d(g(EdcgE

A

The Journal of Emergency Medicine, Vol. 40, No. 6, pp. 682–686, 2011Copyright © 2011 Elsevier Inc.

Printed in the USA. All rights reserved0736-4679/$–see front matter

doi:10.1016/j.jemermed.2009.08.066

Administration ofEmergency Medicine

PILOT STUDY ON DOCUMENTATION SKILLS: IS THERE ADEQUATE TRAININGIN EMERGENCY MEDICINE RESIDENCY?

Moshe Weizberg, MD, Bartholomew Cambria, PA, Yusra Farooqui, MD, Barry Hahn, MD,

Francesca Dazio, PA, Eric Matthew Maniago, MD, Nicole Berwald, MD, Dara Kass, MD,

and Brahim Ardolic, MD

Department of Emergency Medicine, Staten Island University Hospital, Staten Island, New YorkReprint Address: Moshe Weizberg, MD, Department of Emergency Medicine, Staten Island University Hospital, 475 Seaview Avenue,

Staten Island, NY 10305

e

e Abstract—Background: Thorough and accurate docu-mentation in the medical record is important, and docu-mentation skills should be an integral component of emer-gency medicine (EM) residency training. Study Objective:We sought to study the documentation skills of EMresidents as they relate to emergency department (ED)reimbursement. Methods: This was a retrospective, cross-sectional study. We reviewed all charts of patients presentingto the adult ED during a 2-week period. We compared threegroups: patients seen primarily by an EM resident, patientsseen primarily by a physician assistant (PA), and patientsseen primarily by an attending emergency physician. Out-come measures were the incidence of downcodes and dol-lars lost to downcodes in all groups. Results: There were212 patients in the resident group, 683 patients in the PAgroup, and 437 patients in the attending group. There were12 downcodes (5.7%, 95% confidence interval [CI] 2.96–9.70) in the resident group, 10 downcodes (1.5%, 95% CI0.70–2.68) in the PA group, and 17 downcodes (3.9%, 95%CI 2.28–6.14) in the attending group (p � 0.002). The mean

ollar lost per patient seen in the resident group was $3.2195% CI 1.41–5.00); $0.91 (95% CI 0.33–1.49) in the PAroup; and $2.23 (95% CI 1.17–3.28) in the attending groupp � 0.002). Conclusion: Charts documented primarily byM residents were more likely to be downcoded than chartsocumented primarily by PAs or ED attendings. This down-ode rate resulted in a greater loss of revenue in the residentroup. We believe this represents an area for improvement inM residency education. © 2011 Elsevier Inc.

RECEIVED: 25 June 2009; FINAL SUBMISSION RECEIVED: 24

CCEPTED: 30 August 2009

682

Keywords—resident; education; documentation; coding

INTRODUCTION

Thorough and accurate documentation in the medicalrecord is important and serves several important func-tions. It communicates information to other care providers,provides medico-legal protection, and it is also critical toemergency department (ED) reimbursement. Sufficientdocumentation is necessary to support various Evaluationand Management (E&M) codes (1). Insufficient docu-mentation can lead to charts being “downcoded”—thatis, being reimbursed for a lower level of service thanthat actually provided. Downcoding leads to loss ofrevenue (1).

Documentation skills should be an integral compo-nent of emergency medicine (EM) residency training.Documentation is one of the physician tasks listed in the“Model of the Clinical Practice of Emergency Medicine”(2). However, residents have reported that they possesspoor confidence in their ability to accurately determinean E&M code, and that they want more education in thisarea (3,4). Additionally, faculty members review residentcases in real time and have the ability to request addi-tions to resident documentation. However, residents re-port that this often fails to occur (3).

t 2009;

Augus

Resident Documentation Skills 683

At our institution, the ED attending physicians (at-tendings) and physician assistants (PAs) are salaried.However, there is an incentive-based bonus pool, themajority of which is based on clinical productivity. Thisproductivity is based directly on reimbursement mea-surements. This is not an uncommon scenario in EM. Infact, in many places, a large portion of the emergencyphysician’s salary (if not all of it) is “at risk” and deter-mined on reimbursement measurements (e.g., relativevalue units [RVU]/hour). Residents may be leaving theconfines of the academic world unprepared to face thisRVU-based reality, as much of reimbursement is basedon documentation skills.

We sought to study the documentation skills of EMresidents as they relate to ED reimbursement. We hy-pothesize that patients seen primarily by an EM residentmay be more likely to be downcoded due to insufficientdocumentation.

METHODS

Study Design

This was a retrospective, cross-sectional study. Thestudy was approved by the Institutional Review Board.

Study Setting and Population

We retrospectively reviewed all charts presenting to theadult ED during the 2-week period from November 1through November 15, 2005. Charts were obtained fromregistration logs.

Inclusion criteria were all patients presenting to theadult ED during the study period. Exclusion criteria werepatients who left before physician evaluation, patientswho eloped before their evaluation was completed, andpatients transferred to or from another institution. Pa-tients under age 21 years were triaged to the pediatric EDand were excluded from this study. Patients triaged to theUrgent Care Center were also excluded.

We compared three groups: patients who were seenprimarily by an EM resident with real-time attendingsupervision, patients seen primarily by a PA with real-time attending supervision, and patients seen primarilyby an attending emergency physician with no residentinvolvement in the documentation.

Our department is an affiliate site of a Residency ReviewCommittee-approved, 1–4 year Emergency Medicine Res-idency. The residents who rotate in our ED are EM-2,EM-3, and EM-4 residents. They rotate at our hospitalapproximately 3 months per academic year. The charting

system used in our institution is a template paper chart,

and is identical to the chart used at the primary hospitalof the residency. Our PAs work side by side with our EMresidents in the main ED, seeing identical acuity patients.

Our institution is located in a suburban setting with ahigh (63%) private insurance payor mix. However, pro-viders are not generally privy to the patient’s insurancestatus at the time of evaluation. It should be noted that atthe time when the charts were reviewed, the residentshad received no formal training in documentation andcoding. The ED attendings and PAs at our institution hadreceived lectures on the topic several months before thereview period.

Study Protocol

All investigators were involved in chart review, and datawere collected using a standardized abstraction form.The form was developed for the purposes of this studyand prompted the investigator to input the followingdata: age and gender, which were recorded from thepatient registration form; diagnoses and disposition,which were recorded from the physician portion of thechart; and whether the patient was seen primarily by aPA, a resident, or an attending. Information on ED re-imbursement was obtained from our ED coders, whowere blinded to the study. Coders regularly reportedwhen a chart’s E&M level was downcoded. Dollarsbilled were calculated using Medicare par fee schedulefor our geographic area.

Data were entered by one author from the standard-ized data collection sheet into a Microsoft Excel (Mi-crosoft Corporation, Redmond, WA; 2002) spreadsheet.

Outcome Measures

Outcome measures were the incidence of downcodes anddollars lost to downcodes in all groups.

Data Analysis

Sample size analysis was performed before study initia-tion. To achieve 80% power, with an alpha of 0.05, astudy group of 64 patients was required to show a dif-ference in reimbursement of 20%. Means were comparedusing analysis of variance. Proportions were comparedusing chi-square. Results are reported as 95% confidenceintervals (CIs). Statistics were calculated using Analyse-Itfor Microsoft Excel (Analyse-It Software, Ltd., Leeds,UK; version 1.71, 2003) and Graph-Pad (GraphPad Soft-ware, San Diego, CA; version 3.05, 32 bit for Win

95/NT, 2000).

684 M. Weizberg et al.

RESULTS

A total of 1409 patients presented to the adult ED duringthe study period. The charts were unavailable for 6patients. Twenty-two patients left without being evalu-ated. Five patients eloped. Six patients were transferredinto or out of the hospital. Thirty-one patients were � 21years old. One patient went directly to Labor and Deliv-ery. Six patients did not receive a bill. This left a studygroup of 1332 patients.

There were 212 patients in the resident group, 683patients in the PA group, and 437 patients in the attend-ing group. Patient acuities were similar in all groups, andthey were well matched for age, gender, percentage ofpatients admitted to the hospital, and percentage of pa-tients admitted to the ICU.

There were a total of 39 (2.9%) downcoded charts.There were 12 downcodes (5.7%, 95% CI 2.96–9.70) inthe resident group, compared to 10 downcodes (1.5%,95% CI 0.70–2.68) in the PA group, and 17 downcodes(3.9%, 95% CI 2.28–6.14) in the attending group (p �0.002) (Figure 1).

A total of $2271.96 was lost to downcodes in thestudy period. The mean dollar lost per patient seen in theresident group was $3.21 (95% CI 1.41–5.00), comparedto $0.91 (95% CI 0.33–1.49) in the PA group, and $2.23(95% CI 1.17–3.28) in the attending group (p � 0.002)(Figure 1).

DISCUSSION

In our study, resident charts were more likely to bedowncoded than PA charts or attending-only charts.More revenue was lost due to documentation on residentcharts than on PA or attending charts. This may identifya potential area for improvement in resident educationregarding documentation.

In the resident group, $3.21 was lost per patient seendue to downcodes. Because there were 212 patients inthe resident group, this accounted for a loss of $680.52

Figure 1. Percentage of downcodes for residents (RES) vs

Dollars lost per chart for residents vs. physician assistants vs. att

over the 2-week study period. If we project this loss over52 weeks, this would result in a loss of approximately$17,693.52 per year. Our residents rotate from an affil-iate institution and provide approximately 16 h of EDcoverage per day. The projected losses were based onMedicare par fee schedule, which may attenuate theactual losses to an ED, depending on the specific payormix in a particular institution.

The incidence of downcodes was higher in the attend-ing group than in the PA group. Mean dollar lost perpatient seen was more than double for the attendinggroup. This suggests that inadequate documentationskills as a resident may translate into poor documentationpractices as an attending. Leaving a residency programwithout a clear understanding of the impact of poordocumentation on reimbursement is a failure of the res-idency program to prepare their graduates for real-worldemergency medicine practice.

In a 1999 resident survey, a majority of residents ratedtheir confidence in their ability to accurately determinean E&M code as “minimal” or “not at all” (3). A morerecent study found that residents want more education inthe areas of documentation and billing (4). Our studysuggests that this deficit remains at the present time.Several options have been employed to improve residentdocumentation, including financial incentives to the res-idents (5). However, we believe enhanced resident edu-cation in the area of coding and reimbursement is criticalfor improvement. Such education could include formallectures by ED coding experts on a regular basis, real-time attending input, and feedback from billing compa-nies regarding downcoded charts.

It is difficult to understand why the PAs demonstratedmore accurate documentation skills than both the resi-dents and the attendings. The PAs did receive somecursory education on ED documentation, and were alsogiven incentives to document more accurately. The res-idents did not receive any formal education or incentive.However, the attendings received the same education asthe PAs, and are subjected to the same reimbursement

ician assistants (PA) vs. attending emergency physicians.

. phys ending emergency physicians.

Resident Documentation Skills 685

productivity incentive program. Changing physician be-havior can be a difficult endeavor, even with a financialincentive program. This reinforces the idea that soliddocumentation skills should be acquired during resi-dency, as it is more difficult to change physician practiceat the level of attending physician.

Limitations

This study was retrospective in nature, introducing apotential source of error in the chart review. However,we believe that doing this study in a prospective mannerwould introduce a significant Hawthorne effect. If pro-viders knew that their documentation was being studied,there could be an increased focus on ensuring they pro-vide documentation to support E&M codes, creatinganother potential source of bias.

We did not break down our results based on payortype. However, our downcode rate over the last fewyears has not varied based on payor type. In addition, theattendings and the residents are usually unaware of thepayor type at the time of documentation.

We studied documentation in one institution involv-ing a single residency program. Our ED is an affiliate sitefor an EM residency whose residents rotate with usapproximately 3 months per year. It is possible thatresidents’ documentation skills are better at their primarytraining site, where they spend more time and feel morecomfortable. However, our chart is a template paperchart that is identical to the chart used at the residents’

primary hospital. This makes it less likely that documen-

tation practices would differ significantly from site tosite. In addition, all residents were grouped togetherdespite year of training. It is possible that junior resi-dents’ charts are more likely to be downcoded thansenior residents’ charts.

CONCLUSION

In this study, charts documented primarily by EM resi-dents were more likely to be downcoded than chartsdocumented primarily by physician assistants or attend-ing emergency physicians. This downcode rate resultedin a greater loss of revenue in the resident group. Webelieve this represents an area for improvement in EMresident education.

REFERENCES

1. Edelberg C. Emergency department coding and billing. Emerg MedClin North Am 2004;22:131–51.

2. Thomas HA, Beeson MS, Binder LS, et al. The 2005 Model of theClinical Practice of Emergency Medicine: the 2007 update. AnnEmerg Med 2008;52:e1–e17.

3. Howell J, Chisholm C, Clark A, Spillane L. Emergency medicineresident documentation: results of the 1999 American Board ofEmergency Medicine In-Training Examination Survey. Acad EmergMed 2000;7:1135–8.

4. Carter KA, Dawson BC, Brewer K, Lawson L. RVU ready? Pre-paring emergency medicine resident physicians in documentation firan incentive-based work environment. Acad Emerg Med 2009;16:423–8.

5. Pines JM, Reiser RC, Brady WJ, et al. The effect of performance

incentives on resident documentation in an emergency medicineresidency. J Emerg Med 2007;32:315–9.

686 M. Weizberg et al.

ARTICLE SUMMARY1. Why is this topic important?

Documentation skills should be an integral componentof emergency medicine residency training. This studyexamines the documentation skills of emergency medi-cine residents as they relate to emergency departmentreimbursement.2. What does this study attempt to show?

This study attempts to show the incidence of down-codes and dollars lost to downcodes for resident chartscompared to charts documented by physician assistants(PAs) and attending physicians.3. What are the key findings?

The incidence of downcodes and the dollars lost todowncodes were significantly higher for resident chartsthan for PA charts and attending charts.4. How is patient care impacted?

Our results suggest that documentation skills are anarea for improvement in emergency medicine residencyeducation.