The association between depression and mortality – a comparison of survey- and register-based measures of depression

Journal of Affective Disorders 210 (2017) 111–114

Contents lists available at ScienceDirect

Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad

The association between depression and mortality – a comparison of survey- and register-based measures of depression

MARK

⁎

Gunhild Tidemann Christensena,b,c, , Solvej Maartenssonb,d, Merete Oslera,b,c a

Department of Public Health, Section of Social Medicine, University of Copenhagen, Denmark Research Centre for Prevention and Health, Rigshospitalet –Glostrup, Copenhagen University, Denmark c Danish Aging Research Center, Department of Public Health, University of Southern Denmark, Denmark d Competence Centre for Dual Diagnosis, Psychiatric Center Sct. Hans, Roskilde, Denmark b

A R T I C L E I N F O

A BS T RAC T

Keywords: Depression diagnosis Self-reported depression Mortality Cohort study

Background: A number of studies have associated depression with a high mortality risk. However, in surveys, depression is often measured by self-reports in selected sub-samples, while register studies have been based on hospital diagnosis or purchase of antidepressants. We examined how different survey- and register-based measures of depression were associated with 7-year mortality in a cohort of middle-aged Danish men. Methods: The study was based on 10,517 men born in 1953. Depression was assessed through hospital diagnosis for the period from 1969 to 2004 and by self-reported information on depression, use of antidepressants and the Major Depression Inventory (MDI) from a survey in 2004, in which 58.8% (n=6292) of the men participated. Information on mortality and cause of death was retrieved from registers for the period between 2004 and 2011. Results: Depression diagnosis from hospital registers as well as self-reported depression, use of antidepressants and having a high MDI-score were significantly associated with mortality from all, natural and unnatural causes. The associations were of a similar magnitude for the register-based measure of depression and for the surveybased measures reflecting past depression, but the strongest association was found for current depression as assessed by the MDI-score. Limitations: The study population consists almost exclusively of white men and the findings may not be generalizable to female populations or other races and ethnicities. Conclusions: Physicians should be aware of male patients with a history of depression from hospital records or self-reported as they have higher mortality risk years after their first symptom.

1. Introduction The relationship between major depression and excess mortality is well-established. Both older and more recent reviews have provided relative risks (RR) for mortality of 1.50 among people with depression (Harris and Barraclough, 1998; Cuijpers et al., 2014a) with the highest risk in men (RR=2.04; 95%CI 1.76–2.37) (Cuijpers et al., 2014b). Most studies have dealt with the issue of mortality and depression diagnosed in clinical settings, while fewer have studied whether depressive symptoms influence mortality in population based samples (Cuijpers et al., 2013). Depression rating scales are the preferred method to measure depression in surveys, but it is also possible to ask directly whether a person has had a depression in the past or is currently diagnosed with depression. However, surveys are most often subject to

selective non-participation and may be costly for large samples. Consequently, a number of more recent studies have used population-based register information, such as hospitalizations with diagnosis of depression or redemption of antidepressants as indicators of depression. Although these indicators address the above limitations they may not capture cases that remain untreated (Thielen et al., 2009). No study seems to have examined how well different self-reported and register-based measures of depression associate with mortality in the same population. The aim of the present study was to examine how a register-based depression diagnosis, self-reported depression, and a depression rating scale associate with all-cause mortality as well as mortality from natural and unnatural causes of death in a sample of middle-aged men.

Abbreviations:MDI, Major Depression Inventory ⁎ Correspondence to: Section of Social Medicine, Department of Public Health, Øster Farimagsgade 5, PO Box 2099, 1014 København K, Denmark. E-mail address: [email protected] (G.T. Christensen). http://dx.doi.org/10.1016/j.jad.2016.12.024 Received 17 June 2016; Received in revised form 11 October 2016; Accepted 17 December 2016 Available online 20 December 2016 0165-0327/ © 2016 Elsevier B.V. All rights reserved.

Journal of Affective Disorders 210 (2017) 111–114

G.T. Christensen et al.

Table 1 The distribution of register- and survey-based measures of depression in 2004 in relation to subsequent 7-year mortality (natural and unnatural causes) and covariables for 10,517 Danish men born 1953. Measure of depression

Number

Covariables % DM

% Other longstanding illness

% BMI≥30

% Low social class

% Smokers

% ≥35 Drinks/ week

% Low leisure time activity

Hospital diagnosis of depression Total cohort 10,517 611(410/30) No 10,263 574(389/25) Yes 254 37 (26/4)

– – –

– – –

– – –

– – –

– – –

– – –

– – –

Non-responders in 2004 survey 4225 356(244/18)

–

–

–

–

–

–

–

337(23014) 19(16/2)

– –

– –

– –

– –

– –

– –

– –

255(170/12)

4.8

30.0

12.7

18.3

42.4

15.0

17.7

237(159/11) 18(19/2)

2.8 4.7

29.7 49.6

12.3 15.7

18.0 29.8

42.2 57.1

14.9 19.1

17.7 19.8

187(128/6) 63(39/6)

4.4 7.5

27.5 49.4

12.4 14.7

16.8 28.9

40.6 56.8

13.7 24.4

16.8 23.9

Self-reported age of depression onset No depression 5528 187(128/2) Before age 40 278 25 (16/2) Age 40 or later 383 28 (17/4)

4.4 8.6 7.1

27.5 54.3 47.9

12.4 13.7 17.1

16.8 28.8 29.4

40.6 59.0 55.6

13.7 25.2 23.0

16.8 25.1 23.4

Self-reported antidepressant treatment No, never 5590 179(122/6) Yes, ≥3 years ago 268 23 (17/1) Yes, < 3 years ago 150 11(7/0) Yes, on-going 233 33(18/4)

4.4 6.7 6.0 7.6

27.7 45.0 46.5 57.1

12.4 11.8 14.5 16.7

16.8 28.0 35.4 29.6

40.6 55.0 56.7 60.8

13.7 18.7 28.7 28.7

17.0 21.9 20.1 26.6

MDI-score 0–24 25+

4.5 12.9

28.9 63.2

12.6 13.0

17.3 41.8

41.5 67.9

14.1 36.4

16.6 45.2

Total

Deaths (n/ un)

Register-based measure of depression

No Yes

4097 128

Participants in 2004 survey 6292 No Yes

6166 126

Survey-based measures of depression Self-reported depression No 5531 Yes 730

6022 217

204(139/33) 42(10/2)

Abbreviations: n=natural causes of death; un=unnatural causes of death; DM=Diabetes Mellitus; BMI=Body Mass Index; MDI=Major Depression Inventory.

Survey-based measures of depression: in September 2004, 6292 (59.8%) of the 10,517 cohort members responded to a mailed health questionnaire. The survey included measures of Self-reported depression based on positive answers to the question: “Has a doctor ever told you that you suffer from depression?”. Subjects with a positive response were also asked “how old were you when the disease occurred for the first time?” and “Do you or have you ever taken medicine for depression” with the following four categories: 1. Yes, I take this medicine at present, 2. Yes, during the last 3 years, 3. Yes, more than 3 years ago and 4. No, never. The survey also included the Major Depression Inventory (MDI), which was used to measure frequency of depressive symptoms (Olsen et al., 2003). The MDI is a frequently used and well-validated self-report screening tool of depressive symptoms, which consists of 12 items assessing the frequency of depressive symptoms during the past two weeks. Each item was rated on a six point scale ranging from zero (not at all) to five (all the time). Ratings were summed to calculate a total score which can range from 0 to 50. Based on a previous study of diagnostic validity a cut-off of 25 points or more was defined as major depression (Bech et al., 2001).

2. Methods 2.1. Study population The Metropolit cohort is defined as the 11,532 men, born in 1953 in the Copenhagen Metropolitan area, who were living in Denmark in 1968 (Osler et al., 2006). The study population of the present study consists of the 10,517 men from the Metropolit cohort, who were still alive in September 2004, at which time they were invited to participate in a postal questionnaire survey.

2.2. Measures of depression Register-based measure of depression: All 10,517 men were followed through register linkages to the Danish Psychiatric Central Registry from 1969 until September 2004 for time of admission to psychiatric wards and diagnosis on discharge. Diagnoses from the hospital register were classified according to the 8th Revision of the International Classification of Diseases (ICD8) for the years from 1969 to 1993, and the 10th revision (ICD10) from 1994. The diagnoses included for depression in the present study were ICD-8: 296.0, 296.2, 296.3, 298.0, 300.4 and ICD-10: F31.3, F31.4, F31.5, F31.6, F38, F32, F33, F34.1. We also recorded year of first hospital admission.

2.3. Ascertainment of mortality All cohort members were followed for all-cause mortality in the 112

Journal of Affective Disorders 210 (2017) 111–114

G.T. Christensen et al.

adjusted model for those having an MDI-score of 25 or higher (HR=3.11(95%CI 2.07–4.66). For self-reported age of onset of depression the association was no longer significant for those who were 40 years or older at onset of depression (HR=1.30(95%CI 0.82–2.06) in the fully adjusted model. Nearly similar patterns of associations were seen for natural and unnatural causes of death. The association between depression and mortality seemed to be strongest for unnatural deaths, but the estimates were very imprecisely estimated due to the small number of cases.

Civil Registration System as well as death from natural and unnatural (suicides, homicides and accidents) causes in the Danish Register of Cause of Death for the period from September 2004 until December 2011. 2.4. Covariables Information on father's social class was available from the national birth register. Self-reported information on other potential confounding variables was available for those cohort members who responded to the questionnaire: Socioeconomic position, lifestyle (smoking, alcohol use and leisure-time activity), information on diabetes and other longstanding illnesses, and height and weight. Body Mass Index (BMI) was calculated by dividing weight in kg by height in meters squared.

4. Discussion This population-based study among middle-aged Danish men compared register- and survey-based measures derived from the same population. The prevalence of depression was almost 7 times higher when based on self-report rather than information from registers. However, similar associations were found for the register-based measure and the survey-based measures reflecting past depression whereas a stronger association was found between depression assessed by the MDI-score and mortality. This stronger association was not surprising given that the MDI-score reflects current depression and thus might have a stronger effect on subsequent mortality than past depressions. The used cut-point for the MDI-score was assumed to reflect major depression (Bech et al., 2001) and this was further supported as a high MDI score was more frequent among subjects with a self-reported depression diagnosis (yes 17.6% versus no 1.6%) or in on-going treatment with antidepressants (yes 26.8% versus never 1.6%). The statistically significant associations between register- and survey-based measures of depression and mortality from all, natural and unnatural causes are in agreement with a number of previous studies (Harris and Barraclough, 1998; Cuijpers et al., 2014a, 2014b, 2013). No other study has compared the association between registerand survey-based measures of depression and mortality. However, one study has examined how different survey-based measures of depression associate with mortality. In contrast to the results of our study, where we found that depression was associated with increased mortality irrespective of the measure used, a study of598 Irish patients with acute coronary syndrome found a higher risk of mortality during 8 years of follow-up (HR=2.64(1.38–5.35)) for those with a score of 7 or higher on the Hospital Anxiety and Depression scale, while the Beck Depression Inventory was not associated with mortality (Doyle et al., 2012).

2.5. Statistical analysis The association between indicators of depression measured in 2004 and mortality from September 2004–2011 was analysed in Cox regression models. Person years of follow-up were accumulated from age at time of study entry (September 2004) and follow-up was terminated at age of death, emigration, or end of follow–up (December 2011), whichever came first. All estimates were first adjusted for fathers social class (from birth registers) (Osler et al., 2006), while those for survey participants were adjusted for own social class, self-reported comorbidity, BMI, heavy alcohol use, smoking, and sedentary leisure-time activity at the time of the survey. The study protocol was approved by the Danish Data Protection Agency. 3. Results In the period from 1969 to 2004, 2.0% of those who responded to the questionnaire and 3.1% of non-responders were registered with a depression diagnosis in hospital registers. In the survey, 13.2% of the men reported that they had at least once had a depression diagnosis, while 3.6% was currently having a major depression according to the MDI depression scale. Men with previous or on-going depression seemed to have more comorbidity and less healthy lifestyle compared to men with no depression history (Table 1). During the 7 years of follow-up, 611 (5.8%) of the 10,517 men died. Mortality was higher for men who did not participate in the survey than among participants (8.4% versus 4.0%). Survival analyses revealed a significantly higher risk of mortality for men who had a register-based hospital diagnosis of depression compared to those with no hospital diagnosis (Table 2). The association was slightly stronger for those who participated in the survey compared to those who did not. Adjusting for father's social class did not change the estimates. For the men who participated in the 2004 survey those with a register-based hospital diagnosis of depression had an almost 4 times higher risk of mortality (Hazard Ratio (HR)=3.95(95%CI 2.44– 6.37)), as compared to men with no depression diagnosis. The elevated risk was seen for depressions diagnosed before (HR=3.02 (1.81– 5.05);n=95) and after age 40 (HR=2.58(95%CI 1.68–3.95);n=159) (data not shown). The association between hospital diagnosed depression and mortality was attenuated when adjusting for the survey measures of own social class, comorbidity, BMI and lifestyle (HR=2.80 (95%CI 1.65–4.76)). The survey-based depression measures were also associated with increased mortality with the strongest association found for those having an MDI score of 25 or higher (HR=6.45(95%CI 4.63–8.99). The estimates were not influenced by adjustment for father's social class, but were somewhat attenuated after adjustment for own social class, comorbidity, BMI and lifestyle and for depression assessed by the MDIscore the association with mortality was reduced from an almost 7 times higher risk in the crude model to a 3 times higher risk in the fully

5. Limitations The present study only includes men and the results may not be generalizable to female populations due to the about two-fold higher prevalence of depression among women compared to men (Noble, 2005). Furthermore, women and men might differ in the way they respond to questions on depression and also in how they seek treatment for depression, and this could affect the association between depression and mortality and might also affect the comparison of register- and survey-based measures of depression. In addition, the Metropolit cohort consists almost exclusively of white men, and the results may not be generalizable to other races or ethnicities. Finally, Denmark has a universal health care system with free access to health care for all Danish citizens and the results might not be generalizable to populations that differ in terms of access to and utilization of health care. 6. Conclusion No studies have, as ours, examined how both register- and surveybased depression measures associate with mortality. The results from the present study suggest that the association between depression and 113

Journal of Affective Disorders 210 (2017) 111–114

G.T. Christensen et al.

Table 2 Associations (Hazard Ratios (HR) and 95% Confidence Intervals) between register- and survey-based measures of depression in 2004 and subsequent 7-year mortality for 10,517 Danish men born 1953. Measure of depression

All-causes HRcrude

Natural causes

Unnatural causes

HRadjusteda

HRadjustedb

HRcrude

HRcrude

Register-based measure of depression Hospital diagnosis of depression Total cohort No Yes

1 2.75(1.97–3.83)

1 2.73(1.96–3.81)

– –

1 2.72(1.81–4.08)

1 6.54(2.28–18.80)

Non-responders in 2004 survey No Yes

1 1.83(1.15–2.91)

1 1.83(1.15–2.91)

– –

1 1.98(1.15–3.40)

1 4.11(0.94–17.98)

Participants in 2004 survey No Yes

1 3.95(2.44–6.37)

1 3.98(2.47–6.43)

1 2.80(1.65–4.76)

1 3.54(1.92–6.52)

1 10.20(2.23–46.55)

1 2.63(4.63–8.94)

1 2.62(1.97–3.49)

1 1.50(1.37–2.62)

1 2.36(1.65–3.38)

1 7.75(2.50–24.03)

Self-reported age of depression onset No depression 1 Before age 40 2.74(1.80–4.16) Age 40 or later 2.20(1.48–3.28)

1 2.65(1.75–4.03) 2.24(1.50–3.34)

1 1.72(1.09–1.72) 1.30(0.82–2.06)

1 2.55(1.51–4.28) 1.94(1.17–3.22)

1 6.77(1.37–33.66) 9.77(2.75–34.62)

Self-reported antidepressant No, never Yes, ≥3 years ago Yes, < 3 years ago Yes, on-going

treatment 1 2.77(1.80–4.29) 2.33(1.26–4.25) 4.68(3.29–6.78)

1 2.78(1.80–4.29) 2.40(0.30–4.41) 4.65(3.20–6.74)

1 1.38(0.68–2.82) 2.06(1.29–3.30) 2.23(1.42–3.51)

1 2.98(1.79–4.95) 2.16(1.01–4.64) 3.68(2.24–6.04)

1 2.28(0.27–18.91) – 16.64(4.69–58.96)

MDI-score 0–24 25+

1 6.45(4.63–8.99)

1 6.28(4.58–8.76)

1 3.11(2.07–4.66)

1 7.75(5.29–1.36)

1 6.08(1.33–27.78)

Survey-based measures of depression Self-reported depression No Yes

Abbreviations: HR= Hazard ratio; MDI= Major Depression Inventory. a The analyses were adjusted for father's social class. b The analyses were adjusted for own social class, diabetes, other longstanding illness, BMI, heavy alcohol use, smoking and low leisure-time activity. studies that measured both. Br. J. Psychiatry 202 (1), 22–27. Cuijpers, P., Vogelzangs, N., Twisk, J., Kleiboer, A., Li, J., Penninx, B.W., 2014a. Comprehensive meta-analysis of excess mortality in depression in the general community versus patients with specific illnesses. Am. J. Psychiatry 171 (4), 453–462. Cuijpers, P., Vogelzangs, N., Twisk, J., Kleiboer, A., Li, J., Penninx, B.W., 2014b. Is excess mortality higher in depressed men than in depressed women? A meta-analytic comparison. J. Affect. Disord. 161, 47–54. Doyle, F., Conroy, R., McGee, H., 2012. Differential predictive value of depressive versus anxiety symptoms in the prediction of 8-year mortality after acute coronary syndrome. Psychosom. Med. 74 (7), 711–716. Harris, E.C., Barraclough, B., 1998. Excess mortality of mental disorder. Br. J. Psychiatry 173, 11–53. Noble, R.E., 2005. Depression in women. Metabolism 54 (5 Suppl 1), 49–52. Olsen, L.R., Jensen, D.V., Noerholm, V., Martiny, K., Bech, P., 2003. The internal and external validity of the major depression inventory in measuring severity of depressive states. Psychol. Med. 33 (2), 351–356. Osler, M., Lund, R., Kriegbaum, M., Christensen, U., Andersen, A.M., 2006. Cohort profile: the Metropolit 1953 Danish male birth cohort. Int J. Epidemiol. 35 (3), 541–545. Thielen, K., Nygaard, E., Andersen, I., Rugulies, R., Heinesen, E., Bech, P., et al., 2009. Misclassification and the use of register-based indicators for depression. Acta Psychiatr. Scand. 119 (4), 312–319.

mortality is quite similar for register- and survey-based depression measures, when the survey-based measures reflect past depression. Our findings also suggest that physicians should be aware of patients with a history of depression from medical records or self-report as they have a higher mortality risk years after their first symptom. Acknowledgements We thank all those who initiated and/or continued the Metropolit study: K Svalastoga, E Høgh, P Wolf, T Rishøj, G Strande-Sørensen, E Manniche, B Holten, IA Weibull and A Ortman. References Bech, P., Rasmussen, N.A., Olsen, L.R., Noerholm, V., Abildgaard, W., 2001. The sensitivity and specificity of the major depression inventory, using the present state examination as the index of diagnostic validity. J. Affect. Disord. 66 (2–3), 159–164. Cuijpers, P., Vogelzangs, N., Twisk, J., Kleiboer, A., Li, J., Penninx, B.W., 2013. Differential mortality rates in major and subthreshold depression: meta-analysis of

114