Integrated Health Interview Series

The variable SAMPWEIGHT is once again supplying the correct values.

The lastest IHIS data release has added data from 2010, includes material from the historical public use files from 1963 to 1968, and brings the number of integrated variables available through IHIS to approximately 12,000.

Users can now analyze IHIS data online, using high-speed tabulation software developed at UC-Berkeley. The online tabulator is especially useful for creating frequency and cross-tabulation tables, calculating means, and significance testing. Experienced data users may wish to use the tabulator to get quick or preliminary results before downloading a dataset. Users who do not have experience using a statistical package can now answer simple research questions by creating tables online. In introductory classes on statistics or public health, students can use the tabulator to formulate and test hypotheses about public health trends and differentials, getting results within seconds. A detailed user note provides explanations and examples of how to analyze IHIS data on-line.

The latest IHIS data release increased the number of variables to over 7000. New variable groups include adult and child use of vitamins and herbs, falls and balance problems, hearing and vision problems, and use of aids to mobility. Also newly available is a weight (MORTWTSA) suitable for use with variables from the sample adult files in conjunction with variables on survey participants' mortality and cause of death.

The latest IHIS data release added data from 2007, 2008, and 2009 and increased the number of variables to over 5000. New variable groups include adult and child mental health, cancer family history, use of complementary and alternative medicine, health knowledge, adult leisure-time physical activity, imputed income and earnings, HPV immunization, and mortality and cause of death of survey participants.

The BMICALC (IHIS-calculated Body Mass Index), which was created by programming, initially included some incorrect values for 1982-1995 due to an error in the programming statements. Respondents with unknown values for height were included in the BMI calculation, which yielded artificially low values. The version of BMICALC available as of August 16, 2010 no longer includes those incorrect values. Researchers who used the BMICALC variable for 1982-1995 prior to August 16, 2010 should redo their analyses using the corrected version of the variable.

The original versions of the linking key variables NHISPID (for person-level variables) and NHISHID (for household-level variables) were incorrect. A coding fix for NHISPID was posted in IHIS news on January 27, 2009. The versions of NHISPID and NHISHID available as of August 16, 2010 are correct, and the coding fix is no longer needed.

REVISED: Feb 20, 2009. The linking key variables NHISPID (for person level variables) and NHISHID (for household level variables) in the IHIS extract system are currently incorrect. We have identified and corrected the problem. However, the corrected NHISPID and NHISHID variables will not be available until the next IHIS data update.

In the interim, a simple coding fix for NHISPID can be implemented by users in their own IHIS data. For Stata users, adapt the following code for the years in your IHIS extract:

gen str16 test = "002006" + substr(nhispid,7,10) if year == 2006
  replace test = "002005" + substr(nhispid,7,10) if year == 2005
  replace test = "002004" + substr(nhispid,7,10) if year == 2004
  replace test = "002003" + substr(nhispid,7,10) if year == 2003
  replace test = "002002" + substr(nhispid,7,10) if year == 2002
  replace test = "002001" + substr(nhispid,7,10) if year == 2001
  replace test = "002000" + substr(nhispid,7,10) if year == 2000
  replace test = "001999" + substr(nhispid,7,10) if year == 1999
  replace test = "001998" + substr(nhispid,7,10) if year == 1998
  replace test = "001997" + substr(nhispid,7,10) if year == 1997
  replace test = "1996" + substr(nhispid,5,12) if year == 1996
  replace test = "1995" + substr(nhispid,5,12) if year == 1995
  replace test = "1994" + substr(nhispid,5,12) if year == 1994
  replace test = "1993" + substr(nhispid,5,12) if year == 1993
  replace test = "1992" + substr(nhispid,5,12) if year == 1992 & hispoversamp !=1
  replace test = "1991" + substr(nhispid,5,12) if year == 1992 & hispoversamp ==1
  replace test = "1991" + substr(nhispid,5,12) if year == 1991
  replace test = "1990" + substr(nhispid,5,12) if year == 1990
  replace test = "1989" + substr(nhispid,5,12) if year == 1989
  replace test = "1988" + substr(nhispid,5,12) if year == 1988
  replace test = "1987" + substr(nhispid,5,12) if year == 1987
  replace test = "1986" + substr(nhispid,5,12) if year == 1986
  replace test = "1985" + substr(nhispid,5,12) if year == 1985
  replace test = "1984" + substr(nhispid,5,12) if year == 1984
  replace test = "1983" + substr(nhispid,5,12) if year == 1983
  replace test = "1982" + substr(nhispid,5,12) if year == 1982
rename nhispid pidbad
rename test nhispid
sort nhispid

For SAS users, adapt the following code for the years in your IHIS extract:

data data2;
set data1;
format test $16.;
  if year=2006 then test="002006"||substr(nhispid,7,10);
  if year=2005 then test="002005"||substr(nhispid,7,10);
  if year=2004 then test="002004"||substr(nhispid,7,10);
  if year=2003 then test="002003"||substr(nhispid,7,10);
  if year=2002 then test="002002"||substr(nhispid,7,10);
  if year=2001 then test="002001"||substr(nhispid,7,10);
  if year=2000 then test="002000"||substr(nhispid,7,10);
  if year=1999 then test="001999"||substr(nhispid,7,10);
  if year=1998 then test="001998"||substr(nhispid,7,10);
  if year=1997 then test="001997"||substr(nhispid,7,10);
  if year=1996 then test="1996"||substr(nhispid,5,12);
  if year=1995 then test="1995"||substr(nhispid,5,12);
  if year=1994 then test="1994"||substr(nhispid,5,12);
  if year=1993 then test="1993"||substr(nhispid,5,12);
rename
  nhispid=pidbad
  test=nhispid;
run;

For additional information about merging variables from the original NHIS public use data files to IHIS data, please see the user note on merging.

Stata and SAS syntax files for merging variables from the original NHIS public use data files to IHIS data are being revised. Syntax files for 1969 to 1981 are currently unavailable.

In the next couple of weeks, new variables will begin to be added to the IHIS system. This will include new variables from the NHIS 2007 data released this summer, mortality variables from the NHIS-Linked Mortality Files, cancer screening, and cancer family history variables. We will also extend many of the current IHIS variables back in time.

On October 15 2008, IHIS introduced a new look to its website. This upgrade also included a significant redesign of the variable browsing and data extraction system. IHIS users now have more control over how they view the variables available, and they can now select variables from any point in the documentation system. The data extraction system also gives users more flexibility in how they select the years of data and variables they want to download.

Further information about the new browsing and data extraction options, in addition to general aspects of the data series, are available on the FAQ page.