Diagnosis of somatoform disorders in primary care: diagnostic agreement, predictors, and comaprisons with depression and anxiety

The burden of disease due to non-specific, functional and somatoform disorders is high and the presence of recurrent or persistent medically unexplained symptoms is associated with impaired physical and mental quality of life, increased utilization of healthcare services and the development of comorbidities like depressive and anxiety disorders [1, 2]. Available data regarding the prevalence of somatoform disorders are inconsistent and vary considerably in dependence on the underlying study population and the diagnostic criteria applied in the single studies (e.g. [3‐6]). A recent meta-analysis of 32 studies investigated the prevalence of somatoform disorders and medically unexplained symptoms in primary care patients using both strict diagnostic criteria according to clinical assessments (International Classification of Diseases; ICD [7], Diagnostic and Statistical Manual of Mental Disorders; DSM [8]) and standardized questionnaires [9]. That study revealed a high heterogeneity of the primary studies and substantial differences in the prevalence rates in the diagnoses of somatoform disorders according to ICD or DSM criteria with prevalence rates of 34.8% and 26.2%, respectively. Further, in 40% to 49% of the primary care patients, at least one medically unexplained symptom was detected using questionnaires. These numbers underline that somatoform disorders pose a highly relevant public health problem, but existing literature suggests that the process of diagnosing somatoform disorders is challenging and that the different diagnostic criteria may have a substantial impact on the detection rate. The low detection rate in primary care is one key problem in the management of somatoform disorders [6]. It has been demonstrated that only 33% to 60% of the patients are correctly diagnosed by their primary care physician (PCP) and referred to a specialist for further treatment [4]. This is particularly alarming as the primary care practice serves as the patient’s first point of entry into the health care system and access to mental health services. Furthermore, according to current guidelines for non-specific, functional, and somatoform bodily complaints [1], management of somatoform disorders is recommended within a stepped care model according to the course of disease, in collaboration with other physicians and therapists and coordinated by the PCP, thereby emphasizing the high responsibility of the PCP in the diagnostic and therapeutic procedure [10].

Several barriers have been identified with respect to the diagnosis of somatoform disorders in primary care settings [11]. Present data indicate that a lack of specific training, a lack of experience and a lack of reliable diagnostic tools may hinder the diagnosis of somatoform disorders [11]. It has been further demonstrated that the application of the diagnostic criteria seem to be problematic for many PCPs and existing classification systems have been described as being difficult to use, impractical, not distinct, overlapping or too restrictive [11]. Besides these conceptual barriers, both patient and PCP-related characteristics might contribute to a correct diagnosis, and it is of importance to identify relevant predictors of a correct PCP diagnosis of a somatoform disorder to improve early detection of the disease.

The aims of the present study are, first, to investigate the level of agreement between diagnoses of somatoform disorders and its comorbidities depression and anxiety disorder obtained from a structured clinical interview and diagnoses reported from PCPs and, second, to identify patient and PCP-related predictors of PCPs’ diagnoses of a somatoform disorder.

In the present population of primary care patients at high risk for somatoform disorders, the diagnostic agreement between PCPs’ diagnoses and diagnoses obtained from a structured clinical interview was highest for depressive disorders, followed by somatoform disorders and anxiety disorders. Our analyses further revealed that somatic symptom severity was a relevant patient-related predictor for PCPs’ diagnosis of a somatoform disorder, while a specialization in internal medicine and working in a solo practice were relevant physician-related predictors for PCPs’ diagnosis of a somatoform disorder.

We showed that PCPs identified 32.3% of the patients with a somatoform disorder using the CIDI diagnosis as reference. This finding is in line with data from previous studies which revealed agreement rates ranging from 18% to 56%, depending on the investigated populations and the methods applied in the single studies [19‐22]. This variety may reflect the difficulties in the terminology and conceptualization of somatoform type disorders which have often been discussed in previous research [11]. Regarding conceptual barriers in the process of diagnosing somatoform disorders, existing classification systems have been described by PCPs as being difficult to use, as providing little information about the illness, or as too restrictive [11].

In the present study, all PCPs participating in Sofu-Net had attended network meetings and quality circles which were an inherent part of the network providing information about diagnostic approaches and management of somatoform disorders. Moreover, participating PCPs were prompted to use the PHQ as screening tool for early detection and contemporary referral to psychotherapeutic co-treatment. We therefore suppose that participation in Sofu-Net has promoted the PCPs’ awareness for patients with somatoform complaints and their diagnostic knowledge and skills. Further, among the participating PCPs, 85% had a training in psychosomatic basis care, suggesting that these PCPs may be particular interested in the field of psychosomatic care which might has enhanced the detection of patients with somatoform complaints. In addition, PCPs were aware of the patients’ positive screening result, giving advice that somatoform complaints were present in their patients. Nevertheless, using CIDI as reference, a large part of the patients with somatoform as disorders remained unrecognized by their PCPs. However, it is of importance to note that among all patients who had been “overlooked” by their PCPs (67.7%), PCPs stated for 54.6% of these patients that they do not consider the patients’ complaints as medically explained and therefore suspect the presence of a somatoform disorder. This finding suggests that PCPs were aware of somatization in their patients, although not having labelled their patients as having a somatoform disorder. A recent qualitative study by our group among PCPs investigated the process of coding somatoform disorders in primary care [23]. In that study, PCPs reported that coding is done for reimbursement purposes, that they use other information in their personal documentation, for example about the patients’ psychosocial background or potential causes about the presented symptoms, and that they do not necessarily need to document a confirmed diagnosis for treatment. Moreover, PCPs reported that they restrain their coding to protect patients from stigma trough certain diagnoses or other negative consequences. Inaccurate coding was further described to arise from uncertainties regarding the diagnostic criteria and that finding the definite diagnosis is seen as the responsibilities of psychiatric specialists [23]. Although we have not investigated these processes in the present study, it might be assumed that the described factors were of relevance among the PCPs in the present study and have contributed to the relatively low detection rate.

In the present study, the diagnostic agreement was highest for depressive disorders (48%). It has been argued that PCPs are better acquainted with depression than with other mental disorders and may therefore have a greater ability to detect depressive disorders [24]. Regarding the presence of an anxiety disorder, the diagnostic agreement was only 25.0%. When interpreting this result, it needs to be considered that the diagnostic classification of anxiety disorders in the present study encompassed conditions of differential severity (i.e. panic disorder, animal phobia). It is possible that PCPs did not to diagnose light cases, given that symptoms might not have been apparent or might not have been associated with severe impairments. In turn, patients might have not disclosed their symptoms when feeling only slightly impaired.

We demonstrated that somatic symptom severity was a relevant predictor for a diagnosis of a somatoform disorder by PCPs, while the patient’s sociodemographic characteristics had no impact on the detection rate. Possibly, an increasing symptom severity presumably has prompted patients to describe their bodily complaints more often and striking to their PCPs, thereby facilitating the detection of a somatoform disorder.

We further showed that a specialization in internal medicine and working in a solo practice were relevant physician-related predictors for a diagnosis of a somatoform disorder. This result is in line with findings from a study demonstrating that physicians who saw themselves as more effective in dealing with somatoform symptoms were more likely to be working in a solo practice [25]. Data from that study further showed that PCPs working in solo practices and those who were in the same practice for 5 years felt most comfortable in dealing with patients with somatoform disorders. It might be assumed that these PCPs had a better chance to establish a close relationship with their patients and had more time to explore their patients’ complaints and psychosocial background facilitating the diagnostic process. With regard to the PCPs’ education, we found that PCPs with training in internal medicine were more successful in diagnosing somatoform disorders compared to PCPs with training in general medicine, suggesting that training in internal medicine encompasses knowledge and skills facilitating the process of diagnosing somatoform disorders. It might be further possible that training in internal medicine makes PCPs more confident of declaring symptoms as medically explained, and thus diagnosing somatoform disorders. Taken together, the present findings underline that structural circumstances are essential for improving early detection and care for patients with somatoform disorders.

Strengths of the present study encompass the well characterized population of primary care patients at high risk for somatoform disorders as identified using established screening instruments and the structured diagnostic procedure using CIDI interviews. Our analyses are further based on different data sources including patient self-report, diagnoses obtained from CIDI interviews, and a PCP questionnaire, thereby integrating different perspectives on somatoform disorders diagnoses. Limitations may arise from the low response rate in the telephone interview, which potentially has introduced a selection bias. However, screening-positive patients who participated in the telephone interview did not differ from those who dropped out with regard to age, gender and PHQ-scores. Another weakness of the present study is the lack of data from screening-negative patients, which may has led to a misclassification bias underrepresenting the rate of patients with somatization. On the other hand, the PHQ has been demonstrated to be a well-validated measure for detecting and monitoring somatization, depression, and anxiety [15]. Nevertheless, as data on screening-negative patients are lacking, no data are available regarding PCPs diagnostic assessment of these patients.