Sage Journals: Discover world-class research

Abstract

Objectives: The aim of this study was to establish the predictive value of an ICD-10 diagnosis of depressive disorder or dysthymia (depressive patients) among 70 years + frail rural community living patients by measuring morbidity, mortality and use of health services. Identical measures were studied over time in general elderly populations.

Methods: Outcome measures: morbidity, mortality and use of health services were registered over 13 years in: (i) a clinical cohort of frail community-living depressive patients (n = 38), a frail control group (n = 116) and non-frail elderly people (n = 575), all living in the same municipality, and (ii) register-based samples of general rural (n = 4 115) and capital living (n = 54 977) elderly populations. The outcome measures were compared using data from Danish national health registers.

Results: Every one of the clinical cohort of depressive patients had died at the end of the study. Compared with both the frail control group and the non-frail elderly people, the depressive patients had significantly more psychiatric hospital days, outpatient home visits, antidepressant use, antipsychotic use, GP service use and more psychiatric diagnoses with higher morbidity. However, greater use of somatic hospital services or more somatic diagnoses among the depressive elderly patients were not observed. The general elderly population living in the capital had, compared with rural equals, significantly more somatic and psychiatric diagnoses, larger use of somatic hospital services, psychiatric hospital days, antipsychotics and anxiolytics, but less use of antidepressants, psychiatric outpatient home visits and GP services.

Conclusions: An ICD-10 diagnosis of depressive disorder or dysthymia predicted increased use of psychiatric services, more psychiatric diagnoses and increased mortality, indicating poor late-life psychiatric outcome. Contrasting with other studies, depression did not predict increased use of somatic hospital services or more somatic diagnoses. The differences in health care status and use between elderly living in the capital and in rural areas elderly are novel findings.

Keywords

depression elderly general elderly population follow up studies morbidity mortality health services utilization

The high prevalence of depressive disorder patients and clinically significant depressive symptom ‘cases’ (i.e. patients with bothersome depressive symptoms that do not fulfil formal criteria for a depressive disorder) among the elderly [1] underlines the relevance of follow-up studies of these patients concerning morbidity, mortality and health care services utilization. Such studies have shown that community-living depressive disorder elderly patients and depressive symptom cases use more somatic in- and outpatient services including emergency medical services, primary care services and medications, but also that they use less psychiatric services including antidepressants [1–9]. Two studies showed higher consumption of somatic services to be particularly prominent and that elderly depressive disorder patients are more likely to be referred to services not suitable for treatment of depression [10,11]. In another two studies, community-living depressive disorder patients had only increased outpatient services utilization [12,13].

Literature on the comorbidity with depressive disorders shows that physical illness (e.g. cardiovascular and cerebrovascular diseases), cognitive disorders and functional impairment are main predictors of depressive disorders in the elderly [1].

Studies of mortality with depression are contradictory. Some studies report depressive symptom ‘cases’ and mild to moderate depression to be associated with no increased risk [14,15] while others report increased mortality [16–18]. Most studies, however, showed increased mortality in severe depression [14,17,19–21], especially chronic depression [22].

Follow-up studies show that depressive disorders among elderly in primary care have a poor prognosis [23] with under-utilization of psychiatric services [24–26]. Inconsistent with guidelines, elderly patients with and without depression commonly receive treatment with benzodiazepines [26,27,28] and it seems well documented that antidepressants are under-used [1,26].

A study of the need for psychiatric services among rural elderly Americans with depressive symptoms suggested that lack of awareness of available services and lack of routine contact with mental health service providers are important factors limiting the use of these services [29].

In an 8-year controlled follow-up study of inpatients with depressive disorder and frail rural community-living patients with depressive disorder or dysthymia [30] diagnosed according to ICD-10 [31], we saw significantly increased rates of recurrent depression, consumption of antidepressants, use of psychiatric in- and outpatient admissions, hospital days and outpatient home visits, but no increased use of somatic services or mortality [32].

The overall objective of the present 13-year follow-up study was to evaluate whether there were differences in morbidity, mortality and use of health care services as a function of depression, frailty and urban/rural living.

Materials and methods

Materials

This study included individuals 70 years of age or older (70+ years) on the date of entry in the index study in 1996, i.e. groups I and II [30] (Figure 1). In 1996 the subgroups I and II were extracted from the local social welfare register system. Frail elderly people are defined as community-living elderly people to whom municipality visiting nurses have assigned municipality home help in their private home, or residence in a public nursing home on mental and/or physical indications.

Figure 1.

Definitions of the five groups of 70 years of age or older (elderly) participants of the study

The study also included two register-identified 70+ years general populations of rural (group IV) and urban populations (group V), respectively. These data, and data on group III (Figure 1), were extracted from the Danish Civil Registration System [33].

The study was approved by the Danish Data Protection Agency and the scientific ethic committee.

Methods

Overall study methodology

We compared morbidity, mortality and use of health services among five subgroups, i.e. (i) the specified clinical groups I, II and III, and (ii) the register [33] identified general elderly rural and capital living populations, groups IV and V (Figure 1). The general elderly rural and capital populations are studied with the purpose to widen the epidemiological perspective of the subject of morbidity, mortality and services utilization among elderly people living in different sociodemographic areas.

Periods of investigation

The study included data on health characteristics from 31 March 1996 and follow-up periods. Specification of the periods of investigation appear from Table 1, column one.

Table 1.

The Table shows results from comparisons of the Depr cohortMunicipality (group I) versus the ControlsMunicipality (group II) and the Depr cohortMunicipality (group I) versus the Non-frailsMunicipality (group III).

Comparisons		Group I versus group II	Group I versus group III
Registry	Measures	IRR (CI)	P value	IRR (CI)	P value
National Somatic Hospital Register 31 March 1996 –10 December 2009	Hospital days	0.7 (0.5–1.0)	0.058	0.8 (0.6–1.2)	0.26
Diagnoses	0.8 (0.5–1.2)	0.27	0.9 (0.6–1.3)	0.64
Outpatient visits	0.9 (0.6–1.2)	0.41	1.0 (0.7–1.5)	0.79
Diagnoses	0.9 (0.6–1.3)	0.46	1.1 (0.8–1.6)	0.57
Psychiatric Central Research Register 31 March 1996–10 December 2009	Hospital days	9.4 (1.8–49)	0.0045	7.1 (2.4–21)	0.0022
Diagnoses	9.3 (1.8–48)	0.0047	7.0 (2.4–21)	0.0024
Outpatient home visits	3.2 (1.7–6.0)	0.0004	2.3 (1.4–3.8)	0.0027
Diagnoses	3.3 (1.7–6.2)	0.00058	2.3 (1.4–3.9)	0.0045
National Health Service Register 1 January 1996–10 December 2009	Office hour services	1.3 (1.1–1.6)	0.0068	1.7 (1.4–2.1)	3.3e–7
Out-of-hours services	1.1 (0.8–1.4)	0.55	1.1 (0.9–1.4)	0.26
		Difference (CI)	P value	Difference (CI)	P value
Prescription Register 1 January 1996–31 December 2008	All somatopharmaceuticals	0.30(− 0.011–0.61)	0.059	0.71(0.43–1.00)	9.0e–7
		OR (CI)	P value	OR (CI)	P value
	Antidepressants	17 (7.7–39)	5.2e–12	13 (6.3–26)	1.3e–12
Antipsychotics	3.9 (1.5–10)	0.0045	3.1 (1.4–7.0)	0.0070
Anxiolytics	1.3 (0.7–2.4)	0.45	1.4 (0.8–2.5)	0.21
Hypnotics	2.5 (1.2–5.1)	0.015	2.7 (1.4–5.2)	0.0025
		HR (CI)	P value	HR (CI)	P value
Civil Registration System 31 March 1996 –10 May 2010	Mortality	1.6 (1.1–2.4)	0.014	1.8 (1.3–2.5)	0.0024

Data

Data were collected from the Danish national disease and health registers that provide prospectively registered data. The registers are: (i) the Danish National Somatic Hospital Register [34], (ii) the Danish Psychiatric Central Register [35], (iii) the Danish National Health Service Register [36], (iv) the Danish Prescription Registry [37], (v) the Centralised Civil Registration System [38] and (vi) the Danish Register of Causes of Death [39].

Variables

The variables measured during follow-up were: (i) hospital days, (ii) outpatient visits, (iii) psychiatric and somatic disease diagnoses, (iv) GP services, (v) medications, (vi) causes and manners of death and (vii) survival time. Note that the National Health Service Register only provided type and quantity of GP services, not diagnoses.

Statistics

Comparison groups

First, the three clinical groups I, II and III from the municipality (Figure 1) are compared with each other. Next, the register identified four rural municipalities (group IV) are compared with the central urban area municipality population of Copenhagen (group V).

Follow-up period

Each individual's period in the study (‘risk time’) was calculated as the time in years from study entry and until death or end of registration. Period length varies between registries as noted in Table 1. The logarithm of the risk time was included as a covariate in all analyses to account for differences in follow-up time except for survival analyses where the dates define the entry and exit time.

Confounders

The subjects’ gender and age on 31 January 1996 were included as covariates in all analyses. Age was centralized by subtraction of the grand mean calculated from the groups to be compared.

Analyses/models

The following quantities (individually aggregated over the entire study period) were analysed using negative binomial generalized linear models (GLMs) with logarithmic link function [40]: number of hospital days, number of outpatient treatments and number of diagnoses (separate analyses for somatic and psychiatric hospital admissions), and the number of GP services (office hours and out-of-hours, respectively). Negative binomial regressions were chosen to account for over dispersion, i.e. extra variation that implies a pure fit of the usual Poisson distribution.

We used the parameterization with constant dispersion (opposed to the parameterization where the dispersion is a function of the expected mean) as this resulted in the best fit. The result from comparing one group with another is shown as an incidence rate ratio (IRR) with a corresponding 95% confidence interval (CI). Survival curves and estimated hazard rate ratios (HR) were obtained from Cox regressions. Natural versus unnatural manner of death (only available until 31 December 2006) was analysed using a binomial GLM with a logarithmic link function [40]. In every case, two-sided likelihood ratio tests (one degree-of-freedom chi-square) were used to test the null hypothesis that the ratio equals one.

Analysis/model for the prescription data

Separate analyses were made for the following groups of Anatomical Therapeutic Chemical (ATC) codes: antidepressants (N06A), antipsychotics (N05A), anxiolytics (N05B), hypnotics (N05C), and somatopharmaceuticals (all other codes). Contrary to the other analyses, data from the prescription registry included a yearly aggregated number of sold defined daily dosages (DDD) for each ATC code. The logarithm of the proportion of a year that the individual was alive was included as a covariate to the models. Records missing ATC (and therefore DDD) do not contribute to the yearly counts but are potential confounders, as the pattern of missing differs between the groups and the number of missing ATC to some degree correlates to the aggregated measure of drug use (results not shown). In addition to the adjustment for gender and age, we therefore included a dichotomous factor giving a change in intercept for those having at least one record missing ATC (in that year) plus a slope parameter for the number of missing ATCs (individually aggregation per year). Moreover, potential change in medication use over years was accounted for by a four degree polynomial in years since 1996. Somatopharmaceuticals were analysed using generalized estimation equations (GEEs) with a gamma distribution, logarithmic link function, and exchangeable correlation structure [41]. Coefficients (or contrast) measuring the difference in DDDs (logarithmic scale) between groups were estimated with 95%CI and Wald's tests (one degree-of-freedom chi-square) were used to test the null hypothesis that the coefficient equals zero. Psychopharmaceutical usage (yes/no) of each of the four medications was analysed by use of binomial GEEs with logit link function [41] and the same covariates and confounders as for somatopharmaceuticals. Results are given as odds ratios (OR) with 95%CI and the null hypothesis, OR = 1, was tested by a Wald's test.

Multiple testing

All analyses were carried out using Stata 11 (StataCorp 2009). A significance level of 5% was chosen but as a number of tests have been carried out the thresholds for significance should be adjusted for multiple testing [42]. We consider each pair of groups compared and each registry a separate family of tests. Thus to keep the family-wise error rate near 5% we (maybe somewhat conservatively) used a Bonferroni threshold (Bt) for four tests on somatic registrations (Bt = 0.0125), four on psychiatric disorders (Bt = 0.0125), two for health service utilization (Bt = 0.025), five concerning medications (Bt = 0.01) and one for mortality (Bt = 0.05). That is, only P values below the corresponding Bt are considered significant. The comparison of natural/unnatural death was considered explorative only. P values and CIs are shown unadjusted.

Results

Median ages with standard deviations in parentheses and age ranges for the groups I, II, III, IV and V (Figure 1) were: 79.7 (6.3); 69.0–93.4, 81.3 (6.8); 69.1–101.1, 76.4 (6.8); 70.0–98.2, 77.7 (5.9); 70.0–101.1 and 79.3 (6.2); 70.0–106.7.

The total risk time measured in person years was 197 for the Depr cohort^Municipality (group I), 717 for the Controls^Municipality (group II), 4,547 for the Non-frails^Municipality (group III), 29,579 for the Gen popul^Rural (group IV) and 362,941 for the Gen popul^Capital (group V). Total number of participants (n) in the five groups and percentages of women (%) were on 31 March 1996: group I n = 38 (74%), group II n = 116 (64%), group III n = 575 (49%), Group IV n = 4,115 (55%) and Group IV n = 54,977 (68%). All patients in the Depr cohort ^Municipality had died at the end of the study.

Results from comparisons of the measured variables among the Depr cohort ^Municipality (group I) versus the Controls ^Municipality (group II) and the Depr cohort ^Municipality (group I) versus the Non-frails ^Municipality (group III) are shown in Table 1.

The results from comparing measures of the Controls ^Municipality (group II) versus the Non-frails ^Municipality group III) are not shown. However, we will note the significant differences: Controls ^Municipality received more daytime services with GPs (p = 0.00011) and had more prescriptions of defined daily doses of all somatopharmaceuticals (p = 0.0001). There were no significant differences as for prescriptions of psychopharmaceuticals and mortality.

Table 2 shows the results from comparisons of the Gen popul^Rural (group IV) versus the Gen popul^Capital (group V).

Table 2.

The Table shows results from comparisons of the Gen populRural (group IV) versus the Gen populCapital (group V).

Comparisons		Group IV versus group V
Registry	Measures	IRR (CI)	P value
National Somatic Hospital Register 31 March 1996–10 December 2009	Hospital days	0.70 (0.68–0.72)	1.0e–124
Diagnoses	0.78 (0.76–0.80)	2.1e–73
Outpatient visits	0.73 (0.71–0.75)	3.7e–111
Diagnoses	0.68 (0.66–0.71)	1.3e–140
Psychiatric Central Research Register 31 March 1996–10 December 2009	Hospital days	0.55 (0.45–0.67)	5.9e–11
Diagnoses	0.54 (0.44–0.66)	2.6e–11
Outpatient home visits	2.73 (2.51–2.98)	3.5e–91
Diagnoses	2.78 (2.54–3.03)	1.7e–91
National Health Service Register 1 January 1996–10 December 2009	Office hour services	1.46 (1.43–1.49)	2.1e–283
Out-of-hours services	1.10 (1.08–1.12)	1.4e–18
		Difference (CI)	P value
Prescription Register 1 January 1996–31 December 2008	All somatopharmaceuticals	0.10 (0.074–0.13)	1.1e–11
		OR (CI)	P value
	Antidepressants	1.18 (1.09–1.26)	1.2e–5
Antipsychotics	0.70 (0.63–0.77)	5.1e–13
Anxiolytics	0.68 (0.64–0.73)	4.6e–26
Hypnotics	0.93 (0.87–0.99)	0.024
		HR (CI)	P value
Civil Registration System 31 March 1996–10 May 2010	Mortality	0.89 (0.86–0.92)	3.5e–11

The proportion of individuals that died during the entire study period in groups I, II, III, IV and V were: 100%, 88%, 86%, 87% and 91%. In the five groups the range of natural death was 94.7% to 97.1% with no statistically significant differences between the groups. The hazard rate ratios comparing mortality between groups can be seen in Tables 1 and 2. Survival curves are shown for groups I, II and III in Figure 2 and for groups IV and V in Figure 3.

Figure 2.

Survival curves for the cohort of frail depressed patients (Group I, Depr cohort^Municipality) (n = 38), the controls (Group II, Controls^Municipality) (n = 116) and the non-frails (Group III, Non-frails ^Municipality) (n = 575), all living in the municipality of Sonderhald, adjusted for age and gender. Hazard rate ratios are shown in Table 1.

Figure 3.

Survival curves for the general elderly rural populations (Group IV, Gen popul ^Rural) (n = 4 115) and the general elderly population of the central municipality of Copenhagen (Group V, Gen popul ^Capital) (n = 54 977), adjusted for age and gender. Hazard rate ratios are shown in Table 2.

Discussion

The overall objective of the study was in a register-based design to evaluate whether there were differences in morbidity, mortality and use of health care services as a function of depression, frailty and urban/rural living.

Main results

During follow up the clinically established Depr cohort^Municipality (group I) had significantly larger use of psychiatric hospital days, outpatient home visits, antidepressants, antipsychotics and GP office hours services compared with the Controls^Municipality (group II) (Table 1). Further, it had more psychiatric diagnoses including larger morbidity. However, there was no increased use of somatic hospital services or more somatic diagnoses. Versus the Gen popul^Rural (group IV), the Gen popul^Capital (group V) had significantly larger use of somatic hospital services including more somatic diagnoses, psychiatric hospital days including more psychiatric inpatient diagnoses, antipsychotics and anxiolytics, but less psychiatric outpatient diagnoses and less use of psychiatric outpatient home visits, antidepressants, GP services and all somatopharmaceuticals (Table 2).

Similar to the 8 year follow up [32] we observed no significant differences in use of somatic hospital services (Table 1) and number of somatic diagnoses between the Depr cohort^Municipality (group I) and the Controls^Municipality (group II) in this 13 year follow up and the same results apply to the Depr cohort^Municipality (group I) versus the Non-frails^Municipality (group III). The results are contrary to other studies showing increased use of somatic services by depressed patients [2–13]. This is seemingly a result of a well-functioning old age psychiatry, intensively as well as extensively focusing on both mental and physical health.

In identical comparison combinations, i.e. the Depr cohort^Municipality versus the Controls^Municipality and the Non-frails^Municipality the number of psychiatric diagnoses and use of psychiatric hospital days, outpatient home visits, antidepressants, antipsychotics and GP office hours services (but not out of hours services) were significantly increased. This is in disagreement with other studies showing under-utilization of psychiatric services [1–13,23–26]. One study found limited use of available mental health services among rural elderly Americans with depressive symptoms [29], but the results of the present study indicate that this does not seem to be the case among Danish equals. Seemingly, the reason is the same as mentioned in the last three lines of the above section.

Like in other studies of depressive elderly patients [26–28], hypnotics were more used by the Depr cohort^Municipality versus Controls^Municipality though not significantly when adjusted for multiple testing.

The risk of death (hazard rate) was significantly higher for the Depr cohort^Municipality both versus the Controls^Municipality and versus the Non-frails^Municipality (Table 1). This is in agreement with studies of severe depression [14,17,19–21] and some studies of depressive symptom cases and mild to moderate depression [16–18] but in disagreement with others [14,15].

However, this end point follow-up study of the Depr cohort^Municipality did not come up with new information concerning the measured health characteristics except that the hazard rate for risk of death was much increased (Table 1). This was not the case in the 8-year follow-up after adjustment for multiple testing [32]. Although there were no significant differences in cardiovascular, cerebrovascular and respiratory disorders between the Depr cohort^Municipality and the Controls^Municipality at baseline [32], the increased mortality rate may be due to comorbid somatic disease [1] even if the Depr cohort^Municipality did not use more somatic hospital days compared with the Controls^Municipality and the Non-frails^Municipality. However, the Depr cohort^Municipality did use more GP office hours services and all somatopharmaceuticals (Table 1) versus both Controls^Municipality and Non-frails^Municipality (1), which suggests that the Depr cohort^Municipality may have had more somatic disorders/diseases during the follow up and that the GPs mainly treat these patients themselves. However, some of the excess mortality may be caused by the depressive disorder itself.

The much increased use of psychiatric hospital services, especially outpatient home visits, in the present study by the Depr cohort^Municipality versus both the Controls^Municipality and the Non-frails^Municipality (Table 1) may be, as we do not know the GP referring pattern prior to this study, a result of the interventions in the municipality (Figure 1) in connection with the study of prevalence of depression among frail elderly rural community-living individuals in 1996 (see Figure 1, group I) [30]. The interventions in continuation of the study targeted both GPs and nursing staff and included lessons on old age psychiatry and presentations and discussions of the results of the study. All participants of the study diagnosed with an ICD-10 depressive disorder or dysthymia were discussed between the patient's GP and the principal investigator (the first author), who worked out a treatment plan to be carried out by the GPs [30].

Overall, the results of the study indicate poor late-life psychiatric outcome of the Depr cohort^Municipality. Somatic morbidity, less use of somatic hospital services and the depressive disorder or dysthymia of the members of the cohort seem to explain increased mortality. However, the study does not allow us to point out the relative importance of each element due to the naturalistic design.

Concerning age ranges of groups I and II (see results, first section) two persons from group I and one from group II were between 69 and 70 years old. This is most likely due to mistakes made during the inclusion process in the prevalence of depression study in 1996 [30].

The frail Controls^Municipality (group II)'s significantly larger use of all somatopharmaceuticals and GP office hour services versus the Non-frails^Municipality (group III) (see Results, fourth section) suggests that the frail Controls^Municipality have more somatic disorders. The fact that the Controls^Municipality did not have significantly larger use of somatic hospital services suggests that the GPs of the municipality mainly treat their somatic disorders themselves.

The significantly increased mortality in the follow-up period among the register identified populations, Gen popul^Capital (group V) versus the Gen popul^Rural (group IV) is conspicuous (Table 2). However, if we assume there were no differences in the general health and need of health services between the two populations, the interpretation of the results of the comparisons would be that the Gen popul^Capital had significant excess use of somatic hospital services including psychiatric hospital days versus the Gen popul^Rural, but used much less psychiatric outpatient home visits and less GP office hours services and out-of-hours services (Table 2). However, the clinical relevance of 10% less use of out-of-hours services is unclear. The Gen popul^Capital's significantly more somatic and psychiatric inpatient diagnoses (Table 2) during the follow-up does not necessarily reflect poorer general health but may be a consequence of more admissions due to the same disorder(s). However, note that the Gen popul^Capital had much less psychiatric outpatient home visits and diagnoses versus the Gen popul^Rural (Table 2).

Rather than just larger use of somatic hospital services and psychiatric hospital days, and less use of psychiatric outpatient home visits and especially GP office hours services by the Gen popul^Capital (Table 2), part of the explanation of the differences in use of the mentioned health services may be insufficient capacity or skills of the central capital municipality home health care services to take over inpatients for post-hospital home-care after discharge. Likewise, there may have been insufficient capacity of the psychiatric outpatient home visit service and GP services. Improved overall health care, i.e. improved access to suitable services, e.g. central capital municipality home health care services and psychiatric outpatient home visits and GP services, may improve the health care of the age group and reduce the need for somatic hospital services and psychiatric hospital days. Alternatively, the general health of the Gen popul^Capital may be significantly poorer than that of their rural living equals as indicated by the increased mortality; however, the results of the present study do not give an answer to this important question. Searches did not come up with similar studies and so the findings seem to be novel.

The Gen popul^Capital's significant lower use of somatopharmaceuticals and GP services compared with their rural equals may reflect insufficient treatment of somatic diseases, which may in turn contribute to the increased mortality. However, the clinical significance of 10% less use of all somatopharmaceuticals is unclear. The Gen popul^Capital's more use of antipsychotics and anxiolytics may be indicated but it still used more psychiatric hospital services except psychiatric outpatient visits. The Gen popul^Capital's more use of anxiolytics may support the results of other studies showing that elderly people commonly receive inexpedient treatment with anxiolytics [1,27,28]. Compared with rural equals the Gen popul^Capital used fewer antidepressants (Table 2), which may lead to increased use of psychiatric as well as somatic services. The latter because depression in the elderly often presents with somatic symptoms [43,44]. The Gen popul^Capital's lower use of antidepressants may even suggest that antidepressants are under-used [1,26–28]. The latter may be due to GP's difficulties with depression diagnostics among elderly patients [1,30,43].

Strengths of the study

The data from the national health registers are prospectively collected and updated within the time-frame of periods; further, the high number of person years regarding the general elderly populations is a strength of the study made possible by the use of the unique Danish health registers. Finally, it was possible to do a follow up of the municipality-living frail depressive patients from index diagnosis till they all had died.

Limitations of the study

The main limitation of the study was the limited number of person years contributed by the municipality-living depressive patients and controls. Data may also be missing in the registries. Overall register data are collected from daily clinical practice, i.e. they have not been collected by means of standardized research criteria.

Summary of results and conclusion

In this study, frail rural community-living depressed elderly patients used more of GP services, psychiatric hospital services, antidepressants and antipsychotics and had increased mortality, indicating poor late-life psychiatric outcome. The results point to the need for a more effective follow-up regime for this patient category, e.g. life-long follow-up by old age psychiatry district service and/or arranged collaborative care models with primary care involving regular follow ups, primarily to effectively observe for recurrence of depressive disorder and to ensure antidepressant medication compliance.

The register-established general population sample of the central capital municipality of Copenhagen showed increased use of somatic hospital in- and outpatient services and inpatient psychiatric services and higher mortality compared with rural equals. This may be due to a composite of poorer general health, insufficient access to outpatient psychiatric home visits, GP services, municipality care services and possibly to some extent due to lower rate of treatment with antidepressants. The results may indicate less sufficient old age psychiatry care in the elderly Gen popul^Capital in some areas. Thus, the perspectives for future research may be intensified focus on studies of treatment and evaluation of extent and quality of the care for elderly urban populations.

Footnotes

Acknowledgements

We are grateful to Gurli Perto and Søren Skadhede for skilled assistance and valuable discussions concerning the databases.

Declaration of interest: Povl Munk-Jørgensen has since January 2009 received unconditional grants, and/or minor honoraria from H. Lundbeck A/S, AstraZeneca, Eli Lilly, Janssen-Cilag A/S and Bristol-Myers Squibb. The other authors have no conflicts of interests.

References

1. Djernes

. Prevalence and predictors of depression in populations of elderly: a review. Acta Psych Scand 2006; 113:372–387.

2. Unützer

Patrick

Simon

. Depressive symptoms and the cost of health services in HMO patients aged 65 years and older. A 4-year prospective study. JAMA 1997; 277:1618–1623.

3. Callahan

Hui

Nienaber

Musick

Tierney

. Longitudinal study of depression and health services use among elderly primary care patients. J Am Geriatr Soc 1994; 42:833–838.

4. Bingefors

KAL

Isacson

DGL

von Knorring

Smedby

. Prescription drugs and healthcare use among Swedish patients treated with antidepressants. Ann Pharmacother 1995; 29:566–572.

5. Badger

. Depression, physical health impairment and service use among older adults. Public Health Nurs 1998; 15:136–145.

6. Brown

Salive

Guralnik

Pahor

Chapman

Blazer

. Antidepressant use in the elderly: association with demographic characteristics, health-related factors, and health care utilization. J Clin Epidemiol 1995; 48:445–453.

7. Huang

Cornoni-Huntley

Hays

Huntley

Galanos

Blazer

. Impact of depressive symptoms on hospitalization risk in community-dwelling older persons. J Am Geriatr Soc 2000; 48:1279–1284.

8. Lee

Conwell

Shah

Barker

Delavan

Friedman

. Major depression and emergency medical services utilization in community-dwelling elderly persons with disabilities. Int J Geriatr Psychiatry 2008; 23:1276–1282.

9. Peytremann-Bridevaux

Voellinger

Santos-Eggimann

. Healthcare and preventive services utilization of elderly Europeans with depressive symptoms. J Affect Disorders 2008; 10:247–252.

10.

10. Beekman

ATF

Penninx

BWJH

Deeg

DJH

deBeurs

Geerlings

van Tilburg

. The impact of depression on the well-being, disability, and use of services in older adults: a longitudinal perspective. Acta Psychiatr Scand 2002; 105:20–27.

11.

11. Roelands

van Oyen

Depoorter

Baro

van Oost

. Are cognitive impairment and depressive mood associated with increased service utilization in community dwelling elderly people? Health Soc Care Community 2003; 11:1–9.

12.

12. Luber

Meyers

Williams-Russo

. Depression and service utilization in elderly primary care patients. Am J Geriatr Psychiatry 2001; 9:169–176.

13.

13. Fischer

Wei

Rolnick

. Geriatric depression, antidepressant treatment, and healthcare utilization in a health maintenance organisation. Am J Geriatr Soc 2002; 50:307–312.

14.

14. Unützer

Patrick

Marmon

Simon

Katon

. Depressive symptoms and mortality in a prospective study of 2 558 older adults. Am J Geriatr Psychiatry 2002; 10:521–530.

15.

15. Pulska

Pahkala

Laippala

Kivelä

S-L

. Six-year survival of depressed elderly Finns: a community study. Int J Geriat Psychiatry 1997; 12:942–950.

16.

16. St John

Montgomery

. Do depressive symptoms predict mortality in older people? Aging Ment Health 2009; 13:674–681.

17.

17. Penninx

BWJH

Geerlings

Deeg

DJH

van Eijk

JTM

van Tilburg

Beekman

ATF

. Minor and major depression and the risk of death in older persons. Arch Gen Psychiatry 1999; 56:889–895.

18.

18. Engedal

. Mortality in the elderly: a 3-year follow-up of an elderly community sample. Int J Geriat Psychiatry 1996; 11:467–471.

19.

19. Schoevers

Geerlings

Beekman

ATF

. Association of depression and gender with mortality in old age. Results from the Amsterdam Study of the Elderly (AMSTEL). Br J Psychiatry 2000; 177:336–342.

20.

20. Pulska

Pahkala

Laippala

Kivelä

S-L

. Major depression as a predictor of premature deaths in elderly people in Finland: a community study. Acta Psychiatr Scand 1998; 97:408–411.

21.

21. Schulz

Beach

Ives

Martire

Ariyo

Kop

. Association between depression and mortality in older adults. Arch Intern Med 2000; 160:1761–1768.

22.

22. Schoevers

Geerlings

Deeg

DJH

Holwerda

Jonker

Beekman

. Depression and excess mortality: evidence for a dose response relation in community living elderly. Int J Geriatr Psychiatry 2009; 24:68–76.

23.

23. Licht-Strunk

Van Marwijk

Hoekstra

Twisk

De Haan

Beekman

. Outcome of depression in later life in primary care: longitudinal cohort study with three years follow-up. BMJ 2009; 338:a3079.

24.

24. Crabb

Hunsley

. Utilization of mental health care services among older adults with depression. J Clin Psychol 2006; 62: 299–312.

25.

25. Cole

McCusker

Sewitch

Ciampi

Dyachenko

. Health services use for mental health problems by community-living seniors with depression. Int J Psychogeriatr 2008; 20:554–570.

26.

26. Crabb

. Utilization of mental health services among older adults with depression in Canada. Diss Abstr Int 2009; 70:3165.

27.

27. Valenstein

Taylor

Austin

Kales

McCarthy

Blow

. Benzodiazepine use among depressed patients treated in mental health settings. Am J Psychiatry 2004; 161:654–661.

28.

28. Jorm

Grayson

Creasey

Waite

Broe

. Long-term benzodiazepine use by elderly people living in the community. Aust N Z J Public Health 2000; 24:7–10.

29.

29. Rokke

Klenow

. Prevalence of depressive symptoms among rural elderly: Examining the need for mental health services. Psychotherapy 1998; 35:545–558.

30.

30. Djernes

Kvist

Gulmann

. Prævalens af depressionssygdom blandt svage ældre, vurderet ved gerontopsykiater, praktiserende læger og geriater [The prevalence of depressive disorders in the frail elderly: evaluation by GPs, a consultant psychogeriatrician, and a geriatrician]. Ugeskr Laeger 2000; 162:2330–2334.

31.

31. World Health Organization. The ICD-10 classification of mental and behavioral disorders. Diagnostic criteria for research. Geneva: WHO; 1992.

32.

32. Djernes

Gulmann

Ibsen

Foldager

Olesen

Munk-Jørgensen

. A follow-up of elderly depressed patients. Nord J Psychiatry 2008; 62:233–241.

33.

33. Pedersen

Gotzsche

Moller

Mortensen

. The Danish Civil Registration System. A cohort of eight million persons. Dan Med Bull 2006; 53:441–449.

34.

34. Andersen

Madsen

Jorgensen

Mellemkjoer

Olsen

. The Danish National Hospital Register. A valuable source of data for modern health sciences. Dan Med Bull 1999; 46:263–268.

35.

35. Munk-Jorgensen

Mortensen

. The Danish Psychiatric Central Register. Dan Med Bull 1997; 44:82–84.

36.

36. Olivarius

Hollnagel

Krasnik

Pedersen

Thorsen

. The Danish National Health Service Register. a tool for primary health care research. Dan Med Bull 1997; 44:449–453.

37.

37. Gaist

Sorensen

Hallas

. The Danish prescription registries. Dan Med Bull 1997; 44:445–448.

38.

38. Frank

. Epidemiology. When an entire country is a cohort. Science 2000; 287:2398–2399.

39.

39. Juel

Helweg-Larsen

. The Danish registers of causes of death. Dan Med Bull 1999; 46:354–357.

40.

40. Cullagh

Nelder

. Generalized linear models, second. London: Chapman & Hall/CRC, 1989.

41.

41. Liang

Zeger

. Longitudinal data analysis using generalized linear models. Biometrika 1986; 73:13–22.

42.

42. Bender

Lange

. Adjusting for multiple testing – when and how? J Clin Epidemiology 2001; 54;343–349.

43.

43. Friedhoff

. Consensus Development Conference statement: diagnosis and treatment of depression in late life. JAMA 1992; 268:1018–1024.

44.

44. Djernes

Gulmann

Abelskov

Juul-Nielsen

Sørensen

. Symptomprofil ved indlæggelseskrævende depression hos ældre, relateret til kliniske undergrupper: en genvej til forbedret diagnostik [A symptom profile in major geriatric depression inpatients related to clinical presentations of the disorder: a way of ameliorating case identification]. Ugeskr Læger 1998; 160:6820–6823.

13 Year Follow Up of Morbidity,Mortality and Use of Health Services Among Elderly Depressed Patients and General Elderly Populations

Abstract

Keywords

Materials and methods

Materials

Methods

Overall study methodology

Periods of investigation

Data

Variables

Statistics

Comparison groups

Follow-up period

Confounders

Analyses/models

Analysis/model for the prescription data

Multiple testing

Results

Discussion

Main results

Strengths of the study

Limitations of the study

Summary of results and conclusion

Footnotes

Acknowledgements

References