krisenchat [37] is a German-speaking messenger-based psychosocial crisis chat counseling service for children, adolescents, and young adults under the age of 25 years. Counselors are volunteers with an education in psychology, psychotherapy, education, or social work and have received a minimum of 2 months of training in chat-based counseling and in screening for a potential clinical indication [32]. The aim of krisenchat is to listen to, calm, and comfort users during acute crises. Additionally, if deemed necessary, users also receive a recommendation to seek further help, either within the health care system (eg, professional local support services) or from a trusted adult person.

Usually, a chat counseling session consists of the following 5 phases: (1) building a trusting relationship, (2) understanding the problem, (3) clarifying the goal for the counseling session, (4) finding a common solution, and (5) properly ending the counseling session. For example, a female user aged 20 years and currently living with her violent parents may start out by mentioning an argument with her father (phase 1), but open up about being physically harmed after some empathic and supportive messages (phase 2). The counselor and user then agree that they want to do something about the situation involving outside help (phase 3). Since the user is not a minor anymore and not previously known to the youth office, this is not a case of child welfare endangerment. After asking the user what help she would like to receive and where she currently lives, the counselor informs her about her rights and options, and researches a local point of contact (in this case, an association called “Frauen Gegen Gewalt e.V.” [“Women Against Violence Association”]) (phase 4). Finally, the counselor and user agree on how to get in touch with the association and consensually end the chat after making sure that no open questions remain (phase 5).

Within the krisenchat counseling process, the first step toward a recommendation to a health service provider or a trusted adult person includes an assessment of age, gender, living conditions, crisis topic, (symptom) duration and severity, social resources, and existing affiliations to and previous experiences with help service providers. Usually, krisenchat users in need of further help are recommended to involve a trusted adult person from their private circle of acquaintances as a first low-threshold option for minor problems and conflicts. In cases where it seems necessary to eventually involve authorities (eg, youth welfare service), users are first recommended to involve a professional trusted adult person like a school psychologist. Other specialized professional counseling services with expertise in specific topics (eg, eating disorders, LGBTQ+, and drug abuse) are recommended when professional help seems necessary. Further, some users are recommended to seek medical help (eg, general practitioner and emergency ward) or psychotherapy treatment. When there are clear indicators of child welfare endangerment, contacting the youth welfare service is recommended.

For the purpose of this study, anonymous data from all chat users between October 1, 2021, and March 28, 2022, were extracted from the operational database. Data contained information collected by the counselors. In addition to this information, 2 online surveys were conducted: a feedback survey shortly after a chat session and a follow-up survey after 4 weeks.

For the feedback survey, each user received an automatically generated invitation via WhatsApp or SMS text messaging, which included a link to an online survey 6 hours after the first counseling session, if the chat session had at least 20 messages and the user was not considered at risk for child welfare endangerment by the psychological team. A minimum of 20 messages was established as an inclusion criterion based on the experiences of krisenchat with (1) the counseling process: after exchanging 20 messages counselors have usually gained enough information like age, gender, and the (primary) chat topic, which are necessary for a meaningful scientific evaluation of the chat; and (2) the user experience: after exchanging 20 messages the user has usually gained enough information to justify a feedback survey regarding their experience with the service.

For the follow-up survey, users received an automatically generated invitation via WhatsApp or SMS text messaging after 4 weeks, which included a link to an online survey, if the following criteria applied: the user had completed the feedback survey and had given their consent to be contacted again for the follow-up survey; a recommendation for a referral to the health care system or to seek further help from a trusted adult person had been given by the counselor and the counseling process was completed; and the user was not considered at risk for child welfare endangerment. Being considered at risk for child welfare endangerment was again an exclusion criterion in this step of the data collection process as these cases are often not labeled as cases of child welfare endangerment during the first counseling session, but only later in the counseling process. Children and young adults in these kinds of situations often contact krisenchat first with a minor problem to test if they can trust the service before opening up in a later counseling session about their real concern. If users attended more than one session, they received the link to the follow-up survey after the first recommendation for further help-seeking was made.

According to the definition used in a previous evaluation of krisenchat [32], a chat session was defined as a series of messages with no gaps between messages of more than 12 hours. It is possible that krisenchat users chat with more than one counselor during the course of one session. Users may or may not be made aware of the change in counselors, depending on the specific case.

Both surveys were set up in German using typeform [38]. At the beginning of each survey, it was offered to conduct the survey in simplified language in case users had limited German literacy.

Between October 1, 2021, and March 28, 2022, krisenchat received 10,614 requests in total. Of these, 3890 (36.6%) were excluded for one of the following reasons: users did not consent to the terms and conditions of krisenchat, they left the chat, or they did not receive a full counseling session due to a lack of capacity because of high demand. This left 6724 (63.4%) initial counseling sessions, of which 1124 (10.6%) were excluded as the chat session contained less than 20 messages (eg, when a user’s issue could be resolved within less than 20 messages) or the user was considered at risk for child welfare endangerment. The remaining 5600 (52.8%) users received an invitation for the feedback survey (see Figure 1 for the data collection process).

Of the 5600 users who received an invitation for the feedback survey, 539 (9.6%) later received an invitation for the follow-up survey. The other 5061 (90.4%) users were excluded from the follow-up survey as they met the exclusion criteria for the follow-up survey. In total, 280 (5.0%) users completed the follow-up survey, and of these, 33 (0.6%) had to be excluded due to technical circumstances, as they had completed specialized counseling within krisenchat due to being considered at risk for child welfare endangerment and had received the follow-up survey, although they met an exclusion criterion (ie, being considered at risk for child welfare endangerment). This resulted in a final sample of 247 users with completed feedback and follow-up surveys.

Informed consent was provided online via an opt-in function before participating in each survey. Ethics approval was granted by the Ethics Committee of the Medical Faculty, University of Leipzig, on August 3, 2021 (file reference: 372/21-ek).

Descriptive statistics were used for sociodemographic variables, recommendations (for a referral to the health care system or to seek further help from a trusted adult person), further help-seeking behavior, and facilitators and barriers to help-seeking. A chi-square test was used to analyze further help-seeking behavior depending on the recommended service or person. To this end, the number of categories of recommended services or persons was reduced by grouping them into the following 6 categories: psychotherapist or social psychiatric service, school psychologist or school social worker, parents, trusted adult person, counseling service, and other.

To analyze the associated factors of further help-seeking, the 2 subgroups of users displaying further help-seeking behavior and those not displaying further help-seeking behavior were compared regarding categorical variables (gender and session topic) using chi-square tests. For continuous variables (age, well-being after the chat, perceived helpfulness of the chat, and self-efficacy), both subgroups were compared using Mann-Whitney U tests owing to nonnormally distributed data, as indicated by Shapiro-Wilks tests (P<.05). Additionally, a logistic regression analysis was performed to analyze the association between various predictor variables (gender, age, session topic, well-being after the chat, perceived helpfulness of the chat, and self-efficacy) and further help-seeking behavior.

Bonferroni correction was used to account for multiple testing. To estimate effect sizes for chi-square tests, the ϕ coefficient was used, while Cramér V (ϕ_c) was used when the contingency table was larger than 2×2, with ϕ_, ϕ_c=0.10 indicating a small effect, ϕ_, ϕ_c=0.30 indicating a medium effect, and ϕ_, ϕ_c=0.50 indicating a large effect [41]. Effect sizes for Mann-Whitney U tests were interpreted as small (r<0.30), medium (r<0.50), and large (r>0.50) [41]. Statistical analyses were performed using IBM SPSS Statistics version 27 (IBM Corp). A 2-tailed α value of .05 was applied to statistical testing.

In the final sample of 247 users, most users were female (female users: 199/242, 82.2%; male users: 11/242, 4.5%; diverse users: 32/242, 13.2%; the data for 5 users were missing), and the mean user age was approximately 17 years (mean 17.31, SD 3.29; range 12-25 years). Nearly half of all users in each survey (feedback survey: 124/247, 50.2%; follow-up survey: 108/247, 43.7%) chose to conduct the survey in simplified language. The main reasons for contacting krisenchat were psychiatric symptoms (184/247, 74.5%), and psychosocial (136/247, 55.1%) or emotional distress (71/247, 28.7%). The mean self-efficacy score of krisenchat users was 2.82 (SD 0.86).

The most frequently recommended services or persons to seek further help from included a psychotherapist or social psychiatric service (75/225, 33.3%), a school psychologist or school social worker (52/225, 23.1%), and the user’s parents (45/225, 20.0%). Nearly half of all users (120/247, 48.6%) indicated that they contacted the recommended service or person. A chi-square test indicated no significant differences in help-seeking behavior depending on the recommended service or person (χ²_5,225=7.87; P=.16; ϕ_c=0.19).

Of the 120 users who indicated having contacted the recommended service or person, 102 (85.0%) gave further information. Specifically, a large proportion stated that they already had an appointment (or talk) with the respective service or person (70/102, 68.6%) or that an appointment or talk was scheduled (17/102, 16.7%), while only 12 (11.8%) reported that there were no appointments available, 2 (2.0%) reported that they had contacted but not yet reached the recommended service or person, and 1 (1.0%) indicated being referred to another service. Of the 87 users who already had an appointment (or talk with a trusted person) or had scheduled an appointment (or talk), 75 (86.2%) provided further information on possible follow-up appointments. Specifically, the majority stated that a follow-up appointment (or talk) was already scheduled (48/75, 64.0%) or was planned but not yet scheduled (12/75, 16.0%), while 5 (6.7%) reported that they had already received sufficient help and another appointment or talk was not necessary and 1 (1.3%) indicated being referred to another service after the first appointment. The remaining 9 (12.0%) users reported that no follow-up appointments were available or that they did not want to schedule another appointment as they did not feel comfortable or had not received the help they needed.

A detailed description of the reported facilitators and barriers to further help-seeking is provided in Table 2. Users displaying further help-seeking behavior reported facilitators to help-seeking in the categories of mental health literacy (54/120, 45.0%), improvement of self-efficacy (55/120, 45.8%), symptom recognition (40/120, 33.3%), and social support (20/120, 16.7%). Users not displaying further help-seeking behavior reported barriers to help-seeking in the categories of stigmatization (60/127, 47.2%), lack of mental health literacy (59/127, 46.5%), self-reliance and autonomy (53/127, 41.7%), negative family beliefs regarding help services (53/127, 41.7%), structural factors (13/127, 10.2%), and other factors (eg, someone else has offered help; 24/127, 18.9%).

Subgroup comparisons indicated significantly higher levels of self-efficacy (ASKU) in users displaying further help-seeking behavior than in those not displaying further help-seeking behavior (P=.01, small effect; Table 3). Moreover, users with further help-seeking behavior reported a higher perceived helpfulness of the chat than users without further help-seeking behavior (P=.02, small effect); however, this effect was not significant after Bonferroni correction. No significant group differences were found for gender, age, chat topics, and well-being after the chat (all P>.05). An overall logistic regression model for further help-seeking behavior was not statistically significant (χ²₁₀=15.04; P=.13; Nagelkerkes R²=0.09). Only perceived helpfulness of the chat was significantly associated with further help-seeking behavior (B=0.56; P=.03; odds ratio 1.75, 95% CI 1.06-2.88). All other predictors in the model were not significant (all P>.05).

Overall, the results imply that children and young adults using krisenchat experience benefits. The 24/7 online counseling service may facilitate seeking further professional help within the health care system or support from others.

It is important to highlight that the referral of users to professional help services or personal contacts is only one of the aims of krisenchat (only in 27.9% of all krisenchat users), along with listening, calming, and comforting in an acute crisis; engaging in collaborative problem-solving; emphasizing and supporting chat users to find their own solutions; and improving mental health literacy [32]. Therefore, the current results only refer to a subgroup of all chat users (for more details on the overall use of krisenchat see [32]).

The results of this study show that children and young adults using krisenchat benefit in terms of seeking further help. Nearly half of all users (120/247, 48.6%) indicated that they contacted the recommended service or person after the professional counseling. This percentage appears to be very high and shows krisenchat to be an important gatekeeper service for young people and adolescents. Nevertheless, owing to novelty, there are at present limited studies on the evaluation of online chat counseling or crisis text line interventions. The few available studies are not comparable to our study owing to different foci or differing measurement approaches (eg, evaluation of users’ perceptions of effectiveness right after the chat or focus on suicide preventive aspects only [42-44]).

Nevertheless, more than half of the users who took part in the follow-up survey reported that they did not seek further help. There were no differences in help-seeking behavior depending on the recommended service or person. Other potential reasons for not displaying further help-seeking behavior remain unclear, and further research is needed in this area.

As mentioned earlier, little is known about the effects or benefits of online counseling or helplines, since studies are methodologically challenging not least because of the anonymity of the services. In a recent systematic review on the state of youth helplines, which is at least partly comparable to the krisenchat service, the authors concluded that helplines may provide a beneficial service to youth and that psychosocial concerns are the main reasons for contacting these services [45]. This result is in line with previous evaluations of krisenchat [32,33]. The authors also concluded from the results of the review that there is a lack of literature owing to a lack of controlled trials on the one hand, and complex methodological/ethical barriers preventing such trials on the other [45]. Few studies have investigated long-term outcomes of crisis hotlines in the past, mainly focusing on suicide rates within a population [46,47].

Nearly half of the users who sought further help reported mental health literacy to be an important facilitator, and reported the improvement of self-efficacy and symptom recognition as an effect of using the chat counseling service. In users who reported no further help-seeking behavior, the most frequent barriers were stigmatization, lack of mental health literacy, the need for self-reliance and autonomy, and negative family beliefs regarding help services. In line with the literature, the results also indicate that improving mental health literacy seems to be a key strategy for improving help-seeking in adolescents and young adults [19,48-50]. There are different approaches for improving mental health literacy in young people, including whole-of-community campaigns; community campaigns aimed at a youth audience; school-based interventions teaching help-seeking skills, mental health literacy, or resilience; and programs training individuals to better intervene in a mental health crisis [51,52]. 

Overall, it is important to highlight again that previous studies revealed that krisenchat may lower the barriers to access help, that is, about 50% of those contacting krisenchat reported that they never had contact with the professional health care system before [32].

The analysis on group differences revealed that users displaying further help-seeking behavior reported higher levels of self-efficacy compared with those not displaying further help-seeking behavior, but both subgroups did not differ in gender, age, recommended service or person, chat topics, perceived helpfulness of the chat, and well-being after the chat. Generally, the mean self-efficacy score of all krisenchat users was 2.82 (SD 0.86), which is considered to be very low [39]. In the logistic regression analysis, only a higher perceived helpfulness of the chat was associated with further help-seeking behavior; however, the interpretability of this result is limited as the overall model was not significant.

Surprisingly, in this study, central aspects of the krisenchat user (age, gender, and well-being) and service usage (chat topic and recommended service or person) seemed to be irrelevant for predicting further help-seeking. This means, in turn, that based on recent data, the only factor relevant for further help-seeking is self-efficacy. In line with the recent literature, this result supports the claim for more research regarding help-seeking in youth and adolescents, in order to be able to target interventions to improve and facilitate help-seeking behavior [19].

This study is the first to examine the impact of using krisenchat on the further help-seeking behavior of young people, and to identify associated factors of further help-seeking. However, a number of limitations must be considered. The final sample used for this analysis consisted of only 13.2% of all users of the service, who received a recommendation for a referral to the health care system or to seek further help from a trusted adult person, with completed feedback and follow-up surveys. There might be further associated factors of help-seeking or different patterns in the help-seeking behavior of users, which could not be detected due to this attrition rate. Additionally, there might be key differences between those who filled in the survey and those who did not (eg, regarding motivation, perceived helpfulness of the chat, chat topics, and other aspects). Further, several other variables, such as cultural background, fluency in the German language, education level, and socioeconomic status, which could not be assessed due to the anonymity of krisenchat, might influence help-seeking behavior and factors associated with further help-seeking. A further limitation includes the use of self-report data for assessing the recommendation for further help-seeking.

Nevertheless, given the fact that there are limited studies on the impact or benefit of 24/7 online counseling, the sample size of 247 is reasonable and can serve as a basis for stimulating further research and advances in measurements.

The results of this study suggest that children and young adults using krisenchat may benefit in terms of seeking further help. Further help-seeking seems to be associated with higher levels of self-efficacy. Further research is necessary to better understand how young people can further benefit from 24/7 online counseling. Longitudinal research seems especially crucial to understand who benefits when and from what kind of service and to assess which factors are associated with help-seeking behavior.