Diagnostic imaging has evolved considerably in the past several decades, giving clinicians an increasing ability to diagnose and treat disease (1). Specifically, examination of the central nervous system using computed tomography scans (CT) and magnetic resonance imaging (MRI) is at increasing frequency over the past several years. Rao et al (2) investigated neuroimaging utilization rates in the United States and observed a 36% increase in tests performed in 5 years, from 13,987 neuroimaging tests per 100,000 Medicare patients in 1993 to 19,431 tests per 100,000 patients in 1998. The number of MRI scanners has more than doubled in the past decade in the United States, growing from 12.3 scanners per million population in 1995 to 26.5 scanners per million population in 2006 (3). Health care costs have also escalated during this time, which is partly attributed to growth in medical technology (1,4).
Although neuroimaging provides valuable information about various conditions of the head and neck, it is done at considerable cost (5,6). In an era fraught with increasing malpractice claims, concerns have been raised about overuse of imaging, which may do little to alter outcome or management (7). As medical technology advances in the setting of limited resources, physicians are posed with the dilemma of choosing a diagnostic test that is high yield and cost-effective.
Although several studies have evaluated the diagnostic yield of imaging in specific neuro-ophthalmic conditions, the diagnostic and economic yield of imaging in neuro-ophthalmic conditions (7–13) overall has not been investigated. We evaluated the diagnostic yield of neuroimaging in patients seen in the neuro-ophthalmology service at our institution and estimated its economic impact on clinical practice.
This retrospective chart review included all patients referred by the neuro-ophthalmology service at Scheie Eye Institute (University of Pennsylvania Health System) for head and neck evaluation for the following studies: CT, CT angiography, MRI, MRA, and magnetic resonance venography (MRV). These data were gathered over a 12-month period from September 2006 to August 2007 from the practice of 3 neuro-ophthalmologists at the Scheie Eye Institute. No specific criteria were used for referral for neuroimaging. Excluded patients were those undergoing follow-up imaging for known lesions and those who had been impaged prior to our evaluation. Study approval was obtained from the University of Pennsylvania Institutional Review Board.
Data collected included demographic information, reason for imaging referral, imaging modality, and findings. Clinical history was obtained, including symptoms, examination results, diagnosis, and management. Imaging findings were described by significance and relevance. Findings were defined as significant if they led to changes in patient management. A finding was defined as relevant if it was a related to the patient's neuro-ophthalmic symptoms or clinical deficits. Results were then classified into 1 of 5 groups: 1) significant and relevant, 2) significant and not relevant, 3) relevant and not significant, 4) not significant and not relevant, or 5) normal. The percentage of tests with a significant and relevant finding was defined as the diagnostic yield.
Subgroup analyses were performed by evaluating the diagnostic yield based on presenting symptoms. The chief complaint of the patient was divided into the following categories: 1) decreased vision, 2) double vision, 3) unexplained visual phenomena, 4) headache, 5) eye pain, 6) face pain, 7) eye bulging, 8) droopy lid, 9) other, and 10) no complaint. Patients without symptoms but found to have a concerning clinical finding also underwent neuroimaging and were included in the analysis. Decreased vision included patients complaining of transient or permanent loss of vision, blurry vision, or scotomatous quality of vision loss. Visual phenomena included patients complaining of positive visual disturbances such as scintillating scotoma. Of note, in patients with multiple symptoms, the chief complaint was used to assign a symptom category. For example, in a patient with optic neuritis who noted eye pain and decreased vision, if the chief complaint was vision loss, the patient was placed in the decreased vision category.
Subgroup analyses were performed for principal neuro-ophthalmic examination findings for which patients were placed in the following categories: 1) optic disc pallor, 2) optic disc swelling, 3) visual field defect, 4) relative afferent pupillary defect (RAPD), 5) decreased color vision, 6) motility deficit, 7) proptosis, 8) ptosis, 9) anisocoria, 10) strabismus, 11) other, or 12) normal. Patients in the strabismus category had an esodeviation, exodeviation, or hyperdeviation without a motility deficit. Patients with multiple examination findings, such as an RAPD, visual field defect, disc pallor, and disc swelling, were placed in one category with the most significant finding based on the following hierarchy from greatest to least significance: disc abnormalities, visual field defect, RAPD, decreased color vision, and decreased visual acuity. For example, a patient with optic disc pallor and RAPD was placed in the disc pallor category and a patient with RAPD and decreased color vision was placed in the RAPD category. Patients were placed under decreased visual acuity if corrected acuity was worse than 20/20 and no other examination findings were present.
Subgroup analyses were also performed based on indication for neuroimaging. Patients were placed into the following categories: 1) optic neuritis, 2) optic neuropathy, 3) cranial nerve palsy, 4) transient visual loss, 5) unexplained visual loss, 6) visual phenomena concerning for intracranial mass, 7) headache concerning for mass lesion, 8) proptosis or motility deficit concerning for thyroid eye disease, 9) ptosis concerning for structural lesion, 10) internuclear ophthalmoplegia concerning for white matter lesions, 11) visual field defect concerning for mass or stroke, 12) disc swelling concerning for mass, or 13) other. Studies were placed in the “other” category if 2 or less imaging studies were performed for a specific indication in the entire series.
The cost of neuroimaging was determined using the Medicare Resource-Based Relative Value Scale obtained from the Centers of Medicare and Medicaid Services (14). Resource-Based Relative Value Units for the professional and technical components of an imaging examination were added and multiplied by the Medicare conversion factor for 2011 ($33.9764) to obtain cost of the imaging study (Table 1). These values do not reflect hospital charges to insurers or patients but estimate actual costs used by Medicare for reimbursement and are universally applied (15). No regional adjustment was performed. Total costs for all imaging studies performed and costs per significant relevant finding were then calculated.
The overall yield rate and its 95% confidence interval (CI) of all studies combined was calculated, and yield rates by chief complaint, clinical findings, and indication for neuroimaging were also calculated. P values were determined using the χ2 test. In these analyses, the unit of statistical analysis was per imaging study rather than per patient. In the calculation of 95% CI and P values for comparisons of imaging yield rate, the correlation among the multiple image studies from a patient was accounted for by using generalized estimating equations (16).
One thousand six hundred eight new patients and 4,380 total patients were seen by the neuro-ophthalmology service (NJV, KSS, and MAT) at the Scheie Eye Institute from September 2006 to August 2007. Two hundred eleven imaging studies in 157 patients met inclusion criteria for this study. Mean age of this cohort was 52.7 years (standard deviation, 17.7 years; range, 17–89 years). There were 77 male patients (49.0%) and 80 female patients (51.0%).
Of the 211 imaging studies performed, 61 imaging studies (28.9%; 95% CI, 22.5%–36.2%) had significant and relevant findings to the patient's neuro-ophthalmic condition (Table 2). The majority of these cases (36/61, 59%) were imaged for evaluation of optic neuropathy, optic neuritis, and cranial nerve palsies. Five of 211 imaging studies (2.4%; 95% CI, 1.0%–5.6%) had significant findings unrelated to the patient's neuro-ophthalmic condition but required further evaluation. They included one patient with chronic pansinusitis and one with an anterior communicating artery aneurysm, both presenting with unrelated visual phenomena; one patient with an asymptomatic frontal meningioma; a pineal cyst in 2 patients who presented with an unrelated visual field defect; and one patient with optic nerve pallor and enlargement of the pituitary gland without compression of visual pathways. In 4 patients (1.9%; 95% CI, 0.7%–4.9%), MRI revealed a finding that was relevant to the diagnosis but did not change the management. These included elevated optic discs in a patient with bilateral optic disc swelling, optic atrophy in a patient with disc pallor, dilated optic nerve sheaths in a patient with bilateral papilledema, and an empty sella in a patient with headaches, papilledema, and a clinical diagnosis of idiopathic intracranial hypertension (IIH). In 19.4% of cases (41/211; 95% CI, 14.6%–25.3%), imaging findings were not significant and not relevant. These cases mostly represented mild sinusitis and small vessel white matter ischemic disease. Of the 211 imaging studies performed, 100 (47.4%) were reported as normal.
Total cost of all imaging studies performed in 2011 US dollars was $107,615.72, and the cost per study that was significant and relevant was $1,764.19. If a normal MRI included a significant and relevant finding in patients ultimately diagnosed with IIH and optic neuritis (patients without high-risk profiles for developing multiple sclerosis), then cost per clinically significant relevant finding was $1,454.27.
For subgroup analysis performed based on the patient's chief complaint (Table 3), those imaged for proptosis had the highest diagnostic yield (66.7%) followed by those with eye pain (60.0%). This latter group mostly consisted of patients with myositis and optic neuritis. Patients undergoing imaging for double vision or decreased vision had a diagnostic yield of 30.6% and 30.5%, respectively. There was no diagnostic yield in patients imaged for headache or facial pain (0% in each group). Normal imaging was reported mainly in patients presenting with facial pain (100%) and headaches alone (63.6%).
Subgroup analysis based on the principal neuro-ophthalmic examination finding is summarized in Table 4. Patients with an RAPD and a normal-appearing optic nerve and patients with proptosis had the highest diagnostic yields of 72.7% and 66.7%, respectively. Findings in the former group were primarily optic nerve enhancement consistent with an optic neuritis or other cause for an optic neuropathy such as mass lesion. In the latter group, results demonstrated an orbital disease processes such as extraocular muscle enlargement due to thyroid eye disease. Two of 15 neuroimaging studies performed on patients with normal neuro-ophthalmic examination findings had significant and relevant findings. These 2 patients had transient visual loss, and imaging revealed small vessel ischemic disease in one case and internal carotid artery stenosis in the other.
Table 5 shows the diagnostic yield based on indication for neuroimaging. Patients with optic neuritis had a diagnostic yield of 72.0%, those with proptosis and/or a motility deficit suggestive of thyroid eye disease had a diagnostic yield of 70.0%, and individual with optic neuropathy had a diagnostic yield of 25.6%. Optic neuropathy was primarily due to a compression lesion. Diagnostic yield of imaging for cranial nerve palsy was 27.0%. Imaging performed for evaluation of optic neuritis, optic neuropathy, thyroid eye disease, and cranial nerve palsy had statistically significant higher diagnostic yield than studies performed for other reasons (P < 0.001).
In our study, 28.9% of neuroimaging tests requested by neuro-ophthalmologists resulted in an abnormal finding relevant to the patient's neuro-ophthalmic condition and was clinically significant for the management of this condition. We found that the cost per clinically significant and relevant finding is $1,764.19, which is similar to the cost per significant finding ($2,304) in high clinical suspicion group for vocal cord paralysis and significantly lower than the cost per significant finding ($34,535) reported in patients with chronic headache (5,6).
In our subgroup analysis based on chief complaint, there was diagnostic yield for decreased vision, diplopia, proptosis, and eye pain. No significant or relevant findings were observed in patients imaged for face pain or headache, which is consistent with other similar studies (2,7,17–19).
In our subgroup analysis based on principal neuro-ophthalmic examination findings, we observed a high diagnostic yield in patients with pyroptosis. Yet, the highest diagnostic yield was observed in patients with an RAPD and otherwise normal eye examination, emphasizing the importance of careful pupillary testing in patients with diminished vision.
In our subgroup analysis based on indication for neuroimaging, we observed high diagnostic yields (27.0%) in patients with cranial nerve palsy. This is in agreement with previous studies. Chou et al (8) found diagnostic yield of CT and MRI in acute isolated third, fourth, and sixth nerve palsies to be 14% (8). Bendszus et al (9) determined the diagnostic yield of MRI in acute isolated sixth nerve palsy to be 63% comparable with the 54.5% noted in our study. Miller et al (20) demonstrated that in patients with sixth nerve palsy, significant cost savings would have occurred if an evidence-based practice pathway had been followed. In our study, high diagnostic yields were also observed in patients with thyroid eye disease, optic neuropathy, and optic neuritis. Currently, imaging is considered the standard of care for patient with optic neuritis and as a means of detecting demyelinating disease (multiple sclerosis). Even if patients with typical optic neuritis were excluded, 47 (24.9%) of 189 remaining imaging studies would still have findings that were significant and relevant.
We recognize the limitations of our study. First, this population reflects patients referred for neuro-ophthalmic evaluation at a tertiary care center and are not representative of the general population. Second, in our analyses based on chief complaint, neuro-ophthalmic findings, and indications for imaging, some subgroups (e.g., face pain, anisocoria) had small sample size. Third, in our study we defined significance as an abnormal imaging finding that elicited changes in management. However, sometimes a normal imaging study also can elicit changes in management. For example, a normal MRI and MRV in a patient with papilledema led to the diagnosis of IIH. Another example would be patients with optic neuritis who had a normal MRI. This result has important therapeutic and prognostic implications. If these cases were also included in our definition of significance, it would further increase our diagnostic yield. Our current results may represent a conservative measurement.
In a period of rising health care costs and constrained resources, several factors contribute to decision making by clinicians and health care policy makers. Our study is an attempt to elucidate the diagnostic yield of neuroimaging based on the patient's chief complaint, neuro-ophthalmic findings, and indications for imaging. Hopefully, our results will help guide clinicians as to which patients would benefit most from evaluation that includes neuroimaging studies.
1. Hunink MG. Decision making in the face of uncertainty and resource constraints: examples from trauma imaging. Radiology. 2005;235:375–383.
2. Rao VM, Parker L, Levin DC, Sunshine J, Bushee G. Use trends and geographic variation in neuroimaging: nationwide medicare data for 1993 and 1998. AJNR Am J Neuroradiol. 2001;22:1643–1649.
3. Organization for Economic Co-operation and Development (OECD). OECD Health Data 2004: Statistics and Indicators for 30 Countries. 13th edition. Paris, France: OECD Publishing, 2004.
4. Dillon E, Sergott RC, Savino PJ, Bosley TM. Diagnostic management by gatekeepers is not cost effective for neuro-ophthalmology. Ophthalmology. 1994;101:1627–1630.
5. Liu AY, Yousem DM, Chalian AA, Langlotz CP. Economic consequences of diagnostic imaging for vocal cord paralysis. Acad Radiol. 2001;8:137–148.
6. Jordan JE, Ramirez GF, Bradley WG, Chen DY, Lighfoote JB, Song A. Economic and outcomes assessment of magnetic resonance imaging in the evaluation of headache. J Natl Med Assoc. 2000;92:573–578.
7. Harooni H, Golnik KC, Geddie B, Eggenberger E, Lee A. Diagnostic yield for neuroimaging in patients with unilateral eye or facial pain. Can J Ophthalmol. 2005;40:759–763.
8. Chou KL, Galetta SL, Liu GT, Volpe NJ, Bennett JL, Asbury AK, Balcer LJ. Acute ocular motor mononeuropathies: prospective study of the roles of neuroimaging and clinical assessment. J Neurol Sci. 2004;219:35–39.
9. Bendszus M, Beck A, Koltzenburg M, Vince GH, Brechtelsbauer D, Littan T, Urbach H, Solymosi L. MRI in isolated sixth nerve palsy. Neuroradiology. 2001;43:742–745.
10. Schultz KL, Lee AG. Diagnostic yield of the evaluation of isolated third nerve palsy in adults. Can J Ophthalmol. 2007;42:110–115.
11. Lee AG, Chau FY, Golnik KC, Kardon RH, Wall M. The diagnostic yield of the evaluation for isolated unexplained optic atrophy. Ophthalmology. 2005;112:757–759.
12. Volpe NJ, Sbarbaro JA, Gendron Livingston K, Galetta SL, Liu GT, Balcer LJ. Occult thyroid eye disease in patients with unexplained ocular misalignment identified by standardized orbital echography. Am J Ophthalmol. 2006;142:75–81.
13. Kurz-Levin MM, Landau K. A comparison of imaging techniques for diagnosing drusen of the optic nerve head. Arch Ophthalmol. 1999;117:1045–1049.
14. Centers for Medicare & Medicaid Services. PFS relative value files. Available at: http://www.cms.gov/PhysicianFeeSched/PFSRVF/list.asp?listpage=1
. Accessed October 1, 2011.
15. Hsiao WC, Braun P, Dunn DL, Becker ER, Yntema D, Verrilli DK, Stamenovic E, Chen SP. An overview of the development and refinement of the resource-based relative value scale. The foundation for reform of U.S. physician payment. Med Care. 1992;30(11 suppl):NS1–NS12.
16. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73:13–22.
17. Dumas MD, Pexman JH, Kreeft JH. Computed tomography evaluation of patients with chronic headache. CMAJ. 1994;151:1447–1452.
18. Medina LS, Kuntz KM, Pomeroy S. Children with headache suspected of having a brain tumor: a cost-effectiveness analysis of diagnostic strategies. Pediatrics. 2001;108:255–263.
19. Larson EB, Omenn GS, Lewis H. Diagnostic evaluation of headache. Impact of computerized tomography and cost-effectiveness. JAMA. 1980;243:359–362.
20. Miller RW, Lee AG, Schiffman JS, Prager TC, Garza R, Jenkins PF, Sforza P, Verm A, Kaufman D, Robinson W, Eggenberger E, Tang RA. A practice pathway for the initial diagnostic evaluation of isolated sixth cranial nerve palsy. Med Decis Making. 1999;19:42–48.