National Invitational Conference on Long-Term Care Data Bases: Conference Proceedings. Summary of Breakout Sessions


Gerry E. Hendershot, Ph.D., National Center for Health Statistics
Susan S. Jack, National Center for Health Statistics
Joseph E. Fitti, National Center for Health Statistics

GERRY HENDERSHOT: Susan Jack was the lead person for writing the edit specifications for the SOA and is also co-authoring a report on functional limitations which will be published by NCHS. The other co-authors are John Fulton and Sidney Katz of Brown University. Joe Fitti is in our Survey Planning and Development Branch, and was the lead person in the development of the SOA including the questionnaire design, and also has had a major role in the LSOA, which is the follow-up to the 1984 SOA.

QUESTION: It would help if you had someone just briefly describe the long term care questions.

GERRY HENDERSHOT: The first section is on the family, whether they have any children, how far away the children live, that sort of thing. The next section is on community services, including formal community services and informal community contacts. The next section is on occupation and retirement. Conditions and impairments is the next section. There is a vision and hearing impairments and functional limitations section. There is the ADL and IADL checklist and one page on long term care. Not very much, actually, but it does include some questions about previous nursing home stays, knowledge of hospice services available, and a health opinion type question.

It is a complex sample design which is representative of the civilian noninstitutionalized population. Each case has a weight which represents the estimated number of people in the population represented by that sample person. To get accurate estimates for any statistic for the population, you have to multiply the characteristic of the sample person by that sample weight. If you've used any of the commonly used software analysis packages, you will know that is a very simple procedure to use.

How is that weight calculated? There are a number of factors that go into it. First of all we select a sample with the probability proportionate to the population size. The most important factor is the inverse of the probability of selection. We know the probability of each person in the sample having been selected. The first rule is to just weight them up by the inverse of that sampling fraction. There are some other ways you correct for non-response and at the end we do what we call post-stratification, where we compare estimates for 60 age, race and sex groups found in the Current Population Survey (CPS) with our estimates. If they differ, we inflate or deflate our estimates by a ratio to make them agree with the CPS.

SUSAN JACK: On the aging supplement itself, the "60 cells" are appropriate to the core questionnaire because, I believe, 65+ is the only age break for those over 65. They did a slightly more precise age weighting for the SOA itself.

If you have the core questionnaire and its weight for an individual who also had the SOA, they may not have exactly the same weight on the SOA as they do on the core because there was a non-response factor, which was less than 10 percent. They had to in effect, re-inflate them slightly. You will find in some groups, anyway, that there are different weights in the core and the SOA for the same individual. You have to use the weight that is actually on the SOA to come up with any estimates.

GERRY HENDERSHOT: The terms "core" and "supplement" refer to the basic health and demographic questionnaire and the special health topics questionnaires.

The SOA public use tape includes all of the information that was on the SOA questionnaire, plus the person record from the core questionnaire. You can also link that data set to the rest of the core data for that sample person. That is, you have an ID which allows you to link them. To do that you would also have to get the core public use data files, of which there are five.

Any of the information that is available from the core questionnaire can be linked to the information that is available from the SOA questionnaire.

SUSAN JACK: In terms of the financing of long term care and income, we actually have no information directly concerning finances. The only thing we have even vaguely related to that is their health insurance status at the time of the interview.

GERRY HENDERSHOT: We do have family income information.

JOSEPH FITTI: And sources of retirement income.

We have residential ownership or rental status. We do have dollar values in that question series which reflect the current mortgage and the estimated value of the property if they own it, or the amount of rent they are paying if they are paying rent.

QUESTION: What was your experience with proxy respondents?

GERRY HENDERSHOT: You may recall that the respondent rule for the supplement itself was primarily self-response. That was our aim in the original design. The sampling scheme that you have heard described was selecting among those persons who are in the NHIS households for that year of 1984 who were age 55 or over, one-half of those 55-64, and everyone who was 65 and older in those households.

The sampling plan was accompanied with the rule saying we would allow proxies in the event of mental or physical incapability for responding to the interview, or if the selected respondent was not going to be available at all during the period that the interviewer was going to be in the area. They work on a weekly sample basis with roughly a 3 week time-frame to gather all the interviews required. There was a temporal definition for gathering the interviews. That was the second condition by which we may not have actually spoken to the selected respondent.

The consequence of the respondent rule was that this impacted only on a small part of the questionnaire which we felt would not be reasonably responded to by somebody other than the individual. Particularly for the health opinion section which is asking for some information about attitudes and practices. About 8 percent of the data is responded to by proxies.

QUESTION: When you look at reliability comparisons when you do proxy interviews, what do you find in them?

JOSEPH FITTI: I do not have the figures, but there are some slight differences in the proxy data versus the household sample. If you look at the variables and then try to get them to line up, there are not large differences.

GERRY HENDERSHOT: If somebody wanted to exclude the proxy respondents there is a code on the tape which would allow you to identify them and exclude them.

QUESTION: Can you describe the coding of morbidity and the relationship to specific diagnostic attributes?

SUSAN JACK: Let me back up a little bit. On the basic demographic questionnaire, conditions arise in a variety of different ways. Any time they would arise, generally there was a special set of questions on the condition record. Those questions are used to code an ICD code. The coding is actually done by our staff in North Carolina, and they were to some extent trained medical coders, that is, trained using our version of ICD codes. We have something called impairment codes, that nobody else has, known as "X" codes. Very similar questions were asked for any condition that arose on the SOA. If you had a limitation of any kind and if you reported a condition in response to any of these condition lists that is another way to get the condition.

The ADL's and the condition list were, again, coded by the same coders that do our regular coding. Some of those ICD codes we created by computer, or re-code. There is a chronic condition re-code, an acute condition re-code and a re-code that covers all ICD codes. The current version is very similar to the one that is actually listed in the ICD book.

There are actually two different files for the SOA. There is the basic questionnaire, one for every person who was interviewed on the SOA. That could or could not generate a condition record. It is a little bit similar to our basic core questionnaire where you can or cannot have a condition record. There was a condition record generated from ADL, IADL and other condition you reported. The actual source is coded on the condition record, which is not true in our core questionnaire. There is a location that tells you if they recorded this from the specific activity, like dressing, eating or whatever. You have this separate file which has a fair amount of detail about onset.

For the ADL's and IADL's specifically, because we thought it was more useful to have everything on one file, we took a subset of the information on the separate condition record and it is placed at what basically is the bottom of the column asking if you have trouble dressing or whatever. There is a small set of information about the condition that the respondent said caused the limitation. Many people will never have to go out and access the other condition file as a part of the record.

GERRY HENDERSHOT: The SOA public use data is two separate files, both of which are included in a single price when you place your order. The price of $275 gets you both data tapes and all of the documentation. The first tape has the person information both from the core questionnaire and the SOA questionnaire. The second tape has the condition information, one record for each condition that was reported by an SOA sample person.

SUSAN JACK: You could have a person who could report ten conditions on his core questionnaire. If he never records anything to require generating a condition on the SOA, you will not have any condition records for him on the SOA file. You can match to those core condition records. In most cases, however, they will show up on the SOA. In the case where they reported a condition causing limitation of activity on core and they did not report it on the SOA, we went ahead and included it on the SOA.

A person can report something that is fairly general for a limitation of activity. If you ask him specifically about why he has trouble eating, they may well report a different condition which is much more specific. If we could not guarantee that they were the same condition, we just went ahead and called them two different conditions and that did leave us with some proportion of the conditions being reported for a cause of limitation on core and not on the SOA. We did move them and they are identified by a field that says they are only here because they indicated limitation of activity on the basic questionnaire.

JOSEPH FITTI: All these files are linkable. There are identification numbers that tie the SOA person's record, the SOA condition record, the survey basic records or the health insurance records for that year.

GERRY HENDERSHOT: If you have all the data sets you have not only the information about the SOA sample person but the other people in the NHIS household. You could relate SOA data to characteristics of other people in the SOA sample person's household.

The respondent on the core and supplement questionnaires may not have been the same person.

QUESTION: Is there anything on informal caregiving?

SUSAN JACK: If you have a problem with an ADL or an IADL, they did ask the category of people who are helping you and whether or not they were paid. You know whether it is a relative or non-relative who lives in or out of the household. You will have that for any ADL or IADL problem.

They also asked a question whether you needed somebody to help you out for a few days, and if you were disabled for a slightly longer period of time, was there somebody who could care for you. As I say you do not know who the individual is. You just know a category. It also could be five people and you would not know that either. You would just know the category of what kind of person they were.

JOSEPH FITTI: That coding on caregivers is generally, as she described, a relative or nonrelative, and whether that individual is living in or out of the household. There is another category that does distinguish those who mentioned a spouse, child or parent who was the caregiver or helped with the ADL or IADL. So that distinction is made in the relative group.

QUESTION: Is there anything on psychosocial conditions like depression and loneliness?

JOSEPH FITTI: Psychosocial conditions are interesting. It was considered; an original document was designed for an earlier draft of the SOA but it was considerably longer and addressed areas that did not appear in the final version of the questionnaire. One of them was an attempt to get a measure on the mental health status of the sample person. We attempted to administer questions on the patient's mental health status. On the pretest, it was very unsuccessful, we discovered. The decision was not to include that and not to include other elements that were associated with an attempt to measure mental health. It became too problematic in the household setting.

In many of the households there were two people eligible; particularly in those households where the respondents were 65 and older, a husband and wife, we attempted to interview each individually and separately. It did not always happen. Either the wife was reluctant to leave her husband or the reverse. To complicate it even more, they would share the answers in some places. We did try to get only the individual who was addressed to provide the information. That did not always work. That was one of the major problems in trying to do the mental health scale. It was unable to be administered in a private setting.

SUSAN JACK: We do not have psychosocial conditions. There were questions about how good a job they think they are doing taking care of their health. There were psychological types of questions which certainly were not close to a depression scale. How is your current health compared to how it was a year ago? Has your overall health caused you a great deal of worry? Do you think you're as active as other people your same age compared to a year ago? How much control do you think you have over your future health? Apparently that is a question that shows you all kinds of things, at least for some people.

QUESTION: About 8 percent of your sample was black. Did you examine this black subset or were they related to the whole?

SUSAN JACK: We have run some tables looking at Black/White differences. The problem is the very small numbers, particularly for Black males. I have forgotten the total number of Black males in the sample population, but it is very small because they died off, to some extent.

QUESTION: Do you have information on retirement, especially male and female differences?

SUSAN JACK: On the retirement issue, we did ask if they were still working; among the 55-64 years olds, most of them are still working. If they were not working they were asked if there were some types of jobs they could do if jobs were available, and then if they thought there were, we asked them if they wanted to work. If they had ever worked, we did ask them some retirement questions. These included: had they retired more than once; was it because of a physical condition; whether they thought they would have a health problem if they continued to work. It goes into retirement income sources in specific form, but I don't know if anybody has looked at it in terms of sex differences.

GERRY HENDERSHOT: An Advance Data report will be available in about June. It has looked at the retirement data that is in the SOA, particularly the issues of interest and ability to work. It is not a complete analysis of the retirement and occupation section, but it does look at these items about whether or not they were able to work as measured by ten tasks and the ability to perform them, plus the items of opinion about whether they would like to work or not. The retirement section itself, is very extensive in some senses; it was designed to permit a classification of the sample as retired or not retired on a number of scales. One was self-defined status of retirement. A second was retirement as defined by income source. A third was a status of retirement that was more officially used in terms of Social Security Administration (SSA) standards. That data can be looked at in a number of ways on the question of retirement status among the population that we have in the sample. We have begun to look at it now, but I do not have any figures on how it breaks out among males and females. It does look as though most of the data on employment status, says that the males are employed more than the females.

QUESTION: What are the validity and reliability of the health status measures?

SUSAN JACK: We have other surveys in NCHS on some of our data, including hospitalizations and doctor visits; they measure them in different ways. In terms of the IADL's and ADL's, I have seen some other surveys that say we are in the same ball park. I think they can be measured in different ways. The first question that is asked is do you have trouble doing X, but they are allowed to answer that they do not do it for other reasons. Some people have not allowed for an answer like that so if you exclude those people from the population you end up with a very different base than if you included them. Some of the difference has been attributed to who you are actually including in your population. If you force them into a yes/no answer, where do you put the people who do not do this activity?.

JOSEPH FITTI: Some of the more general characteristics of the sample have been looked at in terms of other sources. The distributions by age and sex, they are basic and compare nicely with other data. Detailed information such as the percentage and characteristics of those who live alone, from our study, have been looked at with other data on those who live alone and there are some nice comparisons there also.

GERRY HENDERSHOT: The short answer is that there have not been thorough validation studies of these data. A lot of comparisons have been made with other data sources that give us confidence that we were measuring what we intend to be measuring.

There is a fairly large literature on the NHIS methodology and evaluation of the quality of the data which can be used, and, so far, not a lot specifically on the SOA.

JOSEPH FITTI: We did reinterviews as a reliability check on this study, which is standard procedure for the NHIS, and selected a number of households who were re-contacted after the initial interview was completed. We asked selected items of the SOA, and we re-asked the ADL's and IADL's and a selection among the occupation and retirement section of the sample person that was designated. That procedure was followed.

QUESTION: Regarding the information that you collected on the conditions, and also on the memory and cognitive difficulty, have any analyses been done on that?

JOSEPH FITTI: Not yet. I have not seen anything on the conditions themselves that were collected in the SOA. There are some conditions in the SOA which were designed to produce published estimates among this population. They are probably a better source of the information than the NHIS itself. A couple of them have been looked at. They are in an Advance Data report. Vision and hearing impairments have been looked at, and there is some preliminary information available on that. Urinary problems and incontinence have also been looked at and an Advance Data report has been produced for that.

QUESTION: Within the list of conditions, is Alzheimer's listed on this?

GERRY HENDERSHOT: Alzheimer's, osteoporosis, and broken hip are specifically designated.

SUSAN JACK: If you were interested in looking at Alzheimer's, we do have marginals on all those things, so we can tell you how many cases are showing up. I have run some tables for the ADL and IADL functional limitation report that I am co-authoring, and they were actually run, crossed by their functional limitation status as re-coded by John Fulton and Sidney Katz. In some cases we have regrouped them because the purpose of asking the broken hip was to catch people who did not know they had osteoporosis on the assumption that the broken hip was caused by osteoporosis.

QUESTION: Did you ask if people had representative payee, guardianship, power of attorney or anything of that nature?

SUSAN JACK: The answer to the last part is no. The answer to the first part is, if you want to do it, we would be delighted to have somebody do it, but no, as far as I know. We have, for our basic information, an SED list. We run our basic data outline for many different demographic variables. We have not done that with the aging supplement. I personally would like to run some of those tables, so we can answer some of these questions.

GERRY HENDERSHOT: Maybe I should mention that our analytic staff is relatively small, seven or eight full-time equivalents (FTE's), and we do a survey every year. In 1984 we did the SOA; in 1985 we did health promotion and disease prevention; in 1986 we did vitamin and mineral, insurance and functional limitations; in 1987 we are doing cancer risk factors and cancer control; in 1988 there is another long list. We do not have the staff to do a complete analysis of any of the surveys we do. We try to get out some preliminary reports and assist other people in analyzing the data, but we expect that the main part of the analysis will be done by people who purchase public use data tapes.

The data tapes became available from the SOA in December 1986. Through yesterday we have sold 35 of them, so they are moving very briskly. I expect that within 6 months or so we will begin to see analyses coming out in publications from people who have purchased those public use data tapes.

QUESTION: Do you include in your questionnaire questions about depression? In my country one out of four who's 65+ is continuously depressed.

JOSEPH FITTI: That is what I was alluding to when I mentioned that we tested that and hoped to include depression scales as a set of items. Our experience in actually trying to gather and pretest was such that we were not going to have good data if we had included it. It was excluded, and we do not have it.

GERRY HENDERSHOT: Let me interject one more thing. We have got a small contract in process now, it will be out shortly, to create a microcomputer public use data set from the SOA. It will not contain all cases. It will not contain all items on the questionnaire, and it is not really intended for a rigorous scientific analysis. You will be getting a rough idea of how things lie, for people who then want to go on to analyze the full public use data tapes. That will be available probably toward the end of the year.

QUESTION: You mentioned earlier two other data sets, I think, if my notes are not wrong, the National Household Longitudinal Surveys (NHLS) and LSOA. Would you define those for us and tell us what the status is again?

GERRY HENDERSHOT: The NHLS is a generic term. We also use another term, targeted follow-up surveys, to refer to the same thing. That refers to the capability that the NHIS now has to do follow-up surveys of people interviewed in any particular NHIS. An example of that is the LSOA, which is a follow-up to a person's first interview in the SOA in 1984. They were reinterviewed in 1986. Plans are to reinterview them again in 1988. Part of the LSOA is also to match information obtained in the original SOA and the LSOA with death records through the NDI and Medicare records.

JOSEPH FITTI: Just to clarify, the LSOA is a three-pronged effort.

We are matching with the NDI for all persons who were in the original SOA file, everyone 55 and older, and will be matched against that every year between the years 1984 and 1990 for a 6 year period. We are also going to be looking at Medicare Part A files for persons who were in the 1984 age 65 and older sample for the same years, 1984-1990. The third aspect of the longitudinal follow-up is the reinterview, occurring in 1986 and 1988. For that set we are selecting a sample of the total SOA respondents--only those persons who were 70 and older in 1984, and that in itself has been sub-sampled. Half of those 70-74 have been sub-sampled, and everybody 75 and older. There are 5,151 individuals in the reinterview set.

We have completed the first reinterview, which is essentially a reading of their current functional status repeating the ADL's and the IADL's, and whether they have died. Also, there is a reading on changes, if they have occurred, in living arrangements: whether they are still living where they were; are still living independently; have moved in with someone else; if so, who that person is; or whether they have become institutionalized.

We find, by the way, in our re-contacts, about 12 percent in a 2 years lapse, among the 70 and older group, that we were being told have died in that period. We were checking change in living arrangements and changes in functional status, primarily. We were getting nursing home experience in the interim time and hospitalizations also. That is the content of the reinterview.

This file, which will include NDI match data from the years 1984 and 1985 plus the first reinterview data, will be available in July of this year.

QUESTION: We have the National Long Term Care Survey (NLTCS), so we have information on the disabled. The SOA can give us information on the non-disabled population. Those are two separate reports. Is there a way to put together the data so we can get the whole set of information?

GERRY HENDERSHOT: The two surveys are complementary. I do not think it would be possible to actually pool the data from the two sources, but certainly analyses of the two data sets ought to be set side by side to get a complete picture of long term care for both the institutionalized and noninstitutionalized population.

JOSEPH FITTI: An interesting comparison or combination of data will be the National Nursing Home Survey (NNHS). The 1984 SOA was designed consciously aware of that study and we matched as many variables as we could that appear in both of the studies, the ADL's and IADL's, for instance.

The 1984 SOA did access questionnaires that were either used or planned to be used on other studies of the aging population in the design of its items. There are comparable items in the NLTCS as well as the SOA that could be looked at for comparable data.

QUESTION: Were veterans identified in the survey?

GERRY HENDERSHOT: We did in the SOA address the population aged 55 and older, and we do have in that set a number of veterans that are available. I think I have the raw sample count here. I do not have what it projects to the total population, but among the total of 16,148 people in the SOA, there are 2,812 veterans of World War II. They are certainly a set that would probably be possible to look at independently, if it was desired.

SUSAN JACK: We have that information on core, and which period of military service they had.

QUESTION: Is there anything about the veteran's income, for example, pensions that you might have gotten out of the income question?

JOSEPH FITTI: We have not specified Veterans Administration (VA) or veterans income as a separate category.

GERRY HENDERSHOT: The health insurance supplement which was asked the same year, has information on coverage for health care by veterans' programs.

QUESTION: I am interested in more information about the sample design, which you mentioned was a targeted sample.

JOSEPH FITTI: The LSOA is a follow-up, essentially, of the SOA sample over time. It has already begun, and the long range plan will carry forward through the year 1990, 6 years. There are three major lines of follow-up that will be incorporated in the full LSOA. One will be a follow-up of all persons in the 1984 SOA aged 55+ through links with the NDI which the NCHS maintains. The objective here is to establish some information on mortality, through the registration of certificates. We will also extract from those files information about cause of death to expand the whole set of illness and disability on the sample.

That was one area of follow-up. The second area of follow-up will be done with those persons who were 65+ in 1984 among the SOA sample who will be followed through matching the Part A records for the period 1984-1990. There is a great deal of information there. We are not at the present time planning to extract all of it. Our primary concern and interest in that file is information on hospital utilization and some information on medical expenditures that will be available in the Part A file. That, too, will be done on an annual basis through the year 1990.

The third line of follow-up that is part of the LSOA is re-contact. This will be done among selected samples out of the 1984 SOA group, the first of which has already been designated and the first re-contact was with persons 70+ in 1984. We have reinterviewed them one time already by telephone in 1986 with an interview that essentially was designed to update their functional status, obtain any information on changes through readings of the ADL's and IADL's at that point. Changes in living arrangements is a second area of interest that is in the reinterview conducted in 1986, and also information about nursing home stays that occurred between 1984 and the re-contact, and hospitalizations that occurred between 1984 and the re-contact.

We have 5,151 persons that fall in this category for the re-contact interview in 1986, of which, by the way, we got 93 percent identified in terms of a status, including approximately 12 percent who were reported as dead during the re-contacts. Those will be validated through our NDI matches, obviously.

SUSAN JACK: Let me tell you the general sections that are covered in the SOA. First of all, separate from the SOA there was another supplemental questionnaire that year on health insurance, so there is health insurance data for these people. It is physically on a different data tape. There is information on family structure relations, support and living arrangements. There is also information on community and social support services. There is some information on occupation and retirement. Conditions and impairments, ADL's, IADL's and nursing home stays.

There is help with care. There are really only two questions about whether or not someone can help you: if there is somebody to help you for a few days, or somebody to help you for more than a few days. There is a question about whether or not they know about hospices. There are some health opinion questions. If there is a need for help with ADL's, IADL's or if they mention a condition on the condition list, a condition record is generated.

There are two different files in the SOA. There is a file that has a person record on the front which is all the core information, including income, education, or anything that is on the core personal tape. There is a condition file which is all the conditions generated as a result of going through the SOA.

GERRY HENDERSHOT: Susan referred to the core and the supplement. The core questionnaire is the questionnaire that is used every year, and contains information on everybody in the NHIS sample household. The supplement, the SOA questionnaire, was for a sample of older people in the households. The information from both sources on an SOA sample person is available. We have the core public use data tapes which are available from NTIS, and the SOA public use data tapes which are available from NCHS. If you get both sets of data you can link any of the data from either of the two questionnaires.

SUSAN JACK: We have problems with Medicaid eligibility even on the core data tapes. It varies from month to month who is actually eligible for Medicaid. We do have Medicare information. We do have some Medicaid information because it is on the health insurance tape. They were asked whether or not they have used Medicaid. A person who does not use Medicaid usually does not know if they are eligible for it.

We do list possible sources of income to remind them of things that are income. It is down to the thousand dollar increments, at least for the lower income groups, which is a lot better than we used to do. A lot of people do not know. Close to two-thirds of them are over $20,000. The poverty index is on core, and that deals partly with whether or not there is a person 65+, the number of children in the household, and their income status.

JOSEPH FITTI: We ask the respondents if they had retirement income and whether it was from SSA, railroad retirement, private employer or union pension, government employee pension, military retirement or some other source. That pertains only to retirement income among the SOA sample. The additional income data is from the NHIS.

QUESTION: How do you match the LSOA to Medicare?

JOSEPH FITTI: We actually asked in the reinterview for their health insurance claim number which will be used against the Medicare file. We can also do matches against Medicare using the Social Security number, which we also have from the 1984 interview.

QUESTION: What do you have on the rural/urban dimension?

SUSAN JACK: There is the non-SMSA, and SMSA information and there is a breakdown of SMSA between central city/non-central city. There is also a farm/non-farm non-SIVISA kind of breakdown which is essentially urban/rural,

QUESTION: I noticed that you did a special project on home care utilization a few years ago. I was wondering whether you have given any thought to perhaps making it a routine collection item.

GERRY HENDERSHOT: I guess you are referring to the 1979-1980 home care supplement?

That has about the same information that is contained in the 1984 SOA on ADL and who provides care, if a person has a limitation in ADL's or IADL's.

SUSAN JACK: The categories are not the same and it was not addressed the same way. We asked if you needed help with something. There are only two different possibilities, if you needed help with an ADL or IADL in general. In 1984, the activities were specific and there are categories of persons who helped you, either a relative or a non-relative who lives in the household, or a relative or non-relative outside the household. You know a category but not a person.

There are also two questions on disability. For a short period of time or if it were a somewhat longer period of time, was there somebody who could help you, and the same four categories of helpers.

In the functional limitation supplement in 1986, those same questions are on, but again, it is for a specific activity which is not the same exactly as it was on the home care supplement. You could make similar inferences from it.

GERRY HENDERSHOT: We have collected data in 1979, 1980, 1984, and 1986, and it is being collected in the LSOA. We do not have any plans at the moment to collect that kind of data on a regular basis.

SUSAN JACK: Technically, there is core information that does relate to it. I do not know how large the numbers are. In the doctor visit file there are questions about whether or not you have seen a medical person in the last 2 weeks. For somebody who is getting routine care from, possibly, a visiting nurse, in theory it should show up in there as a home visit, because we know the place of that visit. You could combine several years and look at the home doctor visits in the last 2 weeks to see what kind of person is providing them. That is not going to pick up somebody who comes in and cleans up your house because you can not do it, but it should get medical personnel, like a visiting nurse or somebody.

QUESTION: What would be involved in weighting the NNHS to reflect the total actual nursing home population and then literally creating some kind of merged file.

JOSEPH FITTI: That would be ideal, but I do not think that is really possible. What is possible will be parallel sets of data for the institutionalized and noninstitutionalized population.

QUESTION: Can you weight them so that both sets of data are appropriately reflective of national estimates?

JOSEPH FITTI: Whether you could combine them to get, say, total persons 65+, including those institutionalized and noninstitutionalized, I do not think that is going to be possible.

QUESTION: We were talking about repeats of the aging supplement, a repeat already done and a repeat projected. There were 5,000-6,000 health persons. Will the same people be interviewed again?

JOSEPH FITTI: Correct. This is part of the LSOA, the reinterview aspect. We have 5,000 from the 1984 SOA who were aged 70+ who are being followed in that set.

QUESTION: What were the questions on social and family support about?

JOSEPH FITTI: These questions on community support, as such, appeared only in the 1984 SOA. They are not being repeated in the re-contacts. The areas of community support specifically asked in the original SOA were whether or not they used a senior center, special transportation for the elderly, whether they had meals delivered to their homes, such as Meals-on-Wheels programs, whether they go to senior centers specifically to eat meals, whether they use a homemaker service, whether they use a service that routinely checks by telephone, visiting nurse service, home health aide, or an adult day care center utilization. These areas are the ones that were specifically asked about in the 1984 SOA. The reinterview in 1986 and 1988, as it is planned, will not repeat these questions. We will ask whether or not they are still living in the same place, whether they have moved to an institution or not, whether they have moved into another location to live with another person or whether someone else has come in to live with them as a living arrangement change.

QUESTION: What other kinds of questions are being repeated?

JOSEPH FITTI: The ADL's and IADL's are asked specifically. The change in living arrangement will be obtained. Whether or not they have had nursing home stays since the 1984 or previous interview, what hospital stays they have had and what physician contacts they have had in that interim period. That essentially is the follow-up reinterview data.

QUESTION: On the LSOA, are people who are 55-64 years old but are SSDI disabled beneficiaries lost, or can they be merged with Part A data files? Second, what are the procedures for getting the Part A data from the NDI match?

JOSEPH FITTI: I do not think anybody but the NCHS would be able to do that, because we have proprietary information which must be protected that we gathered in the 1984 interview that only we can handle. What you would need to make the links, we could not give to you. It is an interesting concept. We have not thought of that, actually. We plan to match only the 65+ set. The identifiers would be needed to make those links and we can not provide those to the public.

QUESTION: You would not want us to get the follow-up data on mortality. How are we going to do that?

JOSEPH FITTI: You would have to make a request to NCHS and see if we will do it.

GERRY HENDERSHOT: We will be providing mortality information from our NDI matches on an annual basis for that sample.

JOSEPH FITTI: The LSOA plans for data releases, by the way, are annual during this 6 year period. The first set of data will be out in July of this year, the public use file. Mortality data will be primarily from the NDI match. That is, we will identify those persons in our original sample who have come up in the NDI files and provide probably a date of death and a cause of death bit of information. We also have another source of mortality data among the set who are being reinterviewed.

GERRY HENDERSHOT: The NDI is a system operated by NCHS which allows researchers who have identifying information on persons to discover whether or not a death certificate has been filed for that person. Then researchers have the information they need to go back to the states where the death certificate was filed to get additional information that is recorded on the death certificate. It is that NDI mechanism that is being used to identify persons in the LSOA who have died.

Another general issue that comes out of this discussion that is just gone on is that there are many things that could be done in the way of follow-up surveys or new surveys. To do follow-up surveys, because of confidentiality limitations, it has to be NCHS that actually does the field work and data collection work. We do have an interest and a real wish to cooperate with people in the research community to mount such special follow-up surveys. We have the mechanisms and the staff to do that. We would be glad to talk to anybody who has an interest in doing special follow-up surveys.

There is also a possibility of repeating the SOA as a cross-sectional survey in future years.

QUESTION: How often do you revise the core items on the survey, and to what degree are insurance related issues covered in the core? I do not see any reference to insurance topics and I am wondering if that might not deserve to be on the permanent core survey.

GERRY HENDERSHOT: The general rule of thumb is that we revise the questionnaire about every decade. It was last revised in 1982. We are currently in the process of designing a new core questionnaire with a target of 1989. We did ask insurance questions every other year, in even numbered years.

SUSAN JACK: In spite of the fact it is not listed in 1984, it was done in 1984. You have to understand we have this timing problem on the questionnaire. There is the argument that if things do not change every year why ask them every year. The health insurance has been asked when there is a specific reason for asking it. For instance, in 1983, the reason that it was asked was that was a year of very high unemployment, and we added this supplement for half a year to check whether there was a problem with less of insurance coverage.

GERRY HENDERSHOT: In a general way, the answer to your question about why we do not collect health insurance information is that we have a limited amount of time that we can spend in the household collecting information. There are lots of competing interests for data. In view of that, you might be distressed to learn that the major purpose of our current revision of the core questionnaire is to reduce its length from about 30 minutes to about 15-20 minutes. The reason for doing that is that because of budget pressures the NCHS can no longer support the full survey from its own budget. The way we are making up the difference between what NCHS can include in its budget and the cost of the survey is to offer our services as a data collection mechanism for health statistics to the research community, primarily, the public health service research community. We are increasing the amount of time that is available for these special health topics questionnaires from the current 25-30 minutes to more like 40-45 minutes, or longer, and making that time available to other agencies who have an interest in collecting data on a particular health topic, such as health insurance.

QUESTION: Have you collected information on insurance for long term care, or do you plan to in the future?

SUSAN JACK: There is no good reason why we could not. We have from time to time added questions on health insurance, the 1983 supplement on unemployment. In 1986 we added a question about dental health care coverage, which we had never asked before. Certainly if we did an aging supplement again, obviously, it is a very good question to ask.

GERRY HENDERSHOT: The SOA was a special supplement to the NHIS in l984. The NHIS is an on-going survey in the field, literally all the time. Each week interview assignments go out and each week's assignments are a nationally representative sample of the civilian noninstitutionalized population in the U.S. It is a personal interview in each sample household. We get information on everybody in the household for certain basic health and demographic items which are the same year after year.

In addition, in any given year we do one or more special surveys on selected public health topics. Usually that is for a single randomly selected sample person in the household.

In 1984 the special health topic on which we focused was health of the aging. We selected all persons in NHIS households who were 65 years of age and older for that questionnaire. A random 50 percent sub-sample of persons 55-64 got the same questionnaire. We have everybody 65+, and half of those 55-64. That is about 16,000 cases total, and 11,000 65+.

The major topics include measures of functional limitation, ADL's and non-ADL's, which have gotten a lot of attention at this Conference. Some other physical conditions such as vision impairment, hearing impairments, and so on, as well as use of community services, and a little bit of information on history of use of nursing homes. Any of that data can be combined with data that are collected on the basic health and demographic questionnaire, which includes a lot of information on health conditions, as well.

You have got two sources of data, two questionnaires for the same sample of people.

On the question of periodicity of our collection of data of this kind, there was a similar survey in both 1979 and 1980, which we have called the home care supplement. That did include questions also on functional limitations, the ADL's and IADL's. Unfortunately, the questions in the two surveys were asked a little differently. In the 1979-1980 survey, the ADL and IADL questions were asked something like, because of a health problem or physical condition, do you need or receive help? In 1984, we asked, because of the health or physical problem, do you have difficulty, first of all, in each activity. If they did report any difficulty, we asked how much difficulty, how difficult it was. Then we asked if they received help. It is difficulty and receipt rather than need for and receipt, in 1984. We thought that was an improvement in asking questions, but it does make it different from the 1979-1980 data.

In 1986, for persons 65+, the questions used in the 1984 survey on ADL's and IADL's were repeated. We now have exactly comparable data on ADL's and IADL's for cross-sectional samples in 1984 and 1986, and similar but not exactly comparable data for 1979 and 1980.

There are no plans to repeat this on a periodic basis as things now stand. Essentially, we do the basic health and demographic questionnaire each year, and then the supplements we do in any particular year are determined by a process of inviting proposals from other federal agencies for data collection, reviewing those proposals, and deciding on the basis of technical merit, public health importance, and the availability of funding which of those topics we will do. Whether or not we do another cross-sectional survey like the SOA in the future depends on a number of things which are not well known at this point.

This is a place where I should mention, too, that there is a LSOA which is a follow-up to the 1984 SOA, which is funded by the NIA. That takes the 1984 SOA sample and they will be followed in the NDI, all of them, for at least 6 years. The NDI is an operation of the NCHS which allows a researcher, if he has certain information about an individual, to determine whether or not a death certificate has been filed for that person. It also tells you what state you go to get the death certificate information.

The 1984 sample will be followed for at least 6 years in the NDI so that we will know if they died, when they died, and can relate that to the information collected in 1984.

Those who are 65+ in 1984 will also be followed in the Medicare records, matching to the Medicare records with the cooperation of the Health Care Financing Administration (HCFA). In addition to those two continuing data collections on the 1984 sample, periodically sub-samples of them will be reinterviewed. The first reinterview has already been completed. That was done about August 1986. A sample of those 70 years of age and older in 1984 was reinterviewed, mostly by telephone. They were followed into nursing homes, or wherever they happened to be. We did not just go back to the household they lived in 1984. Part of the purpose of that reinterview was to find out what happens to people over time; do they go into nursing homes, die or what? Each year there will be a new public use data tape released containing the information for the reinterview group on the original SOA, the reinterview data, the NDI information and the Medicare information. That will be updated each year. The first of those data tapes will be available in July of this year. I have been told by Richard Suzman of NIA that his agency is very interested in receiving applications for grants to do research on both the SOA and the LSOA.

We plan to reinterview that same sample again in 1988. Beyond that, it is difficult to say. I mean, it is getting too far in the future, but the hope is that we can follow them for a long period of time. There were about 5,000 in the first reinterview.

Comparability to the NLTCS? I do not really know enough about the NLTCS to say much about that. I guess an important difference is that the SOA and the LSOA are representative samples of the whole community living elderly population, whereas the NLTCS sample was screened for disability. The NLTCS, obviously, has more kinds of data input and has more frequent follow-up.

On insurance, periodically we do a special supplement to the NHIS on insurance. We do not get very much information, but the insurance data were collected in 1984. We had both the insurance supplement and the SOA, os it is possible to get the insurance information and link it to the information from the SOA.

Incidentally, there is an annual publication from the NHIS which is entitled Current Estimates from the National Health Interview Survey. In health insurance, we ask whether they are covered by Medicare; does the plan pay any part of hospital expenses; does it pay doctor or surgeon bills; was it obtained through an employer or union. It also tells whether they are on AFDC, Supplemental Security Income, Medicaid (SSI), any other public health system's program that pays for health care; also, military types of health care coverage such as CHAMPUS. Those have been asked roughly every 2 years for a long time. They were asked again in 1986 so they could be linked to the functional limitation data which were collected in 1986.

We have hospice care. We asked if they are aware of what a hospice is, and we also asked them if there is hospice care available to them in their area. That would be a perception of the availability of that particular kind of service in the area. We have, in some supplements, asked questions about distance from medical doctors and usual source of care.

We decide each year what special health topics are going to be covered. We are planning right now for 1989. We just received proposals a couple of months ago for topics for 1989. That is a wide open process. We receive proposals from anybody who cares to submit them. This is a relatively new procedure for the NHIS and it arose, frankly, because of budget restrictions. NCHS is not able from its own budget to fund the complete costs of the survey. It can fund only about two-thirds of the cost. The other third we generate by, in effect, selling time on the survey to anybody who needs to have health data collected, provided it is in our judgment within the mission of our agency and serves an important public health need.

We have received some proposals from the private sector organizations and expect to receive more in the future. We got one from the American Speech and Hearing Association (ASHA) this year. I think it was the first one we have gotten from a private sector organization.

In 1984 there were approximately 16,000 people interviewed, 65 years of age and older. In 1986 a sample of those who were 70 years or older in 1984 was reinterviewed. There were about 5,000 of them. The plan is to reinterview those 5,000 or their survivors, again in 1988. The capability exists, of course, to reinterview any of the others, as well, the younger ones. Depending on interest and the availability of funds, probably some of the others will also be reinterviewed.

The SOA refers to the cross-sectional 1984 survey. All of the follow-up activities which are being conducted come under the rubric of the LSOA.

I guess the strength of the SOA is that it is a large nationally representative sample of all persons living in the community over the age of 55 and the data were collected in a standard way. They were collected in sufficient detail to allow you to define a limitation for your particular purposes, because we get whether or not they have difficulty, how much difficulty, and whether they receive the help of another person.

You can use difficulty as your cut-off, a lot of difficulty as your cut-off, or receiving the help of another person as your cut-off. It gives you some options on how you would want to measure it. Of course, the option you choose will have a big effect on your estimate of prevalence. A lot more people find things difficult than receive help.

As I said earlier, I am not an expert in long term care, but sitting through 2 days of hearing experts talk about it I think that standardization in the measure of disability and functional limitation is something that is very much in need. If we could agree on how we want to measure it, then I think we could get some standardization in the way it is actually measured.

View full report


"87cfproc.pdf" (pdf, 2.52Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®