Adolescent Depression Screening : Not So Fast

Methods: The evaluation of depression screening by the US Preventive Services Task Force is compared and contrasted with those of comparable agencies in the UK and Canada, and articles arguing for and against screening are reviewed. Internal pharmaceutical industry documents declassified through litigation are examined for conflicts of interest. A case is presented that illustrates the substantial diagnostic limitations of self-administered mental health screening tools.


INTRODUCTION
In early 2018, Dr. Karen Wagner (2018), in her Presidential Address to the American Academy of Child and Adolescent Psychiatry (AACAP), and the American Academy of Pediatrics (AAP) in its Guidelines for Adolescent Depression in Primary Care (Zuckerbrot et al, 2018), called attention to the need for improved identification and treatment of adolescent depression. Wagner cited the "devastating impact of depression on children's emotional, social, and cognitive development," and noted that the prevalence rates of depression have been increasing in youth, such that they are now being reported to be at 11% (2018, p. 6). She stated the goal of her presidential initiative was "to increase *Address correspondence to the author at: 2424 Dwight Way, Suite 2, Berkeley, CA 94704, USA; Tel: (510) 548-1655; E-mail: eclevin@earthlink.net awareness of and screening for depression in children and adolescents." She argued in favor of routine screening for depression in adolescents with the use of the Patient Health Questionnaire (PHQ-A; Wagner, 2018, p. 6). The AAP, in making a similar call for action in its guidelines, included a reference to the PHQ-9 as well as identifying other screening tools, but without suggesting a preference (Zuckerbrot et al. 2018).
While depression during adolescence is a serious condition that can impair social and educational functioning, as well as carry the risk of suicide, I see reasons to be concerned about the possibility of over-diagnosis of this disorder in adolescents and a corresponding over use of medications. Over-diagnosis needs to be considered as a contributor to the trend of increasing percentages of youth being diagnosed and medicated for a variety of mental health conditions, including de-pression. The possibility of over-diagnosis is magnified if screening is developed and promoted as a marketing tool by pharmaceutical companies, which in fact may be the case of calls for the use of depression screening devices.
My misgivings about these calls for universal depression screening were confirmed by my experience with a mid-adolescent female whom I saw. She had been diagnosed as severely depressed, hospitalized, and placed on antidepressant medication based on her receiving the highest possible score on a depression screen. Upon evaluation over the course of treatment, it became clear that the patient's problems were in fact reactions to a series of difficult life circumstances. I will discuss the details of this case later in this article.
The PHQ-A, which is the instrument being recommended by Dr. Wagner and others, calls for the patient to respond to written leading questions by checking boxes. It is derived from the PHQ-9 (Table 1; Kroenke & Spitzer, 2002).
Algorithms supplied by the pharmaceutical corporation that financed the development of PHQ-9 and then offered it free of charge to clinicians use numerical scores to determine the diagnosis. Table 2 reveals that the algorithms, in addition to making a diagnosis, also determine its severity and instruct the user as to when and how to initiate a treatment program. Pharmacotherapy is specified in cases deemed by the algorithm to be moderately or more severe.   Richardson et al., (2010);and Lyon, Maras, Pate, Igusa & Stoep (2016); while others question their use.
The U.S. Preventive Services Task Force (USPSTF) is the agency in the United States comparable to similarly tasked ones in Canada and the UK (Zuckerbrot et al., 2018). In 1998 the US Congress Agency for Healthcare Research and Quality decided, "…to convene the Task Force and to provide ongoing scientific, administrative, and dissemination support to the Task Force." The website of the USPSTF (2018), states that, "Members are screened to ensure that they have no substantial conflicts of interest that could impair the scientific integrity of the Task Force's work." The qualifier "substantial" is not defined.
More positive than negative, the USPSTF gave adolescent depression screening in 2016 a B grade. The essential difference between an A or a B rating hinges on whether the level of certainty that the net benefit of service is substantial or moderate. The USPSTF further qualified the B grade by adding, "Screening should be implemented with adequate systems in place to ensure accurate diagnosis, effective treatment, and appropriate follow-up." I believe it is unlikely that this recommendation would be carried out in everyday practice. Beck, Singer, Busche, Manderscheid & Buerhaus (2018) also support the use of screening instruments. Acknowledging that there is a critical shortage of mental health workers, they justify screening in order to make the mental health workforce more efficient. Reasoning similarly to Beck, the authors Dowdy, Ritchey & Kamphaus (2010) advocate the use of screening tools in schools, though not specifically for clinical reasons. They hold screening devices to be valuable in their ability to efficiently generate populationbased data, which would allow schools to identify such items as SES risk factors. They believed that in turn would then lead to better delivery of services and the more effective utilization of school psychologists.
Other experts focused on school mental health services support the use of school screening, while adding the proviso that "appropriate family, school, and community involvement" are necessary. For example, Weist and colleagues (2007) state: …mental health screening in schools has the potential to be a corner-stone of a transformed mental health system. Screening, as part of a coordinated and comprehensive school mental health program, complements the mission of schools, identifis youth in need, links them to effective services, and contributes to positive educational outcomes valued by families, schools, and communities (p. 53). McCormick et al. (2009) argue for the use of screening instruments in school systems for public health reasons. After citing a high incidence of depression in youth, they recommend the use of a specific instrument, the Reynolds' Multi-stage Depression Screening, and point out: Parents, teachers, and school counselors can readily identify some forms of emotional distress, especially those that manifest as disruptive behavior problems. However, the "quieter" forms of distress that manifest their effects on thoughts and feelings, rather than behavior often go unrecognized in children and adolescents until serious manifestations, such as suicidal ideation or behavior are evident. Systematic assessments that target internalizing symptomology can assist in identifying at-risk students (p. 92).
On researching the issues, Lyon and colleagues (2016) predicted that school mental health services would not be overwhelmed were universal depression screening to be done, and therefore advocated its use. They researched the issue with the use of a system dynamics approach, which is done by computer simulation. Their findings, however, appear to be in conflict with the field studies cited below which failed to demonstrate improvement in outcomes in a variety of settings.

SOME CAVEATS
A number of authors object to the use of screening devices for depression or other mental health disorders for a variety of reasons. Some point out that diagnoses need to be made by experts trained in the art, not by a checkbox instrument, while others say that there is no gain and some potential harm when a diagnosis is made. In addition to its possibly being stigmatizing, it may be a false positive and it does not necessarily lead to appropriate treatment. Some have raised concerns about their encouraging the over diagnosis of psychiatric disorders, which could then lead to over medication of youth and the overwhelming of an already struggling workforce.
Importantly, the Canadian Task Force on Preventive Health Care and the UK National Screening Committee have both recommended against screening, and in so doing, noted that there is no direct clinical trial evidence of benefit to the patient from screening (Jerant et al., 2014;.

LACK OF EVIDENCE THAT SCREENING LEADS TO IMPROVED CARE
Routine screening has not been demonstrated to lead to appropriate treatment. An early study found no impact of screening in primary care on physicians' practices (Hoeper, Nyez, Kessler, Burke & Pierce, 1984). Patients across a wide range of ages in a large primary care clinic (n=1453) were screened using the General Health Questionnaire (GHQ). Those identified by the GHQ as having a psychiatric condition were divided into an experimental and a control group. No significant difference was noted when one group of primary care doctors were informed of the screening scores while the other group remained blind to the data. Bergus et al (2005) screened 861 adult patients in two private rural clinics in Idaho and followed up with a study to see if this led to improved care. The conclusion of their research was that, "The information from the PHQ-9 did not improve clinicians' active management of depressive symptoms or result in improved patient outcomes. Our study is consistent with earlier studies undertaken in urban settings…." (p. 303).
Prochaska, Le, Baillargeon & Temple (2016) tested an unspecified device which screened adolescent students for depression, anxiety and PTSD and found that those who were identified as having scored positively for anxiety had a greater chance of having received treatment in the preceding year. They voiced concern, however, that the percentages of those treated did not match the percentages of those afflicted. How many of those screened positively for depression or PTSD was not cited. The authors commented that: While some may call for more widespread screening in the adolescent population, such a reform independently of broader mental health system improvements would likely overwhelm the currently overtaxed system in place. Further, for screening programs, the benefits of such an effort (including effective treatment) need to clearly outweigh the potential harms.
Ham & Allen (2012), in a paper on adolescent health screening and counseling note "…no trials have examined whether screening improves outcomes in adolescents " (p.1114). The article is a valuable source of information as the authors provide an eloquently detailed description of how best to engage adolescents in discussions of high-risk behavior. Mitchell, Yadegarfar, Gill & Stubbs (2016) conducted a meta-analysis on 26 publications that evaluated the PHQ. The studies they reviewed included some 40,000 adult patients. While finding, "The PHQ has potential to be used to rule out those without depression with few false posi-tives…," they concluded that "…neither the PHQ-9 nor the PHQ-2 can confirm a diagnosis of MDD when used alone as a one-off measure and this is independent of the scoring method" (p. 135). Smith & Erford (2001) wrote, in their comprehensive review of the BDI-II, "The fakability of the inventory has been an issue with all three versions of the BDI. This should always be kept in mind during the administration and interpretation of the test." The BDI-Y contains 21 questions that are similar to the 9 used in the PHQ-9 (Fig. 1). The PHQ-A was modified for use with adolescents, with the elimination of the 9 th question. The BDI was first published in 1961, revised in 1978 and 1996, and likely influenced the development of the PHQ-series, which was developed by Pfizer in 1999. It is reasonable to assume that the PHQwould be subject to the same problems as the BDI.

RISK OF OVER-DIAGNOSIS
Jerant and colleagues (2014) expressed concern that the routine screening of youth for depression with the use of self-administered tools could easily add to inflated diagnoses and the over prescribing of medication. They additionally noted that, "Despite trial evidence of little benefit, antidepressants are frequently prescribed to patients with few or no depression symptoms" (as would likely be the case with routine screening) "resulting in unnecessary costs and potential detrimental effects (e.g., labeling, medication toxicity). Discussions about antidepressants with such patients may also burden office visits, distracting from more salient issues." (p. 612) To the list of harms cited by Jerant, one may add those reported by Jha, Rush & Trivedi (2018). These authors detail the many symptoms which may follow the discontinuation of an SSRI, some of which may take as long as a year to resolve.
Allen Frances, already concerned by the DSM's expanding diagnostic categories, has been one of the strongest critics of universal mental health screening. In a Wall Street Journal article (2016), he argues, "Screening for depression is one of Fig. (1). BDI scoring sheet, copied with patient's permission from hospital records. those ideas that is terrific in theory but terrible in practice." He expresses doubt that a careful evaluation will follow an adolescent identified as depressed and is skeptical that a screening device can distinguish between normal sadness and clinical depression. He warns that transient and selflimited cases, not requiring diagnosis or treatment, will be misidentified. He adds that the influence of substance abuse would be missed. Finally, he argues that, "Drug companies disease-monger depression and aggressively advertise the extremely misleading message that all depression is due to a chemical imbalance that requires a pill solution." (2016), in their systematic review of depression screening to detect major depression in children and adolescents, question the accuracy of such tools. They note that no clinical trials have evaluated the benefits and harms of screening programs and conclude that there is, "…insufficient evidence that any depression screening tool and cut-off accurately screens for MDD in children and adolescents. Screening could lead to overdiagnosis and the consumption of scarce health care resources" (p. 746). Thombs and colleagues (2017) questioned the assessment of the USPSTF position by writing:

INADEQUATE ATTENTION TO POTEN-TIAL HARMS
The USPSTF was recently criticized for relying upon indirect evidence and for not adequately considering potential harms in recommending depression screening…. Experts pointed out that there are numerous examples where the use of insufficient and indirect evidence has led to ineffective and harmful screening programs and argued that guideline makers should refrain from recommending new screening services based on only indirect evidence. In the context of questionnaire-based screening programs, this concern is heightened because, when randomized controlled trials have directly tested these programs, they have not found evidence of health benefits.

Links to Pharmaceutical Companies
The PHQ-9 was developed by Pfizer, the pharmaceutical corporation, which wrote : … research partners Robert Spitzer, M.D. and Janet Williams, D.S.W. from Columbia University and Kurt Kroenke, M.D. from Indiana University, recognized the need for and supported the independent development of dimensional measurement tools for mental disorders to be used by health care professionals….The PHQ-9 and GAD-7 tools really are standard measures for physicians to use (Pfizer, 2010).
As noted previously the PHQ not only offers a diagnosis, "depression," but it also provides the person administering the test with an algorithm that instructs when to prescribe medication, as shown in Table 2.
After funding the development of the PHQ, Pfizer (2010) granted free access to it, allowing providers to use it without cost to the administering clinician or agency. A reasonable concern is that a pharmaceutical company which manufactures an antidepressant might increase the potential market for its product by developing a method of identifying potential consumers and promoting the use of this method by providers who would then prescribe its product.

The Role of Key Opinion Leaders
Key opinion leaders (KOLs) have become increasingly important in influencing health care practices. Both Jureidini (2012) and Meffert (2009) identify KOLs as thought leaders who are experts in their field due to their research and publications. Jureidini notes that what we commonly see as reasons for conflicts of interest, such as bribes, kickbacks, meals and other gifts are commonly part of pharmaceutical companies attempts to influence prescribers, but these may be less important to KOLs than support for their careers, for example help in being published and obtaining grants.

Meffert points out:
KOLs have become intimately entwined with the marketing of pharmaceuticals and medical devices, used not only to lend credibility to claims of efficacy and safety but also to promote anecdotal and off-label use of these medications to increase industry profits. Identification and marketing of the KOLs themselves is being done more and more often by KOL management companies who are hired by industry to turn those involved in medical education and research into efficient and productive members of the sales force (p. 262).
Acting as a key opinion leader, Dr. Wagner has played a major role for the last twenty years in the promotion of SSRI antidepressants for children and adolescents. She has made major contributions to the psychiatric literature as an investigator in clinical trials for SSRI treatment of pediatric depression. However, much of her work has been funded by pharmaceutical companies, and her recommendation for universal depression screening and the use of an instrument developed by one of these companies needs to be considered in the light of her long association with the industry. This issue was not addressed by Dr. Wagner in her Presidential Address.
Furthermore, several of her papers have been challenged as misrepresenting the data in GlaxoSmithKline's study 329 and Forest's study CIT-MD-18 (Amsterdam, McHenry & Jureidini, 2017;Jureidini, Amsterdam & McHenry, 2016;LeNoury et al., 2015). Allegations of data manipulation were based on evidence made available in internal corporate documents released by the courts in the course of successful suits against GSK and Forest.

A CASE OF MISDIAGNOSIS
A mid-adolescent female was referred by her pediatrician, who described her as being a very difficult case. She had been discharged a year previously after a month-long psychiatric hospitalization, where she had been given a diagnosis of major depressive disorder, with comorbid mixed eating disorder and self-mutilation. The hospital confinement had been precipitated by a reported overdose with a benzodiazepine which had been pre-scribed for anxiety. A psychologist who saw her administered the BDI-Y, on which the patient scored 100%. The scoring sheet minus identifying data is shown in Fig. (1).
A score of 70% on the BDI-Y is all that is required to signal a severe level of depression. The patient answered every relevant question in such a manner that it indicated the highest possible score for depression. Her score of 100% played a role in her being hospitalized.
The patient was discharged on fluoxetine 10 mg. per day and did well in outpatient treatment for a year, with her pediatrician providing medication management, until another episode occurred which led her parents again to be concerned about her suicidality. They sought advice from their mental health care system. It was decided in the hope of preventing another hospitalization to refer her for psychiatric evaluation and treatment on an urgent basis.
I was greatly reassured upon first meeting the adolescent and her mother, as the case did not appear so challenging as was initially presented to me. Additionally, not long after I began meeting privately with the patient, she began confiding historical material not previously revealed. She had been very unhappy about having to move with her mother to another city three years earlier, but she had strained to deal adaptively with the situation until her mother announced that she was extending her job there for a fourth year.
Shortly after the initial contact, a nearly oneinch thick stack of out-patient and in-patient records beginning over a year prior to her hospitalization became available for study. Aside from the chart notes made by her hospitalizing psychiatrists and one out of three hospital social workers, a review of records indicated that: • There was no evidence of in-hospital behaviors consistent with a diagnosis of depressive disorder or self-mutilation.
• There was no evidence of abnormal changes of weight prior to hospitalization or of inhospital weights or behaviors consistent with a diagnosis of an eating disorder.
• None of the notes charted by hospital nurses or by 2 of the 3 hospital social workers were consistent with any of her three diagnoses.
• The hospital chart included a print-out of her BDI-Y testing, completed just prior to her admission (Fig. 1). It indicated that the patient was not only severely depressed, but it also showed her with clinically significant levels of anxiety and anger. Yet, no symptoms of those issues were ever charted by nursing staff or by 2 out of 3 social workers as occurring while she was in-hospital. On the contrary, her unit adjustment was consistently reported as positive throughout her confinement. Her only medication was fluoxetine, with the highest dose being 30 mg. Her dose at discharge from the hospital was 10 mg. That dose was maintained for over a year until I began work with her. Shortly after the transition to my care, the medication was successfully tapered and discontinued without adverse consequences. Antianxiety medication was never requested or considered indicated.
In our work together since first meeting, in addition to my review of her medical chart and my contacts with her parents and pediatrician, the patient's dietary habits and weight never emerged as a problem. While she did acknowledge a history of attempting to cut herself, visual inspection of the sites where she says she tried revealed no evidence of scaring; and she acknowledged that she had never drawn blood. While it was my impression that she could do better in school, she did well enough to be in high potential classes and on a college track. She participated in extra-curricular activities and has been employed since age 16 in a job which requires her being responsible for the safety of others.
Sessions with the patient and her parents over the past three years have informed me of two additional episodes somewhat resembling these which occurred prior to my seeing her. These have led me to become aware of significant attachment issues, none of which were mentioned in her outpatient or in-patient records. She has consistently denied in our work any suicidal intent. She did have two episodes of taking excessive doses of medication. Her stated reason for the first ingestion, which precipitated her hospitalization the year before I met her, was to help her sleep, as she felt anxious due to receiving a phone call she felt was derogatory from a girl whom she had considered a best friend. The second ingestion had very similar elements, although that time it involved a boy. In both of the episodes which led to family and professionals considering hospitalization, she had perceived the threat of loss of a peer attachment. It was likely significant that at the time of each, no parent was immediately available to her.
When my patient answered positively on the BDI, "I feel irritated all the time," my sense is the response was over determined by the phone call which she had received and which activated attachment concerns. Rather than being asked to check a box, had she been interviewed, her state of mind might have been better determined by an inquiry as to how long or how often she felt that way.
My basic impression remains that, rather than being depressed, she had been a sad and protesting adolescent with attachment issues who had found herself in what she considered to be an intolerable living situation. I view her maximum depression score a cry for help rather than evidence of a mood disorder.

DISCUSSION
Without question, there is a need for better recognition and care for depressed adolescents. But there is a question as to how best to meet this need. One solution would be to provide more advanced education of generalists, as is specifically detailed and advocated by Zuckerbrot (2018). Additionally, we would be able to provide better care were more medical students trained in psychiatry. The solution should not lie in reducing the quality of care we make available to patients and their families. Nor should it involve making seemingly easy the identification of depression in youth, plus the decision as to when to prescribe medication by algorithm, with the use of a set of leading questions to be checked off by vulnerable patients. Hoffman (2018) addresses the problem of what to do in face of a mental health workforce which is very limited in size relative to the needs of those with significant mental health concerns. This problem has led some to suggest the restructuring of previously proven treatment approaches to depression. Hoffman's solution to the problem is to maintain the quality of care by increasing and activating the workforce. He argues: Most child and adolescent psychiatrists address the biomedical needs of the child 3 and leave the psychosocial care of the patient and family to others. Too often, child and adolescent psychiatrists see their role primarily as a diagnostician who can prescribe the appropriate medication depending on which categorical rubric fits the patient. A biopsychosocial model that includes intensive teaching of psychodynamic principles can help novice child and adolescent psychiatrists understand the psychology of children, adolescents, and their families and help trainees understand their emotional reactions to their patients and their patients' families. This approach helps child and adolescent psychiatrists appreciate the importance of understanding the meaning of the child's symptoms to the child and the child's family. Such a model would promote greater clinical and intellectual excitement not only in the fellowship but also in the field in general. In contrast, the current zeitgeist in too many child and adolescent psychiatry programs does not promote sufficient attention to the importance of listening carefully to patients and their families to understand the individual, familial, academic, and social factors contributing to a child's symptom presentation (p. 977). While Hoffman's focus is on child and adolescent psychiatry, I believe it is equally applicable for adolescent and general psychiatric specialists as well as all others who serve youth.
A similar position is taken by Parry & Levin (2012), in an article they wrote referring to "mindless psychiatry." They stated, "One aspect of this paradigm shift has been an emphasis on structured interviews and rating scales, which are necessary in research. However, this comes at the expense of introspection and reflection about the presenting phenomenology of patients in their life narrative and context" (p. 56).
I agree with the conclusions of Thombs and Ziegelstein (2017) in their aptly entitled article, "Primary care doctors should not screen their patients for depression." After stating multiple objections to screening for depression, such as lack of evidence that it is effective and without harm, the authors conclude: As part of standard care, clinicians should provide good mental health assessment (for depression). They should be alert to clinical cues that depression may be present, such as low mood, insomnia, anhedonia, or fatigue. They should be particularly aware with patients who may be at greater risk, including those with a family or personal history or those with a chronic medical condition, chronic pain, a history of traumatic life events, and drug or alcohol abuse…. There are some conditions, like diabetes mellitus, that almost always require screening or other special laboratory testing to diagnose. Depression is not one of those conditions. …a lack of knowledge is no excuse for using a screening tool with unacceptable test characteristics on all patients, and the best available evidence suggests that doing so would likely lead to more harm than good. Screening is not a substitute for good medical care. (p. 645.)

CONCLUSION
Dr. Wagner asserts that members of the AACAP have a responsibility to increase awareness of and screening for depression in children and adolescents (Wagner, 2018). For the reasons presented above I believe that this imperative is misguided. There is a high cost to overpathologizing and then medically treating patients, of any age, for simply acknowledging emotions by checking a box.
Tools for depression screening cannot make diagnoses. That is a task for skilled diagnosticians. Whether we are psychiatrists or primary care physicians, when we use a self-administered set of leading questions and a set of algorithms to diagnose and to determine when to initiate medication; we are encouraging the deprofessionalization of medicine and disconnecting ourselves from a fuller relationship with our patients. This results in the dumbing down of our profession. Psychiatrists and others need to be trained to diagnose and treat depression appropriately with a variety of tools, not just pharmaceuticals. Admittedly, diagnosing by interviewing takes more time than a selfadministered questionnaire. But there is no better way to screen than to use ourselves in an alliance with our patient and the family. If we are prepared to use our listening skills in the course of a re-spectful, empathic, comprehensive evaluation, we get a much superior product. By the end of listening, taking a history and establishing contacts with collateral sources; we are far more likely to have a strong working relationship with our patient and her family. The positives which flow out of that process include a richer understanding of the patient, a more accurate diagnosis, a more workable treatment plan, and better treatment compliance than one can obtain with a depression screening device.

ABOUT THE AUTHOR
Edmund C. Levin, M.D. is a child and adolescent psychiatrist in private practice in Berkeley, California