Estimation of Years Lived with Disability Using a Prevalence-Based Approach: Application to Major Psychiatric Disease in Korea

To help develop policies concerning the prevention of psychiatric disease in Korea, we reviewed the literature on this topic in different countries and used a prevalence-based approach to estimate the years lived with disability (YLDs) in Korean patients with major psychiatric diseases. We calculated YLDs by extracting data on the number of patients with mild, moderate, and severe cases of schizophrenia, bipolar disorder, and major depressive disorder, as classified by International Statistical Classification of Disease (ICD) codes. YLDs were highest for patients with major depressive disorder (1190.6; 73.9%), schizophrenia (303.3; 18.8%) and bipolar disorder (117.9; 7.3%). Men had higher YLDs for schizophrenia, 2502 (20–24 years); bipolar disorder, 477 (40–44 years); and major depressive disorder, 2034 (75–79 years). Women had higher YLDs for schizophrenia, 484 (45–49 years); bipolar disorder, 214 (≥80 years); and major depressive disorder, 3541 (75–79 years). The prevalence-based approach and severity distribution is useful for estimating long-term psychiatric disease burden and YLDs. However, YLD-estimation studies must compensate for the shortcomings of the ICD-10 by referencing the Diagnostic and Statistical Manual of Mental Disorders 5th edition, as well as updating the disability weight score according to disease severity.


Introduction
The global burden of disease (GBD) study estimates that 7.4% of disability-adjusted life years (DALYs), 22.9% of years lived with disability (YLD), and 0.5% of years of life lost, globally, are related to mental and substance-use disorders [1]. In terms of disability ranking, major depressive disorder ranked first, schizophrenia was fifth, and bipolar disorder was sixth. Major psychiatric disease accounted for 3.3% of global YLDs [2]. According to the Korea National Burden of Diseases (KNBD) study, DALYs for mental and behavioral disorders accounted for 6.4% among all diseases [3]. Traditionally, the burden of diseases has been identified using an incidence-based approach, based on new diagnostic case units [4]. Since 2012, the Institute for Health Metrics and Evaluation has changed from the incidence-based approach to a prevalence-based approach. In addition, the method by which the disability weight (DW) is calculated was changed from specialized groups to the general population, and the distribution of disease severity has been added. The reduction for future health status and the weight for age were removed [5].
The burden of disease in the population can be calculated using the incidence or prevalence approach [6]. Australia and New Zealand changed their calculation method in the GBD 2010 [7,8] and have used the prevalence-based approach since 2007. While KNBD studies have hitherto calculated YLDs using the incidence-based approach [9][10][11][12], these values have recently been reported according to the prevalence-based approach. Gong et al. reported Korea's cancer burden using the prevalence-based approach [13]. Park et al. reported Korea's burden of injury by comparing a prevalence-based approach and an incidence-based approach [14]; the former provided data that were more appropriate for decision-making concerning healthcare resources [13].
In 2016, a KNBD study reported the DALYs of mental and substance-use disorders; however, the results were based on the incidence-based approach [15]. A Brazilian study calculated DALYs of mental and behavioral disorders using the prevalence-based approach and estimated the true global burden of mental illness in 2016 [16,17]. Their data have helped prioritize resource allocation and decision-making related to healthcare policies [18].
Estimating YLDs using the prevalence-based approach can yield data on the prevalence of psychiatric disease, which is needed to develop policies concerning primary or secondary prevention. Therefore, in this study, we set out to review the literature on this topic in different countries and use a prevalence-based approach to estimate the YLDs in Koreans with major psychiatric diseases.

Data Source and Definition of Major Psychiatric Disease
In this study, we used data from the Korean Health Insurance Review and Assessment (HIRA) Services on Healthcare Big-Data Hub. The Healthcare Big-Data Hub includes information on disease prevalence and the use of medical services (http://opendata.hira. or.kr/op/opc/olap3thDsInfo.do accessed on 24 June 2021). The data in the database can be categorized by sex and age, from the age of 5 years. This database has been increasingly used in healthcare research in Korea [19,20].
We obtained the prevalence of patients receiving treatment and diagnosed with specific major psychiatric disease-related codes from January to December 2018 from the Healthcare Big-Data Hub. We defined major psychiatric disease as a main diagnosis of schizophrenia and schizoaffective disorder, bipolar disorder, or major depressive disorder. We reviewed the data using the diagnostic codes (International Statistical Classification of Disease 10th edition [ICD-10]) defined in the GBD 2017 study as causes of death and nonfatal conditions [21]. The ICD-10 codes were as follows: schizophrenia (F20−F20.9), schizoaffective disorder (F25−F25.9), bipolar disorder (F30, F31−F31.9), and depressive disorder and recurrent depressive disorder (F32−F32.9, F33−F33.9).

Literature Review Regarding Major Psychiatric Disease
Diagnostic classifications, such as ICD-10 codes, involve a classification of diseases and clinical syndromes with consideration for the occurrence of disease. As a preliminary study on severity estimation based on the prevalence-based approach, we investigated the rationale for disease classification through the literature. Data were collected from original articles published in academic journals from 1-31 August 2020. The search was performed using the Medical Subject Headings database to select articles meeting the study keywords. The keywords were selected and the search strategy was built according to the participants, intervention, comparison, and outcomes (PICO) criteria. However, we only considered the participants, intervention, and research design in the review. The databases PubMed, Embase, and Google Scholar were searched using the following keyword search algorithm: "mental disorder" or "severe mental illness" or "schizophrenia" or "depressive disorder" or "depressive disorder major" or "bipolar and related disorder" or "bipolar disorder", and "ICD-10" or "International Classification of Diseases". The literature was limited to research studies that used ICD-10 codes and subdivisions. The classification of mild, moderate, and severe symptoms in the major psychiatric diseases according to the ICD-10 codes was reviewed twice by two psychiatrists. We confirmed the validity of code mapping in the originally assigned code. Finally, we estimated the severity distribution of major psychiatric diseases based on the classified codes.

Application of Disability Weight
YLDs are calculated by multiplying the number of patients by the DW of those sequelae. The GBD 2010 study used the incidence-based approach, focusing on disease and injury. The development of health state was not necessary since the DW was investigated by experts [22]. However, the prevalence-based approach focuses on the sequelae of disease or injury, allowing for the evaluation of the general population. Therefore, the severity of a condition can be reflected through the health state [23].
When measuring YLDs for mental and behavioral disorders, it is important to apply updated DWs according to disease severity in the country. In Korea, the burden of disease has been estimated using DW, which has recently been updated [23,24]. In this study, we used the health state for severity-specific states of mental illness. Furthermore, we calculated YLDs using the DW reported by Ock et al. [22]: The DW for acute state of schizophrenia was 0.836, while that for the residual state was 0.742; the DW for manic episodes of bipolar disorder was 0.658, while that for the residual state was 0.248; the DW of mild, moderate, and severe episodes of major depressive disorder were 0.551, 0.756, and 0.838, respectively [23]. We applied a DW of 0.742 for mild schizophrenia, 0.789 for moderate schizophrenia (average value of the acute and residual state), and 0.836 for the severe (acute) state. For bipolar disorder, we used a DW of 0.248 for mild, 0.453 for moderate (average value of a manic episode and the residual state), and 0.658 for severe (manic) bipolar disorder. For major depressive disorder, we used a DW of 0.551, 0.756, and 0.838 for mild, moderate, and severe episodes, respectively.

Computation of YLD
To estimate the YLDs of the disease, we used a cause-sequelae-health state scheme introduced in the GBD 2010 study. Sequelae are defined as the consequences of diseases, and the health state is designed to reflect the common sequelae [25]. YLDs were calculated by extracting the number of patients, based on ICD-10 code, with diagnoses of schizophrenia and schizoaffective disorder, bipolar disorder, and major depressive disorder. For each disease, the number of patients with mild, moderate, and severe, disease was determined from the ICD-10 codes. Moreover, after reviewing the severity distribution of the disease, we calculated the YLDs per 100,000 individuals according to severity by multiplying the number of patients in each age group and the severity distribution by DW.

Classification Review by Psychiatrists
ICD-10 codes classified for each disease based on a classification review by psychiatrists are shown in Table 2.

Estimation of Severity Distribution of Major Psychiatric Disease
Mild, moderate, and severe symptoms were observed in 13%, 46%, and 41% of the patients receiving treatment for schizophrenia, respectively; in 9%, 61%, and 30% of the patients treated for bipolar disorder, respectively; and in 43%, 40%, and 17% of individuals receiving treatment for major depressive disorder, respectively.
Severe schizophrenia involved a higher proportion of females than males, moderate bipolar disorder involved a similar proportion of males and females, and moderate major depressive disorder affected a higher proportion of females than males (Table 3). Table 3. Proportion of disease severity and sex in each group of major psychiatric diseases using ICD-10 codes in Korea.

Comparison of YLD by Severity of Major Psychiatric Disease
The YLDs for each major psychiatric disease are summarized by disease severity in Table 4  The YLDs due to schizophrenia were higher for females than for males. The YLDs' gap between males and females was 5.5 YLDs per 100,000 for mild disease, 15.0 YLDs per 100,000 for moderate disease, and 26.1 YLDs per 100,000 for severe disease. Bipolar disorder had higher YLDs in females than in males, and the YLDs' gap between males and females was 1.1 YLDs per 100,000 for mild, 22.6 YLDs per 100,000 for moderate, and 17.9 YLDs per 100,000 for severe bipolar disease. Major depressive disorder also resulted in higher YLDs in females than in males, and the YLDs' gap between males and females was 256.6 YLDs per 100,000 for mild, 378.3 YLDs per 100,000 for moderate, and 165.9 YLDs per 100,000 for severe major depressive disorder (Table 4).

Age Distribution of YLD for Each Major Psychiatric Disease among Males
The age distribution of YLDs for each major psychiatric disease among males is shown in Table 5 (Table 5).

Age Distribution of YLD for Each Major Psychiatric Disease among Females
The age distribution of YLDs for each major psychiatric disease among females is shown in Table 6. YLDs of schizophrenia was the highest (484 YLDs per 100,000) between the ages of 45-49 years, followed by 461 YLDs per 100,000 between the ages of 50-54 years, 455 YLDs per 100,000 between the ages of 40-44 years, and 455 YLDs per 100,000 between the ages of 30-34 years. The YLDs of bipolar disorder was the highest at 214 YLDs per 100,000 in the ≥80-years group, second highest (202 YLDs per 100,000) between the ages of 25-29 years, third highest (196 YLDs per 100,000) between the ages of 30-34 years, and fourth highest (182 YLDs per 100,000) between the ages of 20-24 years. The YLDs of major depressive disorder was the highest (3541 YLDs per 100,000) between the ages of 75-79 years, second highest (3259 YLDs per 100,000) between the ages of 70-74 years, third highest (2795 YLDs per 100,000) between the ages of 65-69 years, and fourth highest (2387 YLDs per 100,000) at ≥80 years (Table 6).

Discussion
Schizophrenia, depressive disorder, and anxiety disorder account for more than 70% of DALYs [35], indicating these psychiatric conditions as among the most damaging [36]. Using a prevalence-based approach, we confirmed that the YLDs of major psychiatric disease was high. The severity distribution analysis of major psychiatric disease revealed that most schizophrenic patients had moderate (46%) or severe (41%) disease, while only a minority had mild (13%) schizophrenia; almost two-thirds of the cases of bipolar disorder were moderate (61%), approximately one-third was severe (30%), and a small proportion was mild (9%); major depressive disorder was mostly mild (43%) or moderate (40%), while some patients had severe (17%) cases of the condition. In the severity and sex distribution analysis, schizophrenia and bipolar disorder were higher among females than in males, while moderate depressive disorders were higher among females. In the sex and severity proportion of major psychiatry diseases, schizophrenia, bipolar disorder, and major depressive disorders showed statistically significant relevance. In the prevalence study, schizophrenia was higher among females than males, and the prevalence of schizophrenia was higher in the age range of 45-54 years and 35-44 years [37]. The prevalence of bipolar disorder was higher among females (0.23%) than males (0.17%) [38]. Further, the symptoms of depression were more severe among females than males [39]. Additionally, a previous study that reported on the Diagnostic and Statistical Manual-5th edition (DSM-5) for evaluation of the severity of schizophrenia and bipolar disorder showed a severity proportion similar to that in our study [40]. The most widely used diagnostic manuals are the ICD-10 and the DSM-5, and these form the basis for mental disease diagnoses in most parts of the world. However, the comparison between the two manuals is limited because severity assessments should take into account the course of the disease and the level of treatment for the individual. YLDs for major depressive disorder were highest, accounting for 73.9% (1190.6 YLDs per 100,000) of total YLDs for major psychiatric disease, followed by schizophrenia at 18.8% (303.3 YLDs per 100,000), and bipolar disorder at 7.3% (117.9 YLDs per 100,000).
Among individuals with schizophrenia, moderate and severe cases accounted for 87% of the total population. Both males and females had higher YLDs in their 40s and 50s. Patients with schizophrenia reported a high rate of treatment discontinuation, low recovery rates, and difficulty in independent living [41]. In terms of the economic burden of schizophrenia in Korea, both males and females had the highest YLDs in the 40-49-year age group. Furthermore, the total death attributable to patients with schizophrenia in this age group was the highest [42]. Although this study reported higher YLDs in women than in men, no conclusions can be drawn from this finding, as sex-based differences in schizophrenia may vary depending on disease progression, hormonal differences, age, and sex behavior patterns [43]. In a Korean study on bipolar disorder, the prevalence was higher among females, and the rate of increase in its prevalence was high for those aged ≥60 years [38]. In another study on bipolar disorder in 2013, females had higher prevalence rates than males, accounting for 0.4% of the total DALY, 1.3% of total YLD, and DALYs in the 20-50 year age group [44]. In this investigation, we observed that a high proportion of individuals with bipolar disorder were moderately affected and that YLDs were higher among older adults (≥80 years). Bipolar disorder involves repeated episodes of mania and depression and is likely to recur [45]. In addition, continuous treatment and care can lead to the worsening of symptoms and are associated with suicide attempts [46]. Major depressive disorder had the highest YLDs among major psychiatric diseases, although the greatest proportion of patients had mild or moderate cases of the condition. Several previous studies reported a relationship between low quality of life and major depressive disorder. We confirmed that major depressive disorder mostly involved cases of low severity, but high YLDs. We surmise that this reflects a disease-induced low quality of life over a long period of time. Indeed, this finding reflects the utility of the prevalence-based approach in measuring the long-term burden of diseases [13]. Vigo et al. reported that approaches to diagnostic classification, such as the ICD-10, as methods for estimating mental illness need to consider the clinical syndrome and incidence [17]. In psychiatry, the patient's behavior and abnormal experience are diagnosed by subjective reports by the physician [47]. In this study, the severity classification using ICD-10 codes was evaluated by psychiatrists. However, classification codes determined through evaluation by physicians may have had a limited impact on the results due to underestimation. In addition, since most psychiatrists in Korea have been educated and trained with the Diagnostic and Statistical Manual-5th edition (DSM-5), the accuracy of assigning ICD-10 codes is unlikely to be high. This means that the severity of the patient's condition may differ from that indicated by the ICD-10 code. We speculate that the psychiatrists categorized the severity using ICD-10 codes to match the patient symptoms as closely as possible; however, in general, the severity of diseases changes with diagnostic codes according to the time point of the episode of diagnostic criteria. Therefore, this study was limited by the difficulty of accurately estimating the severity distribution among the psychiatric patients.
The DSM-5 diagnoses of patients with schizophrenia, bipolar disorder, and major depressive disorder have been validated in the reported Korean populations: 79.0%, 89.3%, and 86.9% for major depressive disorder, bipolar disorder, and schizophrenia, respectively [48]. This implies that symptom changes in patients with major psychiatric disease are not significant. In comparing the ICD-10 and DSM-5 for bipolar disorder, both systems showed some consistency in the diagnosis of bipolar disorder and related diseases. However, direct comparisons between the ICD-10 and DSM-5 were limited because there was a time gap of 20 years and a distinct difference in criteria, classification, and details between the systems [47]. Although, we point to a realistic problem arising from the difference between DSM-5 and ICD-10, the use of ICD-10 is not a major factor affecting the diagnoses of schizophrenia, bipolar disorder, and major depressive disorder. Therefore, we propose a future follow-up study that combines DSM-5 and ICD-10 to clearly classify patients' symptoms. Further, this study stresses the importance of physicians adequately recording the patients' symptoms to increase the reliability of YLD results using a prevalence-based approach.
DW is an important value for calculations in the prevalence-based approach. We used the DW reported by Ock et al. to calculate the YLDs for major psychiatric disease [22]. However, we applied the mean values for mild and severe DWs to moderate severity because the DW for moderate disease had not been reported. Recently, a study from Thailand reported DW to estimate the burden of disease in patients with major depressive disorder and alcohol disorder [49]. In future research, more detailed YLD calculations could be conducted to calculate the severity of DW separately for mental illnesses.
The severity of major psychiatric disease did not change significantly and is assumed to be maintained from the onset of the disease until death. In this study, we considered literature as a basis for the classification of major psychiatric disease severity, and psychiatrists reclassified ICD-10 codes based on the literature. However, major psychiatric disease can be assessed more reasonably through prevalence duration, drug use, and the number of hospitalizations. Future studies will require an evaluation that reflects the indicators affecting severity.
No study to date has classified the severity of major psychiatric disease according to the ICD-10 codes. The prevalence-based approach used in this study confirmed the severity distribution within the estimates under this disease classification system. We expect that this study will serve as a basis for future observation and evaluation of the epidemiological distribution of major psychiatric disease.

Conclusions
The GBD is a very important issue, and its significance will be emphasized more in the future. The prevalence-based approach was useful for estimating major psychiatric diseases' burden in South Korea. YLD estimation studies need to verify the credibility of clinical experts' ICD-10 coding and carefully classify patients' symptoms by matching ICD-10 and DSM-5 criteria

Informed Consent Statement:
The need for informed consent was waived by the board due to the nature of the study.

Data Availability Statement:
Publicly available data sets were analyzed in this study. This data can be found here: [http://opendata.hira.or.kr/op/opc/olap3thDsInfo.do accessed on 24 June 2021].

Conflicts of Interest:
The authors declare no potential conflict of interest with respect to the research, authorship, and/or publication of this article.