Understanding Practice: Factors That Influence Physician Hand Hygiene Compliance

Janet E. Squires, PhD; Stefanie Linklater, MSc; Jeremy M. Grimshaw, PhD; Ian D. Graham, PhD; Katrina Sullivan, MSc; Natalie Bruce, MScN; Kathleen Gartke, MD; Alan Karovitch, MD; Virginia Roth, MD; Karen Stockton, MHSc; John Trickett, BScN; Jim Worthington, MD; Kathryn N. Suh, MD


Infect Control Hosp Epidemiol. 2014;35(12):1511-1520. 

In This Article


Design and Participants

This was a descriptive qualitative study. All medical and surgical staff physicians and residents at a large Canadian tertiary care hospital at 2 campuses that are 8.4 kilometers apart were eligible to participate. A list of all eligible physicians (divided by campus and specialty) was obtained. From this list, 6 new participant lists were created: (1) staff physicians, medicine campus 1; (2) staff physicians, medicine campus 2; (3) staff physicians, surgery campus 1; (4) staff physicians, surgery campus 2; (5) residents medicine; and (6) residents surgery. Residents were not separated by campus since they worked equally across both campuses. A quasi-experimental sampling strategy was used to randomly select key informants from each list. The first key informant on each list was chosen at random, and the subsequent key informants were then selected according to regular intervals, known as periods. Selected key informants were informed of their selection via e-mail from a study investigator; individuals were instructed to contact the study research assistant if they were interested in participating.

Data Collection

We conducted semistructured key informant interviews. The TDF informed the development of the interview guide. Questions were open ended, and standardized prompts were available to the interviewer if needed. The interview guide was pretested with a medicine and surgery physician, resulting in some wording changes. All interviews were conducted by the study research assistant who was trained in interviewing skills, using mock interviews to ensure she was comfortable with the interview guide. Interviews were conducted by telephone between October and December 2012 and digitally recorded. Interviews were conducted until data saturation was achieved (ie, until no new themes or concepts emerged).[25]

Data Analysis

The recordings were transcribed verbatim and verified by the interviewer prior to analysis. Analysis occurred in 3 steps and was facilitated using NVivo 10 software.

Coding. Two team members trained in qualitative analysis, using thematic content analysis, independently coded the transcripts into the 14 TDF domains, meeting frequently to review their coding and seek consensus. First, the 2 team members read the first 4 transcripts to determine a coding scheme (comprised of codes, definitions of the codes, and examples of quotes that fall under the codes) by consensus, which was then used to analyze the remaining transcripts. Inter-rater reliability was calculated using Kappa statistics in NVivo and SPSS software.

Generation of Specific Beliefs. Specific beliefs were generated for each utterance (coded interview quote) in all TDF domains by 1 team member and double-checked for accuracy by a second team member. A specific belief is a collection of participant responses with a similar underlying theme that suggests a problem and/or influence on the target behavior.[26] Belief statements were initially written to be very specific for each code; later, similar belief statements were merged to form broader statements (ie, themes). The number of interviews (n = 42) in which each belief statement was mentioned was tabulated to create a frequency score for the belief statements. Belief statements were counted only once per interview, even if the statement was mentioned more than once in the interview.

Identification of Relevant Domains. Consistent with previous studies,.[26,27] TDF domains were classified as relevant (ie, of high importance) to physician hand hygiene compliance if there were (1) belief statements in the domain that had relatively high frequencies (ie, 5 or more physicians identified the belief in their interview), (2) conflicting beliefs in the domain (ie, where different physicians identified opposite beliefs; eg, aware of evidence linking hand hygiene to infection and not aware of evidence linking hand hygiene to infection; in case of conflicting beliefs, frequencies less than N = 5 were accepted), and (3) evidence of strong beliefs in the domain that may impact hand hygiene compliance (this was determined by consensus in a team meeting with the clinicians on the research team).


Ethical approval was obtained from the hospital's research ethics board (protocol 2012040801H).