About This Early Release
Early Release of Selected Estimates Based on Data From the 2009 National Health Interview Survey
In this release, the Centers for Disease Control and Prevention’s (CDC) National Center for Health Statistics (NCHS) updates estimates for 15 selected health measures based on data from the 2009 National Health Interview Survey (NHIS) and presents estimates from 1997 through 2008 for comparison. The 15 Early Release measures are being published prior to final data editing and final weighting, to provide access to the most recent information from NHIS. The estimates will be updated as each new quarter of NHIS data becomes available.
Two additional reports are published through the Early Release Program. Health Insurance Coverage: Early Release of Estimates From the National Health Interview Survey (1) is published quarterly and provides additional estimates of health insurance coverage. Wireless Substitution: Early Release of Estimates From the National Health Interview Survey (2) is published in May and December and provides selected estimates of telephone coverage.
The 15 measures included in the present report are lack of health insurance coverage and type of coverage, usual place to go for medical care, obtaining needed medical care, receipt of influenza vaccination, receipt of pneumococcal vaccination, obesity, leisure-time physical activity, current smoking, alcohol consumption, human immunodeficiency virus (HIV) testing, general health status, personal care needs, serious psychological distress, diagnosed diabetes, and asthma episodes and current asthma.
For each selected health measure, a figure is presented showing the trend over time from 1997 through 2009 for the total population, followed by figures and tables showing estimates by age group and sex, based on data from the 2009 NHIS. Also, estimates (adjusted by age, sex, or both, where appropriate) are provided for three race/ethnicity groups (Hispanic; non-Hispanic white, single race; and non-Hispanic black, single race) using data from the 2009 NHIS. Key findings are highlighted by bullets, and data tables containing the values displayed in the figures are included at the end of each section. In addition to providing age-adjusted estimates for Early Release measures that are also Healthy People 2010 (3) Leading Health Indicators (i.e., lack of health insurance coverage, usual place to go for medical care, receipt of influenza vaccination, receipt of pneumococcal vaccination, obesity, leisure-time physical activity, and current smoking), this release provides overall age-adjusted and crude estimates by sex for all indicators. The NHIS questions used to define the selected health measures are provided in the Appendix.
The race/ethnicity categories for data years beginning in 2003 are defined by using the 1997 Revisions to the Standards for the Classification of Federal Data on Race/Ethnicity (4) promulgated by the Office of Management and Budget (OMB). Subsequent to the Early Releases based on data through 2002, the categories ““non-Hispanic white” and “non-Hispanic black” were changed to “not Hispanic white, single race” and “not Hispanic black, single race.” The term “Hispanic” was changed to “Hispanic or Latino,” and “black” was changed to “black or African American.” However, the text and figures in this report use shorter terms, for conciseness. For example, the category “not Hispanic or Latino, white, single race” in the tables is referred to as “non-Hispanic white” in the text. Race/ethnicity-specific estimates for years prior to 2003, released previously, were based on the 1977 OMB standards and therefore are not strictly comparable with estimates for 2003 and later. However, the changes in the OMB standards have little effect on the health estimates reported here. See Tables XI and XII in Health, United States, 2003 (5) for a comparison of estimates for cigarette smoking and private health insurance coverage using both the 1977 and 1997 OMB standards. In addition, beginning with the 2003 NHIS (first implemented in the September 2004 Early Release), NHIS editing procedures were changed to maintain consistency with the U.S. Census Bureau procedures for collecting and editing data on race/ethnicity. These changes reflect the elimination of “other race” as a separate race response. This response category is treated as missing, and race is imputed if this was the only race response. In cases where “other race” was mentioned along with one or more OMB race groups, the “other race” response is dropped, and the OMB race group information is retained. This change is not expected to have a substantial effect on the estimates.
Data are derived from the three components of NHIS from 1997 through 2009: (i) the Family Core, which collects information on all family members in each household; (ii) the Sample Adult Core, which collects information from one randomly selected adult aged 18 years or over in each family; and (iii) the Sample Child Core, which collects information on one randomly selected child in each family with a child. Data analyses for the 2009 NHIS were based on 88,364 persons in the Family Core (88,129 persons for health insurance estimates), 27,858 adults in the Sample Adult Core, and 11,198 children in the Sample Child Core. The NHIS sample size was reduced to approximately half of the usual quarterly sample in January-March 2009 because of budget shortfalls. Visit the NHIS website for more information on the design, content, and use of NHIS.
Transition to 2000 census-based weights
Estimates were calculated by using the NHIS sample weights, which were calibrated to 2000 census-based totals for sex, age, and race/ethnicity of the U.S. civilian noninstitutionalized population. In Early Release reports prior to September 2003, the weights for the 1997-2002 NHIS data were derived from 1990 census-based postcensal population estimates. Beginning with the 2003 data, NHIS transitioned to weights derived from 2000 census-based population estimates. The impact of this transition was assessed for data from the 2000-2002 NHIS by comparing estimates that used the 1990 census-based weights with those that used the 2000 census-based weights. The results are presented in Tables II and III in the Appendix of the September 2003 Early Release (6). Although the changes for all selected measures are no more than 1 percentage point, the 2000-2002 estimates for all measures have been recalculated in the present report by using weights derived from the 2000 census. An error was made in the poststratification component of the NHIS calculation of weights from January 2006-2008. The error affected “nonminority” person weights. Compared with the corrected weight estimates, those made with the original weights generally differ by 0.01 percentage points, and variances generally differ by 0.001 percentage points.
Implementation of a new sample design
A new sample design was implemented with the 2006 NHIS. In addition to the continued oversampling of black and Hispanic persons carried out in the 1995-2005 NHIS sample design, persons of Asian descent are oversampled in the new design. Also, for the Sample Adult Core, a new level of oversampling began in 2006 for persons aged 65 years and over who are black, Hispanic, or Asian; these older adults are now more likely than other adults in the family to be selected as the sample adult. Some of the differences between estimates for 2006 and later and estimates for earlier years may be influenced by the new sample design. However, the impact of the new design on estimates presented in this report is expected to be minimal.
NCHS creates weights for each calendar quarter of the NHIS sample. The NHIS data weighting procedure pdf icon[PDF – 300 KB] is described in more detail elsewhere. Because the estimates for 2009 are being released prior to final data editing and final weighting, they should be considered preliminary and may differ slightly from estimates that will be made later using the final 2009 data files. Estimates from the 1997-2008 NHIS are based on previous reports and are therefore also based on preliminary data files and not final data files. Differences between estimates calculated using preliminary data files and final data files are typically less than 0.1 percentage point. For 2008, differences may be as high as 1.5 percentage points due to a larger-than-usual number of records being removed for insufficient quality in the final data files. As mentioned previously, estimates for 2000-2002 were recalculated in this report using the 2000 census-based weights that were not included in the final files. See “Lack of health insurance coverage and type of coverage” (Section 1 in this release) for details on special data editing for health insurance variables. For NHIS announcements and more detailed information, visit the NHIS website .
Point estimates, and estimates of their variances, were calculated using the SUDAAN software package to account for the complex sample design of NHIS. The Taylor series linearization method was chosen for variance estimation. All estimates shown meet the NCHS standard of having less than or equal to 30% relative standard error. Point estimates in some figures and tables are accompanied by 95% confidence intervals. Beginning with the June 2006 release, confidence intervals are shown to two decimal places to improve the precision of further calculations. The June 2007 Early Release report used final in-house design variables for estimating variance. Early Release reports other than the June 2007 report use Early Release interim design variables to estimate variance. Trends were assessed by using Joinpoint regression (7). Joinpoint regression characterizes trends as joined linear segments. A joinpoint is the year when two segments with different slopes meet. Joinpoint software uses statistical criteria to determine the fewest number of segments necessary to characterize a trend and the year(s) when segments begin and end. Differences between percentages or rates for current estimates were evaluated by using two-sided significance tests at the 0.05 level. Terms such as “higher than” and “less than” indicate a statistically significant difference. Terms such as “similar” and “no difference” indicate that the statistics being compared were not significantly different. Lack of comment regarding the difference between any two statistics does not necessarily suggest that the difference was tested and found to be not significant. Due to small sample sizes, estimates based on less than a year of data may have large variances, and caution should be used in analyzing these estimates. Patterns for such estimates may change as more data become available. Age-sex-adjusted percentages were calculated for three race/ethnicity groups. For the prevalence of obesity, only age-adjusted, sex-specific percentages are presented because the race/ethnicity patterns in obesity prevalence differ by sex. Similarly, only sex-adjusted, age-specific prevalences are presented for the asthma measures because the race/ethnicity patterns in asthma episodes and current asthma differ by age. Direct standardization was used for adjustment, using the projected 2000 U.S. population as the standard population (8) and using age groups that varied depending on the impact of age on the specific measure. Rates presented are crude rates unless otherwise stated.
Additional estimates for health insurance can be found in the Early Release Program’s other quarterly report, Health Insurance Coverage: Early Release of Estimates From the 2009 National Health Interview Survey (1). To reduce potential errors in reporting Medicare and Medicaid status, two additional questions were added to the health insurance section of NHIS beginning in the third quarter of 2004. Persons aged 65 years and over not reporting Medicare coverage were asked explicitly about Medicare coverage. Persons under age 65 with no reported coverage were asked explicitly about Medicaid coverage. For the present report, estimates that exclude the two additional questions are labeled “Method 1,” and estimates that include the additional questions are labeled “Method 2.” Estimates for years 1997-2003 in this report are generated using Method 1. Estimates for 2004 are presented using Method 2 in figures and both Method 1 and Method 2 in tables. Estimates for 2005 and beyond are calculated using Method 2. Statements about trends or comparisons for 1997-2003 are based on estimates calculated using Method 1. Comparisons between 2004 and going forward are calculated using Method 2. Statements assessing trends that encompass years both prior to and after 2004 take both methodologies into account. Conclusions regarding trends are not made in cases where using one method yields a different result than the same trend analysis using the other method. Note that although both methods may yield the same conclusion, the extent of the increasing or decreasing trend may be larger using one method compared with the other method.
Estimates for 2004 were calculated using both methods, to assess the effect of adopting Method 2. From July through December 2004 (third and fourth quarters combined), with the use of Method 2, the estimates (weighted) for persons without health insurance coverage decreased from 10.4% to 9.9% for persons under age 18 years, from 19.7% to 19.5% for adults aged 18-64 years, and from 1.7% to 1.2% for persons aged 65 and over. Also with the use of Method 2, the estimates for public coverage increased from 28.1% to 29.6% for children under age 18, from 11.3% to 11.4% for adults aged 18-64, and from 89.5% to 93.3% for persons aged 65 and over. There is no impact of the two additional questions on the estimates for private coverage. Additional information on the impact of these two questions on health insurance estimates can be found in “Impact of Medicare and Medicaid probe questions on health insurance estimates from the National Health Interview Survey, 2004” (9).
Future plans for early release of NHIS estimates
The NCHS Early Release Program updates and releases estimates by means of this Early Release report and the health insurance report about 6 months after NHIS data collection has been completed for each quarter. For these two reports (1,10), releases are tentatively scheduled for March, June, September, and December. For the wireless substitution report (2), releases are planned for May and December. New measures may be added as work continues and in response to changing data needs. Feedback on the Early Release mechanism and on the estimates is welcome (e-mail).
Announcements about Early Releases, other new data releases, publications, or corrections related to NHIS will be sent to members of the HISUSERS Listserv. To join, visit the CDC website.
A list of previous Early Release Program reports can be found at
Barnes PM, Heyman KM, Freeman G, Schiller JS. Early release of selected estimates based on data from the 2009 National Health Interview Survey. Hyattsville, MD: National Center for Health Statistics. June 2010. Available from /nchs/nhis.htm.