Measuring Quality of Life in Nursing Homes

              Measuring Quality of Life in Nursing Homes M. Powell Lawton Presentation at U. Amsterdam & Heidelberg University My discussion today is based on research in progress, not on neat and completed findings. My conclusions are questions rather than answers. My goal is to have you think with me as I try to deal with research and measurement issues that are highly resistant to a satisfactory resolution. But all of us who have worked in the environmental areas are, by now, thoroughly resigned to the sad fact that our medium is so difficult to work. But at the same time, we probably get a special sense of enjoyment at the challenge posed by our choice to investigate person-environment relationships. I hope, therefore, that my concentration on unresolved issues will contribute to that challenge for all of us. Our current research task is to design a set of instruments that can be used by governmental inspectors as they monitor the quality of care and quality of life provided by nursing homes in the U.S. The measurement methods that we are using were based on concepts and measures developed over the recent years of development of this area of inquiry. The conceptual background q the areas of E and B and QOL are necessary starting points. I shall indicate briefly how person and environment fit into the concept of quality of life and then describe the way I have partitioned environment. The history of attempts to measure personenvironment constructs is then used to identify 4 recurrent issues that must be addressed in constructing any measures. I'll then describe several existing environmental measures and note how they illustrate one or more of those 4 issues. Our own work on QOL in nursing homes will be the main focus of the presentation. A MODEL OF QOL c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (1/11/99) | Quality of life is the multidimensional evaluation, by intrapersonal a^d social-normative standards, of the total life of the person and her or his environmental context. There are four sectors of QOL, each with its own structural differentiation into domains. Two sectors are objective, in the sense that the attributes to be evaluated are directly observable and capable of being assessed in terms of absolute or social-normative standards. The first, behavioral competence, subsumes successively more complex levels of health, functional health, cognition, time use, and social behavior. The second, environment, is suggested as being composed of the physical and social worlds exterior the body of the person. Two sectors are subjective, depending on people's own evaluations of their quality or lack of quality. Perceived quality of life (PQOL) is the person's satisfaction with, or other evaluation of, specific domains of life; there is no finite list of such domains, but frequently included (e.g., in Andrews & Withey, 1976; Campbell, Converse, & Rodgers, 1976) are such domains as family, marriage, housing, economic state, and spare-time activities. Psychological well-being (PWB) represents the person's most global evaluation of self in environment, including both affective and cognitive manifestations of positive and negative mental health. This conception thus views each sector and each domain within each sector as potentially composing a facet of QOL. Objective facets (e.g., high cognitive performance or a resource-rich environment) are aspects of high QOL because they are socially valued and rewarded. Although such evaluations may not reflect a single individual's evaluation, in the aggregate, people in general prize these characteristics and are more likely to experience higher-quality subjective QOL if they have them. PQOL is what is frequently referred to simply as quality of life~the subjective evaluation of limited aspects of a person's life. Psychological well-being, a global, summary quality is the most general index of QOL, to which quality in the other 3 sectors contribute but do not account for completely. c:\Lawton\Kane\Measuring Quality of Life in Nursing Home^(l/11/99) The present view is that any domain within the four sectors of QOL may be used legitimately as an indicator of QOL in the appropriate circumstance, but that it is important to recognize the domain-specificity of each measure and to choose the right indicator for the specific circumstance of QOL assessment. Although the author's preference is for a model whereby the objective sectors are antecedents of both PQQL and PWB, and PQOL an antecedent of PWB (thereby defining PWB as an ultimate outcome criterion, as in Figure 1), such a causal model is not essential to the assertion that we may look anywhere among thedomajns of QOL for appropriate measurement indicators, depending on the need of the research. Environment thus appears to represent only a limited portion of all that is included in QOL. I have suggested how the objective environment may be partitioned (SLIDE) into domains representing the physical environment, the personal environment (one might also designate the small-group environment), the suprapersonal environment, and the social (or megasocial) environment. (PL elaborate). FOUR CORE ISSUES IN PERSON-ENVIRONMENT RESEARCH 4iberately avoided commentmg extensively on the meaning of the QOL sectc th pts 4 t classific-atian -together may be U3cd to illustfatc(Jwhat I consider ttH)trthe"eore issues of the study of pcrSorP environment relations. (SLIDE) (PL elaborate. Note Barker's tesserae vs. mosaic). (Note that social vs. physical was detailed in previous figure and objective vs. subjective in first. Also that QOL is explicitly evaluative, while E may or may not be. Examples of room dimensions, color, natural vs. built environment). c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (l/l 1/99) PERSON VARIABLES AND ENVIRONMENTAL VARIABLES These 4 core issues are themselves joined into the single mega-issue of our field, how person characteristics and environmental characteristics are related to one another. I begin by acknowledging the philosophical perspective, espoused by conceptual leaders in our field such as Barker, Ittelson, Altman, Stokols, and many others, that person and environment are inseparable; to ask how person and environment are related, as if they are 2 separate constructs, ignores the fact that they are definable only in terms of their transactional togetherness. As die empirical 2? researcher with a compulsion to measure, however, I see no way to avoid measuring them separately if we wish to understand behavior in designed environments. Nonetheless, even granting the heuristic necessary of separate measurement, we still must specify where person and environment are most likely to interact. Some years ago, I suggested the statistical interaction term (P x E) as a possible representation of environmental cognition, environmental perception, environmental preference, and other links between P and E (the Lewin ecological equation). -He&sy- Murray termed the combination of- a pcisun need andean pi ess a~"thcma " I use the term "P-E node" to represent human needs that have this special probability of being linked with E barriers. By specifying such P-E nodes we avoid I A diffusing our research efforts^ the impossible task of measuring everything about both individual and environment. In other words, our focused task is to study person-environment interaction in limited chgaaks defined jointly by critical needs of people and by environmental domains that are both relevant to these needs and responsive to environmental-design intervention. PERSON-ENVIRONMENT NODES c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (1/11/99) In fact, most research devoted to measuring person-environment transaction has used P-E nodes as its beginning point. Some examples are privacy, interaction distance (or proxemics), A and way-finding each of which was the focus of large amounts of research in the early days of person-environment research. As the field developed and measurement efforts progressed, these single-node research efforts have been extended to more comprehensive systems of addressing the full range of such P-E nodes. Despite the wish to frame nodes in terms that recognize the equality of person and environment, such P-E node structures inevitably emphasize one aspect or the other. One of the first was articulated by Brill and Kraus (1970), whose nodes woenamed with obvious environmental referents, such as Communality vs. Privacy or Ambiguity vs. Legibility. By contrast, a personal-deficit orientation guided the design of the Weiss Institute, A our care facility for AD patients in 1973. The physical-design was programmed to account for deficits in orientation, memory, self-care, cognition, discretionary leisure-time behaviors, and social behaviors--a clearly person-focused basis for organizing P-E nodes and an environmental program designed to compensate for disability in a very limited user group. More recently, P-E node systems have been suggested for this same user group in a way to specify personal needs in a more generalized form, enabling the specifics of demented persons' needs to be anchored within more general need categories that apply to people in general (SLIDE). The most recency research of this type resulted in the Professional Environmental Assessment Protocol (PEAP) o? Weisman, Sloane, Norris-Baker, Calkins, and myself, which selected 10 personal needs for measurement of the extent to which dementia-care environments afforded satisfaction (SLIDE). THE MEAP (Moos & Lemke) Before moving to the case study of our current research, I'd like to describe Moos and Lemke's MEAP in a little more detail. The MEAP is by far the most psychometrically jy^1^^ sophisticated environmental assessment device now available (SLIDE). All the 4 core issues are c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (1/11/99) 5 represented: Physical and social environment and objective versus subjective features are discernible in this slide. The 446 items also include a mix of evaluative and purely descriptive items. The majority of individual items are rather discrete, but they also are organized into subscales that include a mix of personal needs, environmental descriptors, and social-process descriptors. Subsets of scales form quality indicators called Staff Resources, Staffing Level, Physical Comfort, Security, Autonomy, Services, Rapport, and Control. Given the breadth and depth of the MEAP, one may wonder why our present team did not simply use, or adapt it, to serve the present purpose. I reproduce our reasoning today because it may be helpful to some of you in future attempts to apply an assessment system to a specific environment, and doing so will constitute an introduction to our nursing-home QOL research goals. 1. The evaluative and descriptive quality of MEAP scales are not differentiated from one another. The unknown relevance of descriptive material to QOL lengthens the instrument unnecessarily, (cite example) 2. Although residents are asked to evaluate some aspects of institutional life, the MEAP includes only 4 quality dimensions to represent the universe of P-E nodes (Physical Comfort, Autonomy, Rapport, and Control). 3. Although objective and subjective elements are each well-represented, many subjective-evaluative items are reported only as opinions given by a single staff member. 4. The subscales and their item components often represent concepts that are too global or abstract either to communicate clearly to external inspectors or to be used by careproviding staff as bases for self-evaluation and self-improvement. 5. Note a highly desirable feature of the MEAP: The use of multiple sources of data--archives, observation, staff reports, and resident reports. NURSING HOME QUALITY OF LIFE ASSESSMENT c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (1/11/99) A I won't spend time here on the background of our study, which is described on page 1 of the handout. But refer now to page 2 for the outline for our approach. 1. Source of information. Note first that multiple sources of data are denoted: ; Observation and informant for institution variables and observation, interview, and proxy informants for the individual resident. As will be seen in the discussion of each measure, QOL as a whole is so difficult to assess as to demand that we do so by multiple approaches. Rather than assuming that the "true" information can be yielded by one source (the "gold standard"), we must proceed by assuming that there is error in every source and that true information may be at best approximated by determining how one source comparefto another. 2. P-E nodes. The second task was to select a limited set of P-E nodes, which you see in the left column marked "domains" on the handout. We clearly depended on earlier research performed by many investigators for selecting personal needs whose satisfaction could be related to environmental 1 attitudes. Note that there is one, labeled "dignity," for which we have been unable to specify a physical environmental support indicator. Two other candidate needs that we mtmtnot represent individually were end-of-life quality and order/structure. End of life we decided was of too limited relevance and order was embedded under functional competence. How well did we do in representing the universe of P-E nodes? environmental researchers to judge. Person and environment, which is represented on p. 2 as/Individual Resident// i ^_--__.--_______ _ A n\ Facility, or unit.f, Presumably everything on the left side is theoretically (a) observable and (b) capable of being evaluated independently of the person being studied. c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (1/11/99) That is a task for each of you and all other 3. This is the aggregate, evaluated by H social-normative standards. On the right side is the Person, whose 3 facets together are designed to represent the quality of life of an individual resident, ~f Aggregated enviroiimentY^. ^Physical environment represents what can be observed in a tour through the physical facilities. Every item (see pp. 3-6) was either gathered from the literature or from a conceptual analysis of its meaning to represent an indicator of positive or negative quality. Their choice was guided by the presumed personal need served by each. To the extent possible the items are objective, but there are subjective terms such as "offensive," "easily controlled," "Attractive," and the like which rely on social-normative judgment rather than simply counting or noting their presence or absence. 7 ^ ^ ^ ^ ^ . Observed Psychfesdcial environment (pages 7-11) is meant to describe aggregate psychosocial characteristics of the institution by means of a research observer's noting, counting instances, and reducing to a single rating indicators of how well each P-E node is served by staff behaviors, care practices, regulations, and policies. To call these psychosocial indicators "objective" is not totally accurate. Most of the indicators require the observer to make a value judgment of some kind. Yet the behaviors on which the judgments are based are openly observable and capable of being tested for interrater reliability. The big problem with assessing the Psychosocial Environment is the dilemma of minimizing subjectivity by repeated observations versus minimizing the time-intensity of assessment by requesting only a single rating. Even as I speak we are undecided about the form of these data. Our original plan was to do repeated time-sampled and setting-sampled 10-minute observations with frequency tallies for each item. That is beyond our research resources and would be totally impossible for the government inspectors to do. We have a possible compromise alternative. Our researcher, like the inspector, will spend several days at the institution, talking to many people and spending time in most areas* It would be possible for the inspector to carry a penciled data form to tally or c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (1/11/99) g describe each relevant indicator whenever it is observed. If standardized for the actual amount of time spent in the institution these tallies could be counted, i^ven this much repeated observation \ and notation time might be too demanding. Therefore: we may end with providing only what is displayed on pp. 7-11, examples of indicators to look for, together with summary Likert-type rating scales. TMs is one of the most difficult cnyiromiiontaUiimonoionG to assess. How would you do it? C. Psychosocial environment as reported by a staff informant. Pages 12 and 13 display a Stclu limited number of questions asked of the most knowledgeable informant (nurse, social worker, administrator). Although their meaning is very similar to the content of Section B, they are limited to fa) difficult-to-observe indicators that (M-refer to relatively well-defined practices or policies. Unlike many items in ^je MEAP, none are attitudinal or evaluative; they may clearly be misreported or distorted, but at least the questions concern matters capable of being defined objectively, and many questions with obviously socially-desirable answers were eliminated. ?f^ -The Person (Resident) Some general comments are needed before discussing the sections representing the resident on page 2. The first is to state explicitly the goal of the Person section: To estimate how well the needs of the P-E nodes are met for individual residents. A basic tenet of QOL v *_ -- ? A research is that the individual is the best judge of what is right for him or her. Therefore we must ask eachpers*n--v first thought, am easy task. I'm sure that most of you here today are aware, however, of the many problems encountered when we ask people how satisfied they are or how they rate various qualities of their current situations There are many factors that produce distortion. I don't have time today to review all of them, but one factor is all-important: Frail and powerless nursing home residents are extremely reluctant to express any criticism of their c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (1/11/99) environment or staff. There is therefore a strong positive response bias when residents evaluate their environments. Another problem specific to nursing homes further reduces our ability to elicit QOL judgments. In the USA, over half of the residents have some cognitive impairment, which makes it difficult or impossible for them to express such judgments. Therefore we need to find some way to counteract this measurement problem. In our approach to the Person aspect of assessment we have therefore adopted a triangulated approach, which seeks similar data directly from the resident, from proxy informants, and from direct behavior observation. D. Staff ratings are not shown in the handout because they are derived from archival sources demanded by the federal government. The Minimum Data Set used in the USA includes periodic staff ratings of Cognitive Performance, ADL, Activities, Social Adjustment, and Depression. Such ratings have their own sources of error. Unless they are part of a regular archive, they are a time-consuming burden on caregiving staff to produce. The rationale for seeking them is clear, however, as a basic principle of consumer psychology recognizes: Behavior is the best guide to consumer preference. In addition, when we seek indicators of input or risk factors in analyzing cross-institutional data, resident levels of cognition and ADL are excellent indicators. When searching for outcome indicators, activity participation, social adjustment, and mental health symptoms are usable/ E. Resident observation. Any attempt to characterize an entire institution in terms of the behaviors, attitudes, or feelings of all residents must deal with the problem of how many residents must be assessed to produce such a characterization. Our samples consist of 50 residents in each of 40 institutions. If we were to attempt to obtain observational data in order to complement self- and other-rated characteristics of residents eyon a- mmi observations on 50 residents would be impossible. Therefore our protocol (shown on p. c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (1/11/99) \Q 14 and top p. 15) was designed as a minimal one-occasion estimate based on behavior observed during or near the time when the resident is interviewed Qfor S&fction FT The environmental indicators (e.g., in Privacy, Individuality and Safety) are fixed attributes of the resident's room and easily observed. The other indicators could be observed primarily during the period (5 minutes) between the time a researcher locates a resident and the time the interview begins. The events to be tallied will be rare. The rationale for measurement is that when one of the rare indicators is observed, it will be relatively discriminating (a "difficult" test item, from a psychometric point of view). Note also that Enjoyment and Comfort will be estimated with the use of the PGC Apparent Affect Rating Scale (SLIDE). We shall get the researcher to perform AARS ratings once during the 5-minute pre-interview period and . observed emotion during the interview. F. The Resident Interview (pages 15-17) will be attempted with all residents, but the researcher will discontinue early if any of several exclusion criteria seem to indicate lack of comprehension or ability to communicate. One rationale for using an instrument on which only half of all eligible respond^ may give valid responses is the possibility that more-intact residents may be able to speak for the less-intact. This is, of course, not known to be true, but it is worth treating as a hypothesis to be tested. The questions included in this draft were assembled with much help from the literature on consumer satisfaction in nursing homes. Several recent studies helped identify and eliminate questions t t ^ 100% of some samples answered as being "very satisfied." The content of the questions includes all aspects of the environment. As we embark on this task, we recognize two guiding principles. One principle is that we must ask the resident. The second principle is that we do so at our own risk, recognizing that the error potential atnd data limitations are great enough to prevent any use of resident evaluations as the "gold standard" for measuring QOL. c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (1/11/99) \\ to represent G. The Family Questionnaire, a potential proxy measure, consists of a subset of the same questions asked in F. Resident Interview. Again, it is a calculated risk to hope that another person is capable of giving an accurate estimate of someone's personal perception of an aspect of the environment. In addition, a family member may not see the resident often enough or in the right circumstances to have knowledge about their experience of some features. The subset of resident questions to be asked of family members excludes items least likely to be answerable; we shall learn which of them elicit more "How should I know?" responses. Our plan is to send the family questionnaire to relatives of all residents. We shall be able to identify the questions on which intact residents and family proxy informants agree and disagree to a significant extent. This information, in turn will help us identify the items for which we can consider the family members' responses an acceptable substitute for the responses of impaired residents WHERE DOES THIS QOL ASSESSMENT LEAD US? In the absence of a gold standard the results of the assessment I have described will necessarily fall short of yielding the definitive indicator of QOL in nursing homes. The strength of our research is that we have searched for the best consensual indicators of QOL and thereby achieve some degree of construct validity. Another strength is that we have given the best chance for multiple dimensions of quality to emerge, which we feel is consistent with the nature of QOL. The final form of the assessment will be necessarily shorter and more simplified than what I have described. Inspectors and self-evaluators will be able to devote less time to the process than the researchers will. Some concepts may be too abstract or complex or poorly defined for operating staff to use. c:\Lawton\Kane\Measuring Quality of Life in Nursing Homes (1/11/99) J2 In addition to the problems with length and demand on the user of the assessment system, there is another major gap, one that is related to the fourth of the core P-E issues, the global versus discrete quality. I hope there are some architects, interior designers, and engineers in our group today. But I worry about how relevant our conception will appear to those of you who have to design, test, or specify very discrete environmental features for older people in either home or institutional environments. We theorists are not the ones to bridge that gap between the tesserae and the mosaic. In our research that produced the Professional Environmental Assessment Protocol, we also used Sloan's TESS (Therapeutic Environment Screening Schedule), which consists of many small design specifics--they are, in fact, tesserae, similar to those you will find in our Section A. Physical Environment. Although we found a good relationship between the PEAP and the TESS as a whole, the discrete TESS items that were related to PEAP were related to many PEAP dimensions, rather than displaying a logical relationship between discrete item content and the meaning of a single PEAP dimension. The present state of the art seems to be that we are better able to assess global quality than its discrete components.