This observational study shows that clinical work performed by family physicians correlates poorly with common codes and fees under the existing coding and billing rules.
The income disparity between primary care and other physicians has been attributed in part to the evaluation and management (E/M) rules written by CMS. The purpose of this study was to examine the relationship between family physicians’ work and their actual coding practices and fees collected under these widely used rules.
This was a direct observational time-motion study.
A diverse group of 15 family physicians were shadowed over consecutive patient visits at their ambulatory practices, usually for a half-day of clinic. Data about each visit were recorded, including time parameters; number of issues covered; number of labs, images, and chronic prescriptions ordered; the physician fee code from the Current Procedural Terminology (CPT) system that was submitted; the actual payer for each patient; and the actual fee collected. The primary outcome was the correlation between the time spent for each patient’s care and coding/financial measures.
The average total time a physician spent per patient including documentation time was 20.0 minutes. The average fee collected was $101.40, including patient co-pays. The correlation between the actual fee collected and the physician’s time spent working on each patient’s behalf was poor (R2 = 0.137, P <.001). There was a wide variation in times and fees for each CPT code category.
The existing E/M rules and CPT coding system have created office visit fees that correlate poorly with family physician work. These findings provide another justification for disruptive primary care payment reform.
Am J Manag Care. 2014;20(11):876-882
This observational study shows that clinical work performed by family physicians correlates poorly with common codes and fees under the existing coding and billing rules. Our findings suggest that:
In 2008, Bruce Steinwald, director of healthcare for the Government Accountability Office, testifying before the Senate Health, Education, Labor, and Pensions Committee, stated “When I say primary care services are undervalued, that does not mean that just increasing the prices paid to primary care is the solution.”The chair of the Medicare Payment Advisory Commission in 2009 also recognized that the current fee structure “does not consider the value to patient or value to society or the shortage of various types of providers.”
In the United States, all primary care services paid for by the Medicare and Medicaid programs, and the vast majority of services for privately insured patients, follow documentation, coding, and billing rules established by CMS based on codes created by the American Medical Association for its Current Procedural Terminology (CPT) manual.The most common CPT outpatient codes are 99212 through 99214.
The hours family physicians work per week is near the median of all physicians,although average annual personal income is approximately $175,000-$221,000,which is nearly 3 times less than the average income of the highest paid physicians and is a little more than half the average income of non—primary care physicians.An analysis of physician incomes concluded that the Medicare fee schedules that arise from CMS’s Evaluation and Management (E/M) rules, not inflated fees to non-primary care physicians in the private market, directly explain this income discrepancy.
The income disparity is further explained by the way physician work is valued. The difference in the hourly payment rate between cognitive and procedural services can be quite dramatic. In Boston, for example, Medicare pays primary care $103 for a 30-minute visit but $449 for a diagnostic colonoscopy. The latter has been reported to take about 30 minutes,though other studies found that a screening colonoscopy actually only takes about 14 minutes to complete.
Little research has been conducted to examine the effect of the existing E/M rules and associated physician fee payments on clinical practice since the rules were published in 1995 and 1998.The purpose of this study was to examine the relationship between observed family physicians’ workload and their actual coding practices and fees collected.
The lead author (RY) observed 127 patient clinic encounters of 15 family physicians. The physicians were chosen to represent a variety of practice locations, practice types, and payers. RY personally knew some of physicians prior to the observation. Others were suggested by mutual colleagues to represent a diversity of ages, medical school and residency education, and practice location. Each potential study physician was called to explain the rationale for the study, and their permission was sought to shadow a typical day of their practice for one-half to 1 day in clinic.
The physicians were observed after consent was obtained from each patient. The patient received a 1-page explanation of the purpose of the project from a clinic staff member prior to the observation. Out of 128 patients who were asked, 127 consented to allow the clinical encounter to be observed.
While the physician cared for the patient, RY tried to be merely a “fly on the wall” and not influence the visit. The observed physician was specifically asked to not change any habits or decisions because of the observa- tion. While in the room, the observer moved to the most unobtrusive corner possible, making an effort to minimize eye contact with the physician and patient. The observer stepped out of the exam room for sensitive physical exams such as gynecological exams.
Times were recorded in minutes, which commenced from the moment the physician started working with the patient’s chart prior to entering the exam room to the time the documentation for the visit was complete. The number of issues that were covered were recorded, as were the numbers of labs, images, and chronic prescriptions that were ordered.
The only time the observer interrupted the physician’s usual work flow was between patient encounters. For a patient just seen, the physician was asked which billing codes were submitted, who the payer was, and how many issues the physician had addressed. No patient-specific in- formation was obtained other than what was addressed in the encounter. Time devoted to office administrative issues or other patient care work that was not directly rela- ted to the observed patient encounters was not captured.
Approximately half of the physicians with electronic medical records (EMRs) said they would finish charting later that night from home. To estimate the total time required for each patient encounter, these physicians were asked to estimate the total amount of time they would spend charting for all incomplete medical records. A mean time-imputation method was conducted for each physician using EMR with delayed charting. The total amount of time reported by the physician was divided by the number of incomplete charts. The calculated value was then applied to each of these visits.
At the end of the clinic day, the lead investigator discussed with each physician aspects of that day’s patient care and further clarified the actual bills that were submitted. If possible, expected payment amounts were obtained for each claim at that time. Often the office manager was contacted at a later time to obtain information about the actual fees collected, including patient co-pays. The only adjustments made to the collected data were that all the collected fees were classified as coming from an indepen- dent practice. If the physician worked in a hospital-owned clinic or federally qualified health center, local Medicare rates were used for a participating physician, so as to exclude separate facility fees (so-called provider-based billing) from the revenue estimates. The fees recorded did not include the material costs for supplies such as vaccines and injectable medications, because the allowable fees are very near the practices’ acquisition costs.
Linear regression was used to assess for associations between the recorded times and collected fees. To correlate the CPT code categories, the following 4 categories were established: 99213, 99214, preventive visits (99391, 99395, etc), and combination codes, which were visits that met one of 2 criteria: the visit was billed as a preventive code plus an E/M code with a —25 modifier (the majority of these visits), or the visit consisted of an E/M service plus an additional fee for a common outpatient procedure. Diagnosis codes accounting for fewer than 5 visits were excluded from analysis of categories. Homogeneity of variance between E/M codes for time and fees was examined using Levene’s test. Statistical analysis was performed with SPSS (SPSS Inc, Chicago, Illinois). This project was approved by the John Peter Smith Health Network Institutional Review Board.
Characteristics of the observed physicians are shown in . They were predominantly mid-career physicians in private practice, though a diversity of locations, physician backgrounds, and practice types were represented. Descrip- tive statistics for some of the features of the actual clinic visits are shown in . The average visit length was 20 minutes and the mean actual fee collected was $101. No unusual or extreme documentation, coding, or billing practices were observed. A few visits that really dealt primarily with several chronic conditions were billed as preventive or well-person visits. This accounted for a small minority of observed visits.
The comparison and correlation of the physician time spent for each patient versus the actual fee collected in the observed clinic visits is shown in (= .137, <.001). The variance between the physician time and actual code used is shown in . Variance between the 4 groups of E/M codes was significantly different for time (<.01) and fee (<.001), though each code category had a wide range of associated times (). However, variances were not sig- nificantly different between E/M codes 99213 and 99214 (= .09) or preventive and combination codes (= .58).
The average fees collected for 10-, 20-, and 40-minute visits were $84, $102, and $135, respectively, which translated to $504/ hour, $306/hour, and $203/hour, respectively.
We found a statistically significant but weak correlation between the actual fee collected and the physician’s time spent working on each patient’s behalf, and a wide range of times for each CPT code category. The net effect of these findings resulted in a payment rate per hour that decreased as the visit length increased.
Previous research has found that although physician work involves many factors—including stress, malprac- tice risk, and equipment costs, among others—time is by far the single most correlated variable with work effort. This was concluded in studies that contributed to the current Resource-Based Relative Value Scale systemand subsequent work noting how efficiencies are gained if more than 1 issue is addressed at each office visit.In practical application, this has proved true. When integrated networks, such as Seattle-based Group Health Cooperative, increased the typical clinic visit from 20 to 30 minutes, they achieved improvements in patients’ experiences, composite measures of quality, and clinician burnout over a 2-year period. Emergency department (ED) visits decreased 29% and hospitalizations decreased 6%, which led to total savings of $10.30 per patient per month in a relatively healthy employed population.These findings help justify our use of time as a surrogate measure for work effort.
The average amount of time per visit observed in our study, 20 minutes, was similar to that observed in previous studies of US and Swedish generalist physicians,but longer than the 7 to 9 minutes per visit observed in British general practitioners, who address fewer issues per visit.A large observational study of family physicians conducted the year before the first E/M rules were published found the average clinic time was 10 minutes, though this did not include documentation time.This time was similar to our observed face-to-face time of 12.5 minutes. One of the significant differences between the US and British systems that further explains these visit time differences is that British general practitioners only have general and brief documentation guidance—they must record relevant clinical findings—whereas the CMS rules are contained within a 98-page document,with fur- ther rules and interpretations written by the CMS inter- mediary payers.
Because the highest cost Medicare and Medicaid patients are those with multiple chronic diseases,the documentation, coding, and billing rules should encourage excellent care for these patients, but the current system does not. Previous studies have shown that for a mostly middle-aged insured population, a family physician addresses 2.5 to 3.1 issues in the average clinic visit.The average number of issues per visit grows to 3.9 to 6 for elderly patients and to 4.6 for patients with diabetes.Our findings were similar to these.
These results help inform the findings of another study that primary care physicians in clinic only spend 1 minute on other patient concerns after the chief complaint is addressed.This pressure to end one visit to start the next one has been associated with reduced provision of common chronic disease services when other patient concerns are addressed.These time pressures have also been associated with decreased job satisfaction among family physicians and general practitioners.Adverse work-flow (time pressure and chaotic environments), low work control, and unfavorable organizational culture were strongly associated with low physician satisfaction, high stress, burnout, and intent to leave.
Our findings have implications for the care quality and experience of the patient. Longer consultations in primary care have been found to improve processes of careand have been associated with greater patient satisfaction and reduced healthcare utilization.Shorter family medicine visits have been found to result in less thorough histories,and they have been associated with lower rates of preventive services, lower patient satisfaction, and a lower rating for the doctor-patient relationship.As a result, commentators have called for longer primary care visits, especially for high-risk patients with multiple comorbidities,which is economically unviable for primary care providers in the current payment system.
Our findings have implications for physician work-force planning. Medical students are aware of the large payment disparity between primary care and the other medical fields. The workload and stress of primary care make students less likely to pursue generalist careers today than 20 years ago.Decreasing this payment disparity is one of the reforms that must occur to attract more medical students into primary care careers.
No unusual billing practices were observed, though the lack of agreement regarding the proper CPT code even among professional coders has been recognized as a problem with the current system.Our finding that the actual payment rate per hour of work decreases with longer visits might be in part a result of the complexity of the existing E/M rules. A study of coding specialists in application of the CPT E/M documentation and coding guidelines for 6 hypothetical cases found that they agreed on the appropriate code only 57% of the time.Other studies found rates of agreements between chart auditors as low as 15%.Coding disagreements have been strongly associated with physician undercoding, which is often the result of not documenting all the of the issues addressed and final diagnoses in the medical record.Undercoding was especially noted in longer visits.Several authors have concluded that the CPT coding guidelines are too complex and subjective to be applied consistently by coding specialists or physicians.Therefore, rigorous precision for determining the correct code is probably not achievable.
This study had several limitations. Only the lead investigator observed the clinic visits, so personal biases could have affected the collection and interpretation of the observed behaviors. However, having only 1 observer eliminates inter-rater variability in the study. A previous study of family physicians in practice audio-recorded all patient encounters, which were later analyzed by several investigators and research assistants.Nonetheless, when comparing those study results with those of the current study, little difference was noted, suggesting limited bias effects.
Time results were kept to the minute, with only 1 person (RY) recording these data. In contrast, the other large observational study broke down the audio recordings to 15-second intervals.However, a strength of this study is that all time related to each patient encounter was captured, including documentation time. Concerns about using time as a measure for work were discussed previously. Another major limitation was the fact that many of the subjects finished documenting their patient encounters at home at night. The time required for this work was estimated, not observed.
Although all efforts were made to minimize the impact of the observations on physician or patient behaviors, a Hawthorne effect cannot be ruled out. Generalizability of the results may be limited, as the practices represented in the current study may not represent practices and behaviors in other regions. Finally, the observational nature of the study precludes definitive statements of causality.
Implications and Future Work
Our findings show that the existing CMS E/M rules and resulting conventional payments simply do not support very well the work of family physicians, especially for longer and more complex patient visits. New documen- tation and payment approaches should be explored with the aim of replacing the existing system.
Perhaps different versions of blended payment could prove successful. Phrases such as “pay for value, not volume” are commonly heard in payment reform conversations. The American Academy of Family Physicians has certainly invested much time and effort into the patient-centered medical home (PCMH) concept.Some demonstration projects have found success applying elements of the PCMH,though a recent review casts doubts on the ability of the PCMH model, as currently structured, to reduce hospitalizations and overall healthcare costs.
Our findings suggest that a more profound and disruptive innovation for primary care payment is required to reduce the current primary care/specialist income disparity and to increase the supply of family physicians by attracting more medical students into the field, therefore creating a US healthcare system that provides better care at a lower cost than the current one.
Portions of this proposal have been presented at the CMS Innovation Advisors Meeting, Baltimore, MD, September 2012; John Peter Smith Research Round Table, Fort Worth, TX, September 2012; University of Louisville Medical School Family Medicine/Geriatrics Grand Rounds, Louisville, KY, February 2013; American Academy of Family Physicians Commission on Quality and Practice, Kansas City, MO, February 2013; Health Resources and Services Administration regional office, Dallas, TX, February 2013; Family Medicine Leaders in Large Multispecialty Groups conference, Grapevine, TX, March 2013; North American Primary Care Research Group, Ottawa, ONT, CAN, November 2013.
John Peter Smith Hospital Family Medicine Residency Program, Fort Worth, TX (RY); Department of Trauma Services, John Peter Smith Health Network, Fort Worth, TX (TLO).
Source of Funding:
This project was supported through Dr Young’s participation in the CMS Innovation Advisors Program.
Dr Young’s institution was paid a stipend for his participation in the CMS Innovation Advisor Program. Ms Overton reports no conflicts of interest.
Concept and design (RY); acquisition of data (RY); analysis and interpretation of data (RY, TLO); drafting of the manuscript (RY); critical revision of the manuscript for important intellectual content (RY); statistical analysis (RY, TLO); provision of study materials or patients (RY); obtaining funding (RY); and administrative, technical, or logistic support (RY).
Address correspondence to:
Richard Young, MD, Director of Research, John Peter Smith Hospital FMRP, 1500 S Main St, Fort Worth, TX 76104. E-mail: email@example.com.
1. Arvantes J. Senate hearing links physician payment rates to primary care shortages. http://drdavelynch.blogspot.com/2008_02_01_archive .html. Accessed November 4, 2014.
AAFP News Now.
2. Arvantes J. MedPAC members characterize RBRVS system as subjective, ‘deeply flawed.’ November 3, 2009.
CPT 2014 Standard Edition (Current ProceduralTerminology).
3. American Medical Association. Chicago, IL: American Medical Association; 2014.
Arch Intern Med.
4. Leigh JP,Tancredi D, Jerant A, Kravitz RL. Annual work hours across physician specialties. 2011;171(13):1211-1213.
5. Mean and median 2012 individual income (before taxes, in thousands of dollars) of family physicians. American Academy of Family Physicians website. http://www.aafp.org/about/the-aafp/family-medicine-facts/table-4.html. Updated 2013. Accessed November 4, 2014.
6. Japsen B. Doctor pay rises to $221K for primary care, $396K for specialists. Forbes. June, 12, 2013. http://www.forbes.com/sites/bruce-japsen/2013/06/12/doctorpay-rises-to-221k-for-primary-care-396k-for-specialists/. Accessed November 4, 2014.
7. Berenson RA, Zuckerman S, Stockley K, Nath R, Gans D, Hammons T. What if all physician services were paid under the Medicare fee schedule? an analysis using Medical Group Management Association data. Washington, DC: Medicare Payment Advisory Commission; 2010. www.urban.org/UploadedPDF/412051_physcian_service.pdf. Accessed October 9, 2012.
8. CMS. 1997 Documentation Guidelines for Evaluation and Management Services. http://cms.gov/Outreach-and-Education/Medicare- Learning-Network MLN/MLNEdWebGuide/EMDOC.html.
9. Medicare Learning Network: Evaluation and Management Services Guide. CMS website. http://cms.gov/Outreach-and-Education/Medi- care-Learning-Network-MLN/MLNEdWebGuide/EMDOC.html. Updated December 2010. Accessed January 12, 2012.
N Engl J Med.
10. Brett AS. New guidelines for coding physicians’ services--a step backward. 1998;339(23):1705-1708.
11. Braun P, Hsiao WC, Becker ER, DeNicola M. Evaluation and management services in the Resource-Based Relative Value Scale. 1988;260(16):2409-2417.
J Gen Intern Med.
12. Goroll AH, Berenson RA, Schoenbaum SC, Gardner LB. Fundamental reform of payment for adult primary care: comprehensive payment for comprehensive care. 2007;22(3):410-415.
Health Aff (Millwood).
13. Reid RJ, Coleman K, Johnson EA, et al.The Group Health medical home at year two: cost savings, higher patient satisfaction, and less burnout for providers. 2010;29(5):835-843.
Ann Intern Med.
14. Davidoff F.Time. 1997;127(6):483-485.
15. Andersson SO, Mattsson B. Length of consultations in general practice in Sweden: views of doctors and patients. 1989; 6(2):130-134.
J Fam Pract.
16. Stange KC, Zyzanski SJ, Jaén CR, et al. Illuminating the ‘black box’. a description of 4454 patient visits to 138 family physicians. 1998;46(5):377-389.
17. Anderson G. Chronic care: making the case for ongoing care. Robert Wood Johnson Foundation website. http://www.rwjf.org/content/ dam/farm/reports/reports/2010/rwjf54583. Published 2010. Accessed November 4, 2014.
Ann Fam Med.
18. Beasley JW, HankeyTH, Erickson R, et al. How many problems do family physicians manage at each encounter? a WReN study. 2004;2(5):405-410.
J Fam Pract.
Health Serv Res.
19. Flocke SA, Frank SH, Wenger DA. Addressing multiple problems in the family practice office visit. 2001;50(3):211-216. 20.Tai-Seale M, McGuireTG, ZhangW.Time allocation in primary care office visits. 2007;42(5):1871-1894.
Ann Fam Med.
21. Fenton JJ, Von Korff M, Lin EH, Ciechanowski P, Young BA. Quality of preventive care for diabetes: effects of visit frequency and competing demands. 2006;4(1):32-39.
J Gen Intern Med.
22. Linzer M, KonradTR, Douglas J, et al. Managed care, time pressure, and physician job satisfaction: results from the physician worklife study. 2000;15(7):441-450.
Soc Sci Med.
23. Groenewegen PP, Hutten JB. Workload and job satisfaction among general practitioners: a review of the literature. 1991; 32(10):1111-1119.
Ann Intern Med.
24. Linzer M, Manwell LB, Williams ES, et al; MEMO (Minimizing Error, Maximizing Outcome) Investigators. Working conditions in primary care: physician reactions and care quality. 2009;151(1):28-36, W6-W9.
Cochrane Database Syst Rev.
25. Wilson A, Childs S. Effects of interventions aimed at changing the length of primary care physicians’ consultation. 2006;1:CD003540.
J Eval Clin Pract.
26. van Walraven C, Oake N, Jennings A, Forster AJ. The association between continuity of care and outcomes: a systematic and critical review. 2010;16(5):947-956.
Soc Sci Med.
27. Camasso MJ, Camasso AE. Practitioner productivity and the product content of medical care in publicly supported health centers. 1994;38(5):733-748.
J Fam Pract.
28. Zyzanski SJ, Stange KC, Langa D, Flocke SA.Trade-offs in high-volume primary care practice. 1998;46(5):397-402.
29. Field S. Leading the way. High-quality care for all through general practice. A manifesto for patient care 2010. Royal College of General Practitioners website. http://www.bgs.org.uk/index.php?option=com_content&view=article&id=479:rcgp-manifesto-lead- ing-the-way-high-quality-care-for-all-through-general-practice&catid=6 :prindex&Itemid=99 Updated. Accessed November 4th, 2014.
Arch Intern Med.
30. Schwartz MD, Durning S, Linzer M, Hauer KE. Changes in medical students’ views of internal medicine careers from 1990 to 2007. 2011;171(8):744-749.
J Fam Pract.
31. Chao J, Gillanders WG, Flocke SA, Goodwin MA, Kikano GE, Stange KC. Billing for physician services: a comparison of actual billing with CPT codes assigned by direct observation. 1998;47(1):28-32.
Arch Intern Med.
32. King MS, Lipsky MS, Sharp L. Expert agreement in Current Proce- duralTerminology evaluation and management coding. 2002;162(3):316-320.
J Fam Pract.
33. Zuber TJ, Rhody CE, Muday TA, et al. Variability in code selection using the 1995 and 1998 HCFA documentation guidelines for office services. Health Care Financing Administration. 2000; 49(7):642-645.
J Am Board Fam Pract.
34. King MS, Sharp L, Lipsky MS. Accuracy of CPT evaluation and management coding by family physicians. 2001;14(3):184-192.
South Med J.
35. Holt J, Warsy A, Wright P. Medical decision making: guide to improved CPT coding. 2010;103(4):316-322.
J Fam Pract.
36. Horner RD, Paris JA, Purvis JR, Lawler FH. Accuracy of patient encounter and billing information in ambulatory care. 1991;33(6):593-598.
37. American Academy of Family Physicians (AAFP), American Academy of Pediatrics (AAP), American College of Physicians (ACP), American Osteopathic Association (AOA). Joint principles of the patient-centered medical home. American Academy of Family Physicians website. http://www.aafp.org/dam/AAFP/documents/practice_ management/pcmh/initiatives/PCMHJoint.pdf. Updated February 2007. Accessed November 4, 2014.
38. Looking to achieve NCQA recognition?TransforMed website. http:// www.transformed.com/PPC/NCQA/PPC_ncqa.cfm?gclid=COaUifTH2ao CFUwS2godxmTx-Q. Accessed Aug 18, 2011.
39. Grumbach K, Grundy P. Outcomes of implementing patient centered medical home interventions: a review of the evidence from prospective evaluation studies in the United States. Patient-Centered Primary Care Collaborative website. http://forwww.pcpcc.net/files/ evidence_outcomes_in_pcmh_2010.pdf. Updated November 16, 2010. Accessed Accessed November 4, 2014.
Ann Intern Med.
40. Jackson GL, Powers BJ, Chatterjee R, et al. Improving patient care: the patient-centered medical home: a systematic review. 2013;158(3):169-178.