Health Insurance

The process for non-fatal estimation begins with the compilation of data sources from a diverse set of possible sources, which include 21 possible Global Health Data Exchange (GHDx) data types ranging from scientific literature to survey data to epidemiological surveillance data. Our collaborator network provided 2842 data sources for GBD 2017. We analysed 21 100 sources of epidemiological surveillance data (country-years of disease reporting) for GBD 2017 and 4734 sources of disease registry data. For non-fatal estimation, we did systematic data and literature searches for 82 non-fatal causes and one impairment, which were updated to Feb 11, 2018. Search terms used for cause-specific systematic reviews, inclusion and exclusion criteria, preferred and alternative case definitions, and study methods detailed by cause are available in the supplementary methods (appendix 1 section 4). This search process contributed to the use of 15 449 scientific literature sources and 3126 survey sources used in non-fatal estimation, reflecting our updated counting criteria for GBD 2017. Household survey data archived in the GHDx were systematically screened together with sources suggested by country-level experts, surveys located in multinational survey data catalogues, and Ministry of Health and Central Statistical Office websites. Primary data sources containing disease prevalence, incidence, mortality risk, duration, remission, or severity were then combined in the estimation process. The supplementary methods section provides further details on gold standard data sources, adjustments, correction factors, and standardisations employed when incorporating these different types of non-fatal data (appendix 1 section 4).
In addition to data sources based on primary literature, surveys, and surveillance, the GBD study has used an increasing number of hospital discharge records, outpatient visit records, and health insurance claims to inform various steps of the non-fatal modelling process. This year, we received hospital discharge records for an additional 30 country-years, specifically discharge records from India (3 country-years), Iran (10), Japan (6), Jordan (1), Nepal (1), Brazil (2), China (1), and Italy (6); inpatient and outpatient claims from Taiwan (province of China); additional years of inpatient and outpatient claims from the USA; and inpatient claims from Singapore, representing an additional 148 842 107 hospital admissions globally and bringing the total number of admissions that inform GBD estimation to more than 2·6 billion. Additionally, we received 10 years of outpatient visit records from Norway, representing a total of 153 351 282 outpatient visits over a 10-year period. Overall, the study now uses hospital data from 335 country-years, outpatient visit data from 45 country-years, and health insurance claims data from 33 country-years between the USA, Taiwan (province of China), and Singapore. These data inform multiple cause models in various ways, mainly by providing incidence and prevalence estimates adjusted for readmission, non-primary diagnosis, outpatient utilisation, or a combination of the above, but also by estimating parameters such as case fatality rates, remission rates, procedure rates, and distribution of disease subtypes. The supplementary methods provide a more detailed description of how the clinical data adjustments are calculated and how admission and outpatient visit data are processed and utilised (appendix 1 section 2).
In the supplementary methods (appendix 1), we show the geographical coverage of non-fatal data, both incidence and prevalence, for GBD 2017. In addition, we illustrate the non-fatal data density and availability for GBD 2017 from 1990 to 2017 by GBD region and year for each of the three Level 1 GBD cause groups. The GHDx provides the metadata for all sources used for non-fatal estimation.

The following distinct sources of the SNIIRAM were used to select persons with depression:

Diagnoses of long-term or costly conditions (Affections de Longue Durée, ALD). Patients with specific long-term or costly conditions may require full coverage for all their condition-related health expenditures upon request by their family doctor and after approval by a health insurance fund medical officer (médecin-conseil) [18 ].

Data from national hospital claims (Programme de Médicalisation des Systèmes d’Information, PMSI) for all inpatient and day-case admissions in public and private general and psychiatric hospitals, containing medical diagnoses defined as ICD-10 codes. In both general and psychiatric hospitals, a principal diagnosis is defined as the main reason for admission, while associated diagnoses provide information about conditions that significantly influenced care during the hospital stay [19 ].

Data concerning all national health insurance reimbursements for drugs, laboratory tests and outpatient medical procedures. Individuals receiving reimbursements for antidepressants (N06A section of the ATC classification except for oxitriptan) can be identified. However, these databases do not contain direct information about the diagnosis justifying the prescription, and these drugs are not specific for depression, as they can also be prescribed for other conditions (bipolar disorders, anxiety or chronic pain). An antidepressant prescription is typically valid 1 month.

All three sources were not considered to be equally reliable for identifying patients with depression. Reliability of the sources was assessed as follows: for the purpose of identifying individuals suffering from depression, full coverage for depression as a specific long-term or costly condition (source 1) was more reliable than the hospital claims database (source 2), which was more reliable than reimbursement for antidepressants (source 3). In the hospital claims database, associated diagnoses reported during general hospital stays were assumed to be less reliable than those reported during psychiatric hospital stays. The reasons underlying this classification of source reliability included (1) the mode of acquisition of the information (diagnoses resulting from medical interviews were regarded as more reliable than hospital diagnostic codes sometimes coded by non-medical staff, themselves regarded as a more reliable diagnostic markers than prescription drugs) and (2) what was as stake when the information was coded (hospital diagnostic codes that had no consequence on costs were regarded as less reliable than codes influencing costs or giving access to benefits). These reasons are described and discussed more thoroughly in the Merits and drawbacks of the various methods section of the Discussion section of this article.
Accordingly, five estimation methods with decreasing order of reliability were defined. ICD-10 codes F32 to F39 were used in all estimation methods to identify depression (either as a full health coverage code or as a principal or associated diagnosis). At least three reimbursements for antidepressants were used to identify treatment by antidepressant. Hospital stays in the last 5 years with a principal or associated diagnosis of depression were used to identify principal diagnosis history and associated diagnosis history of depression respectively.

Method A (Full coverage for depression): Selection of individuals with full coverage for depression as a specific long-term or costly condition during the study (source 1);

Method B (Hospitalisation for depression): Selection of individuals with depression as principal or associated diagnosis in a psychiatric hospital stay or as principal diagnosis in a general hospital stay using two timeframes: (a) the current calendar year and (b) the last two calendar years (source 2). Calendar years were used for technical reasons.

Method C (Current antidepressant treatment + History of hospitalisation during the past 5 years): Selection of individuals treated by antidepressant and with a general hospital principal diagnosis history of depression or a psychiatric hospital principal or associated diagnosis history of depression (combination of sources 2 and 3);

Method D (Hospitalisation in a general hospital with an associated diagnosis of depression): Selection of individuals with depression as associated diagnosis in a general hospital stay using two timeframes: (a) the current calendar year and (b) the last two calendar years (source 2);

Method E (Current antidepressant treatment + History of hospitalisation in a general hospital with an associated diagnosis of depression during the past 5 years): Selection of individuals treated by antidepressant and with a general hospital associated diagnosis history of depression (combination of sources 2 and 3).

Individuals with a hospital diagnosis of bipolar disorder (ICD-10 codes F30 or F31) in the last 5 years or a specific treatment for bipolar disorder (lithium, divalproex or valpromide) were not included in the study.

We selected a series of control variables that may be associated with depressive symptoms, including demographic characteristics [36 (link)–38 (link)] (age, gender, marital status, residence, education), health status and health behaviors [39 –42 ] (self-reported health, activities of daily living scale (ADL), smoking, drinking, sleep duration, chronic disease status), and protective factors [31 (link), 43 (link)–46 ] (health insurance, pension, employment status). For age, we selected people aged 45 and above; for marital status, we reclassified them according to the answers of the questionnaire, and considered married and living with spouse, married but not living with spouse temporarily as married; separated and no longer living with spouse, divorced, widowed, and never married as unmarried. Educational attainment was classified into five categories: no education, elementary school, middle school, high school, and college and above. Since the sleep time showed a skewed distribution, we logarithmically processed the sleep time. For chronic disease prevalence, we divided the population into five categories: no disease, one chronic disease, two chronic diseases, three chronic diseases, and four or more chronic diseases. The detailed coding of the variables is shown in Table 1.Table 1

Coding of variables

Variable	Coding
Depression	< 10 = 0, ≥10 = 1
Levels of depression	0 ~ 30
WeChat usage	Not using the WeChat =0, Using the WeChat =1
Social participation	No = 0, Yes = 1
Levels of social participation	0 ~ 10
Voluntary activities	No = 0, Yes = 1
Levels of voluntary activities	No = 0, One kind = 1, Two kinds = 2, Three kinds = 3
Recreation	No = 0, Yes = 1
Levels of recreation	No = 0, One kind = 1, Two kinds = 2, Three kinds = 3
Cultural activities	No = 0, Yes = 1
Levels of cultural activities	No = 0, One kind = 1, Two kinds = 2
Other activities	No = 0, Yes = 1
Levels of other activities	No = 0, One kind = 1, Two kinds = 2
Age	≥45
Gender	Female = 0, Male =1
Marital status	Unmarried = 0, Married = 1
Residence	Rural = 1, Urban = 2
Education	No formal education = 1, Elementary school = 2, Middle school = 3, High school = 4, College or above = 5
Self-reported health	Very poor = 1, Poor = 2, Fair = 3, Good = 4, Very good = 5
ADL	No impaired = 0, Impaired = 1
Smoke status	Still have = 1, Quit = 2, No = 3
Drink status	No = 0, Yes = 1
Sleep time	Take the log of sleep time
Employment	No = 0, Yes = 1
Pension insurance	No = 0, Yes = 1
Medical insurance	No = 0, Yes = 1
Chronic diseases	No = 0, One kind = 1, Two kinds = 2, Three kinds = 3, Four kinds and more = 4

We used data from the China Health and Retirement Longitudinal Study (CHARLS) in wave 4 of 2018 for cross-sectional analysis. CHARLS aims to collect high-quality data on households and individuals aged 45 and older in China to analyze aging and promote research on healthy aging. CHARLS surveyed participants for basic information, health status, health insurance and health care, and retirement. CHARLS was approved by the Ethics Review Committee of Peking University, and all participants signed an informed consent form before the investigation and voluntarily participated in the survey [35 ].
The target population selected was people aged 45 and above. The missing rate of the dependent variable values was 18.18%, and 15,636 participants were included in the study after excluding those with missing key variable values. The total missing rate for the remaining values of all variables was 1.33%, so we directly excluded participants with missing variable values, and the final number of samples included in the study was 15,428. Data inclusion process are shown in Fig. 1.Fig. 1

Data inclusion process

The social cognitive theory of mass communication provides an agentic conceptual framework to analyze the determinants and psychosocial mechanisms through which symbolic communication influences human thought, affect, and action [15 ]. This theory emphasizes the various pathways in which communications systems operate to influence individuals’ engagement in health behaviors. In the direct pathway, health behavior changes are promoted by informing, enabling, motivating, and guiding participants. In the socially mediated pathway, media influences link participants to social networks and community settings that provide natural incentives and continued personalized guidance for the desired change. Structural interconnectedness provides potential diffusion paths; sociocognitive factors determine what diffuses through those paths.
The Theory of Gender and Power is a model that has been used to examine HIV-related exposures, risk factors, and effective preventive interventions for women [16 ]. This model describes three structures: (1) the division of labor manifests as economic exposures such as poverty, poor access to health insurance, being uninsured or underinsured, and being unemployed or having a high demand, low control work environment; (2) the division of power manifests as physical exposures, such as having a partner or partners at high risk of HIV acquisition, history of substance abuse, and limited perceived control; and (3) the structure of cathexis, which refers to social norms and affective attachments, manifests as social exposures such as the desire to conceive, and the lack of knowledge of HIV prevention. Three major social structures characterize the gendered relationships between men and women: the division of labor, the division of power, and the structure of cathexis (the process of allocating mental or emotional energy to a person, an object, or an idea). These theories informed how focus groups were conducted and how questions were dispersed in the web-based survey.