We used data from two separate studies to provide more robust evidence.

Study 1 is the German data linkage project Optimised Data for Public Mental Health, methods described elsewhere [33]. It comprises survey data (paper and pencil interviewing) collected between October and December 2021 linked on the person level to insurance claims data from n = 6,558 participants randomly sampled among adults insured with a major German statutory health insurance company (BARMER).

We repeated some analyses with data from the representative telephone survey German Health Update (GEDA, Study 2), methods described elsewhere [34]. We used data from n = 10,676 randomly sampled participants from the adult population living in Germany interviewed between July 2021 and October 2022.

To counteract possible biases due to selective participation, adjustment weighting by population distributions of gender, age, region (2020 statistics from the Federal Statistical Office) and education (Microcensus 2018) was applied to both datasets across analyses.

Sample characteristics are summarised in Table 1. We defined five age groups (18–29, 30–44, 45–64, 65–79 and 80+ years). The CASMIN Educational Classification (Comparative Analyses of Social Mobility in Industrial Nations) was used to categorise participants’ level of education into low, middle and high based on self-reported information on school-leaving and post-school qualifications [35].

We calculated total percentages with 95% confidence intervals of those who reported a perceived need for mental healthcare in Studies 1 and 2 as well as percentages within sex, age and level of education subgroups in order to assess the prevalence of perceived need and its distribution in the population (aim 1). To examine whether perceived need is associated with psychopathological symptoms, functional impairment and mental disorder diagnoses, we compared these indicators of psychopathology in those with and without perceived need (aim 2). Specifically, we calculated a) percentages of those above the cutoff scores on each of the specific screeners named above as well as any psychopathology screener for each group (Study 1 & 2), b) percentages of individuals with any documented F-diagnoses as well as with each of the selected F-diagnoses for both groups (Study 1) and c) mean functional impairment scores for both groups. To additionally examine the association between perceived need and MHL (aim 3), we also calculated mean MHL by perceived need status. Chi-squared tests and t-tests were used for statistical comparisons for categorical and metric variables, respectively.

Using data from Study 1, we also calculated a multivariate logistic regression model in order to examine the association between perceived need and sociodemographic characteristics, measures of mental health (positive on any screener), functional impairment and MHL whilst controlling for each of the respective other variables. This analysis targets all three study aims, providing an improved understanding of the distribution of perceived need by sociodemographic group (aim 1) as well as of the association between perceived need and more objective measures of need (aim 2) and MHL (aim 3) by elucidating whether observed patterns are explained by the respective other factors. Age was included in all analyses as a categorical variable owing to the relevance of information on particular age groups within public health surveillance as an evidentiary basis for policy as well as to potential non-linear relationships between age and the other variables examined. Due to their unique relationship to the outcome variable – with diagnoses often being a result of care seeking and care seeking often indicating perceived need, we did not include documented diagnoses in this analysis. We tested for multicollinearity among the predictor variables and found a variance inflation factor (VIF) below 2 across variables. Because all scales used comprised a limited range of values, no extra analyses for outliers were necessary. We also checked the dataset for implausible values. All analyses using data from Study 1 were computed with SAS 9.4 [48], while all analyses using data from Study 2 were computed with STATA 17 [49].

One-sixth of the population reported a perceived need for mental healthcare in the previous 12 months in 2021/2022 (Study 1: 18%, Study 2: 16%). This figure is remarkably consistent with findings from other high-income countries from (over) a decade ago (Australia: 14% with perceived need in 2007 [27]; Canada: 17% with perceived need in 2012 [50]) as well as a previous 3-year prevalence estimate for Germany (14%) based on 2019 survey data [31]. One-sixth of adults with perceived mental healthcare needs in the previous year stands in contrast to a 12-month mental disorder prevalence of approximately one-third among adults in Germany as reported based on clinical interview data from 2009–2012 [12].

In line with previous studies, perceived need for mental healthcare was reported more frequently by females [31, 51], younger adults [52, 53] and those with higher educational attainment [26, 53]. The gender gap in perceived need is consistent with higher prevalence of mental disorders among women [12] as well as higher treatment seeking and mental healthcare use among women than men in Germany [15, 54–56]. Notably, the association between sex and perceived need was not statistically significant in the multivariate analysis, suggesting that the bivariate relationship is explained by other related factors such as higher symptom burden, higher functional impairment or differences in MHL (found in our data [results not shown] and in a previous German study [57]), rather than by a more immediate gender role effect on the expression of needs.

Age group differences in perceived need also match age group differences in psychopathology, 2009–2012 survey estimates showing a declining prevalence of mental disorders with age [12]. Consistent with our findings of particularly low levels of perceived need among older adults, the oldest adults with a mental disorder were least likely to report contact to mental healthcare services according to the same data [15]. Higher perceived need among the youngest adults, however, may not always translate into increased treatment seeking: The same clinical interview study did not find increased self-reported service use among younger compared to middle-aged adults affected by a mental disorder [15]. A nationally representative survey asking generally about mental healthcare use in the previous year reported peak use in middle age [56]. Barriers to treatment seeking among young adults with perceived need and barriers to perceiving need among older adults therefore warrant attention. Importantly, unlike sex, age remained a strong predictor of perceived need even after controlling for psychopathology, functional impairment and MHL (which is higher among younger adults, results not shown), suggesting that other age-specific factors explain the observed differences. Long-term surveillance should aim to disentangle age effects from possible cohort effects on the prevalence of perceived need for care [58].

In contrast to patterns observed by sex and age, prevalence of psychopathology and perceived need come apart in the educational attainment groups. While individuals with a low level of education are more frequently affected by mental health problems [59], this group is less likely to report need for mental healthcare than those with a high level of education, with and without control for symptoms, functional impairment and even MHL – which is a strong candidate factor in explaining this discrepancy [26, 60]. This result is in line with findings of equal self-reported service use despite differences in psychopathology by education [15] and socioeconomic status [56] in Germany, as well as national studies showing lower utilisation rates of psychotherapy among those with low levels of education according to self-reports [31] and registry data from statutory health insurance [61]. It has been proposed that disadvantaged groups may avoid mental healthcare due to a greater fear of stigmatisation, lack of trust in mental health professionals as well as a lack of resources negatively impacting healthcare access [60], which is a potential barrier even when treatment itself is covered by insurance. Lower socioeconomic status individuals may also be more likely to attribute their symptoms to their life circumstances rather than to a mental disorder [26]. This underscores the importance of efforts aimed at reducing social inequality in health [59, 62] alongside improving equality in healthcare access, which includes needs awareness.

The relevance of perceived need as a mental health surveillance indicator used to inform evidence-based healthcare planning depends on its association with measures of need that are meant to determine the allotment of healthcare resources – namely, indicators capturing psychopathology and functional impairment, which are also the intended basis for clinician diagnosis of a mental disorder. Its association with these other measures is also a prerequisite for meaningful triangulation between needs indicators within surveillance. Indeed, we find strong associations between perceived need and psychopathological symptom burden (4-5 times as many above-cutoff scores on any screener among those reporting need compared to those not reporting need) as well as an independent association with mental health-related functional impairment, consistent with the literature [6, 13]. The fact that by far not all individuals reporting a need for care screened positive for psychopathological symptoms is to be expected given the limited scope of the screening instruments as well as the differing temporal reference periods of these instruments (2/4 weeks) and the perceived need item (12 months).

Our finding of a nearly twofold prevalence of documented F-diagnoses (F00-99) in the previous 12 months in the group with perceived need compared to the group without and between two to six times the individual diagnoses examined sheds light on the relationship between perceived need and “normative need” [8]: 61% of those with perceived need have had need identified by a health professional. Of course, these results are also bound to be heavily impacted by the close relationship between perceived need and help seeking. Notably, 33% of those without a perceived need for treatment in the previous 12 months have a documented F-diagnosis in this period. This may be because F-diagnoses cover a broad scope and not all diagnoses may be perceived as “problems with emotions, nerves or use of alcohol or drugs.” Generally, perceived need likely varies by disorder just as treatment seeking does [15]. This means that even among those who have had contact to health services that results in diagnosis, identification and reporting of need is likely to vary. However, we find a non-negligible proportion with a documented diagnosis among those without perceived need for each of the specific diagnosis groups examined. In line with the pursuit of patient-centred approaches to mental healthcare [27] and given potential for improvements in diagnostic practice [10, 11], this mismatch should be addressed in healthcare practice.

While it bears repeating that MHL does not (fully) explain the fact that those with a low level of education are less likely to report need but more likely to have mental health problems, an association between MHL and perceived need was found. Furthermore, those with symptoms but no perceived need exhibited significantly lower MHL than those who had symptoms and reported need. These findings are in support of the idea of improving MHL as an approach for improving access to care and reducing the treatment gap. However, MHL score differences between those with and without perceived need were surprisingly small, suggesting that future research could benefit from exploring whether other measures of MHL reveal more variance, particularly as MHL is likely ever evolving in populations.

The following limitations should be considered: 1) The comparability between Study 1 and Study 2 is limited by differences in survey mode (written self-report versus telephone interview) as well as how representative of the population of adults living in Germany the data can be assumed to be (random sampling from members of an insurance company comprising over 10% of the German population versus from the general population). However, differences between the insurance sample and the German population according to age, gender, region and education are minimised using weighting factors. Further differences in certain combinations of age, gender, region and education in the sample are assumed not to have a large impact on the overall results [63]. A strength of our study is that we were able to use a sample randomly drawn from the general population to reproduce some analyses from Study 1. The similarity in results from the two studies can be seen as evidence for the representativeness of the results based on insuree data. 2) While the employed single-item measure of perceived need has been used in several large-scale studies internationally, it has not been validated using longer perceived need assessments such as the one used in Meadows and colleagues [27] or qualitative assessments of item interpretation. 3) The employed screeners cover a limited range of symptoms and do no refer to the same time period as the perceived need measure, limiting the extent to which outcomes can be used to assess the meaning of perceived need. 4) All self-report measures may be subject to social desirability and recall bias effect.

The present study contributes to a better understanding of a single-item measure of perceived need and provides evidence for its suitability as a measure to be monitored within public health surveillance. The integration of different indicators including prevalence of mental disorders, symptoms and functional impairment in the population, prevalence of F-diagnoses in the healthcare system as well as the patient-centred indicator perceived need may be the best approach to the difficult task of evidence-based healthcare planning, as has been previously argued [6, 13].

Research should continue to examine possible factors explaining the absence of perceived treatment need among those with mental health problems with the aim of informing public health measures seeking to improve access to care. These measures should be conceived with the distribution of perceived need within different sociodemographic groups, particularly by age and education, in mind. Improving MHL in the population may be one approach to reducing the treatment gap, but our finding that perceived need differences by age and education persist after controlling for MHL suggest that this may not be sufficient. Another approach is to circumvent perceived need and introduce symptom screening in general healthcare practice, particularly among high-risk groups [64, 65]. However, efforts to overcome the mental health treatment gap should not undermine the patient perspective on need for care. For various reasons, professional care may not always be the desired or most impactful approach to addressing symptoms of mental ill health – particularly with regard to the social determinants of mental health [59]. Future research should examine met versus unmet perceived need and factors linked to unmet perceived need in Germany.