Writing and reading in the electronic health record

Abstract

Han and Lopp investigated physicians' self-assessed EHR reading and writing competence and documentation format preferences. Respondents, with an average of 39 months of experience, reported low competence. Reading and writing competence positively correlated with EHR navigation skills. The authors proposed navigation training and structured education for new users, alongside job aids and just-in-time learning for experienced users. The study concludes that education should be adapted to help physicians and students navigate distributed information systems to rapidly locate key patient information and understand clinical reasoning.

Introduction

Electronic health records (EHRs) became more widespread with the expectation that they would improve healthcare quality. The rate of full implementation among US physicians was 3.6% in 2006, 3.8% in 2007, 4% in 2008, and 6.9% in 2009. Research from 1995 to 2005 focused on EHR ownership, adoption, functionality, and disparities, later expanding to workflow, interoperability, communication, educational use, and organizational support. Few studies examined how physicians write and read EHR documentation.

Paper charts have a linear narrative structure that records patient information chronologically in one place. In EHRs, information entered by multiple people and data from laboratory and hospital systems are distributed across locations connected by hyperlinks. Structured data entry facilitates data reuse and analysis but places substantial learning demands on users. Free-text entry and dictation allow flexible expression but may be less efficient for data reuse and workflow.

Reading and writing in EHRs require navigation skills to locate relevant data across distributed sources and reconstruct the flow of clinical reasoning. Previous studies found that even fourth-year medical students with one year of EHR experience made errors when retrieving information. In 2009, approximately 64% of US medical education programs allowed students to use EHRs, primarily for reading, although some also permitted them to write documentation. Existing education focused on writing accurate, thorough, and logical records, while training in reading and navigation was limited.

The study investigated physicians’ self-assessed competence in reading and writing EHR documentation, their preferences for documentation formats, and the demographic characteristics and experience with EHR use associated with each.

Methods

The study participants were physicians using EHRs in hospitals and outpatient training institutions in the Midwestern United States. Purposive and convenience sampling were used to recruit physicians with several years of experience. Participants represented general and specialty fields in inpatient and outpatient care. The study received an exemption from the Institutional Review Board.

Human resources personnel at three hospitals emailed the survey link and sent two follow-up reminders. Because most participating physicians worked concurrently at the three institutions, they received up to six reminders. No financial compensation was provided, and no personally identifiable information was collected. Data were collected over two months, from February to April 2011.

The research team used 14 five-point Likert items and 12 additional questions to examine self-assessed competence, format preferences, navigation skills, computer skills, and priorities for EHR training in medical education. The team used SPSS (PASW Statistics 18) to calculate descriptive statistics and conduct factor, reliability, correlation, and regression analyses. The analysis considered the risk of Type I errors from multiple correlation analyses and Type II errors from Bonferroni correction. Correlations were interpreted using a coefficient magnitude greater than 0.30 and p<0.005 as criteria.

EHR usage survey instrument

The items were developed from the literature on documentation and EHR functionality, along with the concepts of perceived ease of use and perceived usefulness from the Technology Acceptance Model. The initial five categories were writing competence, reading competence, input format preference, output format preference, and note length preference. An advisory panel of 10 members—including four physicians, professors of medical education, psychology, and computer science, an EHR training expert, a director of clinical assessment, and a research assistant—reviewed the items.

The analysis yielded four components with eigenvalues greater than 1; reading and writing competence formed a single component. Cronbach’s α was 0.840 for reading and writing competence, 0.928 for narrative preference, 0.914 for concise record preference, and 0.698 for typing preference.

Table 1. Factor loadings of survey items

Category	Item	Component 1	Component 2	Component 3	Component 4
EHR writing competence	It is easy to write a thorough patient chart in the EHR.	0.76
	EHR templates help complete the chart thoroughly.	0.67
	*I sometimes feel I write incomplete notes in the EHR.	0.68
	*It takes too much time to enter information into the EHR.	0.60
	*It is difficult to find the appropriate template for a note in the EHR.	0.70
EHR reading competence	*It takes too much time to find information in the EHR.	0.73
	*Patient charts written by other physicians in the EHR are often very difficult to understand.	0.59
	Reading EHR charts makes it easy to understand how diagnostic evaluations and plans were determined.	0.58
Narrative reading preference	Narrative charts are easier to understand than itemized charts.		0.90
	Narrative charts are more informative to the reader than itemized charts.		0.90
Concise record reading preference	In the EHR, concise charts are easier to understand than long charts.			0.92
	In the EHR, concise charts are more useful than long charts.			0.91
Typing writing preference	I primarily use text box entry rather than checkboxes on EHR forms when writing patient charts.				0.89
	I prefer typing over using checkboxes in the EHR.				0.83

The scale is 5=Strongly agree, 4=Agree, 3=Neutral, 2=Disagree, 1=Strongly disagree. * indicates reverse-coded items.

Results

Of approximately 730 physicians, 91 were included in the analysis, yielding a 12.5% response rate. The number of respondents increased from 58 initially to 95 after reminders. One EHR non-user, two incomplete responses, and one outlier were excluded.

The mean age was 49 years, with a range of 28 to 69 and a standard deviation of 10.02. Men comprised 58% of the sample. Participants were affiliated with one to three clinical institutions and used one to two EHR systems, with an average of 39 months of EHR experience. EHR navigation skills averaged 3.54 (SD=1.07), and typing skills averaged 3.55 (SD=1.32). General computer navigation skills were the only variable measured on a 4-point scale, averaging 3.20 (SD=0.82).

Despite an average of 39 months of experience, self-assessed EHR reading and writing competence was low, with a mean of 2.74 (SD=0.76). Physicians reported difficulty finding appropriate templates, felt their notes were incomplete, and struggled to understand other physicians’ diagnoses and treatment plans. The difference in competence between users of the two main EHR systems was not significant (F=0.23, p=0.88), nor was the correlation between the number of systems used and competence.

Preference for narrative records averaged 3.87 (SD=1.09), and preference for concise records averaged 3.64 (SD=1.16). There was no clear preference for either typing or checkboxes (mean 3.01, SD=0.99). Higher self-assessed competence correlated with weaker preferences for concise records (r=−0.33, p<0.001) and narrative records (r=−0.36, p<0.001).

Table 2. Spearman correlation coefficients

Variable	Concise record reading preference	Narrative reading preference	EHR navigation skills	EHR experience (months)	Specialty–Surgery	Age	Gender (M1F2)	Typing skills	General computer navigation skills
EHR reading and writing competence	−0.33** (n=90)	−0.36** (n=90)	0.55** (n=90)
Typing writing preference								0.29** (n=88)
Concise record reading preference		0.24* (n=90)			0.25* (n=90)
Narrative reading preference						0.22* (n=85)	−0.21* (n=89)		−0.26* (n=90)
EHR navigation skills							0.26* (n=89)	0.32** (n=90)	0.28** (n=90)
EHR experience (months)						−0.34** (n=79)
EHR training needs			−0.50** (n=88)	−0.23* (n=80)				−0.30** (n=88)
Age							−0.30** (n=84)	−0.28** (n=85)	−0.30** (n=85)
Gender								0.40** (n=89)
Typing skills									0.32** (n=90)

* p<0.05, ** p<0.01, two-tailed test. Significance indicators follow the criteria of the original table.

EHR navigation skills were positively correlated with reading and writing competence (r=0.55, p<0.0001). The regression model predicting navigation skills had an R² of 0.19 (F=6.4, p<0.001). General computer navigation skills were the only significant predictor (t=2.88, p<0.005). The contributions of gender and typing skills were not significant. Older age correlated with a shorter duration of EHR use (r=−0.34, p<0.002) and lower general computer navigation skills.

The regression model for training needs had an R² of 0.35 (F=14.6, p<0.001). Duration of use (t=−2.5, p<0.02) and navigation skills (t=−5.1, p<0.001) were significant predictors. Longer use and greater navigation proficiency correlated with lower training needs. Duration of use was not positively correlated with reading and writing competence or navigation skills. Respondents’ priorities for medical education included EHR navigation (59%), writing thorough records (58%), and using a computer during patient care (56%).

Discussion

The study proposed providing navigation training before general EHR training. Learning to develop a spatial cognitive map was suggested to help users locate information and templates in nonlinear systems. The weaker preference for narrative formats among physicians with higher competence was linked to their ability to find relevant data across distributed sources.

The automatic inclusion of vital signs, medications, test results, diagnostic reports, and consultation data increases the length of the record. The study suggested that students practice reading long, fragmented records and retrieving relevant information during clinical clerkships. Physicians can help determine record formats and convey key information by placing the assessment, plan, and diagnostic rationale at the top.

The study proposed structured education focused on navigation for new users. Help functions offering just-in-time learning and job aids were proposed for physicians who had received initial training and used the system for an extended period.

The study’s findings are best interpreted in light of its self-assessed measures of competence and navigation skills, small sample, and focus on the Midwestern United States. Follow-up studies with actual performance measures and larger samples were recommended. The small sample size could increase the risk of Type II errors, and results might differ in rural areas or other regions.

Conclusions

EHR competence was characterized as an integrated ability to read and write digitally within the workflow of a distributed information system. The study recommends education tailored to help physicians and students rapidly locate key patient information and follow clinical reasoning in long, nonlinear records.

Reading notes

Because the reading and writing competence scale includes the time required to enter information and the difficulty of locating appropriate forms, this score is best interpreted as a self-assessment reflecting both the ease and the burden physicians experience when performing tasks within the system. The correlation between EHR navigation skills and reading and writing competence (r=0.55) is based on self-reported data collected at the same time.

References

Han, H., & Lopp, L. (2013). Writing and reading in the electronic health record: An entirely new world. Medical Education Online, 18(1). https://doi.org/10.3402/meo.v18i0.18634