The design for this study was cross sectional and responses represent the reports of gastroenterologist members of the American College of Gastroenterology using a survey instrument administered either on-line via email or by hard copy sent through the postal mail between September 2010 and March 2011. All study-related materials were pilot tested among a sample of attending gastroenterologists, residents, and fellows in the Division of Digestive and Liver Disease at Columbia University Medical Center (CUMC). Materials were modified accordingly and approved by the CUMC Institutional Review Board prior to implementation.
The sampling frame was the full membership list supplied with permission by the organization and included a total of 10, 228 members. Those with missing information, non-MDs, non-US residents, and inactive members were excluded, resulting in 6, 777 active members. From this list, a sample of 20% (n= 1, 355) of physicians (MD's and DO's) were selected at random for inclusion in this study. One identical entry was removed, and two members who participated in a pilot test were also removed. Through survey responses and telephone follow up, an additional 353 participants were found to be ineligible (deceased, unable to locate or relocated out of the country, and retired) and were removed for a total of 999 active gastroenterologist members surveyed. Of all eligible gastroenterologists who were sent surveys, 288 responded (29%).
A cover letter accompanying the survey instrument detailed the purpose of the study, explained the selection process, assured that responses would be kept confidential, and urged the respondent to complete the survey and return it in a timely fashion. The on-line version of the survey was created in Qualtrics™, and a link was e-mailed to each participant. After the initial e-mail, follow up emails were sent at two and four weeks. In addition, a postal mailing was completed with hard copies of the cover letter and survey. After the initial postal mailing, a follow up mailing was sent to non-respondents approximately four weeks later.
The survey was comprised of four sections. Sociodemographic characteristics included age, gender, race/ethnicity, country of medical school, specialty/board certification, and years of experience performing colonoscopy. Practice characteristics questions consisted of geographic location and setting (urban, suburban, and rural), teaching hospital affiliation, practice type (private and hospital/university), number of colonoscopies performed per week, and self-reported rate of suboptimal bowel preparations encountered weekly (<10% or ≥10%).
To ascertain barriers to suboptimal bowel preparation, participants were asked about their agreement with a series of statements related to barriers to bowel preparation instruction communication in their practice (6 items). Items based on key informant input and a review of the literature included limited time to discuss information, the volume and complexity of information, lack of reimbursement for patient education, lack of patient educational materials written in languages other than English, and lack of staff to communicate instructions to patients. Participants were also queried about perceived patient barriers using yes/no responses. Patient barriers included not understanding the importance of following the bowel preparation instructions thoroughly, having problems with altering their usual diet, confused about which foods were permissible, unable to tolerate the full course of the purgative, lack of translated/culturally sensitive written instructions, and having problems related to the preparation such as the duration and convenience of the bowel preparation regimen and palatability of the purgative. Scores to each barrier measure (practice-related and perceived patient barriers) were summed to obtain an aggregate score.
Descriptive statistics, including frequencies and percentages, and Pearson’s Chi square test of association, were determined. We calculated internal consistency reliability for practice barriers (Cronbach’s alpha 0.78) and for perceived patient barriers (Cronbach’s alpha 0.63). Mean and standard deviation were calculated and Student’s t test was used to assess differences in means by self-reported proportion of suboptimal bowel preparations per week (<10% vs. ≥10%) and level of perceived patient barriers (low <4 and high ≥4). We used multivariate analysis logistic regression to identify factors predictive of higher perceived patient barriers. All analyses were performed using SAS version 9.2 (SAS Institute, Cary, NC).