Scores were not normally distributed.
A, Positive responses on the ROS, comparing patients with significant clinical findings vs those with unexplained symptoms. B, Positive responses on ROS for group A (hearing loss, nasal obstruction, and hoarseness) and group B (tinnitus, dizziness, and globus sensation) patients. Both plots demonstrate differences in medians, with shaded areas representing 1st to 3rd quartiles and error bars representing data ranges.
A, Clinical findings vs ROSS for all patients. Percentage of significant and marginal or absent findings for all patients grouped by ROSS. B and C, Percentage of significant and marginal or absent findings for group A and group B grouped by ROSS. Marginal findings and absent findings (MUS) had parallel and overlapping trends, so are presented together.
A, Prevalence of psychiatric diagnosis vs review of systems, grouped by review of systems score (ROSS). B, Median ROSS for all patients by the number of different psychiatric diagnoses present in the medical record. One patient with 4 psychiatric diagnoses (ROSS, 12) was included in the correlation analysis but not shown in the figure.
eAppendix. The 69-Point Review of Systems Form
Customize your JAMA Network experience by selecting one or more topics from the list below.
Okland TS, Gonzalez JR, Ferber AT, Mann SE. Association Between Patient Review of Systems Score and Somatization. JAMA Otolaryngol Head Neck Surg. 2017;143(9):870–875. doi:10.1001/jamaoto.2017.0671
What is the relationship between medically unexplained symptoms, psychiatric illness, and the number of positive responses on a medical review of systems in otolaryngology patients?
In this retrospective medical record review that included 605 adults, the number of positive responses to a 69-item standardized review of systems was significantly higher in patients with medically unexplained symptoms, a history of psychiatric diagnoses, and symptoms associated with somatization (dizziness, globus sensation, and tinnitus).
The review of systems may be useful in identifying somatization as an underlying source of the symptoms in otolaryngology patients.
Somatization is a condition in which psychological distress is manifested by medically unexplained symptoms, and it is prevalent in all medical specialties, including otolaryngology. Recognition of somatization can be difficult, and there are limited methods available.
To determine whether patients with somatization respond differently to the review of systems (ROS) portion of the patient interview and whether the ROS can be used to identify patients with somatization.
Design, Setting, and Participants
A retrospective review of medical records of 2120 consecutive consultations of English- or Spanish-speaking patients aged 18 to 89 years who presented to the otolaryngology clinic from January 1, 2014, to November 10, 2015, was conducted to compare how the ROS of patients with chief complaints associated with somatization (group B: globus sensation, dizziness, and tinnitus) differs from those with symptoms more often associated with objective findings (group A: nasal obstruction, hoarseness, and hearing loss); a total of 605 patients were included. Objective clinical findings after physical examination and related testing were reviewed and classified as either significant, marginal, or absent. Current or past psychiatric comorbidities were also examined.
Main Outcomes and Measures
Number of affirmative responses on a standardized, 69-point ROS was recorded as a ROS score (ROSS). Objective clinical findings, symptoms, and psychiatric comorbidities were recorded.
Of the 605 patients included in the analysis, 346 (57.2%) were women, and the mean (SD) age was 51.6 (15.7) years. Among patients with medically unexplained symptoms (median, 11; range, 0-39), the ROSS was higher compared with those with objective clinical findings (median, 6; range, 0-31) (median difference, 4; 95% CI, 3 to 6). Group A (hoarseness, nasal obstruction, and hearing loss: median ROSS, 6, range, 0-41) exhibited lower ROSS than group B (dizziness, globus sensation, and tinnitus: median ROSS, 9; range, 0-39) (median difference, −2; 95% CI −3 to −1). Psychiatric comorbidity (median, 10; range, 0-41) was associated with higher ROSS than patients without psychiatric comorbidity (median, 5.5; range, 0 to 36) (median difference, 5; 95% CI, 3 to 6).
Conclusions and Relevance
The manner in which patients respond to a standardized ROS differs in those with medically unexplained symptoms and in those with psychiatric disease. The ROS offers information beyond the actual systems review, and may be useful in the identification of somatization.
Medically unexplained symptoms (MUS) are somatic symptoms for which a pathophysiologic mechanism cannot be identified. Patients with MUS are frequent utilizers of the health care system and account for up to one-third of all primary care evaluations and 21% of specialty consultations.1,2 This level of use represents a significant burden on the health care system, including otolaryngology. A community survey conducted by Kirmayer et al3 reported that 10.5% of participants had an unexplained symptom in the prior year. Ear, nose, and throat symptoms were the third most common MUS among those surveyed. Otolaryngologists are frequently consulted for symptoms that have limited physical examination findings. Some of these symptoms remain unexplained after further testing and are challenging to relieve. For instance, while a report of hearing loss will often be corroborated with findings on audiologic testing, objective abnormalities in those presenting with dizziness or tinnitus are less commonly found.4
There is, however, a clear connection between these types of symptoms and psychiatric disease. Patients reporting tinnitus or dizziness are more likely than others to experience anxiety or depression.5,6 Patients presenting with globus sensation (another symptom often associated with few objective findings) are more likely to have a diagnosis of major depression, generalized anxiety disorder, somatization disorder, posttraumatic stress disorder, and drug abuse.7 When a MUS is determined to be a manifestation of underlying psychological distress, it is called somatization.
Recognition of somatization can provide significant benefit for patients with MUS and prevent unnecessary or invasive testing. Even with extensive workups, these patients often perceive that their physicians have given insufficient medical explanation or treatment options, and this ambiguity perpetuates a cycle of distress and worsening of the somatic symptom.8,9 Prolonged somatization may result in debilitating symptoms refractory to treatment. Accumulating evidence suggests early recognition and intervention with cognitive behavioral treatment regimens can lead to significant improvement or remission.10
Early recognition also benefits the health care system. Patients with somatization disproportionately utilize medical services and have more emergency, primary care, mental health, and specialty visits; more outpatient procedures; and more hospital admissions.10 The annual health care costs attributed to somatization in the United States is an estimated $256 billion.11
Unfortunately, recognition of somatization can be challenging for the clinician.12 There are limited tools available to assist in the diagnosis. Some sources advocate for identifying risk factors with screening questionnaires or identifying characteristic patterns of care utilization, but these methods require resources and detailed medical records review that are not always feasible in a specialty clinic environment.
One common belief among physicians is that patients with somatization respond differently during a review of systems (ROS). Because these patients often present with wide-ranging symptoms affecting multiple organ systems, they may manifest a “pan-positive” ROS during the patient interview. An increased number of reported symptoms on the ROS has been shown to correlate with characteristics of borderline personality disorder.13 It has also been shown to have predictive value when delineating epileptic from psychogenic nonepileptic seizures.14 In a study of patients with gastrointestinal symptoms, an increased number of symptoms reported on the ROS was found to be sensitive and specific in indicating which patients were eventually deemed to have a functional gastrointestinal disorder.15
We hypothesized that the manner in which patients respond to a ROS could be useful in identifying somatization in a general otolaryngology practice. Specifically, we sought to understand how responses to a standardized ROS would differ in patients with symptoms associated with somatization (globus sensation, dizziness, and tinnitus) compared with those with symptoms more often associated with objective findings (nasal obstruction, hoarseness, and hearing loss). In addition, we investigated how the number of positive responses during a ROS related to subsequent objective findings. Current or past psychiatric comorbidities were also examined.
We performed a retrospective medical records review of 2120 consecutive outpatient consultations at a community otolaryngology clinic seen from January 1, 2014, to November 10, 2015. We included patients who presented with 1 of 6 predetermined chief complaints: hearing loss, hoarseness, and nasal obstruction (group A), and globus sensation, tinnitus, and dizziness (group B). Only English- or Spanish-speaking patients, aged 18 to 89 years, seen by 1 of 3 attending otolaryngologists (including SEM) were included in the study. In total, the records of 605 patients were examined, 346 (57.2%) of whom were women, and 259 (42.8%) were men. All data were collected under the approval of Colorado Multiple Institutional Review Board and entered into a database for statistical analysis; data were deidentified.
A detailed medical records review of the initial consultation and all subsequent or related testing was performed. All patients had completed a standardized 69-point ROS questionnaire (eAppendix in the Supplement). This form was available in English and Spanish. We recorded the number of affirmative responses as a ROS score (ROSS). The patient records were also examined for history of psychiatric illness. The medical records review was completed in a systematic fashion to minimize reviewer bias with acquisition of information in the following order: chief complaint, clinical findings, ROS analysis, and presence of psychiatric comorbidities. The determination of clinically significant findings was defined prior to medical records review as presence of any clearly documented physical examination (including endoscopy), audiometry, or imaging study abnormalities that were causally related to the symptoms by the treating physician. Marginal findings were defined as nonspecific physical examination results or findings documented as very mild or minor. Medically unexplained symptoms were defined as those in which no related objective findings were found in initial or subsequent evaluations.
For investigation of trends, a cluster analysis was used to separate patients into groups formed by the distribution of ROSS. A 2-step cluster analysis was performed, revealing that the data fit into 4 groupings: 0 to 3, 4 to 7, 8 to 13, and higher than 14. The quality of these cluster groupings was confirmed with a goodness-of-fit and pseudo R2 analyses. Statistical analysis was conducted in MATLAB (The MathWorks Inc). The ROSS data were used to compare various groups; median differences and 95% CIs were calculated.16
A total of 605 patients were included, with 387 comprising group A (nasal obstruction [n = 111], hoarseness [n = 74], and hearing loss [n = 202]), and 218 in group B (globus sensation [n = 54], dizziness [n = 83], and tinnitus [n = 81]). The mean (SD) age of all participants was 51.6 (15.7) years, with 346 women (57.2%) and 259 men (42.8%). A total of 523 (86.4%) patients were English speaking, and 82 (13.6%) were Spanish speaking. There were no differences in age between group A (median, 53; range, 16-90) and B (median, 54; range, 19-84) (median difference, −1 year; 95% CI, −4 to 2 years). Group A had fewer women with 204 (52.7%) vs 142 (65.1%) in group B. Of the 605 total patients, 95 (15.7%) had no objective findings in initial or subsequent evaluations that explained their presenting symptom. A total of 170 (28.1%) patients were found to have marginal objective findings that could explain their chief complaint. Three hundred forty (56.2%) patients presented with significant objective findings.
As expected, the ROSS data were not normally distributed (Figure 1). A Spearman rank-order correlation test indicated that ROSS was not related to age (rs = −0.016; 95% CI, −0.083 to 0.051). Median ROSS for patients presenting with significant findings was 6 (range, 0-31) vs 11 (range, 0-39) for those with MUS (Figure 2A). Patients with significant findings exhibited a lower ROSS (median difference, −4; 95% CI, −6 to −3). Median ROSS for patients in group A was 6 (range, 0-41) compared with 9 (range, 0-39) in group B (Figure 2B); median ROSS was lower for patients in group A (median difference, −2; 95% CI, −3 to −1). The prevalence of MUS was 7.5% in group A patients and 30.1% in group B patients.
An inverse relationship was found between the ROSS and the significance of clinical findings. Of patients who had a ROSS of 0 to 3, significant objective findings were identified in 115 of 153 (75.2%); of those who had a ROSS of 4 to 7, significant objective findings were identified in 98 of 156 (62.8%); of those who had a ROSS of 8 to 13, significant objective findings were identified in 83 of 162 (51.2%); and of patients who had a ROSS of 14 or higher, significant objective findings were identified in 44 of 134 (32.8%) (Figure 3A). Group A patients with no clinical findings had a median ROSS of 8 (9 for marginal findings and 5 for significant objective findings). Group B patients with no clinical findings had a median ROSS of 11 (8 for marginal findings and 6 for significant objective findings).
When only group A patients were examined, of those with a ROSS of 0 to 3, significant objective findings were identified in 96 (87.3%); of those with a ROSS of 4 to 7, significant objective findings were identified in 86 (77.5%); of those with a ROSS of 8 to 13, significant objective findings were identified in 63 (67.7%); and of those with a ROSS of 14 or higher, significant objective findings were identified in 38 (52.1%) (Figure 3B). When only group B patients were examined, of those with a ROSS of 0 to 3, significant objective findings were observed in 19 (44.2%); of those with a ROSS of 4 to 7, significant objective findings were observed in 12 (26.7%); of those with a ROSS of 8 to 13, significant objective findings were observed in 20 (28.99%); and of those with a ROSS of 14 or higher, significant objective findings were observed in 6 (9.84%) (Figure 3C). Overall, 233 (38.5%) patients included in this study had psychiatric comorbidity (163 [47.1%] women and 79 [27.0%] men).
The presence of psychiatric comorbidity was associated with an increased ROSS. The median ROSS was 11.5 for those with a history of depression (n = 156), 12 for anxiety (n = 72), 9 for fibromyalgia/chronic pain (n = 91), and 12 for posttraumatic stress disorder (n = 4). This was in contrast to a median ROSS of 5.5 for patients with no psychiatric history. Analysis revealed a difference in ROSS scores between patients with (median, 10; range, 0-41) and without (median, 5.5; range, 0-36) psychiatric disease, where patients with psychiatric disease exhibited higher a ROSS (median difference, 5; 95% CI, 3 to 6). This difference was present and in both groups A and B, although, of patients in group A, only 31.2% possessed psychiatric comorbidity, compared with 51.4% in group B. In group A, patients with a psychiatric history were found to have a greater median ROSS of 10 (range, 0-41), compared with 5 (range, 0-36) in those without (median difference, 5; 95% CI, 3-6). Group B patients with psychiatric comorbidity were found to have a greater median ROSS of 11 (range, 0-39), compared with 7 (range, 0-32) in those without (median difference, 4; 95% CI, 2-6).
The prevalence of psychiatric illness was directly related to ROSS (Figure 4A). For patients in group A who reported 0 to 3, 4 to 7, 8 to 13, and 14 or more symptoms, the prevalence of psychiatric diagnoses was 16.4%, 24.3%, 33.3%, and 61.6%, respectively. For patients in group B who reported 0 to 3, 4 to 7, 8 to 13, and 14 or more symptoms, the prevalence of psychiatric diagnoses was 37.2%, 35.6%, 55.1%, and 68.9%, respectively. In addition, a Spearman rank-order correlation indicated a moderate positive association between the number of psychiatric diagnoses and the ROSS seen in all patients (rs = 0.3511; 95% CI, 0.291-0.408). Median ROSS for patients with no psychiatric diagnosis was 5.5; for those with 1 psychiatric diagnosis, 10; for 2 diagnoses, 9; and for 3 diagnoses, 22 (Figure 4B).
Somatization is an important phenomenon to recognize in otolaryngology. In addition to the associated health care costs, it can be debilitating for patients. Extensive investigations into MUS can be counterproductive for the patient and lead to worsened symptoms. It is important that specialists be aware of somatization as a possible source of the symptoms and include it in their differential diagnosis. With this study, we investigated whether the ROS could be helpful in identifying somatization in otolaryngology patients and to identify trends in regard to MUS or psychiatric illness. The chief complaints investigated were chosen based on the previous associations of somatization with dizziness, tinnitus, and globus sensation.4-7 As a comparison group, we chose 3 symptoms that we considered to have frequent objective findings in a similar patient population: nasal obstruction, hearing loss, and hoarseness. Within these chief complaints there was a 15.7% rate of MUS (no objective findings on subsequent evaluation/testing). When the ROS responses were examined, they were found to be different in patients with MUS. The number of positive responses was higher than in those who had definite objective findings explaining their symptoms (Figure 2A). These results support the notion that the ROS responses are different in patients with MUS. In our 2 cohorts, group A (nasal obstruction, hoarseness, and hearing loss) had fewer positive responses on the ROS compared with group B (globus sensation, dizziness, and tinnitus) (Figure 2B). This finding suggests further that the ROS may be answered differently by patients with somatization. However, this variation could be confounded by differences in psychiatric comorbidity between groups. There was a higher rate of psychiatric history in group B (112 [51.4%] vs 121 [31.3%] patients in group A). Our results revealed that psychiatric illness had an effect on the ROSS, which was higher in patients with psychiatric comorbidity and showed a trend of increasing further with multiple psychiatric diagnoses (Figure 4B). This finding suggests that the ROS could potentially be helpful in the evaluation of other psychiatric conditions as well. In our study, there was a much higher number of MUS in group B (18.81% vs 4.91% in group A), and in both groups there was an apparent inverse relationship between responses on the ROS and objective findings (Figure 3).
In patients with somatization, psychiatric distress manifests as MUS. Our analysis showed an increase in positive responses on the ROS with psychiatric comorbidity and also with MUS. This finding suggests that patients with somatization will have a ROS that has more positive responses than those without.
There were MUS, psychiatric disease, and high ROSS present within all patient groups. Despite the trends observed, somatization may be present in any clinical scenario and our results support this understanding. This study and future work hopefully will aid otolaryngologists in the clinical evaluation of patients with MUS. When faced with MUS, a clinician should consider somatization in the differential diagnosis, especially if the ROS has a high number of positive responses.
There are several clear limitations of this study. None of the patients were formally evaluated for somatization at the time of their consultations, so the actual incidence of somatization in the study population is unknown. Also, the patient population included was from a “safety-net” medical center, and the results may not be applicable in other populations. Our overall prevalence of psychiatric diagnoses was elevated at 38.8%. In addition, this was a retrospective review of medical records that likely included inherent biases of the treating physicians. Patients with a multitude of symptoms on the ROS may have received a different style of evaluation and subsequent documentation. Prospective controlled studies would be needed to validate any predictive value of ROS as a diagnostic tool for somatization.
Medically unexplained symptoms and somatization are common to all medical fields, including otolaryngology. Identifying a somatic symptom as a manifestation of underlying psychiatric distress offers significant benefit to patients as well as to the health care system. The manner in which patients respond to a standardized ROS differs in those with MUS and in those with psychiatric disease. Increasing positive responses on a patient’s ROS was associated with less objective evidence of disease. The ROS offers valuable information beyond the actual systems review and may be useful in the identification of somatization.
Accepted for Publication: April 7, 2017.
Corresponding Author: Scott Edward Mann, MD, Department of Otolaryngology, University of Colorado School of Medicine, 12631 E 17th Ave, B205, Aurora, CO 80045 (firstname.lastname@example.org).
Published Online: June 15, 2017. doi:10.1001/jamaoto.2017.0671
Author Contributions: Mr Okland and Dr Mann had full access to all of the data in this study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Okland, Gonzalez, Mann.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Okland, Gonzalez, Mann.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Okland, Ferber, Mann.
Study supervision: Mann.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: Dr Ferber received support from the National Institutes of Health/National Institute on Deafness and Other Communication Disorders fellowship F30-DC013932.
Role of the Funder/Sponsor: The National Institutes of Health/National Institute on Deafness and Other Communication Disorders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Hannah Dauber, MS (University of Colorado), provided statistical support; there was no financial compensation.