Keywords

Documentation time, Error, Keyboard charting, Voice recognition

 

Authors

  1. Mayer, LeAnn DNP, RN, CNE
  2. Xu, Dongjuan PhD, RN
  3. Edwards, Nancy PhD, RN, APN
  4. Bokhart, Gordon PharmD, BS

Abstract

The purposes of this study are threefold: (1) compare the document times between a voice recognition system and keyboard charting, (2) compare the number of errors between the two methods, and (3) identify factors influencing documentation time. Voice recognition systems are considered a potential solution to decrease documentation time. However, little is known to what extent voice recognition systems can save nurses' documentation time. A pilot, simulation study was conducted using a voice recognition system and keyboard charting with 15 acute care nurses. A crossover method with repeated measures was utilized. Each nurse was given two simple and two complex assessment scenarios, assigned in random order, to document using both methods. Paired t-tests and multivariate linear regression models were used for data analysis. The voice recognition method saved the nurses 2.3 minutes (simple scenario) and 6.1 minutes (complex scenario) on average and was statistically significant (P < .001). There were no significant differences in errors or factors identified influencing documentation times. Eighty percent reported a preference of using voice recognition systems, and 87% agreed this method helped speed up charting. This study can show how a voice recognition system can improve documentation times compared with keyboard charting while still having thorough documentation.

 

Article Content

As an essential part of nursing practice, documentation accounts for more than one-third of nurses' time (35.3%).1 Decreasing documentation burden is the first recommendation of the American Medical Informatics Association report on the EHR 2020 Task Force.2 With recent voice recognition (VR) technology advances, healthcare professionals and settings have increasingly adopted VR systems for patient documentation. Despite the widespread use, two recent systemic reviews of VR program for clinical documentation concluded research on this topic mostly involved the radiology and emergency departments and a lack of studies in nursing.3,4 More research on VR program for nursing documentation is particularly important because nurses are one of the largest users of health information technology and suffer from high documentation burden.

 

Generally, most studies found implementing VR technology had greater clinician productivity in terms of reducing turnaround times (often by 90%) than traditional dictation and transcription.3 However, previous research findings are mixed regarding whether using VR system reduced documentation time. For example, a study found there was a significant reduction in the radiology reporting time using VR compared with the traditional tape dictation-transcription method.5 In contrast, another study indicated using VR required an additional 200% of physician dictation and correction time (9 vs 3 minutes), compared with traditional dictation and transcription in an outpatient pediatric specialty practice.6 Collectively, the findings clearly support the need for more research to further examine the relationship between documentation time and VR technology. Moreover, there is a paucity of evidence regarding whether incorporating VR technology into the documentation process would increase or decrease nursing documentation time.

 

Besides turnaround time and documentation time, another common research topic was the accuracy of reports generated by VR technology. Accurate, legible, and complete documentation is very important for care quality and patient safety. Incorrect information in the EHR was the top EHR-related contributing factor and accounted for 20% of medical malpractice cases.7 A recent study found the error rate was more than 7% in VR-generated clinical documents, nearly 16% of them involved clinical information, and 1 in 250 words contained clinically significant errors.8 Relative to traditional dictation transcription, automated VR technology had a much higher error rate (23% vs 4%) in breast imaging report.9 It is essential to investigate the accuracy of nursing document generated by VR technology.

 

Besides efficiency and quality of VR for documentation, user's satisfaction, experiences, and preferences of VR system are also very critical. A clinical survey of using VR for clinical documentation suggested that 79% reported satisfaction with VR and 77% agreed VR improved efficiency; meanwhile, 29% did not choose VR as one of their preferred documentation methods.10 They also found satisfaction was negatively related to error and editing time but positively related to efficiency.10 A small simulation study with three nurses found all preferred the use of VR and felt it used less time than computerized keyboard documentation.11 Voice recognition software and technical issues were the main reasons for nurse participants preferring keyboard documentation.12 More information is needed to understand nurses' experiences with VR for clinical documentation.

 

The purposes of this study are threefold: (1) to compare the documentation times between a VR program and traditional keyboard charting, (2) to compare the number of errors between the two methods, and (3) to investigate users' experiences with VR and identify factors influencing document time.

 

METHODS

Study Design and Sample

This was a crossover study design utilizing a simulation format. The study participants were recruited from May to September 2019 from two medical-surgical departments of a large, Midwestern hospital. The inclusion criteria were as follows: (1) >=18 years old, (2) a licensed nurse, and (3) a hospital employee. This study was approved by the university, the hospital, and the institutional review board. Written informed consent was obtained from each nurse before any data were collected. The survey was completed anonymously, and the nurses were assured their responses would be kept confidential. A $25 gift card was provided to participants who completed all data collections as compensation for their travel and time.

 

Measures

Patient Scenarios

This was a simulation study and no actual patients were involved. Nurses were given four author-developed common patient scenarios: two simple scenarios and two complex scenarios. Regarding the amount of document charting, the simple scenarios were similar, as well as the two complex scenarios. Prior to the beginning of the study, an RN was timed using keyboard charting for all four scenarios to validate the similarity between scenarios. One of the simple scenarios was designated to document using a keyboard (Cerner), and the other simple scenario was designated to use the VR program (Dragon Medical 360). One of the complex scenarios was designated to document using a keyboard, and the other complex scenario was designated to use the VR program. The sequence of the four scenarios and whether they are documented via VR or keyboard was identified in a random order. Each scenario was timed from when the participant started the program until submitted or saved.

 

Training

The participants interacted with a 15-minute demonstration on the use of the VR program. After the training, the participant was provided with a scenario to practice while using the commands associated with the VR program. The participants practiced with the demonstration scenario until a comfort level was achieved. Almost all the participants were ready to start the study after the second practice attempt. The total training for the participants lasted 25-30 minutes.

 

Surveys

Before the simulation, a survey including age, sex, race/ethnicity, education, shift work, work experience, and use of VR was filled. Upon completion of the simulation, nurses were asked a 5-point Likert scale with four questions about the training as well as their experiences and preferences of VR. Nurses were also asked the likelihood of using VR program for documentation if they had more experience on a 0-10 scale.

 

Statistical Analysis

Descriptive statistics were used to describe the characteristics of nurse participants. Paired sample t-tests were conducted to compare the documentation times and the number of errors between VR program and the traditional keyboard charting for simple and complex patient scenarios, respectively. In sensitivity analyses, we performed the Wilcoxon signed rank tests and found same results. Multivariate linear regression models were performed to explore factors (age, education, shift work, work experience, and whether using the independent VR systems for personal use) influencing the documentation time of VR program. We conducted the Shapiro-Wilk W test for normality (simple scenario, P = .075; complex scenario, P = .069) and the Breusch-Pagan test for heteroscedasticity (simple scenario, P = .342; complex scenario, P = .681). Because the P values were greater than .05, both normality and homoscedasticity assumptions were not violated. In collinearity diagnostics, the variance inflation factor values greater than 10 suggest multicollinearity. In this study, multicollinearity was not an issue because the variance inflation factor values of all variables were less than 3 (range, 1.70-2.61).

 

RESULTS

All 15 participants were female and White, non-Hispanic nurses. More than half (53%) were 40-49 years old. Sixty percent had a BSN degree, and the rest had an Associate of Science degree in nursing. Approximately 53% worked full-time, 20% worked part-time, and 27% worked PRN (pro re nata). Most of them (53%) had less than 10 years of experience. Sixty percent of nurses had used the independent VR systems for personal use, such as Apple Siri and Google Assist; however, none of them had used dependent VR systems for medical documentation. The nurses' characteristics are presented in Table 1.

  
Table 1 - Click to enlarge in new windowTable 1

Regarding simple patient scenarios, the nurses used 4.44 minutes on average by traditional keyboard charting (see Table 2). In contrast, they used 2.14 minutes by VR system. Using VR program for simple patient scenario nursing documentation saved nurses 2.30 minutes on average, and the difference was statistically significant (P < .001; Table 2). Regarding complex patient scenarios, nurses used 8.98 minutes by traditional keyboard charting and 2.89 minutes by VR system. Using VR program for complex patient scenario nursing documentation saved nurses significantly more than 6 minutes on average (6.09, P < .001; Table 2). There were no differences in errors between VR program and the traditional keyboard charting, regardless of simple or complex patient scenarios (Table 2). There were three errors in simple patient scenarios and five errors in complex patient scenarios using VR program. The most common error was the command was documented into the patient record when using the VR program. For example, "next field" was documented in the patient EHR. By contrast, there was one error in simple patient scenarios and four errors in complex patient scenarios using traditional keyboard charting. Table 3 illustrates the factors influencing document time using VR program. Nurses aged 40-59 years tended to use more time to document simple patient scenarios than those aged 20-39 years ([beta] = 0.88, P = .071). Compared with full-time or part-time nurses, those who worked PRN used more time to document complex patient scenarios ([beta] = 1.14, P = .046). No significant differences were noted between education, work experience, and document time using VR program, regardless of simple or complex patient scenarios. In addition, whether nurses used the independent VR systems for personal use had no association with documentation time (Table 3).

  
Table 2 - Click to enlarge in new windowTable 2 Comparison of the Document Times and Errors Between VR Program and the Traditional Keyboard Charting, for Simple and Complex Scenarios, Respectively
 
Table 3 - Click to enlarge in new windowTable 3 Factors Influencing the Time of Using the VR Program for Nursing Documentation, for Simple and Complex Scenarios, Respectively

As presented in Table 4, all nurses felt they had quality training prior to using the VR program for simulation. Only 20% of nurses preferred to use the keyboard for nursing documentation, whereas 80% of nurses preferred to use the VR program for nursing documentation. Almost 87% of nurses felt the VR program was helpful in "speeding up charting." Regarding the likelihood of using VR program for documentation if they had more experience on a 0-10 scale, all nurses gave a score of 5 or more, with an average score of 8.8.

  
Table 4 - Click to enlarge in new windowTable 4 Reponses of Statements Relate to Keyboard and VR Program for Nursing Documentation

DISCUSSION

This simulation study makes several important contributions to the small body of knowledge regarding VR program for nursing documentation. First, compared with traditional charting method, VR program saved nurses time for documentation regardless of the complexity of the patient scenarios. Second, there was no significant difference in accuracy between VR program and traditional charting method. Third, age and shift work were related to documentation time of using VR program with older nurses and those working PRN requiring more time. Fourth, the majority of nurses (80%) preferred to use VR program for nursing documentation. Although the sample size was relatively small, these findings lay the groundwork for more research on VR-assisted nursing documentation.

 

Given the heavy documentation burden faced by nurses, VR technology seems to be a potential faster approach needed for nurse documentation. The VR program saved the nurses more than 2 and 6 minutes on average for one simple scenario and one complex scenario, respectively. If there are 24 patients in a unit, and each patient received two head-to-toe complex assessments, the total documentation time saved would be 9 hours and 44 minutes per day (two 12-hour shifts) for complex documentation. The VR technology reduced nurses' time spent documenting, which may help nurses to increase time spent on direct patient care. This is particularly important when there is an impending nursing workforce shortage throughout the country13 and nurses are burdened with increasing patient acuities and high workloads.12

 

Although there were more errors in documentation generated by VR program than traditional charting method, the difference was not statistically significant. Previous research has found clinicians usually underestimate errors in VR-generated documentation14 and manual editing and review dramatically reduced the error rate from 7.4% to 0.4%.8 Given concerns of accuracy and potential impact on documentation quality, the importance of manual editing and review, education and training about integrating VR programs with EHR, particularly about VR-associated errors, is needed. It is also necessary to develop methods to automatically detect and correct errors in nursing documentation generated by VR technology. It is important to determine whether errors could be reduced as the VR method becomes acclimated to the users' specific voice characteristics.

 

Regarding user characteristics impacting VR performance, previous studies found sex,15,16 native language,16 and experience level with VR17 were influencing factors. Our study found only age and shift work were associated with documentation time of using VR program. This may be due to the small sample size and lack of enough power to detect a significant difference. Further research with a larger sample size is necessary in order to identify subgroups who may need more training or education to improve VR performance. In addition, as nursing documentation is different from the physicians' documentation, more research is warranted to investigate how VR technology can effectively adapt to nurses' specific requirements and the compatibility and effectiveness of VR technology with existing EHR and nursing documentation systems.

 

This study has several limitations. First, this study has a small sample size, and all participants were female and White, non-Hispanic nurses in one hospital. The generalizability of the findings is limited. Second, this study was in a simulation format. The findings may be different when conducing in the actual practice environment; however, the results show the use in an actual patient environment may be feasible. It is necessary to assess the accuracy and utility of VR technology for nursing documentation with a larger scale in real work situations. Third, only two simple scenarios and two complex scenarios were tested. Moreover, scenarios were researcher-developed. Future research involving a broad range of documentation scenarios is needed to understand the impact of VR technology on the quality and efficiency of nursing documentation.

 

CONCLUSION

Voice recognition technology has emerged as an effective strategy to potentially reduce nurses' documentation burden, while persevering documentation quality. More research remains necessary to understand how VR technology for nursing documentation can be appropriately and successfully intergraded with the EHR in acute care settings.

 

References

 

1. Hendrich A, Chow MP, Skierczynski BA, Lu Z. A 36-hospital time and motion study: how do medical-surgical nurses spend their time?The Permanente Journal. 2008;12(3): 25-34. [Context Link]

 

2. Payne TH, Corley S, Cullen TA, et al. Report of the AMIA EHR-2020 task force on the status and future direction of EHRs. Journal of the American Medical Informatics Association. 2015;22(5): 1102-1110. [Context Link]

 

3. Blackley SV, Huynh J, Wang L, Korach Z, Zhou L. Speech recognition for clinical documentation from 1990 to 2018: a systematic review. Journal of the American Medical Informatics Association. 2019;26(4): 324-338. [Context Link]

 

4. Joseph J, Mooe SE, Patton D, O'Conner T, Nugent LE. The impact of implementing speech recognition technology on the accuracy and efficiency (time to complete) clinical documentation by nurses: a systematic review. Journal of Clinical Nursing. 2020;29(13-14): 2125-2137. [Context Link]

 

5. Rana D, Hurst G, Shepstone L, Pilling J, Cockburn J, Crawford M. Voice recognition for radiology reporting: is it good enough?Clinical Radiology. 2015;60(11): 1205-1212. [Context Link]

 

6. Issenman RM, Jaffer IH. Use of voice recognition software in an outpatient pediatric specialty practice. Pediatrics. 2004;114(3): e290-e293. [Context Link]

 

7. Ruder DB. Malpractice claims analysis confirms risks in EHRs. Patient Safety & Quality Healthcare. 2014;11(1): 20-23. [Context Link]

 

8. Zhou L, Blackley SV, Kowalski L, et al. Analysis of errors in dictated clinical documents assisted by speech recognition software and professional transcriptionists. JAMA Network Open. 2018;1(3): e180530. [Context Link]

 

9. Basma S, Lord B, Jacks LM, Rizk M, Scaranelo AM. Error rates in breast imaging reports: comparison of automatic speech recognition and dictation transcription. AJR. American Journal of Roentgenology. 2011;197(4): 923-927. [Context Link]

 

10. Goss FR, Blackley SV, Ortega CA, et al. A clinician survey of using speech recognition for clinical documentation in the electronic health record. International Journal of Medical Informatics. 2019;130: 103938. [Context Link]

 

11. Carter-Wesley J. Voice recognition dictation for nurses. The Journal of Nursing Administration. 2009;39(7-8): 310-312. [Context Link]

 

12. Fratzke J, Tucker S, Shedenhelm H, Arnold J, Belda T, Petera M. Enhancing nursing practice by utilizing voice recognition for direct documentation. The Journal of Nursing Administration. 2014;44(2): 79-86. [Context Link]

 

13. Juraschek SP, Zhang X, Ranganathan V, Lin VW. United States registered nurse workforce report card and shortage forecast. American Journal of Medical Quality. 2019;34(5): 473-481. [Context Link]

 

14. Quint LE, Quint DJ, Myles JD. Frequency and spectrum of errors in final radiology reports generated with automatic speech recognition technology. Journal of the American College of Radiology. 2008;5(12): 1196-1199. [Context Link]

 

15. Rodger JA, Pendharkar PC. A field study of database communication issues peculiar to users of a voice activated medical tracking application. Decision Support Systems. 2007;43(1): 168-180. [Context Link]

 

16. Suominen H, Johnson M, Zhou L, et al. Capturing patient information at nursing shift changes: methodological evaluation of speech recognition and information extraction. Journal of the American Medical Informatics Association. 2015;22(e1): e48-e66. [Context Link]

 

17. Kauppinen TA, Kaipio J, Koivikko MP. Learning curve of speech recognition. Journal of Digital Imaging. 2013;26(6): 1020-1024. [Context Link]