Published online Jul 18, 2023.
https://doi.org/10.3349/ymj.2023.0121
Association between Depression and Mortality in Patients with Pain Conditions: A South Korean Nationwide Cohort Study
Abstract
Purpose
Patients with pain conditions may experience depression that greatly complicates treatment. In this study, we examined risk factors for depression in patients with pain conditions and associations between depression and long-term mortality in such patients.
Materials and Methods
Data from the National Health Insurance Service database in South Korea were used in this cohort study. A total of 2.5% of adult patients diagnosed with pain conditions in 2010 were selected using a stratified random sampling technique and included for analysis. We performed multivariate logistic regression modelling to identify risk factors associated with depression and multivariate Cox regression modelling to determine whether depression is associated with 10-year survival outcomes in patients with pain conditions.
Results
In total, data from 1808043 patients with pain conditions in 2010 were analyzed. Among them, 70148 (3.9%) patients had depression. Multivariate logistic regression modelling identified older age, comorbidities, analgesics, female sex, living in an urban area, and other underlying psychiatric morbidities as potential risk factors for depression in patients with pain conditions. Multivariate Cox regression revealed that 10-year all-cause mortality in patients with depression was 1.13-fold (hazard ratio, 1.13; 95% confidence interval, 1.11–1.16; p<0.001) higher than that in patients without depression and pain conditions.
Conclusion
We identified a few potential risk factors for depression among South Korean patients with pain conditions. Depression was associated with elevated 10-year all-cause mortality in patients with pain conditions.
Graphical Abstract
INTRODUCTION
Depression is a common psychiatric illness that limits psychosocial functioning and decreases quality of life.1 The lifetime prevalence of depression in communities from 30 countries from 1994 to 2014 was 10.8%.2 In South Korea, the prevalence of depression increased from 2.8% in 2002 to 5.3% in 2013.3 As the World Health Organization has projected that depression will be the primary cause of disease burden worldwide by 2030,4 depression will be a critical public health issue in the near future.
Chronic pain is a pervasive health issue that exerts substantial social and economic burden on both the affected individual and society.5 Patients with chronic pain may experience depression that greatly complicates treatment.6 According to a previous study that analyzed a national cohort of 1514 individuals taking opioids for chronic pain in Australia, 6 in 10 patients with chronic pain reported lifetime depression.7 However, information regarding risk factors associated with the diagnosis of depression in patients with chronic pain is still lacking. Moreover, information on the effect of depression on long-term mortality in patients with chronic pain is lacking.
Therefore, using the South Korean National Health Insurance Service (NHIS) database, we examined risk factors associated with depression in adult patients with pain conditions. In addition, we examined whether depression in patients with pain conditions affects their long-term mortality.
MATERIALS AND METHODS
Study design and ethical statement
We assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. All procedures involving human subjects/patients were approved by the Institutional Review Board (IRB) of Seoul National University Bundang Hospital (IRB approval number: X-2105-685-901). The requirement for informed consent was waived by the IRB because the data were analyzed anonymously after masking the individual and sensitive information of the study population.
Data source
The NHIS database was used for data extraction. The NHIS is the sole public health insurance system in South Korea and contains information on all disease diagnoses and prescriptions for any drug and/or procedure. The International Classification of Diseases and Related Health Issues 10th Revision (ICD-10) is used to register disease diagnoses in the NHIS database. A medical record technician from the Big Data Center in the NHIS extracted and provided the data for this study following the approval of the study protocol by the NHIS Ethics Committee (NHIS approval number: NHIS-2021-1-615).
Study population
The ICD-10 codes in Supplementary Material 1 (only online)were used to identify all patients with pain conditions. Pain conditions included rheumatoid arthritis, osteoarthritis, low back pain, neck pain, gout, and other pain conditions. The medical record technician initially screened all patients using the ICD-10 codes for pain conditions from 2010 to 2019, and the data of approximately 800000000 patients over a 10-year period were extracted from the NHIS database. Due to the large sample size, the data of 2.5% of adult patients (aged ≥20 years) were newly extracted using a stratified random sampling technique, considering age and sex as exclusive strata for sampling. Through the sampling process, the sampled study population was drawn from each of the strata, with sizes proportional to the strata of the overall patients. A stratified random sampling method was employed using SAS version 9.4 (SAS Institute, Cary, NC, USA).8
Patients diagnosed with cancer as a comorbidity were excluded to focus on those diagnosed with pain conditions. Next, patients who underwent surgery in 2010 were excluded because acute pain after surgery could be a bias in determining patients with pain conditions.
Among screened patients with pain conditions during 2010–2019, patients with pain conditions in 2010 were selected as a cohort. Patients who died in 2010 were excluded to focus on 10-year survival from January 1, 2011. Finally, patients with pain conditions in 2010 were included in the analysis (Fig. 1).
Fig. 1
Flowchart representing the selection process of patients with pain conditions.
Exposure variable (depression)
As the main exposure variable, diagnoses of depression were extracted using the ICD-10 codes F32, F33, and F34.1. Patients with pain conditions who were diagnosed with depression were considered the depression group, whereas those without a diagnosis of depression were considered the no depression group. As the sole public insurance system in South Korea, appropriate financial coverage for psychological diseases, such as depression, is provided only when the psychological disease is registered in the NHIS database by a physician/psychiatrist. Moreover, information regarding the prescription of antidepressants (ADs) was extracted. The ADs included moclobemide, toloxatone, mirtazapine, bupropion, nefazodone, trazodone, venlafaxine, milnacipran, duloxetine, desvenlafaxine, tianeptine, fluoxetine, fluvoxamine, paroxetine, sertraline, sertraline, citalopram, escitalopram, amineptine, amitriptyline, amoxapine, clomipramine, dosulepin, doxepin, imipramine, maprotiline, nortriptyline, and quinupramine. Individuals prescribed ADs for continuous use of 30 days or more were classified as AD users in this study.
Study outcome
First, we examined factors associated with the diagnosis of depression in patients with pain conditions in 2010. Second, we examined whether depression affected 10-year all-cause mortality in patients with pain conditions in 2010. Third, we examined whether depression affected 10-year disease-specific mortality in patients with pain conditions.
For time-to-event analysis, any death from January 1, 2011 to April 30, 2021 was considered an event, whereas the survival period from January 1, 2011 was considered the duration. For example, if a patient was diagnosed with depression in February 2010 and another was diagnosed with depression in October 2010, the survival times of both patients were calculated from January 1, 2011 in the time-to-event analysis. In addition to 10-year all-cause mortality, 10-year mortality according to specific causes was also evaluated. All physicians register the main causes of death for all individuals using ICD-10 codes in the Statistics Korea database as mandated by the central government. Statistics Korea approved the sharing of information regarding the main causes of death for this study, and the ICD-10 codes of the main causes of death are presented in Supplementary Material 2 (only online): the main causes of death from January 1, 2011 to April 30, 2021 were provided by Statistics Korea using ICD-10 codes.
Covariates
Age and sex were collected as demographic information. To reflect socioeconomic status information, data on employment status, household income level, and residence for the patients were collected. Household income level was nationally registered in the NHIS database to determine the insurance premium of the South Korean population: individuals who are too poor to pay their insurance premiums or have difficulty in supporting themselves financially are covered under the Medical Aid Program, in which the government covers almost all medical expenses to reduce the burden of medical costs. Except for patients who belonged to the Medical Aid Program, all patients were divided into four groups using quartile ratios. The residences of all patients were classified into two groups: urban areas (Seoul and other metropolitan cities) and rural areas (all other areas).
To reflect physical comorbidity status, Charlson Comorbidity Index (CCI) score and underlying disability were collected. CCI score was calculated using registered ICD-10 codes, as shown in Supplementary Material 3 (only online). All individuals with disabilities should be registered in the NHIS database to receive benefits from South Korea’s social welfare system. Disability was divided into six severity grades, and we considered two severity groups (grades 1–3: severe disability; grades 4–6: mild-to-moderate disability). In addition, long-term (≥ 90 days) opioid, gabapentin or pregabalin, paracetamol, and non-steroidal anti-inflammatory drug (NSAID) prescription information was collected as covariates.
As depression can occur in several other psychiatric morbidities,9 other underlying psychiatric morbidities were collected as covariates, such as bipolar disorder (F31), anxiety disorder (F40, F41), insomnia disorder (G47, F51), substance abuse (F10-19), post-traumatic stress disorder (PTSD) (F43.1), and schizophrenia (F20). All covariates were collected based on 2010 data, except for CCI, because they are based on a cohort study in 2010. However, CCI scores were collected using ICD-10 codes registered in the NHIS database during 2009–2010 because a longer evaluation period is required to include all chronic diseases or underlying diseases and calculate CCI scores for the study population.
Statistical analyses
Clinicopathological characteristics are presented as mean values with standard deviations (SD) for continuous and numbers with percentages for categorical variables. For comparison of clinicopathological characteristics between the depression and no depression groups among the 2010 cohort with pain conditions, the t-test and chi-squared test were used for continuous variables and categorical variables, respectively. First, we constructed a multivariate logistic regression model for the diagnosis of depression in the 2010 cohort with pain conditions. All covariates were included in the model for adjustment, and results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). The Hosmer–Lemeshow test was used to confirm that the goodness of fit in the model was appropriate.
Next, we conducted multivariate Cox regression analysis for 10-year all-cause mortality to examine whether depression affected 10-year all-cause mortality in the 2010 cohort with pain conditions. All covariates were included in the adjusted model. For sensitivity analysis, we divided the depression group into two groups (depression group with no AD use or AD use) to examine whether AD use affected the association between the depression group and 10-year all-cause mortality. Additionally, we constructed 17 multivariate Cox regression models for 10-year disease-specific mortality. A variance inflation factor <2.0 was used to confirm that there was no multicollinearity between variables in all models, and the results of Cox regression are presented as hazard ratios (HRs) with 95% CIs. Log-log plots were used to confirm that the central assumption of the Cox proportional hazard model was satisfied. All statistical analyses were performed using IBM SPSS version 25.0 software (IBM SPSS Statistics for Windows, version 25.0, Armonk, NY, USA), and statistical significance was set at p<0.05.
RESULTS
Cohort and depression
The sampled study population was drawn from each of the strata, with sizes proportional to the strata of the overall patients (approximately 800000000 patients). Finally, 19645161 patients were sampled using a stratified random sampling. Among these, 874171 patients diagnosed with cancer as a comorbidity and 756 patients who underwent surgery in 2010 were excluded. Among 18770234 screened patients with pain conditions during 2010–2019, 1808043 patients with pain conditions in 2010 were selected as a cohort. A total of 4024 patients who died in 2010 were excluded to focus on 10-year survival from January 1, 2011. Finally, 1804019 patients with pain conditions in 2010 were included in the analysis (Fig. 1).
Table 1 shows the results of the comparison of clinicopathological characteristics between the depression and no depression groups of patients with pain conditions. The mean ages of the depression and no depression groups were 57.2 (SD, 16.3) and 45.5 (SD, 15.8) years, respectively (p<0.001). Table 2 shows the results of the multivariate logistic regression model for the diagnosis of depression in the 2010 cohort with pain conditions. Older age (OR, 1.02; 95% CI, 1.02–1.02; p<0.001), underlying mild-to-moderate disability (OR, 1.17; 95% CI, 1.14–1.21; p<0.001), severe disability (OR, 1.62; 95% CI, 1.55–1.68; p<0.001), increased CCI score (OR, 1.23; 95% CI, 1.23–1.24; p<0.001), opioid use (OR, 1.25; 95% CI, 1.12–1.39; p<0.001), gabapentin or pregabalin use (OR, 2.08; 95% CI, 1.97–2.20; p<0.001), and NSAID use (OR, 1.27; 95% CI, 1.10–1.46; p<0.001) were associated with an increased prevalence of depression in patients with pain conditions. Male sex (OR, 0.68; 95% CI, 0.67–0.69; p<0.001) and living in a rural area (OR, 0.98; 95% CI, 0.96–0.99; p=0.005) were associated with a decreased prevalence of depression in patients with pain conditions. Moreover, all underlying psychiatric morbidities (bipolar disorder, anxiety disorder, insomnia disorder, substance abuse, PTSD, and schizophrenia) were associated with an increased prevalence of depression in patients with pain conditions (all p<0.05).
Table 1
Comparison of Clinicopathological Characteristics between the Depression and No Depression Groups of Patients with Pain Conditions (n=1804019)
Table 2
Multivariate Logistic Regression Model for the Diagnosis of Depression in the 2010 Cohort with Pain Conditions
Survival analyses among cohort
Table 3 shows the multivariate Cox regression model for 10-year all-cause mortality in patients with pain conditions. In multivariate model 1, the depression group showed a 1.13-fold (HR, 1.13; 95% CI, 1.11–1.16; p<0.001) higher 10-year all-cause mortality than the no depression group. All other covariates in multivariate model 1 are presented in Supplementary Table 1 (only online).
Table 3
Multivariate Cox Regression Model for 10-Year All-Cause Mortality in Patients with Pain Conditions
In sensitivity analysis (multivariate model 2), depression groups with no or with AD use showed 1.14-fold (HR, 1.14; 95% CI, 1.11–1.17; p<0.001) and 1.12-fold (HR, 1.12; 95% CI, 1.09–1.16; p<0.001) higher 10-year all-cause mortality, respectively, than the no depression groups.
Table 4 shows the results of survival analyses for 10-year disease-specific mortality. The depression group showed higher 10-year mortality due to mental disease (HR, 1.46; 95% CI, 1.28–1.67; p<0.001); nervous system disease (HR, 1.85; 95% CI, 1.70–2.01; p<0.001); circulatory disease (HR, 1.13; 95% CI, 1.09–1.18; p<0.001); respiratory disease (HR, 1.23; 95% CI, 1.16–1.30; p<0.001); skin disease (HR, 1.60; 95% CI, 1.09–2.35; p=0.017); events during pregnancy, childbirth, and the puerperium (HR, 10.24; 95% CI, 2.04–51.25; p=0.005); symptoms, signs, and abnormal clinical and laboratory findings (HR, 1.12; 95% CI, 1.04–1.19; p=0.001); and injury, poisoning, and certain other consequences of external causes (HR, 1.55; 95% CI, 1.45–1.66; p<0.001) than the no depression group.
Table 4
Survival Analyses of 10-Year Disease-Specific Mortality among Patients with Pain Conditions
Prevalence of depression in patients with pain conditions (2010–2019)
Fig. 2 shows the prevalence of depression and AD use in patients with pain conditions from 2010 to 2019. The 10-year overall prevalences of depression and AD use were 5.1% (958466/18770234) and 4.8% (891937/18770234), respectively. The prevalence of depression was 3.9% (70833/1808043) in 2010 and gradually increased to 6.3% (119592/1892913) in 2019. The prevalence of AD use was 4.1% (73371/1,808043) in 2010 and gradually increased to 5.1% (96753/1892913) in 2019.
Fig. 2
Prevalence of depression and AD use in patients with pain conditions from 2010 to 2019. AD, antidepressant.
DISCUSSION
This cohort study showed that old age, comorbid status, analgesic use (opioids, gabapentin or pregabalin, and NSAIDs), female sex, living in an urban area, and other underlying psychiatric morbidities (bipolar disorder, anxiety disorder, insomnia disorder, substance abuse, PTSD, and schizophrenia) were potential risk factors for depression in patients with pain conditions. Notably, 10-year all-cause mortality was associated with depression with or without AD use in patients with pain conditions.
Various factors have been identified as potential risk factors for depression in patients with pain conditions. Female sex is a well-known risk factor for depression,10 as we reported in patients with pain conditions. Increased CCI score and underlying disability are well-known risk factors, and depression usually co-occurs with other medical diseases.11 Most importantly, all underlying psychiatric morbidities (bipolar disorder, anxiety disorder, insomnia disorder, substance abuse, PTSD, and schizophrenia) were associated with depression in patients with pain conditions. Psychiatric morbidity could co-occur with depression;12 thus, other psychiatric morbidities might be both co-occurring diseases and risk factors for depression in patients with pain conditions. Patients with pain conditions who took analgesics, such as opioids, gabapentin or pregabalin, and NSAIDs, were more likely to have depression than those who did not take analgesics. Patients with pain conditions who took analgesics might experience more severe pain than those who did not take analgesics, resulting in an increased risk of depression. Moreover, a recent multivariate Mendelian randomization analysis reported that the prescription of opioids itself is a risk factor for increased depression.13
Notably, our study analyzed 10-year follow-up data to identify associations between depression and 10-year all-cause and disease-specific mortality rates in patients with pain conditions. Although the causal effect of depression on mortality remains controversial and unproven, an association between depression and increased all-cause and cause-specific mortality has been reported previously.14 Patients with depression might be at high risk for worsening long-term mortality because depression can co-occur with other medical diseases and adversely affect the course of medical disease.15 In addition, depression can cause immune system dysfunction due to hypercortisolemia and proinflammatory reactions,16 possibly resulting in poorer long-term survival in patients with pain conditions. Interestingly, AD use did not affect the association between depression and increased 10-year all-cause mortality in the present study. As an important treatment for depression, ADs can improve the symptoms of depression. However, as ADs are usually prescribed for patients with moderate-to-severe symptoms of depression in South Korea,17 the effect of AD treatment on mortality in patients with depression and pain conditions might be limited. Therefore, further studies are required to confirm the effect of AD prescription on long-term prognosis in patients with pain conditions and depression.
The prevalence of depression and AD use in patients with pain conditions gradually increased from 2010 to 2019. Specifically, the overall prevalence of depression in patients with pain conditions was 5.1%, whereas that among AD users was 4.8%. Two previous studies have reported that 9% and 20% of patients with chronic pain experience major depression.18, 19 Smith, et al.7 reported that 60% of patients with pain conditions experienced lifetime depression; however, they focused on patients with pain conditions who required opioid prescriptions for the treatment of their pain conditions, whereas we included all patients with pain conditions regardless of opioid prescription. In a Danish register-linkage cohort study, the prevalence of patients with pain conditions was 6.1% (95% CI, 5.5–6.6), which was slightly higher than that in our study.20 In addition, the literature shows that the prevalences of depression and AD use in patients with pain conditions have only gradually increased in South Korea. Two factors might account for the lower prevalence of depression in patients with pain conditions in South Korea.7, 18, 19, 20 First, we used registered ICD-10 codes of depression with confirmation of diagnosis, not a questionnaire, which would also detect depressive symptoms. Therefore, patients with mild symptoms of depression might have been excluded. Second, the prevalence of a major depressive disorder was reported to be 2.7% (95% CI, 2.2–3.3) in the general population in South Korea,21 which might account for the relatively lower prevalence of depression in patients with pain conditions compared with that in other countries.
This study has some limitations. First, a 2.5% stratified random sampling technique was used for data extraction, and there may have been some differences between the sampled patients with pain conditions and all patients with pain conditions in South Korea. Second, the severity of each pain condition was not assessed. The duration, stage, and severity of pain conditions might affect the diagnosis of depression and lead to poorer long-term survival outcomes. Third, important information that could reflect the lifestyle of patients, such as alcohol consumption and smoking history, was not included in this study because of the lack of information contained in the NHIS database. Fourth, the severity and duration of depression in patients with pain conditions were not considered in this study. Lastly, as we focused on patients with pain conditions in 2010 to evaluate the association between depression and 10-year all-cause mortality from 2011, the time of occurrence of depression (exposure time) may vary between the patients. Therefore, immortal time bias may have been present and may have affected the results of this study.
In conclusion, we identified a few potential risk factors for depression in patients with pain conditions. Our results showed that increased 10-year all-cause mortality was associated with depression with or without AD use in patients with pain conditions. Furthermore, patients with pain conditions appeared to be at high risk for depression and poorer long-term survival outcomes, regardless of AD use. Future studies are required to prevent the development of depression and improve long-term survival outcomes in patients with pain conditions.
SUPPLEMENTARY MATERIALS
ICD-10 CodesSUPPLEMENTARY MATERIAL 1
ICD-Code for Disease Specific MortalitySUPPLEMENTARY MATERIAL 2
The ICD-10 Codes Used by Comorbidity to Compute the Charlson Comorbidity IndexSUPPLEMENTARY MATERIAL 3
All Other Covariates in Multivariate Model 1Supplementary Table 1
The authors have no potential conflicts of interest to disclose.
AUTHOR CONTRIBUTIONS:
Conceptualization: Tak Kyu Oh and In-Ae Song.
Data curation: Hye Yoon Park.
Formal analysis: Tak Kyu Oh.
Funding acquisition: In-Ae Song.
Investigation: Tak Kyu Oh and In-Ae Song.
Methodology: Tak Kyu Oh and In-Ae Song.
Project administration: In-Ae Song.
Resources: Hye Yoon Park.
Software: Tak Kyu Oh.
Supervision: In-Ae Song.
Validation: Tak Kyu Oh and In-Ae Song.
Visualization: Tak Kyu Oh and In-Ae Song.
Writing—original draft: Tak Kyu Oh.
Writing—review & editing: In-Ae Song and Hye Yoon Park.
Approval of final manuscript: all authors.
DATA AVAILABILITY
Data will be available upon reasonable request to corresponding author.
References
-
World Health Organization. The global burden of disease: 2004 update. Geneva: World Health Organization; 2008.
-
-
Parsons VL. In: Stratified sampling. Hoboken, NJ: Wiley StatsRef: Statistics Reference Online; 2014.
-
-
Thaipisuttikul P, Ittasakul P, Waleeprakhon P, Wisajun P, Jullagate S. Psychiatric comorbidities in patients with major depressive disorder. Neuropsychiatr Dis Treat 2014;10:2097–2103.
-
-
Hölzel L, Härter M, Reese C, Kriston L. Risk factors for chronic depression--a systematic review. J Affect Disord 2011;129:1–13.
-
-
Søndergaard S, Vægter HB, Erlangsen A, Stenager E. Prevalence of depression and anxiety in patients with chronic non-malignant pain–a Danish register-linkage cohort study. Eur Psychiatry 2017;41(S1):S313
-
MeSH Terms
Figures
1 / 2
Tables
1 / 4
