The study was performed at the Public Health Department of Manado State University, Indonesia, and included all the undergraduate students from year 1 to year 4 at the time of research. Permission to carry out survey was sought from the appropriate university authority (Surat Izin No. 031/6-IKM). Informed consent was obtained from the participants. To ensure reliable and generalizable results, a total sampling technique was applied by distributing the questionnaires to all the students after a class session, apart from those that were unwilling to participate; no credits nor punishment were given for agreeing or refusing to participate in the study. For the qualitative study, a group of students at the extreme end of late chronotype with high depression scores were interviewed.

To assess the chronotypes, we employed an Indonesian version of the MCTQ translated and validated in our center. We choose the MCTQ because it was shown to be sensitive in detecting chronotype shifts between high and low latitudes [13]. The MCTQ included questions on participants’ sleep duration and pattern during weekdays and weekends, and reported the sleep-deprivation corrected mid-sleep time at free days (MSFsc) and average sleep duration per week (SDw) as the main parameters. To assess the DAS level, a validated Indonesian version of DASS was used [14]. The DASS was chosen because it is reliable and easy to administer, especially in conjunction with MCTQ or other relatively long questionnaires, therefore preventing respondents’ survey fatigue [11].

Since there is a fundamental difference in the categorical interpretation of chronotype based on MSFsc between countries, geographical locations, and cultures, we decided to treat MSFsc as a linear scale, rather than categories (e.g. early, middle, or later phenotype). DASS were considered ordinal data as described [11]. MSFsc, the main parameter in determining the chronotype, is calculated by subtracting the mid-sleep time during free days from the sleep debt accumulated over a week, while SDw is the average of one week night’s sleep duration.

We conducted an in-depth interview to delineate whether chronotype shift preceded depression symptoms or vice versa. We only interviewed the subjects who fell into the extreme ends of both spectra, i.e. the later chronotype and high depression scores (n = 5). Interviews were performed either face-to-face or by telephone upon appointment. The questions were prepared beforehand but kept open-ended to facilitate data gathering concerning the phenomenon. The main questions were: 1) how long he/she has been having a late chronotype (“sleep late”); 2) whether the “sleeping late” preceded the feeling of being depressed; 3) was there any major event that influenced the sleeping pattern or induced depression; and 4) the difference between activities done before sleeping when they started to sleep late compared to when they slept “normally.”

Quantitative data were entered into a master table, recording both the MCTQ and DASS scores, and the demographic characteristics. After removing incomplete entries (n = 8), the statistical analyses were performed using SPSS version 20.0 (IBM Corp., Armonk, NY, USA). As appropriate, chi-square, Mann-Whitney U-test, and Spearman’s rank correlation tests were employed to calculate the statistical significance. Significance is set to be less than 0.05. Qualitative data such as the interview results were transcribed and analyzed qualitatively, by identifying common keywords and patterns among the respondents.

While there are many studies on mental health and chronotype, data from tropical and equatorial countries are lacking; to the best of our knowledge, so far, the comparable reports from the tropical areas came from limited sources such as Brazil and Taiwan, in which the chronotype and depressive symptoms were the main parameters [7,15-17]. Since the daylight duration is constant throughout the year in the equatorial area, the effect of seasonal sunshine deprivation on the mood may be negligible, thus rendering a very different pattern of the effects on mental health compared to the subtropical area. In the present research study, we replicated the findings of the previous studies, showing that even in the environment where the sun shines constantly, human chronotype varies, and more importantly, it correlates with mental health parameters, in particular with depressive symptoms. We observed a significant correlation between chronotypes (MSFsc and SDw) and depression score in both sexes, a phenomenon that has been well documented in studies on other subtropical areas [18-22]. It has been well established that seasonal affective disorders are related to the circadian cues from the sunlight, and many depression symptom outbreaks occur during winter, hence the term ‘winter blue’ [23]. The mood swing throughout the year may be a confounding factor in examining the depression symptoms, especially when using self-reported questionnaires such as DASS. Therefore, we thought that replicating such studies in an equatorial area may diminish the confounder and provide extended knowledge on depressive symptoms in relation to the other variables. Indeed, our current findings add to the available body of knowledge that, in the equatorial area where the seasonal daylight fluctuation is negligible, the correlation between chronotype and depression is preserved. It is notable that although there are strong correlations between DASS sections (D vs. A: R = 0.660, p < 0.001; D vs. S: R = 0.675, p < 0.001, A vs. S: R = 0.684, p < 0.001), the correlation with the chronotype is mostly existent in the depression section, which emphasized the validity of the currently used questionnaires in distinguishing subtypes of mental health symptoms.

There was a striking difference in the SDw between the present study and the previous studies. We found that our respondents had SDw of 6:44 ± 1:17 hours, lower than that reported in previous research studies from sub-tropical areas (multinational average around 7:48 hrs). However, on a closer examination, people from Asian countries have shorter SDw. Peltzer and Pengpid [24] reported a mean sleep duration of 6.59 hrs (range: 6.49–6.68 hrs) for Indonesia, comparable to Singapore (6.67 hrs), the Philippines (6.53 hrs), and South East Asia in general (6.82 hrs; range 6.78–6.87 hrs). In addition, Romadhon and Abdussalaam [25] reported that sleep duration in the Moslem community in Indonesia is around 5:58 to 6:15 hours with substantial variation across ages, thus adding a layer of complexity in determining a nationwide SDw. However, on top of this unique relatively short SDw, the effect of chronotype on depression symptoms is preserved.

Across different cultures and countries, females have consistently been shown to be more likely to develop depression and anxiety disorders [26-28]. Our results also showed that female participants consistently scored higher in all parameters of DASSs as measured by DASS. The cause of this skewness across the literature has never been satisfactorily explained. Some explanation comes from socioeconomic factors [28] and the observation that females are more vulnerable to past adverse experiences [29]. Notably, our data showed that females relatively possess an earlier chronotype than males, despite having the same sleep duration. We would address this chronotype difference based on previous observations that females tend to sleep and wake up earlier than males [30,31] and the effect of testosterone in shaping late chronotype among males [32]. Thus, there seems to be an intrinsic uniqueness of self-perception of being depressed, anxious and stressed among females that cannot be explained by chronotype alone. Further research is necessary to confirm this. Some prospective research directions include providing different scoring criteria for women or tailoring specialized questionnaires based on gender.

Based on our findings, the respondents who scored high on the depression scale were unaware of being at the extreme severity scale during the qualitative analysis interview. Since DASS measures the symptoms of depression—instead of acting as a diagnostic tool as mentioned in the Diagnostic and Statistical Manual of Mental Disorders, there is a possibility that this could result in the inaccuracy of the self-reporting process. However, as discussed above, given the intrinsic validity of the DASS questionnaire between sections, it is unlikely that this unawareness could result from false-positive reporting. Instead, we would propose that subclinical depression may be imminent on these subjects, which is a predictor of future major depressive disorder [33,34]. In addition, the fact that these respondents are on the extreme late chronotype, as supported by previous research, places them at a high risk of developing depression [18-20,22].

While our qualitative study is not sufficient to conclude that smartphone use is responsible for chronotype lateness and depression, it is well known that gadgets such as smartphones or tablets emit blue light that disrupts the circadian system putatively by interfering with the melatonin secretion [35]. Eventually, this lead to an increased incidence of a wide range of psychiatric and somatic disorders [36]. Sadly, gadget use is capable of inducing habituation and addictive behavior especially among youths and undergraduate students [37,38]. Furthermore, gadget use, as many other types of addiction, can provide a dopamine release in the brain reward system, thus providing a temporal release of stress, anxiety, or even subclinical depression as discussed above. In combination, this can lead to a vicious cycle of addiction-depression-coping that may need to be professionally intervened. This highlights the need for an initiative for controlling such behaviors and more effort to increase the student’s awareness of the imminence of gadget use. It has also been demonstrated that chronotype disturbance is related to a wide range of somatic diseases such as cancer, cardiovascular, and metabolic disorders [39], thus adding more reasons to include circadian rhythm management as a preventive measure against future pathologies.

Our present findings highlight the positive correlation between depression severity and chronotype, as well as a negative correlation between depression and sleep duration. While these three variables are not necessarily dependent, late-onset and short duration of sleep may reflect a low sleep quality. Thus, one of the immediate follow-up to this finding, and may be a future research topic, would be to evaluate whether depression indeed correlates with sleep quality. In addition, in the future, it is important to incorporate the circadian biomarker (e.g. plasma melatonin) examination in the subjects with extreme chronotype and perform experimental corrective studies such as assisted-chronotype restoration. While the present research study involved a relatively small population from a specific, homogenous group (college students) it may represent the societies in the equatorial area. However, further replication studies would be necessary to test the generalizability.

There is a correlation between chronotype and depression severity. Subjects with later chronotype tend to have higher depression score without being necessarily aware of it. The use of smartphones during or before bedtime may contribute to the shifting of the chronotype towards a later chronotype.