Health agencies within the United States (U.S.) federal government have a strong history of gathering and organizing a wealth of data on a variety of health indicators. The data, collected mainly through survey methods and medical-record reviews, are complied into various datasets and are usually available free of charge for analysis. These datasets, commonly referred to as ‘population-based health databases,’ have been used to conduct original research studies within various fields, such as epidemiology, health services research, nursing, sociology, and demography. Research studies conducted with these databases have documented health disparities (Braveman, Cubbin, Egerter, Williams, & Pamuk, 2010; Centers for Disease Control and Prevention, 2005; Powe, Tarver-Carr, Eberhardt, & Brancati, 2003; Scharoun-Lee, Adair, Kaufman, & Gordon-Larsen, 2009); monitored the progress of various health indicators (Bernert et al., 2010; Dietz, Callaghan, Morrow, & Cogswell, 2006; Jones et al., 2009); and provided recommendations for health policy (Honberg, McPherson, Strickland, Gage, & Newacheck, 2005; Kogan et al., 2010; McPherson et al., 2004; Ormand, Spillman, Waidmann, Caswell, & Tereschenko, 2011; Parish, Shattuck, & Rose, 2009).

Population-based health databases are beneficial not only to nurse researchers, but also to nurses within practice and educational settings who can use health indicators for developing evidence-based interventions and health policy. The purpose of this Legislative Column (Part 1) is to increase understanding of the benefits and limitations of population-based health databases as a possible reference source for developing interventions and health policy. The next Legislative Column. Part 2 will provide an example of how one database, the National Survey of Children’s Health, was used to determine prevalence rates of children with a medical home and explore associations between race/ethnicity and medical homes. In this present column I will discuss the benefits and limitations of population-based health databases, describe available resources and considerations in using these resources to answer health-related questions, and present an original research study using the National Survey of Children’s Health Database.

Benefits and Limitations

Two major benefits of using population-based health databases include the ability to generalize findings and the cost effectiveness of using these databases. The term ‘population-based’ usually implies the data were collected through rigorous, probability-sampling methods. Results from probability sampling can be generalized to an entire population. Samples in some databases, such as the National Survey for Children’s Health, can be generalized to the population of an individual state. However, the majority of population-based health databases represent a national sample and cannot be generalized to people residing in a state or a smaller area. The major advantage of probability sampling is that this method surpasses a convenience sample or a sample limited to a distinct subset of participants because one can generalize the findings of studies using probability sampling.

The second benefit to using databases is related to cost-effectiveness. It would be quite expensive and somewhat prohibitive for a researcher, practitioner, or policy analyst to collect all the data compiled within a health database (Moriarty et al., 1999). The collection and organization of data through stringent data collection protocols have already been done by federal government researchers, or their contractors (McArt & McDougal, 2005; Moldanado, 1991). A majority of these released databases are available for anyone to download from government websites and are either free or available for a small fee. Many databases managed by agencies within the federal government also offer free technical advice and consultation to individuals regarding analyses. For example, the Centers for Disease Control and Prevention (CDC) regularly provides in-person and web-based training sessions. It is advisable to check these government websites on a consistent basis and join relevant listservs to receive announcements of upcoming events, such as training sessions.

While there are noted benefits to population-based health databases, there are also limitations. One limitation is that a database may not contain the variables needed to answer the research, clinical, or policy question you are asking. A match is needed between the research question(s), proposed study variables, and the actual variables in a selected database. A second limitation involves the sampling method. A researcher may want state or local-level information, but only national data are available since probability sampling methods often are not based on smaller subsets, such as individual states. It is usually too costly to conduct probability sampling on a smaller subset. A third limitation involves the accuracy of the data. Databases may reflect self-reported data; the health information provided by participants may not have been validated through a review of medical records or physical examination.

One database, however, the National Health and Nutrition Examination Survey (NHANES), does collect health data from participants through physical examinations, interviews, and an assortment of laboratory tests (Centers for Disease Control and Prevention, 2009). The NHANES is considered an important resource for documenting and tracking the health status of U.S. adults and children.

Despite these limitations, an analysis done through the use of population-based health databases can provide insight and perhaps lead to further research at a local level. For example, if an analysis of national and state data found that certain groups in our society have limited access to healthcare (perhaps associated with their discontinuous health insurance coverage), then one might ask whether this finding would be any different at a local level if major components of the local healthcare system were similar.

Resources and Considerations

A previously published article by Zeni and Kogan (2007) presented an overview of selected, population-based resources from the U.S. Department of Health and Human Services. Since it is not the intent of this column to outline and explain various databases, the reader is referred to Table 1 in this 2007 article for information on the names of databases, descriptions of data files, years available, expected frequency of data collection, and general sample descriptions. The article also addressed an important consideration regarding the need to analyze the databases with appropriate statistical software. Since many of the databases utilize a weighted sampling technique, the statistical software selected for an analysis needs to be capable of handling weighted samples. The researcher may need the expertise of a programmer to address these important analyses issues.

However, not all databases require a programmer for access and analysis. Two population-based health databases, the National Survey of Children’s Health (NSCH) and the National Survey of Children with Special Health Care Needs (CSHCN), are available on a website that includes an interactive search feature. The website, operated by The Data Resource Center for Child and Adolescent Health (DRC) of the Child and Adolescent Health Measurement Initiative (CAHMI) at the Oregon Health Sciences University is available at the website <www.childhealthdata.org>.

The interactive feature of the DRC website allows users to select, view, compare, and download national survey data from two NSCH and two CSCHN databases. This data can be analyzed at individual state and regional levels due to the sampling methods used in these surveys (Data Resource Center for Child and Adolescent Health, n.d.). Previous publications have described details of the surveys, including survey methodologies (Blumberg et al., 2003; Blumberg et al., 2005; Blumberg et al., 2008; Blumberg et al., 2009; Kogan & Newacheck, 2007).

The search feature eliminates the need for a programmer because the user can conduct selected descriptive analyses to investigate over 100 health indicators pertaining to children, their families, communities, access to healthcare, and other issues. However, a programmer will likely be needed for advanced analysis and to access the complete databases which are available on both the DRC and CDC websites.

Original Research Using National Survey of Children’s Health

A study conducted by the author in 2010 used the DRC interactive search feature to compare health indicators of Ohio children with asthma to U.S. children with asthma across the nation. Asthma is one of the leading pediatric chronic conditions. The Centers for Disease Control and Prevention, based on 2007 prevalence rates from the National Health Interview Survey, reported 9.1% of U.S. children less than 18 years old have asthma; health disparities were noted among Blacks, select Hispanic groups, and lower socio-economic groups (Centers for Disease Control and Prevention, 2008).

This study was based on the Aday Framework for Studying Vulnerable Populations (2001). Aday conceptualized certain community and individual characteristics as risk factors associated with poor physical, psychological, or social health. Risk factors are attributes or exposures associated with, or leading to increased probability of selected health outcomes (Aday, 2001). Aday proposed all people are at potential risk of poor health, but the risk of harm is greater for those in poor health with few material and nonmaterial resources for assistance. Aday’s theoretical framework guided the identification of study variables for the study related to a vulnerable group, i.e., children living with asthma (described below).

Four research questions were addressed using data from the 2007 NSCH database and retrieved with the interactive search features from the DRC website:

1. What is the general prevalence of parent-reported asthma among Ohio Children compared to U.S. children in general?
2. Is there an association between family income level and parent-reported asthma among Ohio children?
3. Is there an association between race/ethnicity and parent-reported asthma among Ohio children?
4. Is there a higher percentage of children who have asthma and who are without a medical home in Ohio than in the US nationally?

General findings for each research question were:

1. The prevalence estimate for Ohio children with asthma was 16.1% compared to 13.5% of U.S. children. The higher rate among Ohio children was significant based on a single-sample Z-test with significance set at 0.05 or below.
2. The association between family income level and parent-reported asthma was noted among both U.S. and Ohio children with a higher percent of children living with asthma coming from lower economic groups based on federal poverty levels. A lambda test determined a weak (or low) association between income level and Ohio children living with asthma.
3. An association between the child’s race/ethnicity and asthma was noted for both U.S. and Ohio children. The results of a lambda test suggested a strong association between racial/ethnic minorities and Ohio children living with asthma.
4. A higher percent of Ohio children, 18.2%, did not have a medical home compared to 15.3% of U.S. children. Based on the results of a single-sample Z-test, a significantly higher percent of Ohio children with asthma did not have a medical home compared to U.S. children.

Overall, the findings noted a weak association between income level and Ohio children with asthma and a strong association between racial/ethnic minority groups and Ohio children with asthma. The findings suggested that a significantly higher percent of Ohio children with asthma did not have a medical home compared to U.S. children with asthma. The establishment of a medical home is important in providing consistent evaluation and treatment of chronic conditions, such as pediatric asthma. Practitioners could use the findings to assess if children with asthma, especially children from racial/ethnic minority groups, lacked a medical home. Practitioners could then assist these families in establishing a medical home for their children. Findings support the advocacy efforts of policy analysts working to assure that Ohio children with a common chronic condition, such as asthma, have access to a medical home. Additional research could be conducted to identify Ohio children with asthma who may be at increased risk by not having a medical home, and by being part of a minority group. For example, a researcher could develop a regression model to determine and quantify odds ratios associated with different risk factors, such as lack of a medical home and minority status.

While this analysis went beyond what was available using the interactive search feature, and included the use of statistics to measure association and test for significance, a nurse can use the NSCH data in the DRC website to examine health indicators to advocate for evidence-based health policy. The next Legislative Column (Part 2) will provide a step-by-step description showing how to access the interactive web site and conduct an analysis related to a child health issue.

Summary

Analyzing population-based health databases can provide nurses with the data needed to advocate for evidence-based health policy. Probability sampling methods allow results from an analysis to be generalized to the entire population and further lend credence to the health needs and continued disparities among vulnerable groups in our society. Results of these analyses can provide a framework for policy development and support data-driven approaches rather than dogma-driven initiatives.

Author

Mary Beth Zeni, ScD, MSN, RN
Email: zenim@ccf.org

Dr. Zeni is a senior nurse researcher at Cleveland Clinic, Cleveland, Ohio. Dr. Zeni completed a doctorate at the University of Pittsburgh Graduate School Of Public Health in 1993. She received her MSN in parent-child nursing from Marquette University (1982) and has served as a clinical nurse specialist at various regional perinatal centers. Dr. Zeni has been a researcher in the public and private sectors since 1992. Her research has focused on the evaluation of HIV prevention and maternal-child health programs. Previous publications, research, and presentations have included original analyses of the National Survey of Children’s Health with a focus on children’s access to medical care related to health insurance coverage. She was a graduate faculty member at University of Pittsburgh School of Nursing and at Florida State University College of Nursing (Tallahassee, FL); she currently has faculty appointments at Ursuline College and Case Western Reserve University (Cleveland, Ohio) and the University of Akron (Ohio).