Dizziness Diagnostic Pathways: Factors Impacting Setting, Provider, and Diagnosis at Presentation

Abstract

Objective.

Ensuring that patients with dizziness present to the most appropriate level of care and provider are key goals of quality and cost improvement efforts. Using a symptom-defined cohort of adults presenting for dizziness evaluations, we aimed to identify patient factors associated with ambulatory clinic vs emergency department (ED) presentations, evaluating provider specialty, and assigned diagnoses.

Study Design.

Cross-sectional study.

Setting.

OptumLabs Data Warehouse (OLDW), a longitudinal, real-world data asset with deidentified administrative claims.

Methods.

We performed a cross-sectional analysis of adults (older than 18 years) who received new dizziness diagnoses (2006–2015) and identified factors associated with setting and provider at initial presentation using multivariable regression models.

Results.

Of 805,454 individuals with dizziness (median age 52 years, 62% women, 29% black, Asian, or Hispanic), 23% presented to EDs and 77% to clinics (76% primary care, 7% otolaryngology, 5% cardiology, 3% neurology). Predictors of ED presentation were younger age, male sex, black race, lower education, and medical comorbidity. Predictors of primary care clinic presentation were older age and race/ethnicity other than white. Nonetiologic symptom diagnoses alone were assigned to 51% and were most associated with age older than 75 years (odds ratio, 2.90; 95% CI, 2.86–2.94).

Conclusion.

Adults with dizziness often present to a level of care that may be higher than is optimal. Differential care seeking and diagnoses by age, sex, and race/ethnicity reflect influences beyond dizziness presentation acuity. Targeted patient resources, triage systems, provider education, and cross-specialty partnerships are needed to direct dizzy patients to appropriate settings and providers to improve care.

Dizziness and vertigo rank among the most common symptoms for which adults seek medical evaluation.¹ An estimated 20.6 million Americans sought care for dizziness and vertigo in ambulatory clinics from 2013 to 2015.² The number of annual emergency department (ED) visits for dizziness and vertigo in the United States reached 3.9 million in 2011 and is growing.^3,4 There is rising concern that health care visits for dizziness generate inordinate health service utilization and costs. For example, the cost of ED dizziness presentations is estimated to be more than $10 billion per year,³ a value attributed to high rates of neuroimaging, cardiac testing, lab work, and resulting hospital admissions.^4,5 There is also substantial geographic, practice pattern, and sociodemographic-level variation in diagnostic services utilization for dizziness that cannot be explained by differences in medical need.^6–9 This unwarranted variation evidences uneven care quality, cost-inefficiency, and potential over- or underuse of services.^10,11

An emerging focal point of efforts to right-size costs and health services utilization for dizziness and vertigo is the site of initial clinical presentation. A recent systematic review concluded, in addition to neuroimaging, increased utilization and direct costs attributable to vertigo arise from overuse of emergency care and repeated, poorly targeted clinical consultations when, in either instance, primary care evaluation would have been sufficient.¹² Thus far, examinations of health services utilization patterns for dizziness and vertigo in the United States have not provided a sufficiently broad picture, as they were limited to single-care settings,^2,13,14 specific age groups, or payers (fee-for-service Medicare)⁶ or were based on patient-reported surveys.¹⁵ In order to direct patients toward optimal care, we need to determine what differentiates patients who seek care in EDs from those who seek care in specialty or primary care clinics, as this choice affects future diagnoses and therapeutic care utilization.

In this study, we conducted a detailed analysis of a large cohort of US adults presenting with dizziness in order to identify factors that were (1) predictive of the setting of first presentation, whether in ambulatory clinics vs EDs; (2) associated with the medical specialty of the first evaluating provider; and (3) associated with the diagnoses assigned at presentation. We used the OptumLabs Data Warehouse, a longitudinal, real-world data asset with deidentified administrative claims.¹⁶ Claims were not limited to a single site, health system, or specialty, granting an overview of care initiation across US health care settings.

Materials and Methods

Data Source

This study used deidentified administrative claims data with linked socioeconomic status information from the OptumLabs Data Warehouse (OLDW), which includes medical and pharmacy claims, laboratory results, and enrollment records for commercial and Medicare Advantage (MA) enrollees. The database contains longitudinal health information on enrollees and patients, representing a diverse mixture of ages, ethnicities, and geographical regions across the United States. Since this study involved analysis of preexisting, deidentified data,¹⁶ it was exempt from review by the University of Minnesota Institutional Review Board.

Population

We selected all individuals aged 18 years or older who were assigned a new International Classification of Diseases, Ninth Revision (ICD-9) diagnosis of dizziness and giddiness (780.4) or a specific vestibular disorder (Table 1) on a provider visit claim (evaluation and management code) in an ambulatory clinic (hereafter “clinic”) or ED between 2006 and 2015. To select new presentations, individuals had at least 365 days of continuous enrollment and no dizziness or vestibular diagnoses prior to the index date (provider encounter date) and, for future claims capture, at least 365 days of continuous enrollment thereafter.

Table 1.

Diagnostic Codes for Dizziness and Underlying Conditions of Interest.

Diagnostic category	Diagnostic subcategories	ICD-9-CM code
Dizziness (symptom)	Dizziness and giddiness	780.4
Vestibular disorders	Benign paroxysmal positional vertigo	386.11
	Vestibular neuritis, labyrinthitis	386.30–386.35, 386.12
	Ménière’s disease	386.00–386.04
	Other peripheral vertigo and labyrinthine disorders	386.1, 386.4, 386.5, 386.8, 386.9
	Central vertigo	386.2
Cardiovascular disorders	Orthostatic hypotension	458.00
	Vasovagal syncope	780.2
	Cardiac dysrhythmia	427.0–427.9
Migraine	Migraine	346.0–346.9
Serious neurologic disorders	Cerebrovascular disease (intracranial hemorrhage, arterial occlusion or dissection, transient ischemia)	430–437, 443.21, 443.24, 438.85
	Tumors (vestibular schwannoma, meningioma, tumors of the cerebellum or brainstem)	225.1, 225.2, 191.6, 191.7
	Other neurologic  Multiple sclerosis, cerebellar ataxias, epilepsy	340, 334.0–334.9, 345.0–345.91
Otologic (nonvestibular) disorders	Hearing loss	387–389
	Diseases of the outer ear, middle ear, and mastoid	380–385

Abbreviation: ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification.

Analyses

The cohort was characterized by age, sex, race/ethnicity, education, insurance product (commercial, MA), and geographic census region. Existing claims data 1 year prior to index date were used to assemble the Charlson Comorbidity Index (CCI) of 19 weighted condition categories.^17–19 The index date visit was characterized by clinic or ED site of service, provider specialty, and any concurrent assignment of prespecified cardiovascular, neurologic, or otologic diseases associated with dizziness (Table 1).^13,20–22 When more than 1 specific diagnosis was assigned on the index date, we used the first (primary) diagnosis.

Factors associated with service site and provider specialty were identified with multivariable regression models that controlled for patient sociodemographic and clinical factors. Analyses were performed with SAS 9.4 (SAS Institute).

Results

Patient Characteristics

We identified 805,454 unique individuals who presented for new dizziness evaluations from 2006 through 2015 (Table 2). The median patient age was 52 (15.9) years. Most patients were under age 65 years (73%) and women (62%), and 50% had at least 1 major comorbidity based on CCI. Patients were from all US census regions; 29% were identified as black, Asian, or Hispanic; and 79% had commercial insurance.

Table 2.

Characteristics of Adult Patients Presenting for Initial Dizziness Evaluations, by Setting.

	Setting
Characteristic	All settings (N = 805,454), No. (%)	ED (n = 185,338; 23%), No. (%)	OP (n = 620,116; 77%), No. (%)	P value (ED vs OP)
Age, y				≤.001
18–39	180,795 (22)	43,082 (23)	137,713 (77)
40–49	157,038 (20)	33,508 (21)	123,530 (79)
50–59	175,021 (22)	37,925 (23)	137,096 (77)
60–64	72,111 (9)	16,012 (22)	56,099 (78)
65–74	57,360 (7)	13,631 (24)	43,729 (76)
75+	163,129 (20)	41,180 (25)	121,949 (75)
Sex				≤.001
Female	502,055 (62)	108,297 (22)	393,758 (78)
Male	303,399 (38)	77,041 (25)	226,358 (75)
Race/ethnicity				≤.001
Asian	32,715 (4)	5235 (16)	27,480 (84)
Black	115,030 (14)	34,098 (30)	80,932 (70)
Hispanic	87,848 (11)	19,847 (23)	68,001 (77)
White	452,494 (56)	99,039 (22)	353,455 (78)
Other or unknown	117,367 (15)	27,119 (23)	90,248 (77)
Charlson Comorbidity Indexa				≤.001
0	402,074 (50)	79,211 (20)	322,863 (80)
1	181,334 (23)	43,585 (24)	137,749 (76)
2+	222,046 (28)	62,542 (28)	159,504 (72)
Insurance type				≤.001
Commercial	634,707 (79)	139,147 (22)	495,560 (78)
Medicare Advantage	170,747 (21)	46,191 (27)	124,556 (73)
Education level				≤.001
<12th grade	8876 (1)	2270 (26)	6606 (74)
High school	252,685 (31)	64,460 (26)	188,225 (74)
Less than bachelor’s degree	400,954 (50)	89,781 (22)	311,173 (78)
Bachelor degree or higher	131,882 (16)	26,203 (20)	105,679 (80)
Unknown	11,057 (1)	2624 (24)	8433 (76)
Census regionb				≤.001
Midwest	100,631 (12)	22,440 (22)	78,191 (78)
Northeast	222,651 (28)	54,904 (25)	167,747 (75)
South	365,085 (45)	84,553 (23)	280,532 (77)
West	116,703 (14)	23,347 (20)	93,356 (80)

Abbreviations: ED, emergency department; OP, outpatient.

^aWeighted conditions include myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic obstructive pulmonary disease, paralysis, diabetes with complications, chronic renal failure, liver disease, rheumatoid arthritis, and acquired immunodeficiency syndrome.

^bLess than 1% of cases were of unknown region.

Presentation Setting

Nearly one-fourth of patients with dizziness first presented to EDs (23%) rather than clinics (77%). For each patient factor considered (Table 2), the proportion of patients presenting to EDs was highest among those identified as black (30%), men (25%), over age 75 years (25%), having higher levels of comorbid illness (28%), Northeasterners (25%), MA enrollees (27%), and high school educated or less (26%). The proportion of patients presenting to EDs was lowest among those identified as Asian (16%), ages 40 to 49 years (21%), without comorbidities (20%), Westerners (20%), and holding at least a bachelor’s degree (20%).

In the multivariable model controlling for patient factors (Table 3), care was more likely to be sought in clinics rather than EDs by patients identified as women compared to men (adjusted odds ratio [aOR] [95% CI], 1.24 [1.23–1.26]), age over 40 years compared to younger (1.22 [1.20–1.24] to 1.33 [1.30–1.35]), Asian compared to white (1.43 [1.39–1.48]), and having higher education compared to not (1.12 [1.10–1.13] to 1.22 [1.20–1.24]). Care was less likely to be sought in clinics (ie, more likely in EDs) by black patients compared to white (0.69 [0.68–0.70]) and those with comorbidity compared to without (0.76 [0.75–0.77] to 0.61 [0.60–0.62]).

Table 3.

Factors Associated With Presentation of Dizziness in an Outpatient Clinic Setting Compared to the Emergency Department.^a

Factors	Odds ratio (95% CI)
Age, y
18–39	Reference
40–49	1.22 (1.20–1.24)
50–59	1.28 (1.26–1.30)
60–64	1.33 (1.30–1.35)
65–74	1.29 (1.26–1.32)
75+	1.25 (1.22–1.27)
Sex
Male	Reference
Female	1.24 (1.23–1.26)
Race
White	Reference
Black	0.69 (0.68–0.70)
Asian	1.43 (1.39–1.48)
Hispanic	0.99 (0.98–1.01)
Other/unknown	0.92 (0.91–0.94)
Charlson Comorbidity Index
0	Reference
1	0.76 (0.75–0.77)
2+	0.61 (0.60–0.62)
Education level
<12th grade	Reference
High school	0.99 (0.94–1.04)
Less than bachelor’s degree	1.12 (1.10–1.13)
Bachelor degree or higher	1.22 (1.20–1.24)
Unknown	1.06 (1.02–1.11)

^aBold values indicate significance of P < .05.

First Providers Seen in Outpatient Clinics

Most patients with dizziness presenting to clinics saw a primary care clinician (76%, n = 470,271). Those who first saw a specialist most often saw an otolaryngologist (7%, n = 45,468), cardiologist (5%, n = 32,929), or neurologist (3%, n = 19,023). In multivariable analyses (Table 4), race/ethnicity, age, and comorbidity were most strongly associated with provider specialty. Compared to white individuals, black, Asian, and Hispanic individuals were less likely to see otolaryngologists (0.61 [0.58–0.65]), 0.79 [0.77–0.82], 0.84 [0.81–0.86], respectively) and more likely to see primary care (1.35 [1.30–1.40]), 1.11 [1.09–1.13]), 1.10 [1.07–1.12], respectively). Compared to younger patients, those over 40 years were more likely to present to otolaryngologists and cardiologists and less likely to present to primary care or neurologists. For each decade after 40, patients were increasingly more likely to see cardiology (1.09 [1.05–1.14] to 1.65 [1.58–1.72]) and less likely to see neurology (0.88 [0.85–0.92] to 0.50 [0.47–0.52]). The likelihood of seeing otolaryngology peaked at ages 60 to 64 years (1.72 [1.65–1.78]). Higher levels of comorbidity were strongly associated with seeing cardiology (2.13 [2.07–2.20]) and neurology (2.30 [2.21–2.39]), negatively associated with primary care (0.62 [0.61–0.64]), but not meaningfully associated with otolaryngology (0.96 [0.93–0.98]).

Table 4.

Factors Associated With Provider Specialty at First Evaluation of Dizziness in the Outpatient Clinic Setting.^a

Patient factor	Primary care, OR (95% CI)	Otolaryngology, OR (95% CI)	Cardiology, OR (95% CI)	Neurology, OR (95% CI)
Age, y
18–39	Reference	Reference	Reference	Reference
40–49	0.84 (0.82–0.86)	1.41 (1.37–1.45)	1.09 (1.05–1.14)	0.88 (0.85–0.92)
50–59	0.74 (0.73–0.76)	1.65 (1.60–1.70)	1.29 (1.24–1.35)	0.80 (0.77–0.84)
60–64	0.70 (0.68–0.72)	1.72 (1.65–1.78)	1.52 (1.45–1.59)	0.75 (0.71–0.79)
65–74	0.80 (0.77–0.82)	1.41 (1.35–1.47)	1.54 (1.47–1.62)	0.62 (0.58–0.66)
75+	0.90 (0.87–0.92)	1.11 (1.07–1.15)	1.65 (1.58–1.72)	0.50 (0.47–0.52)
Sex
Male	Reference	Reference	Reference	Reference
Female	1.09 (1.07–1.11)	1.13 (1.11–1.16)	0.65 (0.64–0.67)	1.10 (1.06–1.13)
Race
White	Reference	Reference	Reference	Reference
Black	1.35 (1.30–1.40)	0.61 (0.58–0.65)	0.92 (0.87–0.97)	0.92 (0.86–0.99)
Asian	1.11 (1.09–1.13)	0.79 (0.77–0.82)	1.04 (1.01–1.08)	0.98 (0.93–1.02)
Hispanic	1.10 (1.07–1.12)	0.84 (0.81–0.86)	0.97 (0.93–1.01)	1.05 (1.03–1.10)
Other or unknown	1.00 (0.98–1.02)	0.99 (0.96–1.01)	0.99 (0.95–1.02)	1.05 (1.02–1.09)
Education level
<12th grade	Reference	Reference	Reference	Reference
High school	1.18 (1.09–1.27)	0.84 (0.74–0.94)	0.90 (0.80–1.01)	0.83 (0.71–0.97)
Less than bachelor’s degree	0.87 (0.85–0.88)	1.27 (1.24–1.30)	1.01 (0.98–1.03)	1.10 (1.06–1.14)
Bachelor’s degree or higher	0.66 (0.65–0.67)	1.81 (1.76–1.86)	1.11 (1.07–1.15)	1.31 (1.25–1.37)
Unknown	0.81 (0.76–0.86)	1.32 (1.21–1.43)	1.04 (0.94–1.15)	1.31 (1.16–1.47)
CCI
0	Reference	Reference	Reference	Reference
1	0.79 (0.78–0.81)	1.01 (0.98–1.03)	1.43 (1.38–1.47)	1.65 (1.59–1.71)
2+	0.62 (0.61–0.64)	0.96 (0.93–0.98)	2.13 (2.07–2.20)	2.30 (2.21–2.39)

Abbreviations: CCI, Charlson Comorbidity Index; OR, odds ratio.

^aBold values indicate significance of P < .05.

Diagnostic Assignments

At the initial encounter, providers assigned most patients in our cohort (84%) the symptom-only diagnosis code for dizziness (93% of ED presentations vs 81% of clinic presentations, P < .001). Diagnosis rates of specific vestibular disorders, given to 16% overall (7% EDs, 19% clinics, P < .001), were 2 to 10 times greater in clinics than EDs (Table 5). Providers concurrently assigned half (49%) of the patients with dizziness (ICD-9, 780.4) at least 1 possible etiologic diagnosis, most commonly migraine (37%), nonvestibular otologic (6%), and cardiovascular disorders (ie, arrhythmia, orthostatic hypotension, or vasovagal syncope) (5%). Cardiovascular diagnosis rates were 3 times higher in EDs than in clinics (11% vs 4%) (P < .001). Concurrent diagnoses of neurologic conditions were uncommon in both settings (0.8% overall). In clinics, the proportions of isolated nonspecific dizziness symptom diagnoses by primary care (54%, 254,340 of 470,271) and neurologists (46%, 8660 of 19,023) were 1.5 to 3 times higher than by cardiologists (31%, 10,075 or 32,929) and otolaryngologists (16%, 7032 of 45,468) (P < .001).

Table 5.

Diagnoses Assigned on Date of Initial Dizziness Presentation, by Setting.^a

Diagnostic categories	All sites, No. (%)	ED, No. (%)	OC, No. (%)	ED vs OC, P value
Cohort inclusion diagnosis				≤.001
Dizziness (symptom only)	673,140 (84)	171,549 (93)	501,591 (81)
Vestibular disorder	132,314 (16)	13,789 (7)	118,525 (19)
BPPV	54,777 (7)	6025 (3)	48,752 (8)
Vestibular neuritis, labyrinthitis	36,668 (5)	4622 (2)	32,046 (5)
Other peripheral vertigo, labyrinthine disorders	25,265 (3)	2517 (1)	22,748 (4)
Ménière’s disease	9288 (1)	456 (0.3)	8832 (1)
Central vertigo	6316 (0.8)	169 (0.1)	6147 (1)
Concurrent diagnoses
None (only dizziness)	407,548 (51)	97,412 (53)	310,136 (50)
Migraine	297,949 (37)	61,452 (33)	236,497 (38)
Otologic disorders, nonvestibular	47,776 (6)	3271 (2)	44,505 (7)
Cardiovascular conditions	44,076 (5)	20,585 (11)	23,491 (4)
Orthostatic hypotension, vasovagal syncope	23,252 (3)	12,750 (7)	10,502 (2)
Cardiac dysrhythmia	22,144 (3)	8311 (4)	13,833 (2)
Serious neurologic conditions	6394 (0.8)	2079 (1)	4315 (0.7)
Cerebrovascular disease	5556 (0.7)	1810 (1)	3746 (0.6)
Tumors	226 (0.03)	64 (0.03)	162 (0.03)
Other neurologic disease	1003 (0.1)	268 (0.1)	735 (0.1)
Multiple specific diagnoses	126,182 (16)	13,547 (7)	112,635 (18)

Abbreviations: BPPV, benign paroxysmal positional vertigo; ED, emergency department; OC, outpatient clinic.

^aPatients may have received more than 1 diagnosis; therefore, totals exceed 100% of the sample size.

In a multivariable model (Table 6), the odds of receiving a nonspecific dizziness symptom diagnosis alone (without any potential etiologic diagnosis from Table 1) increased each decade over 40 (1.52 [1.50–1.54] in the fifth decade to 2.90 [2.86–2.94] over 75) and were higher for ED presentations compared to clinics (1.14 [1.13–1.15)).

Table 6.

Factors Associated With Nonspecific Dizziness Diagnostic Assignment at the First Provider Encounter for Dizziness.^a

Factor	Dizziness (symptom only)
	OR	(95% CI)
Age, y
18–39	Reference
40–49	1.52	(1.50–1.54)
50–59	1.99	(1.96–2.02)
60–64	2.37	(2.32–2.41)
65–74	2.65	(2.60–2.71)
75+	2.90	(2.86–2.94)
Sex
Male	Reference
Female	0.88	(0.87–0.89)
Race
White	Reference
Black	1.08	(1.07–1.10)
Asian	0.83	(0.81–0.85)
Hispanic	0.93	(0.91–0.94)
Other/unknown	0.98	(0.96–0.99)
Education level
<12th grade	Reference
High school	0.94	(0.89–0.98)
Less than bachelor’s degree	0.95	(0.93–0.96)
Bachelor’s degree or higher	0.90	(0.88–0.91)
Unknown	0.99	(0.95–1.03)
Site
OP	Reference
ED	1.14	(1.13–1.15)

Abbreviations: ED, emergency department; OP, outpatient; OR, adjusted odds ratio.

^aBold values indicate P < .05.

Discussion

Among a diverse cohort of 805,454 insured US adults with dizziness, nearly 1 in 4 patients (23%) presented to an ED rather than an outpatient clinic for initial evaluation. Among those (77%) who initiated care in clinics, most (76%) were evaluated by primary care providers rather than specialists. In multivariable analyses, both the setting and provider selected for the initial dizziness presentation were significantly influenced by patient characteristics, including age, sex, race/ethnicity, education, and comorbidity. The diagnosis assigned to a patient on the day of presentation was most often nonspecific dizziness, with only 16% of dizzy adults receiving a diagnosis of a vestibular disorder. Diagnostic assignments were associated with patient age, presentation site, and provider specialty. In particular, patients with dizziness sought more specialty care with age but also received less specific diagnoses.

This cohort of patients with dizziness provides a uniquely broad view of initial care-seeking patterns among insured US health care consumers of all ages. In contrast to prior work, analyzed claims were not limited by setting, health system, provider specialties, or patient recall. Furthermore, as symptoms, rather than diagnoses, drive people to seek health care and set diagnostic pathways in motion,²³ we employed a symptom-based approach to cohort selection. This approach captured a range of dizziness presentations, yielding observations more generalizable to diagnostic process improvement than would purely disease-based constructs.²³ Our cohort’s characteristics, including the predominance of women (62%), are consistent with prior analyses in clinic or ED settings.^2,13,24,25 Notably, 73% of dizzy patients were under age 65 years. Suitable emphasis is placed on optimizing dizziness care among the aging population, particularly in relation to fall prevention.^15,26 The large proportion of younger patients suggests diagnostic care improvements will also benefit working-age individuals who have dizziness-related financial^27,28 and care-taking burdens in addition to falls and morbidity.²⁹

Ensuring that patients of all ages present at the most appropriate level of care is one goal of efforts to improve health care quality and control costs. Dizziness is a potentially preventable reason for ED visits,^12,30 as it can often be effectively evaluated and managed in ambulatory clinics. ED visits for dizziness are associated with high rates of (often unwarranted) health service utilization, including neuroimaging, estimated at 40% per ED visit compared to 5.5% per clinic visit.^2,4,5 Imaging acquisition is not associated with an increased likelihood of identifying the cause of dizziness.¹⁵ Beyond financial costs, excess ED utilization precludes establishment of continuity of care and shifts resources away from other patients.³¹ Certainly, some ED visits for dizziness are appropriate, as at least 15% of US ED dizziness presentations have underlying life-threatening conditions.¹³ Accordingly, we observed that patients with higher comorbidity scores were 1.24 to 1.39 times more likely to present to EDs than their healthier peers. However, our observed rate of ED use (23%) is likely to be excessive and warrants investigation of potential underlying factors. Among those who presented to clinics, we observed rates of primary care evaluation at first presentation (76%) were higher and of specialist presentations were lower (otolaryngology [7%], cardiology [5%]) than a National Ambulatory Medical Care Survey analysis that found 52% of all dizziness visits (not restricted to new evaluations) were to primary care, followed by otolaryngologists (13.3%) and neurologists (9.6%).² Thus, our findings also emphasize the substantial opportunity to improve dizziness care from the outset by engaging at the primary care level.

Patients may choose a setting in which to seek care for dizziness based on their own perceptions of urgency, influenced by the timing and severity of symptoms, but are also influenced by considerations of access and quality of care, provider referrals, and costs in relation to alternative sites.^31,32 Our multivariable analyses revealed sociodemographic phenotypes associated with care-seeking patterns for dizziness. Individuals over 40 years were more likely to seek care in clinics and from specialists, even though they are, in theory, at greater risk of stroke or other life-threatening etiologies. This finding may reflect how younger individuals are less likely to have a primary provider, using the ED instead for health care.³³ Race/ethnicity significantly influenced presentation setting and provider. Controlling for comorbidity, compared to white individuals, black individuals were 1.33 times more likely to present in EDs, while Asian individuals were 1.43 times more likely to present to clinics. Black, Asian, and Hispanic individuals were also significantly more likely to see primary care and less likely to see otolaryngologists for dizziness. Consistent with our findings, a Medical Expenditure Panel Survey analysis (2007–2015) found, compared to non-Hispanic whites, non-Hispanic African Americans were more likely to use EDs for otolaryngologic conditions in general, and non-Hispanic African Americans and Hispanics were less likely to use otolaryngology clinic services.³⁴ We also previously observed that black Medicare beneficiaries with dizziness were less likely than whites to receive vestibular testing.⁶ Without known medical justification, racial and ethnic differences in ED vs clinic utilization and specialty care services may result from systemic racism, barriers to establishing care, and differential care provision and trust in health care systems.^34,35 Investigation of the causes and consequences of observed disparities within specific communities is needed to promote equitable care provision.

The observed sociodemographic patterns can inform strategies to direct patients with dizziness to the most appropriate level of care. In the most affected communities, potentially preventable ED visits could be curbed by improvements in access and quality of ambulatory and, particularly, primary care.³⁶ Primary care providers may take a broad view, avoiding pitfalls of subspecialty diagnostic silos. However, the range of potential etiologies is challenging, prompting high rates of specialist referral.² Improved provider education in dizziness evaluation and cross-specialty partnerships would improve ambulatory care quality. With dizziness, patients and physicians likely have different perspectives on what constitutes an emergency, something ambulatory care improvements will not change.³⁷ Rather, the triage systems becoming widely available through health system and payer initiatives, including those employing artificial intelligence,³⁸ could assist in screening and referral without compromising outcomes.³⁷ Directed patient education materials also facilitate care navigation.³⁷ As there are 550,000 monthly Google searches in the United States for the words dizziness, dizzy, and vertigo, welldeveloped online resources are likely to be accessed.³⁹ Community education and outreach should also be tailored to populations we identified as most likely to present to EDs.

Our analyses of diagnoses on initial provider claims also revealed patterns that may inform quality improvement efforts. The cause of dizziness may go undiagnosed at presentation, pending further testing or symptom evolution.⁴⁰ Thus, diagnostic assignments at care initiation typically drive decisions regarding future tests and consults as providers seek to refine diagnoses and initiate effective therapies. Consistent with studies of US ED or clinic dizziness presentations, the most common diagnostic groupings in our cohort were otologic/vestibular, followed by cardiovascular, with a lower proportion due to cerebrovascular disease,^4,13,24 and vestibular disorders were more commonly diagnosed in clinics and by otolaryngologists.² Strikingly, we observed the likelihood of receiving a nonspecific dizziness diagnosis alone increased markedly with age (odds ratio, 2.90 over age 75), even as the use of ambulatory and specialty care increased. In addition to the etiologies we selected, medication side effects cause dizziness among older adults in 11% of cases¹⁵ but are difficult to capture in claims data. However, this is unlikely to be the sole explanation for the magnitude of the age effect. Providers may have greater diagnostic uncertainty among older individuals, and this clinical ambiguity raises concern for diagnostic error, such as a missed stroke⁴¹ and delays in care. Nonspecific diagnoses were also associated with ED and primary care² presentations. Targeted frontline clinician education, geared toward older patient populations, is essential to ensure we are providing optimal care and reducing dizziness-related morbidity. Future analyses will seek evidence of factors predictive of successful diagnostic refinement following presentation.

Our study has several limitations. Claims data cannot reveal symptom character⁴² or duration, nor could diagnoses be adjudicated for accuracy or the appropriateness of evaluation or treatment. We could not capture causes of dizziness lacking ICD-9 codes. International Classification of Diseases, 10th Revision (ICD-10) was not in use during the study period. While not subject to recall bias that limits surveybased work, claims do not capture chief complaints. Clinicians may differ in their coding practices and may not code dizziness for all symptomatic individuals. Thus, our cohort selection process had greater specificity than sensitivity for dizziness, excluding patients who received nonvestibular etiologic diagnoses alone. We anticipate this affected all settings and sociodemographic categories. The observed relative proportions of symptom and vestibular diagnoses are similar to those in other studies using our inclusion criteria.^2,6 The population represents US adults with commercial or managed care Medicare plans, and findings may not be applicable to enrollees of commercial plans not included in OLDW,⁴³ fee-for-service (FFS) Medicare, Medicaid, or the uninsured. Commercial beneficiaries in OLDW have been shown to be similar in age, race/ethnicity, and sex to the broader US commercially insured population.⁴⁴ Compared to MA enrollees, FFS Medicare beneficiaries exhibit greater utilization of potentially inappropriate health care resources and have a higher probability of inappropriate ED visits.⁴⁵ Their absence may have decreased the number and odds of ED visits among older individuals. When comparing Medicaid and commercial enrollees, no consistent association was found between insurance type and low-value care,⁴⁶ but Medicaid beneficiaries were more likely to seek ED care for primary-care treatable conditions.⁴⁷ Associations between insurance type and dizziness presentation approaches could be considered for future research. Despite these limitations, we believe our results emphasize important characteristics of dizziness diagnostic pathways in both ED and clinic settings that deserve more attention.

Conclusion

Ensuring that patients present to the most appropriate setting and provider is a key goal of efforts to improve health care quality and control costs. In a diverse patient cohort, we observed how patient sociodemographics, settings, and providers interact to form initial patterns of care. Targeted patient resources, triage, provider education, and cross-specialty partnerships may improve care quality for dizziness. Future work should further characterize care pathways to define what level of testing and consultation add sufficient value to justify costs for specific patient populations.