Validity of self‐reporting depression in the Tabari cohort study population

Abstract Aims Depression is a common cause of mortality and morbidity worldwide. To detect depression, we compared BDI‐II scoring as a valid tool with participants' self‐reporting depression. Methods The sample size was determined to include 155 participants with positive self‐reporting of depression in a total of 1300 samples with 310 healthy participants were included in the study through random selection. In order to evaluate the diagnostic value of self‐reporting, BDI‐II was completed by blind interviewing to the case group as well as to another group who reported that they were not depressed, as control. Results Sensitivity, specificity, accuracy, false positive, false negative, positive, and negative predictive values of self‐reporting were calculated 58.4%, 79.1%,73.4%, 20.8%, 41.6%, 51.8%, and 83.2% for the total population, respectively, as well as, sensitivity, specificity, accuracy, positive, and negative predictive values of self‐report in males were 83.3%, 77.2%, 77.1%, 43.8%, and 95.6% and 53.7%, 78.1%, 71.2%, 49.2%, and 81.1% for females, respectively. Conclusion The positive predictive value and sensitivity of self‐reporting are insufficient in total population and females, and therefore self‐reporting cannot detect depressed patients, but regarding to its average positive predictive value, perhaps, it can be used to identify nondepressant individuals.


| INTRODUC TI ON
The World Health Organization (WHO) has identified depression as the fourth reason of disability in the world, accounting for the greater portion of nonlethal diseases, and predicts it to be the second cause of death by 2020 [1][2][3] . In a review study, the prevalence of lifetime depression varied from 1.5 percent in Taiwan to 19 percent in Lebanon. The average in western Germany was 9.2 percent, and in Edmonton in Canada, it was reported at 9.6 percent 1 . An international research by the WHO, reported the prevalence of major depression in the general population to be from 1 percent in the Czech Republic to 16 The prevalence of depression in the Iranian adult population is assessed at 21 percent 4 . Regarding the high importance of this disorder, screening of this serious condition and timely management would be an important subject. There are several assessments for diagnosis of depression, namely Hamilton Depression Rating Scale (HAM-D), Zung Self-Rating Depression Scale 5 , Montgomery-Asberg Depression Rating Scale, HADS 6 , Geriatric Depression Scale, and the General Health Questionnaire (GHQ). They have few items for depression, except the HAM-D 4 , these depression assessment tools were developed as a measure of treatment outcome rather than a diagnostic or screening depression 1 . However, the Beck Depression Inventory (BDI) assesses both the psychosomatic and the physical symptoms, and its effectiveness has been discussed in many studies 7 . This tool has been used in more than 7000 researches so far.
The theoretical assumption of the BDI relied upon the negative believes that distorted cognition is the core of depression characteristic 2 . This inventory is a valuable instrument, with high reliability to discriminate depressed and nondepressed participants, and its content, structural, and concurrent validity have been approved 2 . This tool has been revised two times, and the latest version (BDI-II) was published in 1996 3 . The available psychometric evidence showed that the BDI-II could be noticed as a valid cost-effective inventory for measuring the depression severity, with wide applicability for research and clinical practice 2 .
We administrated BDI-II as screening tool for assessment of depression after the self-reporting of depression in the Persian cohort in Mazandaran, Iran. As in some studies, it has been indicated that the prevalence of depression measured through diagnostic scales by patients has been higher than the self-report results 2 , the researchers decided to compare the diagnostic value of the depression with BDI-II in Mazandaran's Persian cohort study with self-reported depression. The study's primary objectives were sensitivity, specificity, accuracy, false positive, false negative, positive, and negative predictive values of self-reporting, and the secondary outcomes were the association of depression according to BDI-II with sex, age, depression, depression in family in the case and control groups.
To our knowledge, this is the first study to compare the prevalence of depression with self-reporting and BDI-II as well as the first study to evaluate depression screening in a general population with the patients' self-report.

| ME THODS
In this cross-sectional study, we used a subset of data collected in Tabari cohort (Mazandaran's Persian cohort study), which is part of the national cohort, entitled as Prospective Epidemiological Research in Iran (Persian) 4,5 .
For conducting this study, 1300 participants with self-reporting depression interview, aged 35-70 years living in urban areas of Sari, Mazandaran, Iran, were enrolled. As part of data collection in Tabari cohort, a standardized questionnaire consisting of general information, socioeconomic status, and occupational history was completed. All the participants were asked a question" Are you depressed?". Among all the participants, 155 cases had a positive history of depression, which were selected as the case group. Among the remaining participants who did not report depression, 310 individuals were selected as control group randomly and matched in age and sex.
In order to evaluate the diagnostic value of self-reporting, BDI-II was completed to the case group as well as to another group who reported that they were not depressed.
Trained interviewers who were blind to the interviewees, dis- times, the sample size was estimated 155 participants that allocated through the census method. In order to increase the test power by 2 times, 310 healthy participants (by self-reporting) were entered the study through random selection (based on the available list) and the following formula:

| Statistical analysis
Data were entered into SPSS (version 22) software for statistical analysis. After filtering, the distribution of characteristics of the studied population was presented through descriptive tests such as frequency, mean, and standard deviation. Comparison between three groups for categorical data were statistically analyzed using chi-square or Fisher-exact test. Also, sensitivity, specificity, positive, and negative predictive value and accuracy of self-report method were determined. A p-value of 0.05 or less was considered significant statistically. Using IBM SPSS12 statistics version 23 and Stata version, the data were analyzed.

| Demographic information Questionnaire
This questionnaire included demographic information such as age, sex, and history of depression.

| The Beck Depression Inventory (BDI-II)
The BDI-II is a multiple-choice self-report inventory, consisting of 21 Weight loss, 20. Somatic preoccupation, and 21. Loss of libido 8,9 .
In this inventory, 4-6 questions are asked concerning each of the mentioned items based on one of the symptoms of the illness, ranging from the mildest to the most severe aspect of the mentioned attribute 9 .
The quantitative values of each item from 0 to 3 are determined as mild to severe disorder. Several forms of this questionnaire have been prepared. Here, the regular form includes 21 items 9 .This questionnaire is a self-assessment instrument and takes 5-10 minutes to complete.
It should be noted that, even though this inventory was designed for use in clinical populations 10 ; besides, it could also be used in normal populations 9 , 11 .

| Cut-off of BDI-II
The cut-off score for screening of depression varied according to the type of sample. In a study in Iran, the best BDI-II cut-off was 14, with sensitivity of 62% (95% CI (43%, 81%)), specificity of 81% (95% CI (72%, 90%)), PPV of 53%, and NPV of 85% 8 . The internal consistency was described as around 0.9 and the test-retest reliability ranging from 0.73 to 0.96 2 . Accordingly, in this study, a score of 14 was considered as the cut-off point for screening of depression. In addition, Table 1 shows the frequency of population characteristics in the case and control groups based on self-report; moreover, Table 2 shows the frequency of depression according to BDI-II (sex, age, depression, depression in family) in the case and control groups according to BDI-II, respectively.

| D ISCUSS I ON
In this study, the prevalence of depression was assessed blindly (being case or control) using BDI-II in two groups. According to the results, the sensitivity and specificity of self-reporting were found to be low, with many of the cases being found not depressed via BDI-II (Table 3).
It was concluded that self-reporting was not suitable for screening for depression in this population, and thus, there is a need to use a scale such as BDI-II as the gold standard 9 for depression screening.
Individual clinical interview is the "gold standard" for diagnosis of depression 14 . However, this approach may be problematic for screening of depression in large populations. In the Persian cohort study, the participants were asked only one question in this case, namely "are you depressed based on physician's opinion?". This study aimed to evaluate the diagnostic value of self-reporting compared with one of the most popular scales for depression screening.
The BDI is one of the most well-known tools for screening of depression in general population and psychiatric patients 14,15 . One of the problems of the BDI is that it did not completely include all of the symptoms in the DSM in depression criteria 16 . This revised instrument does not rely on any certain theory of depression 14 . The BDI-II has a good reliability and validity 10 . The correlation between BDI-II and BDI-I has been described strong 17 .
The correlation between BDI-II and BDI-I has been reported high 2 With respect to the Multiscale Depression Inventory as a "gold standard" 18 , the curve of receiver operational characteristic showed BDI-II to be an adequate diagnostic measure 19 and that the optimal total cut-off score was 18.5 17 With this cut-off score, 25% of multiple sclerosis patients were positively identified as having clinically relevant depression. The result of this study showed that the BDI-II is a valid, reliable, and simple tool for depression detecting and grading 17  BDI-II instrument depends on the clinical and social context of the assessment 24 . This questionnaire was used in our study for depression screening in the general population.
Concerning the self-report in the research, in a study, self-re-  as depression, is a barrier to self-reporting of these problems, and a valid and reliable instrument is required to be arranged and conducted for detecting depression. In addition, sensitivity in our study was low by self-reporting compared with BDI-II as a gold standard.
The positive predictive value and sensitivity of self-reporting are low, and therefore self-reporting cannot help in detecting depressed patients; however, concerning its average positive predictive value, perhaps, it can be used to identify nondepressant people.

| LI M ITATI O N S
The present study does have some limitations. First, it is difficult to launch a causal association in a cross-sectional study. Second, depression was measured by the BDI-II and self-reporting rather than a psychiatric structured interview. The BDI is a self-report questionnaire, which might underestimate a person's grade of depression. Moreover, many persons were excluded because they lacked a self-reporting depression. There was a possibility that those persons who were affected by mild or moderate depression did not identify as depressed person.

ACK N OWLED G M ENTS
The authors are grateful to The Iranian Ministry of Health and Medical Education, Mazandaran University of Medical Sciences, all the patients who participated in this study, and the research assistants and colleagues who kindly cooperated in the conduct of the study.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
MZ developed the original idea for the trial and attracted funding.

I N FO R M E D CO N S E NT
The clients were explained on the purpose and method of the study.
They were asked to complete a consent form.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly