Rapid Responses to:

PRIMARY CARE: Bruce Arroll, Natalie Khin, and Ngaire Kerse Screening for depression in primary care with two verbally asked questions: cross sectional study BMJ 2003; 327: 1144-1146 [Abstract] [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Two Questions Does Not Work Accurately

Alex J Mitchell   (14 November 2003)

WHERE DOES OUR REMIT END?

mackenzie graeme   (14 November 2003)

screening for depression or dumbing down?

J Martin Dace   (15 November 2003)

It's just not that simple!

James C D McMillan   (16 November 2003)

Two or five verbally asked screening questions for depression?

Kaj Sparle Christensen   (17 November 2003)

Re: Two Questions Does Not Work Accurately

Adam Burrows   (17 November 2003)

Depression by GPs: another point of view

Alfonso SAURO   (18 November 2003)

We need realistic information on the time involved in screening

Christopher G Timmis   (19 November 2003)

Authors' reply

Bruce Arroll, Natalie Khin, Ngaire Kerse   (25 November 2003)

Screening can lead to the medicalisation of distress

Richard Byng   (4 December 2003)

Poor subjective well-being is a health hazard

Heli-Tuulie J Koivumaa-Honkanen   (17 December 2003)

Computing specificity

Jan MA Mens   (25 April 2004)

Two Questions Does Not Work Accurately 14 November 2003
Alex J Mitchell, Consultant in Liaison Psychiatry Leicester General Hospital (UK) Send response to journal: Re: Two Questions Does Not Work Accurately	The issue of general practioners’ (and other non-specialists’) ability to accurately identify depression is very important and the appeal of a quick and acceptable screening tool is extremely high. However, in our enthusiasm to identify such a tool, we are in danger of skipping over the scientific analysis of a diagnosis test (Steurer et al, 2002). An article in the BMJ has recently shown that a single question screening test for depression, doesn’t work very well (Lloyd-Williams et al, 2003). Maybe the answer is two questions instead? Arroll et al (2003) examined whether 1. answering “yes” to one of two common questions about depression is an accurate screening test for the presence of depression and 2. whether answering “no” to both of the these questions rules out depression. What they found, contrary to their summary box and the accompanying editorial by Del Mar (BMJ 2003;327:1117) is that the tests works in second direction but not the first. This is clearly illustrated in table 1 where asking any combination of the two questions (except both questions together to “rule in” depression whose data was not presented), leads to a phenomenally high false positive rate, out numbering true positives approximately five to one. But wait, the authors have two answers to this problem, which they of course spotted themselves. They suggest referring these cases to “another health professional”, presumably a psychiatrist in secondary care. This suggestion of referring not only the truly depressed patients, but five times as many incorrectly diagnosed as depressed is not going to help specialists deal with serious mood disorders efficiently, especially when many services already have issues with a high rate of inappropriate referrals from primary care and inadequate resources (Evans et al, 2002). The authors second suggestion for dealing with the low positive predictive value of this test is to “ask further questions”. Intuitively, this is a brilliant suggest akin to saying if the inaccurate method of assessment doesn’t work, why not do a proper assessment? Of course, the authors forgot that in the context of a screening tool in the clinic, the “two question test” will appear to work (ie generate false positives anyway because there will be no other reference standard in the field unlike this research study) unless, that is, all cases are automatically given the full reference standard, defeating the object of the concise test! In summary, yes the two verbally asked questions appears to detect most cases of depression, but at a cost of misidentifying a third (129/392) of people who are not depressed as depressed (and presumably if the test were actually relied upon assigning that 32% to inappropriate treatment). Obviously, this test is NOT a valid and clinically useful test and does NOT offer a reasonable trade off between true positives and true negatives. References Arroll B, Khin N,Kerse N (2003) Screening for depression in primary care with two verbally asked questions: cross sectional study BMJ 2003;327:1144-1146 (15 November). Evans J, Wilkinson E, Brindle L, et al (2002) Clinician opinions about the appropriateness and severity of general practitioner referrals to specialist mental health services: a cross-sectional survey. Primary Care Psychiatry 8 (3): 91-94 SEP 2002. Lloyd-Williams M, Dennis M, Taylor F, et al. (2003) Is asking patients in palliative care, "Are you depressed?" appropriate? Prospective study BRIT MED J 327 (7411): 372-373 AUG 16 2003. Steurer J, Fischer JE, Bachmann LM, et al (2002) Communicating accuracy of tests to general practitioners: a controlled study BMJ, Apr 2002; 324: 824 - 826. Competing interests: None declared
WHERE DOES OUR REMIT END? 14 November 2003
mackenzie graeme, GP Whitehaven CA28 7RG Send response to journal: Re: WHERE DOES OUR REMIT END?	Hypotheses 1. "Periods of depression are part of the human experience" 2. These periods will often resolve with no intervention and the mind and body will have learned what resolved them making the person stronger next time the arise. 3. Eventually "normal" depression gets extinguished with age Screening for depression may result in more and more people using antidepressants instead of life to resolve their depression therefore denying themselves and their "soul" a chance to learn what would have made them better. Could it be that antidepressants are like appetite suppressants? They work as long as you take them but on stoppingthem you are more likely to relapse because little learning has gone on. As an experienced GP I am increasingly sceptical of all screening. I think to ask people about depression, where clearly it is not indicated,is probably wrong Competing interests: None declared
screening for depression or dumbing down? 15 November 2003
J Martin Dace, freelance GP London SE14 5NJ Send response to journal: Re: screening for depression or dumbing down?	This article suggests that if you ask patients if they are depressed they will tell you, and if you're not sure how depressed then you have to ask some more questions. I think I knew that before. As GPs, we build on this common-sense understanding to develop more sensitive (in the human sense) ways of understanding our patients and perhaps helping them. This research risks dumbing down the art of medicine into one-size-fits-all inventories. Sir William Osler said, 'Listen to the patient, he is trying to tell you the diagnosis.' Isn't that a better way to do it? Competing interests: None declared
It's just not that simple! 16 November 2003
James C D McMillan, GP Registrar Perth, Scotland Send response to journal: Re: It's just not that simple!	I feel the biggest limitation here is its lack of applicability to my day to day surgeries. Each of my consultations are directed towards the problem in hand. As a trainee I am developing the tools for extracting the true problems from the multitude of vagueries with which patients present. This requires (amongst other things) various communication skills, intuition, life experience, as well as the background of medical knowledge required to recognise severe and disabling psychological illnesses. Am I going to ask each consecutive patient that walks through the door two questions on the off-chance? I think not. If I do and am reassured by negative responses, did I need to ask them at all? So does this help? Not really.The diagnosis of depression in Primary Care (and I suspect elsewhere) is just not tick-box material. Competing interests: None declared
Two or five verbally asked screening questions for depression? 17 November 2003
Kaj Sparle Christensen, PhD student, General Practitioner Research Unit for Functional Disorders, Aarhus University Hospital, 8200 Aarhus N, Denmark Send response to journal: Re: Two or five verbally asked screening questions for depression?	After reading the paper by Arrol and colleagues I ask myself: how many questions are actually being asked? Two? 1. During the past month have you often been bothered by feeling down, depressed, or hopeless? 2. During the past month have you often been bothered by little interest or pleasure in doing things? Five? 1. During the past month have you often been bothered by feeling down? 2. During the past month have you often been bothered by feeling depressed? 3. During the past month have you often been bothered by feeling hopeless? 4. During the past month have you often been bothered by little interest in doing things? 5. During the past month have you often been bothered by little pleasure in doing things? Further research in this area may benefit from clarifying which verbally asked questions or combination of questions demonstrate the maximum diagnostic accuracy. Competing interests: None declared
Re: Two Questions Does Not Work Accurately 17 November 2003
Adam Burrows, Assistant Professor of Medicine, Boston University School of Medicine Upham's Elder Service Plan, 1140 Dorchester Avenue, Boston, MA, USA, 02125 Send response to journal: Re: Re: Two Questions Does Not Work Accurately	Please keep in mind that the investigators evaluated the utility of a screening approach, not that of a diagnostic test. For a screening test, one wants a very high sensitivity (to rule out a condition in its absence), with an acceptable specificity (to rule in a condition in its presence). No one (I hope) is proposing that we diagnose depression with one or two questions. Rather, the study provides reassurance that we can comfortably rule out depression if the questions are answered in the negative. If the questions are answered in the affirmative, then depression may be present and the practitioner must take a full symptom inventory. Only if further history confirmed the presence of sufficient signs and symptoms would one make a diagnosis of depression. Competing interests: None declared
Depression by GPs: another point of view 18 November 2003
Alfonso SAURO, Responsable SNAMID South Italy - General Practitioner - EURACT and WONCA Member SNAMID CE - 81100 Caserta Italy Send response to journal: Re: Depression by GPs: another point of view	Depression by GPs: another point of view A. Sauro SNAMID - EURACT - WONCA Member GPs in Italy take in care a number of patients which remains quite the same for a long time, often all along their life. It happens that GPs know not only the clinical history of their patients but their life: family problems, religious beliefs, social relations, economic situations and even sexual tendences and problems. Because GPs know their patients deeply it becames easy to notice what they have changed in their way of doing. In a situation like that the GPs is often able to make a diagnosis of Depressive Syndrome easily and nearly always he recognises the starting causes too. In this case two questions could be enough: 1. During the past month have you often been bothered by feeling down, depressed, or hopeless? 2. During the past month have you often been bothered by little interest or pleasure in doing things? But we have to remember that the patient himself plays the most important part in every contact with his doctor (GPs or specialist). The patient tends often to refuse the idea and the diagnosis: he refuses to recognize he is affected by a "mental" disease. In my opinion it means that the real problem is: how to let the patients understand that depression is a common disease? How to let him understand that depression is curable? Depressive Syndrome needs to be attacked in another way: to get knowledge of what depression is restate the problem itself; more you know less you risk. Competing interests: None declared
We need realistic information on the time involved in screening 19 November 2003
Christopher G Timmis, General Practitioner Eastcote Health Centre, Abbotsbury Gardens, Eastcote, Pinner. HA5 1TG Send response to journal: Re: We need realistic information on the time involved in screening	Despite the brevity of the two question test for depression (Arroll et al) the poor specificity of the test leads to significant workload implications. 37 per cent of the patients screened positive for depression, of whom only 6.6 per cent were actually depressed. Having asked the two rather emotive questions I would be left with no option but to explore further the feelings of those patients who had responded positively . During my average working day this will amount to some 9 patients, of whom only one or two will actually be depressed. Fuller assessment of the mental state of these patients cannot be done with a few rapid-fire questions, but requires sensitivity and patience. Time for this is not easy to find in nine consultations each day. It seems that the GPs in the study did go on to discuss with some patients issues such as suicide risk, but no information is given on the effect of the test and the subsequent exchanges on average consultation times. It would be even more interesting to know the additional time taken when the screening test was employed without the reassurance of knowing that the patients were shortly to have a follow up ‘gold standard’ test for depression. In the busy consultations of present day General Practice information such as this is essential to decide on the applicability of the test to one’s own practice. Competing interests: None declared
Authors' reply 25 November 2003
Bruce Arroll, Associate Professor of General Practice and Primary Health Care University of Auckland Private Bag 92019 Auckland New Zealand 1001, Natalie Khin, Ngaire Kerse Send response to journal: Re: Authors' reply	Some important issues have been raised by the rapid responses to our article (Arroll B et al BMJ 2003;327:1144-6). In response to AJ Mitchell most screening tests (for most conditions) have a high rate of false positives compared with true positives. For example for mammography in women 50 -65 years there are 10 false positive tests for every true positive test. We only considered major depression (prevalence in this group 7%) yet there would be an equal number of people with minor depression in the group and hence the rate of true to false positives would be lower if both conditions were considered important. The spoken version of the two questions may have a better specificity than the written form1 because it raised the issue of depression as part of a conversation rather than filling in a piece of paper. It may be the conversation (and the doctor-patient relationship that occurs with the conversation) that is useful to accurate detection rather than the information in itself. Yet a more efficient use of the two questions would be to use the system advocated in the MacArthur foundation project.2 They advocate using the two questions and if either is positive to go on to the PHQ which is a 9 question gold standard3 and advocated by some as a screening tool in its own right.4 Patients could be given the (written) two question screening tool in the waiting room and invited to complete the written PHQ if they answered yes to either question. This would go some way to answering the response by CG Timms who was concerned about the time taken to deal with these issues. Our response to G Mackenzie and JM Dace would be to be aware that GPs miss a large proportion of depressed patients and that screening is recommended (on good evidence) by the US Preventive Services Task Force.5 My clinical colleagues and I were constantly surprised in another study (unpublished) by the apparently happy patients who responded yes to one of the two depression screening questions. Without the screening questions that information may never have seen the light of day. Our response to JCD McMillan is that he can be reasonably assured by a negative response as there was only one false negative in our study. The chance of having a major depression given a negative test (post test likelihood of a negative test) is 0.4%. 1. Whooley MA, Avins AL, Miranda J, Browner WS. Case finding instruments for depression two questions as good as many. J Gen Intern Med 1997;12:439-45. 2. Macarthur foundation. www.depression-primarycare.org, 3. Kroenke K, Spitzer RL, Williams JB. The PHQ-9 Validity of a brief depression severity measure. J Gen Intern Med 2001;16:606-13. 4. Nease DE, Malouin JM. Depression screening:a practical strategy. J Fam Pract 2003;52:118-26. 5. Pignone MP, Gaynes BN, Rushton JL, Burchell CM, Orleans CT, Mulrow CD, et al. Screening for depression in adults: a summary of the evidence for the US preventive services task force. Ann Intern Med 2002;136:765-76. Competing interests: None declared
Screening can lead to the medicalisation of distress 4 December 2003
Richard Byng, Lecturer Department of General Practice and Primary Care, GKT,Kings College London Send response to journal: Re: Screening can lead to the medicalisation of distress	Arroll et al’s study provides some useful new knowledge about screening for depression. It is useful to know that the two questions have a high sensitivity. My concerns are about how the results can be embedded in clinical practice. Two important points need to be made. Firstly, as a screening project, the GPs asked all the patients, regardless of any concern about mental illness. It is significant that nearly all the GPs actually remembered to ask the questions! It is probably unrealistic to expect GPs in routine practice to ask all patients. It would be useful to know the effect of training GPs to routinely use the 2 questions when having any level of suspicion of depression, as compared to recognition as usual. The impact of training to use screening should be compared with training to recognise possible recognition and engage with the patient in a dialogue to achieve a mutually agreed diagnosis. Secondly as has been discussed, the 18% positive predictive value emphasises that the two questions are only the first phase of a diagnostic process. A positive response should lead clinicians towards a process of attempting to disconfirm the diagnosis. The research did not report the impact on physician diagnosis and in particular the potential effect of wrongly labelling large numbers of patients with depression. I wonder if the authors have data on the rates of physician diagnosis. The potential negative impact of the medicalisation of distress should not be overlooked. Perhaps that is why screening programmes are not as effective as predicted. Competing interests: None declared
Poor subjective well-being is a health hazard 17 December 2003
Heli-Tuulie J Koivumaa-Honkanen, psychiatrist, clinical researcher Kuopio University Hospital, Pl 1777, 70211 Kuo, Finland Send response to journal: Re: Poor subjective well-being is a health hazard	Clinicians agree that it would be important to efficiently screen for depression. However, the high number of false positives of brief screening tests makes clinicians sceptical. They also might hesitate to ask about depression without clear indication and wonder whether brief screening jeopardizes the art of medicine. Indeed, a respectable clinician should seek for disabling psychological illnesses without giving opportunity for medicalisation of distress. And this should be done with sensitivity and patience, and not with few-rapid-fire questions! Moreover, screening and further evaluation requires time, which is not available. Thus, busy practitioners try to avoid any delays and patients might help them by avoiding reporting symptoms that might be interpreted as a mental disease. Unfortunately this short-cut seems not to work in a long run. In the Finnish Twin Cohort, 30,000 subjects were asked about happiness, easiness, interest in life and feelings of loneliness. All the four questions were provided by 96% of the respondents to a health questionnaire (response rate 84%) indicating the eagerness of reporting life satisfaction. The study subjects were followed for up to 20 years. Among the healthy, life dissatisfaction was not only associated with concurrent health hazards such as poor health behavior, but the dissatisfied (13% of the healthy) also had increased risk of mortality (especially in men) and subsequent morbidity compared to the satisfied. Furthermore, the 4-item life satisfaction scale was strongly associated with 21-item Beck Depression Inventory, which was less willingly responded. Screening is important due to huge disease burden of depression, but also due to the chronic course and poor health outcomes of sub-threshold depression. Brevity is not bad and whether the questions are written or verbally asked is not the main issue. Early detection and intervention is possible. People should be educated that subjective well-being should be paid more attention. It can be done in individual level, in health care system and in all other sectors of society responsible for building up mental health. Competing interests: None declared
Computing specificity 25 April 2004
Jan MA Mens, Rehabilitation Medicine Erasmus University Rotterdam 3000 DR Rotterdam Send response to journal: Re: Computing specificity	Dear colleague, I have read your publication with a lot of interest. I am sure I will use it in my practice. I still have a question. Could you explain how specificity (67%) had been computed? To my best knowledge, the specificity of your test is 18% (28)/157). Perhaps you use the term sensitivity in an alternative way as I do. Jan Mens Competing interests: None declared