Headache Neuroimaging

A Survey of Current Practice, Barriers, and Facilitators to Optimal Use

Evan L. Reynolds PhD; James F. Burke MD, MS; Lacey Evans MPH; Faiz I. Syed MD, MS; Eric Liao MD; Remy Lobo MD; Wade Cooper DO; Larry Charleston IV MD, MSc; Brian C. Callaghan MD, MS

Disclosures

Headache. 2022;62(1):36-56.

Abstract and Introduction

Abstract

Objective: The objective of this study was to understand current practice, clinician understanding, attitudes, barriers, and facilitators to optimal headache neuroimaging practices.

Background: Headaches are common in adults, and neuroimaging for these patients is common, costly, and increasing. Although guidelines recommend against routine headache neuroimaging in low-risk scenarios, guideline-discordant neuroimaging is still frequently performed.

Methods: We administered a 60-item survey to headache clinicians at the Veterans Affairs health system to assess clinician understanding and attitudes on headache neuroimaging and to determine neuroimaging practice patterns for three scenarios describing hypothetical patients with headaches. Descriptive statistics were used to summarize responses, stratified by clinician type (physicians or advanced practice clinicians [APCs]) and specialty (neurology or primary care).

Results:The survey was successfully completed by 431 of 1426 clinicians (30.2% response rate). Overall, 317 of 429 (73.9%) believed neuroimaging was overused for patients with headaches. However, clinicians would utilize neuroimaging a mean (SD) 30.9% (31.7) of the time in a low-risk scenario without red flags, and a mean 67.1% (31.9) of the time in the presence of minor red flags. Clinicians had stronger beliefs in the potential benefits (268/429, 62.5%) of neuroimaging compared to harms (181/429, 42.2%) and more clinicians were bothered by harms stemming from the omission of neuroimaging (377/426, 88.5%) compared to commission (329/424, 77.6%). Additionally, APCs utilized neuroimaging more frequently than physicians and were more receptive to potential interventions to improve neuroimaging utilization.

Abstract and Introduction
Methods
Results
Discussion
Conclusion
References

Table 1. Clinician opinions regarding harms and benefits of MRI, stratified by clinician type and specialty
Survey question	Overall (N = 431)	Advanced practice clinicians (n = 103)	Physicians (n = 326)	Wilcoxon rank sum test p value: APCs vs. physicians	Neurologists (n = 177)	Primary care clinicians (n = 140)	Wilcoxon rank sum test p value: neurologists vs. primary care clinicians
In patients with headaches and a normal neurologic examination, how often does an MRI identify …
Any abnormality at all
<1%	64 (14.9%)	19 (18.5%)	45 (13.9%)	0.039	14 (7.9%)	30 (21.4%)	<0.001
1%–5%	116 (27.0%)	38 (36.9%)	78 (24.0%)		38 (21.5%)	38 (27.1%)
6%–10%	78 (18.2%)	15 (14.6%)	63 (19.4%)		32 (18.1%)	30 (21.4%)
11%–30%	86 (20.1%)	10 (9.7%)	76 (23.4%)		53 (29.9%)	20 (14.3%)
31%–50%	33 (7.7%)	6 (5.8%)	27 (8.3%)		17 (9.6%)	10 (7.1%)
51%–70%	19 (4.4%)	3 (2.9%)	15 (4.6%)		10 (5.7%)	4 (2.9%)
71%–89%	16 (3.7%)	4 (3.9%)	12 (3.7%)		7 (4.0%)	5 (3.6%)
90%–94%	9 (2.1%)	4 (3.9%)	5 (1.5%)		4 (2.3%)	1 (0.7%)
95%–99%	4 (0.9%)	2 (1.9%)	2 (0.6%)		1 (0.6%)	1 (0.7%)
>99%	4 (0.9%)	2 (1.9%)	2 (0.6%)		1 (0.6%)	1 (0.7%)
Missing	2	0	1		0	0
Brain tumor
<1%	346 (82.0%)	73 (71.6%)	272 (85.3%)	0.001	152 (87.4%)	113 (82.5%)	0.196
1%–5%	57 (13.5%)	19 (18.6%)	38 (11.9%)		20 (11.5%)	18 (13.1%)
6%–10%	2 (0.5%)	1 (1.0%)	1 (0.3%)		1 (0.6%)	0 (0.0%)
11%–30%	0 (0.0%)	0 (0.0%)	0 (0.0%)		0 (0.0%)	0 (0.0%)
31%–50%	2 (0.5%)	1 (1.0%)	1 (0.3%)		0 (0.0%)	1 (0.7%)
51%–70%	1 (0.2%)	0 (0.0%)	1 (0.3%)		0 (0.0%)	0 (0.0%)
71%–89%	3 (0.7%)	1 (1.0%)	2 (0.6%)		1 (0.6%)	1 (0.7%)
90%–94%	3 (0.7%)	2 (2.0%)	1 (0.3%)		0 (0.0%)	1 (0.7%)
95%–99%	5 (1.2%)	3 (2.9%)	2 (0.6%)		0 (0.0%)	2 (1.5%)
>99%	3 (0.7%)	2 (2.0%)	1 (0.3%)		0 (0.0%)	1 (0.7%)
Missing	9	1	7		3	3
An abnormality that changes management of the patient
<1%	168 (39.6%)	29 (28.4%)	139 (43.3%)	0.009	65 (37.1%)	70 (51.1%)	0.027
1%–5%	170 (40.1%)	49 (48.0%)	120 (37.4%)		74 (42.3%)	44 (32.1%)
6%–10%	55 (13.0%)	10 (9.8%)	45 (14.0%)		25 (14.3%)	18 (13.1%)
11%–30%	14 (3.3%)	3 (2.9%)	11 (3.4%)		10 (5.7%)	1 (0.7%)
31%–50%	7 (1.7%)	7 (6.9%)	0 (0.0%)		0 (0.0%)	0 (0.0%)
51%–70%	5 (1.2%)	1 (1.0%)	4 (1.3%)		1 (0.6%)	2 (1.5%)
71%–89%	1 (0.2%)	1 (1.0%)	0 (0.0%)		0 (0.0%)	0 (0.0%)
90%–94%	1 (0.2%)	1 (1.0%)	0 (0.0%)		0 (0.0%)	0 (0.0%)
95%–99%	3 (0.7%)	1 (1.0%)	2 (0.6%)		0 (0.0%)	2 (1.5%)
>99%	0 (0.0%)	0 (0.0%)	0 (0.0%)		0 (0.0%)	0 (0.0%)
Missing	7	1	5		2	3
How often do you believe headache patients …
Are harmed by MRI due to subsequent decisions based on MRI results?
Never	31 (7.2%)	11 (10.8%)	20 (6.1%)	0.204	10 (5.7%)	7 (5.0%)	0.004
Rarely	217 (50.6%)	52 (51.0%)	164 (50.3%)		101 (57.1%)	60 (42.9%)
Occasionally	149 (34.7%)	32 (31.4%)	117 (35.9%)		59 (33.3%)	56 (40.0%)
Often	31 (7.2%)	7 (6.9%)	24 (7.4%)		7 (4.0%)	16 (11.4%)
Almost always	1 (0.2%)	0 (0.0%)	1 (0.3%)		0 (0.0%)	1 (0.7%)
Missing	2	1	0		0	0
Benefit from MRI due to improved management decisions based on MRI results?
Never	8 (1.9%)	3 (2.9%)	5 (1.5%)	<0.001	2 (1.1%)	3 (2.1%)	0.816
Rarely	153 (35.7%)	18 (17.7%)	134 (41.1%)		74 (41.8%)	59 (42.1%)
Occasionally	179 (41.7%)	49 (48.0%)	130 (39.9%)		77 (43.5%)	51 (36.4%)
Often	74 (17.3%)	28 (27.5%)	46 (14.1%)		18 (10.2%)	23 (16.4%)
Almost always	15 (3.5%)	4 (3.9%)	11 (3.4%)		6 (3.4%)	4 (2.9%)
Missing	2	1	0		0	0
Among patients I have cared for receiving MRI for headache, how often have the following harms occurred?
Difficulty tolerating the MRI scan itself
Never	29 (6.8%)	18 (17.5%)	11 (3.4%)	0.008	3 (1.7%)	7 (5.0%)	0.105
Rarely	137 (32.1%)	31 (30.1%)	105 (32.5%)		50 (28.7%)	51 (36.4%)
Occasionally	216 (50.6%)	44 (42.7%)	172 (53.3%)		104 (59.8%)	66 (47.1%)
Often	45 (10.5%)	10 (9.7%)	35 (10.8%)		17 (9.8%)	16 (11.4%)
Almost always	0 (0.0%)	0 (0.0%)	0 (0.0%)		0 (0.0%)	0 (0.0%)
Missing	4	0	3		3	0
Unnecessary tests due to false positive findings
Never	37 (8.7%)	20 (19.4%)	17 (5.3%)	<0.001	10 (5.7%)	6 (4.3%)	0.086
Rarely	165 (38.6%)	49 (47.6%)	115 (35.6%)		66 (37.7%)	46 (33.1%)
Occasionally	174 (40.8%)	27 (26.2%)	147 (45.5%)		82 (46.9%)	62 (44.6%)
Often	50 (11.7%)	7 (6.8%)	43 (13.3%)		17 (9.7%)	24 (17.3%)
Almost always	1 (0.2%)	0 (0.0%)	1 (0.3%)		0 (0.0%)	1 (0.7%)
Missing	4	0	3		2	1
Unnecessary procedures due to false positive findings
Never	76 (17.7%)	30 (29.1%)	46 (14.2%)	<0.001	26 (14.8%)	19 (13.6%)	0.228
Rarely	213 (49.7%)	50 (48.5%)	162 (49.9%)		91 (51.7%)	66 (47.1%)
Occasionally	115 (26.8%)	17 (16.5%)	98 (30.2%)		53 (30.1%)	43 (30.7%)
Often	24 (5.6%)	6 (5.8%)	18 (5.5%)		6 (3.4%)	11 (7.9%)
Almost always	1 (0.2%)	0 (0.0%)	1 (0.3%)		0 (0.0%)	1 (0.7%)
Missing	2	0	1		1	0
Unnecessary medications due to false positive findings
Never	86 (20.1%)	27 (26.2%)	59 (18.2%)	0.264	39 (22.2%)	19 (13.6%)	0.020
Rarely	237 (55.3%)	51 (49.5%)	185 (56.9%)		100 (56.8%)	80 (57.1%)
Occasionally	89 (20.8%)	21 (20.4%)	68 (20.9%)		33 (18.8%)	34 (24.3%)
Often	16 (3.7%)	4 (3.9%)	12 (3.7%)		4 (2.3%)	6 (4.3%)
Almost always	1 (0.2%)	0 (0.0%)	1 (0.3%)		0 (0.0%)	1 (0.7%)
Missing	2	0	1		1	0
Unnecessary consultations due to false positive findings
Never	37 (8.6%)	16 (15.5%)	21 (6.5%)	<0.001	16 (9.1%)	4 (2.9%)	0.002
Rarely	153 (35.7%)	46 (44.7%)	106 (32.6%)		66 (37.5%)	36 (25.7%)
Occasionally	166 (38.7%)	34 (33.0%)	132 (40.6%)		63 (35.8%)	68 (48.6%)
Often	63 (14.7%)	7 (6.8%)	56 (17.2%)		29 (16.5%)	25 (17.9%)
Almost always	10 (2.3%)	0 (0.0%)	10 (3.1%)		2 (1.1%)	7 (5.0%)
Missing	2	0	1		1	0
How emotionally bothersome would you find it if one of your patients …
Was harmed by treatment for a finding discovered by a guideline-supported headache MRI
Not at all	14 (3.3%)	2 (2.0%)	12 (3.7%)	<0.001	7 (4.0%)	5 (3.6%)	0.701
Slightly	81 (19.1%)	8 (8.0%)	73 (22.6%)		41 (23.3%)	31 (22.5%)
Moderately	131 (30.9%)	29 (29.0%)	101 (31.3%)		51 (29.0%)	47 (34.1%)
Very much	145 (34.2%)	43 (43.0%)	102 (31.6%)		56 (31.8%)	41 (29.7%)
Extremely	53 (12.5%)	18 (18.0%)	35 (10.8%)		21 (11.9%)	14 (10.1%)
Missing	7	3	3		1	2
Was harmed by a guideline-supported decision not to perform headache MRI
Not at all	9 (2.1%)	1 (1.0%)	8 (2.5%)	0.002	6 (3.4%)	2 (1.4%)	0.376
Slightly	40 (9.4%)	6 (5.9%)	34 (10.5%)		12 (6.8%)	21 (15.1%)
Moderately	114 (26.8%)	21 (20.8%)	93 (28.7%)		53 (30.1%)	38 (27.3%)
Very much	175 (41.1%)	43 (42.6%)	131 (40.4%)		72 (40.9%)	54 (38.9%)
Extremely	88 (20.7%)	30 (29.7%)	58 (17.9%)		33 (18.8%)	24 (17.3%)
Missing	5	2	2		1	1

Abbreviations: APCs, advanced practice clinicians; MRI, magnetic resonance imaging.

Table 2. Expected improvement in MRI utilization upon implementation of various interventions, stratified by clinician type and specialty
Survey question	Overall (N = 431)	Advanced practice clinicians (n = 103)	Physicians (n = 326)	Wilcoxon rank sum test p value: APCs vs. physicians	Neurologists (n = 177)	Primary care clinicians (n = 140)	Wilcoxon rank sum test p value: neurologists vs. primary care clinicians
How much do you think your headache MRI use would improve if the following environmental changes were implemented …
Link to clinical practice guidelines in the electronic health record
Not at all	51 (11.9%)	2 (1.9%)	49 (15.1%)	<0.001	38 (21.6%)	10 (7.2%)	0.002
Slightly	103 (24.1%)	12 (11.7%)	90 (27.8%)		50 (28.4%)	39 (28.1%)
Moderately	132 (30.8%)	37 (35.9%)	95 (29.3%)		47 (26.7%)	44 (31.7%)
Very much	117 (27.3%)	38 (36.9%)	79 (24.4%)		34 (19.3%)	42 (30.2%)
Extremely	25 (5.8%)	14 (13.6%)	11 (3.4%)		7 (4.0%)	4 (2.9%)
Missing	3	0	2		1	1
Better clinical practice guidelines
Not at all	35 (8.2%)	1 (1.0%)	34 (10.6%)	<0.001	27 (15.4%)	6 (4.4%)	<0.001
Slightly	82 (19.3%)	10 (9.7%)	72 (22.4%)		44 (25.1%)	27 (19.6%)
Moderately	142 (33.3%)	32 (31.1%)	109 (33.9%)		56 (32.0%)	49 (35.5%)
Very much	142 (33.3%)	44 (42.7%)	98 (30.4%)		42 (24.0%)	53 (38.4%)
Extremely	25 (5.9%)	16 (15.5%)	9 (2.8%)		6 (3.4%)	3 (2.2%)
Missing	5	0	4		2	2
Restriction on imaging ordering within the electronic health record
Not at all	135 (32.0%)	20 (19.4%)	114 (35.9%)	<0.001	72 (41.4%)	39 (28.9%)	0.008
Slightly	96 (22.8%)	19 (18.5%)	77 (24.2%)		39 (22.4%)	37 (27.4%)
Moderately	89 (21.1%)	25 (24.3%)	64 (20.1%)		41 (23.6%)	21 (15.6%)
Very much	84 (19.9%)	29 (28.2%)	55 (17.3%)		19 (10.9%)	33 (24.4%)
Extremely	18 (4.3%)	10 (9.7%)	8 (2.5%)		3 (1.7%)	5 (3.7%)
Missing	9	0	8		3	5
Streamlined specialty referral protocols
Not at all	58 (13.7%)	3 (2.9%)	55 (17.1%)	<0.001	52 (30.1%)	2 (1.4%)	<0.001
Slightly	74 (17.4%)	9 (8.7%)	64 (19.9%)		42 (24.3%)	21 (15.1%)
Moderately	108 (25.4%)	25 (24.3%)	83 (25.9%)		39 (22.5%)	41 (29.5%)
Very much	143 (33.7%)	52 (50.5%)	91 (28.4%)		35 (20.2%)	53 (38.1%)
Extremely	42 (9.9%)	14 (13.6%)	28 (8.7%)		5 (2.9%)	22 (15.8%)
Missing	6	0	5		4	1
Increased availability of informal neurologist input prior to headache MRI
Not at all	78 (18.6%)	5 (4.9%)	72 (22.8%)	<0.001	65 (38.2%)	5 (3.7%)	NA
Slightly	64 (15.2%)	7 (6.8%)	57 (18.0%)		34 (20.0%)	23 (16.8%)
Moderately	90 (21.4%)	21 (20.4%)	69 (21.8%)		37 (21.8%)	29 (21.2%)
Very much	123 (29.3%)	41 (39.8%)	82 (26.0%)		28 (16.5%)	50 (36.5%)
Extremely	65 (15.5%)	29 (28.2%)	36 (11.4%)		6 (3.5%)	30 (21.9%)
Missing	11	0	10		7	3
Patient education materials regarding risks/benefits of headache MRI
Not at all	38 (9.0%)	4 (4.0%)	34 (10.6%)	<0.001	21 (12.0%)	11 (8.0%)	0.111
Slightly	122 (28.8%)	18 (17.8%)	103 (32.0%)		56 (32.0%)	46 (33.3%)
Moderately	124 (29.3%)	26 (25.7%)	98 (30.4%)		60 (34.3%)	34 (24.6%)
Very much	105 (24.8%)	38 (37.6%)	67 (20.8%)		28 (16.0%)	37 (26.8%)
Extremely	35 (8.3%)	15 (14.9%)	20 (6.2%)		10 (5.7%)	10 (7.3%)
Missing	7	2	4		2	2
Clinical decision support in the electronic health record to help with headache MRI decisions based on the patient's clinical scenario
Not at all	48 (11.2%)	1 (1.0%)	47 (14.6%)	<0.001	35 (19.9%)	10 (7.3%)	<0.001
Slightly	89 (20.8%)	14 (13.6%)	74 (22.9%)		51 (29.0%)	23 (16.7%)
Moderately	130 (30.5%)	31 (30.1%)	99 (30.7%)		49 (27.8%)	45 (32.6%)
Very much	127 (29.7%)	42 (40.8%)	85 (26.3%)		30 (17.1%)	53 (38.4%)
Extremely	33 (7.7%)	15 (14.6%)	18 (5.6%)		11 (6.3%)	7 (5.1%)
Missing	4	0	3		1	2
Shared decision-making tool to help patients with headache neuroimaging decisions
Not at all	43 (10.1%)	5 (4.9%)	38 (11.8%)	<0.001	25 (14.3%)	11 (7.9%)	0.002
Slightly	112 (26.2%)	16 (15.5%)	95 (29.4%)		61 (34.9%)	34 (24.5%)
Moderately	120 (28.1%)	29 (28.2%)	91 (28.2%)		47 (26.9%)	41 (29.5%)
Very much	123 (28.8%)	37 (35.9%)	86 (26.6%)		35 (20.0%)	47 (33.8%)
Extremely	29 (6.8%)	16 (15.5%)	13 (4.0%)		7 (4.0%)	6 (4.3%)
Missing	4	0	3		2	1
Less time pressure in clinic
Not at all	47 (11.0%)	2 (1.9%)	44 (13.6%)	<0.001	33 (18.9%)	9 (6.5%)	<0.001
Slightly	90 (21.1%)	13 (12.6%)	77 (23.8%)		55 (31.4%)	22 (15.8%)
Moderately	104 (24.4%)	32 (31.1%)	72 (22.3%)		31 (17.7%)	38 (27.3%)
Very much	111 (26.0%)	28 (27.2%)	83 (25.7%)		41 (23.4%)	38 (27.3%)
Extremely	75 (17.6%)	28 (27.2%)	47 (14.6%)		15 (8.6%)	32 (23.0%)
Missing	4	0	3		2	1
Reminders about optimal practice patterns in the electronic health record
Not at all	79 (18.8%)	7 (6.9%)	72 (22.6%)	<0.001	44 (25.6%)	26 (18.8%)	0.066
Slightly	128 (30.5%)	18 (17.8%)	109 (34.3%)		61 (35.5%)	47 (34.1%)
Moderately	111 (26.4%)	37 (36.6%)	74 (23.3%)		40 (23.3%)	32 (23.2%)
Very much	79 (18.8%)	26 (25.7%)	53 (16.7%)		21 (12.2%)	29 (21.0%)
Extremely	23 (5.5%)	13 (12.9%)	10 (3.2%)		6 (3.5%)	4 (2.9%)
Missing	11	2	8		5	2

Abbreviations: APCs, advanced practice clinicians; MRI, magnetic resonance imaging.

Table 3. Clinician attitudes and confidence towards MRI utilization, stratified by clinician type and specialty
Survey question	Overall (N = 431)	Advanced practice clinicians (n = 103)	Physicians (n = 326)	Wilcoxon rank sum test p value: APCs vs. physicians	Neurologists (n = 177)	Primary care clinicians (n = 140)	Wilcoxon rank sum test p value: neurologists vs. primary care clinicians
How would you rate your knowledge of guideline recommendations for headache neuroimaging?
Poor	62 (14.5%)	22 (21.4%)	40 (12.3%)	<0.001	9 (5.1%)	29 (20.7%)	<0.001
Fair	165 (38.5%)	50 (48.5%)	114 (35.1%)		42 (23.9%)	69 (49.3%)
Good	127 (29.6%)	23 (22.3%)	104 (32.0%)		66 (37.5%)	34 (24.3%)
Very good	60 (14.0%)	7 (6.8%)	53 (16.3%)		46 (26.1%)	7 (5.0%)
Excellent	15 (3.5%)	1 (1.0%)	14 (4.3%)		13 (7.4%)	1 (0.7%)
Missing	2	0	1		1	0
Evaluate your agreement with the following statements:
MRI is overused in patients presenting with headaches
Strongly disagree	7 (1.6%)	2 (2.0%)	5 (1.5%)	<0.001	1 (0.6%)	2 (1.4%)	0.723
Disagree	31 (7.2%)	16 (15.7%)	15 (4.6%)		9 (5.1%)	5 (3.6%)
Neutral	74 (17.3%)	30 (29.4%)	44 (13.5%)		25 (14.1%)	16 (11.4%)
Agree	208 (48.5%)	38 (37.3%)	170 (52.2%)		87 (49.2%)	81 (57.9%)
Strongly agree	109 (25.4%)	16 (15.7%)	92 (28.2%)		55 (31.1%)	36 (25.7%)
Missing	2	1	0		0	0
It is important to reduce overall use of MRI in patients presenting with headaches
Strongly disagree	4 (0.9%)	0 (0.0%)	4 (1.2%)	0.007	1 (0.6%)	2 (1.4%)	0.227
Disagree	37 (8.7%)	15 (14.9%)	22 (6.8%)		13 (7.4%)	6 (4.3%)
Neutral	107 (25.1%)	31 (30.7%)	76 (23.4%)		46 (26.1%)	27 (19.3%)
Agree	202 (47.3%)	42 (41.6%)	160 (49.2%)		81 (46.0%)	77 (55.0%)
Strongly agree	77 (18.0%)	13 (12.9%)	63 (19.4%)		35 (19.9%)	28 (20.0%)
Missing	4	2	1		1	0
Given how headaches are currently evaluated, it is common that an underlying headache-causing diagnosis is missed during initial evaluation
Strongly disagree	22 (5.2%)	1 (1.0%)	21 (6.5%)	<0.001	15 (8.5%)	6 (4.3%)	<0.001
Disagree	160 (37.7%)	25 (25.0%)	135 (41.7%)		88 (50.0%)	46 (33.1%)
Neutral	116 (27.3%)	31 (31.0%)	84 (25.9%)		41 (23.3%)	39 (28.1%)
Agree	103 (24.2%)	37 (37.0%)	66 (20.4%)		25 (14.2%)	38 (27.3%)
Strongly agree	24 (5.7%)	6 (6.0%)	18 (5.6%)		7 (4.0%)	10 (7.2%)
Missing	6	3	2		1	1
It is my responsibility to never miss a brain tumor causing headache
Strongly disagree	9 (2.1%)	1 (1.0%)	8 (2.5%)	0.211	5 (2.8%)	3 (2.2%)	0.286
Disagree	26 (6.1%)	5 (4.9%)	21 (6.5%)		11 (6.2%)	10 (7.2%)
Neutral	49 (11.5%)	6 (5.9%)	42 (12.9%)		19 (10.7%)	21 (15.1%)
Agree	213 (49.8%)	58 (56.9%)	155 (47.7%)		84 (47.5%)	66 (47.5%)
Strongly agree	131 (30.6%)	32 (31.4%)	99 (30.5%)		58 (32.8%)	39 (28.1%)
Missing	3	1	1		0	1
When it comes to medical care, in general, more is usually better
Strongly disagree	64 (16.1%)	8 (8.5%)	56 (18.5%)	0.010	29 (17.1%)	27 (21.6%)	0.457
Disagree	222 (55.9%)	54 (57.5%)	168 (55.5%)		96 (56.5%)	69 (55.2%)
Neutral	77 (19.4%)	17 (18.1%)	60 (19.8%)		39 (22.9%)	18 (14.4%)
Agree	31 (7.8%)	14 (14.9%)	17 (5.6%)		5 (2.9%)	10 (8.0%)
Strongly agree	3 (0.8%)	1 (1.1%)	2 (0.7%)		1 (0.6%)	1 (0.8%)
Missing	34	9	23		7	15
When it comes to diagnostic tests, more is usually better
Strongly disagree	89 (21.0%)	12 (12.0%)	77 (23.9%)	0.020	40 (23.1%)	37 (26.4%)	0.896
Disagree	217 (51.3%)	55 (55.0%)	162 (50.3%)		92 (53.2%)	68 (48.6%)
Neutral	86 (20.3%)	25 (25.0%)	61 (18.9%)		33 (19.1%)	24 (17.1%)
Agree	26 (6.2%)	7 (7.0%)	19 (5.9%)		7 (4.1%)	9 (6.4%)
Strongly agree	5 (1.2%)	1 (1.0%)	3 (0.9%)		1 (0.6%)	2 (1.4%)
Missing	8	3	4		4	0
I am confident in my ability to:
Take a history to identify "red flags" suggesting a possible intracranial mass lesion
Strongly disagree	1 (0.2%)	0 (0.0%)	1 (0.3%)	<0.001	1 (0.6%)	0 (0.0%)	<0.001
Disagree	4 (0.9%)	4 (3.9%)	0 (0.0%)		0 (0.0%)	0 (0.0%)
Neutral	24 (5.6%)	12 (11.8%)	12 (3.7%)		2 (1.1%)	9 (6.4%)
Agree	242 (56.4%)	61 (59.8%)	180 (55.2%)		75 (42.4%)	99 (70.7%)
Strongly agree	158 (36.8%)	25 (24.5%)	133 (40.8%)		99 (55.9%)	32 (22.9%)
Missing	2	1	0		0	0
Identify abnormalities on neurologic examination that suggest a possible intracranial mass lesion
Strongly disagree	2 (0.5%)	1 (1.0%)	1 (0.3%)	<0.001	1 (0.6%)	0 (0.0%)	<0.001
Disagree	6 (1.4%)	4 (3.9%)	2 (0.6%)		0 (0.0%)	2 (1.4%)
Neutral	43 (10.1%)	17 (16.7%)	25 (7.7%)		2 (1.1%)	23 (16.4%)
Agree	240 (56.1%)	69 (67.7%)	171 (52.6%)		71 (40.3%)	92 (65.7%)
Strongly agree	137 (32.0%)	11 (10.8%)	126 (38.8%)		102 (58.0%)	23 (16.4%)
Missing	3	1	1		1	0
Make decisions about whether to obtain MRI in headache patients
Strongly disagree	1 (0.2%)	0 (0.0%)	1 (0.3%)	<0.001	1 (0.6%)	0 (0.0%)	<0.001
Disagree	12 (2.8%)	7 (6.9%)	5 (1.5%)		0 (0.0%)	5 (3.6%)
Neutral	27 (6.3%)	14 (13.7%)	13 (4.0%)		0 (0.0%)	13 (9.3%)
Agree	275 (64.1%)	68 (66.7%)	206 (63.2%)		96 (54.2%)	102 (72.9%)
Strongly agree	114 (26.6%)	13 (12.8%)	101 (31.0%)		80 (45.2%)	20 (14.3%)
Missing	2	1	0		0	0
Discuss whether to obtain MRI with a patient that wants MRI, but I believe it is not clinically indicated
Strongly disagree	3 (0.7%)	2 (2.0%)	1 (0.3%)	0.016	1 (0.6%)	0 (0.0%)	<0.001
Disagree	19 (4.5%)	6 (5.9%)	13 (4.0%)		2 (1.1%)	10 (7.3%)
Neutral	58 (13.7%)	16 (15.8%)	41 (12.7%)		16 (9.0%)	25 (18.3%)
Agree	258 (60.7%)	64 (63.4%)	194 (60.1%)		98 (55.4%)	89 (65.0%)
Strongly agree	87 (20.5%)	13 (12.9%)	74 (22.9%)		60 (33.9%)	13 (9.5%)
Missing	6	2	3		0	3
Interpret and appropriately act on brain MRI reports
Strongly disagree	2 (0.5%)	1 (1.0%)	1 (0.3%)	<0.001	1 (0.6%)	0 (0.0%)	<0.001
Disagree	27 (6.3%)	14 (13.7%)	13 (4.0%)		1 (0.6%)	11 (7.9%)
Neutral	37 (8.7%)	16 (15.7%)	21 (6.5%)		2 (1.1%)	19 (13.6%)
Agree	224 (52.3%)	60 (58.8%)	164 (50.5%)		69 (39.2%)	90 (64.3%)
Strongly agree	138 (32.2%)	11 (10.8%)	126 (38.8%)		103 (58.5%)	20 (14.3%)
Missing	3	1	1		1	0

Abbreviations: APCs, advanced practice clinicians; MRI, magnetic resonance imaging.

Table 4. Other considerations for MRI ordering: Patient, time, and other concerns, stratified by clinician type and specialty
Survey question	Overall (N = 431)	Advanced practice clinicians (n = 103)	Physicians (n = 326)	Wilcoxon rank sum test p value: APCs vs. physicians	Neurologists (n = 177)	Primary care clinicians (n = 140)	Wilcoxon rank sum test p value: neurologists vs. primary care clinicians
How commonly do you order MRI for a patient presenting with headaches …
When a patient requests the test
Never	27 (6.3%)	8 (7.8%)	19 (5.8%)	0.480	5 (2.8%)	13 (9.3%)	0.003
Rarely	140 (32.6%)	32 (31.4%)	108 (33.1%)		55 (31.1%)	51 (36.4%)
Occasionally	192 (44.8%)	50 (49.0%)	142 (43.6%)		79 (44.6%)	60 (42.9%)
Often	51 (11.9%)	8 (7.8%)	42 (12.9%)		27 (15.3%)	13 (9.3%)
Almost always	19 (4.4%)	4 (3.9%)	15 (4.6%)		11 (6.2%)	3 (2.1%)
Missing	2	1	0		0	0
When a patient requests the test, but you believe it is not clinically indicated
Never	74 (17.4%)	14 (13.9%)	60 (18.5%)	0.992	30 (17.1%)	28 (20.1%)	0.095
Rarely	203 (47.7%)	54 (53.5%)	149 (46.0%)		78 (44.3%)	69 (49.6%)
Occasionally	120 (28.2%)	28 (27.7%)	91 (28.1%)		51 (29.0%)	36 (25.9%)
Often	19 (4.5%)	3 (3.0%)	16 (4.9%)		11 (6.3%)	5 (3.6%)
Almost always	10 (2.4%)	2 (2.0%)	8 (2.5%)		6 (3.4%)	1 (0.7%)
Missing	5	2	2		1	1
For your patients that have received an MRI for headaches, how often did you believe it was the right decision to order the MRI
Never	2 (0.5%)	0 (0.0%)	2 (0.6%)	0.021	1 (0.6%)	1 (0.7%)	0.329
Rarely	37 (8.8%)	3 (3.0%)	34 (10.7%)		21 (12.2%)	12 (8.7%)
Occasionally	111 (26.5%)	23 (23.2%)	88 (27.6%)		42 (24.4%)	46 (33.3%)
Often	199 (47.5%)	54 (54.6%)	144 (45.1%)		76 (44.2%)	61 (44.2%)
Always	70 (16.7%)	19 (19.2%)	51 (16.0%)		32 (18.6%)	18 (13.0%)
Missing	12	4	7		5	2
On average, how much time does it take
To discuss, order, interpret and follow-up an MRI for a patient presenting for headaches
0–5 min per patient	37 (8.7%)	3 (3.0%)	34 (10.5%)	0.023	16 (9.1%)	17 (12.1%)	0.969
6–10 min per patient	132 (31.0%)	28 (28.0%)	104 (32.0%)		62 (35.2%)	39 (27.9%)
10–20 min per patient	168 (39.4%)	44 (44.0%)	123 (37.9%)		62 (35.2%)	59 (42.1%)
20+ min per patient	89 (20.9%)	25 (25.0%)	64 (19.7%)		36 (20.5%)	25 (17.9%)
Missing	5	3	1		1	0
To NOT order an MRI for a patient presenting with headaches? (e.g., discussing decision with patient, implementing alternative management strategies)
0–5 min per patient	62 (14.6%)	9 (9.1%)	53 (16.3%)	0.055	41 (23.3%)	12 (8.6%)	<0.001
6–10 min per patient	141 (33.2%)	30 (30.3%)	111 (34.2%)		67 (38.1%)	40 (28.6%)
10–20 min per patient	141 (33.2%)	39 (39.4%)	101 (31.1%)		42 (23.9%)	55 (39.3%)
20+ min per patient	81 (19.1%)	21 (21.2%)	60 (18.5%)		26 (14.8%)	33 (23.6%)
Missing	6	4	1		1	0
How often are the following considerations relevant to your decisions to order MRI in patients presenting with headache
Malpractice concerns
Never	70 (16.4%)	15 (14.6%)	55 (17.0%)	0.334	33 (18.8%)	20 (14.4%)	0.038
Rarely	125 (29.2%)	27 (26.2%)	98 (30.3%)		57 (32.4%)	40 (28.8%)
Occasionally	118 (27.6%)	31 (30.1%)	87 (26.9%)		50 (28.4%)	34 (24.5%)
Often	88 (20.6%)	24 (23.3%)	63 (19.4%)		28 (15.9%)	33 (23.7%)
Almost always	27 (6.3%)	6 (5.8%)	21 (6.5%)		8 (4.6%)	12 (8.6%)
Missing	3	0	2		1	1
Financial incentives
Never	360 (84.7%)	77 (75.5%)	282 (87.6%)	0.002	155 (88.6%)	119 (86.2%)	0.564
Rarely	45 (10.6%)	14 (13.7%)	31 (9.6%)		15 (8.6%)	16 (11.6%)
Occasionally	15 (3.5%)	8 (7.8%)	7 (2.2%)		4 (2.3%)	3 (2.2%)
Often	4 (0.9%)	3 (2.9%)	1 (0.3%)		1 (0.6%)	0 (0.0%)
Almost always	1 (0.2%)	0 (0.0%)	1 (0.3%)		0 (0.0%)	0 (0.0%)
Missing	6	1	4		2	2
Requirements prior to specialist referral
Never	111 (26.4%)	11 (10.7%)	99 (31.3%)	<0.001	93 (55.4%)	6 (4.3%)	<0.01
Rarely	56 (13.3%)	6 (5.8%)	50 (15.8%)		40 (23.8%)	9 (6.5%)
Occasionally	71 (16.9%)	17 (16.5%)	54 (17.1%)		26 (15.5%)	28 (20.1%)
Often	107 (25.5%)	42 (40.8%)	65 (20.6%)		7 (4.2%)	55 (39.6%)
Almost always	75 (17.9%)	27 (26.2%)	48 (15.2%)		2 (1.2%)	41 (29.5%)
Missing	11	0	10		9	1

Abbreviations: APCs, advanced practice clinicians; MRI, magnetic resonance imaging.

Authors and Disclosures

Evan L. Reynolds PhD¹, James F. Burke MD, MS^1,2, Lacey Evans MPH², Faiz I. Syed MD, MS^3,4, Eric Liao MD³, Remy Lobo MD³, Wade Cooper DO¹, Larry Charleston IV MD, MSc^5,6 and Brian C. Callaghan MD, MS^1,2

¹Department of Neurology, University of Michigan, Ann Arbor, Michigan, USA
²VA Center for Clinical Management Research, Ann Arbor, Michigan, USA
³Department of Radiology, University of Michigan, Ann Arbor, Michigan, USA
⁴Department of Radiology, VA Ann Arbor Health System, Ann Arbor, Michigan, USA
⁵College of Human Medicine, Michigan State University, East Lansing, Michigan, USA
⁶Jefferson Headache Center, Philadelphia, Pennsylvania, USA

Correspondence
Brian C. Callaghan, Department of Neurology, University of Michigan, 109 Zina Pitcher Place, 4021 BSRB, Ann Arbor, MI 48104, USA. Email: bcallagh@med.umich.edu

Conflict of Interest
Dr. Reynolds reports no disclosures. Dr. Burke has received compensation from Astra Zeneca for his role on the adjudication committee of the SOCRATES trial. Ms. Evans reports no disclosures. Dr. Syed reports no disclosures. Dr. Liao reports no disclosures. Dr. Lobo reports no disclosures. Dr. Cooper reports no disclosures. Dr. Charleston has served as consultant, advisory board member, or has received honoraria from AbbVie/Allergan, Alder/Lundbeck, and Biohaven; is on the advisory board of BrainWeekend and Ctrl M Health; has performed medical legal consultation for the Vaccine Injury Compensation Program; and is an Associate Editor of Headache: The Journal of Head and Face Pain. He reports no relevant conflict of interest for this manuscript. Dr. Callaghan consults for a PCORI grant, DynaMed, receives research support from the American Academy of Neurology and performs medical legal consultations including consultations for the Vaccine Injury Compensation Program.

Author Contributions
Study concept and design: Evan L. Reynolds, James F. Burke, Brian C. Callaghan. Acquisition of data: James F. Burke, Lacey Evans, Faiz I. Syed, Brian C. Callaghan. Analysis and interpretation of data: Evan L. Reynolds, James F. Burke, Lacey Evans, Faiz I. Syed, Eric Liao, Remy Lobo, Wade Cooper, Larry Charleston IV, Brian C. Callaghan. Drafting of the manuscript: Evan L. Reynolds, James F. Burke, Brian C. Callaghan. Revising it for intellectual content: Evan L. Reynolds, James F. Burke, Lacey Evans, Faiz I. Syed, Eric Liao, Remy Lobo, Wade Cooper, Larry Charleston IV, Brian C. Callaghan. Final approval of the completed manuscript: Evan L. Reynolds, James F. Burke, Lacey Evans, Faiz I. Syed, Eric Liao, Remy Lobo, Wade Cooper, Larry Charleston IV, Brian C. Callaghan.