The Affordable Care Act (ACA) of 2010 requires non-profit hospitals to conduct a Community Health Needs Assessment (CHNA) to identify health issues in its service area and adopt an implementation plan at least every three years. Under Internal Revenue Service rules,1 a charitable hospital may gather input from persons who represent the broad interests of the community it serves through meetings, focus groups, interviews, surveys, and/or written comments. Hospitals are expected to assemble sufficient information on the medically underserved, low-income, or minority populations to identify and prioritize health needs and develop a strategy to improve the health of these groups in their communities. Hospitals are free to decide how to collect data given their financial and technical support resources.
When conducting its CHNA, a hospital may collaborate with other charitable, for profit, or government hospitals and state and local public health departments. In 2012, the health department and three medical centers serving 45,000 residents in rural Polk County, Wisconsin conducted a community health needs assessment. It engaged more than 1,800 county residents through coordinated surveys and community forums.2 Studies have found that when hospitals collaborated with the local health department, their CHNAs were of higher quality, scarce resources could be more effectively and efficiently deployed, and common goals helped achieve positive population health outcomes3,4,5 In addition, it is important to include community members in the CHNA process to identify health needs of groups experiencing health disparities.4,6
In 2013, Wake County, North Carolina, with its mix of large cities (Raleigh, Cary), and rural small towns, employed cluster sampling of census blocks to generate 35 face-to-face interviews in each of the county’s eight health zones, for an anticipated total of 280 interviews. One limitation was that households selected have an unequal probability of selection, and the analysis included a mathematical weight for probability of selection to reduce bias.7 In 2016, the county collected data through focus groups, telephone, and internet-based community and key leader surveys.8
Thumb Rural Health Network
The Thumb Rural Health Network (TRHN) includes eight community hospitals and three county health departments in Michigan’s “Thumb.” The Thumb refers to the three counties that extend up between Lake Huron on the east and Saginaw Bay on the West, approximately 100 miles or 161 kilometers due north of Detroit which form the thumb of the mitten-shaped Lower Peninsula (Figure 1).
Figure 1 Thumb Regional Counties Map
TRHN was formed in 2003 and incorporated as a non-profit in 2007 to address the highest priority health care needs of Michigan’s Sanilac, Huron, and Tuscola Counties through collaborative projects. It focuses on access to health care for underserved populations, improving community health status, educational opportunities, and leadership development. Each hospital pursues its community-based mission through meaningful relationships and collaborations with other community organizations. The estimated population of the three rural counties in July 2016 was Huron 31,481, Sanilac 41,409, and Tuscola 53,338 for a total of 126,228.
In August 2015, the Michigan Center for Rural Health, the Hospital Council of East Central Michigan, and TRHN convened a group of TRHN members and three consultants to identify and use common data collection instruments and methodologies for each hospital’s 2016 CHNA. The group agreed to adapt and administer the University of North Dakota’s Center for Rural Health’s community survey instrument, and focus group and stakeholder interview questions. 9,10 The Center had conducted CHNAs for 21 critical access hospitals in rural communities in North Dakota utilizing a mixed methods design involving surveys, focus groups, and stakeholder interviews.
Each of the eight thumb area hospitals received the results from survey respondents in its service area as defined by zip codes, the focus group held at the hospital, and their county stakeholder interviews. The findings enabled each hospital to identify health needs, establish priorities, and create a plan to address those needs for its service area. Since each Thumb area hospital used the same survey instrument, it was possible to aggregate, analyze, and report the data by county and the tri-county region which generated reports that could facilitate countywide and regional collaboration for improving health care and delivery.
The mixed methods consisted of a community health needs assessment survey, focus group meetings, and stakeholder interviews conducted from January through December 2016. The focus groups and key stakeholder interviews were used to corroborate and enrich the survey findings.
The gold standard for surveys is the random sample in which each individual is chosen entirely by chance and each member of the population has an equal chance of being included in the sample. However, limited budgets preclude most rural hospitals from conducting a random sample survey. When data is difficult to collect, an exploratory survey using mixed methods can identify health issues and improve healthcare delivery.11 The task is to collect and analyze data from respondents that can be statistically inferred to a defined population.
The Thumb Area CHNA survey employed a non-probability purposive and convenience sampling design. In a purposive sample, respondents are recruited based on some characteristic that will be useful for a CHNA, for example, seniors residing in independent or assisted living facilities as well as patients, clients, and employees at health care and social service facilities. Convenience sample respondents can be anyone who has access to the survey such as people in a mall, friends and family of volunteers who distribute the survey, or individuals who access the survey online. If the mixed sampling design results in a reasonable number of low-income, low-education, and senior citizen respondents, this permits an analysis of their health concerns and views on health care services.
The Thumb Area survey asked respondents about their community’s quality of life, the availability and delivery of health services, barriers to care, and concerns about the health of adults, youth, and seniors. Demographic information included gender, educational attainment, income, age, employment status, residential ZIP code, primary health insurance, race/ ethnicity, marital status, and whether currently employed by a hospital, clinic, or health department.
Adjustments were made to align the proportion of low-income, low-education, and senior citizen respondents with county level demographics available from the Census Bureau’s American Community Survey (ACS).12 Each year the ACS contacts more than 3.5 million households across the county, including about 4,800 in Michigan. The use of the hospital surveys, focus groups and stakeholder interviews, and ACS data was exempted by the Michigan State University Institutional Review Board.
Each hospital distributed surveys to business and community venues in their service area. Community organizations that were familiar with the health care system and/or vulnerable populations also received surveys to distribute. Respondents could leave printed surveys at central collection boxes or mail them in business-reply prepaid envelopes to the Institute for Public Policy and Social Research (IPPSR) at Michigan State University. An online version of the survey was administered using SurveyMonkey. The web link was included in press releases and public service announcements on local radio stations. Links were sent to hospitals and service providers who, in turn, forwarded it to their employees and emailed their patients or clients. Paper surveys were entered by IPPSR and then combined with the online surveys. A total of 1,758 usable surveys were received.
Focus Groups and Key Stakeholder Interviews
Each hospital held a focus group comprised of five to 12 community members, for a total of 56 participants. They represented agricultural interests, the faith community, schools, civic organizations, local government, police, health care professionals, and hospital volunteers, staff, and board members. In each county, interviews were conducted with four to five key stakeholders who were familiar with the community and its health care services and needs. The 14 stakeholders were from county departments of health and human services, county mental health agencies, an outpatient mental health clinic, a community collaborative of public and private members working together to coordinate health and human services across systems, an intermediate school district, county government, and an economic development corporation.
Survey data were analyzed using the Statistical Package for the Social Sciences (SPSS) Version 22. This version includes a multiple response analysis which is used when more than one response is allowed for a survey question. For example, one survey question asked “Think about the AVAILABILITY OF HEALTH SERVICES in your community. What are your top three concerns? (Choose up to THREE)” Several of the choices included ability to get appointments, availability of specialists, availability of mental health services, availability of in-home assistance, and availability of immunization services. Multiple response analysis allows the set of responses to this question to be treated as a single variable to generate frequencies and cross-tabulations.
A total of 1,758 useable surveys were received. Respondents were assigned to a hospital when their residential zip code was within its service area. The eight hospitals received a total of 3,270 surveys, with a range of 145 to 844, and a mean of 408.75. The difference of 1,512 between the total surveys used by the hospitals and the total of actual useable surveys received (3,270 - 1,758) reflects the number of respondents who resided in a zip code claimed by more than one hospital.
The total usable responses for each county were Huron 706, Sanilac 845, and Tuscola 364 for a total of 1,915 and a mean of 638.3. The difference of 157 between the total surveys for each county and the total of actual useable surveys (1,915 - 1,758) reflects the number of respondents who resided in a zip code that crossed county lines. If a respondent resided in a zip code that crossed county lines, they were excluded from the county analysis if less than 25 percent of the zip code was in the county of interest, unless they reported using a hospital in that county in the past two years. These criteria avoided double counting and increased the generalizability of the findings on the county and regional levels.
After eliminating missing data, the initial frequencies for gender and educational attainment revealed a strong selection bias. For example, the percent female for the three counties was between 76.6 and 79.2 percent, and the percent of respondents with an associate level college degree or higher was between 60.6 and 63.1 percent. In order to reduce bias, weights were assigned by comparing the frequency distributions of these variables in the survey with their distributions in the American Community Survey (ACS).12 Since they were highly skewed, sex and age were combined into one weight. Unfortunately, between 9.3 and 14.2 percent of the respondents across the three counties did not report their annual household incomes, so income was not weighted. Instead, education was weighted. This meant that the answer to a specific survey question by a respondent in the under-represented group, e.g. males, received a weight greater than 1 and the answer of those in the over-represented group, e.g. females, received a weight less than 1.
The focus groups and stakeholders were asked to respond to a set of 16-19 items. Several items were lists of services provided by the focus group’s hospital or county health department. Participants highlighted the services they did not know about. The lists were collected and the number of participants who did not know about a listed service was recorded and rank ordered. They also received a list of health-related issues that may affect their community: the environment, health services, the physical health, mental health, and substance abuse of adults, of youth and children, and the aging population. They highlighted those they felt were a concern and then starred the five they thought were of the greatest concern. These were then rank ordered by concern.
Responses to open ended questions were recorded and grouped. Some were very easy to group, e.g. Where do people turn for trusted health information? or What particular populations or groups in the area are medically underserved? Others generated a variety of responses, e.g. What are the major challenges facing your community? or, What would help remove barriers that may be affecting the use of local health by the community as a whole? These were recorded, grouped by similar words or phrases, and rank ordered.
As shown in Table 1, the weighted adjustments brought the valid percent females in the Thumb Area survey down from 78.4 to 50.6 percentpwexwnr, just above the ACS estimated range from 49.8 to 50.3 percent. The valid percent males increased from 21.6 to 49.4 percent, just under the ACS range from 49.7 to 50.2 percent. The adjustments increased the percent of respondents with a high school degree or less from 19.0 to 55.4 percent and those with some college but no degree from 18.4 to 23.2 percent. Correspondingly, the percent with an associate degree or higher decreased from 62.6 to 21.4 percent. These are almost identical with the ACS estimates for high school degree or less (55.4%), some college no degree, (22.2%) and associate degree or higher (22.4%).
The percent of respondents with household incomes less than $25,000 rose from 21.0 to 38.5 percent while those with incomes $50,000 or more decreased from 54.0 to 35.9 percent. The adjustments barely changed the percent of households with incomes between $25,000 and $49,999 (25.0% to 25.6%). The ACS estimate for income between $25,000 and $49,999 is higher at 30.7 percent. However, the survey results do not include the 13.5 percent who did not report their household incomes.
Table 1 Unadjusted and Adjusted Regional Control Demographics (N=1,758)
Age was grouped into quartiles. The adjustment lowered the youngest quartile from 38 or younger to 37 or younger. The second quartile increased from 39-51 years of age to 38-53; the third increased by two years from 52-61 to 54-63, and the oldest quartile increased by two years from 62 and older to 64 and older. The ACS placed the median age in the three counties between 43.8 and 48.3 which below the top of the adjusted second quartile at 53.
The adjusted regional demographics are in Table 2. Almost three-fifths (59.1%) of the Thumb Area respondents were either married or remarried. A little less than one-fifth (18.7%) were single, never married. Only three-eights (37.2%) of households had children under 18. A little less than half (46.6%) worked full time, 12.3 percent worked less than full time, 16.8 percent were retired, and 12.1 percent reported being disabled. Those who reported being unemployed and seeking a job accounted for 5.1 percent of the respondents and 1.5 percent were unemployed but not seeking a job.
Insert Table 2 Regional Demographics after Adjustment (N=1,758)
The number one self-reported source for health insurance was an employer or union (44.1%), followed by Medicare for older people (17.1%) and Medicaid for low-income, financially needy people (15.8%). Only 9.0 percent of Thumb Area respondents individually purchased a plan directly from a health insurance company or through the Affordable Care Act website, healthcare.gov. Finally, 1.6 percent reported not having any health insurance.
With respect to vulnerable populations, the adjusted demographics (not shown) revealed that a little more than half (55.4%) had a high school diploma or less, and about three-eighths (38.5%) earned less than $25,000 per year. Approximately one-quarter (23.6%) of Thumb Area respondents were seniors 64 years of age or older. However, almost all (94.4%) of the respondents self-reported that they were White, which accounts for the absence of any analysis of ethnic minorities.
Finally, given the role of the hospitals in distributing the surveys, it is not surprising that approximately one-fifth (19.3%) of the respondents worked for a hospital, clinic, or public health department. This skewed measurement of education and income.
Table 3 summarizes the concerns of approximately half or more of the survey respondents by county as revealed in frequencies and crosstabulations. The top concerns in all three counties were: cost of health insurance, cost of medications for the elderly, not enough specialists, not enough primary care physicians, cancer in adults, and job with livable wages. The top concerns of respondents in two of the three counties were not enough evening or weekend hours, youth drug use and abuse including prescription abuse, and not able to get appointments.
Table 3 Top Concerns of Half or More of Survey Respondents by County
About half or more of respondents in one county were concerned about access to healthy food; assistance for low income families; obesity/overweight in adults; adult drug use and abuse including prescription abuse; don’t know about local services; retaining doctors, nurses, and health professionals; resources to keep the elderly in their homes; attracting/ retaining young families; and the cost of health care services.
Regional Focus Group Findings
The hospital-based focus groups reflected the personal concerns of the individual participants. Their greatest concerns about adult health were drug use and abuse, alcohol use and abuse, and obesity/overweight. They were also concerned about the costs of health insurance and prescription drugs. Much lower on their list was the need for evening and weekend medical appointments and retaining doctors and nurses in the community. The top economic challenges were attracting and retaining young families, and not enough jobs with livable wages.
Regional Stakeholder Interview Findings
The stakeholders, who were from the health, social services, education, and community development sectors, added a broader communitywide perspective. Stakeholders identified alcohol and drug problems, child and domestic abuse, and lack of mental health services as issues to address. They indicated that the top economic challenges facing the region were good jobs and public transportation to access health care. They hoped that hospitals would provide health and wellness programs and collaborate with other providers to cover mental health services. They noted that some people believed that one has to be poor to use public health department clinics and services.
Internal Revenue Service regulations covering Community Health Needs Assessments do not require a rigorous data gathering methodology, thereby allowing each hospital to decide how much to spend to collect data and whether to use the same instruments and methodology as neighboring hospitals.1 While a random sample is necessary to ensure the statistical accuracy of a survey, it is possible to adjust for selection bias and assign respondents who live in a shared service area to an appropriate hospital. The non-probability sampling design resulted in selection bias, and therefore the survey findings were adjusted using the American Community Survey.12 Self-reported information and opinions can be inaccurate as a result of social desirability, recall, and concerns about confidentiality of income and health information.
Shortcomings of focus groups and stakeholder interviews include participant selection, the skills of the facilitator or interviewer, and the consistency of researchers when coding a wide range of topics and opinions. Finally, participants in a needs assessment will vary in their perceptions of what others in the community would say they need and what problems others would say the community should address.
Discussion and Implications
The analysis revealed that respondents in each county identified the same set of top concerns: physician supply, the costs of health insurance and medications, as well as cancer and obesity among adults, and drug use and abuse among youth. The qualitative findings from the focus groups and stakeholder interviewers confirmed these survey findings.
This mixed methods approach provided a set of reasonably accurate, interpretable data enabling each hospital to establish its own priorities and implementation strategies. Knowing their own priorities and resources, hospitals can then participate in health planning and policy decision-making at the county and regional levels along with county agencies, community groups, and regional authorities.
The goal of implementation plans is to improve population health and health care delivery. Highest priorities should be those that the hospitals and health departments can directly address: shortages of health personnel, clinic hours, and staff training in cultural awareness. Second level priority should be those for which hospitals and health departments can join forces with social service, mental health and other health agencies to address substance abuse, mental health services, wellness and rehabilitation services, and resources for the elderly.
Medium priority should be given to working with community and civic organizations to provide assistance for low-income families, and with employers to provide jobs with livable wages and affordable health insurance. Some issues, like transportation to health facilities and services, require county level support and advocacy on the part of non-government health organizations and groups.
We acknowledge and thank Steve Barnett, President/CEO of McKenzie Health System, Sandusky, MI for his thorough review of the original manuscript. We thank all those who responded to the survey, attended focus groups, or participated in the key stakeholder interviews.
1. Internal Revenue Service (IRS). Additional Requirements for Charitable Hospitals; Community Health Needs Assessments for Charitable Hospitals. Department of the Treasury, Internal Revenue Service. 79 Fed. Reg. 78954 (December 31, 2014).
2. Gretchen Sampson, Kim J. Miner Gearin, and Mary Boe, “A Rural Local Health Department–Hospital Collaborative for a Countywide Community Health Assessment,” Journal of Public Health Management and Practice. 21, no.1, (2015): 23–30.
3. Cara L. Pennel, Kenneth R. McLeroy, James N. Burdine, and David Matarrita-Cascante, “Nonprofit Hospitals’ Approach to Community Health Needs Assessment,” American Journal of Public Health, 105, no. 3, (2015): e103-e113.
4. Cara Pennel, Kenneth R. McLeroy, James N. Burdine, David Matarrita-Cascante, and Jia Wang, A Mixed-Methods Approach to Understanding Community Participation in Community Health Needs Assessments.” Journal of Public Health Management and Practice, 23, no. 2, (2017): 112–121.
5. Kate E. Beatty, Kristin D. Wilson, Amanda Ciecior, and Lisa Stringer, Collaborations Among Missouri Nonprofit Hospitals and Local Health Departments: Content Analysis of Community Health Needs Assessments,” American Journal of Public Health, 105, no. S2, (2015): S337-S344.
6. Thomas K. Bias, Christiaan G Abildso, Emily Vasile, and Jessica Coffman, “The Impact of Community Input in Community Health Needs Assessments,” Journal of Public Health Management and Practice, 23, no. S4. (2017): S29-S33.
7. Edie Alfano-Sobsey, Sue Lynn Ledford, Kasey Decosimo, and Jennifer A. Horney, “Community Health Needs Assessment in Wake County, North Carolina: Partnership of Public Health, Hospitals, Academia, and Other Stakeholders,” North Carolina Medical Journal 75 no. 6, (2014):376-383.
8. Wake County, North Carolina. 2016 Wake County Community Health Needs Assessment: Opportunities & Challenges. Executive Summary. Available at www.rexhealth.com/app/files.pdf
9. Karin L. Becker, Emerging Health Trends in North Dakota: Community Health Needs Assessments Aggregate Data Report. Grand Forks, ND: Center for Rural Health, University of North Dakota, School of Medicine & Health Sciences. 2013. Available at: ruralhealth.und.edu/projects/community-health-needs-assessment.pdf
10. Karin L. Becker, “Conducting Community Health Needs Assessments in Rural Communities: Lessons Learned,” Health Promotion Practice, 16, no. 1 (2015):15-19.
11. William L. Miller, Benjamin F. Crabtree, Michael I. Harrison, and Mary L. Fennell, “Integrating Mixed Methods in Health Services and Delivery System Research,” Health Services Research, 48, no. 6 Pt2 (2013): 2125-2133.
12. U.S. Census Bureau. American Community Survey 2010-2014 ACS 5-year estimates.
Harry Perlstadt, Professor Emeritus of Sociology at Michigan State University, has evaluated health programs and initiatives for the Kellogg Foundation, USDHHS Health Resource Services Administration, and World Health Organization. He chaired the Commission on Accreditation of Programs in Applied and Clinical Sociology. He was a Fulbright lecturer at Semmelweis University, Budapest, Hungary. Harry received the 2014 American Sociological Association Distinguished Career Award for the Practice of Sociology.
He has served on the governing council, science board, and joint policy committee of the American Public Health Association and on the National Board and Research/ Scientific Advisory Committee of the American Lung Association.
Harry has a BA in political science and an MPH in health planning and administration from the University of Michigan, and a PhD in sociology from the University of Chicago.
Debra L. Rusz is a Senior Project Manager at the Office for Survey Research (OSR) within the Institute for Public Policy and Social Research at Michigan State University. She Joined OSR in 1996 as a graduate research assistant and was hired as a full-time staff member in 2001. During her research career she has predominately worked on health-related projects, but has worked on additional studies ranging from wildlife management to crime victimization. She is experienced with all aspects of phone, self-administered, and multi-mode surveys.
She holds a B.S. in Wildlife Management and a M.A. in Environmental Sociology with an emphasis in Applied Research from Michigan State University, and has done doctoral coursework in Environmental Sociology.
Travis Fojtasek, PhD, is an independent survey research consultant. He earned his Ph.D. in medical sociology from Michigan State University in 2003. Prior to moving to Michigan, he had a successful 23-year career as a sales representative in Texas for Whirlpool home appliances. His business background brings an added dimension to his social research interest on rural health issues. Travis has worked with the Michigan Center for Rural Health for about 12 years on a number of studies on rural health issues including physician recruitment and retention, the state rural health plan (MiSORHI), physician workforce needs, an assessment of rural health clinics, the student loan repayment program (SLRP) as well as the J-1 Visa waiver program, and the rural EMS system. When not pursuing his interests in rural health issues, Travis is known in his community as an “activist” in politics, the environment, domestic violence, and promoting arts and culture as an economic driver. He also serves on several community boards.
Kay Balcer is a Community Consultant dedicated to supporting nonprofit organizations. In 2000, she began a consulting business with the mission of “Helping Nonprofit Organizations Maximize their Potential.” She has a varied background with a degree in secondary education and more than 20 years working in the health and human services arena. She is currently the owner and operator of Balcer Consulting and Prevention Services and has worked with various nonprofits including critical access hospitals, public health departments, cross sector community coalitions, education entities, early childhood coalitions, emergency medical services, and rural businesses. Her responsibilities include strategic planning, coaching and technical assistance, needs assessments, community organizing, program development, system change efforts, local level evaluation, program development, developing and delivering training programs, grant writing, and project implementation.
She is an alum of Michigan State University and has received training in various best practices such Mobilizing Action Through Community Partnerships, Collective Impact, ABLe Change, and Logic Model Evaluation Strategies. She is a certified Bridges out of Poverty instructor and strives to move communities to adopt a system-based approach to addressing poverty and other complex health and social issues.
Darcy A. Czarnik Laurin is Executive Director of the Thumb Rural Health Network (TRHN) and has filled that role since 2011. TRHN works to improve comprehensive health services in Michigan’s Thumb region by collaborating to improve health and promote wellness of the population in Huron, Sanilac, and Tuscola Counties through access, public awareness, education, advocacy, and leadership. By setting aside historical competitive issues and focusing on emerging collaborative opportunities, network members are better able to address the complex health issues of the Thumb’s rural communities. Most recently, Darcy added the role of Project Director for TRHN’s Thumb Opioid Response Consortium project to her job duties.
A native of rural northern Michigan, Darcy has more than 15 years of non-profit experience working with both federal- and state-level grant funded organizations. She began her non-profit career working as a watershed coordinator designing and implementing watershed and wetland restoration initiatives. Soon after, Darcy’s focus switched to rural health, and she has since dedicated her work to helping alleviate the health challenges and disparities found in rural communities.
Darcy holds a bachelor’s degree in health sciences from Central Michigan University. She also completed the first ever NCHN and Saybrook University partnership Change Our World: Leadership and Transformation Engagement (LOTE) program.