Objective.—Explore the impact of proactive outreach to a health plan population during COVID-19 pandemic in New Mexico.
Background.—By March 2020, the 2019 novel coronavirus (COVID-19) was a global pandemic, circulating in more than 114 countries. As more information about virus transmission, symptoms, and comorbidities were reported over time, recommendations for reducing the spread of the virus within communities was provided by leading health organizations like the Centers for Disease Control and Prevention (CDC).
Methods.—Criteria were developed to identify health plan members most at risk for virus complications. Once members were identified, a health plan representative contacted each member to inquire about member needs, questions, and provide them with resources. Members were then tracked for COVID-19 testing results and vaccination status.
Results.—Overall, 50,000+ members received an outreach call (during 8-month timeframe), and 26,000 calls were tracked for member outcomes. Over 50% of the outreach calls were answered by the health plan member. Of the members who were called, 1186 (4.4%) tested positive for COVID-19. Health plan members that could not be reached represented 55% of the positive cases. A chi-square test of the two populations (reached vs unable to reach) showed a significant difference in COVID-19 positive test results (N = 26,663, X2(1) = 16.33, P<0.01).
Conclusions.—Community outreach was related to lower rates of COVID-19. Community connection is important, especially in tumultuous times, and proactive outreach to the community provides an opportunity for information sharing and community bonding.
The 2019 novel coronavirus (COVID-19), originating in Wuhan, China, at the end of 2019, turned into a global pandemic by March 2020 affecting more than 114 countries. In the United States, a national emergency was declared on March 13, 2020, as the number of positive cases for COVID-19 and associated deaths, rose exponentially.1 Initial symptoms of COVID-19 infection included those seen in mild respiratory infections, but in some cases, the symptoms rapidly progressed to respiratory distress requiring immediate intensive care treatment and mechanical ventilation.2 Older adults have an elevated risk for hospitalization, eight-fold higher risk for people 75-84 years old compared to 18-29-years old, and higher risk of mortality.3,4
As the virus spread, researchers learned more about virus transmission, symptoms, and comorbidities linked to poor clinical outcomes. The Centers for Disease Control and Prevention (CDC) provided recommendations for reducing the spread of the virus within the community,5 and given coronaviruses are not new (they have been around since the 1960s), individuals susceptible to previous coronavirus outbreaks (SARS, MERS) would once again be susceptible to this current outbreak. Using previous research publications as a guide, cohorts based on demographics, health conditions, and physical health were developed for proactive outreach, targeting the most vulnerable first should they contract COVID-19.
Health promotion has been defined as the process of enabling people [and communities] to increase control over, and to improve their health.6 Health promotion programs aim to empower through education, providing tools to modify behaviour, and providing guidance for setting healthy goals and priorities. Health promotion at the individual level can work, but it can be ineffective during an epidemic where the community can play a big role in infection rates and community spread. Previous research has shown that people act on health warnings, if they: 1) believe they are susceptible to infection against which protection is required, 2) perceive the condition as severe, 3) acknowledge the preventive action as effective to reduce the threat, and 4) believe they can perform the preventive action.7 With COVID-19, some communities did not perceive the risk as severe and did not engage in preventive actions, like mask wearing, social distancing, and limiting exposure by avoiding going out in the community. Information sharing is key to communicate risk, but the abundance of information available on COVID-19 from both official and unofficial websites with recommendations, instructions, and statistics, has created confusion within communities.7
To be proactive in messaging and communication and to get ahead of the virus in New Mexico, criteria were developed to identify health plan members most at risk for virus complications. Member demographics and historical claims data were the main data sources, in addition, the Johns Hopkins Adjusted Clinical Group (ACG) system8 was leveraged to flag members for prospective utilization and condition flags. The ACG grouper combines administrative claims data with electronic medical record (EMR) data; the addition of timelier EMR data provided an opportunity for earlier risk detection. Once members were identified, a health plan care coordinator contacted each member to inquire about member needs, answer questions, and provide the member with informational or community resources. Examples of community resources included information for and contact with food banks, medical equipment and/or supplies, and meals (if qualified), among many others. Members were then tracked to see if they later tested positive for COVID-19 and/or if they received an initial dose of the COVID-19 vaccine when they became eligible, and the vaccine became available.
Many COVID-19 surveillance studies exist in the literature, with many looking at the impacts of COVID-19 exposure on physical and mental well-being. This study analyzed COVID-19 impacts upstream, focusing on prevention activities and evaluating: 1) if the prevention activities were effective in preventing COVID-19 infection, and 2) did the health promotion piece of the messaging assist later with COVID-19 vaccination rates.
DESIGN and METHODS
Criteria for the tiered outreach.
Study Location and Design
This study is a retrospective analysis of the COVID-19 outreach effort employed during the height of the COVID-19 pandemic in New Mexico by an integrated health care system. Founded in 1908, Presbyterian Healthcare Services (PHS) has grown to a health care system that consists of 9 hospitals across the state, a multi-specialty medical group with more than 900 providers, and a state-wide health plan.
Health Plan Members and Data Collection
Starting in March 2020 and continuing through October, a health plan population of 599,477 members were segmented into outreach tiers (6 high-risk, distinct groups) based on their demographics, social determinants of health (SDoH), known health conditions, medications, geographic location, and likelihood for hospitalization. Member demographics and historical claims data were the main data sources; in addition, the Johns Hopkins Adjusted Clinical Group (ACG) system (Johns Hopkins Bloomberg School of Public Health, 2019) was leveraged to flag members for prospective utilization and condition flags. Criteria for the tiered outreach was based on these two factors: 1) Centers for Disease Control (CDC) guidance on the most vulnerable populations based on data available at the time, and 2) health plan medical director input based on the New Mexico population and the state's health care infrastructure (access to health care if needed). Specific criteria are outlined in Figure 1.
Based on the member's response to a phone script, the outreach outcome was bucketed into 5 categories.
Based on the member's response to a phone script, the outreach outcome was bucketed into 5 categories.
COVID-19 testing results were collected from near real-time HL7 messages between the health plan and large laboratories within the state, including: TriCore Reference Laboratories, Quest Diagnostics, LabQuest, and laboratories within the integrated health system. For reference, HL7 is a standard for exchanging health information between medical applications and between health care providers.9 COVID-19 vaccination status was compiled from multiple data sources, including: Centers for Medicare & Medicaid Services (CMS) files, the New Mexico State-wide Immunization Information System (NMSIIS), health plan claims, records in the integrated health care delivery system, and pharmacy encounters.
Outreach Scripts
Standardized scripts were provided to each health plan representative for the outreach calls. The script was purposely kept at a high-level, leaving opportunity for the health plan member to express their needs and ask questions. Health plan representatives were provided with guidance should certain topics come up in the conversation, like medications, nursing advice, and specific questions regarding the COVID-19 virus (see Figure 2). A sample script used for the outreach effort:
“This is {Name}, from Presbyterian Health Plan's Care Coordination team. I am reaching out to see if you have any questions or if you need assistance due to the changes that are happening right now. I want to ensure you have everything you need to stay at home and to see if you have any concerns we need to discuss.”
Based on the member's response to the above script, the outreach outcome was bucketed into 5 categories:
(1) member reached – answered questions, (2) member reached – referred for more information, (3) member reached – refused to speak to caller, (4) member reached – no needs identified, and (5) member not reached.
Data Analysis
Descriptive statistics for health plan member characteristics, COVID-19 positive test results, and COVID-19 initial vaccination status are presented using simple means, counts, and proportions (Table 1). Bivariate and multivariable analyses were conducted to examine what factors were associated with: 1) a positive COVID-19 test result, and 2) receiving the initial COVID-19 vaccine dose. Bivariate and multivariable analyses were conducted using a logistic regression model (both dependent variables were binary). A stepwise procedure using the SAS command PROC LOGISTIC was employed to determine which covariates enhanced the model. Evaluated covariates included: member age (years), member sex, member location (central vs non-central), outreach outcome (member reached vs member not reached), engagement with the member's primary care provider (PCP) in the last 12-months, and the outreach tier for the health plan member (see Figure 1 for outreach tier criteria). To test the fit of the two models, a Hosmer-Lemeshow chi-squared value and associated p-value were calculated and evaluated. All analyses were completed using SAS (v. 7.15).
Characteristics of Study Population by Outreach Outcome (N = 26,663 Health Plan Members)
Ethical Considerations
Approval and permission for this study were obtained from the Presbyterian Healthcare Services Human Research Protections and Institutional Review Board (IRB). The IRB determined that this analysis was classified as a quality improvement project and did not involve clinical/human subject research.
findings
Demographically, the overall outreach population on average were 65 years old, female (51%), resided in the central region of the state (68%), and saw a general practitioner in the 12-months prior to outreach (91%). Over 50% of the outreach calls were answered by the health plan member (Table 2). Most of the members who were reached did not have questions about the coronavirus. Only 488 members were referred to further programs for medications, care coordination, or additional information (eg, nurse advice line). Less than 1% of the calls resulted in refusal to speak to the health plan representative. Of the members who were called, 1339 later tested positive for COVID-19 (as of October 11, 2021). Health plan members that could not be reached represented 54.3% of the positive cases.
Outreach Outcomes by Tiered Group, New Mexico
Comparing members by outreach status (reached vs not reached), members reached were more likely to identify as female, live in the central region of NM, average 67 years-of-age, have seen their general practitioner in the last 12 months, less likely to have tested positive for COVID-19, more likely have received their first COVID-19 vaccine, and have had the largest representation in Tier 5 (Table 1). Univariate analyses comparing the demographic variables by outreach status showed that all the variables were significantly different between the two outreach populations at a p-value of 0.05 except for member gender.
COVID-19 Positive Testing
Overall, the percentage of the outreach population, which tested positive for coronavirus, was relatively low (5.0% test positivity in entire population). The outreach subpopulation with the highest positivity rate was members who were referred for more information at 6.6%. Members who did speak with a health plan representative had a positivity rate of 4.5%; a chi-square test (X2) comparing coronavirus positive test results by outreach outcome population (members reached vs members not reached) indicated a significant difference between the two populations (N = 26,663, X2(1) = 13.25, p<0.01).
In unadjusted analyses, factors significantly associated with increased odds of having a COVID-19 positive test result included outreach outcome and outreach tier (Table 3 ). Other factors not included in the adjusted logistic regression model included age, gender, territory, and general practitioner visit. These factors, when included in the adjusted model, caused a poor fit, as determined by the Hosmer-Lemeshow chi-squared value. A stepwise procedure for identifying the most influential factors resulted in the inclusion of outreach outcome and outreach tier (Hosmer-Lemeshow gof X2 = 0.892).
Health Plan Member Characteristics and Odds of Testing Positive for COVID-19 (N = 26,663 Health Plan Members)
COVID-19 Initial Vaccination
Initial vaccination results were high among the entire outreach population (>50%). Certain outreach groups had higher initial-dose vaccination results; 67.8% of the outreach population that was reached with no identified needs received their first dose of the COVID-19 vaccine. The group with the lowest initial-dose vaccination results were the outreach members who refused to speak with caller at 58.9%. A chi-square test (X2) comparing initial-dose vaccination rates by outreach outcome population; members reached vs members not reached, indicated a significant difference in vaccination results between these two groups (N = 26,663, X2(1) = 94.65, p<0.01).
In unadjusted analyses, factors significantly associated with an increased odds of receiving the first dose of the COVID-19 vaccine included outreach outcome and outreach tier (Table 4 ). Other factors not included in the adjusted logistic regression model included age, gender, territory, and general practitioner visit. These factors when included in the adjusted model caused a poor fit as determined by the Hosmer-Lemeshow chi-squared value. A stepwise procedure for identifying the most influential factors resulted in the inclusion of outreach outcome and outreach tier (Hosmer-Lemeshow gof X2 = 0.612).
Health Plan Member Characteristics and Odds of Receiving First COVID-19 Vaccination (N = 26,663 Health Plan Members)
discussion and Implications
By October 12, 2021, New Mexicans testing positive for COVID-19 since the beginning of the pandemic reached 261,370 (12.4% of the state population).10 In comparison, less than 5% of the health plan members who received a call from a health plan representative tested positive for COVID-19 by the same date. The outreach population that answered the call represented less than half of those who tested positive. Vaccination rates in general were high for this health plan member population (>50%) but was even higher among those health plan members who interacted with a health plan representative. According to the Mayo Clinic's Coronavirus vaccine tracker website,11 65.7% of the US population by October 12, 2021, received their first dose of the vaccine. In comparison in this study, 64.7% of the members reached by a health plan representative were vaccinated by this same date. Health plan members who were not reached during the outreach effort were 16% less likely to have received their first COVID-19 vaccination by October 2021 (OR = 0.84; 95% CI: 0.80, 0.89; p<0.05). This large-scale health plan outreach initiative provided an opportunity for health plan members to get answers to coronavirus questions and for the health plan to show its connection with the community.
Previous studies of populations affected by disasters indicated that proactive outreach efforts provide an opportunity to address unmet needs by providing psychosocial support, monitoring for health needs, and providing health care guidance.12,13 Health plan members for this outreach effort had positive experiences with the outreach interaction. Health plan representatives reported back that members were appreciative of the effort and thankful for the provided clarity and guidance that they received regarding their questions. At the time of this outreach effort, COVID-19 provoked fear, and having the ability to talk to a health care professional helped to alleviate some of the fear and anxiety14 that the members were experiencing. With the unknowns and the state shutting down, navigating the health care system became a journey fraught with obstacles.15 The outreach calls provided an opportunity for the members to ask pointed questions about their health care needs and provided guidance on how to navigate the health care system, especially given the many state-mandated procedural changes that occurred due to the pandemic. In addition to providing navigation guidance, for those members who tested positive, other health plan programs were offered to help manage the member's COVID-19 symptoms and address some access issues due to quarantining requirements (eg, 14-day food delivery initiative for members with identified food insecurity).
Studies have shown that targeted outreach to specific populations produce beneficial outcomes.16 The health plan population was very large, over 500,000 members, and outreaching to each member was not feasible. By identifying those members most at risk (due to their demographics, health conditions, health maintenance, and probabilities for utilization derived from previous utilization), the health plan was able to target the available staff resources in a logical method using an analytic approach.
The health implications of COVID-19, both physical and psychosocial, will be studied for many years to come. Some health outcomes are currently being studied; notably the impact of COVID-19 on isolation and suicide rates,17 women's health,18 families,19 and those in nursing homes.20 As more studies are conducted, and the many layers of COVID-19 impact are peeled back, the goal is to learn from this pandemic so that lessons learned can be applied to the next pandemic.
Qualitative, Retrospective Feedback from Health Plan Representatives
In general, health plan representatives felt that they were as prepared as they could be for the outreach effort, given the worry, fright, and unknowns surrounding the coronavirus. Due to the ever-changing health care landscape, health plan representatives stated that even though sometimes they answered a member's question with “I don't know,” the members appreciated the phone call; being considered, heard, and valued. Many of the members lived alone and the opportunity to connect with another person was invaluable and provided an emotional benefit to the member.
One of the prevalent themes from health plan representatives was the wish for more resources specifically around food or groceries. As more resources became available within the community, information was shared between the care teams to then pass on to members during the outreach effort. A second theme identified was through conversations with members receiving community benefits; health plan representatives were able to assure members that there would be no interruptions to receiving these benefits. Health plan representatives were also able to identify members who needed more assistance going forward in the pandemic, noting that future touchpoints were needed to ensure the member was doing well.
Lastly, health plan representatives noted that the outreach effort left them with a positive experience. Yes, there were extra work hours required, but the health plan representatives were able to strengthen their relationships with members and provide value to each member and the community.
Strengths and Limitations
There are several limitations to this study. First, although vaccination status for this analysis was gathered from multiple data sources, if a health plan member was vaccinated outside of the state of New Mexico, then that information was not available for this analysis. Second, for health plan members who did their COVID-19 testing outside of the laboratories that report to the health plan, like at New Mexico Department of Health clinic sites or sites outside of the state, the test results may not have been included in this analysis. Third, factors outside of the control of the health plan may have influenced COVID-19 test results. At different times during calendar year 2020, New Mexico went on state required lock-down; the entire state was asked to stay in their homes unless there was a life-threatening emergency. Social distancing and mask mandates influenced viral spread amongst populations across the state potentially impacting COVID-19 testing prevalence and results.
conclusion
This study has shown that targeted outreach to those most vulnerable during times of health disasters, like a pandemic, can have beneficial results both for those who are outreached as well as to the larger community. The outreach interaction provided an opportunity for health care needs to be met and questions to be answered that ultimately kept health plan members out of hospitals during a tumultuous time period. Secondly, this outreach effort provided an opportunity for health plan representatives to engage with members on a 1:1 basis; not all health plan members were in care coordination program prior to the outreach call.
Thanks to the health plan representatives who tirelessly worked on contacting health plan members; being a friendly voice to so many who were scared and lonely during the first months of the pandemic.