3.1. Methods for the benefit assessment

The target population of the benefit assessment was adult SAD patients (≥ 18 years). The experimental interventions were phototherapy and vitamin D therapy. The comparator intervention was the respective other intervention, a second-generation antidepressant, psychotherapy, placebo, or no treatment. For studies on phototherapy, “placebo” was defined as follows: light boxes with 0 to a maximum of 300 lux, infrared light, or light boxes classified as a placebo in the original study, as well as inactive negative ion generators. Phototherapy with less than 300 lux was considered a placebo even if it had been classified as active treatment in the study itself because no light-related effects can be expected at such low lux levels [29]. Dawn simulation was viewed as an active intervention even at light intensities below 300 lux. However, at less than 5 lux or a duration of less than 15 minutes, it was classified as sham dawn simulation [29].

Based on research conducted in advance and interviews with 2 patients (see 3.6), the following patient-relevant outcomes were examined for the investigation:

• Mortality
• Morbidity
  • Response
  • Remission of depression
  • Severity of depressive symptoms
  • Functioning
• Health-related quality of life
• AEs

Only RCTs were included in the benefit assessment. To eliminate any carry-over effects or impact of seasonal changes, only the first phase of any cross-over RCTs was used in the HTA. In addition, only studies with an intervention duration of at least 2 weeks were included since, according to the clinical expert acting as advisor for this HTA, no intervention-related effect can be expected from shorter studies. If studies used multiple scales to calculate depression scores, the HTA extracted the depression scores measured with the scale most commonly used in the included studies.

In the MEDLINE, Embase, Cochrane Central Register of Controlled Trials, and PsychINFO databases, a systematic search for primary literature was conducted. In parallel, a search for relevant systematic reviews was conducted in the databases MEDLINE, Embase, Cochrane Database of Systematic Reviews, HTA Database, and PsycINFO. In addition to the primary literature, the reference lists of the systematic reviews identified for the search of reference lists were viewed. The following additional sources of information and search techniques were additionally used: study registries, publicly available documents from regulatory authorities, and requests to authors. If additional relevant documents are identified in the commenting procedure for the preliminary HTA report, they will be taken into account before the report is finalized.

The present project is part of an IQWiG study investigating the efficiency of study selection [30]. Therefore, the study selection from the hits of the bibliographic search was conducted both on the title/abstract level and on the full-text level by 3 persons (2 external experts, 1 IQWiG employee) independently from one another, each using his or her own screening tool. The results of the selection were summarized after the full text assessment. Any discrepancies were solved by the external expert managing the project. The hits from the study registries were viewed by 2 people, while the search results from any further information sources were screened by 1 person and reviewed by a 2^nd person.

One person extracted data into standardized tables, and a 2^nd person reviewed them. To evaluate the qualitative certainty of results, 2 persons, independently of one another as per IQWiG procedure, assessed the risk of bias on the study and outcome levels and rated it as low or high in each case. Any discrepancies were resolved by discussion.

To the extent that the studies were comparable in terms of their research questions and relevant characteristics and no meaningful heterogeneity was observed, the individual results were quantitatively summarized in metaanalyses using the statistics software R [31–33]. If at least 5 studies were available, a model with random effects was calculated (Knapp-Hartung method including the Paule-Mandel estimator for the heterogeneity parameter τ). If 2 to 4 studies were pooled or subgroups calculated, a model with fixed effect was used. The originally planned subgroup analyses by age, sex, and severity of depression at baseline were omitted because the studies were similar in terms of the average severity of depression as well as sex and age distributions and the studies themselves did not report on such subgroups. Instead, a post-hoc subgroup analysis was conducted, which distinguished between phototherapy with a patient-facing light box versus phototherapy with a head-mounted unit (HMU). For any metaanalyses pooling data from at least 10 studies, an additional funnel plot was generated in order to assess the risk of publication bias.

For each outcome, a conclusion was drawn on the evidence for (greater) benefit and (greater) harm, with 4 levels of certainty of conclusions: proof (highest certainty of conclusions), indication (moderate certainty of conclusions), hint (lowest certainty of conclusions), or neither of the above. The latter is the case if no data are available or the available data do not permit classification into one of the other categories. In that case, there is no hint of (greater) benefit or (greater) harm. The evidence was rated following the standardized definitions of IQWiG.

3.2. Methods for the health economic assessment

3.3. Methods for evaluating ethical aspects

The evaluation of ethical aspects was based on the overarching questions of the simplified questionnaire from Hofmann et al. 2014 [40]. The scoping search was performed on the information sources of Belit, CINAHL, EthxWeb, ETHICSWEB, MEDLINE, PsycINFO, and Scopus. Additional sources were the studies included for the benefit and economical assessment as well as interest group-based information sources such as stakeholder websites were used.

The target population of the ethical evaluation was expanded to include patients with nonseasonal depression. This is a legitimate approach because the ethical challenges faced by SAD patients and those suffering from nonseasonal depression tend to be similar.

One person viewed information from all sources identified in the searches for conclusions on ethical arguments and aspects of the intervention to be investigated, extracted them into standardized tables, and also provided a narrative. The result was scrutinized for quality by a 2^nd person.

3.4. Methods for evaluating social and organizational aspects

Scoping searches in MEDLINE, the AMWF (Association of the Scientific Medical Societies in Germany) guideline database, and on the internet (Google) focused on the evaluation of social and organizational aspects. In addition, both the studies included for the benefit assessment and health economic evaluation as well as the studies viewed for ethical aspects were screened for social and/or organizational aspects. Further, findings from patient interviews were integrated (see 3.6) and the “reflective thoughts” method was applied following INTEGRATE-HTA instructions [41].

The information processing on social aspects followed the conceptional framework proposed by Mozygemba et al. 2016 [42], while the information processing on organizational aspects followed the grid template proposed by Perleth 2014 [43] for the assessment of the organizational consequences of treatment methods.

One reviewer screened the information from all sources found in the scoping searches for conclusions on social and/or organizational arguments and aspects of the technologies to be investigated. The results were scrutinized for quality by a 2^nd person. All arguments and aspects necessary for information processing were extracted into tables and summarized as a narrative.

3.5. Methods for evaluating legal aspects

The databases of the German Federal Court, the Federal Legal Information System, the Manz legal database, the Lexis Nexis legal database, and the EUR-Lex legal database of the European Union were searched for relevant literature. In addition, a Google search and a search on websites of the Medical Service of the National Association of SHI Funds, the Bavarian Association of SHI Physicians, and the European Commission were conducted.

Information processing on legal aspects was based on the guideline developed by Brönneke 2016 [44] for the identification of legal aspects.

One person screened information for statements about legal aspects of the technologies to be investigated. The result was scrutinized for quality by a 2^nd person. All arguments and aspects necessary for information processing were extracted into tables and summarized as a narrative.

3.6. Patient interviews

After obtaining a favourable opinion from the internal review board of Danube University Krems (EK GZ02/2018-2021), 2 interviews with SAD patients were conducted using a predefined interview guide, posing questions on the disorder, experiences with therapies (particularly phototherapy and vitamin D therapy), and on expectations for an ideal therapy. The interviews were intended to determine which aspects of the disorder are experienced as particularly burdensome and which outcomes are most important to patients.