Original Article
Proxy Use of the Canadian SF-36 in Rating Health Status of the Disabled Elderly

https://doi.org/10.1016/S0895-4356(98)00090-0Get rights and content

Abstract

Professionals have often depended on proxies to assess health status of elderly individuals unable to provide this information. This study set out to estimate agreement between elderly subjects and their proxies on perceptions of health status, when patients were inpatients in a rehabilitation setting or outpatients in a day hospital, and to determine if there were overestimations or underestimations by proxies. Eighty-three consenting elderly patients living in Montreal, Canada, completed a generic health-status questionnaire, the SF-36, and named a significant other and a health care provider who knew them well. These individuals completed the same questionnaire on behalf of the patients. Agreement between patient and proxy was assessed pairwise (patient/health professional and patient/significant other) using intraclass correlation coefficients. The possibility of a biased rating by type of proxy was determined via paired t-tests. When concordance within respondent pairs was examined, results indicated only poor to moderate agreement regardless of setting and type of proxy for all scales. A trend toward underestimating by proxies was found. Proxy completion of health status questionnaires did not provide similar information, but these preliminary findings need replication.

Introduction

To describe outcomes of care, traditional measures of morbidity and mortality are currently being supplemented by measures of health status. Such measures better reflect the way individuals feel and function in their daily lives [1], as they are made up of items important to, and often generated by, patients. Usually, health status measures incorporate dimensions such as physical, psychological, and social functioning; role performance; and perception of well-being, and they might or might not include the presence of symptoms 2, 3, 4. The literature contains a large number of instruments designed to assess health status 5, 6, 7 and hundreds of studies of diverse patient populations employing these devices.

There is a general consensus among researchers that health status is at least partially subjective, making the individual whose life is being assessed the best source of information 1, 8. As very old and/or ill individuals may not be able to provide information about their health status [9], investigators are confronted with whether or not to use alternative sources of information such as a health care provider or someone close to the patient [8]. This is an important issue. In one study, 20% of elderly community dwellers and 50% of nursing home residents were unable or unwilling to provide information about health status [10]. In other studies, exclusion rates varied from 22% to 36% because subjects with cognitive or linguistic impairments could not complete the questionnaires 11, 12. Moreover, studies of the elderly that fail to use proxies often have a high rate of nonresponse or missing data [13]. The use of a proxy potentially offers a solution to nonresponse; however, whether the information collected is similar to that provided by the subject is not definitive.

Two studies that included elderly subjects compared ratings of physicians and their patients on the Spitzer Quality of Life Scale [14]. Slevin and colleagues [15] concluded that physicians could not adequately rate on behalf of the patients. Pearlman and Uhlmann [9] found that the physicians consistently underestimated the patients’ health status regardless of medical condition and suggested that this may have been caused by the physicians’ difficulty in detecting the more subjective aspects of quality of life. Another study [16], using part of the Functional Status Questionnaire [17], reported that physicians underevaluated or even failed to recognize functional disabilities that were revealed by the patient. Yet dysfunction was overstated by physicians for patients relating no disability.

Significant others as proxies have been more commonly studied. Four investigations used the Sickness Impact Profile (SIP) [18]. In a study by McCusker and Stoddard [19], individuals were administered the SIP while a family member, their primary caretaker, completed the questionnaire on their behalf. Correlations were moderate, with patients tending to higher estimates for the physical scales and those of the family member indicating higher scores for scales in the psychosocial domains. When the SIP was used by Rothman et al. [20], patients’ and informal caregivers’ scores on the physical dimension were highly correlated; again, patients evaluated themselves higher than did the proxies. The correlations between psychosocial dimension scores were weak, suggesting that proxy scores were not valid substitutes for those of the patient in this area. Epstein et al. [21] found that patients’ and significant others’ ratings were moderate to high for social activities, overall health, and functional status but lower for emotional health and satisfaction with care. Sneeuw et al. [22] reported fairly close agreement between patients and their proxies but described systematic differences for several SIP scales in which proxies noted that patients had more disablement than did the patients themselves.

In a study of elderly patients with hip fracture and their significant others, Magaziner et al. [12] examined patient-proxy responses related to health and functional status before surgery. Agreement was higher for observable tasks, such as walking, than for less observable functions such as handling money or for general ratings of health. Overall, proxies tended to underrate patients’ performance. The authors concluded that proxies may have exaggerated patient dysfunction as a consequence of perceived burden or to encourage placement.

Although, in general, there appears to be an understimation of the patients’ health status by proxies, the extent of the disagreement is not consistent across studies and is influenced by a number of factors. For example, agreements were higher for better educated and noncognitively impaired patients and their proxies 19, 20, but gender, socioeconomic status, and marital status did not influence concordance 4, 21. Advanced age on the other hand, resulted in lower correlations than when younger patients were studied [20]. Living arrangements and caregiver duties also affect the levels of agreement. For significant others, there was better concordance if the proxy lived with the subject but was not providing care 12, 19, 20.

In summary, literature on the use of proxies in assessing health status of elderly individuals is limited [4], and results are influenced by both patient and proxy characteristics as well as the domain of quality of life being tapped. A need exists for further examination of the value of using proxy information in specific situations. In particular, the living situation and the intensity of the relationship between the proxy and the patient have not been adequately investigated. This study, thus, set out to determine the degree to which a proxy could provide accurate information about the health status of an elderly individual and whether the living situation of this individual was influential in determining which type of proxy, significant other or professional caregiver, should be selected. Specifically, the objective was to estimate the agreement between elderly patients and their proxies when the patients were inpatients in a rehabilitation hospital or outpatients in a day hospital and the proxies were professional caregivers or significant others in the lives of patients. A secondary objective was to determine whether proxies consistently reported higher or lower scores than patients.

The Medical Outcome Study 36-Item Short-Form (SF-36) [23] was chosen as the measure of health status in this study because it is increasingly being selected to assess outcomes of care. During the past few years, it has been translated, culturally adapted, and psychometrically evaluated in more than 20 countries, including Canada [24], and this process is ongoing [25]. In addition, it has been found suitable for use with elderly subjects 26, 27, 28, 29, 30 and has been suggested for use with patients undergoing rehabilitation 31, 32.

Section snippets

Subjects

The study was an interrater reliability study in which the self-perceived health status scores of elderly subjects were compared with their health status scores as provided by two different proxies. Elderly subjects were recruited from three health care institutions in Montreal. At the Centre Hospitalier Côte des Neiges, subjects from the inpatient rehabilitation unit and the day hospital were included; the Jewish Rehabilitation Hospital provided inpatients; and the Royal Victoria Hospital

Overview of the Sample

One hundred twenty respondent pairs (79 patient/professional and 41 patient/significant other) participated in the study. Table 1 classifies the type of respondent pairs by health care setting. Table 2 groups the 83 elderly subjects by health care setting and language of questionnaire completion and demonstrates that there were similar numbers of individuals responding in French and English divided between inpatient rehabilitation and day hospital settings. In only one pair did the patient

Discussion

Despite uncertainty surrounding the use of proxy respondents, researchers and clinicians have been depending on them to assess health status of elderly individuals unable to provide this information. In this study, we set out to examine the extent to which significant others or health care providers, identified by the patients as knowing them well, could complete a measure of health status that corresponded to the information provided by the patients themselves. In addition, we were interested

Acknowledgements

The authors acknowledge technical support for the data collection, verification, coding, and analyses: S. Beaulieu, C. Corrigan, A. Levy, S. Mayo, F. Rice, V. Sakadis, and S. Scott.

References (52)

  • A. Bowling

    Measuring HealthA Review of Quality of Life Measurement Scales

    (1991)
  • C.S. Burckhardt et al.

    Quality of life with chronic illnessA psychometric study

    Res Nurs Health

    (1989)
  • I. McDowell et al.

    Measuring HealthA Guide to Rating Scales and Questionnaires

    (1987)
  • N.K. Aaronson

    Quality of life assessment in clinical trialsMethodological issues

    Control Clin Trials

    (1989)
  • R.A. Pearlman et al.

    Quality of life in chronic diseasesPerceptions of elderly patients

    J Gerontol Med Sci

    (1988)
  • J. Coroni-Huntley et al.

    Established Populations for Epidemiological Studies of the ElderlyResource Data Book

    (1986)
  • R. DeHaan et al.

    Measuring quality of life in stroke

    Stroke

    (1993)
  • D. Blazer et al.

    The epidemiology of depression in an elderly community population

    Gerontologist

    (1987)
  • M.L. Slevin et al.

    Who should measure quality of life, the doctor or the patient?

    Br J Cancer

    (1988)
  • D.R. Calkins et al.

    Failure of physicians to recognize functional disability in ambulatory patients

    Ann Intern Med

    (1991)
  • A.M. Jette et al.

    The Functional Status QuestionnaireReliability and validity when used in primary care

    J Gen Intern Med

    (1986)
  • M. Bergner et al.

    The Sickness Impact ProfileDevelopment and final revision of a health status measure

    Med Care

    (1981)
  • J. McCusker et al.

    Use of a surrogate for the Sickness Impact Profile

    Med Care

    (1984)
  • M.L. Rothman et al.

    The validity of proxy-generated scores as measures of patient health status

    Med Care

    (1991)
  • A.M. Epstein et al.

    Jr. Using proxies to evaluate quality of life. Can they provide valid information about patients’ health status and satisfaction with medical care?

    Med Care

    (1989)
  • K.C.A. Sneeuw et al.

    Assessing quality of life after strokeThe value and limitations of proxy ratings

    Stroke

    (1997)
  • Cited by (59)

    • Preadmission quality of life can predict mortality in intensive care unit-A prospective cohort study

      2014, Journal of Critical Care
      Citation Excerpt :

      One could therefore argue that the information obtained may not accurately reflect the HRQOL of the patient. Although the use of proxies in assessing HRQOL in an ICU setting has been validated [11-13], some studies performed in other patient populations have mixed results [23,24]. Although the use of proxies may bias our results, we believe that this inclusion technique is the most appropriate in an ICU setting, in which patients are often mentally or physically incapable of completing a questionnaire unaided.

    View all citing articles on Scopus
    View full text