CORONAVIRUS (COVID-19) RESOURCE CENTER Read More

Survey of the Health of All the Population and Environment (SHAPE)

State: MN Type: Promising Practice Year: 2019

Hennepin County Public Health is located in the most populous county in Minnesota, with 1.2 million persons according the most recent 2012-2016 American Community Survey estimates.  Hennepin is a diverse county with a population that is 12.3% black/African American, 6.8% Asian, 0.6% American Indian/Alaskan Native, 3.2% two or more races, and 6.8% Hispanic/Latino. Nearly 12% of the population lived below the poverty level in the past 12 months. Since 1998, every four years, Hennepin County has conducted a population health survey. The Survey of the Health of All the Population and the Environment (SHAPE) provides data on health status and the factors that affect health in Hennepin County. A broad range of organizations use results to support health equity data analyses, surveillance of health disparities and community health assessments.

A major issue for SHAPE and similar surveys are decreasing response rates and concerns about non-response bias. Residents of lower educational attainment, low income, racial and ethnicity minorities, and young adults are often under-represented in survey research. The public health issue addressed was reducing non-response bias to ensure health is measured for all populations to accurately identify health disparities.  Objectives included 1) identify new ways to collaborate with other county agencies that service representing many hard-to-reach populations; 2) exploration of new ways to collect data such as in-person; and 3) employ targeted marketing strategies towards under-represented populations.

To achieve this goal, SHAPE 2018 had to expand beyond the traditional approach of a mail-only survey.  SHAPE 2018 had two components: 1) an address-based random sample mail survey, which over-sampled census-tracts with higher concentrations of hard-to-reach populations, and 2) a non-random, convenience sample, which surveyed county clients in-person at various locations. In-person data collection sites included six county Human Service Centers, a County-partner Federally Qualified Health Center clinic, and the Office of Multicultural Services, an office connecting immigrants and refugees to county and community resources.  In-person data collection provided data for the county client's population, but it also provided an opportunity to augment the SHAPE 2018 mail sample. Respondents were asked for their home address. Demographically, many of those populations under-represented in 2014 completed in-person surveys. Those addresses were cleaned, geocoded, and a those that matched a nonresponding household address from the mail survey were ‘replaced' with an in-person survey at the same address. Final survey results are not yet available, but to date over 3,300 in-person surveys were collected. Preliminary results show 171 surveys were exact matches to addresses in the mail sample for which there was no mailed survey response. Not only did this effort reach populations that may not have otherwise responded to the mail survey, it also led to invaluable collaboration opportunities.

A critical aspect to the success of in-person data collection was identifying who would approach clients to complete the survey. The SHAPE project team collaborated with the Volunteer Communities Partnership Program (VCPP) coordinators early in planning, which strengthened the project in critical ways.  Coordinators recruited volunteers to conduct a pilot test of in-person data collection and worked closely with the SHAPE team to analyze results to inform full-scale implementation. A key observation, critical to clients completing the survey, was the importance of having fully engaged volunteers who represented the communities where sites were located. Coordinators led volunteer recruitment and onboarding of staff for data collection. More than 20 volunteers were part of this project representing more than 70 hours of direct client time, plus additional hours of training.  VCPP coordinators led training of volunteers, using detailed protocols developed in consultation with SHAPE team members.

Targeted marketing of the mail survey was accomplished by employing trusted voices in diverse communities to endorse the survey via mailed postcards.  A similar strategy was undertaken in 2014 and showed an increase in response in those areas following the targeted postcard mailing, so the strategy was expanded in 2018.

              While final survey results are pending, all participation/response objectives for SHAPE 2018 were achieved. The impact of this practice has practical implications for local health departments providing a new model for collaborating with partners to reach under-represented populations in surveys. This project provides a model for sample augmentation and replacement strategies into a random household survey that local public health departments could adopt in their own jurisdictions.  More information on SHAPE 2018 can be found at https://www.hennepin.us/your-government/research-data/shape-surveys.

The public health issue addressed with this practice is the need for reliable local surveillance data to inform public health practice. A primary tool used for this purpose is the population-based survey; however, survey participation has been declining over time, which threatens the validity of data obtained from this method. Of particular concern are low response rates from populations at-risk for experiencing disparities, including low-income individuals and people of color. This must be addressed to ensure health is accurately measured for all populations, in order to be able to measure and identify health disparities to make informed programming and policy decisions that can lead to health equity.  The target populations for this effort are the traditionally under-represented groups in survey research, such as persons who are non-white, Hispanic or Latino, low-income, low-education and young adults 18-24 years old.

The most recent American Community Survey 5-year estimates from 2012- 2016 show the following demographics for these hard to reach groups in Hennepin County:  29.9% of the population reported their race as non-white, 6.8% reported their ethnicity as Hispanic or Latino, 11.9% live below poverty, 24.6% of persons 25 years and older have only a high school education or lower, 9.0% of the population are young adults age 18-24 years.  Overall, these percentages are for Hennepin County, which has a total population of 1.2 million according to the latest ACS 5-year estimates.

Our preliminary analysis shows that demographically, in-person client surveys were completed by many of the hard-to-reach populations that were under-represented in earlier iterations of the survey.  Early analysis of the first 2,490 in-person client surveys show 29.8% reported their race as White compared to 82.5% from the 9010 completed mail surveys, 57.6% reported their race as Black-African American compared to 6% from the mail survey. Low-income persons accounted for nearly half of in-person survey respondents (46.6%) compared to 4.4% on the mail survey and 12% of the in-person surveys were from adults 18-24 compared to 2.4% of the mail survey respondents.


One of the new features of SHAPE 2018 will be the ability to produce reports about the health status of clients for each of the Human Service Center   locations. Various approaches have been taken in the past to reach and report on hard-to-reach populations. In 2002, the survey was conducted via the phone and in four different languages. This came at a large financial cost and was possible through a partnership with the University of Minnesota's Heath Services Research Center to manage and complete the interviews, a center that no longer exists. Survey outreach via targeted marketing materials such as postcards in Spanish and posters in low-income multi-unit buildings, oversampling census tracts that had higher proportions of person in poverty or non-white populations were a few strategies employed in the past.  In 2014, SHAPE produced a disparities report at the county level to highlight disparities of various populations such as persons various racial and ethnic minority groups, low-income and low education, persons with disabilities, the LGBT population, and persons with frequent mental distress.

Hennepin County Public Health is excited about the new approach SHAPE 2018 has taken to reach hard-to reach populations. In particular, the opportunity to collaborate with so many county partners has created shared buy-in for the data for the populations they work with. Additionally, partners such as health systems in the county that utilize SHAPE data for Community Health Needs Assessments, are looking forward to using data they know was gathered from those hard-to reach populations that we all collectively serve, but are often under-represented in assessments. The approach taken to collect this data has practical implications for local public health departments by providing a new model for collaborating with internal and external partners to conduct surveys that reach traditionally hard to reach populations and incorporating their data into a traditional random sample household mail survey.


This practice used tools from Healthy People 2020 topics and objectives were reviewed and incorporated when possible during the survey question development stage in order to have survey results to compare to national objectives.  Wherever possible, question design and formats were used from state and national sources such as the Minnesota Student Survey, Behavioral Risk Factor Surveillance System and National Health Interview Survey. In addition, survey methodologies from these sources were consulted as well. Survey experts and publications were consulted regarding sample replacement strategies of the in-person surveys into the household random sample survey. The survey team maintains active membership in the American Association for Public Opinion Research (AAPOR), and attends annual conference of that group and participates in relevant webinars featuring relevant content.

This practice used evidence-based survey practices of convenience and random probability sampling, both well-established survey practices (e.g. Dillman, Don. Internet, Phone, Mail and Mixed-Mode Surveys, 2014; Office of Management and Budget, Standards and Guidelines for Statistical Surveys, 2006).  However, complimenting the SHAPE 2018 mail survey with data collected via the convenience sample by pulling in matched addresses in order to increase the proportion of traditionally under-represented populations is an innovative practice for a large household survey surveillance project.

The major goal for SHAPE 2018 was to reduce non-response bias by strategizing ways to reach traditionally under-represented populations. The target population for this effort are groups such as persons who are non-white, Hispanic or Latino, low-income, low-education and young adults 18-24 years old.Objectives related to this goal included 1) identify new ways to collaborate with other county agencies that service representing many hard-to-reach populations; 2) exploration of new ways to collect data such as in-person; and 3) employ targeted marketing strategies towards under-represented populations. To achieve this goal and objectives, SHAPE 2018 had to expand beyond the traditional approach of a mail-only survey.SHAPE 2018 used an address-based random sample mail survey as in previous iterations of the survey, but in 2018 a non-random, convenience sample was added, which would survey 3200 or more county clients in-person at eight locations.

To complete this large-scale in-person data collection, collaboration and partnership with a variety of county agencies was key to success. In-person locations included six county Human Service Centers, a County-partner FQHC clinic, and the Office of Multicultural Services. Conversations were held with the managers of each of the in-person data collection sites to determine what our shared interest were and how best to meet those together. The team also attended several meetings of the department Model Integrity Team, made up of managers from service areas across the county, who provided input on survey content and data collection strategies. The team also developed a partnership with the Office of Multicultural Services, creating a temporary assignment of a staff person from that area to act as a liaison to develop strategies to reach non-English speaking populations, improving the language and cultural responsiveness of the survey. The SHAPE team also met with managers from the Public Health Department, the Community Health Improvement Partnership (CHIP), and Mental Health Local Advisory Committee. Topic-specific advice was sought from staff working on issues such as homelessness and sexual violence.

In-person data collection sites were chosen based on the clients they serve, which are overwhelmingly comprised of the traditionally under-represented groups in previous SHAPE surveys. Client demographic data was analyzed to inform the SHAPE team about the populations utilizing county services including volume of clients at each location in order to assist with logistical planning. The Assessment Team worked with data analysts in Human Services to obtain client count data and important demographic information such as race/ethnicity, language spoken, and other factors to assist in planning for in-person data collection. This partnership also resulted in establishing methods for reporting more detailed client counts during data collection to help develop survey participation rates and determine how representative the survey respondents were relative to client population.

As the SHAPE survey team is composed of only five staff, one intern, a manager and supervisor, volunteers were key to the success of the large-scale in-person data collection. The SHAPE project team collaborated with the county's Volunteer Communities Partnership Program (VCPP) coordinators early in planning, which was vital to the project in a number of critical ways.VCPP Coordinators recruited volunteers to assist with a pilot test of in-person data collection and provided valuable insight to inform full-scale implementation. A key observation, critical to clients completing the survey, was the importance of having fully engaged volunteers who represented the communities where sites were located. Coordinators led volunteer recruitment and onboarding of staff for data collection. More than 20 volunteers were part of this project representing more than 70 hours of direct client time, and additional hours of training.VCPP coordinators led training of volunteers, using detailed protocols developed in consultation with SHAPE team members.

Planning for this effort started in 2016 and included numerous meetings with managers and staff at a variety of county agencies. A pilot for data collection was completed in the summer of 2017 with revisions to protocols occurring shortly after. The SHAPE mail survey was first mailed in May of 2018 and lasted through the end of October 2018.In person data collection occurred at two sites simultaneously during weeks in June, July and August.Data collection at the Office of Multi-cultural services was over a period of six weeks as data collection was dependent upon their staff to provide language support on the survey, and as a result lasted throughout the month of August and early September.

All objectives for SHAPE 2018 were achieved with final survey results pending. The impact of this practice has practical implications for local health departments providing a new model for collaborating with partners to reach under-represented populations in surveys. Combined with providing a model for sample augmentation and replacement strategies into a random household survey that local public health departments could adopt in their own jurisdictions. 


One of the main goals of SHAPE 2018 was to reduce non-response bias by reaching a representative cross-section of residents to gather generalizable data for health status and other measures at a county and sub-county level. The SHAPE survey is a primary data source for Hennepin County.  The Hennepin County Public Health Assessment Team was responsible for survey development and planning.  Survey questions were developed using a variety of sources including previous versions of SHAPE to continue valuable trend data, various program staff and managers as well as internal and external stakeholders were consulted, Healthy People 2020 goals and objectives and national and state surveys were reviewed. Two methods were used to collect data targeting hard-to-reach populations 1) an address-based random sample household mail survey, which over-sampled census-tracts that have higher concentrations of hard-to-reach populations, and 2) a non-random, convenience sample utilizing in-person data collection from Hennepin County clients.  The new approach of in-person data collection from Hennepin County clients was one of the primary methods to accomplish the main goal of reaching a representative sample of residents and reduce non-response bias. Objectives related to the overall goal included 1) identify new ways to collaborate with other county agencies that serve hard-to-reach populations; 2) exploration of new ways to collect data such as in-person; and 3) employ targeted marketing strategies towards under-represented populations.

The ultimate performance measures for the project include the reliability of data collected, reduction in non-response bias, and, most importantly, the community buy-in to the applicability of the results. Supporting measures include the overall response rate, response rate from low-income communities, cooperation rate of those approached with in-person data collection, and the number of completed surveys that come from populations that are traditionally survey non-responders respondents (e.g. low income, persons of color, language). Process measures of success include data-collection sites successfully identified and implemented, and data collection staff, volunteers, and interpreters successfully identified, trained, and utilized.

Matching addresses gathered during in-person data collection back to the random sample is one practice that we will evaluate for the impact on non-response bias.  In addition, using sample replacement strategies such as nearest neighbor replacement utilizing GIS to measure distances of a Hennepin County client from an address from the 30,000 households that were sent a survey but did not respond is currently being conducted and once complete will be evaluated for the impact on the demographics of our survey respondents and non-response bias. Final survey results are not yet available as analysis is in process, but to date over 3,300 in-person county client surveys were collected. Utilizing GIS, we have found that more than 170 of the in-person client surveys matched to the outstanding mail household survey sample, and over 800 of the 3300 in-person surveys were completed by a person living in the same building as a unit that was sent a mail survey but failed to return it. Further work using GIS and survey demographics will be done to determine how many of those 800 surveys, as well as other close proximity addresses should be added to the household survey results.

Survey results are being analyzed utilizing STATA software and point estimates with confidence intervals will be produced for each survey question by various demographics and geographies in Hennepin County.  Preliminary analysis shows that demographically, many of the hard-to-reach populations completed in-person surveys that were under-represented in SHAPE 2014.  Early analysis of the first 2490 in-person client surveys show 30% reported their race as White compared to 83% from the 9010 completed mail surveys, 58% reported their race as black/African American compared to 6% from the mail survey. Persons with very low income (<$10,000 annually) accounted for nearly half of in-person survey respondents (47%) compared to four percent on the mail survey and 12% of the in-person surveys were from adults 18-24 compared to two percent of the mail survey respondents. Another new effort to reach under-represented populations was the partnership with the Office of Multicultural Services (OMS) to assist persons unable to complete the survey in English with Language support and translation services.  Staff and volunteers primarily at OMS, but also at the other client survey locations provided English translation support to 223 clients in 11 different languages.  Additionally, the survey was translated into Spanish and Somali, 282 person completed the Spanish version of the survey, and 93 completed the Somali version.  

While analysis of the final data sets is ongoing, the in-person data collection at county sites already is proving to have been an invaluable addition to SHAPE 2018. Based on preliminary data, not only did this effort accomplish the goal of reaching many of under-represented populations, this approach had the result of extraordinary collaboration opportunities between Public Health, Human Services, Office of Multicultural Services and North Point Health and Wellness Clinic, a county FQHC.

The steps taken to complete in-person data collection from Hennepin County clients at Human Service Centers will provide a valuable model for public health departments that could replicate such data collection efforts. Combined with a model for sample augmentation and replacement strategies into the random household mail sample that local public health departments could adopt in their own jurisdictions. Once all data analysis has been completed, a debrief and review of what worked and what did not, will be conducted to determine what changes to make for SHAPE 2022. A similar debrief was conducted leading up to SHAPE 2018, which is where the need to reduce non-bias was prioritized as goal for SHAPE 2018 was first identified and strategies to do so, such as in-person data collection, were first considered. 

The sustainability of this practice is evaluated by the degree to which resources were successfully identified, coordinated, and deployed to meet the project goals. Project results demonstrate that this practice is sustainable through augmenting a project team with ongoing responsibility to local public health surveillance through developing strong partnerships to make in-person data collection feasible. More than 20 volunteers were successfully recruited, trained, and deployed, and data collection targets were met at each site. Time to meeting data collection targets was faster than anticipated; seven of eight sites achieved 400 surveys in under 2 weeks, and many sites met goals in just one week of data collection. Most clients approached agreed to take the survey, and the majority were able to make time to complete the survey in a way that was not disruptive to their schedule.

There were several lessons learned that assisted in making the practice sustainable. First, a pilot test of the survey instrument and methodology at two sites was instrumental in planning for appropriate staffing and estimating field-time for data collection. External data collection sites were of critical importance; therefore, meeting with partners early was necessary to develop project buy-in. Project partners were encouraged to review the survey instrument and suggest additional question that would be of benefit to their work area. Site visits were extremely helpful in planning for space needs related to data collection. At each data collection site, the logistics of data collection such as space and workflow were successfully negotiated with site staff in a way that was minimally disruptive. Sufficient numbers of volunteers were identified, many of whom were recruited through partnerships with local colleges and universities, giving students a tangible hands-on experience in public health data collection. Non-governmental agencies operating in co-located sites also provided support through offering data collection volunteers and modifying client workflows to make space for data collection. During data collection, there were few barriers during client encounters, and no significant disruptions to site workflows were found.

Costs of the project included one on-site data coordinator at each site during survey collection. Each site also had one to two additional volunteers or interpreters to assist in data collection. Other costs included printing of surveys, marketing banners, and basic office supplies. The result was also a 33% increase in the volume of surveys needing to be processed and scanned, which was taken on by the project team.

The final sustainability of this practice will be determined through community engagement to support the dissemination of the data findings with the survey partners and the communities they serve. Through this experience, we hope to learn about the additional value that the in-person strategy brought to the project. The specific plans for sustainability will be to develop a revised project methodology based on these results and through continued review of best-practices in local health surveillance.

I am a previous Model Practices applicant