136

© Schattauer 2015

Electronic Health Records and Patient Safety Co-occurrence of early EHR implementation with patient safety practices in primary care settings C. Tanner1; D. Gans4; J. White1; R. Nath2; J. Pohl3

1Michigan Public Health Institute, Okemos, Michigan, United States; 2Colorado Hospital Association, Center for Health Information and Data Analytics, Greenwood Village, Colorado, United States; 3The University of Michigan, School of Nursing, Ann Arbor, Michigan, United States; 4Medical Group Management Association, Englewood, Colorado, United States

Keywords Electronic health records and systems, safety culture, organizational change management, work- flows and human interactions, sociotechnical aspects of information technology

Summary Background: The role of electronic health records (EHR) in enhancing patient safety, while sub- stantiated in many studies, is still debated. Objective: This paper examines early EHR adopters in primary care to understand the extent to which EHR implementation is associated with the workflows, policies and practices that promote patient safety, as compared to practices with paper records. Early adoption is defined as those who were using EHR prior to implementation of the Meaningful Use program. Methods: We utilized the Physician Practice Patient Safety Assessment (PPPSA) to compare pri- mary care practices with fully implemented EHR to those utilizing paper records. The PPPSA measures the extent of adoption of patient safety practices in the domains: medication manage- ment, handoffs and transition, personnel qualifications and competencies, practice management and culture, and patient communication. Results: Data from 209 primary care practices responding between 2006–2010 were included in the analysis: 117 practices used paper medical records and 92 used an EHR. Results showed that, within all domains, EHR settings showed significantly higher rates of having workflows, policies and practices that promote patient safety than paper record settings. While these results were ex- pected in the area of medication management, EHR use was also associated with adoption of pa- tient safety practices in areas in which the researchers had no a priori expectations of association. Conclusions: Sociotechnical models of EHR use point to complex interactions between technology and other aspects of the environment related to human resources, workflow, policy, culture, among others. This study identifies that among primary care practices in the national PPPSA database, hav- ing an EHR was strongly empirically associated with the workflow, policy, communication and cul- tural practices recommended for safe patient care in ambulatory settings.

Research Article

C. Tanner et al.: EHR and Patient Safety

Correspondence to: C. Tanner Michigan Public Health Institute 2501 Jolly Road, Suite 180 Okemos, MI 48864 Email: [email protected]

Appl Clin Inform 2015; 6: 136–147 http://dx.doi.org/10.4338/ACI-2014-11-RA-0099 received: November 6, 2014 accepted: January 23, 2015 published: March 11, 2015 Citation: Tanner C, Gans D, White J, Nath R, Pohl J. Electronic health records and patient safety – co-oc- currence of early EHR implementation with patient safety practices in primary care settings. Appl Clin Inf 2015; 6: 136–147 http://dx.doi.org/10.4338/ACI-2014-11-RA-0099

For personal or educational use only. No other uses without permission. All rights reserved. Downloaded from www.aci-journal.org on 2015-08-29 | ID: 1001059516 | IP: 168.156.40.175

137

© Schattauer 2015

The 1999 IOM report To Err is Human presented evidence that errors and adverse events occur widely in health care [1]. It was widely expected that adoption of Electronic Health Records (EHRs) would substantially ameliorate threats to patient safety. More recently, models posit that the effect of EHRs on patient safety is mediated by multiple interacting factors [2, 3]. Specifically the Interactive Sociotechnical Analysis (ISTA) framework calls attention to the effects of EHR design, technical in- frastructure, and the work environment (workflows, policies, procedures) [2]. Our study contributes to the debate about the effect of EHR on patient safety through an examination of the empirical rela- tionship between EHR adoption and mediating aspects of work environment. Specifically the study utilized a national survey to compare practices with and without an EHR according to the following patient safety work environment domains: medications, handoffs and transitions, personnel qualifi- cations and competencies, practice management and culture, and patient education and communi- cation. We limited the study to pre-meaningful use (MU) era adopters in order to control for aspects of the adoption environment that may be related to monetary incentives. In other words we wanted to look at adoption that was motivated by anticipated intrinsic value of EHR.

1. Background Much of the patient safety research has focused on inpatient care, but there are significant safety concerns in ambulatory care as well [4, 5]. While hospital care can be more technologically complex, ambulatory care involves challenges with information exchange across multiple settings and in- herent risks when patients are expected to assume an administrative role in coordinating their own care [6, 7]. Ambulatory patient safety risks also affect a much larger population, as there are 300 times more patients seen in ambulatory settings than admitted to hospitals [8].

In 2011, the American Medical Association’s (AMA) Center for Patient Safety released a report summarizing ambulatory safety research from the decade following the IOM report. In this report, “Research in Ambulatory Patient Safety 2000–2010: A 10-Year Review,” the AMA outlined six pri- mary areas of potential harm to patients in the ambulatory setting: medication errors (such as pre- scriptions for incorrect drugs or incorrect dosages), diagnostic errors (such as missed, delayed, and wrong diagnoses), laboratory errors (such as switched or lost specimens and delays in communicat- ing test results or following up with patients), clinical knowledge errors (such as clinical tasks errors, misdiagnosis, and errors in treatment decisions), communication errors (such as delays in transfer- ring information, language barriers, and low health literacy), and administrative errors (such as fail- ure to protect patient information and errors with scheduling and management of patient records) [8, 9]. Among ambulatory settings, assessing patient safety in primary care is critical. For example, laboratories, imaging facilities and other diagnostic services are generally separate from the primary care practice, and communication of results to primary care practitioners and patients is subject to many sources of failure [6, 15, 16]. The same holds true for referrals and transitions of care that occur frequently in primary care.

When well-designed EHRs are properly implemented and effectively used, they offer the poten- tial to address many of these areas of concern, particularly those related to medication safety, diag- nostic errors, and communication issues. Computerized physician order entry (CPOE) has been shown to reduce medication-related errors [12]. Research suggests that proper implementation of interoperable health information technology (HIT) systems can enhance patient safety by establish- ing more effective communication methods [12]. EHRs can also enhance patient safety by detecting missed diagnoses, producing diagnostic error alerts to prevent misdiagnosis, and assisting the prac- titioner in gathering and synthesizing patient information [17].

However, new concerns have arisen about unintended consequences of EHRs, which may include dosing errors, failure to detect fatal illnesses, and delays in treatment [2, 3, 12, 18]. Several models posit that complex interactions between technical and non-technical factors influence the effects of EHR. Harrison et al. describe how unintended consequences may result from complex and recursive interactions between the following: the design of the EHR, the work environment, the technical and physical infrastructure, and interactions between these elements [2]. Further elaborating the rela- tionships between HIT and safety, Sittig and Singh present eight interacting domains: hardware and software computing infrastructure, clinical content of the HIT, human computer interface, people,

Research Article

C. Tanner et al.: EHR and Patient Safety

For personal or educational use only. No other uses without permission. All rights reserved. Downloaded from www.aci-journal.org on 2015-08-29 | ID: 1001059516 | IP: 168.156.40.175

138

© Schattauer 2015

workflow and communication, internal organizational features, external rules and regulations, and measurement and monitoring [3].

While interactions between policies, workflow, people and technology are posited to interact with EHRs, potentially leading to unpredictable outcomes, it is unknown how often EHRs may indeed undermine, rather than support, needed workflows, communications, policies, and other practices. The Physician Practice Patient Safety Assessment (PPPSA) database, described in a previous publi- cation, provides an opportunity to study the empirical association of EHR with the implementation of patient safety practices in six domains: medications, handoffs and transitions, invasive procedur- es, personnel qualifications and competencies, practice management and culture, and patient edu- cation and communication [19, 20, 21, 22]. The research team hypothesized that EHRs enable im- plementation of some patient safety workflows and practices – particularly in the areas of medi- cation management, handoffs and transitions, and patient communications – and would thus be empirically associated with those workflows and practices.

2. Methods Data presented in this paper pertain to primary care practices that submitted data to the Medical Group Management Association’s PPPSA benchmarking database between 2006 and 2010. This was done in order to focus on early EHR adopters as the research team believed that there would be a difference between practices that adopted EHR before the existence of widespread Meaningful Use incentives compared to practices that adopted later.

Practices were included that clearly identified their level of HIT adoption as “EHR-based” or “paper-based”. This classification was determined by responses to: “Describe how the health/medical records system stores information for the majority of patients served by the specific practice/lo- cation. If the specific practice/location uses multiple technologies, choose the system used for the majority of your patients’ medical records.” The 117 practices that answered “Paper medical records/ charts filed in record cabinet” were labeled as “paper-based practices” and the 92 that selected “An EHR that stores patient medical and demographic information in a database accessible by computer terminals or other electronic means…may also incorporate features of a document imaging manage- ment system (DIMS)” were classified as “EHR-based practices”. Practices that replied “other”, or re- ported using DIMS only, were excluded from the analysis.

The research team was affiliated with a regional extension center (REC) for EHR adoption, and utilized extension center staff – with exposure to a wide range of actual EHR utilization in a variety of settings – to rate each survey item as employing workflows that are: 1) substantially enabled by EHRs, 2) potentially enabled by EHRs but dependent on product and setting, and 3) unrelated to EHRs.

Two REC staff independently classified the items. In case of divergence, the item was discussed with REC and research staff and a consensus classification was assigned. Based on this classification, items in ▶ Table 1 are noted as to whether an empirical association is expected (yes, possible, or no, respectively). Responses for items within each section ranged on a five-point ordinal scale: A. Unaware of this issue or aware but there has been no activity to implement this item. B. This item has been formally discussed and considered, but it has not been implemented. C. This item has been partially implemented in the practice for some areas, patients, drugs, D. This item is fully implemented in the practice for some areas, patients, drugs, procedures and/or

staff. E. This item is fully implemented in the practice for all areas, patients, drugs, procedures and/or

staff.

The gamma statistic was computed to test for differences between paper-based and EHR-based practices on all survey items, excepting those related to surgery which would not apply to primary care settings. Gamma (γ) is a measure of rank correlation used for cross-tabulated data when both

Research Article

C. Tanner et al.: EHR and Patient Safety

For personal or educational use only. No other uses without permission. All rights reserved. Downloaded from www.aci-journal.org on 2015-08-29 | ID: 1001059516 | IP: 168.156.40.175

139

© Schattauer 2015

variables are ordinal. It is estimated by γ=(P−Q)/(P+Q). Analysis was performed using IBM SPSS statistics version 21.

3. Results

3.1 Sample Two-hundred-nine primary care practices responded to the PPPSA during the years 2006–2010. Of these, 117 used a paper medical records system and 92 used an EHR. A strong majority of the prac- tice respondents (70%) were part of a hospital integrated delivery system with almost all (99%) single specialty practices. The number of physician FTEs in these practices was primarily 10 or fewer (96%) and the same held true for non-physician providers. Practices were located across varied populations including non-metropolitan areas of under 50,000 population (27.6%), moderate sized metropolitan areas of 50,000–250,000 (27.6%), larger metropolitan areas of >250,000 to 1,000,000 and very large metropolitan areas of >1,000,000 (14.3%).

3.2. Data Overview ▶ Table 1 presents a comparison of responses from practices utilizing electronic vs. paper records across five patient safety domains that are relevant for primary care. Overall, primary care practices that were early EHR adopters were more likely to adopt a series of patient safety practices. In fact, of all items included in the study, there were no items in which paper-based practices out-performed practices using an EHR. There were some areas in which no substantive or statistically significant differences were present.

3.3 Medication Safety Of 16 medication-related patient safety practices, all but three (medication information for non- English speakers, warfarin tracking system, and pregnancy testing teratogen prescribing) had statis- tically significant and positive relationship to full EHR adoption.

3.4 Handoffs and Transitions The differences between EHR and paper-based practices were statistically significant for six out of 11 safety practices related to hand-offs and transitions. Tracking consultations and imaging test re- sults, as well as communicating about medications during transition care were associated with EHR use, but were not statistically significant. The following differences were statistically significant in favor of EHR sites: tracking of laboratory tests, communicating test results to patients, and having a process to learn and record new information from outside the practice. EHR impact on cross-setting communication was dependent on interoperability and data exchange. ▶ Table 2 presents additional data from the PPPSA showing that manual processes for consult tracking, lab ordering, and results tracking were widespread even in EHR-based settings.

3.5 Personnel/Qualifications/Competency Items in this section of the survey related to education of practice staff, competency assessment, and orientation of new practitioners and staff. Although these items do not have a clear connection to EHR use, all differences between practices were statistically significant.

3.6 Practice Management/Culture Twenty-two questions were asked in the domain of overarching management practices that indicate a culture of safety. Consistent with other results, practices with EHR outperformed those on paper across most items – 13 of which were statistically significant. As with other areas of the survey, there

Research Article

C. Tanner et al.: EHR and Patient Safety

For personal or educational use only. No other uses without permission. All rights reserved. Downloaded from www.aci-journal.org on 2015-08-29 | ID: 1001059516 | IP: 168.156.40.175

140

© Schattauer 2015

were no a-priori expectations for EHR-based practices to outperform paper-based practices on items such as: patients instructed in proper use of devices, or patients encouraged to share safety concerns. There was a strong emphasis on staff and patient safety in the EHR-based practices. In contrast, in an area where differences might have been expected – essential patient information rec- orded on a separate intake form in such a way that it is clearly evident and easily accessible – com- pliance was equally high (nearly 96% adoption) in both types of settings while compliance was low in both settings on using current literature on adverse events for practice improvement.

3.7 Patient Education and Communication Differences between EHR and paper-based practices were statistically significant in seven of the 13 items. Seventy-nine percent of practices with EHR also provide translation services for individuals for whom English is a second language or who have a hearing impairment – compared to 63% of practices without (p=0.001). Early EHR adopters did not report a statistically significantly higher level of identification of patients with chronic disease and providing special education and monitor- ing services.

4. Discussion This paper took advantage of a large national dataset to investigate the empirical association be- tween early EHR adoption and implementation of workflows, policies, and practices that promote patient safety. Our findings confirm that among the primary care settings studied, EHR adoption was associated with adoption of numerous patient safety practices, including those beyond the ex- pected relationship to an EHR. It appears that these early adopters were specifically adopting EHR as part of a larger strategy: an overall emphasis on patient safety. Since sociotechnical models posit that the ultimate effect of EHR on safe patient care is mediated in part through complex interactions with workflow, policies, and other aspects of the environment, it is good news that the empirical as- sociation between EHR use and patient safety workflows and policies was found to be strongly posi- tive. It is unknown whether later adopters of EHR – motivated in part by MU financial incentives and penalties [23] – are adopting EHR along with patient safety workflows.

As expected the most consistent differences between EHR and paper-based settings were in the areas of medication documentation and decision support. However, in some areas where a com- puter might be expected to offer the most advances over paper – such as in chronic disease manage- ment – significant differences were absent. This could be attributed to the fact that registry func- tions, preventive care reminders, integrated search, and other functions are more recent advances in EHR software.

This paper reports the empirical association of EHR use with a broad range of patient safety prac- tices but does not address causation. It is possible that other underlying factors, such as being part of a hospital system that is more heavily resourced, are related to both EHR and broad patient safety practice adoption. Moreover, safe patient care can be offered with or without EHR, as very high re- sponses on a number of items indicate. Finally, an important limitation is that the PPPSA is based on survey data rather than objective verification of having the patient safety workflows, policies and practices in place.

Overall, however, these data support expectations that EHR adoption enhances patient safety generally. On nearly all items, practices with an EHR scored higher. This is particularly noteworthy in relation to patient communications, an area that some posit to suffer when an EHR is adopted.

5. Conclusions Sociotechnical models of EHR use point to complex interactions between technology and other as- pects of the environment related to human resources, workflow, policy, and culture, among others. This study identifies that among primary care practices in the national PPPSA database, having an EHR was strongly empirically associated with the workflow, policy, communication and cultural

Research Article

C. Tanner et al.: EHR and Patient Safety

For personal or educational use only. No other uses without permission. All rights reserved. Downloaded from www.aci-journal.org on 2015-08-29 | ID: 1001059516 | IP: 168.156.40.175

141

© Schattauer 2015

practices recommended for safe patient care in ambulatory settings. This association held across do- mains measured: medication safety, handoffs and transitions, practice management and culture and patient education and communication. The consistent and strong empirical association of EHR use with adoption of other safety practices suggests that early adopters were using the EHR as part of a broader strategy to provide safe care.

Clinical Relevance Statement The PPPSA is a tool designed specifically to measure patient safety activities and to serve as a direc- tional instrument for ambulatory care practices interested in adopting extensive safety practices across all aspects of care-provision. When practices implement EHR, thought should be given as to how the EHR will perform as a tool to support patient safety practices, and what additional work- flows and resources may be necessary to maximize the usefulness of EHR to promote safety.

Conflict of Interest Clare Tanner, Joanne Pohl, and Joanne White have no financial or personal relationships with people or organizations that would influence or bias the objectivity of the study. David Gans and Radhika Nath do not report any conflicts of interest.

Human Subjects Protections This study was conducted solely with de-identified archival data. No human subjects were involved that would make the study subject to the provisions of 45 CFR 46.

Research Article

C. Tanner et al.: EHR and Patient Safety

For personal or educational use only. No other uses without permission. All rights reserved. Downloaded from www.aci-journal.org on 2015-08-29 | ID: 1001059516 | IP: 168.156.40.175

142

© Schattauer 2015

Table 1 Patient safety practice adoption by primary care settings with and without EHR, years 2006–2010, n=209.

Expected relationship to EHR

Medications (16 items)

Yes

Yes

Yes

Yes

Possible

Possible

Possible

Possible

Possible

No

No

No

No

No

No

No

Handoffs and Transitions (11 items)

Yes

Possible

Possible

Possible

Possible

Possible

Possible

Items

Patients receive up-to-date medication list at each visit

E-prescribing system in place

Documentation system (manual or electronic) for all prescribed medications and renewals

All prescribing practice staff have access to drug in- formation and CDS for medications

Detailed records/log on all vaccines dispensed

Review of all medications and side effects with docu- mentation at every visit

Complete medication history including over-the- counter medications

List of high-alert drugs relevant to practice that require direct physician-pharmacist contact

Tracking system in place for all patients on warfarin

Indication for all medications on all prescriptions

Medications in practice checked regularly for expir- ation dates

Detailed labeling of all multi-dose injectable vials

Detailed labeling of all dispensed medications in- cluding samples

External medications labeled and stored separately

Up-to-date medication information for non-English speakers

All child bearing age females required to have preg- nancy test before teratogens prescribed

Practice tracks all laboratory and pathology testing

Process for learning and recording essential new in- formation about patient from outside practice

Patients have easy access to all laboratory and con- sult results

All test results are communicated to patient in timely manner (24–48 hours)

Confirm/record patient discharge from hospital/facil- ity

System in place to track critical versus routine lab- oratory and pathology tests

Practice has process to communicate all medications when patient admitted to hospital/facility

% Fully implemented n=209*

Paper n=117

10.3%

17.9%

57.3%

88.0%

74.8%

70.1%

82.9%

37.6%

71.0%

19.7%

88.9%

64.9%

57.0%

52.1%

30.8%

60.7%

68.4%

63.2%

70.1%

72.6%

71.8%

88.0%

79.5%

EHR n=92

44.6%

94.6%

91.3%

96.7%

90.6%

85.9%

92.4%

54.3%

81.6%

44.6%

98.9%

83.3%

81.7%

71.7%

43.5%

60.9%

85.9%

85.9%

83.7%

87.0%

81.5%

93.5%

87.0%

Sig.**

0.000

0.000

0.000

0.001

0.001

0.005

0.016

0.020

0.119

0.000

0.000

0.000

0.000

0.009

0.072

0.382

0.013

0.001

0.006

0.010

0.027

0.060

0.110

Research Article

C. Tanner et al.: EHR and Patient Safety

For personal or educational use only. No other uses without permission. All rights reserved. Downloaded from www.aci-journal.org on 2015-08-29 | ID: 1001059516 | IP: 168.156.40.175

143

© Schattauer 2015

Expected relationship to EHRPossible

Possible

Possible

No

Personnel/Qualifications/Competency (6 items)

No

No

No

No

No

No

Practice Management/Culture (20 items)

Possible

Possible

No

No

No

No

No

No

No

No

No

No

No

Items

Transfer of patient is clear with new clinician respon- sibility accepted including patient records

System in place for imaging test tracking

Consultation tracking system in place

Practice identifies in-office emergent situations with process in place to address them

Employees educated about new drugs/products by staff/practitioners (not sales representatives)

System to periodically assess physician competency

Orientation to policies and procedures for nursing/ technical staff

Orientation to policies and procedures for all new physicians/PAs/NPs

System to periodically assess nursing/support staff competency

Practice maintains system for continuing education

Patients instructed on proper use/maintenance of prescribed devices

Essential patient information clear and easily access- ible to appropriate office personnel

Practice encourages patients to share safety con- cerns in writing or in surveys

When errors/near misses occur, open educational ef- forts used that include all personnel <