Using the Nottingham Health Profile (NHP) among older adult inpatients with Varying Cognitive Function

Quality of Life Research (2006) 15: 575 585 Ó Springer 2006 DOI 10.1007/s11136-005-3691-0 Using the Nottingham Health Profile (NHP) among older adult inpatients with Varying Cognitive Function Eva Baro 1,2,3, Montse Ferrer 1, Olga Vázquez 4, Ramo n Miralles 3,4, Angels Pont 1, Asuncio n Esperanza 4, Antoni Ma Cervera 4 & Jordi Alonso 1,3 1 Health Services Research Unit, Institut Municipal d Investigacio Me`dica, IMIM-IMAS, Barcelona, Spain (E-mail: MFerrer@IMIM.ES); 2 Health Outcomes Research Department, 3D Health Research, Barcelona, Spain; 3 Universidad Auto noma de Barcelona, Barcelona, Spain; 4 Geriatric Department of the Centro Geria trico Municipal Hospital de la Esperanza Hospital del Mar, Instituto de Atencio n Geria trica y Sociosanitaria (IAGS) IMAS, Barcelona, Spain Accepted in revised form 2 October 2005 Abstract Background: High rates of missing, non-applicable items and insufficient reliability have been frequently reported as limitations of the generic Quality of Life questionnaires for older patients. The Nottingham Health Profile (NHP) might be more suitable as it contains easy to respond (yes/no) items covering moderate-to-severe health deterioration. Objectives: To assess feasibility, reliability and validity of the NHP in disabled, older patients. Design: Cross-sectional study. Setting: Acute care hospital. Subjects: 134 inpatients aged 65 with severe disability, abnormal cognitive function, or other persistent health problems precluding their discharge. Methods: The (interviewer-administered) NHP, Mini-Mental State Examination (MMSE), Barthel Index, and diagnostic information were recorded. Results: Completion rates varied from 98% of the 49 patients with normal cognition (MMSE 21) and 86.3% of the 51 with moderate cognitive impairment (MMSE 10 20), to 5.9% of the 34 with severe cognitive impairment (MMSE<10). Cronbach s alpha of the total NHP score was near 0.9 (0.82 and 0.87 for patients with MMSE 21 and 10 20, respectively; p=0.291). The correlation between Physical Mobility of the NHP and Barthel Index was also similar in both cognitive groups (0.39 and 0.40). Conclusion: Interviewer-administered NHP is suitable, reliable and valid, even in patients with moderate cognitive function. Key words: cognitive disorders, frail elderly, geriatric assessment, Quality of Life, questionnaire Introduction With the aging of society, the maintenance of Quality of Life for older adults is increasingly important. Although disability scales based on Activities of Daily Living are the traditional outcome measures in older adults, the use of questionnaires developed in the last decades to measure Health-Related Quality of Life (HRQoL) provides both conceptual and practical advantages, such as a greater ability to discriminate between groups or individuals, and to track changes over time. A number of studies carried out among older adults using generic questionnaires, such as the Short- Form 36 [1 12], the Duke Health Profile [13] or the Nottingham Health Profile (NHP) [14], generated some doubts about their suitability. Table 1 summarises these limitations. The most frequently reported limitation is a high proportion of missings. This could be partially due to difficulties with self-completion, related to the high prevalence of specific impairments (such as vision problems or manual dexterity) among older adults. In fact, completion

576 Table 1. Limitations found in the last 10 years, concerning the use of generic HRQoL questionnaires in older adults [Reference] Author, year Age (sample size) Population Instrument (response rate) Administration Inapplicable/ (missing data) a,b lack of relevance Metric properties: reliability#, validity and responsiveness Hayes et al. (1995) [1] Hill et al. (1996) [2] Brazier et al. (1996) [3] Hobson and Meara (1997) [4] Stadnyk et al. (1998) [5] Mallinson (1998) [6] 65 y (n=195) 75 85 y (n=47) 75 y (n=380) >65 y (n=66) 65 y (n=146) 65 y (n=56) Hospital outpatients & general practice (patients with MMSE>19). Mental or continence service Randomised controlled trial of hip fracture (women) Parkinson s disease from community register (patients with MMSE >30) Frail rehabilitation inpatients and outpatients Communitybased occupational therapy or physiotherapy SF-36 outpatients (68%) general practice(90%) self-completed (61%) a 41% of missing interviewed (12%) a data concentrated on 9 of 36 items of work or vigorous activities. SF-36 interviewed Functional tasks were considered as inappropriate for older people by authors SF-36 self-completed(11 32%) b Missings EQ-5D self-completed (<10%) b concentrated on 10 items of PF*, V & GH Perception of SF-36 dimensions SF-36 self-completed (24%) a 80% of missing data concentrated on physical activity and daily and work activities Poor sensitivity to change (compared with in-depth interviews). Cronbach s alpha < 0.7 for SF & GH dimensions SF-36 interviewed (34%) a ICC < 0.7 for RE, V & SF Poor sensitivity to change (compared with Spitzer Quality of Life Index). SF-36 self-completed by post (40%) a Issues concerning lack of relevance and misunderstanding are reported by authors Issues concerning face validity and formatting are reported by authors

577 O Mahoney et al. (1998) [7] Parker et al. (1998) [8] Carver et al. (1999) [9] Seymour et al. (2001) [10] Iglesias et al. (2001) [11] Novella et al. (2001) [12] >45 y (n=92) 65 y (n=463) 65 y (n=333) 58 95 y (n=314) 70 y (n=422) >60 y (n=148) Stroke patients from general practice Medical or surgical inpatients and outpatients Communitydwelling population (patients with a modified MMSE score >50) Rehabilitation patients from day hospital Randomised trial of hip protectors for fracture prevention in general practice Mental disorder inpatients SF-36 self-completed (>25% RP & E) b SF-36 Inpatients (46%) Outpatients (71%) Ambulatory (93%) self-completed (38%-8% inpatients) b SF-20 interviewed Important domains are lacking (i.e. memory, environmental adaptation) SF-36 interviewed Cronbach s alpha and/or ICC were < 0.7 for: RP, SF & RE dimensions in normal and cognitively impaired patients, GH & MH dimensions only in cognitively impaired patients SF-12 (6.6%) a York SF-12 (8.5%) a SF-36 73% of patients needed help from interviewer MMSE<10 (>20%) a MMSE>9 (<7%) a (0.6 8.8%) b Cronbach s alpha <0.7 in 2 out 8 dimensions ICC < 0.7 in 7 out of 8 dimensions.

578 Table 1. Continued Metric properties: reliability#, validity and responsiveness Administration Inapplicable/ (missing data) a,b lack of relevance Population Instrument (response rate) Age (sample size) [Reference] Author, year Cronbach s alpha and/or ICC <0.7 in 7 out 80% of patients needed help from interviewer MMSE<10 (>25%) a MMSE >9 (<5%) a Duke Health Profile Mental disorder inpatients >60 y (n=148) Novella et al. (2001) [13] of 10 dimensions Cronbach s alpha and/or ICC below 0.7 in Energy, Social isolation and Emotional Reactions NHP 16% self-completed & 84% interviewed Hospitals or centres specialised indementia care 60 y (n=145) Bureau-Chalot et al. (2002) [14] (14 33%) b # Unsatisfactory reliability has been defined as having a Cronbach s alpha coefficient or an intraclass correlation coefficient <0.7. a Patient proportion with at least one missing item in the questionnaire. b Patient proportion with at least one missing item in each dimension. PF Physical Functioning, V Vitality, GH General Health, SF Social Functioning, RE Role Emotional, RP Role Physical, MH Mental Health, BP Bodily Pain. rates tend to improve when the questionnaire is interviewer-administered [1, 8, 15, 16]. Another limitation is that some items, or even health domains, are non-applicable to individuals with severe disability. Unsatisfactory reliability has also been described for some domains of these questionnaires. Finally, and just as importantly, it has been argued that individuals with cognitive impairment are not capable of a subjective HRQoL assessment. However, some empirical evidence suggests that older adult patients with a moderate degree of cognitive impairment can carry out reliable and consistent HRQoL assessments [17, 18]. In the context of a clinical trial, we needed to select an instrument suitable for disabled older adult inpatients to assess the effectiveness of a comprehensive geriatric intervention programme [19] after acute-care hospital discharge. The NHP was chosen because of its theoretical advantages: (a) items covering moderate-to-severe health status deterioration [20, 21]; (b) the yes/no response format makes it easier to administer, compared to the multiple choice Likert Scales of other questionnaires; and (c) it had shown acceptable results in a study conducted in older adults with dementia [14]. The present study explored the feasibility, reliability and preliminary validity of the NHP in disabled older adult inpatients, and compared these properties according to cognitive impairment, in order to provide new evidence about its hypothesised suitability for use in this population. Methods Data was obtained before planned hospital discharge. This study was approved by the hospital review board, and informed consent was obtained from all patients. Study design and participants All consecutive older adult inpatients (aged 65 or over) visited by the geriatric consultation team between October 1999 and June 2001, of a public, teaching, acute-care hospital, meeting the inclusion criteria were recruited. Inclusion criteria were that the subject must suffer from at least one of the following conditions: 1. Inability to stand from a chair/bed or to walk

579 Table 2. Characteristics of patients and Nottingham Health Profile (NHP) completion according to cognitive function (n=134) Overall, N=134 Cognitive status p-value Normal (MMSE 21), N=49 Moderate impairment (MMSE 10 20), N=51 Severe impairment (MMSE <10), N=34 Gender (women), n (%) 87 (65.4%) 28 (57.1%) 38 (74.5%) 12 (36.4%) 0.183* Age, mean (SD) 79.77 (7.33) 78.82 (7.26) 81.57 (6.63) 78.42 (8.09) 0.081 # Diagnostic categories, n (%) Locomotor 43 (32.6%) 18 (36.7%) 19 (37.3%) 6 (18.8%) 0.134* Pulmonary 20 (15.2%) 10 (20.4%) 6 (11.8%) 4 (12.5%) Cardiovascular 21 (15.9%) 10 (20.4%) 7 (13.7%) 4 (12.5%) Neurological 20 (15.2%) 3 (6.1%) 8 (15.7%) 9 (28.1%) Others 28 (21.2%) 8 (16. 3%) 11 (21. 6%) 9 (28.1%) Length of stay Mean (SD) 30.26 (23.54) 29.92 (21.18) 25.69 (22.98) 38.06 (26.40) 0.064 # {observed range} {1 160} {6 82} {1 160} {3 109} Charlson Index, n (%) Absent comorbidity (0) 23 (17.4%) 11 (22.4%) 12 (23.5%) 0 (0.0%) 0.018* Low comorbidity (1) 40 (30.3%) 16 (32.7%) 16 (31.4%) 8 (25.0%) High comorbidity (>1) 69 (52.3%) 22 (44.9%) 23 (45.1%) 24 (75.0%) Barthel Index, mean (SD) 25.79 (19.32) 34.37 (15.85) 25.61 (16.69) 12.94 (21.25) <0.001 # observed range 0 75 7 68 0 75 0 73 NHP completion, n (%) Completed 94 (70.1%) 48 (98.0%) 44 (86.3%) 2 (5.9%) Non completed 40 (29.9%) 1 (2.0%) 7 (13.7%) 32 (94.1%) P-value was calculated by applying T-test (#) or Chi-Square test (*) depending on the nature of the variable. MMSE: Mini Mental State Examination ranges from 0 (severe cognitive impairment) to 30 (normal cognitive function) [23]. Barthel Index ranges from 0 (high dependence) to 100 (high independence). without assistance; 2. Abnormal mental status, assessed by the confusion assessment method [22]; or 3. Persistent medical, functional or psychological problems that interfered with their discharge to home. Patients with terminal illnesses were excluded. Patient evaluation Functional status was evaluated using the Barthel Index (BI) (ranging from 0 100) [23], and the Charlson Index [24] was used to assess Comorbidity. Patients were categorised according to cognitive function as measured with the Mini- Mental State Examination (MMSE) [25], into: (a) normal cognitive function (MMSE 21); (b) moderate cognitive impairment (MMSE 10 20); and (c) severe cognitive impairment (MMSE<10, or subjects unable to perform the MMSE owing to aphasia, severe mental status deterioration or decreased awareness [19, 26]). HRQoL was assessed by means of the Nottingham Health Profile (NHP), which was interviewadministered by a trained geriatrician [20, 21]. The NHP contains 38 items divided into six domains (Energy, Pain, Emotional Reactions, Sleep, Social Isolation, Physical Mobility). Each item in the questionnaire has a weight attached, providing an estimate of the distress associated with the state described. Weights for the Spanish items were obtained from a sample of 1123 individuals by using Thurstone s method of paired comparisons [27] as with the original questionnaire. A score was calculated for each domain by applying Spanish weights and following the procedures recommended by the developers of the questionnaire [28]. The overall NHP score was obtained by averaging the domain scores. NHP total and domain scores range from 0 (no perceived distress) to 100 (maximum perceived distress). Statistical analysis Chi-square and Fisher s Exact Test (when appropriate) were used to compare categorical variables

580 among patients according to cognitive function. In order to assess the feasibility of the NHP, the completion rates were calculated. According to feasibility results, patients with severe cognitive impairment were not included in the subsequent analysis. Suitability of items in this population was assessed in terms of the percentage of patients with not applicable and missing items. For those items considered not applicable by more than 5% of patients, the association with disability was evaluated by testing differences between groups defined by Barthel Index tertiles. The NHP domain distributions were studied, and in order to compare NHP scores between Table 3. Percentage of patients with not applicable items and missing information on the Nottingham Health Profile (NHP) by cognitive status Overall, N=92 Cognitive Status Chi-Square test p-value Normal (MMSE 21), N=48 Moderate impairment (MMSE 10 20), N=44 Missing items, % a Energy 0.0 0.0 0.0 n.a. Emotional 6.5 4.2 9.1 0.339 Pain 1.1 0.0 2.3 0.294 Sleep 4.3 2.1 6.8 0.266 Social 5.4 8.3 2.3 0.200 Physical mobility 2.2 4.2 0.0 0.171 Not applicable, % b Energy 0.0 0.0 0.0 n.a. Emotional 2.2 4.2 0.0 0.171 Pain 64.1 62.5 65.9 0.733 Sleep 2.2 2.1 2.3 0.950 Social 0.0 0.0 0.0 n.a. Physical mobility 1.1 2.1 0.0 0.336 Not applicable, % c pain at night 0.0 0.0 0.0 n.a. unbeareable pain 0.0 0.0 0.0 n.a. constant pain 0.0 0.0 0.0 n.a. painful to change position 1.1 0.0 2.3 0.294 pain when I m sitting 1.1 2.1 0.0 0.336 pain when I m standing 39.1 39.6 38.6 0.926 Barthel index >36 9.1 9.1 9.1 1.000 Barthel index 24 36 44.4 53.8 35.7 0.343 Barthel index <24 65.6 76.9 57.9 0.266 p-value <0.001 <0.001 0.029 pain when I walk 40.2 39.6 40.9 0.897 Barthel index >36 12.1 13.6 9.1 0.706 Barthel index 24 36 51.9 53.8 50.0 0.842 Barthel index <24 59.4 69.2 52.6 0.348 p-value <0.001 0.002 0.046 pain... stairs or steps 64.1 62.5 65.9 0.733 Barthel index >36 36.4 36.4 36.4 1.000 Barthel index 24 36 74.1 84.6 64.3 0.228 Barthel index <24 84.4 84.6 84.2 0.975 p-value <0.001 0.003 0.028 a Proportion of patients with any missing item in each dimension. b Proportion of patients with any not applicable item in each dimension. c Proportion of patients with not applicable in each item of pain dimension.

581 patients with normal cognitive status and those with moderate cognitive impairment, the t-test or Mann Whitney U- test (when distribution was not normal) were used. Floor and ceiling effects were calculated (proportion of patients with the worst (100) and the best (0) possible scores, respectively). Scaling success was calculated as the percentage of times in which an item scale correlation was higher for the hypothesised domain than for the other domains of the questionnaire. Reliability of the NHP was assessed using Cronbach s alpha coefficient of internal consistency, which measures homogeneity between items. The Alpha test program [29] was used to test statistically significant differences in Cronbach s alpha coefficients. Based on an expected Cronbach s alpha of 0.85 for patients with normal cognitive status [21], and 0.65 for patients with moderate impairment [14], the sample size of each group was calculated to include 49 patients in order to detect differences with a statistical power of at least 80% at a significance level of 5% [30]. To assess validity, the association between the Barthel Index and the NHP Physical Mobility (the NHP domain which measures the concept closest to functional capacity) was tested. We hypothesised that they would be moderately correlated, and Spearman coefficients were duly calculated for each cognitive group. Analyses were performed using the Statistical Package for the Social Sciences [31], except where otherwise indicated. Results The study included 134 patients, of which 49 (36.6%) showed normal cognitive function, 51 (38%) moderate cognitive impairment, and 34 (25.4%) severe cognitive impairment. It was possible to administer the NHP to 70.1% of the total sample: to 98% of the patients with normal cognitive function; 86.3% of those with moderate cognitive impairment; and to just 5.9% of patients with severe cognitive impairment (p<0.001) (Table 2, shows patients characteristics). The mean age was 80 years old, with a high comorbidity rate. Average hospital stay was 30 days. The most common diagnoses on admission were for disorders of the locomotive system. The higher the cognitive impairment, the lower the Barthel Index score, with a mean ranging from 34.4 to 12.9 (p<0.001). Of the 92 patients with normal cognitive status or moderate cognitive impairment who completed the NHP, 64% considered some of the items not applicable (see Table 3). With regard to this high percentage, it should be noted that items considered not applicable were almost all concentrated in the Pain dimension. More than 60% of the patients, regardless of their cognitive status group, considered at least 1 of the 8 Pain items as not applicable. In fact, the presence of not applicable items in the NHP was very low overall (<3%), if we exclude items from the Pain dimension. Nevertheless, 3 of the 8 items from the Pain dimension ( Pain when walking, Pain when standing, Pain when going up/down stairs ) accounted for nearly all the not applicable items, with no differences according to cognitive status. The percentage of patients who considered these three items to be not applicable in each group, defined by the BI, showed association with functional status. The lower, or worse, the BI score, the higher the probability of an item being considered not applicable. For example, only 9% of the patients with a BI score >36 considered the item Pain when standing as not applicable, whilst 44% of the patients with a BI of 24 36, and 66% of the patients with a BI <24 considered it not applicable. There were no differences between patients with normal cognitive function and those with moderate cognitive impairment. Patients with normal cognitive function showed a less deteriorated HRQoL than those patients with moderate cognitive impairment (46.8 vs. 60.6 for total NHP score; p=0.087), although statistically significant differences were showed only for Energy and Emotional domains of NHP (see Table 4). None of the patients in the sample obtained the minimum (floor) or maximum (ceiling) total score possible in the NHP. Floor effects of >20% were observed only for Energy and Physical Mobility in patients with moderate cognitive impairment, whilst ceiling effects were present for Energy and Social domains only amongst the group with normal cognitive function. Scaling success was high (80% or over) in both groups for most NHP domains. Cronbach s alphas of the total NHP score were 0.82 and 0.87 respectively

582 for those patients with normal cognitive function, and moderate cognitive impairment. Also, Cronbach s alphas of NHP domains presented by the group of patients with moderate cognitive impairment were similar to those found amongst patients with normal cognitive function (no statistical significant differences). The Spearman correlation coefficient between the Physical Mobility score of the NHP and the Barthel Index was 0.39 (p<0.01) among patients with normal cognitive status. This correlation was very similar among those patients with moderate cognitive impairment (r=0.40, p<0.01). Discussion On the whole, the NHP could be administered by an interviewer to more than two-thirds (70.1%) of the study sample. Specifically, it was successfully administered to nearly all patients with normal cognitive function or moderate cognitive impairment (98% and 86.3%, respectively), but to hardly any patients with severe cognitive impairment (<6%). These results indicate good acceptability and feasibility of the interviewer-administered NHP amongst older adults with disabilities, and are consistent with previously reported high completion rates of NHP amongst older adults living in nursing homes (>95% [14]). The difficulty of administering HRQoL questionnaires to patients with severe cognitive impairment, defined as an MMSE score of <10 points, is also consistent with previous studies [12 13]. Thus, the evidence would support this MMSE score as the cut-off point for obtaining an HRQoL assessment directly from the patient. A fundamental point for discussion is the degree of reliability and validity of the NHP answered by those patients with moderate cognitive impairment. The reliability of the NHP was adequate for Emotional and Pain, but Energy, Sleep, Social Isolation, and Physical Mobility were below the standard of 0.7 [32]. The low level of reliability for Social Isolation and Energy had previously been reported in patients with dementia [14], as well as in some groups of a comprehensive evaluation of 5578 patients [20] (primary care, musculoskeletal and connective tissue diseases, chronic renal failure and cardiovascular disease). Nevertheless, the Cronbach s alphas of total scores (0.82 and 0.87) were close to 0.9 (the highest standard recommended for allowing individual comparisons [29, 32]), both in patients with normal cognitive function and those with moderate cognitive impairment. Overall, our results support the conclusion that older adults are reliable and consistent when answering the NHP [17, 18], but the use of the overall score, instead of the interpretation of isolated dimension scores, should be seriously considered. The correlation shown by Physical Mobility with the Barthel Index was statistically significant and very similar between the two cognitive status groups (0.39 and 0.40). However, the strength of the correlation, which was slightly weaker than hypothesised, merits comment. The conceptual differences between Functional Capacity, as measured by the Barthel Index, and the Physical Mobility dimension of NHP could explain this weak-moderate correlation: whilst almost all the items of this NHP dimension focus on mobility, the Barthel Index measures mainly the self-care activities of daily living, including only three items of mobility (sitting, walking, and going up and down stairs). In summary, our findings support the validity of the NHP in disabled older adults with moderate cognitive impairment, as well as the relevance of including HRQoL questionnaires in geriatric evaluations, as they add information to that covered by the traditional indices based on Activities of Daily Living. The number of missing items in the NHP was low, with no differences between those patients with normal cognitive function and those with moderate cognitive impairment. These results could be due to the fact that the items were simply stated (designed for a minimum reading age of nine years), and required only yes/no answers. Unsuitable content is a relatively common problem in studies where a generic HRQoL instrument is administered to older adults populations [1]. Although it might appear discouraging that around two-thirds of the patients in our study considered at least 1 item not applicable to their situation, these responses were in just 3 of the 8 items of the Pain domain, addressing physical activities not performed by a high proportion of dependent patients. It is important to note that the percentage of patients in the study with an item considered not applicable was insignificant in the other NHP

583 domains (<3%). And there were no repercussions on the NHP measurement of Pain, taking into account its good reliability (Cronbach s alpha=0.71). On the other hand, the low number of unsuitable items in the NHP (3 out of the 38) is particularly remarkable, compared with the nearly 10 items of the SF-36 which have been identified as poorly applicable to older adults [1, 3, 4]. The suitability of the NHP domains for covering the range of severity also merits comment. Since the patients studied were severely disabled, most could be clustered around the worst extreme of the scale. Table 4. Distribution characteristics, and reliability of the Nottingham Health Profile (NHP) by cognitive status Overall, N=92 Cognitive Status p-value* Normal (MMSE 21), N=48 Moderate impairment (MMSE 10 20) (N=44) Mean score (SD) Total 58.9 (34.4) 46.8 (19.5) 60.6 (22.4) 0.087 Energy 49.8 (28.7) 51.0 (35.4) 67.5 (31.4) 0.040* Emotional 41.7 (30.8) 41.6 (26.3) 58.8 (28.9) 0.005 Pain 53.3 (32.3) 40.8 (29.3) 42.9 (33.6) 0.852 Sleep 42.1 (28.8) 51.2 (32.5) 55.7 (32.4) 0.571* Social 78.0 (20.4) 37.6 (27.4) 46.7 (29.8) 0.140 Physical mobility 53.0 (21.6) 74.3 (21.0) 81.8 (19.4) 0.057* Floor effect, % a Total 0.0 0.0 0.0 n.a. Energy 28.3 20.8 36.4 0.098 Emotional 7.6 0.0 15.9 0.004 Pain 2.2 2.1 2.3 0.948 Sleep 15.2 14.6 15.9 0.862 Social 5.4 0.0 11.4 0.016 Physical mobility 23.9 14.6 34.1 0.029 Ceiling effect, % b Total 0.0 0.0 0.0 n.a. Energy 15.2 20.8 9.1 0.118 Emotional 2.2 4.2 0.0 0.169 Pain 2.2 2.1 2.3 0.948 Sleep 7.6 8.3 6.8 0.786 Social 17.4 22.9 11.4 0.146 Physical mobility 0.0 0.0 0.0 n.a. Scaling Succes, % c Total Energy 40.0 40.0 40.0 1.000 Emotional 86.7 82.2 80.0 0.787 Pain 100.0 90.0 100.0 0.040 Sleep 88.0 90.0 64.0 0.017 Social 88.0 76.0 80.0 0.733 Physical mobility 85.0 80.0 75.0 0.592 Cronbach s alpha Total 0.85 0.82 0.87 0.291 Energy 0.50 0.52 0.42 0.644 Emotional 0.73 0.67 0.74 0.485 Pain 0.71 0.67 0.75 0.421 Sleep 0.61 0.61 0.62 0.520 Social 0.48 0.42 0.52 0.607 Physical mobility 0.44 0.46 0.41 0.687 a Proportion of patients with the maximum score (100). b Proportion of patients with the minimum score (0). c Percentage of times an item scale correlation is higher for hypothesised domain than for the other domains of the questionnaire. * P-value was calculated by applying the t-unpaired test or Mann Whitney U-test (marked with an asterisk) when distribution was not normal. The Chi-Square test was used for categorical variables.

584 However, none of the patients obtained the worst or the best possible total NHP score, which suggests that the questionnaire satisfactorily covers the Quality of Life range presented by these patients. The limited size of the groups analysed is a relatively common factor in studies with disabled older adult inpatients [33], and the extent to which the lack of statistically significant differences in some of the hypotheses tested are due to insufficient statistical power should be considered. However, the direction of the differences in Cronbach s alpha between the two groups under comparison (higher reliability for the group with moderate cognitive impairment in almost all NHP domains), and their small size (ranging from 0.01 to 0.10) would suggest a lack of any real difference. Secondly, because of the difficulties previously reported in literature concerning the self-completion of HRQoL questionnaires in this population [1, 8, 15, 16], the NHP was interviewer-administered to all patients in our study, thus no comparison of reliability could be made with the self-completion method. Finally, because we recruited older adult inpatients with disabilities, the study sample was not representative of older adults patients from other settings. This factor may not be a limitation but rather an advantage, because NHP characteristics are expected to be better in healthier older adults populations. To summarise, if administered by a trained interviewer, the NHP could be a feasible, reliable and valid option for assessing HRQoL in disabled older adults patients with mild to moderate cognitive impairment (MMSE 10). These findings allow us to confirm the theoretical advantages of using the NHP with older adults patients, mainly because of its ease of completion, and its good coverage of a broad health-status range. Further research is needed to test the advantages of NHP compared to other generic HRQoL questionnaires, by including various instruments in a single study. However, head-to-head comparative studies are especially difficult in this population because of the high burden the completion of more than one questionnaire would represent for these patients. On the other hand, it is important to note the additional information provided by these measures, beyond the traditional scales of disability generally used for geriatric evaluation, and we would recommend the use of both, whenever feasible. Acknowledgments This study was supported by the Fondo de Investigaciones Sanitarias of Spain (FIS no 97/1103), Instituto de Salud Carlos III (network of excellence RCESP - C03/09), the Agencia d Avaluacio de Tecnologies I Recerca Me` diques (AATM 038/ 06/02), and DURSI Government of Catalonia (2001 SGR 00405).The authors would like to thank Maxine Hollewell for general and editorial assistance in the preparation of this article. References 1. Hayes V, Morris J, Wolfe C, et al. The SF-36 Health Survey Questionnaire: is it suitable for use with older adults. Age Ageing 1995; 24: 120 125. 2. Hill S, Harries U, 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(1): 94 98. 3. Brazier JE, Walters SJ, Nicholl JP, et al. Using the SF-36 and Euroqol on an elderly population. Qual Life Res 1996; 5(2): 195 204. 4. 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; 6(3): 213 216. 5. Stadnyk K, Calder J, Rockwood K. Testing the measurement properties of the Short Form-36 Health Survey in a frail elderly population. J Clin Epidemiol 1998; 51(10): 827 835. 6. Mallinson S. The Short-Form 36 and older people: some problems encountered when using postal administration. J Epidemiol Community Health 1998; 52(5): 324 328. 7. 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(1): 19 22. 8. Parker SG, Peet SM, Jagger C, et al. Measuring health status in older patients. The SF-36 in practice. Age Ageing 1998; 27(1): 13 18. 9. Carver DJ, Chapman CA, Thomas VS, et al. Validity and reliability of the Medical Outcomes Study Short Form-20 questionnaire as a measure of quality of life in elderly people living at home. Age Ageing 1999; 28(2): 169 174. 10. Seymour DG, Ball AE, Russell EM, et al. Problems in using health survey questionnaires in older patients with physical disabilities. The reliability and validity of the SF-36 and the effect of cognitive impairment. J Eval Clin Pract 2001; 7(4): 411 418. 11. Iglesias CP, Birks YF, Torgerson DJ. Improving the measurement of quality of life in older people: the York SF-12. Q J Med 2001; 94(12): 695 698. 12. Novella JL, Jochum C, Ankri J, et al. Measuring general health status in dementia: practical and methodological issues in using the SF-36. Aging 2001; 13(5): 362 369.

585 13. Novella J, Ankri J, Morrone I, et al. Evaluation of the quality of life in dementia with a generic quality of life questionnaire: the Duke Health Profile. Dement Geriatr Cogn Disord 2001; 12(2): 158 166. 14. Bureau-Chalot F, Novella JL, Jolly D, et al. Feasibility, acceptability and internal consistency reliability of the nottingham health profile in dementia patients. Gerontology 2002; 48(4): 220 225. 15. Ferrer M, Alonso J. The use of the Short Form (SF)-36 questionnaire for older adults. Age Ageing 1998; 27(6): 755 756. 16. Lyons RA, Perry HM, Littlepage BN. Evidence for the validity of the Short-form 36 Questionnaire (SF-36) in an elderly population. Age Ageing 1994; 23(3): 182 184. 17. Mozley CG, Huxley P, Sutcliffe C, et al. Not knowing where I am doesn t mean I don t know what I like : cognitive impairment and quality of life responses in elderly people. Int J Geriatr Psychiatry 1999; 14(9): 776 783. 18. Feinberg LF, Whitlatch CJ. Are persons with cognitive impairment able to state consistent choices? Gerontologist 2001; 41(3): 374 382. 19. Miralles R, Sabartes O, Ferrer M, et al. Development and Validation of an Instrument to Predict Probability of Home Discharge from a Geriatric Convalescence Unit in Spain. J Am Geriatr Soc 2003; 51(2): 252 257. 20. Mauskopf JA, Austin R, Dix LP, et al. Estimating the value of a generic quality-of-life measure. Med Care 1995; 33(4 Suppl): AS195 202. 21. Lamarca R, Alonso J, Santed R, et al. Performance of a perceived health measure in different groups of the population: a comprehensive study in Spain. J Clin Epidemiol 2001; 54(2): 127 135. 22. Inouye SK, Dyck CH van, Alessi CA, et al. Clarifying confusion: the confusion assessment method. A new method for detection of delirium. Ann Intern Med 1990; 113(12): 941 948. 23. Granger CV, Albrecht GL, Hamilton BB. Outcome of comprehensive medical rehabilitation: measurement by PULSES profile and the Barthel Index. Arch Phys Med Rehabil 1979; 60: 145 154. 24. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40(5): 373 383. 25. Folstein MF, Folstein SE, McHugh PR. Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12(3): 189 198. 26. Erkinjuntti T, Kurz A, Gauthier S, et al. Efficacy of galantamine in probable vascular dementia and Alzheimer s disease combined with cerebrovascular disease: a randomised trial. Lancet 2002; 359(9314): 1283 1290. 27. Prieto L, Alonso J, Viladrich MC, et al. Scaling the Spanish version of the Nottingham Health Profile: evidence of limited value of item weights. J Clin Epidemiol 1996; 49(1): 31 38. 28. Mckenna SP, Hunt SM, McEwen J. Weighting the seriousness of perceived health using Thurstone s method of paired comparisons. Int J Epidemiol 1981; 10: 93 97. 29. Lautenschlager GJ. ALPHATST: Testing for differences in values of coefficient alpha. Appl Psychol Measurement 1989; 13: 284. 30. Bonett DG. Sample size requirements for comparing two alpha coefficients. Appl Psychol Measurement 2003; 27(1): 72 74. 31. SPSS Inc.. SPSS X User s Guide., 2 ed.nd ed., Chicago: McGraw-Hill, 1986. 32. Scientific Advisory Committee of the Medical Outcomes Trust. Assessing health status and quality-of-life instruments: attributes and review criteria. Qual Life Res 2002; 11(3): 193 205. 33. Guyatt GH, Eagle DJ, Sackett B, et al. Measuring quality of life in the frail elderly. J Clin Epidemiol 1993; 46(12): 1433 1444. Address for correspondence: Dr. Montse Ferrer, Health Services Research Unit, Institut Municipal d Investigacio Mèdica, IMIM-IMAS, C/Doctor Aiguader, 80, Barcelona 8003, Spain E-mail: MFerrer@IMIM.ES