Helsinki University Hospital (HUS) provides original Finnish-language nationwide iCBT (HUS-iCBT) programs for various psychiatric disorders including GAD, SAD, PD, and OCD. The therapy programs are free of charge for patients. Referring patients to the therapy is possible for any licensed physician in Finland. The referring physicians (most often, general practitioners) receive support from web-based instructions but retain overall responsibility throughout the patient’s treatment. For each iCBT program, HUS performs observational, nationwide, open-label, real-world studies, as described by Ritola et al [16]. This study is a cross-diagnostic investigation using specific HUS-iCBT data sets together.

The recruitment period in HUS-iCBT for GAD was June 1, 2014, to December 31, 2017, that for OCD was October 1, 2015, to December 31, 2017, that for PD was June 1, 2014, to December 31, 2017, and that for SAD was February 1, 2016, to January 31, 2018.

The inclusion criteria for HUS-iCBT programs were being 18 years of age or older and having a diagnosis of GAD, SP, PD, or OCD verified by the referring physician. Patients with suicidal intentions; current drug misuse; diagnosis of psychotic, bipolar, or serious personality disorder; or impaired cognitive performance due to neurologic or neuropsychiatric disorder were excluded.

Observational studies have been or are currently being conducted for all 4 therapies and included all patients who gave their informed consent, with no additional inclusion or exclusion criteria.

In this study, we performed secondary analyses of the data from those 4 observational studies. No additional informed consent was required.

The HUS-iCBT program for GAD included 12 sessions, that for SAD included 7 sessions, that for PD included 10 sessions, and that for OCD included 10 sessions. Participants were required to complete self-report questionnaires during treatment.

Therapists providing HUS-iCBT were clinical psychologists, psychology students, or nurses with additional therapeutic training employed by HUS. The training of the therapists is described by Ritola et al [16]. The therapists provided support and feedback for the patients throughout the treatment process.

Each treatment program was investigated separately; however, the same statistical methods were used. Total scores were calculated for each symptom scale. Analyses were limited to the comparison of associations and differences between the first (baseline) and last (end point) sessions for each treatment program.

The requirements of parametric tests (homogeneity of error variance and normally distributed error terms) were assessed using graphical methods such as histograms, Q-Q-plots, and scatter plots, as suggested, for example, by Mage [33] and Ernst and Albers [34]. A graphical approach was chosen over normality and homogeneity tests because the sample size was rather large and statistical tests might be overpowered and show discrepancies that are not practically significant. 

Changes in the mean scores of OASIS, GAD-7, PSWQ, OCI-R, PDSS, and SPIN between the first and last sessions were assessed for therapy completers, using a standard univariate ANOVA F test.

Bivariate correlation analysis (Pearson product-moment correlation coefficient) was used to estimate correlations of OASIS with GAD-7, PSWQ, OCI-R, PDSS, and SPIN in each session; analysis was performed for each treatment program separately.

A general linear model for repeated measures (mixed models) and interaction analysis were used to investigate the changes and relationships in mean scores of OASIS and GAD-7, PSWQ, OCI-R, PDSS, and SPIN from the first to the last session. Separate models were built for every pair of OASIS and a disorder-specific scale.

The model was chosen on the basis of Bayesian information criteria. The number of bootstrapped random samples was 5000. 

In addition, we calculated the δ variables as a difference between the values of the first and last session of each specific measurement and OASIS. The δ pairs of GAD-7 and OASIS, PSWQ and OASIS, OCI and OASIS, PDSS and OASIS, and SPIN and OASIS were then compared using correlation analyses. The correlations between change scores were analyzed for descriptive purposes only and to show how different variables of interest behave. 

The study protocol and its amendments were approved by the ethics committee of HUS and by pertinent institution authorities (HUS’s chief medical officer; 179/13/03/03/2014). The study was conducted in compliance with the International Council for Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use–Good Clinical Practice guidelines, the tenets of the Declaration of Helsinki, and current national regulations. Personal data were pseudonymized. The researchers who had access to the data were unable to reidentify individual patients. In accordance with the Finnish legislation, patients were not compensated for neither their participation in the abovementioned observational studies nor the secondary analyses of the data in this research.

To our knowledge, this is the first study investigating the feasibility of OASIS as an outcome measure in iCBT for ADs. In all iCBT programs at the HUS included in this study (those for GAD, PD, SAD, and OCD), OASIS mean scores significantly decreased from baseline to the end point, as did those of all disorder-specific symptom severity measures. OASIS had clear and relatively stable associations with disorder-specific symptom measures. According to estimates of fixed effects in mixed models, associations of OASIS with OCI-R and PDSS were significant and stable within the treatment programs. By contrast, relationships of OASIS with SPIN, PSWQ, and GAD-7 did not demonstrate such stability. Associations between OASIS and SPIN, PSWQ, and GAD were stronger at the end of treatment than at the beginning. Thus, in SAD and GAD therapies, OASIS appears to be differentially sensitive to the severity of disorder-specific symptoms across the treatment phase. However, the lower first-session associations between OASIS and SPIN, PSWQ, and GAD-7 were 69%-93% of the higher last-session associations, suggesting moderate stability of the association from first to the last session. Excluding PSWQ, which targets worry rather than anxiety, the lower first-session association was 82% (SPIN) or 93% (GAD-7) of the higher last-session association. Thus, coherent interrelation of the OASIS and disorder-specific scales at the end of treatment indicates the capability of OASIS, especially as a follow-up instrument.

For face-to-face treatments, OASIS is a valid transdiagnostic outcome measurement instrument in both nonclinical and clinical samples [36,37] in different therapeutic interventions for ADs. Our results suggest that OASIS might have potential as a transdiagnostic outcome measure in iCBT as well.

Recent studies demonstrate that in iCBT, patient-reported lack of sufficient time or being too busy was associated with decreased user satisfaction and weakened adherence to the iCBT [38,39]. In addition, extensive text content of the therapy modules (including long questionnaires), requiring high levels of concentration and reading skills, appears to be an obstacle to successful treatment [40]. Thus, OASIS as a short, easy-to-use instrument may offer a worthwhile option to enhance adherence and, thereby, the overall effectiveness of internet-based therapy for ADs. Overall, we assume that OASIS can become a worthwhile alternative to the disorder-specific scales used in this study. The specificity of disorder-related scales should be weighed against the lightness and easiness of OASIS when choosing measures for iCBT for ADs on a case-by-case basis, depending on concrete needs of researchers or practitioners.

This study was conducted on a nationwide scale with a large number of participants, and the data were gathered in a previously underexplored routine clinical practice setting.

The absence of a control with clinician-applied measures might be seen as a limitation. However, all scales with which OASIS was compared are valid, reliable, and widely used psychometric measures. Moreover, it should be recognized that the use of self-rating scales is the only realistic form of outcome measurement in large-scale iCBT programs in busy routine care with high patient volumes [41].

Our study did not apply gold-standard measures of functional impairment, and the feasibility of the impairment subscale of OASIS could not, thus, be separately elucidated.

Our study included HUS-iCBT programs for many but not all ADs. For instance, specific phobias, agoraphobia, and separation anxiety disorders were not included. This leaves the question of feasibility of OASIS for iCBT for these disorders unresolved.

OASIS includes items for assessment of both symptom severity and anxiety-related functional impairment. Gold-standard functional impairment measures should be used in future studies to assess whether OASIS in iCBT could replace not only other symptom severity measures but also separate functional impairment scales. If this should prove to be the case, use of OASIS could facilitate the measurement battery in a range of diagnosis-specific and plausibly even transdiagnostic iCBT programs for ADs.

Future research is needed to evaluate the feasibility of OASIS for iCBT programs for ADs not included in our study. Furthermore, more research is required to assess the utility of OASIS for between-disorder comparison of the effectiveness of iCBT for various ADs. Psychometric properties of OASIS across different iCBTs could be subsequently investigated.

OASIS demonstrates clear and relatively stable associations with disorder-specific symptom measures in iCBT for a range of ADs. Considering the benefits of OASIS as a short, easy-to-use self-rating scale, it might have an implementation as an outcome measurement instrument for disorder-specific and possibly also transdiagnostic iCBT programs for ADs in regular clinical practice. Further research is needed to elucidate the feasibility of OASIS for iCBT for ADs other than GAD, PD, SAD, and OCD and to gain an overview of the use of OASIS’ functional impairment subscale.