Depression is a major burden in modern post-industrial countries (1). In Germany, only approximately a third of the people with depressive symptoms seek mental health care and usually with considerable delay after symptom onset (2). This issue is found in many countries with well-developed primary care, even when there is a specialised mental health care system (3). Furthermore, during the Covid-19 pandemic rates of depressive symptoms became more prevalent, exacerbating the issue (4).

Biddle et al. (5) defined help-seeking for psychological complaints as a social process whereby the focus lies on how symptoms are interpreted and managed by individuals and their communities. They argued that in a “cycle of avoidance” [(5), p. 988] the layperson will renegotiate what is normal so as to avoid the label of “illness” as long as possible and postpone seeking professional health care, wanting to manage the illness by themselves.

Avoiding the label of mental illness is understandable when considering that those labelled as mentally ill may experience stigmatisation for their mental illness (6, 7) and are often seen as dangerous, unpredictable, and unreliable (8). Mental health-related stigma is part of a network of beliefs deterring professional help-seeking behaviour. In a systematic review, Clement et al. (9) found that when all types of barriers are considered, anticipated stigma, especially regarding disclosure and confidentiality concerns, is the fourth highest barrier and has a negative effect on help-seeking (d = −0.27, (9), p. 15). Another study reported that the largest reported treatment-seeking barriers are wanting to handle the problem on one’s own and a low perceived need for care (10).

To overcome stigmatising attitudes and empower help-seeking, we suggest that it is essential to strengthen an individual’s biopsychosocial causal and self-efficacy beliefs. We have defined help-seeking as a process with four consecutive steps, from symptom awareness, over self-identifying as having a mental illness, to help-seeking intention and, finally, behaviour (11).

Depending on what a person believes to be the cause for their mental health complaints, they are more or less likely to seek help. For example, a biogenetic explanation will increase the likelihood that a person seeks help from a general practitioner (12) or a psychiatrist (13) compared to a psychosocial explanation. However, the biogenetic causal point of view is debated when considering its possible effect on stigmatising attitudes (14). On the one hand, biogenetic beliefs (e.g., heredity) have been found to reduce some aspects of stigma [e.g., (15)] because they imply that the person is not to blame for their illness. On the other hand, biogenetic beliefs facilitate an internal and stable illness attribution (14, 16). This essentialist attribution not only increases beliefs in dangerousness and differentness (17), it will also likely produce the idea that therapeutic help is not effective as the cause is assumed to be too stable to change, reducing help-seeking behaviour.

In contrast, a balanced biopsychosocial model from the perspective of vulnerability-stress research shows promising results both to counteract stigma (15, 17) and facilitate help-seeking intention (18). Such a model encompasses biogenetic as well as psychological and social causal beliefs and allows a person to regard the cause of their issues holistically and not just as the result of one (type of) cause (e.g., biomedical).

Causal beliefs can be seen as conceptually related to another important predictor for help-seeking behaviour: agentic self-efficacy (19). The Attribution Theory (20) with its three main attributional dimensions, namely, “stability,” “locus,” and “control,” implies this. For example, if a person believes a cause for their depression to be their “weak will” (i.e., stable, internal, and non-controllable), then they are also more likely to have lower self-efficacy beliefs and vice versa. Research on self-efficacy has demonstrated that people are more likely to engage in certain behaviours if they believe their efforts will be successful (19, 21). This has been shown for a variety of health behaviours (22–24). Additionally, findings support the notion that general self-efficacy is a universal construct yielding significant relations with other psychological constructs (25). Because help-seeking behaviour for mental illness is a complex form of psychosocial health behaviour, involving planning and execution, the role of self-efficacy has to be considered within this context (5, 26, 27).

Furthermore, self-efficacy should be considered domain specific (21) and will be operationalised as such in the current study. The somewhat complementary self-efficacy forms under observation pertain to self-help and seeking professional help. First, the specific form of self-efficacy to self-help was included in the study because previous research indicated that it could negatively influence the help-seeking process, specifically the intention to seek help (27). Furthermore, it is rarely examined separately from general self-efficacy. Only two studies were identified in which similar task-specific self-efficacy was investigated (28, 29).

Second, the specific form of self-efficacy to seek professional help was included because we suggest that it could help bridge the “intention-behaviour gap” (23). This “gap” describes the problem that people do not show help-seeking behaviour despite their intent to do so, even when they have positive attitudes towards help-seeking and few perceived barriers [e.g., (30, 31)]. One reason for this gap is mental health stigma, which is negatively associated with self-efficacy [e.g., (32)], wherefore it seems prudent to try and bolster self-efficacy beliefs, specifically concerning help-seeking behaviour, as a way to support health care utilisation.

In this context, previous treatment experience was identified as an important distal predictor of help-seeking, positively influencing self-identification as having a mental illness (15). However, what possible impact satisfaction with previous treatment has on help-seeking has not been considered in the context of different stigmatising attitudes, causal beliefs, and self-efficacy beliefs.

The aim of the current study is twofold and corresponds to the three primary research questions formulated in the study protocol (11). First, we analysed the effects of informational, vignette-based, and online-administered interventions, respectively, endorsing a balanced biopsychosocial causal model of mental illness, self-efficacy to self-help, and self-efficacy to seek professional help. Second, and in addition to the study protocol, we investigated the conceptual associations and predictive values of these variables on help-seeking intention and behaviour. Because of the conceptual relatedness between causal and self-efficacy beliefs, possible interactions were considered within the analyses. The respective research questions were:

Does a balanced biopsychosocial causal model of depression positively influence help-seeking intention?

In this article, we examined how causal and self-efficacy beliefs facilitate help-seeking intention and behaviour, both through interventional manipulation and by analysing their conceptual associations. Specifically, how and in what ways help-seeking is influenced by causal beliefs and self-efficacy beliefs and how these beliefs interact with each other. We also examined how previous treatment experience and the satisfaction therewith influenced the different process variables.

Does a balanced biopsychosocial causal model of depression positively influence help-seeking intention? The intervention on a biopsychosocial causal belief model had no significant effect on help-seeking intention, neither by itself nor compared to the control groups. However, a significant effect on intention emerged when including interaction terms for both self-efficacy interventions. While this indicates that this specific intervention did not have an effect on the biopsychosocial diversity of a person’s causal beliefs or on the intention to seek professional help, it also points to the potential of more complex interventions that consider additional aspects, such as self-efficacy. It should be noted that the control groups were active controls that received a similar combination of other interventions [e.g., depression literacy, which is conceptually closely related to causal beliefs; (40)] even if they did not receive the causal belief interventional message. Additionally, participation in such a study could have heightened awareness of one’s own symptoms and started reflection processes irrespective of group allocation.

However, on a conceptual level, there was a clear association between a more heterogeneous belief system and the intention to seek help from a professional, as indicated by significant correlations (see Supplementary Table S3) and regression weights. Interestingly, these associations were stronger for the intention to seek help in a counselling centre or from a social worker than from a GP or an MHP. Because the BPS-CM index measures the heterogeneity of a person’s causal beliefs, those with higher scores were more likely to believe that biological, psychological, and social causal antecedents could be responsible for their issues. As the literature shows, this leads to reduced help-seeking from GPs and MHPs (12, 13), possibly explaining why the associations were weaker for these more formal health care sources. Furthermore, this could lead to an openness to seek help from a source outside of the health care system, explaining the predictive value of β = 0.23 with the intention to seek help in a counselling centre or from a social worker. The associations remained significant even in the context of other predictors, indicating a robust finding. The research question was answered insofar as that a more balanced biopsychosocial causal belief system positively predicts help-seeking intention. However, the experimental manipulation did not work as planned, i.e., compared with the active controls.

Does a higher self-efficacy to self-help negatively influence help-seeking intention? The intervention in self-help strategies had a small, positive effect on help-seeking intention (both from a GP and from MHPs). This was surprising and counter-hypothetical. However, even though the items had a very good internal consistency they were not psychometrically validated and their sensitivity to change was unknown. On the other hand, the measure correlated strongly (r = 0.48) with the second self-efficacy measure used in this study, indicating construct validity. In addition, the measure correlated negatively with depression severity, which was also in line with self-efficacy literature (19, 25). This taken together pointed to a robust measure and, therefore, indicated that the intervention did in fact have an effect on the intention to seek help even if not on the efficacy beliefs themselves.

Additionally, on a conceptual level, there were clear positive associations between self-efficacy to self-help and different intentions to seek help, which again contradicted the research question. However, the correlations (see Supplementary Table S3) were small and, when considered in the context of other predictors, the values were non-significant (with the exception of β = 0.07 on intention to seek help in a counselling centre or from a social worker). Interestingly, the interaction term between both self-efficacy beliefs had a negative predictive value β = −0.10 on the intention to seek help from an MHP. After considering this, the same multiple regression analyses were conducted, but this time separately for three levels of depression severity, namely, mild (PHQ-9 score 8 to 10), moderate (PHQ-9 score 11 to 15), and moderately severe/severe (PHQ-9 score 16 to 27), according to Kroenke et al. (48). The results of the multiple and logistic regressions can be seen in the supplementary material (Supplementary Table S4). The interaction effect disappeared in the groups with moderate or moderately severe/severe depression and was only significant in the group with mild depression (β = −0.17). Self-efficacy to seek help became increasingly important with greater symptom severity (mild: β = 0.14, moderate: β = 0.16; moderately severe/severe: β = 0.27). We interpreted these findings to indicate that self-efficacy to self-help is more relevant for participants with mild depressive symptoms and will inhibit help-seeking intention, which is plausible, when considering that one of the largest reported treatment seeking barriers is wanting to handle the problem on one’s own (10). With increased symptom severity, self-efficacy to self-help loses relevance and self-efficacy to seek help becomes increasingly relevant, assumedly because the need for help becomes more pressing with higher illness severity. However, because the analyses were explorative and not confirmative, further research is needed to elucidate these interactional findings of how self-efficacy influences help-seeking intention. The hypothesis was rejected.

Does a higher self-efficacy to seek help positively influence help-seeking intention and does it make help-seeking behaviour more likely? The intervention on help-seeking strategies had a small, positive effect on help-seeking intention (both from a GP and MHPs). It had a significant effect on its target variable self-efficacy to seek help (d = 0.14), and when participants received all three intervention elements there was a small significant effect on intention to seek help (both from a GP and MHPs). Especially considering that the other two interventions (i.e., on causal beliefs and self-efficacy self-help) did not influence their intended target variable it is likely that the intervention designed to influence self-efficacy to seek help is the main contributor to the effect on intention. In addition, the interventions were very short (M_{word count} = 170) and were administered online. Because of the online setting and the short and informational character of the intervention we only expected small effect sizes (11). Taking this into account, the results indicated that even a very low threshold, namely, an online intervention designed to strengthen self-efficacy to seek help, especially in combination with additional informational messages, such as biopsychosocial causal beliefs and vulnerability-stress, can strengthen self-efficacy beliefs and support a person in forming their intention to seek help from a GP or MHP.

On a conceptual level, there were clear positive associations between self-efficacy to seek help and different intentions to seek help (see Supplementary Table S3). These correlations remained significant predictors when considered in the context of other variables such as self-efficacy to self-help or stigmatising attitudes. Therefore, the research question was answered insofar as self-efficacy to seek help positively predicted help-seeking intention.

On the other hand, help-seeking behaviour was not predicted by self-efficacy to seek help, the main predictor being the respective intention to seek professional help, as has been discussed in current research on the “intention-behaviour gap” (23, 31). It could also be that people with depressive complaints not only suffer from reduced self-efficacy (49) but that when seeking professional help other factors such as perceived accessibility to the health care systems are of greater importance than their own self-efficacy beliefs.

The interaction effects for different levels of depression severity (see Supplementary Table S5) were difficult to interpret because there were no main effects, which is why further research is needed to explore possible negative effects self-efficacy might have on actual help-seeking behaviour. The hypothesis was, therefore, rejected concerning actual behaviour but could be accepted concerning help-seeking intention.

Is treatment experience positively associated with both self-efficacy beliefs, as well as help-seeking intention and behaviour? In addition, how is satisfaction with previous treatment associated with these variables? The results differed depending on whether treatment experience as such was examined or the satisfaction therewith.

Concerning self-efficacy, treatment experience did not seem to influence self-efficacy beliefs regarding self-help or seeking help. There was no significant difference between the two groups (see Table 1). However, the significant, positive correlations of satisfaction with previous treatment experience with both self-efficacy beliefs indicated that there was a subjectively felt quality of therapy that influenced self-efficacy (see Supplementary Table S3). Possibly, higher satisfaction with previous treatment reflected better therapeutic results, which in turn showed itself in higher self-efficacy beliefs.

Concerning help-seeking intention, the findings clearly indicated that treatment experience was an important factor. People were more likely to have a higher intention to seek help for their current complaints if they had done so before. The findings remained significant even in the context of other variables influencing intention.

Concerning behaviour, the findings were mixed. People with treatment experience were more likely to seek help from a professional, especially a GP, than if they had no previous treatment experience. Satisfaction with previous treatment, however, did not influence behaviour. This indicates that it is the processes of help-seeking, e.g., knowing where to go, what the doctor will ask, what to expect, and less the quality of previous experiences, i.e., satisfaction, that seems to have an influence on help-seeking behaviour. We suggest that it is important, then, to support people in their behaviour by strengthening their knowledge of how to seek help and what structural barriers to expect. In addition, the possibility of simulation interventions, e.g., using virtual reality technology, could train participants to overcome actual barriers when seeking help.

There are some important strengths and limitations that should be highlighted. The study was preregistered (11) and the aims and procedure were transparently documented in the study protocol. The drop-out rate between the baseline assessment (N = 1,368) and both the 3-month or 6-month follow-ups together (N = 983) was acceptable with 28.14% and there were only a few missing data for treatment experience (3.95%). The data control procedure was very high, especially our criteria to only include participants with a stable PHQ-9 score over the first 36 h (baseline and intervention assessment) and to exclude all participants who finished the study too quickly. We did this to counteract the problem that in online settings outside distractions are more probable than in a laboratory setting, even if the level of attention does not seem to be impaired (50). Overall, however, the online setting was a strength. First, the sample was more representative since we were able to reach people across Germany. Second, it seemed likely that people with current depressive symptoms who were currently not seeking help would be more willing to participate in a low-threshold online study than a full clinical trial. However, one drawback was that the participants had to be registered in an online panel and it was, therefore, likely that our population was more motivated to participate in surveys and were younger than the general population. Another issue was that drop-out was more likely when a person was younger, came from a multi-person household, and was less educated; wherefore, there was a bias in our study and possibly the intervention material, which seemed to be implicitly preferred by people with higher education living in more individualistic households.

The study controls were active control groups, with all participants receiving at least a vignette describing a person with depressive complaints and most receiving some type of intervention. The vignette alone likely triggered a reflective process on their current complaints. Therefore, on the one hand, the intervention effects discussed in this article can be seen as very robust. Possibly, some effects stayed undetected due to the complexity of the fractioned factorial design. On the other hand, the fractioned design might have led to a confounding of the main effects. Subsequent research could then analyse the factors more insularly, now that the positive tendency has been established.

In conclusion, the results show promising interventional effects strengthening self-efficacy to seek help. Less clear are the effects of the interventions concerning help-seeking intention. Therefore, we think that an intervention would be effective if designed to strengthen self-efficacy beliefs as well as causal beliefs. Furthermore, the person should be encouraged to reflect on both how and when to manage their own symptoms, for example, when their symptoms are still mild, and when it would be better to seek professional health care. This seems a sensible approach when considering the conceptual associations and especially the interaction effects found concerning the intention to seek help dependent on depression severity levels. In addition, when considering the effect that previous treatment has on help-seeking behaviour, interventions should include concrete suggestions of how to actually seek help and which structural barriers have to be overcome in the process. This could reduce the potential reach of such an intervention, making it less generalisable and more local than the current study.

Questions relating to the exact interactions of self-efficacy and causal beliefs on help-seeking remained unanswered and should be considered in future studies examining the help-seeking process of people with depressive complaints.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. The intervention material is available in the supplement of the study protocol (11).

The studies involving humans were approved by Ethics Commission of the University Medicine Greifswald (BB 061/18). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

TM developed and drafted the manuscript, and analysed and interpreted the data. TM and L-JP contributed to the design of the study and the acquisition of the data. ST and HM contributed to the conception of the work. L-JP, ST, HM, SS and GS revised the draft. GS and SS managed the project. All authors contributed to the article and approved the submitted version.