National Health Interview Survey (NHIS)

Database Name: National Health Interview Survey (NHIS)

Description

The NHIS is a government cross-sectional survey of non-institutionalized civilians administered annually and conducted by household interview. The in-person interviews are conducted weekly on an ongoing basis throughout the year by staff at the US Census Bureau.

Data Source

The survey developed out of the National Health Survey Act of 1956, and is intended to provide accurate national estimates of self-reported illness/disease and healthcare utilization. The survey is designed by the National Center for Health Statistics (NCHS) as part of the CDC, and is administered by the US Census Bureau. The NCHS trains approximately 400 interviewers, who conduct the interviews.

Overview of data contents

Broadly, the intention of the survey is to assess the self-reported prevalence and impact of illness and disability in the United States and the need for medical services as a result of the illness or disability. In any given year, one can get an estimate of disease burden for a variety of self-reported health conditions, some particular to otolaryngology (i.e. hearing status, allergy symptoms, dizziness). Since the survey is administered annually and the question set remains relatively constant over time (there are significant revisions to the core questions every 10-15 years), it allows assessing national trends in illness.

Patients ages included

Any adult in the contacted household age 18 and older is eligible. Interviewers also survey children age 17 and younger by proxy (the sample adult in the household answers in their stead).

Practice setting

The information is collected from non-institutionalized civilians, and is conducted by in-person household interview by trained staff of the US Census Bureau.

Date range available

The data is published annually and has been administered since 1957. The survey content does change periodically to reflect new national health focuses. Intermittent (non-annual) specialized supplements are occasionally added to the core questions. For instance, in 2008, a balance/dizziness supplement was added for one cycle. Major revisions to the core questions occurs every 10-15 years. The last major core revision was in 1997.

Relevant Work

Example publications
 

Prevalence and Impact of Bilateral Vestibular Hypofunction: Results From the 2008 US National Health Interview Survey. BK Ward, Y Agrawal, HJ Hoffman, JP Carey, CC Della Santina
JAMA Otolaryngology—Head & Neck Surgery 139 (8), 803-810.

Ward BW, Schiller JS. Prevalence of multiple chronic conditions among US adults: estimates from the National Health Interview Survey, 2010. Prev Chronic Dis. 2013 Apr 25;10:E65. doi: 10.5888/pcd10.120203. PubMed PMID: 23618545;PubMed Central PMCID: PMC3652717.

Access 

Cost estimate(s)

The data is available free of charge, and data from 1997 to present is available for download via the NHIS website.

Contact/website information


http://www.cdc.gov/nchs/

Practicalities 

Ease of use

Each year's data include a downloadable readme file, descriptions of the data, samples of the survey, and sample code for SAS, SPSS and Stata to import the data for analysis. Descriptions for applying sample weights are included as well. You need to be familiar with one of these statistical programs, but the process is quite user friendly.

Data analysis

The data uses stratification, multistage sampling, and a probability cluster sampling technique with oversampling of minorities to improve statistical estimates. You therefore need to apply sample weights to the data to get appropriate national estimates. Weights are provided for each dataset, and instructions on applying these weights to multiple years may require assistance from a statistician. limitations for longitudinal analyses would occur if the same questions you were investigating were not asked over time. Additionally, the same individuals are not surveyed in subsequent years.

Pros

It is a nationally representative sample of community-dwelling US civilians, allowing accurate estimates of self-reported disease/healthcare utilization. Interviews are conducted by personal visit, not telephone. The sampled population is large, with approximately 35,000 households (87,500 individuals) surveyed annually. One could assess temporal trends in disease burden over time. There are data available on children as well, with survey questions answered by proxy. It also oversamples under-represented minorities to improve prevalence estimates.

Cons

The data on health and disease burden are self-reported, and therefore subject to associated bias. They are also anonymous, and the same individuals are not followed sequentially over time.