Strengths and Limitations of the 2007 Survey of Adult Transition and Health (SATH) data

The 2007 SLAITS Survey of Adult Transition and Health (SATH) data have a number of strengths and limitations that analysts be aware of as they may impact use of the data. The 2007 SATH used a follow-back survey design in an attempt to locate and contact 10,933 eligible cases originally included in the 2001 National Survey of Children with Special Health Care Needs (NS-CSHCN). In 2001 these children were 14 to 17 years of age and lived in English-speaking households; in 2007, the same subjects were young adults 19 to 23 years of age. The primary goals of the SATH are to examine the current health care needs of the original survey subjects and to better understand their transition from pediatric health care providers to adult health care providers.

STRENGTHS

Largest survey to date conducted with this subgroup. To our knowledge, this file contains the largest publicly available sample to study this transition period for children with special health care needs at the national level. The topics covered in this survey are important and policy-relevant, and yield a rich source of information on a rare population.

Identical variables assessed at two points in time, six years apart. Because of the follow-back design feature, we were able to include many variables from the 2001 NS-CSHCN public use file (PUF) in the 2007 SATH PUF for each case.

Single data file. The 2001 and 2007 variables are included in a single PUF to make the file easier to use.

Self-reported data from the young adults. In 2001, the survey respondent was the parent or guardian who was most knowledgeable about the health and health care of the sampled child with special health care needs. In 2007, data were collected directly from this subject.

Cooperation and refusal rates. Although this survey has a low response rate (17.5% using standards of the American Association of Public Opinion Research), this rate is primarily driven by the inability to contact original households and not due to a high refusal rate. We were unable to contact approximately 7,409 (68%) of the original 2001 parents or guardians. Once were able to contact the 2001 respondent to gather current contact information on the young adult, many eligible subjects chose to participate. Of the young adults we were able to contact, almost all of them cooperated; the AAPOR cooperation rate was just over 98%.

LIMITATIONS

Sample design. There were 10,933 cases originally identified in the 2001 NS-CSHCN as eligible for follow-up. The 2001 survey was a cross-sectional random-digit-dial (RDD) telephone survey, and was not originally designed as a longitudinal survey. Very limited six-year-old contact information was initially available to locate the 2001 households. Telephone numbers sampled in 2001 were first dialed to locate the 2001 respondent. If this person could be located, current contact information was requested for the SATH-eligible young adult so he/she could be contacted directly. If the telephone number was not useful, the case was researched using commercial means to try to find the original respondent. Many of these attempts were not fruitful.

Response rate. We located and completed interviews with 1,916 young adults, of which 1,865 cases are included in the SATH data file. The unweighted SATH interview completion rate was 17.5%.

Non-response bias. Sampling weights for the 2007 SATH dataset were derived from the original 2001 NS-CSHCN sampling weights. These SATH weights were then adjusted for non-response to the SATH interview, using demographic information from the 2001 NS-CSHCN. However, no analysis of non-response bias has been completed. Young adults from households that did not move between 2001 and 2007 may be overrepresented in the dataset, and these young adults may differ in important ways from young adults living in households that did move between 2001 and 2007.

Sample design variables in the 2007 SATH public use file (PUF) are STATE (strata), IDNUMR (primary sampling unit or PSU), and WEIGHT_SATH (final SATH interview weight).

Because of the sample design, response rate, and the possibility of non-response bias, analysts may wish to consider the SATH sample as a purposive or convenience sample rather than a random sample. Data from a purposive or convenience sample should not be used to conclusively test hypotheses. However, these data are worthwhile for:

hypothesis generation, or develop ideas to be substantiated in data analyses with other most robust datasets (e.g., those drawn from random samples)

exploratory analysis of relationships between variables, recognizing that results may only be narrowly extrapolated and may not be generalizable to the national population of young adults with special health care needs.

Analysts should include a disclaimer that credits any analyses, interpretations, or conclusions reached to the author(s) (i.e., recipient of the data file) and not to NCHS, which is responsible only for the initial data.

This page last reviewed September 09, 2008