Changes in self-rated health, disability and contact with services in a very elderly cohort: a 6-year follow-up study

Transcription

1 Age and Ageing 98; 2: Changes in self-rated health, disability and contact with services in a very elderly cohort: a 6-year follow-up study TOM R. DENING, LJN-YANG CHI 1, CAROL BRAYNE 1, FEUCIA A. HUPPERT 2, EUGENE S. PAYKEL 2, DANIEL W O'CONNOR 3 Department of Psychiatry, Addenbnooke's NHS Trust, Box 311, Fulboum Hospital, Cambridge, CBI EF, UK Departments of 'Community Medicine and 2 Department of Psychiatry, University of Cambridge, Cambridge, UK 3 Department of Psychological Medicine, Monash University, Australia Address correspondence to I R. Dening. Fax (+44) Abstract Objectives: to study the relationships between global self-rated health, reported physical symptoms and depressive symptoms and the receipt of community services by very elderly people, and to examine changes in these variables over time. Design: three-wave study with follow-up at 2.4 and 6 years after first interview. Structured interview, incorporating cognitive examination (Mini-Mental State Examination) and enquiring specifically about overall self-rated health, physical symptoms and depressive symptoms. Setting: community setting in city of Cambridge, UK. Participants: 2609 were initially recruited: all patients aged years and over from lists of six general practices (and one in three from a seventh practice). At 2.4 years, 113 individuals re-examined and at 6 years 628 individuals. Measurements: general health self-rated in comparison to others of similar age and individual physical and depressive symptoms self-rated as present or absent. Symptoms were added to produce physical health and depressive symptom scores. Data presented from cross-sectional analysis of 6-year sample; also examined longitudinal data from all three waves of study for ageing and cohort effects. Finally the effect of health variables on the receipt of services was examined. Statistics used included \ 2 and non-parametric statistics for continuous data, also odds ratios for likelihood of receiving services. Results: at 6 years, 0% rated their overall health as good or very good. Overall self-rated health showed both ageing and cohort effects, improving with increasing age and especially with more recent cohort. Reported physical symptoms increased with ageing. Depression scores also increased with ageing but the relationship between depressive symptoms and ageing was less clear-cut. Receipt of services was associated with poor self-rated health and reported physical symptoms as well as with ageing. Higher depression scores at 2.4 years were associated with increased service receipt at 6 years, indicating a lag between the symptoms and the service response. Individuals in the more recent cohort were less likely to receive services, but those who did so received more frequent contact. Conclusions: although very elderly people have a high prevalence of reported physical symptoms, they often rate their overall health as good. There was a stronger relationship between ageing and physical symptoms than with depressive symptoms. Symptoms of both kinds influenced the likelihood of receiving services, although there was a lag between depressive symptoms and service response. Cohort effects on service receipt may reflect changes in public service policy. Keywords: community services, dementia, depression, elderly people, self-rated health Introduction As the population ages, by far the greatest increase is in the number of very elderly people, aged 80 years or older. Accurate information on the physical and mental health of very elderly people is essential to anticipate the future needs for health and social care. Epidemiological studies to date have concentrated on survival, physical health or dementia; but have seldom attempted to examine the relationships between 23

2 T. R. Dening et al. physical health, mental health and the receipt of services by very elderly people or how these variables change over time [1, 2]. This report presents data from a community study of a large cohort of elderly people in Cambridge, UK (the Cambridge city cohort), followed up over 6 years. The report is one of several from the third wave of interviews, and other publications will examine changes in cognition, activities of daily living and social support. Here, we examine subjectively-rated physical and mental health, both cross-sectionally and longitudinally and the receipt of services and the relationships between health and service receipt. Subjects and methods The sample was originally recruited in 8 and 86 and comprised people aged and over from the lists of seven general practices in the city of Cambridge. The initial study, the Hughes Hall Project for Later Life [3], included 2609 participants. The second study, the Cambridge Project for Later life [4], which took place 2.4 years later, re-examined 113 individuals who were not demented at the time of the initial study, with particular emphasis on the incidence of new cases of dementia and on depressive symptoms. The third-wave project, named the Cognition, Activities and Services Study, aimed to study all those individuals who were seen in the first study and still surviving. For the purposes of this paper we refer to the three phases as TO, T2.4 and T6, to indicate the interval (in years) between them. Ethical approval was obtained from the Cambridge research ethics committee. Participants in T6 were identified after checking names of survivors with general practitioners and full consent was obtained from participants and/or their families. Those who had moved were traced and approached for interview. All those who were contacted and agreed to be interviewed were seen, i.e. there were no exclusion criteria. A total of 628 people were interviewed at their normal place of residence during 92. A structured screening interview was administered by four trained lay interviewers. The interview was similar to that used for the T2.4 study, with additional items relating to physical health, social contacts and contact with medical and social services. The interview also included the items of the Mini-Mental State Examination (MMSE) [], enabling MMSE scores to be calculated. In all cases, interviewers were instructed to administer the cognitive examination. Proxies were interviewed for the non-cognitive data where cognitive performance of the participant was so impaired that information was likely to be missing or inaccurate. The key variables examined were: 1. Global self-rated health: participants were asked to rate their general health compared with others of the same age. 2. Reported physical symptoms: physical health was further examined by questions about a list of nine common symptoms (poor vision, poor hearing, arthritis, back pain, chest pain, shortness of breath, limb weakness, unsteadiness on feet, tendency to fall). These were rated as to whether in the last month they had caused interference with day-to-day activities, were present but not interfering or were not a problem at all. A physical health score (PHS; range 0-9) was derived by adding together the items and used as a variable for other parts of the data analysis. 3. Depressive symptoms: these were rated by means of depression symptom items derived from CAMDEX [6] as previously used in the T2.4 interview schedule. A depression symptom score (DSS) was derived by adding together the items [] (see Appendix), with a range of scores from 0 to Contact with services: service contacts in the week before interview were recorded, including home help, meals on wheels, community nursing, day centre and day hospital. Data from the T6 interviews were analysed crosssectionally and also combined with data from TO and T2.4 to examine changes over time. Where the ages of participants were used as a variable, the sample was split into three age groups: 80-84, 8-89 and 90+. Ageing effects (following the same individuals over time) were examined by analysing data at all three time points. For some variables, slight differences between the interview schedules used for TO, T2.4 and T6 meant that comparisons could only be made for two of the three waves of the study. Thus, individual physical and depressive symptoms could only be compared between T2.4 and T6, and numbers of service contacts could only be compared between TO andt6. Cohort effects (comparing individuals of similar age at different time points) were investigated by comparing different individuals, aged years old, at TO andt6. Missing questions were not assigned any further values, except for the cognitive examination where 0 was assumed if an item was not answered by the participant. Percentages are given as 'valid percentages', i.e. omitting missing values. Results Of the 628 individuals in the T6 sample, 611 had been interviewed at both TO and T2.4, but 1 were omitted from T2.4, the commonest reason being that they had dementia at TO. Of the T2.4 sample (n = 113), 62 24

3 Self-rated health and contact with services were not studied again at T6. This figure included 389 people who had died and 13 who declined to participate again or who were too ill, together with others who had moved away or were lost to follow-up. Among those who did not participate again, 29% had MMSE scores at T2.4 of, so were probably too severely demented for re-interview. The refusal rate at T6 was 8 out of 8 people approached (14.3%), compared with.6% at TO [3] and 12.9%atT2.4 [4]. T6 interview: cross-sectional data Demographic data are summarized in Table 1. Global self-rated health Overall, comparing themselves with others of the same age, 181 individuals (32%) reported very good health, 28 (46%) reported good health, (%) rated their health as fair and (4%) as poor or very poor. One hundred and fourteen women (30%) rated their health as very good, compared with 6 men (34%), but this difference was not significant. Older people were significantly more likely to report very good health relative to their contemporaries (for very good health versus other categories, x 2 =.9, 2 clf,, P = 0.0). hi the age group 80-84, 8 people (2%) reported very good health compared with 22 (39%) of those aged 90+; none of the latter group rated their health as poor or very poor. For 61 people (9% of the total T6 sample), this item was missing. The two commonest causes of missing data were that subjects did not know anyone of the same age to compare themselves with (n = 39) or were too demented to respond (n = ). Reported physical symptoms As mentioned above, physical symptoms were scored in three categories (interfering with daily activities, present but not interfering or not a problem). For the purposes of analysis, we combined the latter two categories thus creating a dichotomy (active problem or not). Individual health problems are summarized in Table 2 and analysed by age group and sex. Locomotor problems,, including arthritis, limb weakness and unsteadiness, caused the most disruption to everyday activities. Some symptoms were significantly more likely to be disabling in older individuals, including poor vision, poor hearing, unsteadiness and tendency to fall. Arthritis, back pain and limb weakness were significantly more common in women. Chest pain was relatively more common in men, but this difference was not statistically significant. Ten percent of the sample had no physical symptoms at all. When ratings for individual symptoms were added to create a PHS, the mean PHS was 3-2 (median = 3, Table I. Demographic data at T6 (6-year follow-up; 92), n = 628 Category Age (years) >90 Mean Sex Male Female Social class" Non-manual Manual Education (age at leaving school) b <14 S1 Marital status b Married Widowed Separated/divorced Single Living arrangements 0 Institution Alone With spouse With other(s) Accommodation 11 House/flat Warden-controlled flat Council residential home Private residential home Long-stay hospital bed Other *Data missing for 13 subjects (2%). b Data missing for one subject c Data missing for nine subjects (1%). "Data missing for two subjects. No. of subjects (%) 31 (0%) 23 (38%) 6 (12%) 8. 9 (32%) 429 (68%) 29 (41%) 3 (%) 399 (64%) 229 (36%) 1 (2%) 390 (62%) 1 (2%) 6 (%) (9%) 3 (%) 126 (%) 83 (13%) 492 (9%) 2(11%) (3%) 28 (%) (1%) 8(1%) interquartile range = 2-). PHS increased significantly with increasing age (Kruskal-Wallis H = 9.8, P < 0.01) and was significantly higher in women (Mann- Whitney U = , P < 0.001). PHS also tended to be higher in those who had had manual occupations (Mann-Whitney U = , Z = -1.86, P = 0.06), but was not significantly related to educational level. When asked about all the physical health problems listed and their effect on everyday life, 2% said that their symptoms had no effect on life at all, 26% that life was slightly affected, 18% that life was moderately affected and % that it was very much affected. For the % with no reported symptoms, this question was not relevant. 2

4 T. R. Dening et al. Table 2. Self-reported physical symptoms at T6 a % showing each symptom Age (years) Sex in = 31) 8-89 in = 23) >90 in = 6) P Male in = 9) Female in = 429) P Poor vision Poor hearing Arthritis Back pain Chest pain Short of breath limb weakness Unsteadiness Tend to fall "6-year follow-up; 92. Depressive symptoms The interview schedule contained items relating to depression. These were rated (usually 0 or 1, except mood which was rated 0, 1 or 2) and then added together to produce a depression symptom score ranging from 0 to 11 (see Appendix). A DSS could be calculated for participants (18 men and 388 women). The mean DSS for the whole sample was 33 (median = 3.0, interquartile range = 1.0-0). There was no relationship between DSS and age, social class or education, but DSS was significantly higher in women than in men (mean value for women = 3.4, mean for men = 2.9; Mann-Whitney U= , Z= -2.80, P= 0.00). The frequencies of individual depressive symptoms by age and sex are shown in Table 3- In contrast to the findings with physical symptoms, there were few significant differences in depressive symptoms with age group and sex. Women were more likely to report tension or decreased energy, and the 90+ age group were more likely to feel life was not worth living. Service contacts Information was obtained from 89 people. Of these, 1 (18.%) were receiving home help, usually once or twice per week in = 66,11.%), with 36 (6%) receiving home help on days or more. Older people were significantly more likely to have home help (x 2 = -3, Table 3. Self-reported depression items at T6 a % showing each symptom Age (years) Sex in = 31) 8-89 in = 23) >90 in = 6) P Male in = 9) Female in = 429) P Sad/depressed Occasionally Most of time More tense Irritable Decisions Less energy Lost interest Alone Bleak future Not worth living Poor sleep "6-year follow-up;

5 Self-rated health and contact with services 2 df, P < 0.001), with 3% of those over 90 receiving such help. Sixty-five people (11%) received meals on wheels, usually at leastfivetimes per week (n = 46, 8%). Again, this service was significantly more often received by older people (x 2 = 16.0, 2d.f.,P< 0.001). Fifty-three people (9%) had been visited by a community nurse within the last week, 40 receiving one visit, receiving two visits and three receiving three visits. Once again, there were significantly more visits made to those over age 90 (x 2 = 8.8, 2 df, P < 0.02). Twenty-five people (4%) had been to a day centre within the last week, the modal number of attendances being two (n = ). Five people had attended a day hospital. No significant association between age and day care attendance was detected, possibly due to small numbers. No significant association was found between sex and the receipt of any of the above services, even when age was controlled for. Changes over time: longitudinal data Where possible, results from T6 interviews were also examined in relation to data from TO and T2.4. Longitudinal data were analysed in two ways. First, by examining those who had been interviewed on all three occasions (n = 611), it was possible to investigate the effects of ageing in the survivors of a followed-up cohort. Second, we examined for cohort effects at two different time points by comparing individuals between the ages of 81 and 86 years at TO (n = 8) and T6 (n = 388), i.e. different individuals of the same age at different times. The first of these analyses thus examined changes over time in individuals; the second looked at differences between groups. Effect of ageing in the followed-up cohort Global self-rated bealtb Results from the 6l 1respondentsseen at all three time points are shown in the upper half of Table 4. At T6, 430 (8%) rated their health as 'good' or 'very good' relative to others of the same age, compared with 39 (6%) of the TO sample. This difference was statistically significant, as was the difference between TO and T2.4. Most of the difference appeared to be accounted for by participants changing their ratings from 'fair* to 'good' or 'very good' over time [in TO l6l (29%) were 'fair', compared with (%) in T6]. Self-ratings of 'poor' or 'very poor' health barely changed across the three interviews. Self-ratings of health were not significantly different between T2.4 and T6. Thus, overall, the two later interviews recorded significantly better ratings of general health compared with the peer group than at the initial contact. Reported physical symptoms hi terms of causing problems, all nine symptoms studied became significantly more common from T2.4 to T6 (see Table ). The largest proportional increases were in poor vision [from 41 (%) to 113 (%) cases] and tendency to fall [from 1 (3%) to (%) cases]. Unsteadiness, limb weakness, shortness of breath and poor hearing were also more than doubled. Depressive symptoms The mean DSS at T6 was 33, significantly higher than that at T2.4, which was 2.8 (Mann-Whitney U = 11 1, P < 0.001). However, examination of the individual symptoms revealed a more complex picture Table 4. Self-ratings of overall health in the followed-up cohort (seen at all three time points*) and in those of the same age (81-86 years at the time of assessment) Self-rating Valid % of respondents TO Followed-up cohort (n = 611) Very good 2 Good 40 Fair 29 Poor/very poor 4 Those of same age (n = 8 at TO; n = 388 at T6) Very good Good 3 Fair 34 Poor/very poor T T b. n Q C - rt r\r\a d 0.01 D ;0.8 c ; < 1 "TO, original study, T2.4, 2.4-year follow-up; T6, 6-year follow-up. "TO vs T2.4; C T2.4 vs T6; To vs T6. 2

6 T. R. Dening et of. Table. Self-reported physical symptoms from T2.4 and T6" for the followed-up cohort (n = 611), showing ageing effect Symptom Poor vision Poor hearing Arthritis Back pain Chest pain Short of breath limb weakness Unsteadiness Tend to fall Other Valid % of respondents reporting T2.4 T * T2.4, 2.4-year follow-up; T6, 6-year follow-up. (Table 6). Four symptoms (tension, loss of interest, pessimism about the future and poor sleep) were significantly more common at T6. Persistent sad mood, irritability, difficulty with decisions and being alone more increased, but these changes did not reach statistical significance. Feeling that life was not worth living did not change; and one symptom (decreased energy) was significantly less common at T6. There thus appeared to be an ageing effect for at least some of these symptoms, but as persistent sadness of mood did not increase significantly it may be asked whether the increase in symptoms was really due to depression or simply to the process of ageing. Service contacts The number of individuals using the various social and nursing services increased from TO to T2.4 and again to T6 for all the services studied (home help, meals on wheels, community nursing, day centre attendance and day hospital attendance; Table ). For the first four of these the increase from TO to T6 was significant at least at the level of P < 0.01, but for day hospital attendance the numbers involved were too small to reach statistical significance. Cohort effects in individuals of similar age Globed self-rated health Self-ratings of overall health are summarized in the lower half of Table 4. The differences between TO and T6 were highly significant, with the proportion rating Table 6. Self-reported depressive items from T2.4 and T6* for the followed-up cohort (n 611), showing ageing effect Symptom Sad/depressed Occasionally Most of the time More tense Irritable Decisions Less energy Lost interest Alone Bleak future Not worth living Poor sleep Valid % of respondents reporting T2.4 T " T2.4, 2.4-year follow-up; T6, 6-year follow-up. 28

7 Self-rated health and contact with services Table. Service contacts within the last week from TO, T2.4 and T6" for the fbllowed-up cohort (n = 611), showing ageing effect Valid % of respondents TO T2.4 T6 P (TO vs T6) Home help Meals on wheels Community nurse Day centre Day hospital 3 4 C 0.6 b 0 b D 0.8 b *T0, original study; T2.4, 2.4-year follow-up; T6, 6-year follow-up. ""Appreciable amounts of missing data (-8%). c Withln the last month. their health as good or very good rising from 6% to %. Unfortunately, differences between the interview schedules and symptom ratings at TO and T6 meant that cohort effects for physical symptoms and depressive symptoms could not be examined. Service contacts Comparisons between services provided at TO and T6 are summarized in Table 8. Home help, meals on wheels and community nurse visits were less often received at T6 than at TO. The proportion of people receiving community nurse visits at T6 was less than half that at TO. There was little change in day centre or day hospital attendance. The numbers of service contacts per week were only recorded for TO and T6. There was some evidence that those people who did receive services were receiving them more frequently at T6 than at TO. For home help, the proportion receiving assistance five or more times per week rose from 16/149 (11%) at TO to 14/ (26%) at T6 (x 2 = 6.9, 1 d.f, P = 0.008). At TO, no participant attended a day centre more frequently than once per week, but at T6, out of the 14 individuals (1%) who attended day centres did so at least twice per week (x 2 = 23.3, 1 d.f, P < 1). For meals on wheels there was a small, non-significant increase in the frequency of meals received per week, but even at TO, 62% of recipients were having meals on wheels at least five times per week, so there may have been a ceiling effect. Community nurse contacts were difficult to compare, as they were counted over the last month for TO and the last week for T6: the proportion receiving two or more visits per week seemed to increase at T6 but it was not possible to be certain of this. In summary, therefore, people over the age of 81 were less likely to be receiving services at T6 (in 92) than at TO (in 8-6), but those who were receiving social services at T6 tended to have more frequent contacts. Relationships between health and service contacts Poor self-rated health was associated with increased service usage. People who rated their general health as poor or very poor compared with others of the same Table 8. Service contacts within the last week from TO and T6 a, for individuals of the same age (>81 years) Valid % of respondents TO («= 8) T6 (n = 388) P Home help Meals on wheels Community nurse Day centre Day hospital 2 12 I4 b C 0.4 c < *T0, original study; T6, 6-year follow-up. ''Within the last month. c Up to 13% of data missing; in all other cases -8% were missing. 29

8 T. R. Dening et al. age were more likely to receive home help [odds ratio (OR) 2.8, 9% confidence interval (CI) = , P < 1], meals on wheels (OR = 2., CI = , P = 0.001), community nurse visits (OR = 2.8, CI = 1.-.1, P = 0.001) and day centre attendance (OR = 2.9, CI = , P = 0.011). For 606 people at T6 for whom a PHS could be calculated, those receiving home help, meals on wheels, community nurse visits, day centre and day hospital care had significantly higher PHS than those who did not receive services. This was also the case at TO and T2.4, but with day hospital attendance reaching significance at somewhat lower levels of probability owing to small numbers. ORs for receiving services given the presence of individual reported physical symptoms are shown in Table 9. For the T6 sample, those with chest pain, shortness of breath, limb weakness and tendency to fall were significantly more likely to be attending a day hospital. Those with poor vision, arthritis, limb weakness, unsteadiness, tendency to fall and other symptoms were more likely to receive home help and, apart from poor vision, these symptoms also increased the likelihood of community nurse visits. Poor hearing and back pain were not associated with significantly increased odds of receiving any of the services. Table 9 also indicates which symptoms were significantly associated with increased likelihood of receiving services in the T2.4 study. In general, the pattern appeared similar to that for the 16 sample, but (presumably due to the larger numbers at T2.4) more comparisons reached statistical significance. There were no significant differences in DSS at T6 between those individuals who received services and those who did not. People attending day hospital did have higher DSS, but the numbers were too small for this to reach statistical significance. Interestingly, DSS at T2.4 was associated with service receipt at T6: People receiving home help and those attending day centres at T2.4 had higher DSS than those who did not (P = and P = respectively). This suggests that there is a lag between the existence of depression, its recognition and service response. ORs for receiving services in the presence of individual depressive symptoms are shown in Table. Only two symptoms were associated with significantly increased receipt of home help and only one with increased provision of meals on wheels. The ORs for day hospital attendance were high for several symptoms (including low mood, loss of interest and pessimism about the future), but these results did not reach significance because of the small numbers of day hospital attendees. By contrast, at T2.4, more symptoms were significantly associated with increased odds of receiving services, e.g. depressed mood with receipt of all types of services. Discussion Changes in population samples may be studied by means of cross-sectional, longitudinal or cohort approaches. Most studies of elderly people have been cross-sectional, although longitudinal studies are also appearing [2, 8]. The sample studied in this paper is notable because of its size, the advanced age of the participants and the length of follow-up, -which enables examination of longitudinal ageing and cohort effects, as well as cross-sectional data. The initial TO sample was a representative one, consisting of all those aged over years from the lists of six general practices and a random selection from a seventh, covering a wide area of Cambridge city. The T6 study attempted to interview as many surviving people from this sample as Table 9. Relationship of service usage to physical symptoms odds ratios of receiving services given the presence of individual symptoms in the T6 sample" (n = 628) Odds ratio (and i^) of receiving service Symptom Home help Meals on wheels Poor vision Poor hearing Arthritis Back pain Chest pain Shortness of breath limb weakness Unsteadiness Tendency to fall Other 2.0 (0.008) 1.(0.1) 2.1 (0.002) c 1.8(0.11) c 2.0(0.12) 1.8(0.094) c 3. () c 4.1 () c 2.8 (0.001) c 2. (0.001) c 0.9 (0.94) c 2.4 (0.02) 1.(0.09) c 1. (0.4) 0. (1.0) 1.1 (O.9) c 1.9 (O.03) c 3.0 () c 2.6 (0.014) 1.8(0.1) Community nurse 0.9 (0.98) c 1.1 (0.80) 1.8 (0.0) c 0.6 (0.46) c 3.4 (0.0) 1.4(0.4) c 2. (0.00) c 4.0 () c 3.4 (0.002) c 3.4 () c Day centre 1.9 (0.18) c 2.9 (O.O3) c 3.2 (0.0l4) c 2. (O.O6) c 2.0 (0.29) 1.3 (0.2) 3.9 (0.002) c 2.2 (0.) c 3. (0.0) 1.1 (0.) Day hospital 1.1 (1.0) 2.6 (0.38) 6.4 (0.02) 2.4 (0.39) 36.1 (0.001) 14.2 (0.008) 1.8 (0.00) 6.6 (0.049) c 6.6 (O.O3) c *6-year follow-up study. ''Using X 2 or Fisher's exact test; bold type indicates P < c Significanl odds ratios in T2.4 (2.4-year follow-up). 30

9 Self-rated health and contact with services Table. Relationship of service usage to depressive symptoms odds ratios of receiving services given the presence of individual symptoms in T6 a sample (n = 628) Symptom Low mood Occasionally Most of the time More tense Irritable Decisions Less energy Loss of interest Alone Bleak future Not worth living Poor sleep Odds ratio (and P 6 ) of receiving service Home help Meals on wheels Community nurse Day centre Day hospital 1.3 (0.32) c 2.8 (0.02) c 0.8 (O.3) c 0.8 (0.1) c 0.9 (0.91) 1.2 (0.42) c 2.1 (0.001) 1.1 (0.8) 2.1 (0.003) 1.4(0.) c 1.3 (0.30) 1. (0.16) 0.6 (0.) c O.6(O.14) C 0.8 (0.4) 1.1 (0.8) 1.(0.) 1. (0.0) 1. (0.22) 1.8 (0.064) 2.3 (0.00) c 0.9 (0.80) 1.0 (1.0) 0.3 (0.34) c 0.8 (0.48) 0.6(0.1) 0. (0.32) 1.1 (0.9) 1.0(1.0) c 0.9 (0.9) 1. (0.30) 1.3(O.1) c 1.1 (0.90) 1.9 (0.22) 1.0(1.0) c 1.0(1.0) 1.2 (0.94) 2.0 (0.1) 1.6 (0.43) 1.4(0.6) 1.1 (1.0) 2.4 (0.068) 2.2(0.11) 1.6 (0.40).1 (0.1) 11.0( (0.) 0. (1.0) c 0. (1.0) 8. (0.040) c.9 (0.061) 6.0 (0.09) 3.(0.1) *6-year follow-up study. 'Using X 2 or Fisher's exact test; bold type indicates P < "Significant odds ratios in T2.4. possible, hi general there is relatively little migration among such elderly people, so the sample available for re-interview should have remained fairly representative of the over-81-year-old population. The longer the initial sample is followed up, the more deaths occur so that those available for re-interview are more obviously a population of survivors. This will tend for all measures to introduce a bias in the direction of good health, particularly if there is evidence that non-responders have a higher level of symptoms, hi particular, those who did not participate in the later interviews at T2.4 and T6 probably included many individuals with dementia and/or physical illness, so that information about the service usage of the more disabled will tend to have been lost. On the other hand, the study was mainly interested in the usage of community services and many people with severe disability would have moved into institutional care. Interpretation of the results must take these considerations into account. There are some further limitations arising from the interview schedule and its administration. For various reasons, the first two waves (TO and T2.4) did not use identical schedules and it was also necessary to make certain amendments for the T6 study, given the particular interests of the study. Every attempt was made to be as consistent as possible with the wording of each item, but nonetheless there were differences which in some instances precluded certain comparisons, e.g. a physical health score could not be calculated from the TO items. Another problem was that limited resources prevented the use of more objective ratings (except for cognitive testing). Thus, the physical and mental health symptoms reported above are subjectively rated, hi theory, such self-rated symptoms might contaminate each other, with low mood being particularly likely to increase all other reported symptoms. However, this does not seem to have been the case: although depression scores did increase with age in the followed-up group, overall ratings of general health improved -when examined both for ageing and cohort effects. The validity of self-reported information on global health and symptoms has been recently discussed [9, ]. However, whether or not such data accurately reflect the presence or absence of disease, selfreported health has been shown to be an important predictor of mortality [11, 12]. Around 0% of the T6 sample rated their overall health as good or very good compared with others of the same age, a figure comparable to that found in the 91 UK General Household Survey [13]. This may suggest a bias towards optimism or stoicism in these surviving elderly people and is perhaps a tribute to their resilience. There was a strong cohort effect on overall self-rated health, with the later cohort reporting better health. hi addition, the T6 sample had quite high levels of reported physical symptoms (Table 2), although even among those aged 90+ no individual symptom was reported by even half the T6 sample. Again, these data are comparable to other studies, such as the UK Health and Lifestyle Survey [14]. For individual physical symptoms, there was a clear ageing effect, with all symptoms increasing in frequency, although when examined cross-sectionally at T6, only certain symptoms were significantly commoner in the older age groups. We were unable to examine for cohort effects upon individual physical symptoms. 31

10 T. R. Dening et al. It was somewhat surprising to see a significant increase in the DSS with ageing. The reasons for this are unclear, particularly as the cardinal symptom of low mood did not show a significant increase. It may be that the increase in DSS reflects not increasing depression, but some other change, perhaps physical ill-health or simply the process of ageing. The items which contributed to the increase in DSS were tension, decreased energy, pessimism about the future and poor sleep. These symptoms might also be worsened by the presence of anxiety rather than depression as such. Anxiety disorders in elderly people have often been neglected [1] and this possibility would be worthy of further investigation. Social services such as home help or meals on wheels were fairly commonly received by these elderly people, with day centre and day hospital attendance being much less common. Older people received more services, both when compared cross-sectionally and longitudinally, but there were significant cohort effects, with a smaller proportion of people at T6 receiving relatively more frequent service contacts. There are several possible explanations for this. First, perhaps a higher proportion of the T6 sample was in institutional care so not requiring community services, hi fact 6% of the TO sample were in institutional care compared with 9% at T6, which could account for the observed pattern of change. Secondly, the T6 sample may reflect increasing functional independence over time, a possibility supported by our data on activities of daily living, to be reported elsewhere. Thirdly, there may have been decreased availability of services either due to increased demand or public spending cuts. A fourth possibility is that services may have become increasingly targeted at those with most dependency, an explanation consistent with the aims of statutory services. New community care legislation was introduced in the UK in 93 and so would not have directly affected the period being studied, though new arrangements in social care were being considered at this time. Having certain reported physical symptoms, particularly limb weakness, unsteadiness and tendency to fall, was associated with a higher chance of service usage. However, when comparing T2.4 and T6 data, more individual symptoms increased the odds of receiving services at T2.4 (see Table 9)- This may be due to the smaller size of the T6 sample. Alternatively, as the sample becomes older and physical symptoms become more prevalent, they may lose some of their predictive power in determining who receives certain services. It is perhaps encouraging that more potentially life-threatening symptoms like chest pain and shortness of breath were associated with day hospital usage. At T6, few depressive symptoms were significantly associated with increased service receipt, whereas at T2.4 more significant associations were found, most notably with depressed mood and the feeling that life was not worth living. Again, this difference may partly have been due to smaller numbers at T6, but why depressive symptoms had less powerful effects on service receipt in the later sample is not clear. As we have not measured the severity of physical and depressive symptoms, it is not possible to ascertain whether physical symptoms were more likely to influence service receipt than were mental symptoms, although this has been suggested [16]. hi conclusion, this large community-based study of elderly people has examined their levels of subjectively-rated physical and mental health symptoms and the levels of formal services received by them. Crosssectionally, 0% rated their health as good or very good compared with others, with this proportion increasing in those aged 90 or more. Physical symptoms were commoner in the older participants, but % did not report any such symptoms. Depression scores were higher in women. Older people received more formal services. Longitudinal data showed significant increases in all physical symptoms with ageing, though overall ratings of subjective health improved with ageing. There was an ageing effect upon DSS, though not all depressive symptoms did become more frequent with increased age. Ageing was associated with increased receipt of services. There was a marked cohort effect in the direction of improved self-rated health in the later cohort. Individuals in the more recent cohort were less likely to receive services, but those who did received services more frequently. Poor overall health and poor physical health were associated with increased receipt of services. The relationship between depressive symptoms and services was less strong, but service receipt at T6 was more strongly predicted by DSS at T2.4 than at T6. We conclude that, although very elderly people have a high prevalence of physical symptoms, they often rate their overall health as good. There is a stronger relationship between ageing and physical symptoms than with depressive symptoms. Symptoms of both kinds influence the likelihood of receiving services, but there seems to be a lag between depressive symptoms and service response. Cohort effects in service receipt may reflect changes in public service policy. Acknowledgements The original Hughes Hall study was funded by the Charles Wolfson Trust. The Cambridge Project for Later life was funded by the Medical Research Council. The Cognition, Activities and Services study was funded by the East Anglian Regional Health Authority. Thanks also to our interviewers, to Eileen Richardson for administrative help, to the local general practitioners for their support and to the elderly residents of Cambridge who agreed to participate. 32

11 Self-rated health and contact with services Key messages This paper reports data from a community study of people aged and over, followed up over a period of 6 years, examining in particular self-rated health, reported physical and depressive symptoms and the receipt of community services. Initially 2609 people were recruited, 113 were reexamined at 2.4 years and 628 after 6 years, allowing analysis of both ageing and cohort effects. Although reported physical symptoms increased with increasing age, global self-reported health improved with age. Poor self-rated health and increased physical symptoms were associated with increased receipt of community services, but there was a time lag between higher depression scores and increased services. Cohort effects were observed with service receipt which may reflect changes in public service policy. References 1. Cullen M, Blizzard R, Livingston G, Mann A. The Gospel Oak project 8-90: provision and use of community services. Health Trends 93; 2: Jagger C, Spiers NA, Clarke M. Factors associated with decline in function, instltutionalization and mortality of elderly people. Age Ageing 93; 22: O'Connor DW, Pollitt PA, Hyde JB et al The prevalence of dementia as measured by the Cambridge Mental Disorders of the Elderly Examination. Acta Psychiatrica Scand 89; 9: Paykel ES, Brayne C, Huppert FA et al. Incidence of dementia in a population older than years in the United Kingdom. Arch Gen Psychiatry 94; 1: Folstein MF, Folstein SE, McHugh PR. 'Mini-Mental State': a practical method for grading the cognitive state of patients for the clinician. J Psychiatric Res ; 12: Roth M, Huppert FA, Tym E, Mountjoy CQ. CAMDEX. Cambridge, UK: Cambridge University Press, 88.. Girling DM, Huppert FA, Brayne C, Paykel ES, Gill C, Matthewson D. Depressive symptoms in the very elderly: their prevalence and significance. Int J Geriatr Psychiatry 9; : McCallum J, Kendig H, Freeman E et al Ageing and Families Seven Years After Data from a Seven Year Follow-up of Older Australians. Canberra: National Centre for Epidemiology and Population Health, Bridgwood A, Savage D. General Household Survey 91. London: HMSO, 92.. Swain VJ. Changes in self-reported health. In: Cox BD, Huppert FA, Whichelow MJ, eds. The Health and Lifestyle Survey: Seven Years On. Aldershot: Dartmouth, 93; Cox BD, Huppert FA, whichelow MJ eds. The Health and Lifestyle Survey: Seven Years On. Aldershot: Dartmouth, Engedal K. Mortality in the elderly: a 3-year follow-up of an elderly community sample. Int J Geriatr Psychiatry 96; 11: Office of Population Censuses and Surveys. General Household Survey 91. Supplement A: People Aged 6 and Over. London: HMSO, Cox BD, Blaxter M, Buckle A et al eds. The Health and Lifestyle Survey. London: Health Promotion Research Trust, Iindesay J ed. Neurotic Disorders in the Elderly. Oxford: Oxford University Press, Ely M, Brayne C, Huppert FA, O'Connor DW, Pollitt PA. Cognitive impairment: a challenge for community care a comparison of the domiciliary service receipt of cognitively impaired and equally dependent physically impaired elderly women in 86. Age Ageing 9; 26: Received 12 February 9 Appendix. Depressive symptom questions and scoring 1. Do you feel more tense and worry more than usual about little things? 2. Do you feel sad or depressed or miserable? (no = 0; occasionally - 1; most of the time = 2) 3. Have you felt more irritable lately (e.g. intolerant of noise)? 4. Would you say that you have more or less energy at the moment than you did two years ago? (more or the same = 0; less = 1). Have you had more trouble sleeping recently than is normal for you? 6. Do you find it more difficult to make decisions than you used to?. Have you lost interest in doing things you usually cared about or enjoyed? 8. Have you preferred to be more on your own recently? 9. How do you feel about the future how do you think things will work out for you? (pessimistic response = 1). Do you sometimes feel that life is not worth living? Except where otherwise noted each question scores 1 for each 'yes' response. 33

12 Lillian Houseman, aged 99, born 189: "I started work in the Mills as a half-timer. I went to work at 6 am until 12 pm, then in the afternoon I went to school." Ian Beesley.