Medical Scribes: How Do Their Notes Stack
Total Page:16
File Type:pdf, Size:1020Kb
Anita D. Misra-Hebert, ORIGINAL RESEARCH MD, MPH; Linda Amah, MD; Andrew Rabovsky, BS; Shannon Morrison, MS; Medical scribes: Marven Cantave; Bo Hu, PhD; Christine A. Sinsky, MD; Michael B. Rothberg, How do their notes stack up? MD, MPH Center for Value-Based Care Research, Medicine Institute, Their outpatient notes stack up well, according to this Cleveland Clinic, Ohio (Drs. Misra-Hebert and Rothberg); small, retrospective review. Scribes’ notes were rated Department of Hospital Medicine, Bridgeport slightly higher in overall quality than physicians’ notes— Hospital-Yale New Haven Health, Bridgeport, Conn at least for certain patient encounters. (Dr. Amah); Case Western Reserve University College of Medicine, Cleveland, Ohio (Mr. Rabovsky); Department of Quantitative Health ABSTRACT scribed and nonscribed notes was similar Sciences (Ms. Morrison, Dr. Hu), Cleveland Clinic, Ohio; Objective u Medical scribes are increasingly (mean words 618, standard deviation (SD) Case Western Reserve employed to improve physician efficiency with 273 for scribed notes vs 558 words, SD 289 for University, Cleveland, Ohio regard to the electronic medical record (EMR). nonscribed notes; P=.12). (Mr. Cantave); American Medical Association, The impact of scribes on the quality of out- Conclusions u In this retrospective review, Chicago, Ill (Dr. Sinsky) patient visit notes is not known. To assess the ambulatory notes were of higher quality effect, we conducted a retrospective review of when medical assistants acted as scribes than [email protected] ambulatory progress notes written before and when physicians wrote them alone, at least The authors reported no potential conflict of interest relevant to this after 8 practice sites transitioned to the use of for diabetes visits. Our findings may not ap- article. medical assistants as scribes. ply to professional scribes who are not part of The data reported here were Methods u The Physician Documentation the clinical care team. As the use of medical presented as a poster presentation at Quality Instrument 9 (PDQI-9) was used to scribes expands, additional studies should ex- the Society of General Internal Medicine’s national meeting in compare the quality of outpatient progress amine the impact of scribes on other aspects Toronto, Canada on April 24, 2015. notes written by medical assistant scribes with of care quality. the quality of notes written by 18 primary care physicians working without a scribe. The notes eam-based models of primary care pertained to diabetes encounters and same- delivery may incorporate medical day appointments and were written during scribes to improve efficiency of elec- T 1-4 the 3 to 6 months preceding the use of scribes tronic documentation. The employment of (pre-scribe period) and the 3 to 6 months after medical scribes has grown rapidly, and it is scribes were employed (scribe period). estimated that within several years there may Results u One hundred eight notes from the be one scribe for every 9 physicians.3 pre-scribe period and 109 from the scribe pe- Accurate documentation is important to riod were reviewed. Scribed notes were rated providing high-quality patient care but can higher in overall quality than unscribed notes take a significant amount of time. Attending (mean total PDQI-9 score 30.3 for scribed physicians have been estimated to spend as notes vs 28.9 for nonscribed notes; P=.01) and long as 52 minutes per day authoring notes.5 more up-to-date, thorough, useful, and com- Medical scribes can help physicians improve prehensible. The differences were limited to the efficiency of electronic documentation6 diabetes encounters. For same-day appoint- and save time.2 Using scribes can also im- ments, scribed and nonscribed notes did not prove physician productivity7-10 and thereby differ in quality. The total word count of all potentially increase access to care. The im- JFPONLINE.COM VOL 65, NO 3 | MARCH 2016 | THE JOURNAL OF FAMILY PRACTICE 155 pact of scribes on the quality of outpatient sible, succinct, synthesized, and internally visit notes, however, is unknown. consistent.11,12 The PDQI-9 has been applied A team-based care delivery model in previously in inpatient12 and outpatient our health system’s primary care clinics uses settings.13 medical assistants to scribe notes during the While the PDQI-9 is a validated tool, it outpatient encounter. We hypothesized that relies on subjective ratings of note quality outpatient notes written by medical assistant by the reviewer. To control for the subjective scribes would be of similar quality to notes nature of the ratings, an experienced inter- written by the same group of physicians with- nist and an internal medicine resident coded out a scribe. 10 progress notes separately using the PDQI-9 and discussed the results. The process was repeated for a total of 20 notes, after which METHODS consensus was reached with >70% agreement Study design and sample on each attribute of the PDQI-9, suggesting We conducted a retrospective review of am- that the resident’s ratings were reliable when bulatory notes from 18 primary care physi- compared with those of an experienced prac- cians at 8 practice sites in our health system ticing physician. who had adopted a care model in which The resident then evaluated a random medical assistants act as scribes. Each physi- sample of notes written by each physician for Scribed notes cian works with 2 medical assistants. To train diabetes or same-day appointments in the were more for the new model, the physician and medi- pre-scribe and scribe periods. Word counts for up-to-date, cal assistants participated in 2 training ses- the entire note were measured. The notes used thorough, sions of 2 hours each and a half day of clinic to establish the reliability of the ratings were useful, and observation and evaluation with a project excluded from the analysis for this study. comprehensible manager. for diabetes Of the 18 primary care physicians in- Data analysis encounters. cluded in this study, none had less than one We used linear mixed-effects models to ex- year of experience in our health system. Ten- amine note quality measures by adjusting for ure ranged from one to 24 years with a mean possible correlations of notes from the same of 11.3 years. physician. Least-squares estimates were de- For each participating provider, we re- rived; the results were not adjusted for mul- quested all available outpatient progress tiple comparisons. notes with either an International Classifica- tion of Diseases, 9th revision (ICD-9) code for diabetes or a designation of “same day” for RESULTS the 3 to 6 months preceding the use of scribes One hundred eight notes from the pre-scribe (pre-scribe period) and the 3 to 6 months period and 109 notes from the scribe period after employing scribes (scribe period). We were reviewed. Compared with notes writ- chose diabetes encounters as examples of ten by a physician alone, scribed notes were notes addressing chronic disease manage- rated slightly higher in overall quality (mean ment and same-day encounters as examples total PDQI-9 score 30.3 for scribe notes vs of problem-focused notes because these 28.9 for pre-scribe notes; P=.01) and more 2 types of encounters are common in outpa- up-to-date, thorough, useful, and compre- tient primary care practice. hensible (TABLES 1 AND 2). The differences Note quality was evaluated using the were limited to diabetes encounters. For Physician Documentation Quality Instru- same day appointments, scribed notes did ment 9 (PDQI-9), a validated instrument de- not differ in quality from nonscribed notes signed for this purpose, comprising 9 items (TABLE 2). Total word count did not vary sig- rated subjectively on a 5-point Likert scale nificantly between all scribe and pre-scribe (1= not at all, 5= extremely). The items as- notes (mean words 618, SD 273 for scribed sess whether notes are up-to-date, accurate, notes vs 558 words, SD 289 for nonscribed thorough, useful, organized, comprehen- notes; P=.12). 156 THE JOURNAL OF FAMILY PRACTICE | MARCH 2016 | VOL 65, NO 3 MEDICAL SCRIBES TABLE 1 Comparison of PDQI-9 scores for pre-scribe and scribe notes Pre-scribed period Scribe period PDQI-9 item score, N=108 score, N=109 P value Mean (SE) Mean (SE) 1. Up-to-date description of the ideal note 3.28 (0.11) 3.53 (0.11) .01 2. Accurate description of the ideal note 3.05 (0.05) 3.11 (0.05) .32 3. Thorough description of the ideal note 3.38 (0.17) 3.86 (0.17) <.001 4. Useful description of the ideal note 3.41 (0.12) 3.74 (0.12) <.001 5. Organized description of the ideal note 3.18 (0.11) 3.25 (0.11) .42 6. Comprehensible description of the ideal 3.1 (0.11) 3.32 (0.11) .01 note 7. Succinct description of the ideal note 3.04 (0.14) 2.97 (0.14) .56 8. Synthesized description of the ideal note 3.48 (0.16) 3.61 (0.16) .30 9. Internally consistent description of the 2.98 (0.09) 2.96 (0.09) .84 ideal note Total PDQI-9 score 28.9 (0.91) 30.3 (0.91) .01 N, number of notes; PDQI-9, Physician Documentation Quality Instrument 9; SE, standard error. Future EMR development might best focus on planned DISCUSSION found that physician notes and scribed notes utilization by In this retrospective review of ambulatory for both diabetes and same-day encounters physician-scribe notes, progress notes written by medical as- often used EMR-based note templates, which teams. sistant scribes were of higher quality than can lead to over-documentation. notes physicians wrote alone, at least for dia- In general, both physician and scribed betes visits.