International patient preferences for physician attire: results from cross-sectional studies in four countries across three continents.

OBJECTIVE: The patient-physician relationship impacts patients' experiences and health outcomes. Physician attire is a form of nonverbal communication that influences this relationship. Prior studies examining attire preferences suffered from heterogeneous measurement and limited context. We thus performed a multicentre, cross-sectional study using a standardised survey instrument to compare patient preferences for physician dress in international settings.
SETTING: 20 hospitals and healthcare practices in Italy, Japan, Switzerland and the USA. PARTICIPANTS: Convenience sample of 9171 adult patients receiving care in academic hospitals, general medicine clinics, specialty clinics and ophthalmology practices. PRIMARY AND SECONDARY OUTCOME MEASURES: The survey was randomised and included photographs of a male or female physician dressed in assorted forms of attire. The primary outcome measure was attire preference, comprised of composite ratings across five domains: how knowledgeable, trustworthy, caring and approachable the physician appeared, and how comfortable the respondent felt. Secondary outcome measures included variation in preferences by country, physician type and respondent characteristics.
RESULTS: The highest rated forms of attire differed by country, although each most preferred attire with white coat. Low ratings were conferred on attire extremes (casual and business suit). Preferences were more uniform for certain physician types. For example, among all respondents, scrubs garnered the highest rating for emergency department physicians (44.2%) and surgeons (42.4%). However, attire preferences diverged for primary care and hospital physicians. All types of formal attire were more strongly preferred in the USA than elsewhere. Respondent age influenced preferences in Japan and the USA only.
CONCLUSIONS: Patients across a myriad of geographies, settings and demographics harbour specific preferences for physician attire. Some preferences are nearly universal, whereas others vary substantially. As a one-size-fits-all dress policy is unlikely to reflect patient desires and expectations, a tailored approach should be sought that attempts to match attire to clinical context. © Author(s) (or their employer(s)) 2022. Re-use permitted under CC BY. Published by BMJ.

With over 9000 participants, this is the largest international study examining opinions on physician dress to date.

Use of a standardised survey instrument allowed direct comparisons across diverse geographic regions, populations, physician types and clinical contexts.

Robust and careful survey design, including randomisation and constant photographic features, mitigated bias and confounding.

Comparative over-representation of the USA and convenience sampling may have contributed to disproportionate representation.

The survey instrument used predefined Likert scales, which may not accurately reflect nuanced patient opinions, and which do not capture other elements of patient–physician interactions.

Introduction

Successful patient–physician relationships are founded on mutual respect, trust, confidence and care. The strength of these connections can directly impact patients’ experiences with healthcare, satisfaction and important health outcomes such as adherence to treatment recommendations,1 2 30-day readmissions3 and mortality.4 Patient–physician interactions are complex and dependent on multiple factors including social definitions and cultural norms. To ensure the highest quality care, it is essential to identify techniques that physicians may use to establish and maintain strong relationships with their unique individual patients while recognising the influence of sociocultural context. From initial introductions, physicians employ verbal and nonverbal communication to form impressions and cultivate partnerships with their patients.5

The clothing worn by a physician is one form of nonverbal communication that may influence the patient–physician relationship. Physician attire is an important element in establishing patient confidence and trust,6 enhancing patient comfort when discussing personal problems7–9 and shaping patient perceptions of physician professionalism,6 intelligence10 and empathy.11 Most prior scholarship has focused on a single geographic region, country or clinical context (eg, primary care clinic, hospital setting)12–15 and has not considered the relative impacts of different physician specialties, contexts of care, geography and patient factors such as age, education and gender. In addition, heterogeneity among prior studies, such as different sampling methodology and survey instruments, has made comparisons across different studies challenging.

The objective of this international, multicentre, cross-sectional study was to use a structured survey instrument to examine patient preferences for physician attire in different regions, countries and continents. The survey instrument allowed direct comparisons among a variety of cultures and contexts, thereby mitigating the heterogeneity of prior studies.16–18 We report comparisons of data from five primary cross-sectional survey research studies conducted in Italy, Japan,19 Switzerland20 and the USA.21 22 Our aim was to identify common themes and differences of patient expectations for physician dress so that we may tailor attire and thus elevate the patient experience and optimise health outcomes.

Methods

Study design and participants

We performed a survey-based study using a convenience sample of patients in 20 hospitals and healthcare practices in Italy, Japan, Switzerland and the USA. These sites were selected based on our research networks and availability of clinicians who would serve as leads in their respective institutions. Sites included academic hospitals (general medicine wards, intensive care units), general medicine ambulatory clinics, specialty ambulatory clinics (dermatology, infectious disease, neurology, orthopaedic surgery) and ophthalmology practices (table 1). Data collection took place between June 2015 and October 2017.

Table 1

Characteristics of participating study sites

Country	Dates of data collection	Types of outpatient clinics	Clinical setting(s)	Hospitals, Practices	Geographic regions sampled	Surveys completed
Italy	10/26/2015-10/21/2016	Infectious Disease, Ophthalmology, Geriatric Intensive Care Unit	Outpatient and Inpatient	1	1*	958
Japan	12/01/2015-10/30/2017	General Medicine, Medicine Specialties, Orthopaedic Surgery	Outpatient and Inpatient	4	3†	2020
Switzerland	06/15/2015-10/31/2016	Dermatology, Infectious Disease, Neurology	Outpatient	1	1‡	834
USA§	06/01/2015-10/31/2016	General Medicine, Medicine Specialties	Outpatient and Inpatient	10	4¶	4062
		Ophthalmology	Outpatient	4	3**	1297

*One site in the Tuscany region.

†Two sites in the Kantō region; one site in the Kansai region; one site in the Chūgoku region.

‡One site in the Canton of Zurich.

§Geographic regions of the USA include Northeast, Midwest, South and West.

¶Three sites in the Midwest, three sites in the South, two sites in the Northeast, two sites in the West.

**Two sites in the Midwest, one site in the Northeast, one site in the West.

At each participating healthcare location, the research team printed and randomly administered a survey instrument, targeting representative adult patients who were receiving clinical care at one of those sites. Participants were presented with a paper-based instrument of 22 questions that included photographs of either a male or female physician wearing various forms of attire and asked to rate their preferences. Respondents could request assistance with form completion from persons accompanying them.

All participants provided informed verbal consent. No identifying information was collected from participants who completed the study. Institutional permission for recruitment and data collection was obtained from each site.

Procedures

The 22-item survey instrument was developed following a systematic review of the literature that examined the role of physician attire on the patient experience.23 The survey instrument was developed and piloted by a multidisciplinary team to gather feedback and refine photographs, questions, rating scale, presentation order and randomisation scheme. Questions were translated into different languages for each country by interpreters at each site: Italian for Italy, Japanese for Japan, German for Switzerland (since the Swiss survey was conducted in Zurich), and English for the USA.

Each question referenced particular preferences and opinions of respondents in relation to photographs of medical providers wearing seven unique forms of attire. The forms of dress presented included: casual, casual with white coat, scrubs, scrubs with white coat, formal, formal with white coat and business suit. Photographs were taken with attention paid to achieving constant physician facial expressions as well as consistent visual cues such as lighting, background and pose. Photographs used at all study sites were identical with the following exceptions: In Switzerland, photographs of physicians in medical attire contextually appropriate to the Swiss health system (ie, white scrubs instead of blue scrubs) were used. All other photographic elements including physician models and other forms of attire were unchanged. In Japan, photographs of physicians of Japanese descent with slightly modified attire were used (online supplemental appendix A).

Each survey instrument had four sections. The first section showed a photograph of either a male or female physician wearing one of the seven unique forms of attire. To avoid biases such as anchoring, priming, order response, and gender conformity, 14 different versions of the survey instrument were created. The gender and attire of the first photograph seen by each respondent were randomised; all other sections of the survey were identical (online supplemental appendix B).

Measurements

Respondents were first asked to rate the standalone, randomised physician photograph using a 1–10 scale across five domains (ie, how knowledgeable, trustworthy, caring and approachable the physician appeared, and how comfortable the physician’s appearance made the respondent feel), with a score of 10 representing the highest rating. Respondents were subsequently given seven photographs of the same physician wearing various forms of attire. Questions were asked regarding preference of attire in varied clinical settings (ie, primary care, emergency department, hospital, surgery) and overall preference. To identify the influence of and respondent preferences for physician dress and white coats, a Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree) was employed. Preferences for attire by respondent characteristics such as age, gender, education level, nationality and number of unique physicians seen in the past year were collected. Unanswered questions and those with multiple responses were excluded.

The primary outcome of attire preference was calculated as the mean composite score of the five individual rating domains (ie, knowledgeable, trustworthy, caring, approachable and comfortable), with the highest score representing the most preferred form of attire. We also assessed variation in preferences for physician attire between countries, by physician type and clinical setting, and by respondent characteristics such as age and gender.

Statistical analysis

Survey data were entered independently and in duplicate by the study teams. Respondents were not required to answer all questions; therefore, the denominator for each question varied. Data were reported as mean and SD, or N and percentage, where appropriate. Differences in the mean composite rating scores between countries were assessed using one-way analysis of variance (ANOVA) with Tukey’s method for pairwise comparisons. Differences in mean composite score within country by sociodemographic factors were assessed using Student’s t-tests. Differences between countries with respect to categorical responses were compared by using χ2 tests. Statistical tests were assessed using p<0.05 considered significant. All analyses were performed using SAS V9.4 (SAS).

Patient and public involvement

Patients were not included in the design of the survey instrument, recruitment or conduct of the study. Patients who participated did so anonymously, and therefore, the study team will be unable to disseminate the results to study participants.

Results

Characteristics of study sites and participants

A total of 9171 patients completed the survey instrument in outpatient and inpatient healthcare settings within a total of 20 hospitals or practices, 13 distinct geographic regions, 4 countries and 3 continents. Patients were examined in age ranges of 18–64 years and 65 years or older. Patients aged 65 years or older comprised 36.0% of all respondents with substantial age variation across countries. For instance, those 65 years or older represented 48.5% of respondents in Japan, 35.6% in the USA, 27.8% in Italy and 16.7% in Switzerland. Among all respondents, 44.9% were female, 39.6% had a college or graduate degree and 26.6% had seen 6 or more physicians in the previous year. Characteristics of study sites are found in table 1, and sociodemographic characteristics of respondents are described in table 2.

Table 2

Sociodemographic information

	Italy (n=958)	Japan (n=2020)	Switzerland (n=834)	USA (n=5359)	Total (n=9171)
Age	n=928	n=2010	n=812	n=5279	n=9029
18–25	61 (6.6%)	67 (3.3%)	50 (6.2%)	241 (4.6%)	419 (4.6%)
26–34	89 (9.6%)	162 (8.1%)	93 (11.5%)	464 (8.8%)	808 (9.0%)
35–54	310 (33.4%)	461 (22.9%)	341 (42.0%)	1299 (24.6%)	2411 (26.7%)
55–64	210 (22.6%)	345 (17.2%)	192 (23.6%)	1393 (26.4%)	2140 (23.7%)
≥65	258 (27.8%)	975 (48.5%)	136 (16.7%)	1882 (35.6%)	3251 (36.0%)
Gender	n=905	n=2011	n=806	n=5194	n=8916
Female	471 (52.0%)	1040 (51.7%)	304 (37.7%)	2184 (42.0%)	3999 (44.9%)
Male	434 (48.0%)	971 (48.3%)	502 (62.3%)	3010 (58.0%)	4917 (55.1%)
Education	n=919	n=2010	n=808	n=5247	n=8984
Less than high school	237 (25.8%)	243 (12.1%)	368 (45.5%)	146 (2.8%)	994 (11.1%)
High school diploma	416 (45.3%)	1236 (61.5%)	82 (10.2%)	2691 (51.3%)	4425 (49.3%)
College degree	77 (8.4%)	487 (24.2%)	340 (42.1%)	1490 (28.4%)	2394 (26.6%)
Graduate degree	189 (20.5%)	44 (2.2%)	18 (2.2%)	920 (17.5%)	1171 (13.0%)
No of unique physicians seen in the past year	n=928	n=2009	n=810	n=5265	n=9012
0	76 (8.2%)	38 (1.9%)	13 (1.6%)	51 (1.0%)	178 (2.0%)
1	126 (13.6%)	140 (7.0%)	83 (10.2%)	377 (7.2%)	726 (8.1%)
2	199 (21.4%)	373 (18.5%)	165 (20.4%)	769 (14.6%)	1506 (16.7%)
3	188 (20.3%)	512 (25.5%)	203 (25.1%)	940 (17.9%)	1843 (20.4%)
4	112 (12.1%)	359 (17.9%)	126 (15.6%)	824 (15.6%)	1421 (15.8%)
5	84 (9.0%)	225 (11.2%)	57 (7.0%)	571 (10.8%)	937 (10.4%)
≥6	143 (15.4%)	362 (18.0%)	163 (20.1%)	1733 (32.9%)	2401 (26.6%)

Ratings of attire types by country

Responses regarding patient preferences for physician attire varied by country. Formal attire with white coat received the highest ratings from respondents in Italy and the USA with mean composite scores of 7.5 (SD 1.8) and 8.1 (SD 1.8), respectively. Conversely, scrubs with white coat received the highest ratings in Switzerland (mean composite score of 7.5 (SD 1.7)) and casual attire with white coat in Japan (mean composite score of 7.1 (SD 1.8)). The forms of attire that received the lowest mean composite ratings were business suit in Italy, Japan, and Switzerland with mean composite scores of 5.6 (SD 2.4), 5.5 (SD 2.1) and 5.2 (SD 2.2), respectively, and casual attire in the USA with a mean composite score of 6.2 (SD 2.5). Within each country, composite scores for attire forms with white coat were higher than those for the corresponding forms without white coat, with only one exception (composite scores for scrubs and scrubs with white coat in Japan were 6.8 and 6.6, respectively). Ratings of different forms of attire by country are found in figure 1 and ratings of physician attire by domain are found in online supplemental appendix C.

Figure 1

Mean composite ratings of physician attire.

Comparisons of patient preferences between countries

Preferences for physician attire by type of attire

Similarities between countries when comparing preferences for different types of physician attire were observed. For instance, there was complete concordance for all types of attire between the European countries of Italy and Switzerland. There was near complete concordance when comparing Italy and Japan, with the only statistically significant difference of Italy more strongly preferring formal attire with white coat compared with Japan (mean composite rating difference 0.54, simultaneous 95% confidence limits 0.06 to 1.01). Similarly, there was near complete concordance when comparing Switzerland and Japan, with the only significant difference of Switzerland more strongly preferring scrubs with white coat compared with Japan (mean composite rating difference 0.90, simultaneous 95% confidence limits 0.36 to 1.44). Among all types of attire, the form with the most concordance across countries was casual attire, with no between-country differences identified.

Just as ratings for physician attire varied by country, preferences for specific forms of attire also differed. For instance, the USA significantly more strongly preferred both forms of scrubs-based attire when compared with Italy and Japan, but not when compared with Switzerland. In addition, the USA significantly more strongly preferred all forms of formal attire (ie, formal attire with or without white coat and business suit) when compared with the other countries. These results are summarised in online supplemental appendix D.

Preferences for physician attire by type of physician

Photographs of either a male or female physician in seven different forms of attire (online supplemental appendix B) were shown, and respondents were asked to select which attire they preferred for different physician types. With respect to primary care physicians, respondents had varying preferences for attire. The highest rated forms in each country were formal attire with white coat in Italy (31.6%) and the USA (46.8%), casual attire with white coat in Japan (34.1%) and casual attire in Switzerland (24.4%). Heterogeneity in patient preferences was particularly noted in Switzerland with nearly equal preference given to casual attire, casual attire with white coat and formal attire with white coat. The highest rated form of attire across all respondents was formal attire with white coat (40.1%).

With respect to hospital-based physicians, preferences again diverged. The highest rated forms in each country were scrubs with white coat in Italy (43.8%) and Switzerland (35.0%), casual attire with white coat in Japan (34.0%) and formal attire with white coat in the USA (37.6%). The highest rated form of attire across all respondents was formal attire with white coat (32.8%).

With respect to both emergency department physicians and surgeons, preferences were more uniform. Among all respondents across all countries, the most preferred form of attire was scrubs (44.2% for emergency department physicians, 42.4% for surgeons) followed by scrubs with white coat (30.4% for emergency department physicians, 25.4% for surgeons).

With respect to the most preferred form of attire overall, differences between countries were noted. The top forms of attire in each country were scrubs with white coat in Italy (41.7%) and Switzerland (31.5%) and formal attire with white coat in Japan (35.3%) and the USA (45.7%). The highest rated form of attire across all respondents was formal attire with white coat (38.6%). Among all respondents, 78.6% preferred some form of attire with a white coat, while 21.4% preferred a form without a white coat. Table 3 shows preferred physician attire by physician type and clinical care setting.

Table 3

Preferred physician attire by physician type and care setting

Physician type	Attire	Italy	Japan	Switzerland	USA	Total
Primary care physician	Casual	103 (11.0%)	33 (1.6%)	199 (24.4%)	158 (3.0%)	493 (5.5%)
	Casual with white coat	165 (17.6%)	682 (34.1%)	183 (22.4%)	518 (9.9%)	1548 (17.2%)
	Scrubs	61 (6.5%)	188 (9.4%)	90 (11.0%)	238 (4.6%)	577 (6.4%)
	Scrubs with white coat	126 (13.5%)	357 (17.9%)	78 (9.6%)	742 (14.2%)	1303 (14.5%)
	Formal	128 (13.7%)	49 (2.5%)	73 (8.9%)	787 (15.0%)	1037 (11.6%)
	Formal with white coat	296 (31.6%)	669 (33.4%)	188 (23.0%)	2451 (46.8%)	3604 (40.1%)
	Business suit	57 (6.1%)	22 (1.1%)	6 (0.7%)	340 (6.5%)	425 (4.7%)
Emergency department physician	Casual	36 (3.9%)	42 (2.1%)	31 (3.8%)	63 (1.2%)	172 (1.9%)
	Casual with white coat	89 (9.6%)	206 (10.3%)	65 (8.0%)	298 (5.7%)	658 (7.3%)
	Scrubs	343 (37.2%)	1131 (56.5%)	382 (46.9%)	2108 (40.2%)	3964 (44.2%)
	Scrubs with white coat	324 (35.1%)	354 (17.7%)	271 (33.3%)	1784 (34.1%)	2733 (30.4%)
	Formal	16 (1.7%)	61 (3.0%)	8 (1.0%)	134 (2.6%)	219 (2.4%)
	Formal with white coat	105 (11.4%)	204 (10.2%)	52 (6.4%)	793 (15.1%)	1154 (12.9%)
	Business suit	10 (1.1%)	5 (0.2%)	5 (0.6%)	60 (1.1%)	80 (0.9%)
Hospital physician	Casual	25 (2.7%)	19 (1.0%)	33 (4.1%)	68 (1.3%)	145 (1.6%)
	Casual with white coat	98 (10.6%)	680 (34.0%)	138 (17.0%)	435 (8.3%)	1351 (15.1%)
	Scrubs	176 (19.1%)	162 (8.1%)	203 (25.0%)	594 (11.4%)	1135 (12.7%)
	Scrubs with white coat	404 (43.8%)	444 (22.2%)	285 (35.0%)	1600 (30.7%)	2733 (30.5%)
	Formal	17 (1.8%)	26 (1.3%)	20 (2.4%)	346 (6.6%)	409 (4.6%)
	Formal with white coat	189 (20.5%)	660 (33.0%)	129 (15.9%)	1964 (37.6%)	2942 (32.8%)
	Business suit	14 (1.5%)	9 (0.4%)	5 (0.6%)	212 (4.1%)	240 (2.7%)
Surgeon	Casual	32 (3.5%)	13 (0.6%)	17 (2.1%)	37 (0.7%)	99 (1.1%)
	Casual with white coat	85 (9.2%)	238 (11.9%)	44 (5.4%)	179 (3.4%)	546 (6.1%)
	Scrubs	289 (31.2%)	942 (47.1%)	345 (42.6%)	2224 (42.5%)	3800 (42.4%)
	Scrubs with white coat	302 (32.6%)	501 (25.0%)	272 (33.6%)	1202 (23.0%)	2277 (25.4%)
	Formal	37 (4.0%)	35 (1.8%)	17 (2.1%)	192 (3.7%)	281 (3.1%)
	Formal with white coat	155 (16.8%)	266 (13.3%)	108 (13.3%)	1102 (21.1%)	1631 (18.2%)
	Business suit	25 (2.7%)	6 (0.3%)	7 (0.9%)	291 (5.6%)	329 (3.7%)
Overall	Casual	20 (2.2%)	17 (0.9%)	46 (5.8%)	70 (1.4%)	153 (1.7%)
	Casual with white coat	94 (10.2%)	606 (30.3%)	136 (17.0%)	367 (7.1%)	1203 (13.5%)
	Scrubs	146 (15.8%)	203 (10.1%)	205 (25.6%)	390 (7.5%)	944 (10.6%)
	Scrubs with white coat	385 (41.7%)	436 (21.8%)	252 (31.5%)	1289 (24.8%)	2362 (26.5%)
	Formal	25 (2.7%)	26 (1.3%)	22 (2.7%)	448 (8.6%)	521 (5.9%)
	Formal with white coat	235 (25.5%)	707 (35.3%)	131 (16.4%)	2370 (45.7%)	3443 (38.6%)
	Business suit	18 (1.9%)	7 (0.3%)	8 (1.0%)	255 (4.9%)	288 (3.2%)

Importance, impact and appropriateness of physician attire and white coats

Respondent opinions were sought using a Likert scale in which a score of 1 indicated ‘strongly disagree’ and 5 ‘strongly agree.’ In response to the prompt ‘how my doctor dresses is important to me,’ mean scores were similar for Italy (3.55), Japan (3.51) and the USA (3.49) and lower for Switzerland (3.05) (p<0.05 for all three pairwise comparisons). In response to the prompt ‘how my doctor dresses influences how happy I am with the care I receive,’ mean scores for Italy were 2.92, Japan 3.22, Switzerland 2.47 and the USA 3.17 (p<0.05 for all pairwise comparisons except for Japan-USA). In response to the prompt ‘it is appropriate for a doctor to dress casually when seeing patients over the weekend,’ all countries differed with mean scores for Italy of 3.15, Japan 2.57, Switzerland 3.37 and the USA 3.27 (p<0.05 for all six pairwise comparisons).

With respect to perceptions of whether white coats should be worn by physicians in various settings, differences emerged. When considering whether physicians should wear a white coat when seeing patients in their office, mean scores for Italy were 3.92, Japan 3.59, Switzerland 3.27 and the USA 3.53 (p<0.05 for all pairwise comparisons except for Japan-USA). When asked if physicians should wear a white coat in the emergency department, mean scores for Italy were 4.06, Japan 3.05, Switzerland 4.02, and the USA 3.34 (p<0.05 for all pairwise comparisons except for Italy-Switzerland). When asked if physicians should wear a white coat in the hospital, all countries differed with mean scores for Italy of 4.16, Japan 3.57, Switzerland 3.89 and the USA 3.63 (p<0.05 for all six pairwise comparisons). In response to the prompt ‘doctors should always wear a white coat when seeing patients in any setting,’ all countries differed with mean scores for Italy of 3.56, Japan 2.99, Switzerland 2.82 and the USA 3.12 (p<0.05 for all six pairwise comparisons). These results are summarised in table 4 and online supplemental appendix E.

Table 4

Respondent opinions regarding importance, influence and appropriateness of physician attire and white coats

	Italy	Japan	Switzerland	USA	Total
How my doctor dresses is important to me.
Strongly disagree	60 (6.4%)	67 (3.3%)	110 (13.4%)	222 (4.2%)	459 (5.1%)
Disagree	87 (9.4%)	280 (13.9%)	151 (18.4%)	531 (10.0%)	1049 (11.6%)
Neither agree nor disagree	220 (23.7%)	430 (21.4%)	260 (31.8%)	1603 (30.2%)	2513 (27.7%)
Agree	410 (44.1%)	1031 (51.3%)	185 (22.6%)	2303 (43.5%)	3929 (43.4%)
Strongly agree	153 (16.4%)	202 (10.1%)	113 (13.8%)	641 (12.1%)	1109 (12.2%)
Mean*	3.55	3.51	3.05	3.49
How my doctor dresses influences how happy I am with the care I receive.
Strongly disagree	132 (14.3%)	124 (6.2%)	223 (27.3%)	334 (6.3%)	813 (9.0%)
Disagree	209 (22.6%)	396 (19.7%)	235 (28.8%)	851 (16.1%)	1691 (18.7%)
Neither agree nor disagree	250 (27.0%)	536 (26.7%)	171 (20.9%)	2088 (39.5%)	3045 (33.7%)
Agree	263 (28.5%)	812 (40.5%)	124 (15.2%)	1633 (30.9%)	2832 (31.3%)
Strongly agree	70 (7.6%)	138 (6.9%)	64 (7.8%)	384 (7.2%)	656 (7.3%)
Mean*	2.92	3.22	2.47	3.17
It is appropriate for a doctor to dress casually when seeing patients over the weekend.
Strongly disagree	81 (8.7%)	209 (10.4%)	104 (12.8%)	182 (3.5%)	576 (6.4%)
Disagree	213 (22.9%)	837 (41.7%)	139 (17.2%)	955 (18.1%)	2144 (23.7%)
Neither agree nor disagree	218 (23.4%)	613 (30.5%)	147 (18.2%)	1761 (33.3%)	2739 (30.3%)
Agree	326 (35.1%)	300 (15.0%)	189 (23.4%)	2047 (38.7%)	2862 (31.7%)
Strongly agree	92 (9.9%)	48 (2.4%)	230 (28.4%)	340 (6.4%)	340 (7.9%)
Mean*	3.15	2.57	3.37	3.27
Doctors should wear a white coat when seeing patients in their office.
Strongly disagree	20 (2.2%)	48 (2.4%)	108 (13.2%)	84 (1.6%)	260 (2.9%)
Disagree	47 (5.1%)	226 (11.2%)	132 (16.1%)	552 (10.4%)	957 (10.6%)
Neither agree nor disagree	139 (14.9%)	437 (21.7%)	170 (20.8%)	1698 (32.1%)	2444 (27.0%)
Agree	504 (54.1%)	1085 (54.0%)	251 (30.7%)	2361 (44.7%)	4201 (46.4%)
Strongly agree	221 (23.7%)	214 (10.7%)	157 (19.2%)	593 (11.2%)	1185 (13.1%)
Mean*	3.92	3.59	3.27	3.53
Doctors should wear a white coat when seeing patients in the emergency department.
Strongly disagree	15 (1.6%)	102 (5.1%)	47 (5.8%)	111 (2.1%)	275 (3.0%)
Disagree	36 (3.8%)	541 (27.0%)	56 (6.9%)	828 (15.6%)	1461 (16.2%)
Neither agree nor disagree	115 (12.3%)	623 (31.1%)	75 (9.2%)	1952 (36.9%)	2765 (30.6%)
Agree	480 (51.2%)	628 (31.3%)	294 (36.0%)	1973 (37.3%)	3375 (37.3%)
Strongly agree	291 (31.1%)	110 (5.5%)	343 (42.1%)	426 (8.1%)	1170 (12.9%)
Mean*	4.06	3.05	4.02	3.34
Doctors should wear a white coat when seeing patients in the hospital.
Strongly disagree	13 (1.4%)	45 (2.2%)	50 (6.1%)	65 (1.2%)	173 (1.9%)
Disagree	19 (2.0%)	236 (11.7%)	45 (5.5%)	401 (7.6%)	701 (7.7%)
Neither agree nor disagree	83 (8.8%)	441 (22.0%)	128 (15.7%)	1507 (28.5%)	2159 (23.9%)
Agree	509 (54.3%)	1114 (55.4%)	311 (38.2%)	2756 (52.1%)	4690 (51.8%)
Strongly agree	314 (33.5%)	174 (8.7%)	281 (34.5%)	560 (10.6%)	1329 (14.7%)
Mean*	4.16	3.57	3.89	3.63
Doctors should always wear a white coat when seeing patients in any setting.
Strongly disagree	23 (2.5%)	109 (5.4%)	179 (21.9%)	181 (3.4%)	492 (5.4%)
Disagree	119 (12.7%)	567 (28.2%)	164 (20.0%)	1140 (21.5%)	1990 (22.0%)
Neither agree nor disagree	269 (28.7%)	682 (33.9%)	202 (24.7%)	2147 (40.6%)	3300 (36.4%)
Agree	361 (38.5%)	550 (27.4%)	169 (20.7%)	1497 (28.3%)	2577 (28.5%)
Strongly agree	165 (17.6%)	103 (5.1%)	104 (12.7%)	326 (6.2%)	698 (7.7%)
Mean*	3.56	2.99	2.82	3.12

*Means calculated with scores of 1 assigned to ‘strongly disagree,’ 3 to ‘neither agree nor disagree’ and 5 to ‘strongly agree.’

Comparisons of patient preferences within countries

Similarities and differences were identified when comparing preferences within countries based on respondent sociodemographic characteristics. When examining respondent gender, men and women rated different types of physician attire similarly within their respective countries. The only significant difference was that men rated formal attire more highly than women in Switzerland (male composite score 6.2, female composite score 5.4, p=0.04) (online supplemental appendix F). When comparing respondents aged 65 years and older with those between 18 and 64 years, there were no significant differences in composite scores for all types of physician attire in both Italy and Switzerland. In contrast, when compared with the younger cohort, respondents aged 65 years and older rated casual attire, formal attire, formal attire with white coat and business suit more highly in both Japan and the USA. When compared with the younger cohort, respondents aged 65 years and older rated casual attire with white coat and scrubs more highly in Japan, a finding that was not significant in the USA (online supplemental appendix G). There was no association between respondent preferences for physician attire and number of physicians seen in the prior year.

Discussion

In this international, multicentre, cross-sectional study, we report preferences of 9171 patients for physician attire across a variety of geographic regions, clinical contexts, physician types and patient sociodemographic characteristics. We found that the highest rated form of physician attire differed across countries, but that all most strongly preferred a white coat with any attire. Respondents from the USA more strongly preferred all types of formal attire compared with those from Italy, Japan and Switzerland. All countries more strongly preferred scrubs-based attire for emergency department physicians and surgeons. Taken together, these findings suggest that how a physician dresses has importance that varies around the world.

Our study adds to the existing literature by demonstrating that patients harbour expectations of how their physicians dress and that these expectations depend on sociocultural norms, context and patient factors. In some clinical care contexts, preferences vary substantially. In others, they are nearly universal such as those for emergency department physicians and surgeons wearing scrubs-based attire. With some exceptions, patients tended to dislike extremes in attire such as casual or business suit. Finally, it was very common for patients to prefer their physicians wear a white coat, a historically traditional aspect of the physician’s uniform and what is often considered a symbol of the profession.24 This was particularly evident when patient preferences for the underlying form of attire were split (eg, primary care and hospital physicians).

Other studies exploring patient perceptions for physician attire have yielded a diverse and often conflicting array of findings, most of which are complicated by different measurement tools and outcomes. Consistent with our results, numerous studies across continents have identified a clear patient preference for white coats.6 7 10 12 14 23 25–41 However, some studies reveal no significant preferences,42–45 and others indicate that the white coat may even cause higher levels of tension in patients.44 Some studies have shown that physician attire carries little importance with patients,46–50 whereas others have shown it has a substantial impact on the patient experience,30 51 congruent with our results. Literature differs on whether preferences for the white coat change after patients are educated about potential risk of microbial transmission, with some studies showing decreased preference14 52 and another showing no change.35 Studies examining attire in countries with bare-below-the-elbow policies have indicated near universal disdain for this infection prevention measure.27 35 Some studies have shown preference for different forms of attire such as scrubs (eg, specific circumstances such as gastroenterology suites18 53 and emergencies5) and informal attire,54 and some have revealed no specific patient preferences.52 55 56 Five studies noted that patient perceptions of compassion, professionalism and credibility were not associated with a physician’s dress.25 32 57–59 Finally, some studies have demonstrated that attire is more important to patients who are older,34 51 60 a finding we noted in Japan and the USA.

Studies conducted around the globe have repeatedly demonstrated that context is crucial when considering nonverbal cues like physician dress. Patient viewpoints are associated with a variety of factors such as type of care delivered, type of physician and even time of day. In one example, Switzerland has a defined healthcare uniform of white scrubs and white coat.20 This relatively unique phenomenon likely caused patients in Switzerland to expect this form of attire and thus strongly prefer it to other forms. In another example from the USA, parents of children being evaluated in the paediatric emergency department were more likely to prefer physicians wearing scrubs but only if their children were experiencing a surgical emergency.46 Likewise, in that same study, parents who visited the emergency department during the day shift preferred formal attire, whereas those who visited during the night preferred less formal attire.46 Finally, preferences have also previously been shown to deviate from cultural norms or established national dress.11 13 30 38 For instance, patients in family medicine clinics in Saudi Arabia were more likely to adhere to medical recommendations and return for subsequent care if the physician was dressed in Western garb60; yet this same population was significantly more willing to discuss personal issues such as psychological problems with a physician wearing Saudi national dress.60 This finding of preferences that varied based on topic of conversation was noted in other studies as well.9 10

A number of strengths distinguish our study from others that have previously investigated patient preferences for physician attire. To our knowledge, this study of over 9000 participants is the largest examination of opinions on physician dress to date. We employed a standardised survey instrument which allowed direct comparisons across diverse geography and contexts. Randomisation of photograph sequence and instrument delivery reduced the risk of ordering, priming and anchoring bias. We also used photographs containing physician models with identical postures, facial expressions, lighting and background, all of which limited the confounding associated with previous studies using models of different backgrounds and appearances.16–18 51 61 Finally, our findings have important policy implications for physician dress code in different care settings and areas of the world.

Our study also has limitations. Our physician models were young, slender and either Caucasian or Asian, and as such were not representative of the various sociodemographic characteristics of physicians. Likewise, purposeful differences among survey instruments, including white scrubs instead of blue scrubs in the Switzerland survey and physician models of Japanese descent in the Japan survey, were introduced to ensure relevance. Our study over-represented the USA more so than Japan and the European countries, which could have contributed to skewed results and greater power in any comparison with the USA. This was particularly evident when examining attire for hospital physicians, for example, in which the highest preference for formal attire with white coat was driven by US respondents. Despite large sample sizes in Italy and Switzerland, only one clinical site was represented in each of these countries, and this may not fully represent the country. When feasible from our convenience sampling methodology, we surveyed multiple clinical sites, because this approach likely achieved better representation of patients’ preferences for different forms of attire in the respective countries. We did not obtain results from other regions including Africa, Australia, the Middle East and South America, which could have contributed noteworthy input. Countries yielded different arrays of respondent sociodemographic characteristics such as age and education, which led to disproportionate representation among some groups. The survey instrument used Likert scales with predefined categories which may not accurately reflect nuanced patient opinions, and the clinical relevance of small but significant differences in these scales is unknown. The instrument did not capture or explore other elements of etiquette-based patient–physician interaction62 such as introductions and smiles,17 18 26 36 45 which are known to be paramount for ensuring effective healthcare relationships. Our study did not compare the relative impacts of physician attire with these and other factors known to influence the patient–physician relationship such as health literacy,63 communication skills64 65 and respect for patient autonomy.64 Finally, the data from several of the individual country-specific studies have been previously published. However, this study is the first instance in which all data are compiled to allow for cross-national comparisons.

In conclusion, the effects of physician attire on the patient experience are complex and multilayered. Our findings suggest that one-size-fits-all physician attire policies which extend to all healthcare specialties and contexts are unlikely to reflect the desires and expectations of patients. Instead, our nuanced results that harness direct patient preferences may be used to inform local, regional and national healthcare policy-makers and leaders in their efforts to define physician uniforms. Given that preferences vary, a tailored approach should be sought that matches attire with acuity, setting and context. This approach is most likely to cultivate the patient–physician relationship and in turn enhance patient satisfaction, trust, confidence and health outcomes.