Short communication The identification in primary care of patients who have been repeatedly referred to hospital for medically unexplained symptoms: A pilot study Benjamin J. Smith, Kelly J. McGorm, David Weller, Christopher Burton, Michael Sharpe��� The Schools of Molecular and Clinical Medicine and Community Health Sciences, University of Edinburgh, Scotland, UK Received 23 March 2008 received in revised form 12 January 2009 accepted 13 January 2009 Abstract Objectives: The study aimed (a) to test a method of identifying patients who have been repeatedly referred (RR) from primary care to medical outpatient clinics where they have received multiple diagnoses of medically unexplained symptoms (MUS) and (b) to describe the prevalence and characteristics of these patients. Methods: RR patients with MUS (RRMUS) were arbitrarily defined as those with (a) five or more referrals in a 5-year period and (b) a specialist final diagnosis of MUS for at least three of these referrals. A two-stage method of identifying these patients was piloted in one primary care practice: Stage 1 used computerized health service data to identify RR Stage 2 used manual case note review to identify referrals that ended with specialist diagnoses of MUS. The RRMUS patients identified were asked to complete a questionnaire, a psychiatric diagnostic interview, and their GPs were asked to rate how ���difficult to manage��� they were. Results: The process was feasible and reasonably accurate. From 6770 registered patients aged 18 to 65 years, 23 (0.3%) were identified as RRMUS. They accounted for 157 referrals over the 5-year period. Sixteen agreed to further assessment, and 8 (50%) had a current anxiety or depressive disorder. GPs rated only 8 (50%) as ���difficult to manage.��� Conclusion: This two-stage procedure offers a practical method of identifying RRMUS patients in primary care as a first step in achieving more cost-effective care. These patients have substantial psychiatric morbidity. �� 2009 Elsevier Inc. All rights reserved. Keywords: Case identification Health care utilization Medically unexplained symptoms Primary care Depression Referral Introduction Patients attending primary care may be referred to specialist services for assessment of physical symptoms. When the hospital specialist finds a disease explanation for the symptoms, the biomedical treatments offered are appropriate. However, when they are deemed to be ���medically unexplained symptoms��� (MUS), the patient is typically simply discharged back to primary care. From a hospital perspective, it is striking that patients are sometimes repeatedly re-referred from primary care to specialist medical services despite receiving multiple diagnoses of MUS [1]. These repeatedly referred (RR) patients with MUS (RRMUS) are important because they incur high health care costs and occupy waiting lists while manifestly not receiving effective treatment for their persistent and distressing symptoms [2���4]. It is surprising to observe that general practitioners are often unaware of which of their patients fall into this category and that they cannot be readily identified from primary care case note summaries, which typically omit mention of referrals that result in a diagnosis of MUS. The aims of this pilot study were to (1) develop and test a system for identifying the most severe RRMUS patients in primary care and to determine its feasibility and accuracy (2) Journal of Psychosomatic Research 67 (2009) 207���211 ��� Corresponding author. Professor Michael Sharpe, Kennedy Tower, Royal Edinburgh Hospital, Edinburgh EH10 5HF, Scotland, UK. Tel.: +44 0131 537 6672 fax: +44 0131 537 6641. E-mail address: michael.sharpe@ed.ac.uk (M. Sharpe). 0022-3999/09/$ ��� see front matter �� 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.jpsychores.2009.01.004

determine the prevalence and characteristics of the RRMUS patients thereby identified. Methods Setting and sample As it was a pilot, the study was conducted in a single general practice. The practice had approximately 6770 registered patients aged 18 to 64 at the time of the study (i.e., end of the retrospective study period). Definition of FRMUS patients We defined RRMUS patients as those who had an average of at least one referral a year over a previous 5-year period, of which three or more had resulted in a specialist diagnosis of MUS. This high threshold was chosen to target the most severe and persistent cases. Identifying FRMUS patients We developed a two-stage system. In Stage 1, we identified RR patients in the practice by using a central national electronic register, the Scottish Morbidity Recording. This system, run by the Information and Statistics Division of the Scottish NHS (ISD) (www. isdscotland.org), stores data on all NHS patients in Scotland. Referrals made from the practice during the latest 5-year period of complete referral data (1995���1999) were counted (with referrals to specialties unlikely to receive referrals for MUS, i.e., clinical genetics, obstetrics, oncology, palliative care, and psychiatry excluded). In Stage 2, we identified which of the RR patients were RRMUS by reviewing the letters from the specialist to the general practitioners filed in the primary care notes. A diagnosis was classified as MUS when the specialist's letter indicated that the symptoms were inadequately explained by disease or were due to ���psychosocial factors.��� Detailed criteria for rating diagnoses are listed in Appendix 1. The accuracy of the computerized referral data was determined by comparing the number of referrals recorded by ISD with the number recorded in the case notes. The reliability of the criteria for diagnosing MUS was tested in a separate reliability study in which three doctors (MS and BS who were psychiatrists and DW who was a general practitioner) rated 80 sets of specialist letters. Describing RRMUS patients Identified RRMUS patients were invited to complete a questionnaire and participate in an interview. The ques- tionnaire assessed (a) functioning using the Medical Out- comes Study 36-item Short Form Survey (SF-36) [5] (b) number of somatic symptoms using the 15-item Patient Health Questionnaire [6]. The interview assessed psychiatric diagnoses using the Structured Clinical Interview for DSM- IV (SCID) [7]. The patient's GP was asked to complete the Difficult Doctor���Patient Relationship Questionnaire (DDPRQ), a measure of how difficult the doctor has found consultations with the patient [8]. Analysis The feasibility of the identification system was assessed by reviewing its implementation. Its accuracy was deter- mined by seeking the correlation of the number of referrals identified by ISD and those recorded in the primary care records. The interrater reliability of identifying a referral as resulting in a diagnosis of MUS was determined using the �� statistic. The prevalence of RRMUS patients in the practice was calculated by dividing the number of RRMUS patients identified by the number of registered patients aged 18���64. The patients' characteristics were described. The GP responses to the DDPRQ were coded as ���difficult��� if above the recommend cutoff of 30 [8]. Ethical approval The Local Research Ethics Committee approved the study. All interviewed patients gave informed consent. The Privacy Advisory Committee of the Scottish ISD gave permission for the referral data to be released to the primary care practice. Results Feasibility and accuracy of the identification procedure There was good correlation (coefficient, 0.84) between the number of referrals recorded by ISD and the number recorded in the primary care notes. As the case notes were located in the practice, they were easy to access and review of the relevant letters took only a short time (average 20 min to locate notes and read correspondence). The operationa- lized criteria for identifying MUS were found to have had good interrater reliability with a mean �� of 0.73 (range, 0.67���0.83). Prevalence of RRMUS The derivation of the sample of RRMUS patients is shown in Fig. 1. The 23 RRMUS patients accounted for 157 referrals between them over the 5-year period and made up approximately a third of those RR. They were 0.3% (23/ 6770) of practice patients aged between 18 and 64 years. Characteristics of RRMUS patients Seventy percent (16/23) of the RRMUS patients agreed to further assessment. The mean age of participants did not differ from that of those who declined, but three of the four men refused interview. Only six (38%) of the 16 patients 208 B.J. Smith et al. / Journal of Psychosomatic Research 67 (2009) 207���211