Provider attire has come under scrutiny in the more recent medical literature. Epidemiologic data have shown that lab coats, ties, and other articles of clothing are frequently contaminated with disease-causing pathogens including methicillin-resistant Staphylococcus aureus , vancomycin-resistant enterococci, Acinetobacter species, Enterobacteriaceae, Pseudomona s species, and Clostridium difficile.1 Clothing may serve as a vector for spread of these bacteria and may contribute to hospital-acquired infections, increased cost of care, and patient morbidity. Prior to February 2015, the dermatology service line at Geisinger Medical Center in Danville, Pennsylvania, had followed a formal dress code that included white lab coats (white coats) along with long-sleeve shirts and ties/bowties for male providers and blouses, skirts, dress pants, and dresses for female providers. After a review of the recent literature on contamination rates of provider attire,2 we transitioned away from formal attire to adopt fitted, embroidered, black or navy blue scrubs to be worn in the clinic (Figure). Fitted scrubs differ from traditional unisex operating room scrubs, conferring a more professional appearance.
Limited research has shown that dermatology patients may have a slight preference for formal provider attire.2,3 In these studies, patients were shown photographs of providers in various dress (ie, professional attire, business attire, casual attire, scrubs). Patients preferred or had more confidence in the photograph of the provider in professional attire2,3; however, it is unclear if dermatology provider attire has any measurable effect on overall patient satisfaction. Patient satisfaction relies on a myriad of factors, including both spoken and unspoken communication skills. Patient satisfaction has become an integral part of health care, and with an emphasis on value-based care, it will likely be one determining factor in how providers are reimbursed for their services.4,5 In this study, we investigated if a change from formal attire to fitted scrubs influenced patient satisfaction using a common third-party patient satisfaction survey.
Patient Satisfaction Survey
We conducted a retrospective cohort study analyzing 10 questions from the care provider section of the Press Ganey third-party patient satisfaction survey regarding providers in our dermatology service line. Only providers with at least 12 months of survey data before (study period 1) and after (study period 2) the change in attire were included in the study. Mohs surgeons were excluded, as they already wore fitted scrubs in the clinic. Residents also were excluded, as they are rapidly developing their patient communication skills and may have a notable change in patient satisfaction over a 2-year period.
The survey data were collected, and provider names were removed and replaced with alphanumeric codes to protect anonymity while still allowing individual provider analysis. Aggregate patient comments from surveys before and after the change in attire were digitally searched using the terms scrub, coat, white, attire, and clothing for pertinent positive or negative comments.
We compared individual and aggregate satisfaction scores for our providers during the 12-month periods before and after the adoption of fitted scrubs. The primary outcome was statistically significant change in patient satisfaction scores before and after the institution of fitted scrubs. Secondary outcomes included summation of patient comments, both positive and negative, regarding provider attire, as recorded on satisfaction surveys.
Overall survey scores and scores on individual survey items were summarized using mean (SD), median and interquartile range, or frequency counts and percentage, as appropriate. The overall satisfaction score and responses to individual survey items were compared using Mantel-Haenszel or Pearson χ2 tests, as appropriate.
Assuming an equal number of surveys would be completed during study periods 1 and 2, an average (SD) satisfaction score of 95.4 (15), we calculated that as many as 2136 surveys would be needed to conclude satisfaction scores are the same for equivalence limits of −1.9 and 1.9 (a 1% difference). As few as 352 surveys would be needed to conclude satisfaction scores are the same for equivalence limits of −4.7 and 4.7 (a 5% difference). Sample size calculations assume 80% power and a significance level of 0.05. Comparison of responses for study periods 1 and 2 were made using the Mantel-Haenszel χ2 test.
Because more than 80% of respondents selected very good for each question, the responses also were treated as dichotomous variables with a category for very good and a category for responses that were lower than very good (ie, good, fair, poor, very poor). Responses of very good versus less than very good were compared for the study periods 1 and 2 using the Pearson χ2 test.
Two versions of an overall score were analyzed. The first version was for patients who responded to at least 1 of 10 survey items. If responses to all the items were very good, the patient was assigned to the category of all very good. If a patient answered any of the questions with a response less than very good, he/she was categorized as at least 1 less than very good. The second version was for patients who responded to all 10 survey items. If all 10 responses were very good, the patient was assigned to a category of all very good. If any of the 10 responses were less than very good, he/she was categorized as at least 1 less than very good. Differences between study periods for both score versions were tested using the Pearson χ2 test.