Learning Culture and Other Factors Affecting the Adoption of Electronic Medical Records in a Tertiary-Care Teaching Hospital

I was pleased to serve as a Member of the Ph.D. Doctoral Committee for Dr. Virgina Chavez. I encourage all of my doctoral students to consider publication of their rssearch results in a peer-reviewed, scholarly journal. Her timely research relating to electronic medical records was published in a recent on-line edition of the AAMA EXECUTIVE (American Academy of Medical Administrators).
Robert E. Hoye, Ph.D. FAAMA, FRSH
AAMA State Director for Kentucky


Learning Culture and Other Factors Affecting the Adoption of Electronic Medical Records in a Tertiary-Care Teaching Hospital

Virginia D Chavis, PhD
Senior Project Analyst
Intel Corporation
Chandler, AZ

Florence-Betty C. Roque, RN, BSN, MSN, ND, CPNP
Primary Care Provider
Rio Grande Medical Group
Deming, NM

Katherine Kenny, DNP, RN, ANP-BC, CCRN
Clinical Associate Professor
Arizona State University
Phoenix, AZ


Background
One of the highest priorities in healthcare reform is to change a paper-based culture to one that relies on electronic record keeping. The shift from paper to electronic files offers a different way for health professionals to think about patient needs. Improvements in the exchange of medical information may lead to better patient care and innovative ways to treat diseases and disorders. Using electronic medical records (EMR) in hospitals, medical centers, and private practices requires system implementation managers to diagnose the level of workers’ resistance to technology and then decide how best to motivate them to use it.
Medical record reform research (Margalit et al., 2006; Menachemi et al., 2007; Menachemi, Hikmet, Stutzman, & Brooks, 2006; Nowinski, Becker, Reynolds, Beaumont, Caprini, Hahn, et al., 2007) captures the impact technology acceptance has on various organizational factors. Accounting for these factors, many EMR implementations continue to fail. A leading cause is the information technology-centric approach, one that overlooks how health information technology (HIT) will affect the organization’s structures and processes (Berg, 2001). Other challenges, such as identifying professionals who will embrace the new technology and identifying the system features that become objects of resistance, also contribute to failure. To reduce the risk of product implementation failure, EMR should “be conceived as organizational development” (p. 147) and not a “mere” technical project (p. 148). An organizational development (OD) approach puts the emphasis on core business processes rather than the technology functions. Information and technology usage and barriers to adoption are vital elements addressed in an OD approach. The approach also can help manage organizational learning culture changes and IT capabilities.
Limited literature exists that examines the role of learning mechanisms. In the past five years, the amount of literature that keeps abreast of EMR implementation factors increased (Anderson, 2007; Anderson & Balas, 2006; Baker, Persell, Thompson, Soman, Burgner, Liss, et al., 2007; Croll & Croll, 2007; Christensen & Grimsmo, 2005; Feldstein et al., 2006; Lapointe & Rivard, 2006). Analysis of the literature showed that little formal data were available to understand the impact technology has on organizational learning. It is therefore difficult to understand if an organizational learning model is an adequate approach in healthcare settings. The literature review also showed that most EMR implementations are managed as information technology (IT) projects rather than using relevant and reliable organizational development strategies. The empirical insight also identifies common barriers to HIT adoption and describes common attitudes towards a healthcare learning environment. Unfortunately, innovation deployment often includes implementation strategies not congruent with the complex professional roles and relationships in medical environments. Moreover, a limited number of studies have significantly examined technology adoption among physicians, nurses and other medical specialists who care for patients in a multi-disciplinary medical setting.
Given the gaps in literature, the purpose of this IRB-approved quantitative survey research was to identify and understand technology adoption barriers by examining technology-usage behavior (e.g., use of EMR), technology adoption factors, and employees’ attitudes towards organizational learning. It also aimed to assess if a significant relationship exists between technology adoption and target variables, such as age, tenure with the organization, and overall level of education. We examined the relationship between technology adoption and principles of learning organization by investigating medical professionals’ perceptions of the learning culture of the research site and other barriers to EMR adoption. We were interested in knowing if a correlation difference existed between technology adoption scores for a learning mechanism and the learning organization scores, if a significant difference existed in the technology adoption scores for medical professionals who have prior EMR experience compared to those with no experience, and if a correlation difference existed in demographic characteristics of medical professionals and technology adoption scores. Data on the barriers to EMR use among medical staff was used to examine these correlations. Lastly, We presented data on learning organization score, which may serve as a benchmark for understanding which learning dimensions and human factors best influence the widespread adoption of an EMR.
Methods
Survey Sample
We selected all employed medical professionals at SJHMC Hospital and Medical Center (SJHMC), in Phoenix, Arizona. Medical professionals were defined as registered nurses (RNs), physician (including employed residents), and other medical professionals such as pharmacists and physical therapists. Using Sample Size Calculator by Creative Research Systems®, sample size and power calculations were determined. Based upon a medical population size of N = 1576, a confidence interval of 5% and a confidence level of 95%, the minimal sample size needed was 309. We are confident that all members of the population would have answered that learning environment influenced technology adoption.
Survey Design and Administration
Surveys were completely anonymous. It was publicized in the hospital’s newsletter and nursing huddles. The study was conducted during two periods: October 2009 – November 2009 and 2 weeks in January 2010. For the first collection, a survey along with a cover letter was distributed to 376 faculty physicians and residents, as well as 1,200 nurses, including mid-level practitioners. Paper versions of the survey were handed out to nurses and residents, and electronic surveys were emailed to faculty physicians. All participants, regardless of distribution method, had the option to either return the completed survey in interoffice mail or complete the survey online. Nonmedical staff such as janitors, volunteers, and operational staff (e.g., human resources, legal, and information technology), contingent medical staff, and administrative medical staff were excluded.
A follow-up announcement reminding staff of the study was sent two weeks after the initial invitation. We tracked respondents by hospital units. For the second period, nonrespondents were identified using the unit categories. These groups were handed a second survey package, which included the cover letter, survey, and return envelope.
The technology and Learning Organization Questionnaire was a two-part, self-administered questionnaire. Usage permissions were obtained prior to data collection. The first part, the technology assessment, captured general information technology use, EMR use, and barriers to EMR use. The instrument was developed by Menachemi et al. (2006) and was adopted with some minor modifications. For example, the word “practice” was changed to “center or clinic.” The second portion of the questionnaire came from the Dimensions of the Learning Organization Questionnaire® (DLOQ-A) developed by Yang (2003). It is a validated questionnaire composed of seven discipline areas that measure “changes in organizational learning practices and culture as perceived by the employees” (Marsick & Watkins, 2003, as cited in Dymock & McCarthy 2006, p. 528). The organization adoption score was correlated to organizational learning scores and other variables.
Participants completing the questionnaire online were required to give consent. Once the consent was electronically acknowledged, the participant gained access to the questions. Participants that completed the paper version returned the questionnaire through interoffice mail in a preaddressed envelope. All mailed responses were entered into the electronic database, and data was verified and cross-checked by an independent statistician. In addition, SJHMC Hospital and Medical Center Institutional Review Board (IRB approval number - 08BN042) and Walden University (IRB approval number - 09-29-09-0287953) evaluated and approved this study.
Statistical Methods
Analyses included standard descriptive and inferential statistics. Descriptive statistics was used to describe respondents’ characteristics such as age range, gender, years of experience as a nurse or physician, and years of employment at SJHMC. Learning and technology adoption and significances were determined through t test analysis. Inferential statistical correlation analysis was performed to validate whether a correlation or association existed between two the types of scores); two-sample t tests compared the means of the two populations, and ANOVA. JMP® software was used for analysis, and significance was considered at the p < 0.05 level. With survey research, response bias is a possibility. According to Menachemi (2006), surveys are a valuable method for gathering information for a geographically diverse population, especially when face-to-face interviews are not possible because of cost and/or time constraints. Surveys of physicians also typically yield lower response rates than other healthcare professionals. To attempt to correct for nonrespondent bias physician respondents were compared to nurse respondents, and the return rates were the same. The response rate for nurses was 17.4% and for physicians it was 17.3%. From these numbers, the percentage of nurses that responded equaled the percentage of physicians that responded. Surveys were distributed to all clinical units; however, demographic profiles of the clinical units are not known. Therefore, it is not possible to correct for nonrespondent bias within the two groups. The study conducted by Menachemi and Brooks (2006), which used the same technology assessment survey, compared responders and nonresponders with respect to known demographics and found “no significant evidence of bias was detected even after employing common techniques used to identify response bias” (p. 85). Results Demographics Study participants included registered nurses (RNs), physician (including employed residents), and other medical professionals such as pharmacists and physical therapists. Of the 337 surveys returned (a 21.4% participation rate), 7 respondents were excluded, 69 respondents (20.9%) reported themselves as physicians, and 220 (66.7%) reported their position as nurse. The final return rate was 21%. The percentage of physicians that responded was equal to the percentage of nurses that responded. Demographic characteristics of the respondents are shown in Table 1. One nonemployed physician completed the survey, and those responses were removed from analysis. In addition, six administrative support staff members that would not use an EMR were excluded. Missing demographic answers were noted on the specific questions and the total respondents revised for that question. Respondents that did not answer or partially answered questions 10 and 17 were excluded from data analysis. The number of respondents for this analysis was adjusted to reflect missing responses. Table 1 Participants’ Demographic Characteristics Category Physicians Nurses Other or unknown Age range: < 35 36 (12%) 79 (26.3%) 2 (0.7%) 35-50 22 (7.3%) 91 (30.3%) 11 (3.7%) > 50 10 (3.3%) 46 (15.3%) 3 (1.0%)

Gender:
Male 35 (11.7%) 26 (8.7%)
Female 33 (11.0%) 189 (63.2%) 16 (5.4%)

Mean yrs. At SJHMC 4.6 (<1 – 33) 8.7 (<1 - 39) 8.3 (<1 - 26)

Mean yrs. since graduation 10.3 (<1 – 48) 12.8 (<1 - 41) 13.2 (<1 - 27)




Descriptive analysis of the demographic data consisted of frequencies and percentages of each variable. The majority of the participants were female (63.2%). More people reported their age range as between 35 and 50 (30.3%).
Barriers to the Use of EMRs
To compare barriers among SJHMC employed medical respondents, chi-square analysis was used, and significance was considered at the p < 0.05 level. Differences existed among physicians, nurses, and other medical staff (see Table 2). For example, no time to learn how to use such a system barrier was dramatically higher for nurses (53.8%) than physicians (34.7%) and other medical staff (18.2%). Another notable differences between nurses and the other groups were system difficulty (50.6% vs. 30% for physicians and 9.1% for other medical staff). Patient confidentiality ranked at 40% for both non-nurse groups, but nurses ranked it at 62.1%. The rank for patient resistance was 43.8% and was considerably lower by physicians (14%) and other medical staff (11.1%). Nurses, more frequently than the other two medical groups, indicated more barriers to using EMR.