Predictors of Non-Rhinogenic Facial Pain or Pressure in Otolaryngology Clinic

Abstract

Objectives:

Identify demographic and clinical characteristics that may help differentiate non-rhinogenic facial pain or pressure (NRFP) from sinusitis.

Study Design:

Retrospective single-institution study

Setting:

Tertiary care center rhinology clinic

Methods:

All patients presenting with a complaint of facial pain or pressure over a three-year period were included. Patients were categorized into either NRFP or sinusitis groups based on CT imaging and nasal endoscopy. Data pertaining to demographics, history, and SNOT-22 questionnaire domains were compared via univariate analysis as well as logistic regression with backwards variable selection.

Results:

296 patients met inclusion criteria, of which 128 had NRFP and 168 had sinusitis. A significantly greater percentage of patients in the NRFP group were women of childbearing age (40.6% vs 28.0%, p = 0.02). Backwards variable selection resulted in a model with four variables predicting a diagnosis of NRFP – female sex (OR = 2.998, p < 0.0001), no history of prior sinonasal surgery (OR = 0.340 for history vs no history, p < 0.01), low nasal domain score (OR = 0.551, p < 0.0001), and high ear/facial domain score (OR = 1.453, p < 0.01).

Conclusion:

Accurately identifying patients with NRFP at initial presentation based on history would help direct patients to the appropriate care pathway and prevent ineffective treatments such as antibiotics and sinus procedures. Our findings suggest that the suspicion for NRFP should be higher in women of child-bearing age as well as patients with greater ear/facial symptoms or lesser nasal symptoms.

Introduction:

Facial pain or pressure is a common presenting symptom in both the primary care and otolaryngology settings. While it may be easily attributed to sinusitis (acute or chronic), evidence suggests that it is more likely a result of neurogenic causes, namely migraine. Eross et al showed that amongst 100 patients diagnosed with “sinus headache”, only 3% actually had sinusitis, whereas 86% met criteria for migraine or other primary headache diagnosis¹. In another study by Schreiber et al of almost 3000 patients with “sinus headache,” 88% of patients were determined to have migraine². Similarly, a study by Barham et al demonstrated that 95% of patients with acute sinus symptoms had no objective findings of sinus inflammation on point of care CT imaging³. Use of pain in the diagnostic criteria for sinusitis has even been called into question recently, with Hirsch et al showing that the elimination of pain improved specificity of clinical diagnostic criteria for CRS from 37% to 65%⁴.

The implications for the misdiagnosis of acute and chronic sinusitis are significant. The direct cost of CRS alone is $10 billion annually in the United States⁵, with the majority of expenditures attributed to outpatient visits and antibiotic prescriptions with primary care providers⁶. Based on the high prevalence of migraine in the general population, it is conceivable that the majority of these patients have facial pain or pressure, and the majority of this subset do not have sinusitis. There is also evidence that procedures such as balloon sinus dilation are sometimes performed for patients with non-rhinogenic pain or pressure⁷. All of this leads to inefficient care and a substantial cost to the healthcare system.

The identification of non-rhinogenic facial pain or pressure (NRFP) and differentiating it from sinusitis is challenging especially without point of care imaging and nasal endoscopy. Neurogenic causes of facial pain and pressure are associated with nasal autonomic symptoms such as congestion and rhinorrhea, further increasing the likelihood of misdiagnosis⁸.

Prior studies have looked to better characterize NRFP. Jones described a diagnosis of ‘midfacial segment pain,’ which has the characteristics of tension type headache in a midfacial distribution⁹. This was the most common diagnosis (65%) found by Agius et al in a cohort of 240 NRFP patients, followed by migraine meeting International Headache Society criteria (25.5%) and other diagnoses including trigeminal neuralgia, cluster headaches, and others¹⁰. A study by Wu et al utilized the SNOT-22 questionnaire to demonstrate that NRFP patients have response patterns distinct from those of sinusitis patients, with a higher ratio of ear/facial symptoms to nasal symptoms¹¹. Despite these prior studies, identification of NRFP based on initial history and examination remains difficult without a larger body of evidence to guide clinical practice. The present study aims to identify clinical features in patients with NRFP that could help differentiate it from sinusitis. We hypothesized that demographic features and symptom patterns as assessed by the SNOT-22 would be significantly different in patients with NRFP.

Methods:

This is a retrospective single-institution study of patients who presented to an academic rhinology clinic between January 2019 and January 2022. The study protocol was reviewed and approved for exemption by the Duke University Health System Institutional Review Board. All adult patients who listed facial pain/pressure as a complaint on the history of present illness or on the SNOT-22 questionnaire and underwent a CT scan of the sinuses in the preceding 12 months were selected. Patients were excluded if they had an underlying neoplastic process, or if they met criteria for both NRFP and CRS at different time points. Patient charts were reviewed and data pertaining to demographics, comorbidities, nasal endoscopy, CT scan, and SNOT-22 were collected. Nasal endoscopy findings and SNOT-22 data from the initial visit were utilized.

CT images and nasal endoscopy were used to categorize patients into the sinusitis group or the NRFP group. Patients with significant mucosal thickening on CT during a symptomatic period or signs of inflammation (polyps, purulent drainage) at the middle meatus or sphenoethmoid recess on nasal endoscopy as determined by an attending rhinologist were categorized as having sinusitis. Non-circumferential sinus mucosal thickening less than 2mm or a single mucous retention cyst were considered non-significant.

Patients with SNOT-22 questionnaires that were less than 50% completed were excluded. For patients with questionnaires with at least 50% of the SNOT-22 questions completed, missing items were imputed with the mean of the completed items based on prior published methodology¹². Univariate analysis of patient characteristics was performed using Kruskal Wallis or Chi-Square tests. Backwards variable selection with logistic regression was then performed with the following variables entered into the model: age, sex, race, ethnicity, BMI, female of child-bearing age, asthma, migraine history, current smoker, prior sinus surgery, and SNOT-22 domains (nasal, ear/facial, sleep, function, emotion)¹³. Child-bearing age was defined as 15–49 years of age in keeping with the World Health Organization criteria. Analyses were performed in SAS 9.4 (SAS Institute Inc., Cary, NC).

Results:

376 patients met inclusion criteria. After patients with incomplete SNOT-22 data and missing race were excluded, 296 patients were included in the analysis. 128 were in the NRFP group and 168 were in the sinusitis group. Table 1 displays demographic and clinical characteristics of each group. Females made up 77% of the NRFP group compared with 52% of the sinusitis group (p < 0.0001). Females of child-bearing age made up a significantly larger proportion of the NRFP group (41%) compared with the sinusitis group (28%) (p = 0.0223). SNOT-22 data is shown in Table 2. The SNOT-22 nasal domain score tended to be higher in the sinusitis group (mean 2.2 ± 1.0) compared to the NRFP group (mean 1.8 ± 1.0) (p = 0.0004). The other SNOT-22 domains were distributed similarly for both NRFP and sinusitis.

Table 1:

Patient Characteristics

	NRFP (N=128)	Sinusitis (N=168)	Total (N=296)	P-Value
Age				0.851
Mean (SD)	49.4 (17.2)	50.1 (17.7)	49.8 (17.4)
Sex				<0.00012
Female	99 (77.3%)	88 (52.4%)	187 (63.2%)
Female of Child-Bearing Age				0.022
Yes	52 (40.6%)	47 (28.0%)	99 (33.4%)
Race				0.742
Asian	5 (3.9%)	3 (1.8%)	8 (2.7%)
Black	24 (18.8%)	32 (19.0%)	56 (18.9%)
Multiracial/Other	8 (6.3%)	11 (6.5%)	19 (6.4%)
White	91 (71.1%)	122 (72.6%)	213 (72.0%)
Ethnicity				0.092
Hispanic	3 (2.3%)	11 (6.5%)	14 (4.7%)
Not Hispanic	125 (97.7%)	157 (93.5%)	282 (95.3%)
BMI				0.452
Obese	49 (38.3%)	75 (44.6%)	124 (41.9%)
Overweight	49 (38.3%)	62 (36.9%)	111 (37.5%)
Underweight/Healthy Weight	30 (23.4%)	31 (18.5%)	61 (20.6%)
Smoking				0.832
Yes	6 (4.7%)	7 (4.2%)	13 (4.4%)
Migraine History				0.112
Yes	30 (23.4%)	27 (16.1%)	57 (19.3%)
Migraine Medications				0.162
Yes	18 (14.1%)	15 (8.9%)	33 (11.1%)
Asthma History				0.842
Yes	36 (28.1%)	49 (29.2%)	85 (28.7%)
Allergy History				0.332
Yes	84 (65.6%)	101 (60.1%)	185 (62.5%)
Polyps History				0.012
Yes	0 (0.0%)	9 (5.4%)	9 (3.0%)
OSA History				0.492
Yes	19 (14.8%)	30 (17.9%)	49 (16.6%)
Sinonasal Surgery History				<0.012
Yes	19 (14.8%)	55 (32.7%)	74 (25.0%)
Laterality				0.822
Bilateral	72 (56.3%)	91 (54.2%)	163 (55.1%)
Unilateral	28 (21.9%)	35 (20.8%)	63 (21.3%)
Missing	28 (21.9%)	42 (25.0%)	70 (23.6%)
Chronicity				0.922
Chronic	13 (10.2%)	19 (11.3%)	32 (10.8%)
Episodic	78 (60.9%)	99 (58.9%)	177 (59.8%)
Missing	37 (28.9%)	50 (29.8%)	87 (29.4%)
Lund Mackay Score				<0.00011
Mean (SD)	0.5 (0.8)	7.5 (5.1)	4.5 (5.2)
Contact Point				0.062
Yes	9 (7.0%)	25 (14.9%)	34 (11.5%)

¹Kruskal Wallis

²Chi-Square

Table 2:

SNOT-22 Responses

	NRFP (N=128)	Sinusitis (N=168)	Total (N=296)	P-Value
SNOT-22 Total Score				0.201
Mean (SD)	42.0 (19.5)	44.5 (20.7)	43.4 (20.2)
SNOT-22 Nasal Domain				<0.011
Mean (SD)	1.8 (1.0)	2.2 (1.0)	2.1 (1.0)
SNOT-22 Ear/Facial Domain				0.111
Mean (SD)	1.9 (1.1)	1.7 (1.1)	1.8 (1.1)
SNOT-22 Sleep Domain				0.681
Mean (SD)	2.3 (1.4)	2.4 (1.4)	2.3 (1.4)
SNOT-22 Function Domain				0.731
Mean (SD)	2.3 (1.5)	2.2 (1.4)	2.2 (1.4)
SNOT-22 Emotion Domain				0.901
Mean (SD)	1.3 (1.3)	1.2 (1.2)	1.2 (1.2)

¹Kruskal Wallis

Table 3 displays the results of the logistic regression analysis after using the backwards variable selection method where the probability modeled is for an outcome of NRFP. Four variables were kept in the final model: sex, prior sinus surgery, SNOT-22 nasal domain, and SNOT-22 ear/facial domain. In this model, females had higher odds of NRFP compared to males (OR (CI): 2.998 (1.733, 5.186), p<0.0001). Patients who had a prior surgery had lower odds of an NRFP diagnosis compared to those who did not have a prior surgery (OR (CI): 0.340 (0.181, 0.639), p=0.0008). As SNOT-22 nasal domain scores increase, the odds of NRFP decrease; that is, higher scores in the SNOT-22 nasal domain were associated with lower odds of an NRFP diagnosis (OR (CI): 0.551 (0.414, 0.733), p<0.0001). In contrast, higher scores in the SNOT-22 ear/facial domain were associated with higher odds of an NRFP diagnosis (OR (CI):1.453 (1.118, 1.888), p=0.0052).

Table 3:

Logistic Regression OR Estimates for Diagnosis Probability Modeled: Outcome is NRFP^**

Predictor	OR	95% C.I.	P-Value*
Sex (ref: Male)
Female	2.998	(1.733, 5.186)	<0.0001
Prior Sinus Surgery (ref: No)
Yes	0.340	(0.181, 0.639)	0.0008
SNOT Nasal Domain	0.551	(0.414, 0.733)	<0.0001
SNOT Ear/Facial Domain	1.453	(1.118, 1.888)	0.0052

^*Chi-Square. N=296

^**Hosmer and Lemeshow goodness of fit chi-squared test 8.94, df=8, p-value = 0.35

C-index = 0.733

Discussion:

This study sought to identify demographic and clinical characteristics that could help to differentiate NRFP from sinusitis in patients presenting with facial pain or pressure. We found that women of child-bearing age were a larger percentage of NRFP cases compared to sinusitis cases. We believe that this is a specific demographic feature that can raise initial clinical suspicion for NRFP. This finding is consistent with the broader literature on migraine and other craniofacial pain disorders. A review by Vetvik and MacGregor notes that migraine is more prevalent in women than men, with greater duration and severity of symptoms¹⁴. Other craniofacial pain disorders have also shown prevalence up to three times higher in women than men, presumably due to the effect of sex hormones on nociceptive signaling^15,16.

Our multivariate model demonstrated that while higher nasal domain scores were associated with sinusitis, higher ear/facial domain scores were associated with NRFP. This is consistent with studies by Wu et al and Sharbel et al, which also reported prominence of ear/facial symptoms in relation to nasal symptoms in patients with NRFP^11,17. Indeed, our clinical experience has been that patients with NRFP will often complain of ear fullness and vestibular problems in the setting of normal ear exams. With increasing evidence from multiple institutions supporting this, we believe that this can serve as an important clinical differentiator in everyday practice. At the same time, it should be noted that patients with migraine or other primary headache disorder may still experience prominent rhinologic symptoms, due to potential associated autonomic dysfunction of the nasal cavity⁸.

Prior sinus surgery was predictive of sinusitis diagnosis in our model. This result seems intuitive given that many patients with rhinosinusitis may present with recurrent symptoms after having prior surgical management, but it is noteworthy that a significant number of patients with NRFP diagnosis did have prior surgery in our series (14.8%), and this rate has been noted as high as 41% elsewhere in the literature⁸. Because the details surrounding the prior sinus procedures are unknown, we cannot say if these NRFP patients truly had CRS at some point in the past, or if these patients had surgery for inappropriate indications. It is also conceivable that an NRFP patient develops sinusitis as a result of unindicated surgery. Assuming that NRFP patients appropriately had sinus surgery for CRS, this finding serves as a reminder that having one of these diagnoses at one time point does not exclude a patient from having the other at a later point in time.

Our study is limited by the retrospective nature of the study design and small sample size. Review of clinical documentation yielded very little information on the characteristics surrounding patients’ facial pain or pressure such as alleviating/aggravating factors, quality, frequency/duration of episodes, and location. However, the findings of our study can help to raise awareness of identifying NRFP early on in the clinical care pathway. Future research on this topic should take a prospective approach using standardized questionnaires to gather more detailed data on the nature and chronicity of facial pain/pressure. While our model suggests that high ear/facial domain and low nasal domain scores are indicative of NRFP diagnosis, our data were insufficient to determine what threshold values should be used to define a ‘high’ or ‘low’ score for these domains using receiving operator characteristic curves, and further research should seek to delineate this to provide a more readily applicable clinical heuristic.

Conclusion

Accurately identifying patients with NRFP at initial presentation based on history would help to direct patients to the appropriate care pathway early on and prevent costly and ineffective diagnostic tests and treatments. The suspicion for NRFP should be higher in women of child-bearing age as well as patients with more ear/facial domain symptoms or fewer nasal domain symptoms, and the suspicion for CRS should be higher for patients with a history of previous sinus surgery as well as patients with higher scores in the SNOT-22 nasal domain. Our findings support the idea that symptom patterns and demographic characteristics can help to differentiate NRFP from CRS.