail individuals may be under sampled, and if the associations between exposures and infection status will be unbiased. Although the investigators note sampling by calendar month to avoid selection bias, they do not comment on the potential for selection bias due to the mechanism of contacting individuals (i.e., location of sampling).

Response 16: See response 2
Analysis of historical (hospital-based samples) records raises concerns of selection bias. Adjustment for factors such as age, sex, and co-morbidities are not described.
Response 17: The hospital based samples are an important comparative baseline data set to compare with the population based data. The hospital data provides a clinical comparison within a population health context. The issue of sample bias (or actually sample difference) will be addressed in several ways, including adjustement for such factors of age, sex, and co-morbidities, which will be obtained for all of our samples.
The sample size calculation for Aim 1 does not adjust for age as a confounding factor, which may be an important confounder given the influx of "snow birds". Response 18: Yes, we expect levels of colonization to be greater in persons >50 years of age. For our sample size calculation, we used a carriage rate slightly less than empirical measures of household contacts. We also used a very conservative estimation of transmission involving non-outside contacts (>0%) as work such as what we propose has not been done before. We expect that >30% of our sample population will include household contacts. This leads to an overall transmission rate of >15% which was used for our sample size calculation. Considering "snowbirds" specifically will increase the number of transmission events that we sample and will increase our power. Our use of general transmission rates therefore serves as a more conservative estimation of the number of participants needed.
The use of logistic regression modeling does not account for the clustered observations within common social and family networks. The correlation among these sub-cluster observations is not accounted for in the analysis and is important to avoid biased estimation. Response 19: The use of a mixed effects logistic regression model (adding a random effect to the previously described fixed effects model) will account for the possible correlation of participants within family networks.

Environment Strengths
The basic science and statistical infrastructure and related core facilities are sufficient to support the work of the project.

Weaknesses
The environmental support does not appear to include close interaction with clinical infectious disease expertise, which is important for the proposed sampling, analysis methods, and inference. Response 20: See response 5.