Documente Academic
Documente Profesional
Documente Cultură
1093/ageing/afn193
The Author 2008. Published by Oxford University Press on behalf of the British Geriatrics Society. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org
Does depression predict adverse outcomes for older medical inpatients? A prospective cohort study of individuals screened for a trial
SARAH CULLUM1 , CHRIS METCALFE2 , CHRIS TODD3 , C AROL BRAYNE4
Division of Psychiatry, Cotham House, Cotham Hill, Bristol BS6 6JL, UK Department of Social Medicine, University of Bristol, Canynge Hall, Whiteladies Road, Bristol BS8 2PR, UK 3 School of Nursing, Midwifery and Social Work, University Place, University of Manchester, Oxford Road, Manchester M13 9PL, UK 4 Department of Public Health & Primary Care, Institute of Public Health, University Forvie Site, Robinson Way, Cambridge, CB2 2SR, UK
2 1
Address correspondence to: Sarah Cullum, Division of Psychiatry, Cotham House, Cotham Hill, Bristol BS6 6JL, UK. Email: sarah.cullum@awp.nhs.uk This work was performed at West Suffolk Hospital and Department of Public Health and Primary Care, University of Cambridge.
Abstract
Objective: to examine the relationship between depressive symptoms and hospital outcomes in an unselected consecutive sample of older medical inpatients. Design: a prospective cohort study of individuals screened for a trial. Setting: medical wards of UK district general hospital in rural East Anglia. Participants: six hundred and seventeen medical inpatients aged 65+ were randomly selected from consecutive admissions. Baseline measures: 15-item Geriatric Depression Scale (GDS-15), the Abbreviated Mental Test Score (AMTS) and the Cumulative Illness Rating ScaleGeriatric (CIRS-G). Main outcome measures: length of hospital stay; discharge to a community hospital (for rehabilitation), institutional care or usual place of residence; dying in hospital. Results: depressive symptoms are independently associated with an increased likelihood of inpatient death and transfer to a community hospital for rehabilitation, but are not associated with longer length of stay. Conclusions: research evaluating effectiveness of identication and treatment of depression in older medical inpatients should consider including inpatient death and use of rehabilitation services as potential outcomes.
Keywords: depression, aged, inpatients, cohort study, elderly
Introduction
Studies suggest that a substantial proportion of older hospital inpatients have depressive symptoms that may hinder their recovery. Adverse outcomes include delayed discharge [1, 2], greater risk of readmission [3], increased need for rehabilitation or institutional care [4] and higher mortality [49]. In order to attempt to reduce the impact of depression on patients health outcomes and healthcare costs, recent UK health policy encourages screening and treatment programmes for older people in mainstream healthcare settings [10, 11]. However, the research evidence cited above derives mainly from countries other than the UK and may not be generalisable to older people in UK hospitals. The aim of
this study is to examine the relationship between depressive symptoms and hospital outcomes in an unselected consecutive sample of older medical inpatients in a UK hospital, thus informing the selection of measurable outcomes that might be impacted by a policy of screening and treatment of depression in this population.
Methods
Participants
Acute admissions to the medical wards of a district general hospital aged 65 and over who had been in hospital between 3 and 7 days were eligible for the study. As
690
Informed consent was obtained from all participants. Participants who answered afrmatively to the screening question regarding suicidal ideas were immediately identied to ward staff with a recommendation for further assessment by a mental health professional. This study was approved by the West Suffolk Hospital LREC.
Results
From 2112 eligible admissions over a period of 15 months, 1009 older medical inpatients were randomly selected into the study. Three hundred and eighteen did not meet the inclusion criteria for the screen, 42 refused consent and 31 were excluded after the screen (including one participant who expressed suicidal ideas), as reported in [20]. Data from one participant who had not been discharged from hospital at the end of the study were not included; thus data from 617 participants were used in the analyses. Of 617 participants, 181 (29%) reported a history of depression and 76 (12%) were already known to psychiatric Table 1. Length of stay in days by covariates
................................................................ All sample 617 19.48 12.0 5.0060.00 Age groups 6569 57 12.05 9.0 4.029.0 7074 104 15.04 9.0 4.046.0 7579 119 21.99 11.0 5.069.0 8084 150 21.65 15.0 5.064.0 8589 122 20.84 13.0 5.060.0 90+ 64 21.05 13.0 5.061.0 P for trend 0.001 Sex Female 363 20.46 12.0 5.064.0 Male 254 18.08 12.0 4.053.0 P for trend 0.20 CIRS-G 27 122 19.40 11.0 5.069.0 89 108 18.33 11.5 4.053.0 1011 132 20.27 12.0 5.059.0 1213 134 18.21 13.0 5.061.0 1421 121 21.12 14.0 4.047.0 P for trend 0.68 AMTS 6/7 94 25.26 15.5 5.080.0 8 104 18.82 13.0 5.046.0 9 184 20.72 5.060.0 12.5 10 235 16.49 10.0 4.047.0 P for trend 0.006 GDS-15 04 347 18.17 11.0 4.053.0 57 133 21.13 13.0 5.069.0 810 89 17.43 12.0 5.043.0 1115 48 28.19 16.5 5.095.0 P for trend 0.13
P-values adjusted for exposure to the trial intervention. n Mean Median 90% range
Materials
Participants were assessed using the 15-item Geriatric Depression Scale (GDS-15) [13], the Abbreviated Mental Test Score (AMTS) [14] and the Cumulative Illness Rating Scale Geriatric (CIRS-G) [15], a measure of physical co-morbidity. A higher score on the GDS-15 indicates that the patient has more depressive symptoms. The CIRS-G scores severity on 13 different organ systems ranged from 0 to 4; a higher score indicates a higher level of physical co-morbidity. The AMTS is a commonly used brief measure of cognitive function scored out of 10; a cut-off of 7 or 8 is suggested to discriminate between cognitive impairment and normality [16]. Participants with moderately severe impaired cognitive function (AMTS<6) were excluded, as their responses to the other rating scales were likely to be unreliable. All rating scales have been validated in older people [16 18]. Patients were tracked on the hospital information system to ascertain discharge date and destination or inpatient death.
Statistical analysis
The associations of length of stay with depressive symptoms and cognitive function were estimated using ordinary leastsquares regression. To accommodate the large positive skew in the distribution of length of stay, condence intervals were obtained using the bias-corrected and accelerated percentile bootstrap method (4999 resamplings), and corresponding P-values were calculated using a robust estimate of the standard error [19]. Logistic regression was used to compare the odds of the patient being discharged to a community hospital (for rehabilitation) or institutional care, or dying in hospital, rather than being discharged to their usual place of residence. Data from 62 participants who subsequently entered the trial and received the active treatment were included, but we allowed for exposure to the intervention in all the regression analyses. Consequently, all analyses were repeated three times with accumulating adjustment for exposure to the trial intervention, age and sex, physical co-morbidity and cognitive function. Adjustment was achieved by adding dummy variables to the regression equation, distinguishing the cate-
691
S. Cullum et al.
Table 2. Association of length of hospital stay with depressive symptoms
Crude (1) Mean difference 95% CI + Age, sex adjusted (2) Mean difference 95% CI Fully adjusted (3) Mean difference 95% CI
................................................................................................................ GDS-15 04 Baseline Baseline Baseline 57 2.96 1.37, 7.29 2.96 1.33, 7.25 2.65 1.61, 6.90 810 0.21 5.01, 5.43 0.31 4.84, 5.46 1.75 5.88, 2.37 1115 11.42 2.84, 25.67 11.06 3.48, 25.61 9.46 1.80, 20.73 P for trend 0.13 0.14 0.17
Adjusted for (1) exposure to the trial intervention, (2) plus age and sex, (3) plus co-morbidity and cognitive function.
................................................................................................................ 04 290 (83.5%) 8 (2.5%) 16 (4.5%) 14 (4.0%) 19 (5.5%) 57 101 (76.0%) 0 (0.0%) 4 (3.0%) 8 (6.0%) 20 (15.0%) 810 61 (68.5%) 2 (2.5%) 6 (6.5%) 9 (10.0%) 11 (12.5%) 1115 30 (62.5%) 1 (2.0%) 6 (12.5%) 5 (10.5%) 6 (12.5%) Total 482 (78.0%) 11 (2.0%) 32 (5.0%) 36 (6.0%) 56 (9%) Odds of hospital outcome by GDS-15 score Crude (1) OR 04 57 810 1115 P for trend 04 57 810 1115 P for trend 04 57 810 1115 P for trend Baseline 1.64 4.20 5.53 0.001 Baseline 0.72 1.42 2.76 0.258 Baseline 3.02 2.41 2.53 0.012 Age and sex adjusted (2) 95% CI OR 95% CI Community hospital (n = 36) 0.67, 4.03 1.66, 10.61 1.69, 18.06 Baseline 1.66 4.64 6.23 0.67, 4.14 1.78, 12.08 1.79, 21.73 Fully adjusted (3) OR Baseline 1.66 3.35 3.98 0.004 Baseline 0.70 1.71 4.05 0.032 Baseline 2.67 2.11 2.80 0.017 95% CI
GDS-15 score
Usual residence
Care home
Other hospital
Death
0.001 Other hospital (n = 64) 0.23, 2.20 0.44, 4.59 0.78, 9.73 Baseline 0.69 1.35 3.02 Death (n = 56) 1.55, 5.89 0.95, 6.09 0.77, 8.37 Baseline 3.19 2.70 2.48 0.010 0.247 0.22, 2.13 0.41, 4.49 0.84, 10.92
Each destination in turn is compared with discharge to usual place of residence, adjusted for (1) exposure to the trial intervention, (2) plus age and sex, (3) plus co-morbidity and cognitive function
services. The high depression score correlated with a lower cognition score (Spearmans r = 0.10, P = 0.013) and a higher co-morbidity score (r = 0.26, P<0.001), whereas there was no evidence of association between cognition and co-morbidity scores (r = 0.04, P = 0.38).
Length of stay
nitive function and higher score on the depression rating GDS-15 but was largely unaffected by gender. Table 2 shows that, after adjustment for potential confounding factors, depressive symptoms were not associated with longer length of stay (P for trend = 0.17).
Hospital outcomes
Table 1 shows that median length of stay in hospital increased with age, greater physical co-morbidity, poorer cog-
Table 3 shows the hospital outcome for inpatients in terms of discharge destination and mortality by a GDS-15 score on admission.
692
Of the 617 participants, 36 (5.8%) were transferred to a community hospital for rehabilitation prior to discharge. Table 3 shows the odds of being transferred to a community hospital (compared with discharge to usual place of residence) by depression rating score on admission. The adjusted odds were higher in participants with a higher GDS-15 depression rating score (P for trend = 0.004).
Institutional care
Only 11 of the 617 (1.8%) participants were discharged from hospital to institutional care (residential or nursing home), which precludes an investigation of the association with depressive symptoms.
Inpatient mortality
both the eligibility and inclusion criteria. Patients who were discharged before 3 days or were too ill/confused to participate in the study might have had different outcomes to those who took part. Nonetheless, the ndings are generalisable to those older medical inpatients who are most likely to be screened for depression in a real world NHS setting, i.e. those who stay in hospital for more than 3 days and are t enough to complete a screening questionnaire. It is important to note that this study examined the effect of depressive symptoms rather than the clinical syndrome of depressive disorder although there is a strong association between the GDS score and clinical diagnosis. An associated prevalence study included one-third of the participants in the present study, with diagnostic interviews allowing the prevalence of ICD-10 depressive disorder to be estimated as 17.7% (95% CI 12.922.5) [20].
Fifty-six of the 617 (9.1%) participants died in hospital. Table 3 shows the odds of death in hospital compared with discharge to usual place of residence by depression rating score on admission. A higher GDS-15 score on admission was associated with a higher risk of inpatient death, even after adjustment for potential confounding factors such as age, physical co-morbidity and cognitive function (P = 0.017).
Discussion
The main ndings of this study are that a higher depression rating on admission to hospital is associated with increased odds of inpatient death and increased odds of being transferred to a community hospital for rehabilitation independent of age, sex and physical status. Depressive symptoms were not associated with longer length of hospital stay. Most of the screen positive participants were invited to take part in a nested trial of liaison psychiatric nursing. The active intervention in the trial was assessment and implementation of a management plan by a liaison psychiatric nurse compared to standard care. It is possible that the active intervention may have biased our estimate of the association between depressive symptoms and adverse outcomes, but the trial intervention had little effect and moreover was adjusted for in the analyses. A Hawthorne effect, due to all trial participants being interviewed again at follow-up, may also have biased the ndings, as trial participants had higher depression ratings. However, this is unlikely as most of the interviews took place after discharge.
Strengths and weaknesses
Studies carried out in the USA [21] and Italy [22, 23] support the ndings of our study that, after adjustment for potential confounding factors, length of stay is not associated with depressive symptoms. However, a Swiss study [24] found that depressed mood (GDS-15>6) was associated with inappropriate hospital days (P = 0.001), indicating that, although depressive symptoms may not increase total length of stay, they may increase inappropriate use of inpatient hospital resources.
Other hospital outcomes
Very few studies have considered whether depressed mood affects the need for rehabilitative or long-term institutional care, but our study found that depressive symptoms were associated with transfer to a community hospital for rehabilitation. In Switzerland, Bula et al. [4] similarly found that depressive symptoms were associated with an increased risk of nursing home placement and increased average costs of both acute and rehabilitation services in elderly medical inpatients. However, after adjustment for demographic factors and physical status, only the association for rehabilitation services remained. This is consistent with the ndings in our study.
Mortality
The main strength of the study was its large sample size, which was greater than that in any other identied study that has measured similar outcomes. The large sample size, plus the measurement of and adjustment for potential confounding factors, permits a more condent interpretation of the ndings. The main weakness of the study was that the ndings could only be generalised to the 45% of admissions that met
A number of studies, with follow-ups ranging from 90 days to 9 years, have reported an association between depressive symptoms and longer-term mortality [49]. However, very few studies have examined the effect of depressive symptoms on inpatient mortality. One study [1] reported that mortality in older medical inpatients was signicantly higher among depressed compared with nondepressed controls (6 vs 0 deaths, P = 0.03), but the analyses were not adjusted for potential confounding factors associated with depressive symptoms such as physical co-morbidity. Another study [25], carried out in a population of younger medical inpatients (mean age
693
S. Cullum et al.
50 years), found that, after adjustment for age, number, type and severity of patients illnesses, the diagnosis of major depressive disorder did predict mortality, whereas demographic variables and medical diagnoses did not. These ndings are supported by the results of our study that inpatient mortality is independently associated with severity of depressive symptoms.
Does identication and treatment of depression make a difference?
tively improves (or does not improve) health outcomes will inuence the development of healthcare policy. Researchers carrying out these studies should consider including inpatient death and use of rehabilitation services as potential outcomes.
Key points
r Depression is common in older medical inpatients, but
We can only speculate about the causal pathways underlying these ndings. Does depression reect an underlying diminution in the bodys ability to recover from serious physical illness, thus increasing the likelihood of inpatient death and/or need for rehabilitation services? Or do hospital staff misdiagnose depression as slower recovery from physical illness and thus recommend rehabilitation when they should be recommending treatment for the depressive illness instead? Frailty is used loosely to describe a range of conditions in older people, including general debility and cognitive impairment associated with increased economic costs [26]. It is generally asserted that a combination of biomedical and psychosocial factors inuences frail peoples physiological state to the extent that its function is reduced, and subsequent exposure to further minor environmental stresses may lead to dependence. The ndings of this study suggest that depressive symptoms may be an important indicator of frailty. If depressive symptoms are a marker for frailty, then the question is: will treatment of depression have any impact on health outcomesmental or physical? Studies in primary care demonstrate that treatment of depression improves mental health, physical health and quality of life [2729], but the evidence for the effectiveness of treatment of depression in general hospital inpatients is less clear [12, 3032]. Two recent RCTs [12, 33] carried out in general hospital settings (in Canada and the UK) were unable to demonstrate that systematic detection and multidisciplinary care of depression was more benecial than usual care for elderly medical inpatients, but both of these studies were underpowered. Another larger UK study [34] evaluated the effectiveness of a nurseled mental health liaison service for older medical inpatients and found that subjects randomised to the intervention arm had signicantly lower GDS scores at follow-up than those receiving usual care.
its recognition is poor and treatment often inadequate, despite research suggesting that depression may be independently associated with adverse hospital outcomes. r The aim of this study is to examine the relationship between depressive symptoms and hospital outcomes in older medical inpatients. r Our study found that depressive symptoms are independently associated with an increased likelihood of inpatient death and transfer to a community hospital for rehabilitation, but are not associated with longer length of stay. r Further research is required to establish whether screening, assessment and treatment of depression identied in this population will reduce the risk of these adverse outcomes.
Acknowledgements
Maggie Bray (research assistant); Barbara Lee (research nurse); Linda Sharples, Lu Gao and Tony Johnson (medical statisticians); Prof. Ann Louise Kinmonth (sponsorship of MRC Fellowship and advice on study design).
Conicts of interest
None.
Funding
This work was supported by a MRC Health Services Research Training Fellowship (G106/950) and a NHS Executive Eastern Research and Development Project Grant (RCC33041). Financial sponsors played no role in the design, execution, analysis and interpretation of data or writing of the study.
References Conclusions
This study has shown that depressive symptoms are associated with an increased likelihood of inpatient death and transfer to a community hospital for rehabilitation, but are not associated with longer length of stay. Currently, we have no robust research evidence regarding the impact and costeffectiveness of screening, assessment and treatment of depression identied in this population. There is clearly a need for further research in the area, as evidence that identication and treatment of depression in older medical inpatients effec1. Koenig HG, Shelp F, Goli V, Cohen HJ, Blazer DG. Survival and health care utilization in elderly medical inpatients with major depression. J Am Geriatr Soc 1989; 37: 599606. 2. Holmes J, House A. Psychiatric illness predicts poor outcome after surgery for hip fracture: a prospective cohort study. Psychol Med 2000; 30: 9219. 3. Koenig HG, Kuchibhatla M. Use of health services by medically ill depressed elderly patients after hospital discharge. Am J Geriatr Psychiatry 1999; 7: 4856. 4. Bula CJ, Wietlisbach V, Burnand B, Yersin B. Depressive symptoms as a predictor of 6-month outcomes and services
694
5.
6.
7.
8.
9.
10.
11.
12.
13.
14. 15.
16.
17.
18.
19.
695