Strengths and limitations of this study
The search for guidelines was extensive.
Inclusion of guidelines was not limited to English texts.
The main limitation is that some guidelines may have been overlooked and that the cut-off for guideline inclusion can be debated on a geographical and local level.
The guideline committee mandate and scope may have varied between guidelines.
Guidelines were searched for all relevant statements about adverse effects and corresponding citations, but some statements or text excerpts may have been overlooked.
The use of antidepressants in children and adolescents with depression is the topic of several national and international guidelines which appear to differ considerably in extent, quality and information on potential harms.
As defined by the US Institute of Medicine in 2011, clinical practice guidelines are statements that include recommendations intended to optimise patient care that are informed by a systematic review of evidence and an assessment of the benefits and harms of alternative care options. Guidelines should provide a clear description of potential benefits and harms for each recommendation.1 2 However, compliance with standards has been shown to be low for several parameters, including more focus on benefits than on potential harms.3 Other researchers have assessed the quality of clinical guidelines regarding benefits of antidepressants4 5 using the Appraisal of Guidelines for Research& Evaluation II (AGREE II) tool.6 We have not identified any specific appraisals of the adverse effects information provided in guidelines on antidepressant therapy.
For fluoxetine, which is the suggested medication of first choice, few high-quality clinical trials are available to assess risk–benefit.7 The largest randomised controlled clinical trial on fluoxetine in children and adolescents with depression is a National Institute of Mental Health (NIMH)-funded study (Treatment for Adolescents With Depression Study (TADS)),8 which is considered a high-quality trial.9 We have previously analysed the adverse effects reporting in the TADS study10 and found that little data had been published concerning adverse effects profile during treatment for more than 12 weeks, with the exception of suicidality.
Based on our previous study,10 we had reason to believe that adverse effects descriptions in many guidelines on treatment of depression in children and adolescents would focus primarily on suicidality, with less attention given to the risk of other adverse effects and the risk-benefit balance. The adverse effects profile of the most commonly used medications, the selective serotonin reuptake inhibitors (SSRIs), is generally assumed to be similar in adults and children/adolescents and similar between the different SSRIs. This includes psychiatric effects such as suicidality, mania, anxiety, agitation and sleep disorders, as well as gastrointestinal effects such as nausea, diarrhoea and anorexia/weight loss. In this review, we aimed to analyse to what extent adverse effects data on SSRIs were mentioned in clinical practice guidelines on treatment of depression in children and adolescents. We also aimed to characterise the documentation provided as references in the clinical guidelines, and to assess to what extent data from the TADS study, with the identified data gaps, was used as basis for information about adverse effects.
To identify guidelines and clinical evidence summarisations on treatment of depression in children and adolescents, a search was performed according to the Norwegian Health Library guidelines on literature searches for development of clinical procedures.11
Literature database search
Searches in PubMed and guideline collections were carried out in the period 30 October 2018 to 17 February 2019. An EMBASE search with time limits 2008–2019 was performed 5 December 2019.
Several PubMed searches were performed, using the search terms:
(“depressive disorder”[MeSH Terms] OR (“depressive”[All Fields] AND “disorder”[All Fields]) OR “depressive disorder”[All Fields] OR “depression”[All Fields] OR “depression”[MeSH Terms]) AND ((“guideline”[Publication Type] OR “guidelines as topic”[MeSH Terms] OR “guideline”[All Fields]) OR (“practice guideline”[Publication Type] OR “practice guidelines as topic”[MeSH Terms] OR “clinical practice guideline”[All Fields])) AND (“humans”[MeSH Terms] AND (“infant”[MeSH Terms] OR “child”[MeSH Terms] OR “adolescent”[MeSH Terms]))
“Depressive Disorder/drug therapy”[MAJR] AND “Antidepressive Agents/therapeutic use”[MAJR] AND (Guideline[ptyp] OR Practice Guideline[ptyp])
((practice guidelines as topic) AND adolescent psychiatry) AND depressive disorder
Filters: Humans, Child: birth-18 years
EMBASE search terms were: (exp adolescent depression/ OR exp major depression/ OR exp depression/) AND practice guideline/. Limits: (child <unspecified age>or adolescent <13 to 17 years>) and yr=“2008–2019”
Searches in clinical guidelines collections
Clinical treatment guidelines were identified through searches in Guidelines International Network, McMaster Plus, Epistemonikos, UpToDate, BMJ Best Practice, DynaMed Plus, International Network of Agencies for Health Technology Assessment, National Institute for Health and Care Excellence (UK), Cochrane library, the Norwegian Health Library, Sundhetsstyrelsen (Denmark), Socialstyrelsen (Denmark), Center for Clinical Guidelines (Denmark), Socialstyrelsen (Sweden), AHRQ website (USA), IQWIQ website (Germany), Psychenet website (Germany), guidelines.gov (USA), SIGN website (Scotland), Health Canada website, CADTH website (Canada). In addition, a manual reference search was performed on identified studies and guidelines. The guideline search and selection process is described in a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart.12
Inclusion and exclusion criteria
We included national guidelines from European and/or English-speaking countries, published in English, German, French and any Scandinavian language since 2008, covering the last 10 years. We also included well-known and widely used international decision support tools. Local hospital or county guidelines were not included.
The guidelines and decision support tools (hereafter referred to as guidelines) should specifically address depression in children and adolescents. In cases where we retrieved multiple guidelines from the same organisation, the most recent version was analysed.
The guidelines were examined to identify information on safety or adverse effects. All authors made an overall assessment of the full guidelines. Text excerpts and corresponding references concerning safety or adverse effects were initially extracted by TW. Data extraction from all guidelines was checked by SN and MK. Text excerpts were assessed in detail by all authors. Information on adverse effects provided in each guideline was analysed and classified by organ system, in accordance with the classification used in the product monograph for Prozac (fluoxetine) by Eli Lilly.13 In order to distinguish suicidality or self-harm from other psychiatric adverse effects, these were registered as a separate category. Analysis was limited to adverse effects occurring during treatment and did not include mention of withdrawal reactions.
We assessed whether the guidelines provided information on the quality of the underlying evidence (high, moderate, low or very low) and strengths of recommendations (strong or weak) by use of the Grading of Recommendations Assessments, Development, and Evaluation (GRADE) tool14 15 or similar criteria for quality ratings.
We also evaluated the guidelines regarding presentation and risk–benefit assessment. To our knowledge, there are no explicit criteria for such evaluations. The guidelines were assessed by all authors independently, according to whether the guideline provided an overall discussion on risks, made an attempt to assess risks and benefits together, gave information on handling adverse effects, provided frequency estimates on risks or gave recommendations on monitoring patients. Any discrepancies in judgement were discussed by all authors.
Classification of adverse effects
References that were provided as basis for statements on adverse effects were extracted from all guidelines, and the most cited references were identified. Adverse effects were mainly classified by System Organ Class according to the MedDRA classification, as used in European Summaries of Product Characteristics.16 In this classification, appetite disorders are classified under Metabolism and nutrition disorders; sleep disorders, restless, anxiety, mania and mood disorders are classified under psychiatric disorders and headache, dizziness, and somnolence are classified under nervous system disorders. Suicidal thoughts and behaviour are classified under psychiatric disorders in MedDRA, but has been noted separately in this review. Reproductive system and breast disorders include sexual dysfunction and erectile and ejaculation disorders. Libido changes are classified under Psychiatric disorders in MedDRA, we have, however, included any such information in the Reproductive system and breast disorders category, in order to show a comprehensive view on all sexual adverse effects. An overview of the System Organ Classes and the included adverse effects can be found in the online supplementary material table S1.
The references were assessed and classified by category. Analysis by category and type of adverse effects was performed to assess to what extent the citations were used to discuss risk of suicidality versus other adverse effects.
Patient and public involvement
No patients were involved in this review.
Inclusion of guidelines
Adverse effects mentioned in guidelines
The 19 guidelines mentioned an average of 5.3 adverse effects categories (median 4.0, range 1–15). All 19 guidelines included information on suicidal risk. Fifteen of 19 mentioned other types of psychiatric adverse reactions, including mention of manic switching or manic/hypomanic episodes in seven guidelines.
Seven of 19 guidelines limited their information on adverse effects to suicidality and/or psychiatric reactions. Three guidelines mentioned risk of suicidal behaviour only (figure 2). Nervous system adverse effects including headache was mentioned by 10/19 guidelines, as was also the case for gastrointestinal reactions. Several possible adverse effects were only mentioned in a few guidelines, as an example, the risk of sexual adverse effects was mentioned in 5/19 guidelines. Four guidelines mentioned adverse effects in 10 organ categories or more.17–20
The overall assessments of somatic and nervous adverse effects varied widely, as illustrated by different descriptions of the SSRI risk profile. While one guideline stated that studies have shown somatic adverse effects to have small significance,21 another guideline stated that SSRI treatment causes significantly more nausea, diarrhoea, anorexia and stimulatory side effects (agitation, insomnia and anxiety) than tricyclic antidepressants.22 In our evaluation of the guidelines regarding presentation and risk–benefit assessment, we judged that 9/19 guidelines could be classified as having a sufficiently extensive and balanced consideration of adverse effects in their overall recommendations, while 10 guidelines did not (see online supplementary material table S2). Nine of 19 guidelines presented grading of evidence and strength of recommendations by the GRADE tool or similar, though the evidence grading mostly reflected efficacy documentation (see online supplementary material table S3).
References in guidelines as basis for adverse effects information
We identified 124 specific references as basis for statements in the 19 guidelines (see online supplementary material table S4). Some guidelines mentioned general sources of information, such as ‘FDA warnings’, ‘Cochrane reviews’, or ‘Product monographs’. The guidelines had a mean number of adverse effects references of 9.3 (range 0–32).
The 124 references were cited 177 times, as some references had multiple citations (table 3). The largest group of references was single studies, followed by review articles, statements or warnings from medicinal authorities and systematic reviews (see online supplementary material table S4). Most references to review articles, authorities, systematic reviews and guidelines concerned risk of suicidality. For the single studies, however, 22/52 citations (42,3%) referred to adverse effects in other organ systems (see online supplementary material table S4). These included 14 clinical trials. Somatic adverse effects were referred to in 3.2 organ categories for each citation (mean value), minimum value 1, maximum value 10.
Forty-one of 177 citations concerned adverse effects other than suicidality and/or other psychiatric adverse effects (see online supplementary material table S4). Most of those referred to somatic adverse effects in few organ systems (mean value 2.8, median 2.0, minimum value 1, maximum value 10).
Overall, 116 citations concerned risk of suicidal behaviour and 54 citations concerned risk of other psychiatric adverse effects, while fewer citations concerned somatic adverse effects. For some known adverse effects according to the product monograph,13 no guidelines provided any citations (figure 3).
Data from the TADS study as basis for information about adverse effects
Of the 19 guidelines, seven referred directly to publications from the TADS study, while 11 referred to sources that cited TADS as part of their assessment. One guideline did not provide any references and consequently did not refer to the TADS study.
Adverse effects data from the TADS study were published by March et al8 and in more detail by Emslie et al in 2006.23 Both publications describe a broad range of adverse effects, including risk of suicidal behaviour, psychiatric adverse effects such as mania, sedation or sleeping problems, abdominal pain, diarrhoea and vomiting during the first 12 weeks of treatment. These articles were cited by seven and two guidelines, respectively. Most guidelines referring to TADS, directly or indirectly, described psychiatric, nervous, gastrointestinal and respiratory adverse effects. However, three guidelines20 24 25 that cited the TADS publication from March 20048 did not cite the study’s findings of somatic adverse effects. No guideline commented on the lack of long-term adverse effects data from TADS beyond 12 weeks.
Our analysis of 19 treatment guidelines of depression in children and adolescents shows that the guidelines vary widely regarding information on adverse effects. The most commonly used medications, the SSRIs, may have multiple adverse effects involving several organ systems, but few guidelines described the full adverse effect profile. Several guidelines mentioned only risk of suicidal behaviour with or without mention of other psychiatric adverse effects. In many guidelines where adverse effects are mentioned, the frequency or the benefit-risk assessment is missing. Overall, somatic adverse effects were mentioned to little extent in the guidelines, and mention was often limited to few organ systems. The reporting of adverse effects in the guidelines appears to be selective and arbitrary. Several guidelines give the impression that suicidality is the only safety issue of consequence, and that there is a minor and limited risk of other adverse effects. We have not been able to identify a current, accepted standard for inclusion of adverse effects information in guidelines. It is possible that the varying descriptions of adverse effects in the guidelines are due to a lack of consensus as to what should be included. It can be argued that a full spectrum of adverse effects should be described. However, limitations may be necessary due to readability and format, in which case selection criteria should be stated.
All guidelines indicated suicidality as a major risk factor, reflecting that this has been a major topic of safety discussions. This was also the case for the underlying literature references. Most citations concerned suicidality with or without other psychiatric adverse effects. Far fewer citations concerned adverse effects in other organ systems. The fact that most references to review articles, authorities, systematic reviews and guidelines concerned risk of suicidality indicate that suicidality risk has been the main focus when addressing adverse effects of SSRIs. There were indications of selective citing, where the underlying articles provided more details of adverse effects than was referred to in the guidelines. The most frequently cited reference, the Cochrane database review by Hetrick et al,7 provides details on several adverse effects observed in the included studies, but most guidelines referred to this review only in the context of suicidality risk. Likewise, the TADS publication by March et al8 describes several somatic adverse effects which were not mentioned in many guidelines that cited the study on suicidality risk. A later publication on the range of adverse effects observed in the TADS study23 was cited by two guidelines only. We found that most of the known somatic adverse effects were not mentioned in the majority of the guidelines, thereby giving users a biased and skewed impression of the risk of harms. This lack of adverse effect information may affect treatment of children and adolescents with depression.
Guidelines should include descriptions of benefits and harms for the recommendations, however, this is a general statement and does not specify to what extent possible harms and adverse drug reactions should be described. The recommended tool for assessment of guideline quality, the AGREE II tool6 includes the criterion ‘The health benefits, side effects and risks have been considered in formulating the recommendations’. This should include descriptions of supporting data and reports of side effects, reports of the balance between benefits and side effects, and recommendations reflecting considerations of both benefits and side effects.26 The tool does not, however, specify what should be considered a sufficient level of risk information. We have not identified any quality assessment studies of adverse effect information in clinical guidelines.
For systematic reviews, which sum up the literature and form a basis for guidelines, a framework has been proposed to include relevant harms data in a more comprehensive way.27 However, research still show failings in the reporting of harms in systematic reviews.28 29 Despite having procedures and checklists for developing guidelines, there is no guarantee that adverse effects identified in underlying clinical trials will be reflected in the finished guideline.
In conclusion, we found that many guidelines on treatment of depression in children and adolescents did not provide a thorough risk assessment with information on well-known and common adverse effects. There is currently no international standard regarding the extent of adverse effects information that should be included in guidelines. Development of such standards would give clinicians better accounts of risks and benefits as basis for therapy decisions.
Strengths and limitations
We conducted an extensive search for guidelines, however, some older guidelines were not found in electronic full text due to closure of the US National Guideline Clearinghouse. In many cases, we were able to identify updates that were available elsewhere. Inclusion of guidelines was not limited to English texts. Due to our geographical location and language issues, many included guidelines are of European or American origin. The cut-off for guideline inclusion can be debated on a geographical and local level. We have included guidelines on a national level, but did not aim to include guidelines from all countries. Local hospital or county guidelines were not included. There is, however, a possibility that we have not identified all relevant guideline collections, and that some guidelines may have been overlooked. We included three well-known, international decision support tools as they are probably being used by professionals to an increasing extent, however, other similar tools may be commercially available. Guidelines were examined for text extracts concerning adverse effects, and the corresponding reference was noted. We cannot exclude the possibility that some statements or references may have been overlooked. Finally, we did not aim to perform a formal assessment of the overall quality of the guidelines and cannot assess whether this is correlated with the extent of adverse effects information provided in the guideline. It is possible that guideline committee mandates and local expectations varied between guidelines.