The research described below was carried out as part of my MSc in Research Methods in Psychology, completed at the University of Warwick in 2010. You can read about it in much more depth here. I am currently developing this research by investigating whether people make judgements about their physical symptoms in the same way as they do their mental symptoms and how this influences help-seeking behaviour.

Introduction

Over- and under-detection of depression and anxiety is quite common when diagnosis is unassisted (i.e., unaided by diagnostic instruments or severity scales, etc.); individuals without clinically significant symptoms may be diagnosed with depression at rates up to 50% and only around 33% of people with anxiety and just under 50% of people with depression presenting in primary care are actually identified and diagnosed (Lecrubier, 2001; Mitchell, Vaze, & Rao, 2009). In order to account for this, previous research has focused on errors in physician judgement such as biases and faulty reasoning (for a review, see Harding, 2004). We argue instead that patient related factors, in particular the beliefs patients hold about the severity of their symptoms, may contribute significantly to the over- and under-detection of depression and anxiety. It is the patient that initiates most consultations in primary care (Kessler, Lloyd, Lewis & Gray, 1999), so for any diagnosis to occur, the patient needs to judge whether their symptoms are clinically significant or not and then actively seek help if they think they are. Furthermore, the outcome of consultation is influenced by what is presented by the patient and how (Weich, Lewis, Donmall, & Mann, 1995), so a correct diagnosis is somewhat reliant on patients being able to accurately recount symptoms, their occurrence, and perhaps most essentially, their severity.

Research Findings

We applied the Decision by Sampling (DbS: Stewart, Chater & Brown, 2006) model (for more information see my research page) to judgements of depression and anxiety symptom severity and showed how these judgements were rank based.

Study 1

In Study 1 we elicited participants’ (n=144) beliefs about the distribution of depression and anxiety symptom occurrence in the general population and asked how many days a month they experienced such symptoms. We used this information to calculate the relative rank position of the number of days each participant experienced the symptom within their elicited distribution (subjective rank) along with the mean of each participant’s subjective distribution function (subjective mean). Using ordinal regression we found that participants’ judgements of whether they thought they suffered from depression or anxiety were mainly predicted by subjective rank and not by subjective mean or the actual number of days they experienced the symptom.

Study 2

In Study 2 we experimentally manipulated the context for judgement to see if simply seeing other numbers of days at the same time as rating a given “target” number (e.g., 10 days/month) influenced how severe the symptom was judged to be. Participants (n=52) were given one of two distributions of the number of days a month 11 hypothetical people experienced six symptoms of depression and anxiety and were asked to give a severity rating for each person based on the number of days they experienced the symptom. These distributions had the same mean, rage and end points and were as follows: unimodal distribution: 3, 10, 12, 13, 14, 16, 18, 19, 20, 22, and 29 days, bimodal distribution: 3, 4, 6, 8, 10, 16, 22, 24, 26, 28, and 29 days. In addition to the endpoints, these distributions have three common points (10, 16 and 22) and using analysis of variance we found that for each symptom participants rated experiencing that symptom 10 days a month as being more severe if they were given the bimodal distribution where it's rank was 5 than if they were given the unimodal distribution where it's rank was 2. Furthermore, participants rated experiencing a symptom 22 days a month as being more severe if they were given the unimodal distribution where it's rank was 10 than if they were given the bimodal distribution where it's rank was 7. There was no difference in severity ratings of the middle common point (16 days a month) or the two end points which had the same rank position in both distributions.

Conclusions and Implications

We concluded that judgements of depression and anxiety symptom severity are based not on the symptoms’ objective severity (the number of days the symptom is experienced) but rather by where this severity ranks in comparison to the believed symptom severity of others. For example, an individual who feels depressed on 15 days each month and who believes that only 10% of people feel depressed on more than 15 days a month may be likely to classify themselves as depressed.

Errors in judgments will occur when the comparison sample constructed is not representative of the relevant population. For example, if the aforementioned individual believes that 80% of people feel depressed on more than 15 days a month they may judge that their own symptoms are normal when in fact they are severe enough to warrant diagnosis. Similarly, if an individual feels depressed on 5 days a month but believes that only 10% of people feel depressed on more than 5 days a month then they may judge that their symptoms are clinically significant when they are perhaps not.

Both these examples show how false beliefs about the context of judgement, in this case the actual symptom occurrence within the population, could lead to an underestimation or overestimation of symptom severity which in turn could affect help seeking behaviour. Many people may be experiencing anxiety and depression but not receiving any help or treatment because they (incorrectly) believe that their symptoms are normal simply because other people around them are also suffering but to a greater extent. This implies that the people who could be the most vulnerable to mental health disorders (i.e., from geographical regions or demographic groups where disorders such as anxiety and depression are high) may be the ones that are most at risk from non-identification and treatment of these disorders.

References

Kessler, D., Lloyd, K., Lewis, G., & Gray, D. P. (1999). Cross sectional study of symptom attribution and recognition of depression and anxiety in primary care. British Medical Journal, 318, 436–39. DOI: 10.1136/bmj.318.7181.436

Lecrubier, Y. (2001). Prescribing patterns for depression and anxiety worldwide. The Journal of Clinical Psychiatry, 62 (Suppl 13), 31–36.

Mitchell, A. J., Vaze, A., & Rao, S. (2009). Clinical diagnosis of depression in primary care: a meta-analysis. Lancet, 374, 609–619. DOI: 10.1016/S0140-6736(09)60879-5

Stewart, N., Chater, N., & Brown, G. D. A. (2006). Decision by sampling. Cognitive Psychology, 53, 1–26. DOI: 10.1016/j.cogpsych.2005.10.003

Weich, S., Lewis, G., Donmall, R., & Mann, A. (1995). Somatic presentation of psychiatric morbidity in general practice. British Journal of General Practice, 45, 143–147.