The use of the Short Form (SF)-36 questionnaire for older adults.
Hayes et al.  reported that 26% of their patients in a study with the anglicised version of the SF-36 had missing item data. The authors found that missing item information was higher in those aged 75 and over. O'Mahony et al.  also described poor completion rates for a postal administration of the SF-36 in stroke patients aged 45 years or over, but unfortunately did not report completion rates by age groups.
Missing information is a relevant issue when selecting the most appropriate questionnaire in a research project. Because the SF-36 has very good measurement properties (in all age groups) and reference values are available, we believe that before rejecting the questionnaire or modifying its content, strategies to decrease the missing data should be considered.
In the pilot test of a study using the Spanish version of the SF-36 in outpatients with chronic obstructive pulmonary disease (COPD) , we identified some difficulties in responding to the original `grid' format of the questionnaire, particularly among older patients. Since this was consistent with previous reports of higher missing response rates for items in `grid' format than for items printed independently (i.e. in a question and answer format) , we decided to use the SF-36 health survey without a `grid' format, individually typing each SF-36 question (in large print) followed by the complete response options. Additionally, we instructed interviewers to check the completeness of self-completed questionnaires, identifying (and pointing out to the patient) any missing item responses.
Out of 321 male COPD patients included in the study, more than half were 65 or over: 131 (41%) were 65-74 years and 49 (15%) were 75 or over. Seventy-three percent of the patients self-completed the questionnaire and only 5% had missed one or more items. This proportion is considerably lower than that reported by Hayes et al. , even though the two studies had very similar proportions of self-administered questionnaires (78% in the latter). We did not find a significant difference in missing item data among subjects younger and older than 75 years (4 vs 8%, P= 0.1). Moreover, the proportion of missing information was reduced to less than 2% of the patients when we applied the imputation algorithm recommended by the developers of the questionnaire .
Historically, Spanish older adults have not had much opportunity to become familiar with printed questionnaires or with `grid'-format questions. Although we ignore the relative contribution of the change in the format (`question and answer' vs `grid') and the instructions to the interviewers, we believe that both contributed to the value of the SF-36 for our patients. Simple strategies such as these (that do not change the content of the questionnaire and therefore do not compromise comparability of results) are effective and make it possible to use a robust health status measure in older adults.
MONTS FERRER, JORDI ALONSO Health Services Research Unit Institut Municipal d'investigacio Medica (IMIM) C/ Doctor Aiguader, 80; E-08003 Barcelona, Spain
[1.] Hayes V, Morris J, Wolfe C, Morgan M. The 5F-36 health survey questionnaire: is it suitable for use with older adults? Age Ageing 1995; 24: 120-5.
[2.] Parker SG, Peet SM, Jagger C et al. Measuring health status in older patients. The SF-36 in practice. Age Ageing 1998; 27: 13-8.
[3.] O'Mahony PG, Rodgers H, Thomson RG et al. Is the SF-36 suitable for assessing health status of older stroke patients? Age Ageing 1998; 27: 19-22.
[4.] Gladman JRF. Assessing health status with the SF-36 (Editorial). Age Ageing 1998; 27: 3.
[5.] Ferrer M, Alonso J, Morera J et al. Chronic obstructive pulmonary disease stage and health-related quality of life. Ann Intern Med 1997; 127: 1072-9.
[6.] Ware JE, Keller SD, Gandek B et al. Evaluating translations of health status questionnaires. Methods from the IQOLA Project. Int J Technol Assessment Health Care 1995; 11: 225-51.
[7.] Ware JE, Snow KK, Keller SD, Kosinski M, Gandek B. SF-36 Health Survey. Manual and interpretation guide. Boston: The Health Institute, New England Medical Center; 1993.
SIR--The role of the SF-36 questionnaire in the assessment of health status in older people is coming under much closer scrutiny, and the work of Parker et al.  contributes to this. While it is increasingly clear that self-completion rates for the SF-36 decline with age (depending on the population under study), its use as an interview-based measure is also called into question by this paper. Although cautious in their criticism, the authors conclude that the interview-based approach produced completion rates which were "still unacceptable in one-quarter of the patients". In fact, this referred to the completion rate for one domain, mental health, while all the other items had completion rates of 85% or more (there being no global score for the SF-36), which would appear rather good, given the nature of the patients under study.
Parker et al. also showed that completion rates decline with higher levels of physical disability, potentially biasing results, but this is based on data amalgamated from self-completed questionnaires and interview administration. We therefore do not know whether disability was a factor in the interview-based sample alone and if so what is the hypothesis behind this? The sample included those with cognitive impairment (over half had a CAPE-IOS score of less than 10) and the odds ratios quoted for cognitive function show that it has a much more profound effect on completion rates than physical disability. Lyons et al. found that only about one-third of a community-based sample of elderly people with an Abbreviated Mental Test Score of less than 8 could complete the SF-36 . Therefore the SF-36 appears to be of little value as a measure in subjects with dementia. However, as an interview-based measure we still require further information on its relationship with disability.
Response rates in hospitals in the future may be improved by the ongoing development of the `acute' form of the SF-36, which replaces the term "during the past 4 weeks" with "during the past week", but this remains to be seen.
CHRISTOPHER DYER, ALAN SINCLAIR Department of Geriatric Medicine, The Hayward Building, Selly Oak Hospital, Raddlebarn Road, Birmingham B29 6JD, UK Fax: (+44) 121 627 8304
[1.] Parker SG, Peet SM, Jagger C, Farhan M, Castleden CM. Measuring health status in older patients. The SF-36 in practice. Age Ageing 1998; 27: 13-8.
[2.] Lyons RA, Crome P, Monaghan S, Killalea D, Daley JA. Health status and disability among elderly people in three UK districts. Age Ageing 1997; 26: 203-9.
SIR--We thank Dyer and Sinclair for their interest in our paper.
We agree with the observation that physical and cognitive dysfunction interfere with response and completion of the SF-36 and that this has important implications for the use of this instrument (and possibly others) in older hospital inpatients. Interview administration in our study was associated with an overall response rate of 77%. Among these, scale completion ranged from 75% (mental health) to a maximum of 96%, implying that between 58% and 74% of interview schedules would produce either no return or an incomplete questionnaire on at least one scale. This phenomenon was affected independently by both physical and cognitive dysfunction, implying an effect in both interview administered and self completed questionnaires. We do not have an explanation for this in our data, but we believe that an examination of the effect of specific impairments--such as vision, hearing or manual dexterity--would be worthwhile, particularly to see if functional barriers to completing and returning questionnaires could potentially be overcome (for example by the provision of aids to completion).
Neither do we have an easy alternative for measuring the impact of ill health on the functional health status of older people. We agree entirely that scales for use with ill older people should be developed and validated in this population, before they are recommended for widespread use.
STUART G. PARKER, SUSAN M. PEET Division of Medicine for the Elderly, University of Leicester, Leicester General Hospital, Gwendolen Road, Leicester LE5 4PW, UK Fax: (+44) 116 258 4666 CAROL JAGGER Department of Epidemiology and Public Health, University of Leicester, UK
SIR--We read with interest the papers of Parker et al, and O'Mahony et al. on the suitability of the SF-36 health-related quality of life (HRQL) instrument for older hospital and community patients and for elderly stroke survivors [1, 2]. Both papers report the deficiencies of the SF-36 as a self-report measure with elderly patients, seen in terms of missing information, suitability and relevance of some items. These are important and valid concerns which echo other investigations [3-6].
Parker and colleagues' findings of poorer response rates of the SF-36 by cognitively impaired patients raises a fundamental question about the validity and use of self-report questionnaires in these patients. They also report that the response rate for the SF-36 improves by interview administration, which in itself introduce interview bias--especially if it is administered by health professionals. The introduction of an interviewer to improve response rates will increase the costs of administration of the SF-36. We are also concerned that O'Mahony et al. do not refer to the validation of the SF-36 in stroke survivors by Anderson et al. . As each comes to differing conclusions, we feel that this would have been worth discussing.
Neither paper used the SF-36 as an outcome measure for the purpose for which it was originally developed. Furthermore, we are not told the patients' perceived HRQL. In the paper of O'Mahony et al., a disease-specific measure used alongside the SF-36 would have been informative in determining how well the SF-36 performs compared with other validated measures. If the SF-36 is unsuitable for measuring the impact of ill health on the HRQL of the elderly of older people, what alternative measures could be suggested? The recently developed and validated HSQ12 (derived from the SF-36) has been used as an HRQL measure for elderly people with chronic illnesses [8, 9]. Its authors do, however, advise that an additional disease- and/or domain-specific outcome measure should be used wherever possible.
Measures of HRQL are being used increasingly in health policy decisions, particularly with increased emphasis on clinical effectiveness. However, although patients and clinicians often report positive outcomes from health care interventions, it is often difficult to quantify these changes in older populations using HRQL measures. Should we not develop more appropriate measures for elderly subjects instead of making do with existing instruments?
PETER HOBSON, BIMAL BHOWMICK, JOLYON MEARA University Department of Geriatric Medicine (North Wales), Glan Clwyd District General Hospital, Rhyl, Denbighshire LL18 5UJ, UK Fax: (+44) 1745 534668
[1.] Parker SO, Peet SM, Jagger C, Farham M, Castleden CM. Measuring the health status in older patients. The SF-36 in practice. Age Ageing 1998; 27: 13-8.
[2.] O'Mahony PG, Rodgers H, Thomson RG, Dobson R, James OFW. Is the SF-36 suitable for assessing health status of older stroke patients? Age Ageing 1998; 27: 19-22.
[3.] Hobson JP, Meara RJ. Is the SF-36 Health Survey Questionnaire suitable as a self report measure of the health status of older adults with Parkinson's disease. Qual Life Res 1997; 3: 6: 213-6.
[4.] Hobson JP, Bhowmich BK, Meara RJ. The use of the SF-36 Questionnaire in Cerebrovascular disease. Stroke 1997; 28: 464-5.
[5.] Hayes V, Morris J, Wolfe C, Morgan M. The SF-36 health survey questionnaire: Is it suitable for use with older adults? Age Ageing 1995; 24: 120-5.
[6.] Hill S, Harries U, Popay J. Is the short form 36 (SF-36) suitable for routine health outcome assessment in health care for older people? Evidence from preliminary work in community based health services in England. J Epidemiol Community Health 1996; 50: 94-8.
[7.] Health Outcomes Institute. Twelve item health status questionnaire (HSQ-12) version 2.0 user guide. Bloomington, MN: Health Outcomes Institute, 1996.
[8.] Bowling A, Windsor J. Discriminative power of the health status questionnaire 12 in relation to age, sex and long-standing illness: finding from a survey of households in Great Britain. J Epidemiol Community Health 1997; 51: 564-73.
SIR--The paper by Anderson et al. was published after our paper was submitted . In that study, the SF-36 was used as an interviewer-administered instrument, with a corrected response rate of 76%. Our response rate to a postal version embedded within a larger questionnaire was higher, suggesting that interview administration is not necessarily associated with better response rates. However, despite our higher response rate, the administration of the SF-36 by postal questionnaire is associated with unacceptably high rates of missing data in those who do respond. Whilst Anderson et al. report that the SF-36 was "not compromised by high rates of missing data", they do not give figures for the completion of individual items of the survey.
We could have reported mean scores for each domain of the SF-36 in our study, but because of the data-quality problems highlighted, we did not feel such results would provide a valid and representative picture of the health status of stroke survivors in our study. Furthermore, the comparison of mean scores between studies can be misleading because of differences in case-mix. Anderson et al. provided limited information on the dependency status of their patients, but the low floor effects and high ceiling effects found in the SF-36 scores suggest that the patients studied may have been a less dependent group.
We did not use a disease-specific measure as suggested, since there is no generally accepted stroke-specific quality of life measure. We agree that there is a need to develop and validate quality of life measures which are suitable for use in elderly subjects. The SF-36 may be suitable for use in the general elderly population [2-4], but less so for assessing the quality of life of those disabled by conditions such as stroke or Parkinson's disease [5, 6]. The EuroQol may be a more appropriate quality of life instrument for stroke survivors [7, 8]. Until this is confirmed, it is important to be aware of the potential limitations of the widely used SF-36 in such patient groups.
PAUL G. O'MAHONY, HELEN RODGERS, RICHARD G. THOMSON, RUTH DOBSON, OLIVER F. W. JAMES Department of Geriatric Medicine, Royal Victoria Infirmary, Newcastle upon Tyne, NEI 4LP Fax: (+44) 191 222 5638
[1.] Anderson C, Laubscher S, Burns R. Validation of the Short Form 36 (SF-36) health survey questionnaire among stroke patients. Stroke 1996; 27: 1812-6.
[2.] Hayes V, Morris J, Wolfe C, Morgan M. The SF-36 health survey questionnaire: is it suitable for use with older adults. Age Ageing 1995; 24: 120-5.
[3.] Lyons RA, Perry HM, Littlepage BNC. Evidence for the validity of the short-form 36 questionnaire (SF-36) in an elderly population. Age Ageing 1994; 23: 182-4.
[4.] Weinberger M, Samsa GP, Hanlon JT et al. An evaluation of a brief health status measure in elderly veterans. J Am Geriatr Soc 1991; 39: 691-4.
[5.] Jenkinson C, Peto V, Fitzpatrick R, Greenhall R, Hyman N. Self-reported functioning and well-being in patients with Parkinson's disease: comparison of the short-form health survey (SF-36) and the Parkinson's disease questionnaire (PDQ-39). Age Ageing 1995; 24: 505-9.
[6.] Hill S, Harries, Popay J. Is the short form 36 (SF-36) suitable for routine health outcomes assessment in health care for older people? Evidence from preliminary work in community based health services in England. J Epidemiol Community Health 1996; 50: 9-98.
[7.] Dorman PJ, Slattery J, Farrell B, Dennis MS, Sandercock PAG and the United Kingdom collaborators in the International Stroke Trial. A randomised comparison of the EuroQol and Short Form-36 after stroke. Br Med J 1997; 315: 461.
[8.] Dorman PJ, Waddell F, Slattery J, Dennis M, Sandercock P. Is the EuroQol a valid measure of health-related quality of life after stroke? Stroke 1997; 28: 1876-82.