Electronic health records: Use, barriers and ... - Wiley Online Library

8 downloads 0 Views 76KB Size Report
2Professor, Department of Health Policy and Management, Harvard School of Public Health; ... Women's Hospital; Harvard Medical School, Boston, MA, USA.
Journal of Evaluation in Clinical Practice ISSN 1356-1294

Electronic health records: Use, barriers and satisfaction among physicians who care for black and Hispanic patients Ashish K. Jha MD MPH,1 David W. Bates MD MSc,2 Chelsea Jenter MPH,3 E. John Orav PhD,4 Jie Zheng PhD,5 Paul Cleary PhD6 and Steven R. Simon MD MPH7 1

Assistant Professor, Department of Health Policy and Management, Harvard School of Public Health; Division of General Internal Medicine, Brigham and Women’s Hospital; The VA Boston Healthcare System; Harvard Medical School, Boston, MA, USA 2 Professor, Department of Health Policy and Management, Harvard School of Public Health; Division of General Internal Medicine, Brigham and Women’s Hospital; Harvard Medical School, Boston, MA, USA 3 Project Manager, Division of General Internal Medicine, Brigham and Women’s Hospital, Boston, MA, USA 4 Associate Professor, Division of General Internal Medicine, Brigham and Women’s Hospital; Harvard Medical School, Boston, MA, USA 5 Senior Programmer, Department of Health Policy and Management, Harvard School of Public Health, Boston, MA, USA 6 Professor, Harvard Medical School, Boston, MA, USA 7 Associate Professor, Department of Ambulatory Care and Prevention, Harvard Medical School, Boston, MA, USA

Keywords electronic health records, physician satisfaction, racial disparities Correspondence Dr Ashish K. Jha Department of Health Policy and Management Harvard School of Public Health 677 Huntington Avenue Boston MA 02115 USA E-mail: [email protected] Accepted for publication: 21 November 2007 doi:10.1111/j.1365-2753.2008.00975.x

Abstract Objectives Electronic health records (EHRs) are a promising tool to improve the quality of health care, although it remains unclear who will benefit from this new technology. Given that a small group of providers care for most racial/ethnic minorities, we sought to determine whether minority-serving providers adopt EHR systems at comparable rates to other providers. Methods We used survey data from stratified random sample of all medical practices in Massachusetts in 2005. We determined rates of EHR adoption, perceived barriers to adoption, and satisfaction with EHR systems. Results Physicians who reported patient panels of more than 40% black or Hispanic had comparable levels of EHR adoption than other physicians (27.9% and 21.8%, respectively, P = 0.46). Physicians from minority-serving practices identified financial and other barriers to implementing EHR systems at similar rates, although these physicians were less likely to be concerned with privacy and security concerns of EHRs (47.1% vs. 64.4%, P = 0.01). Finally, physicians from high-minority practices had similar perceptions about the positive impact of EHRs on quality (73.7% vs. 76.6%, P = 0.43) and costs (46.9% vs. 51.5%, P = 0.17) of care. Conclusions In a state with a diverse minority population, we found no evidence that minority-serving providers had lower EHR adoption rates, faced different barriers to adoption or were less satisfied with EHRs. Given the importance of ensuring that minorityserving providers have equal access to EHR systems, we failed to find evidence of a new digital divide.

Introduction Racial and ethnic minorities often fail to receive high-quality health care [1], although the reasons for these disparities are not fully understood. One potential explanation is that minorities and whites receive care in different settings, and providers that care for blacks and Hispanics may face more barriers to providing highquality care. A recent examination of the Medicare population found that just 22% of primary care physicians care for over 80% of elderly black Americans, and the physicians who care for blacks 158

were more likely to report difficulties providing high-quality care [2]. The concerns about inadequate quality are not limited to providers who care for minority populations [3,4]. Many clinicians and policy makers believe that increasing the use of electronic health records (EHR) will improve the quality of medical care [5–7] through reductions in medical errors [8], increased availability of real-time information and decision support [9]. However, given that disparities in health care might be due in part to where minorities receive their health care, policy makers worry that

© 2008 Blackwell Publishing Ltd. No claim to original US government works, Journal of Evaluation in Clinical Practice 15 (2009) 158–163

A.K. Jha et al.

differential adoption of health information technology (HIT) may worsen disparities [10]. If providers that care for large-minority populations have less financial means to afford these expensive systems, they may not be able to adopt them, which would hamper these physicians’ ability to provide high-quality care. Whether providers that care for large-minority populations are less likely to work in practices that adopt EHRs systems is not known. Furthermore, few empirical data are available regarding whether physicians who care for large-minority populations face greater financial barriers to adopting HIT in general or EHRs in particular. Therefore, we sought to answer three questions: First, do providers that care for large-minority populations have lower rates of adoption of EHRs? Second, do these providers face different barriers to adoption of EHR systems? Finally, is their satisfaction with EHR systems comparable with providers who predominantly care for white Americans?

Methods Overview We conducted a statewide survey of Massachusetts physicians. We asked them about the racial and ethnic composition of their patient panel, whether they had adopted an EHR, what barriers they face with respect to adoption or expansion of IT in their office practice, as well as their satisfaction with EHR systems. The details of the survey design and administration have been detailed previously [11] and are further described below.

Electronic health records

The survey also measured practice characteristics, including number of physicians, primary care versus specialty and number of visits per week. We asked each respondent to estimate the percentage of patients seen in a typical week who were of Hispanic ethnicity (Hispanic vs. non-Hispanic) and who were of one of five different race groups (whites, blacks, Asians, Native Americans or other). The survey also assessed the financial stability of the office practice, the availability of capital for investment in HIT, and the presence of financial incentives for the use of HIT. Respondents were asked about perceived barriers to EHR adoption and on characteristics of the office practice environment that may impede or enhance the adoption of this technology into office practices.

Survey administration A professional survey firm administered the survey between June and November 2005. The initial survey was sent by express mail with a $20 cash honorarium to sampled physicians. Subsequently, second and third mailings were sent to non-respondents, without remuneration. Between these mailings, multiple telephone contacts were attempted; 17 respondents (1%) completed the survey by telephone. A total of 94 physicians in the sample were deemed ineligible for the following reasons: 30 relocated to a different practice site; one was deceased; 62 had retired or closed their practice; and one had an address that was a corporate office, not a clinical practice.

Practice demographics Sample Using a database from a private vendor (Folio Associates, Hyannis, MA, USA), supplemented with data from the Massachusetts Board of Registration in Medicine, we identified physicians from all specialties practising in Massachusetts in spring 2005. After excluding physicians who were residents in training, retired or without direct patient care responsibilities, the total population of physicians was 20 227. These physicians practised in 6174 unique sites in Massachusetts. We drew a stratified random sample of 1921 practices. We oversampled hospital-based and larger practices, as well as practices in rural areas, to ensure that they would be adequately represented. We then randomly selected one physician from each practice for a survey. If we learned that the initially selected respondent/practice was unavailable, we randomly re-sampled a new physician or practice whenever possible. We re-sampled 137 physicians because the physician was no longer at the practice, retired or deceased; we did the same for 57 practices if the initially selected practice had closed. Thirty-seven could not be re-sampled, resulting in a final sample size of 1884 physicians.

Survey questionnaire We developed a questionnaire based on a systematic review of the literature regarding factors related to the adoption of EHRs in ambulatory practices. We cognitively tested the draft instrument with seven physicians. The final questionnaire included items designed to assess organizational characteristics and factors related to the use of EHRs.

© 2008 Blackwell Publishing Ltd. No claim to original US government works

Because we were concerned about the validity of physicians’ self-reported practice demographics, we examined the demographic make-up of the census tract in which the practice was located. We used the address of each practice to identify the census tract in which the practice was located using a linkage algorithm provided by Mapping Analytics (http://www.mappinganalytics. com). Then, we linked each practice’s census tract to the Census Bureau’s data on the demographics of that tract, including its percentage of residents who are blacks and Hispanics, percentage of residents with incomes below the poverty line, and the percentage of residents with a college degree.

Measurement of minority-serving practice We examined two sets of predictor variables. In the first set of analyses, we examined physicians’ self-reported percentage of patients seen in that practice that were black or the percentage who were Hispanic. We also examined a combined variable that was the sum of the percentage black and percentage Hispanic. The findings were qualitatively similar when we examined blacks and Hispanics separately and so for simplicity, we present the results of the analyses where we use the combined variable (sum of percentage of black patients and percentage of Hispanics). We performed a second set of analyses to address our concern that physicians may not be able to accurately report the percentage of patients in their practice who are minorities. Here, we examined the percentage of residents in the census tract of the practice who were black and the percentage who were Hispanic. Again, we also 159

Electronic health records

examined a combined variable that was the sum of the percentage black plus the percentage Hispanic in that tract. The two sets of predictor variables were moderately correlated. The correlation between a physician’s self-reported percentage of black patients and the percentage of black residents in that census tract was high (correlation coefficient 0.63), but it was only 0.37 for the measures of Hispanics and 0.50 for the combined measures of minorities (blacks plus Hispanics). Finally, we examined our primary predictor, the percentage of reported minorities in the practice, as both a continuous and a categorical variable. We examined the effect of dichotomizing the variable at different cut points, including using tertiles, and found that different cut points yielded similar results. Therefore, we chose to present results in which the physician practices are divided into three categories: those practices with relatively few minorities (less than 10%), moderate number of minorities (10– 39%) and those with a high fraction of minorities (40% or greater). We defined those physicians who reported that 40% or more of their panel was composed of minorities as ‘minority-serving’. We also examined each set of results by substituting the percentage of minorities in the census tract for a physician’s selfreported racial composition and the results were almost identical. Therefore, we present only the analyses using the physician reports of racial composition.

Measures of EHR adoption rates, barriers to adoptions and perceived benefits We examined three types of outcome variables. The first focused on whether the practices of minority-serving physicians and nonminority-serving physicians had comparable levels of EHR adoption. Our dependent variable was the response to the question that asked, ‘Does your practice have components of any EHR, that is, an integrated clinical information system that tracks patient health data and may include such functions as visit notes, prescriptions, lab orders, etc.?’ The next set of outcome variables focused on the barriers to EHR adoption and whether barriers differed between minority-serving physicians and other physicians. Finally, we examined whether minority-serving physicians had the same satisfaction with EHR systems and with their clinical practice in general compared with other physicians.

Statistical analysis Data were weighted by specialty, practice size, hospital affiliation and geography to make the sample representative of all Massachusetts physicians. We made bivariate comparisons using chisquared tests. We used logistic regression to analyse whether the percentage of self-reported minorities in physicians’ practices was a predictor of the adoption of EHRs, barriers to EHR use or perceived benefits of an EHR. In these models, we adjusted for practice characteristics that were potentially confounding variables: practice type, size, urban or rural location, and whether the practice was a teaching practice or hospital based. For most of the analyses, the unadjusted and the adjusted results were qualitatively similar. Therefore, we present rates of EHR adoption adjusted for the above characteristics unless the results between the unadjusted and adjusted rates were different (in which case we present both results). 160

A.K. Jha et al.

The study protocol was approved by the Partners Health Care Human Research Committee. The data were analysed using the SUDAAN 9.0 (RTI, Research Triangle Institute, NC, USA) statistical software package with appropriate corrections for the complex survey design.

Results Of 1884 physicians surveyed, 1345 returned completed surveys for a response rate of 71%. We excluded respondents who reported that they provided no ambulatory care (n = 157, 12%) or who did not report the racial make-up of their patient population (n = 48, 4%). If the respondents’ reported racial composition of patients totalled less that 90% or greater than 110%, we assumed the values were erroneous and excluded them from analysis (n = 8, 1%). Our final sample included 1132 surveyed physicians. A total of 185 respondents (16%) reported that at least 40% of the patients they saw in the previous week were black or Hispanic (high minority) while 558 respondents report that between 10% and 39% of the patients they had seen were minorities (medium minority). The rest were designated as low-minority practices. High-minority practices were more often located in urban settings, hospital based, affiliated with teaching hospitals, and more likely to have seven or more physicians in the practice (P < 0.01 for each comparison, see Table 1). In unadjusted analyses, physicians whose patient panels were more than 40% black or Hispanic were more likely to have some component of an EHR in their practice (31% vs. 18%, P = 0.048), a difference which become non-significant when we adjusted for practice location and size (27.9% and 21.8%, respectively, P = 0.46, see Table 2). Of practices that had adopted some component of an EHR, high- and low-minority practices had comparable levels of electronic visit notes (85.0% vs. 85.5%, P = 0.37), laboratory results management (88.6% vs. 84.7%, P = 0.86) and electronic prescribing (25.8% vs. 18.8%, P = 0.39). There were large differences in the use of laboratory order entry (56.9% among high-minority practices vs. 28.0% among low-minority practices, P = 0.04) and in computerized scheduling (62.5 vs. 54.5, P = 0.02, Table 2). High- and low-minority practices reported similar barriers to EHR implementation (Table 3). All physicians cited costs as a primary barrier to implementation, and this barrier did not vary by the percentage of minorities seen in that practice. Of all barriers examined, only one was significantly related to the proportion of minority patients: high-minority providers were less likely to be concerned about the privacy and security aspects of EHR adoption than other practitioners (47.1% vs. 64.4%, P = 0.01, Table 3). There were few differences in the perceived benefits of computers in health care among physicians in high- versus low-minority practices. The majority of physicians in each category of practice perceived EHRs to have a positive effect on the costs, quality and efficiency of care (Table 4). Physicians in low-minority practices were more likely to believe that EHRs had a negative effect on patient privacy than physicians in other practices (41.5% and 55.2%, respectively, P = 0.048).

Discussion We examined whether practices in Massachusetts with largeminority populations had lower levels of EHR adoption or faced

© 2008 Blackwell Publishing Ltd. No claim to original US government works

A.K. Jha et al.

Electronic health records

Table 1 Characteristics of practices by the percentages of African-American and Hispanic patients seen in a typical week

Practice type Primary Care Specialty/mixed Hospital-based Yes No Urban Urban Rural Teaching practice Yes No Practice size 1 physician 2 or 3 physicians 4–6 physicians 7 or more physicians

Low minority (0–9%) n = 389 (36%)

Medium minority (10–39%) n = 558 (48%)

High minority (ⱖ40%) n = 185(16%)

n

Weighted %

n

Weighted %

n

Weighted %

169 220

29 71

197 361

26 74

69 116

27 73

62 327

8 92

151 407

13 87

63 122

15 85

241 148

89 11

442 116

93 7

166 19

97 3

138 248

25 75

282 276

42 58

115 70

41 59

187 84 68 50

74 15 8 4

211 129 109 109

62 21 10 6

71 32 24 58

70 14 6 10

P-value 0.78 0.006