Validity of the Self Reporting Questionnaire-20 for Depression Based on National Health Survey


 Background: Adequate data is required to assess the validity of mental emotional disorder for depression based on Indonesia Basic Health Research. The purpose of this study is to assess the validity of mental-emotional disorder using Self Reporting Questionnaire-20 (SRQ-20) to the depression evaluated through the means of a MINI (Mini International Neuropsychiatric Interview) questionnaire. Methods: The data were obtained from a total sample of 555,066 subjects analyzed from the National Health Survey (NHS) in 2018. These subjects were at least 15 years old with their mental-emotional disorders and depression assessed by using the SRQ and MINI, respectively. This study used the calculation of sensitivity, specificity, positive and negative predictive value, positive and negative likelihood ratio, ROC (receiver operating characteristic), kappa Brennan, and Prediger with the STATA statistical program version 15.00 to analyzed the data. Results: The cut-off point in each group was different, ranging from 4 and 6 that almost all of the Area Under Curve (AUC) values were above 0.90 and the SRQ agreement with MINI depression are good because they all have values above 0.80. Conclusion: The results obtained are used as material to predict the rate of depression in Indonesian residents aged ≥ 15 years.


Introduction
Depression is a disorder discovered to be contributing to the burden of disease and Disability-Adjusted Live Years (DALYs).
Moreover, certain improvements have been reported about its contribution to DALYs [1], as observed in its 20th position in 2006 and 19th in 2016 [2]. Burden of disease are caused by the reduction in the productivity of life of patients and the severe circumstances accompanied by the threat of suicide. Depression symptoms are expected to be exhibited by someone experiencing a sad event, but when drags on for a long time, it causes interference or even death [3] The continuous increase in the burden of disease and the threat of death associated with suicide has led the Indonesian government to conduct a survey to obtain data on the prevalence of depression disorder in the country. However, the National Health Survey (NHS) conducted in 2007 and 2013 only displayed the data on the mental-emotional disorder or psychological distress which is also called Common Mental Disorder (CMD) that are assessed using self-reporting questionnaire-20 . The term "mental-emotional disorder" has been used in the Indonesian household survey since 1995 and has continuously being applied even though it is not completely right. Meanwhile, distress is known as the neurotic state that causes anxiety and depression for people in the society. However, SRQ-20 is not the speci c instrument to measure depression and this means that a special measurement is needed for the next NHS.
The assessment of the depression in Indonesia NHS 2018 was conducted using a special questionnaire obtained from mini international neuropsychiatric interviews (MINI) [4][5][6], and the results showed the prevalence among the population aged ≥ 15 years was 6.1% [7]. This tool has also been applied in the mental health surveys of several other countries such as India [8].
Also, some mental emotional disorders apart from depression were assessed in NHS 2018 using SRQ-20 due to its more practicality [9].
Therefore, the information above shows SRQ-20 as easy-to-use measurement tools that consider the answers designed to be yes or no but it does not provide speci c results to determine if an individual has depression or anxiety. In contrast to the use of MINI, this questionnaire requires good training for enumerators to determine the right answer for the respondent besides it has been discovered to be the speci c measurement tool to assess depression.
The health program implementers and practitioners expect Indonesian to have a speci c or diagnostic number of mental disorders. This can only be produced through surveys with diagnostic instruments but not possible in the country because it consists of many islands and geographic di culties, in the NHS 2018. Also, it is allowed to assess with two measuring instruments, namely the SRQ-20 and MINI, but only one for the next NHS for the fact that an extrapolation tends to be needed to answer it needs in the future. Therefore, this analysis was conducted to determine the validity of mental-emotional disorder assessed by SRQ-20 to the depression evaluated using the MINI questionnaire through the use of the NHS 2018 data.

Method Study Design
This study was a sub-analysis of Indonesia NHS 2018 conducted in all the provinces and districts or cities in the country.
The NHS sample was carried out using two stages that in the rst one, implicit strati cation, probability pro-portional to size (PPS) were carried out to determine the census block while in the second stage, 10 households in each census block were selected using systematic sampling. The method used in this study was written in the NHS's report [7]and socioeconomic level data were retrieved from the National Socio-Economic Survey (Susenas) conducted by the Central Statistics Body and integrated into the NHS. Therefore, the data used was sourced from the NHS in 2018 that analyzed a large number of subjects.

Subject
The respondents used were selected based on certain inclusion criteria which include being at least 15 years old, physically available, has the ability to answer all the questions, and not being represented by other people when answering the questions. The sample framework used in the NHS was the 2018 National Socioeconomic survey and the data were obtained from the NHS questionnaire.

Self Reporting Questionnaire
Mental Emotional Disorders were assessed using the 20 questions originally formulated in Self-Reporting Questionnaire (SRQ) [10] and the cutoff point for Indonesia requires six "Yes" answers [11]. The SRQ was conducted since 1995 and has been periodically used in national surveys that assessed neurotic disorders which is more popular with the term mental emotional disorder in Indonesia within the last 1 month. However, it was recommended by WHO as a mental disorder screening tool for developing countries, and has been proven to have good face, content, criterion, validity, construct validity and also used in various surveys and research studies [10].

Mini-International Neuropsychiatric Interview (MINI):
Depression was evaluated using questions based on module of the Mini-International Neuropsychiatric Interview version 6 [4][5][6]. The MINI is a interviewed diagnostic tool assessing depression in past 2 weeks or life time. In this study, the depression module from MINI was used as a diagnostic reference (considered the gold standard). Depression module can be seen Appendix 1. Both SRQ-20 and MINI questionnaires were read out by enumerators who had previously been trained on the content of the questionnaire and interview techniques.

Data analysis
The SRQ-20 in this study was applied as a screening test for validity while MINI depression was used as a reference. The sensitivity, speci city, true and false positive, positive and negative predictive value, positive and negative likelihood ratio, ROC (Receiver Operating Characteristic), PCA (Principal Component Analysis) and Kappa Brennan, and Prediger [12] were calculated using the statistical program, STATA version 15.00.

Subjects
Following the inclusion criteria, a dataset consisting of 560,472 respondents was obtained but a total of 5,361 were excluded because they were represented by other persons at the time of the interview. Also, 45 others were excluded due to the provision of incomplete answers related to mental-emotional disorders or depression. Therefore, the total number of those used as samples was 555,066 subjects and their characteristics are presented in Table 1.

Analysis of the mean scores
The mean score of SRQ-20 for the entire population in Indonesia was 1,857, at a 95% con dence interval (CI) 0.004-1.865 and those identi ed not to be depressed by the MINI questionnaire had a mean score of SRQ 1.431 (95% CI 1.425-1.437) while those con rmed depressed had 8,677 (95% CI 8.629-8.725). The quality of the measurement instruments used was assessed through several validity parameters as shown in Table 2 and several cut-off points were simulated to obtain the best from the results.

Prevalence
The table above shows a lower SRQ score that produces a higher sensitivity and NPV while those with higher scores gave higher speci city and PPV. Moreover, the positive likelihood ratio (LR) was observed to be higher at lower SRQ scores and vice versa. There are various cut-off point shown in Table 2 which tends to be used as a reference for various purposes. Subsamples based on the characteristics of the respondents such as male, female, age group and, also the resulting ROC shows almost the same shape. Figure 1 -5 are ROC curve of the SRQ-20 score on depression that shows high sensitivity and speci city but these two parameters are not the determinants required to obtain the optimum value of 6 used as the cut-off point for depression in SRQ-20 and also, the area under the curve value was found to be 93% and well based on the analysis.

Factorial validity
In PCA, 2 factors that form SRQ are found, as shown in Table 3. These are factors that have an Eigen value> 1.
This study identi es factor 1 in the table above as depression, while factor 2 as a symptom of somatic and anxiety. The depression factors consist of thinking clearly, unhappy, cry more, di culty to enjoy an activity, di culty to make decisions, daily work suffering, unable to play a useful part, losing interest, and worthless person. Meanwhile, the somatic-anxiety factor consists of poor appetite, sleep badly, easy frightened, handshake, tense worried, and poor digestion.

Discussion
The results showed various validity parameters at various cut-off point and several groups and it can be seen that the SRQ has good validity against depression as assessed by the MINI. The optimum cut-off point was different for each group and that the SRQ agreement with MINI depression is good because almost all have values above 0.80. Based on PCA analysis, there are two structural factors on the SRQ, namely the depression and a mixture of somatic and anxiety. The prevalence of depression obtained by 5.9% was slightly different from the NHS 2018 report, which was 6.1% because the sample analyzed only the subjects that answer the questions themselves, while the NHS report still included subjects represented by a companion or caregiver [7].
The resulting ROC for total subjects, male, female, young, and old shows a similar image with AUC values and they are all above 0.90. This AUC value is even higher than previous studies in China, India, Vietnam, and Afghanistan which only had an AUC of around 0.80, even though it uses different reference instruments such as the Clinical Interview Schedule Revised (CIS-R), Composite International Diagnostic Interview (CIDI), Hopkins Symptom Checklist-25 (HSCL-25) [13][14][15][16]. Research using the same questionnaire such as SRQ and MINI was conducted in South Africa with an emergency department setting, the results also showed the AUC value of around 0.80 [17].
Regarding the forming factors, research in South Africa also showed 2 factors as depression-anxiety and somatic, while research in China showed 3 factors, including depression, anxiety, and somatic. In this study, there were two factors formed, and they include; speci cally anxiety-somatic and depression. This shows that there is not much difference with other populations in various countries in terms of the factors forming the SRQ.
Various validity parameters have been provided at various cut-off point and in groups by gender and age by showing the AUC. The agreement between SRQ and MINI depression on this nding is good because the values are above 0.70 and 0.80.
These results can be used as a basis for extrapolating, for example, the previous NHS data will be able to calculate the depression rate. A more important need is that the future NHS will also be able to obtain depression rates in the population if they tend not to use the MINI depression mode anymore. Regarding the cut-off point, all groups showed that the optimal cutoff point was between 4 -5 and this is slightly different from previous studies which determined it to be about 5 or even 6 (in women) for any diagnosis [17].
This validity assessment procedure has been studied and the short and easy questionnaires were shown to be preferable to special ones requiring special skills from the enumerator. These methods have been used in other studies, for example, a patient care questionnaire (PHQ) which was validated with MINI [18,19]. Besides MINI, the Beck Depression Inventory (BDI) has also been used as a reference by several studies [20]. Moreover, the Center for Epidemiological Study for Depression questionnaire (CES-D) have also been used with certain populations according to the target of the study [21,22]. SRQ has been used as a CMD screening tool, for example, in Eritrea and Afghanistan [23,24]. Therefore, this means that both SRQ and Patient Health Questionnaire (PHQ) are good enough but require con rmation for positive screening results from health care facilities [25]. A prevalence of 5.90% and 9.88% was obtained for depression and mental-emotional disorder respectively. However, a survey conducted in Iraq with the use of SRQ-20 as initial screening produced a CMD percentage of 35.5% [26]. In Mexico, SRQ was also used to assess people to be deported for violating the border area and the CMD was found to be 16%. [27] The variation in these results despite the use of the same instrument is associated with the characteristics and conditions of different populations. However, the use of MINI in measuring the prevalence of depression in India was able to produced 2.68% [28].
The validity assessed is a diagnostic test or a screening tool which is also referred to as convergent and concurrent validity in other studies [29]. Meanwhile, some qualitative research has also been conducted to support the assessment of validity [30] with the focus on the appropriateness of tools, processes, and data [31].
The strength of this study was the use of a large sample which is an indication of high representation of the Indonesian population and this means it has the ability to re ect the real situation of the country. However, it was limited for the fact that NHS does not only assessed mental health, but various diseases and matters related to the subject's health. Through all the questions asked in this study, there is the potential for fatigue in both enumerators and respondents.

Page 6/12
The ndings showed SRQ-20 with cut-off point ≥6 (cut-off point that has been used by the NHS) is able to screen depression, especially to detect those that are not depressed with a speci city of 94.12%. However, a person with an SRQ-2O score <6 tends to be 98.3% of depression with MINI questions and it is therefore, recommended that when using a cut-off point ≥ 6, it will obtain sensitivity at 73.75%, speci city at 94.12%, positive predictive value at 44%, and negative predictive value at 98.3%.
The implications for survey policies in Indonesia in the future are to make it possible that MINI is used to assess depression like the survey in India, but if this is not possible then the results of this study can be used as a material to predict the rate of depression in Indonesians aged ≥ 15 years. Authors' contributors SI was the study coordinator, proposed the data, and drafted the manuscript. IY represented and helped SI. RM analyzed the data. LI and ID prepared the data set at the beginning of the study.