Elsevier Editorial System(tm) for International Journal of Nursing Studies Manuscript Draft
Manuscript Number: IJNS-D-10-00676R3
Title: Diabetes mellitus and risk of subsequent depression: A longitudinal study Article Type: Research Paper
Keywords: depression; diabetes mellitus; incidence; longitudinal studies. Corresponding Author: Dr. pei-chun Chen,
Corresponding Author's Institution: Chinal Medical University and Hospital First Author: Yi-Min Hsu
Order of Authors: Yi-Min Hsu; Li-Ting Su; Hui-Mei Chang; Fung-Chang Sung; Shu-Yu Lyu; pei-chun Chen
Abstract: Background: Findings of previous studies on the association between diabetes and the risk of depression are contradictory. Furthermore, much less is known concerning the association among young adults.
Objective: To investigate whether diabetes is associated with an increased risk of subsequent development of depression, with emphasis on age-specific variations.
Design: A cohort study.
Setting: Claims data of one million subjects randomly selected from 23 million people covered by the Taiwan National Health Insurance program.
Participants: From the claims data, we identified 14,048 patients aged ≥ 20 years with newly diagnosed diabetes in 2000~2002 and randomly selected 55,608 non-diabetic subjects for comparison, that were frequency-matched by calendar year, age, and gender. Incidence rates of depression to the end of 2007 were identified, and risks were compared between the two groups. Results: The incidence of depression was 1.80-times higher in the diabetic group than in nondiabetic subjects over a median follow-up of 6.5 years (adjusted hazard ratio [HR] = 1.46, 95% confidence interval [CI]: 1.24~1.71). Age-specific HRs for incidence of depression in relation to diabetes were not statistically different between the patient subgroups aged 20-39, 40-49, 50-59, 60-69 and ≥ 70 years (p value for age-diabetes interaction=0.33). Stratified analyses showed that the association was much stronger for subjects without comorbid cardiovascular disease than for those with this comorbidity. Insulin treatment was associated with a 43% reduced risk of depression in diabetic patients. Conclusions: In this population-based study, diabetic patients were at a higher risk for subsequent depression. Adequate treatment reduced the risk.
Response to Reviewers: EDITORIAL COMMENTS
Thank you for your comprehensive response to the reviewers comments and further analysis and interpretation of the data from your study. There are still some aspects of the data analysis and presentation of the results that are unclear to the reader and would benefit from further clarification. Reviewer #4: Review for IJNS10-00676R2
I am not satisfied with how the authors have described the null interaction term with age in the text - in the response they state that they tested that there was NO SIGNIFICANT INTERACTION between age
and diabetes status, and yet in the text there is NO MENTION of this null result, and instead they continue to describe the findings as though there is a diminishing effect with age (the interaction term says this is not true). The statements "In unadjusted and partially adjusted models the association between diabetes and risk of depression DECREASED slightly with aging." and "test for age-diabetes interaction, 0.33") are INCOMPATIBLE. Only the latter statement is correct - there is NO EVIDENCE THAT THE RELATIONSHIP between depression and diabetes varies with age.
The authors need to state explicitly in the METHODS that they test for the diabetes-age interaction. The abstract, results & discussion need to be re-written to properly state what they actually found (that is, no effect of age) and eliminate statement such as "decreased slightly", "diminished with age", "more pronounced" - these are not true statements according to the interaction term.
Reply:
Thank you for your suggestions. We have withdrawn all statements with regard to differences in diabetes-depression association among age groups in this revision and re-written the relevant descriptions. The statistical methods used to test for interaction effect are also described in the “methods” section.
In ABSTRACT: Deleted-
The age-specific risk of depression decreased slightly with aging, as the significance diminished for those aged ≥ 70 years.
Added-
Age-specific HRs for incidence of depression in relation to diabetes were not statistically different between the patient subgroups aged 20-39, 40-49, 50-59, 60-69 and ≥ 70 years (p value for age-diabetes interaction=0.33). (please see page 1, the last 2 lines, and page 2, line 1, in the revision). In “What this paper adds” box:
Deleted-
The risk of depression associated with diabetes diminishes in patients ages ≥ 70 years. Added-
Diabetes was associated with increased risk of subsequent depression, and the association did not vary by age. (please see page 3 in the revision)
In METHODS: Deleted-
We also performed regression analyses among patient subgroups of CVD to assess whether there was an effect of an interaction between comorbid CVD and diabetes.
Added-
To assess interaction effect between age and diabetes, we performed the likelihood ratio test comparing models with and without interaction terms between the 5 age groups and diabetes. The interaction effect between comorbid CVD and diabetes was also examined. (please see page 9, lines 4-7 from the last line)
In RESULTS: Deleted-
The age-stratified analysis showed that among all age groups of < 70 years, diabetes was significantly associated with a greater risk of developing depression. In the unadjusted and partially adjusted models, the association between diabetes and risk of depression decreased slightly with aging, from an HR of 1.98 for the 20~29-year group to 1.84 for the 60~69-year group and 1.23 for the oldest group (model 1; test for age-diabetes interaction, p=0.33). In the model further adjusted for cardiovascular comorbidity, the HR became statistically insignificantly in patients aged 20~39 years. We found no
statistically significant association between diabetes and subsequent depression among subjects ≥ 70 years old in any model.
Added-
We found no significant interaction between age and diabetes on the risk of subsequent depression (likelihood ratio test for age-diabetes interaction in model 2, p=0.33), which indicated the estimated HRs for the diabetes-depression association were not significantly different between age groups. (please see age 11, lines 2-5, in the revision)
Deleted-
Diabetic subjects aged 20~39 and 40~49 years had respective HRs of 1.93 (95% CI: 1.10~3.40) and 1.85 (95% CI: 1.16~2.95), and the risk was weaker for those aged ≥ 70 years (HR: 1.58, 95% CI: 0.36~6.94).
Added-
The corresponding figures for patients aged 20~39, 40~49 and ≥ 70 years were 1.93 (95% CI:
1.10~3.40), 1.85 (95% CI: 1.16~2.95), and 1.58 (95% CI: 0.36~6.94), respectively. But the differences in these five age-specific HRs were not statistically significant (The p-value for age-diabetes
interaction, 0.48). (please see page 11, paragraph 2, last 4 lines)
In DISCUSSION: Deleted-
The association was more pronounced in persons who were aged < 70 years and those without comorbid CVD.
Added-
No significant age-specific variation was observed in the association. (please see page 12, paragraph 2, the last 2 lines)
Deleted-
This study revealed an appreciably increased risk of developing depression associated with diabetes, particularly for younger patients. People in this age group with diabetes could most effectively benefit from screening for depression, as they are at higher risk and have long-term economic and social burdens due to concurrent diabetes and depression.
Added-
This study revealed an appreciably increased risk of developing depression associated with diabetes, and the association did not vary by age. (please see page 16, lines 3-4)
IJNS AUTHOR CHECKLIST Authors of all papers should submit this checklist together with their manuscript. The checklist will be made available during the submission process online to all authors and full step-by-step guidance given.
Part 1 identifies basic requirements for the manuscript submission (mandatory for all submissions)
Part 2 identifies recognized guidelines for scientific reporting, which you should use to prepare your manuscript (required for systematic reviews and original research)
Part 3 Is a self assessment checklist that is designed to help to ensure that your research or review manuscript meets basic standards and the journal’s Guide for Authors. (optional only)
PART 1 Basic requirements
Author response or further detail Tick
Word count 258 in the abstract; 2483 in the text Was ethical approval
given and by whom? (give any reference number)
This study used claims data containing scrambled personal identifications. Institutional review board approval was exempted for conducting this study.
Please state any conflicts of interest
None.
Please state sources of funding and the role of funders in the conduct of the research
This study was supported partly by the National Science Council, Executive Yuan, Taiwan, Republic of China (NSC 97-2625-M-039-003), China Medical University Hospital (grant number 1MS1) and Taiwan Department of Health Clinical Trial and Research Center for Excellence TD-B-111-004) and Cancer Research Center of Excellence (DOH99-TD-C-111-005). The funding sources had no involvement in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication.
Please state any study registry number (e.g. ISRCTN)
Not applicable.
Title The title is in the format ‘Topic / question: design/type of paper’ and identifies the population / care setting studied.
(e.g. The effectiveness of telephone support for adolescents with insulin dependant diabetes: controlled before and after study: the structure is optional for discussion papers, editorials and commentaries)
X
Abstract A structured abstract appropriate to the design (see guidelines for authors). Reports of controlled trials should follow the CONSORT format (does not apply to editorials or commentaries, Abstracts for discussion papers need not be structured)
X
Key words Between four and six key words have been provided in alphabetical order, which accurately identify the paper's subject, purpose, method and focus. Use the Medical Subject Headings (MeSH®) thesaurus or Cumulative Index to Nursing and Allied Health (CINAHL) headings where possible (see http://www.nlm.nih.gov/mesh/meshhome.html).
X
What the paper adds Bullet points have been included that identify existing research knowledge relating to the specific research question / topic (what is already known) and a summary of the new knowledge added by this study (see Guide for Authors, does not apply to editorials or commentaries)
X
References Citations accord to the journal’s format (Author, date) and reference list includes full details of all cited references in the proper format and alphabetical order (see Guide for Authors)
X
Other Published accounts
All published and in press accounts of the study from which data in this paper originate are referred to in the paper and the relationship between this and other publications from the same study is made clear(see Guide for Authors)
X Author Checklist
The study is referred to by a distinctive name which will be used in any future publications to identify that it as the same study.
Please upload copies of all previous, current and under review publications from this study and / or give full details below
PART 2 Standards of reporting
The editors require that manuscripts adhere to recognized reporting guidelines relevant to the research design used. These identify matters that should be addressed in your paper. Please indicate which guidelines you have referred to.
These are not quality assessment frameworks and your study need not meet all the criteria implied in the reporting guideline to be worthy of publication in the IJNS. The checklists do identify essential matters that should be considered and reported upon. For example, a controlled trial may or may not be blinded but it is important that the paper identifies whether or not participants, clinicians and outcome assessors were aware of treatment assignments.
**You are encouraged (although not required) to submit a checklist from the appropriate reporting guideline together with your paper as a guide to the editors and reviewers of your paper.
Reporting guidelines endorsed by the IJNS are listed below:
G ui de li ne r e fe rr e d to Che c k li s t s ub m itte d ** Observational cohort, case control and cross sectional studies
STROBE Strengthening the Reporting of Observational Studies in Epidemiology http://www.equator-network.org/index.aspx?o=1032
X
Quasi experimental / non-randomized evaluations
TREND - Transparent Reportingof Evaluations with Non-randomized Designs http://www.equator-network.org/index.aspx?o=1032
Randomised (and quasi-randomised) controlled trial
CONSORT – Consolidated Standards of Reporting Trials http://www.equator-network.org/index.aspx?o=1032
Study of Diagnostic accuracy /
assessment scale
STARD Standards for the Reporting of Diagnostic Accuracy studies http://www.equator-network.org/index.aspx?o=1032
Systematic Review of Controlled Trials
PRISMA - Preferred Reporting Items for Systematic Reviews and Meta-Analyses http://www.equator-network.org/index.aspx?o=1032
Systematic Review of Observational Studies
MOOSE Meta-analysis of Observational Studies in Epidemiology http://www.equator-network.org/index.aspx?o=1032
Qualitative researchers might wish to consult the guideline listed below
Qualitative studies COREQ: Consolidated criteria for reporting qualitative research Tong, A., Sainsbury, P., Craig, J., 2007. Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups. International Journal for Quality in Health Care 19 (6), 349-357. (http://dx.doi.org/10.1093/intqhc/mzm042)
Other (please give source)
Not applicable (please elaborate)
Pei-Chun Chen, PhD
Graduate Institute of Clinical Medical Science China Medical University
91 Hsueh-Shih Road Taichung 404, Taiwan September 12, 2011 Sharon McKinley, PhD
Editor
International Journal of Nursing Studies Ms. Ref. No.: IJNS-D-10-00676 (Revision 3)
Title: Diabetes mellitus and risk of subsequent depression: a longitudinal study Dear Prof. McKinley:
Thank you very much for your letter of September 8, 2011 and the reviewer’s comments for the above referred manuscript. We have revised the manuscript and responded to all comments point-by-point. All additions to the manuscript have been put in bold blue. All deletions of the text also have been indicated in the “response to reviewer” document.
All authors have agreed to this submission and no similar paper has been submitted for publication elsewhere. There is no interest confliction involved for this submission.
Thank you again for your review of this revision and consideration for publication in the journal. Sincerely, Pei-Chun Chen, PhD Assistant Professor Tel: 886-4-2205-3366 ext 6119 Fax: 886-4-2201-9901 e-mail: peichun@mail.cmu.edu.tw Cover Letter
Response to reviewers Ms. Ref. No.: IJNS-D-10-00676 (Revision 3)
Title: Diabetes mellitus and risk of subsequent depression: A longitudinal study
EDITORIAL COMMENTS
Thank you for your comprehensive response to the reviewers comments and further analysis and interpretation of the data from your study. There are still some aspects of the data analysis and presentation of the results that are unclear to the reader and would benefit from further clarification.
Reviewer #4: Review for IJNS10-00676R2
I am not satisfied with how the authors have described the null interaction term with age in the text - in the response they state that they tested that there was NO
SIGNIFICANT INTERACTION between age and diabetes status, and yet in the text there is NO MENTION of this null result, and instead they continue to describe the findings as though there is a diminishing effect with age (the interaction term says this is not true). The statements "In unadjusted and partially adjusted models the
association between diabetes and risk of depression DECREASED slightly with aging." and "test for age-diabetes interaction, 0.33") are INCOMPATIBLE. Only the latter statement is correct - there is NO EVIDENCE THAT THE RELATIONSHIP between depression and diabetes varies with age.
The authors need to state explicitly in the METHODS that they test for the diabetes-age interaction.
The abstract, results & discussion need to be re-written to properly state what they actually found (that is, no effect of age) and eliminate statement such as "decreased slightly", "diminished with age", "more pronounced" - these are not true statements according to the interaction term.
Reply:
Thank you for your suggestions. We have withdrawn all statements with regard to differences in diabetes-depression association among age groups in this revision and re-written the relevant descriptions. The statistical methods used to test for interaction effect are also described in the “methods” section.
In ABSTRACT: Deleted-
The age-specific risk of depression decreased slightly with aging, as the significance *Response to Reviewers
2 diminished for those aged ≥ 70 years.
Added-
Age-specific HRs for incidence of depression in relation to diabetes were not statistically different between the patient subgroups aged 20-39, 40-49, 50-59, 60-69 and ≥ 70 years (p value for age-diabetes interaction=0.33). (please see page 1, the last 2 lines, and page 2, line 1, in the revision).
In “What this paper adds” box: Deleted-
The risk of depression associated with diabetes diminishes in patients ages ≥ 70 years.
Added-
Diabetes was associated with increased risk of subsequent depression, and the association did not vary by age. (please see page 3 in the revision)
In METHODS: Deleted-
We also performed regression analyses among patient subgroups of CVD to assess whether there was an effect of an interaction between comorbid CVD and diabetes.
Added-
To assess interaction effect between age and diabetes, we performed the likelihood ratio test comparing models with and without interaction terms between the 5 age groups and diabetes. The interaction effect between comorbid CVD and diabetes was also examined. (please see page 9, lines 4-7 from the last line)
In RESULTS: Deleted-
The age-stratified analysis showed that among all age groups of < 70 years, diabetes was significantly associated with a greater risk of developing depression. In the unadjusted and partially adjusted models, the association between diabetes and risk of depression decreased slightly with aging, from an HR of 1.98 for the 20~29-year group to 1.84 for the 60~69-year group and 1.23 for the oldest group (model 1; test for age-diabetes interaction, p=0.33). In the model further adjusted for
cardiovascular comorbidity, the HR became statistically insignificantly in patients aged 20~39 years. We found no statistically significant association between diabetes and subsequent depression among subjects ≥ 70 years old in any model.
We found no significant interaction between age and diabetes on the risk of subsequent depression (likelihood ratio test for age-diabetes interaction in model 2, p=0.33), which indicated the estimated HRs for the
diabetes-depression association were not significantly different between age groups. (please see age 11, lines 2-5, in the revision)
Deleted-
Diabetic subjects aged 20~39 and 40~49 years had respective HRs of 1.93 (95% CI: 1.10~3.40) and 1.85 (95% CI: 1.16~2.95), and the risk was weaker for those aged ≥ 70 years (HR: 1.58, 95% CI: 0.36~6.94).
Added-
The corresponding figures for patients aged 20~39, 40~49 and ≥ 70 years were 1.93 (95% CI: 1.10~3.40), 1.85 (95% CI: 1.16~2.95), and 1.58 (95% CI:
0.36~6.94), respectively. But the differences in these five age-specific HRs were not statistically significant (The p-value for age-diabetes interaction, 0.48). (please see page 11, paragraph 2, last 4 lines)
In DISCUSSION: Deleted-
The association was more pronounced in persons who were aged < 70 years and those without comorbid CVD.
Added-No significant age-specific variation was observed in the association. (please see page 12, paragraph 2, the last 2 lines)
Deleted-
This study revealed an appreciably increased risk of developing depression
associated with diabetes, particularly for younger patients. People in this age group with diabetes could most effectively benefit from screening for depression, as they are at higher risk and have long-term economic and social burdens due to concurrent diabetes and depression.
Added-This study revealed an appreciably increased risk of developing depression associated with diabetes, and the association did not vary by age. (please see page 16, lines 3-4)
Ms. Ref. No.: IJNS-D-10-00676 (Revision 3)
Title: Diabetes mellitus and risk of subsequent depression: A longitudinal study Yi-Min Hsua, Li-Ting Sub,c,d, Hui-Mei Changa, Fung-Chang Sungc, Shu-Yu Lyue, and Pei-Chun Chenb,f,g
a
Department of Nursing, China Medical University Hospital, Taichung 404, Taiwan
b
Management Office for Health Data, China Medical University Hospital, Taichung 404, Taiwan
c
Department of Public Health, China Medical University, Taichung 404, Taiwan
d
Trauma and Emergency Center, China Medical University Hospital, Taichung 404, Taiwan
e
School of Public Health, Taipei Medical University, Taipei 110, Taiwan
f
Department of Health Risk Management and gGraduate Institute of Clinical Medical Science, China Medical University, Taichung 404, Taiwan
Running head: Diabetes and depression risk
Correspondence :
1. Pei-Chun Chen, China Medical University Graduate Institute of Clinical Medical Science, 91 Hsueh-Shih Road, Taichung 404, Taiwan.
Tel.: +886 4 22053366, ext 6119; fax: +886 4 22339216; E-mail address: peichun@mail.cmu.edu.tw
2. Shu-Yu Lyu, School of Public Health, Taipei Medical University, 250 Wu-Hsing Street Taipei 110, Taiwan.
Tel: 886-2-2736-1661 ext 6518; Fax: 886- 2-2738-4831; E-mail address: sylyu@tmu.edu.tw
Sources of funding:
This study was supported partly by the National Science Council, Executive Yuan, Taiwan, Republic of China (NSC 97-2625-M-039-003), China Medical University Hospital (grant number 1MS1) and Taiwan Department of Health Clinical Trial and Research Center for Excellence (DOH100-TD-B-111-004) and Cancer Research Center of Excellence
(DOH100-TD-C-111-005). The funding sources had no involvement in study design, data collection, analysis and interpretation of data, writing of the report, and the decision to submit the paper for publication.
Conflict of Interest: None declared. *Title Page (with author details and affiliations)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 1
Ms. Ref. No.: IJNS-D-10-00676 (Revision 3)
Title Diabetes mellitus and the risk of subsequent depression: a longitudinal study
Abstract
Background: Findings of previous studies on the association between diabetes and the risk of
depression are contradictory. Furthermore, much less is known concerning the association
among young adults.
Objective: To investigate whether diabetes is associated with an increased risk of subsequent
development of depression, with emphasis on age-specific variations.
Design: A cohort study.
Setting: Claims data of one million subjects randomly selected from 23 million people
covered by the Taiwan National Health Insurance program.
Participants: From the claims data, we identified 14,048 patients aged ≥ 20 years with newly
diagnosed diabetes in 2000~2002 and randomly selected 55,608 non-diabetic subjects for
comparison, that were frequency-matched by calendar year, age, and gender. Incidence rates
of depression to the end of 2007 were identified, and risks were compared between the two
groups.
Results: The incidence of depression was 1.80-times higher in the diabetic group than in
nondiabetic subjects over a median follow-up of 6.5 years (adjusted hazard ratio [HR] = 1.46,
95% confidence interval [CI]: 1.24~1.71). Age-specific HRs for incidence of depression in
relation to diabetes were not statistically different between the patient subgroups aged *Manuscript (without author details and affiliations)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 2
20-39, 40-49, 50-59, 60-69 and ≥ 70 years (p value for age-diabetes interaction=0.33).
Stratified analyses showed that the association was much stronger for subjects without
comorbid cardiovascular disease than for those with this comorbidity. Insulin treatment was
associated with a 43% reduced risk of depression in diabetic patients.
Conclusions: In this population-based study, diabetic patients were at a higher risk for
subsequent depression. Adequate treatment reduced the risk.
Keywords: depression; diabetes mellitus; incidence; longitudinal studies.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 3
What is already known about the topic?
• Studies have demonstrated an association between depression and diabetes.
• Whether the incidence of depression is subsequently elevated among patients diagnosed with diabetes is controversial.
• The effect of diabetes on subsequent depression may differ with age.
What this paper adds?
• Diabetes was associated with increased risk of subsequent depression, and the association did not vary by age.
• The association between diabetes and an elevated risk of developing depression was stronger in patients without cardiovascular disease than in those with the disease.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 4 1. Introduction
Diabetes is an increasingly prevalent chronic disease often comorbid with medical conditions
that lead to health and economic burdens (Wild et al., 2004; Campbell and Martin, 2009). It is
recognized that patients with type 2 diabetes have a higher prevalence of depression compared
to nondiabetic subjects (Aliet al., 2006; Anderson et al., 2001). Several prospective studies
examined the bidirectional temporal association between the two diseases, indicating an
excess risk of diabetes incidence in subjects with depression (Golden et al., 2008; Mezuk et
al., 2008). However, whether the incidence of depression is subsequently elevated among
patients diagnosed with diabetes is controversial.
A meta-analysis study examined this issue and suggested that the association may differ
with age (Mezuk et al., 2008). Most previous studies assessed the relationship between
diabetes and the incidence of depression in middle-aged and older populations (de Jonge et al.,
2006; Kim et al., 2006; Maraldi et al., 2007; Palinkas et al., 2004; Polsky et al., 2005). Little
information is available on the risk for young adults compared to older groups. A
cross-sectional study reported that the association between diabetes and the prevalence of
depression was stronger in patients younger than 40 years old than in older patients (Zhao et
al., 2006), but data on the incidences are not available. Given the increasing prevalence of the
onset of diabetes in young adults (Lee et al., 2007), it is important to public health to
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 5
burdens could be particularly important for young patients, as they have longer life spans, and
the presence of depression deteriorates the quality of life of patients with diabetes (Eren et al.,
2008).
Using a nationwide database from the National Health Insurance (NHI) program in
Taiwan, we conducted a population-based retrospective cohort study to examine the
association between diabetes and the risk of subsequent depression, emphasizing age-specific
variations. We also assessed whether cardiovascular disease (CVD), which usually coexists
with diabetes, modifies the association between diabetes and depression. This knowledge
could provide recommendations for primary care practice on identifying targets of prevention
efforts.
2. Methods
2.1. Data Source
Taiwan’s NHI, a government-operated, single-payer health insurance program reformed in 1995, covered approximately 99% of the total 23 million people in Taiwan by 2007 (Lu and
Hsiao, 2003; NHI profile, 2009). For research and administrative purposes, the National
Health Research Institute maintains computerized claims data, which include files of
ambulatory care, inpatient care, prescription drugs, and a registry of beneficiaries, and
releases the database for public access. This study used a subset consisting of longitudinal
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 6
NHI. Using this dataset, we were able to select study subjects and obtain longitudinal
healthcare information for each subject. In order to protect patients’ privacy, all patient-level information can be retrieved and linked only through scrambled personal identification, and
thus, this study was exempt from institutional review board approval.
2.2. Patient Population and Subjects in the Comparison Cohort
We defined diabetes mellitus (DM) using the International Classification of Diseases,
9th Revision, Clinical Modification (ICD-9-CM) code 250. Patients with diabetes included in
this study were individuals aged 20 years and above, newly diagnosed in 2000~2002
identified from ambulatory care visits or admission records. Subjects with a prior diabetes
history were excluded. In an attempt to reduce the misclassification of the diabetes status, we
included only patients who received medical care at least three times for diabetes during the
3-year period.
A comparison cohort was randomly selected from individuals with no history of
diabetes. For each patient with diabetes, four persons free of diabetes were randomly selected
for the comparison cohort which were frequency matched with distributions of the year of
diagnosis, age (at 10-year intervals), and gender. The date of entering this study for both
patients with diabetes and subjects in the comparison cohort was the date when the person
was identified by the attending physician. We excluded persons with a history of depression
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 7
cohort for the data analysis. A patient was defined as having a depression history if he or she
had at least one outpatient and/or inpatient claim for depression from 1996, when the
computerized claims data were available, until the date of entering this study.
2.3. Measures of Study Outcomes and Comorbid Conditions
Study subjects were followed up to the end of 2007 to measure the incidence of
depression. We identified subjects as having developed depression if they had at least two
treatment claims for depression in outpatient visits and/or hospitalizations for ICD-9-CM
code 296.2 or 296.3 during the follow-up period. There were three diagnosis fields on each
outpatient claim and five diagnosis fields for inpatient claims. Follow-up person-years were
determined by calculating the time interval between the entry date and the earliest of one of
the following: a diagnosis of depression, the date of withdrawal from the NHI program
including loss to follow-up or death, or December 31, 2007.
In the data analyses, we took into account several health conditions that are reported to
be associated with diabetes and depression (Brown et al., 2006). These medical histories
included hypertension (ICD-9-CM codes 401-405), hyperlipidemia (ICD-9-CM code 272),
coronary artery disease (CAD; ICD-9-CM codes 410-414, 425-429), stroke (ICD-9-CM codes
430-438), dementia (ICD-9-CM codes 290.0, 290.1, 290.2, 290.3, and 331.0), Parkinson's
disease (ICD-9-CM code 332), and cancer (ICD-9-CM codes 140-209). We considered these
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 8
claims and/or hospitalizations records of study subjects prior the date of their entering this
study. In the data analysis, we defined CVD as at least one diagnosis of hypertension,
hyperlipidemia, CAD, and/or stroke present before the date of entering the study. We also
identified insulin use among patients with diabetes, as previous studies reported an
association between insulin use and the incidence of depression (Ali et al., 2006; Brown et al.,
2006).
2.4. Statistical Analysis
We compared differences in sociodemographic characteristics including age, gender,
occupation, level of urbanization of residential area, and monthly income, and comorbidities
between patients with diabetes and the nondiabetic comparison group. Levels of urbanization
were determined by the population density (persons/km2) in the region where the study
subject was registered for insurance. Areas in the lowest and highest quartiles of population
density were classified as areas of low and high urbanization, respectively; those in the other
two quartiles were categorized as moderately urbanized areas.
The Kaplan-Meier method was used to compare the cumulative incidence of depression
in the diabetes and comparison groups, and a log-rank test was used to examine the
significance level of differences between groups. We used Cox proportion hazard regression
analyses to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) of depression for
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 9
of diabetes in young patients, we performed all regression analysis by stratifying subjects into
five age groups (20~39, 40~49, 50~59, 60~69, and ≥ 70 years) based on the age at entry into
the study. We combined patients aged 20~29 and 30~39 years to avoid imprecise estimates
due to the small numbers of subjects in these two age groups. A univariate analysis estimated
the crude association with diabetes. The multivariable-adjusted models included age, gender,
occupation, income, and medical history, such as hypertension, hyperlipidemia, CAD, and
stroke. We pre-selected several potential confounding factors based on knowledge from a
literature review and then used several approaches to select variables for inclusion in the
models for adjustment. We examined whether these variables were substantially associated
with diabetes and depression in descriptive analyses. The crude and adjusted associations
were also compared to evaluate whether the covariates influenced the strength of the
association between diabetes and the incidence of depression. To assess interaction effect
between age and diabetes, we performed the likelihood ratio test comparing models with
and without interaction terms between the 5 age groups and the diabetes. The
interaction effect between comorbid CVD and diabetes was also examined. In addition,
regression analyses were repeated for patients with diabetes stratified by indication of insulin
use. The proportional hazard assumption underlying the Cox regression models was examined
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 10
performed using SAS statistical software (vers. 9.1 for Windows; SAS Institute, Cary, NC),
and all statistical tests were two-sided.
3. Results
Compared to subjects in the comparison cohort, patients with diabetes were less likely
to be white collar workers and had lower incomes (Table 1). Patients in the diabetes group
were much more likely to have hypertension (49.0% vs. 22.3%, p<0.001), hyperlipidemia
(29.7% vs. 5.2%, p<0.001), CAD (27.8% vs. 14.8%, p<0.001), and stroke (12.1% vs. 6.4%,
p<0.001).
The Kaplan-Meier analysis showed that the cumulative probability of depression
incidence was higher for patients with diabetes than for those without diabetes during the
study period (p<0.001, Figure 1). The cumulative incidence of depression was higher in
diabetic patients than in nondiabetic subjects in all age groups (Figure 2). The difference in
the incidences between each age group pair declined with increasing age, and was not
significant only for the subgroup aged ≥ 70 years.
With a median follow-up of 6.5 years in both groups, the overall incidence of
depression was 1.80-times higher in the diabetic group than in the comparison group (28.5 vs.
15.9 per 10,000 person-years) (Table 2). Relative to the non-diabetic comparison group,
patients with diabetes had an HR of 1.77 for developing depression (95% CI: 1.53~2.06,
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 11
slightly attenuated to 1.46 (95% CI: 1.24~1.71, model 2) after further adjusting for the
medical history. We found no significant interaction between age and diabetes on the risk
of subsequent depression (likelihood ratio test for age-diabetes interaction in model 2,
p=0.33), which indicated the estimated HRs for the diabetes-depression association were
not significantly different between age groups.
Table 3 presents the adjusted age-specific HR for the risk of depression in relation to
diabetes for patients with and without CVD. We found a statistically significant effect of an
interaction between CVDs and diabetes (p<0.001). Diabetes was associated with an
appreciably increased risk of depression in patients without CVDs (HR: 2.21, 95% CI:
1.70~2.86) but not in those with the disease. Among diabetes patients without CVD, those
aged 60~69 years had the greatest HR (3.82, 95% CI: 2.07~7.07), followed by the 50~59-year
group (HR: 2.24, 95% CI: =1.33~3.75). The corresponding figures for patients aged 20~39,
40~49 and ≥ 70 years were 1.93 (95% CI: 1.10~3.40), 1.85 (95% CI: 1.16~2.95), and 1.58
(95% CI: 0.36~6.94), respectively. But the differences in these five age-specific HRs were
not statistically significant (The p-value for age-diabetes interaction, 0.48).
Table 4 shows that diabetes patients with insulin treatment were less likely to have a
risk of subsequent depression. The beneficial effect was the greatest for diabetic patients with
comorbid CVD, with an HR of 0.57 (95% CI: 0.38~0.86) relative to comparison subjects.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 12
depression than those prescribed with insulin. Further analyses of other antidiabetes drugs,
including biguanide, sulfonylurea, thiazolidinedione, and alpha-glucosidase inhibitors,
revealed similar findings (data not shown).
4. Discussion
In this population-based retrospective cohort study in Taiwan, newly diagnosed
diabetes was associated with an increased incidence of depression over a median follow-up of
6.5 years. The association was more pronounced in persons without comorbid CVD. No
significant age-specific variation was observed in the association.
Previous studies from Western populations reported a marginal or no association
between diabetes and the incidence of depression (Brown et al., 2006; Engum, 2007; Palinkas
et al., 2004; Polsky et al., 2005). Differences in methodology and inclusion of study subjects
between those studies and ours may explain the inconsistent findings. First, previous studies
assessed diabetes using self-reported data, blood tests for fasting glucose, or claims data that
may have included patients with different severities of diabetes, thereby yielding inconsistent
findings. (Brown et al., 2006; Engum, 2007; Mezuk et al., 2008; Polsky et al., 2005). Second,
most previous studies were limited to Western populations (Brown et al., 2006; Engum, 2007);
few data are available for Asian populations. Ethnic differences in the risk of having
depression may exist among patients with diabetes (Ali et al., 2009). Kim et al. (2006)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 13
line with our observations for the older age group. The Multi-Ethnic Study of Atherosclerosis
included Chinese-Americans in their analysis, but the authors did not report the strength of
ethnicity-specific associations (Golden et al., 2008).
Some possible mechanisms may explain the association between diabetes and the
incidence of depression. Complications of diabetes and psychological pressure resulting from
coping with diabetes may induce subsequent depression (Golden et al., 2008; Talbot and
Nouwen, 2000). de Jonge et al. (2006) found that diabetes was associated with an increased
prevalence of depression only in subjects without comorbidity, but the incidence of
depression was more likely to be pronounced in those with the presence of a comorbidity. In
the present study, diabetes did not add an appreciably risk to depression in subjects with
preexisting CVD. A possible explanation for this observation is that we included subjects with
new-onset diabetes. They had not yet developed serious complications. The impact may be
more pronounced in patients with diabetes for a longer duration. It is important to note that
insulin treatment has a protective effect in patients with the comorbidity of CVD by reducing
the risk of depression by 43%. We also found a similar beneficial effect from taking other
antidiabetic medicines (data not shown). This finding indicates the importance of adherence
to medical treatment recommendations for patients with diabetes.
We also found in this study that among all age groups, the incidence of depression in
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 14
was the lowest for this age group. This discrepancy yielded a smaller HR for developing
depression in these old diabetic patients. Other risk factors may be more prominent than
diabetes in the association with depression for older patients.
The following limitations merit consideration before concluding this study. First,
relying on administrative data of the diagnosis may include some extent of misclassification
of diseases. In an attempt to reduce the likelihood of misclassification, we used at least two
consecutively identical diagnoses to define diabetes and depression. In addition, we were able
to capture disease events at any time point when a subject had a medical care visit, which
reduced the likelihood of underestimating incident events that occurred during the interval
between follow-up visits (de Jonge et al., 2006). Second, we were unable to include patients
with mild symptoms who were less likely to seek health care, and patients who had not
reported their mood symptoms. Our study results thus apply to patients with detected and
treated diabetes and depression. Previous studies reported that the prevalence of depression is
lower in South Asians than in white Europeans, suggesting a possibility of culture differences
in the presentation of depression (Ali et al., 2009; Ineichen, 1990; Yeung et al., 2004). Asians
may have greater difficulty or may be more reluctant to express depressive symptoms because
of the stigma associated with mental disorders, which could lead to a lower sensitivity of the
diagnosis of depression. Thus, in this study, the incidence of depression may have been
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 15
similar magnitudes in both the diabetes and comparison groups and thus would likely have
little effect on the estimated relative risk (hazard ratio). Third, we did not differentiate type 1
from type 2 diabetes in the claims data. The relationship with depression might differ between
these two types of diabetes, as the pathophysiology of these conditions differs. In this study,
most patients were type 2 diabetes because 97% of patients with diabetes have type 2 diabetes
in Taiwan (Chuang et al., 2001). Fourth, some potential confounding factors for the
association between diabetes and depression, such as the body-mass index and smoking status,
are not included in the NHI database. Residual confounding might have occurred in the
observed association. Fifth, because the computerized claims data were unavailable before
1996, we were unable to identify depression diagnosed earlier. Thus, we could not exclude
the possibility of including recurrent cases of depression, particularly among older subjects.
Last, the small number of subjects in the stratified analyses of age groups, CVD, and diabetes
may have caused imprecise estimates and insufficient statistical power to detect a moderate
association.
Findings of this study have important implications for nursing and clinical practice.
Evidence has shown that among patients with both diabetes and depression, those who receive
depression treatment are at a lower risk of mortality than those without an intervention
(Bogner et al., 2007). Depression is an important predictor of suicide, and the risk of suicide
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 16
intervention strategies to help patients cope with psychological pressure in the early stage
when caring for patients with diabetes.
This study revealed an appreciably increased risk of developing depression associated
with diabetes, and the association did not vary by age. Further investigation of the role of
CVD in the association between diabetes and depression is warranted. In addition, insulin and
antidiabetic medications may be associated with a reduced risk of depression. Further studies
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 17 References
Ali, S., Davies, M . J., Taub, N. A., Stone, M. A., Khunti, K., 2009. Prevalence of diagnosed
depression in South Asian and white European people with type 1 and type 2 diabetes
mellitus in a UK secondary care population. Postgraduate Medical Journal 85(1003), 238–
243.
Ali, S., Stone, M. A., Peters, J. L., Davies, M. J., Khunti, K., 2006. The prevalence of
co-morbid depression in adults with type 2 diabetes: a systematic review and meta-analysis.
Diabetic Medicine 23(11), 1165–1173.
Anderson, R. J., Freedland, K. E., Clouse, R. E., Lustman, P. J., 2001. The prevalence of
comorbid depression in adults with diabetes: a meta-analysis. Diabetes Care 24(6),
1069-1078.
Bogner, H. R., Post, E. P., Morales, K. H., Bruce, M. L., (2007). Diabetes, depression, and
death. A randomized controlled trial of a depression treatment program for older adults
based in primary care (PROSPECT). Diabetes Care 30(12), 3005-3010.
Brown, L., Majumdar, S., Newman, S., Johnson, J., 2006. Type 2 diabetes does not increase
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 18
Campbell, R. K., Martin, T. M., 2009. The chronic burden of diabetes. American Journal
Management and Care 15(9 Suppl), S248-254.
Chuang, L. M., Tsai, S.T., Huang, B. Y., Tai, T. Y., DIABCARE (Taiwan) Study Group,
2001. The current state of diabetes management in Taiwan Diabetes Research and Clinical
Practice 54(Suppl 1), S55-65.
de Jonge, P., Roy, J., Saz, P., Marcos, G., Lobo, A., 2006. Prevalent and incidence depression
in community-dwelling elderly persons with diabetes mellitus: results from the
ZARADEMP project. Diabetologia 49(11), 2627–2633.
Engum A., 2007. The role of depression and anxiety in onset of diabetes in a large
population-based study. Journal of Psychosomatic Research 62(1), 31-38.
Eren, I., Erdi, O., Sahin, M., 2008. The effect of depression on quality of life of patients with
type II diabetes mellitus. Depression and Anxiety 25(2), 98-106.
Golden, S. H., Lazo, M., Carnethon, M., Bertoni, A. G., Schreiner, P. J., Diez Roux, A. V.,
Lee, H., Lykesos, C., 2008. Examining a bidirectional association between depressive
symptoms and diabetes. Journal of American Medical Association 299(23), 2751-2759.
Ineichen, B., 1990. The mental health of Asians in Britain: little disease or under-reporting?
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 19
Kim, J., Stewart, R., Kim, S., Yang, S., Shin, I., Yoon, J., 2006. Vascular risk factors and
incident late-life depression in a Korean population. The British Journal of Psychiatry
189(1), 26-30.
Lee, J. M., Okumura, M. J., Freed, G. L., Menon, R. K., Davis, M. M., 2007. Trends in
hospitalizations for diabetes among children and young adults: United States, 1993-2004.
Diabetes Care 30(12), 3035-3039.
Lu, J. F., Hsiao, W. C., 2003. Does universal health insurance make health care unaffordable?
Lessons from Taiwan. Health Affairs 22(3), 77–88.
Maraldi, C., Volpato, S., Penninx, B. W., Yaffe, K., Simonsick, E. M., Strotmeyer, E. S.,
Cesari, M. , Kritchevsky, S.B., Perry, S., Ayonayon, H.N., Pahor, M., 2007. Diabetes
mellitus, glycemic control, and incident depressive symptoms among 70- to 79-year-old
persons: the health, aging and body composition study. Archives of Internal Medicine
167(11):1137-1144.
Mezuk, B., Eaton, W. W., Albrecht, S., Golden, S. H., 2008. Depression and type 2 diabetes
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 20
NHI profile (updated 2009.06.05). Available at
http://www.nhi.gov.tw/webdata/AttachFiles/Attach_13787_1_NationalHealthInsurancein
Taiwan2009.pdf. Accessed June 4, 2010.
Palinkas, L., Lee, P., Barrett-Connor, E., 2004. A prospective study of type 2 diabetes and
depressive symptoms in the elderly: the Rancho Bernardo Study. Diabetic Medicine
21(11): 1185-1191.
Polsky, D., Doshi, J. A., Marcus, S., Oslin, D., Rothbard, A., Thomas, N., Thompson, C. L.,
2005. Long-term risk for depressive symptoms after a medical diagnosis. Archives of
Internal Medicine 165(11), 1260-1266.
Talbot, F., Nouwen, A., 2000. A review of the relationship between depression and diabetes in
adults: is there a link? Diabetes Care 23(10), 1556-1562.
Wild, S., Roglic, G., Green, A., Sicree, R., King, H., 2004. Global prevalence of diabetes:
estimates for the year 2000 and projections for 2030. Diabetes Care 27(5), 1047-1053.
Yeung, A., Chan, R., Mischoulon, D., Sonawalla, S., Wong, E., Nierenberg, A. A., Fava, M.,
2004. Prevalence of major depressive disorder among Chinese-Americans in primary care.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 21
Zhao, W., Chen, Y., Lin, M., Sigal, R. J., 2006. Association between diabetes and depression:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 22 Figure Legends
Figure 1. Kaplan-Meier curves of the time to occurrence of depression for patients with
diabetes and the comparison group.
Figure 2. Kaplan-Meier curves of the time to occurrence of depression for patients with
log-rank test, p <0.001 Figure1
log-rank test, p =0.001 log-rank test, p< 0.001 40-49 years 20-39 years 50-59 years 60-69 years log-rank test, p<0.001 ≥ 70 years
log-rank test, p<0.001 log-rank test, p=0.23
Table 1
Socio-demographic characteristics and comorbidity in patients with diabetes and in comparison cohort Comparison group (n=55,608) Pubjects with diabetes (n=14,048) p-value Women, n (%) 26782 (48.2) 6808 (48.5) 0.52 Age, years, n (%) 1.00 20-39 6567 (11.8) 1662 (11.8) 40-49 13128 (23.6) 3310 (23.6) 50-59 14692 (26.4) 3707 (26.4) 60-69 12589 (22.6) 3178 (22.6) ≥70 8632 (15.5) 2191 (15.6) Mean (SD) 55.1 (13.5) 55.6 (13.1) <0.001 Occupation, n (%) <0.001 White collar 21504 (38.7) 5000 (35.6) Blue collar 22838 (41.1) 6069 (43.2) Others 11266 (20.3) 2979 (21.2) Urbanization, n (%) 0.95 Low 2356 (4.2) 587 (4.2) Moderate 17282 (31.1) 4363 (31.1) High 35969 (64.7) 9097 (64.8) Income, NTD, n (%) <0.001 <15,000 23075 (41.5) 6210 (44.2) 15,000-29,999 23657 (42.5) 5822 (41.4) ≥30,000 8876 (16.0) 2016 (14.4) Comorbidity, n (%) Hypertension 12581 (22.6) 6876 (49.0) <0.001 Hyperlipidemia 2945 (5.3) 4178 (29.7) <0.001 Coronary artery disease 8380 (15.1) 3905 (27.8) <0.001
Stroke 3621 (6.5) 17.2 (12.1) <0.001 Dementia 244 (0.4) 98 (0.7) <0.001 Parkinson's disease 511 (0.9) 199 (1.4) <0.001 Cancer 896 (1.6) 267 (1.9) 0.017 Diabetes treatment, n (%) Insulin - 3816 (27.2) Biguanide - 9352 (66.6) Sulfonylurea - 9270 (66.0) Thiazolidinedione - 2390 (17.0) Alpha-glucosidase inhibitor - 2400 (17.1)
NTD=new Taiwan dollars. Table 1
Table 2
Age-specific incidence and hazard ratio of depression associated with diabetes
Comparison group Patients with diabetes Hazard ratio (95% confidence interval) Age, years n, depression Person-years at risk Incidencea n, depression Person-years at risk Incidencea
Unadjusted Model 1 Model 2
All 572 360870 15.9 258 90460 28.5 1.79 (1.54-2.07) 1.77 (1.53-2.06) 1.46 (1.24-1.71) 20-39 67 42425 15.8 34 10633 32.0 1.97 (1.30-2.99) 1.98 (1.30-3.03) 1.39 (0.84-2.30) 40-49 130 85661 15.2 59 21412 27.6 1.82 (1.34-2.48) 1.86 (1.36-2.54) 1.68 (1.18-2.39) 50-59 153 95117 16.1 73 23808 30.7 1.92 (1.45-2.53) 1.89 (1.43-2.50) 1.48 (1.08-2.01) 60-69 119 81970 14.5 59 20542 28.7 1.93 (1.41-2.65) 1.84 (1.35-2.53) 1.60 (1.15-2.25) ≥70 103 55697 18.5 33 14065 23.5 1.27 (0.86-1.89) 1.23 (0.83-1.83) 0.98 (0.65-1.47) a per 10,000 person-years
Model 1 was adjusted for age, sex, occupation and income.
Model 2 was adjusted for variables in model 1 and comorbidiy including hypertension, stroke, hyperlipidemia and coronary artery disease. Table 2
Table 3
Adjusted age-specific incidence and hazard ratio of depression associated with diabetes by history of cardiovascular disease
Patients with diabetes Comparison group Hazard ratio (95% CI) in relation to diabetes Age, years No. of
depression Incidence No. of depression incidence Patients with cardiovascular diseasea 20-39 17 3.7 14 6.4 0.69 (0.33-1.44) 40-49 37 3.1 32 2.6 1.18 (0.73-1.91) 50-59 55 3.4 80 2.9 1.19 (0.85-1.68) 60-69 44 2.7 81 2.2 1.18 (0.82-1.71) ≥70 31 2.5 88 2.5 0.99 (0.65-1.49) All 184 3.0 295 2.6 1.10 (0.91-1.33) Patients without cardiovascular diseasea 20-39 17 2.8 53 1.3 1.93 (1.10-3.40) 40-49 22 2.3 98 1.3 1.85 (1.16-2.95) 50-59 18 2.3 73 1.1 2.24 (1.33-3.75) 60-69 15 3.4 38 0.8 3.82 (2.07-7.07) ≥70 2 1.1 15 0.7 1.58 (0.36-6.94) All 74 2.5 277 1.1 2.21 (1.70-2.86) CI = confidence interval.
Models were adjusted for age, sex, occupation and income.
a
Patients with cardiovascular disease were identified if they had been diagnosed with any one of the following medical conditions prior to the date of entering this study:
hypertension, hyperlipidemia, coronary artery disease and stroke. Table 3
Table 4
Adjusted hazard ratio of depression associated with diabetes and insulin use by age and history of cardiovascular disease Age, years
All 20-39 40-49 50-59 60-69 ≥70
HR (95%CI) HR (95%CI) HR (95%CI) HR (95%CI) HR (95%CI) HR (95%CI)
Patients with CVDa No diabetes 1.00 1.00 1.00 1.00 1.00 1.00 Diabetes with insulin use 0.57 (0.38-0.86) 0.20 (0.03-1.53) 0.56 (0.20-1.58) 0.83 (0.41-1.65) 0.66 (0.31-1.44) 0.37 (0.14-0.998) Diabetes without insulin use 1.30 (1.06-1.58) 0.81 (0.39-1.69) 1.37 (0.84-2.23) 1.31 (0.91-1.88) 1.39 (0.94-2.05) 1.32 (0.85-2.03) Patients without CVDa No diabetes 1.00 1.00 1.00 1.00 1.00 1.00 Diabetes with insulin use 1.85 (1.13-3.03) 1.94 (0.76-4.95) 1.35 (0.49-3.68) 2.83 (1.23-6.52) 1.61 (0.38-6.73) - - Diabetes without insulin use 2.34 (1.76-3.13) 1.93 (1.02-3.67) 2.01 (1.21-3.34) 2.02 (1.10-3.73) 4.84 (2.54-9.23) 2.86 (0.65-12.52)
HR=hazard ratio; CI=confidence interval.
Models were adjusted for age, sex, occupation and income.
a
Patients with cardiovascular disease (CVD) were identified if they had been diagnosed with any one of the following medical conditions prior to the index date: hypertension, hyperlipidemia, coronary artery disease and stroke.