Letters

intervention, residents were aware that albuterol neb treat- Analysis and interpretation of data: Moriates, Novelero, Khanna, Mourad. ments were more expensive than albuterol MDIs (82%, pre- Drafting of the manuscript: Moriates, Novelero. Critical revision of the manuscript for important intellectual content: Quinn, test; 94%, postintervention [P = .11]). Prior to the interven- Khanna, Mourad. tion 13 of the residents (26%) answered incorrectly that neb Statistical analysis: Khanna. treatments were more efficacious than MDIs, in contrast to only Obtained funding: Novelero. 1 resident (3%) following exposure to our intervention (P < .01). Administrative, technical, and material support: Novelero, Quinn, Mourad. Study supervision: Khanna, Mourad. At baseline, none of the residents agreed that “patients re- Published Online: July 22, 2013. ceive adequate inpatient MDI teaching”; however, this rate im- doi:10.1001/jamainternmed.2013.9002. proved to 16% after the first 2 months of implementation Conflict of Interest Disclosures: None reported. (P < .01). Additional Contributions: Theodore Omachi, MD, MBA (Department of Medicine, University of California, San Francisco), and Sumant Ranji, MD Discussion | Our multifaceted intervention was associated (Department of Medicine, University of California, San Francisco), contributed with a simultaneous decrease in unnecessary neb treat- to the design and implementation of this project. They did not receive compensation. ments, an increase in evidence-based resident physician 1. Turner MO, Patel A, Ginsburg S, FitzGerald JM. Bronchodilator delivery in knowledge, and potentially an improvement in MDI patient acute airflow obstruction: a meta-analysis. Arch Intern Med. 1997;157(15):1736- education. This concurrent improvement in quality of care 1744. with a decrease in cost maximizes the “value equation” 2. Dolovich MB, Ahrens RC, Hess DR, et al; American College (defined as quality divided by costs). The approximately of Chest Physicians; American College of Asthma, , and 50% decrease in nebs following our intervention highlights . Device selection and outcomes of aerosol therapy: evidence-based guidelines: American College of Chest the degree of wasteful usage of this resource-intensive Physicians/American College of Asthma, Allergy, and Immunology. Chest. therapy previously on our pilot medical ward. Reducing 2005;127(1):335-371. inappropriate nebs represents a straightforward way for 3. Mandelberg A, Chen E, Noviski N, Priel IE. Nebulized wet aerosol treatment institutions to reduce health care costs through a simple in emergency department: is it essential? comparison with large spacer device intervention. for metered-dose inhaler. Chest. 1997;112(6):1501-1505. Our study has some limitations. Owing to the nature of 4. Press VG, Arora VM, Shah LM, et al. Misuse of respiratory inhalers in hospitalized patients with asthma or COPD. J Gen Intern Med. our intervention and the significant crossover of our physi- 2011;26(6):635-642. cians, RTs, and nurses, it was not possible to create a control group at our medical center during this pilot study. Also, our financial model may overestimate our cost savings since RT Documentation and Diagnosis of Overweight time is a semifixed cost and our intervention has not yet led and Obesity in Electronic Health Records to a decrease in actual RT full-time equivalents. However, of Adult Primary Care Patients RT daily staffing is based on volume at our large hospital; Almost 69% of US adults are either overweight or obese thus, if the project is successfully scaled medical center– (body mass index [BMI], calculated as weight in kilograms wide, then it would likely result in a decrease in daily divided by height in meters squared, ≥25),1 yet clinicians RT staffing. Currently, this saved time is being repurposed often fail to diagnose overweight and obesity or discuss for our RTs to perform other important job duties at our weight management with their patients.2-6 Many clinicians hospital, such as MDI training and smoking cessation use electronic health records (EHRs), and adoption of EHRs counseling. has been increasing since the introduction of the Health In conclusion, our pilot study illustrates that a multifac- Information Technology for Economic and Clinical Health eted effort may be successful in dramatically decreasing the (HITECH) Act in 2009.7 Electronic recording of vital signs— overuse of neb therapies on an inpatient medicine service. Re- including height, weight, and BMI—is now one of the ducing utilization of these resource intensive and unneces- requirements for achieving “meaningful use” of EHRs,8 but sary treatments may provide an ideal target for improving few studies have examined rates of BMI documentation and health care value. diagnosis of overweight and obesity in EHR data. We con- ducted a retrospective study to examine these rates in the Christopher Moriates, MD EHRs of adult primary care patients before the passing of Maria Novelero, MA, MPA the HITECH Act in 2009. Kathryn Quinn, MPH Raman Khanna, MD Methods | We evaluated patients at 25 primary care practices Michelle Mourad, MD within a large academic care network in Boston, Massachu- setts. We included adult patients (≥18 years) who had at Author Affiliations: Department of Medicine, University of California, San least 2 visits with the same clinician between 2004 and Francisco, San Francisco (Moriates, Novelero, Quinn, Khanna, Mourad). 2008 and were not pregnant at the time of the visit. The Corresponding Author: Christopher Moriates, MD, Department of Medicine, study was approved by the Partners Human Research University of California, San Francisco, 505 Parnassus Ave, M1287, San Committee. Francisco, CA 94143-0131 ([email protected]). Data were extracted from coded fields in the EHR. The Author Contributions: Study concept and design: Moriates, Novelero, Quinn, Mourad. primary outcome was documentation of at least 1 BMI in the Acquisition of data: Moriates, Novelero. appropriate coded EHR field at any time during the study

1648 JAMA September 23, 2013 Volume 173, Number 17 jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/29/2021 Letters

Table 1. Documentation of Body Mass Index (BMI) in the Electronic Health Records of 219 356 Adult Primary Care Patients

No. in ≥1 BMI, Adjusted OR Characteristic Group No. (%) (95% CI) P Value Overall 144 522 (65.9) Patient Characteristics Age at first visit, y 18-29 42 170 27 934 (66.2) 1 [Reference] 30-39 46 382 29 851 (64.4) 0.93 (0.90-0.97) 40-49 45 156 30 427 (67.4) 1.04 (1.00-1.08) <.001 50-59 38 822 26 438 (68.1) 1.00 (0.96-1.04) 60-69 24 877 16 729 (67.3) 0.94 (0.89-0.99) ≥70 21 949 13 143 (59.9) 0.60 (0.56-0.63) Sex Male 81 742 48 233 (59.0) 1 [Reference] <.001 Female 137 614 96 289 (70.0) 1.45 (1.41-1.48) Race/ethnicity White 145 391 96 081 (66.1) 1 [Reference] Black 17 814 11 851 (66.5) 1.01 (0.97-1.05) Hispanic/Latino 29 432 19 935 (67.7) 1.05 (1.00-1.09) <.001 Asian 8885 6033 (67.9) 0.97 (0.92-1.03) Other or missing 17 834 10 622 (59.6) 0.84 (0.81-0.88) Primary insurance Private 167 479 111 906 (66.8) 1 [Reference] Medicare 36 734 23 759 (64.7) 0.94 (0.90-0.98) <.001 Medicaid 5764 3979 (69.0) 0.93 (0.87-1.00) No insurance or self-pay 9379 4878 (52.0) 0.64 (0.61-0.68) Visits, No. 2-5 75 868 41 572 (54.8) 1 [Reference] 6-9 52 940 35 002 (66.1) 1.87 (1.81-1.92) <.001 10-14 38 295 27 491 (71.8) 2.78 (2.68-2.87) ≥15 52 253 40 457 (77.4) 4.66 (4.50-4.83) Obesity-related comorbidities, No. 0 109 051 65 300 (59.9) 1 [Reference] 1 52 549 36 603 (69.7) 1.34 (1.30-1.38) <.001 2 30 213 21 844 (72.3) 1.48 (1.42-1.53) ≥3 27 543 20 775 (75.4) 1.73 (1.66-1.80) Clinician Characteristicsa Age, y <30 20 336 14 042 (69.1) 1 [Reference] 30-39 76 896 52 394 (68.1) 1.04 (0.88-1.24) 40-49 59 452 39 948 (67.2) 1.14 (0.91-1.41) .23 50-59 49 581 30 319 (61.2) 1.10 (0.88-1.37) ≥60 12 462 7475 (60.0) 0.77 (0.52-1.15) Sex Male 100 487 60 247 (60.0) 1 [Reference] .01 Female 118 869 84 275 (70.9) 1.20 (1.04-1.38) Abbreviations: NP, nurse practitioner; Type OR, odds ratio; PA, physician Staff physician 188 273 124 744 (66.3) 1 [Reference] assistant. a NP or PA 20 307 13 058 (64.3) 0.88 (0.84-0.92) <.001 Counts refer to the number of patients who had visits with clinicians Resident or fellow 10 776 6720 (62.4) 1.02 (0.95-1.08) with these characteristics.

period. Body mass index is calculated if patients have both patients with at least 1 BMI of at least 25 (overweight) or at height and weight; once a height has been entered, it is car- least 30 (obese), we also examined whether they had a diag- ried forward and used in subsequent calculations. Among nosis of “overweight,” “obesity,” “weight gain,” or “weight

jamainternalmedicine.com JAMA Internal Medicine September 23, 2013 Volume 173, Number 17 1649

Downloaded From: https://jamanetwork.com/ on 09/29/2021 Letters

Table 2. Diagnosis of Overweight and Obesity Among 98 762 Adult Primary Care Patients With BMI of at Least 25

Diagnosis of Overweight/Obesity No. in on Problem List, Adjusted OR Characteristic Group No. (%) (95% CI) P Value Overall 98 762 16 926 (17.1) Patient Characteristics Age at first visit, y 18-29 15 336 2347 (15.3) 1 [Reference] 30-39 19 309 3416 (17.7) 1.00 (0.93-1.07) 40-49 21 845 4107 (18.8) 0.86 (0.80-0.92) <.001 50-59 19 862 3910 (19.7) 0.76 (0.70-0.82) 60-69 12 913 2140 (16.6) 0.63 (0.57-0.69) ≥70 9497 1006 (10.6) 0.41 (0.37-0.46) Sex Male 39 303 5279 (13.4) 1 [Reference] <.001 Female 59 459 11 647 (19.6) 1.48 (1.41-1.55) Race/ethnicity White 63 554 11 013 (17.3) 1 [Reference] Black 9601 2002 (20.9) 0.91 (0.84-0.98) Hispanic/Latino 16 247 2739 (16.9) 0.89 (0.82-0.96) <.001 Asian 2641 198 (7.5) 0.67 (0.57-0.80) Other or missing 6719 974 (14.5) 1.03 (0.94-1.12) Primary insurance Private 74 217 12 993 (17.5) 1 [Reference] Medicare 18 061 2822 (15.6) 0.88 (0.82-0.94) <.001 Medicaid 3162 591 (18.7) 0.82 (0.73-0.93) No insurance or self-pay 3322 520 (15.7) 0.99 (0.88-1.12) Visits, No. 2-5 24 659 3065 (12.4) 1 [Reference] 6-9 22 576 3526 (15.6) 1.09 (1.02-1.16) .08 10-14 19 309 3397 (17.6) 1.04 (0.97-1.11) ≥15 32 218 6938 (21.5) 1.03 (0.97-1.10) Obesity-related comorbidities, No. 0 37 147 3716 (10.0) 1 [Reference] 1-2 26 513 4575 (17.3) 1.62 (1.53-1.71) <.001 3-4 17 609 3897 (22.1) 1.92 (1.80-2.05) ≥5 17 493 4738 (27.1) 2.24 (2.10-2.40) Highest BMI <27 21 537 434 (2.0) 1 [Reference] 27-29.9 27 765 1530 (5.5) 3.09 (2.76-3.46) 30-34.9 27 819 4965 (17.9) 13.23 (11.91-14.70) <.001 35-39.9 12 468 4590 (36.8) 40.31 (36.15-44.96) ≥40 9173 5407 (58.9) 110.85 (99.04-124.08)

(continued)

management” on the EHR problem list at any time during practices. All analyses were conducted using SAS statistical the study period. software (version 9.3). We computed frequencies of patient and clinician char- acteristics, documentation of BMI, and diagnosis of over- Results | A total of 219 356 patients were included in the weight and obesity. Multivariate logistic regression models with analysis. The average age of patients at their first visit dur- generalized estimating equations were used to examine asso- ing the study period was 45.7 years, and the median number ciations of patient and clinician characteristics with documen- of visits per patient was 8. Among these patients, 65.9% had tation of BMI and diagnosis of overweight and obesity, ac- at least 1 BMI in the EHR (Table 1). Almost all of the missing counting for clustering within health care providers and BMI information was due to missing data on height; only

1650 JAMA Internal Medicine September 23, 2013 Volume 173, Number 17 jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/29/2021 Letters

Table 2. Diagnosis of Overweight and Obesity Among 98 762 Adult Primary Care Patients With BMI of at Least 25 (continued)

Diagnosis of Overweight/Obesity No. in on Problem List, Adjusted OR Characteristic Group No. (%) (95% CI) P Value Clinician Characteristicsa Age, y <30 9467 1747 (18.5) 1 [Reference] 30-39 36 094 6403 (17.7) 1.11 (0.93-1.33) 40-49 25 728 4207 (16.4) 1.01 (0.82-1.25) .38 50-59 21 965 3712 (16.9) 1.07 (0.87-1.34) ≥60 5270 817 (15.5) 1.45 (1.01-2.09) Sex Male 45 373 7203 (15.9) 1 [Reference] Abbreviations: BMI, body mass index .79 (calculated as weight in kilograms Female 53 389 9723 (18.2) 0.98 (0.85-1.13) divided by height in meters squared); Type NP, nurse practitioner; OR, odds ratio; Staff physician 85 212 14 640 (17.2) 1 [Reference] PA, physician assistant. a NP or PA 8655 1475 (17.0) 0.96 (0.89-1.04) .002 Counts refer to the number of patients who had visits with clinicians Resident or fellow 4895 811 (16.6) 0.81 (0.72-0.91) with these characteristics.

66.0% had at least 1 height in the EHR, whereas 90.6% had In conclusion, many primary care patients lack documen- at least 1 weight in the EHR. Factors that were associated tation of BMI in the EHR, and most overweight and obese pa- with documentation of BMI are shown in Table 1. Among tients do not have a diagnosis on the problem list. Further re- patients with BMI in the EHR, 68.3% had at least 1 BMI of at search should focus on interventions to improve least 25, and 34.4% had at least 1 BMI of at least 30. Of documentation of BMI and diagnosis and management of over- patients with a BMI of at least 25 or BMI of at least 30, 17.1% weight and obesity in the primary care setting. and 30.1%, respectively, had a diagnosis on their problem list. Factors that were associated with a diagnosis of over- Heather J. Baer, SD weight and obesity are shown in Table 2. Andrew S. Karson, MD, MPH Jane R. Soukup, MSc Discussion | Approximately one-third of adult primary care Deborah H. Williams, MHA patients in this population had no BMI in the EHR. Further- David W. Bates, MD, MSc more, very few overweight or obese patients had a diagnosis on the problem list. Our findings are consistent with those Author Affiliations: Division of General Internal Medicine and Primary Care, of previous studies showing that overweight and obesity are Brigham and Women’s Hospital, Boston, Massachusetts (Baer, Soukup, Williams, Bates); Harvard Medical School, Boston, Massachusetts (Baer, Karson, poorly documented and diagnosed by primary care Bates); Department of Epidemiology, Harvard School of Public Health, Boston, 2-6 clinicians, but few prior studies have focused specifically Massachusetts (Baer); Clinical Decision Support Unit, Department of Medicine, on documentation and diagnosis of overweight and obesity Massachusetts General Hospital, Boston (Karson); Partners HealthCare, Boston, in electronic health records. In a study by Rose et al,9 the Massachusetts (Bates); Department of Health Policy and Management, Harvard School of Public Health, Boston, Massachusetts (Bates). rates of height, weight, and BMI documentation in the EHRs Corresponding Author: Heather J. Baer, SD, Division of General Internal of primary care patients were very similar to those observed Medicine and Primary Care, Brigham and Women’s Hospital, 1620 Tremont St, in our study, but they did not examine diagnosis of over- Boston, MA 02120 ([email protected]). 3 weight and obesity. In another study, Bardia et al found Published Online: July 8, 2013. that19.9%ofprimarycarepatientswithaBMIofatleast30 doi:10.1001/jamainternmed.2013.7815. had obesity documented as a diagnosis in the EHR, which is Author Contributions: Dr Baer had full access to all of the data in the study and lower than in our study. However, to be included in our takes responsibility for the integrity of the data and the accuracy of the data analysis. study population, patients had to have seen the same pri- Study concept and design: Baer, Karson, Bates. mary care clinician at least twice during the 5-year study Acquisition of data: Baer, Karson. period. Analysis and interpretation of data: Baer, Karson, Soukup, Williams, Bates. Our study has several limitations. The analyses were Drafting of the manuscript: Baer. Critical revision of the manuscript for important intellectual content: Baer, cross-sectional, and we examined documentation of BMI Karson, Soukup, Williams, Bates. and diagnosis of overweight and obesity only in coded EHR Statistical analysis: Baer, Soukup. fields. In addition, we did not examine management of Obtained funding: Baer. Administrative, technical, and material support: Baer, Karson, Williams, Bates. overweight and obesity because this is not typically docu- Study supervision: Baer, Bates. mented in coded fields. However, in the study by Bardia et Conflict of Interest Disclosures: Dr Bates is a coinventor on patent No. 3 al, diagnosis of obesity was a strong predictor of formula- 6029138, held by Brigham and Women’s Hospital, on the use of decision tion of an obesity plan. support software for medical management, licensed to the Medicalis Corp. He

jamainternalmedicine.com JAMA Internal Medicine September 23, 2013 Volume 173, Number 17 1651

Downloaded From: https://jamanetwork.com/ on 09/29/2021 Letters

holds a minority equity position in the privately held company Medicalis, which a develops web-based decision support for radiology test ordering. He serves on Table. Key Program Elements the board for SEA Medical Systems, which makes intravenous pump Element Implementation Month technology. He serves as an advisor to Calgary Scientific, which makes Discharge summary template in the July (month 1 of program) technologies that enable mobility within EHRs. He is on the clinical advisory electronic medical record board for Zynx Inc, which develops evidence-based algorithms, and Patient Safety Systems, which provides a set of approaches to help hospitals improve We incentivized 2 template fields: (1) whether the patient expressed safety. He is a consultant for EarlySense, which makes patient safety monitoring wishes for care during the systems. hospitalization and (2) whether the Funding/Support: Dr Baer was supported by a Mentored Research Scientist patient identified a surrogate decision maker; additional fields Career Development Award from the Agency for Healthcare Research and allowed residents to enter detailed Quality (K01HS019789). Some funding for this work also was provided by the information Eleanor and Miles Shore 50th Anniversary Fellowship for Scholars in Medicine Financial incentive program July (month 1 of program) from Harvard Medical School. For residents to receive the $400 Role of the Sponsors: The funding organization had no role in the design and financial incentive, documentation conduct of the study; in the collection, analysis, and interpretation of the data; of the 2 incentivized fields was or in the preparation, review, or approval of the manuscript. required within 48 h of discharge for 75% of patients in at least 3 of 1. Flegal KM, Carroll MD, Kit BK, Ogden CL. Prevalence of obesity and trends in 4 quarters of the year the distribution of body mass index among US adults, 1999-2010. JAMA. Audit and feedback Varied (see below) 2012;307(5):491-497. General information about the July (month 1 of program) 2. Abid A, Galuska D, Khan LK, Gillespie C, Ford ES, Serdula MK. Are healthcare program, including targets and professionals advising obese patients to lose weight? a trend analysis. requirements, in monthly reminder MedGenMed. 2005;7(4):10. e-mails and at education sessions 3. Bardia A, Holtan SG, Slezak JM, Thompson WG. Diagnosis of obesity by Feedback at the team level in August (month 2 of program) biweekly e-mails primary care physicians and impact on obesity management. Mayo Clin Proc. 2007;82(8):927-932. Standardized graphical and tabular November (month 5 of program) data comparing overall, team, and 4. Ko JY, Brown DR, Galuska DA, Zhang J, Blanck HM, Ainsworth BE. Weight individual residents’ loss advice U.S. obese adults receive from health care professionals. Prev Med. documentation rates in biweekly 2008;47(6):587-592. e-mails 5. Ma J, Xiao L, Stafford RS. Underdiagnosis of obesity in adults in US a Our intervention included 3 aspects: (1) discharge summary template, (2), outpatient settings. Arch Intern Med. 2009;169(3):313-314. financial incentive, and (3) staged performance feedback to residents. 6. McAlpine DD, Wilson AR. Trends in obesity-related counseling in primary care: 1995-2004. Med Care. 2007;45(4):322-329. 7. American Recovery and Reinvestment Act: HR1 2009. http://www.gpo cial incentive. Internal Medicine residents selected the goal of .gov/fdsys/pkg/BILLS-111hr1enr/pdf/BILLS-111hr1enr.pdf. Accessed October 17, improving documentation of advance care planning discus- 2012. sions on the basis of pilot work and experience that inconsis- 8. Blumenthal D, Tavenner M. The “meaningful use” regulation for electronic tent documentation was a barrier to honoring patients’ wishes health records. N Engl J Med. 2010;363(6):501-504. on transitions of care. Input from key stakeholders, including 9. Rose SA, Turchin A, Grant RW, Meigs JB. Documentation of body mass index emergency department, outpatient, hospital, and palliative care and control of associated risk factors in a large primary care network. BMC Health Serv Res. 2009;9:236. providers, informed the intervention, especially location and content of documentation. The project included 3 key ele- ments; the details of each of these are included in the Table. Incentivizing Residents to Document Inpatient To assess documentation rates, program residents reviewed Advance Care Planning charts of a random sample of recently discharged patients on Discussing preferences for care near the end of life increases a biweekly basis. The UCSF institutional review board ap- the likelihood that patients will receive care consistent with proved the project. their preferences.1-4 Recent work5 demonstrates that medi- cal professionals infrequently ask about and document pref- Results | The Figure shows implementation of key aspects of erences for patients upon hospitalization. Because most end- the intervention and the percentage of discharge summaries of-life discussions occur in hospitals,6 we implemented a that included the required documentation by project month. quality improvement program incentivizing resident physi- Documentation rates are based on medical record review of cians to consistently document key information about inpa- 1474 patients, comprising 55.5% of those discharged from the tient advance care planning discussions in a timely manner in medical service during the project period. Rates rose from an accessible location. 22.2% at the beginning of the program to more than 90% by October and remained near this level through May. In com- Methods | We conducted the project between July 1, 2011, and parison, documentation rates for patients discharged from an May 31, 2012, on the medical service at the University of Cali- attending-only service, which used the electronic template but fornia, San Francisco (UCSF), where the Medical Center and did not receive the financial incentive or feedback, were 0% departments of medicine and graduate medical education col- to 50% with a yearly mean of 11.7%. laborated to form the Housestaff Incentive Program. In this pro- gram, trainees choose quality improvement goals and faculty Discussion | We implemented a multifaceted intervention to im- mentor trainee champions through design and implementa- prove resident documentation of advance care planning dis- tion of projects. If goals are met, all trainees receive a finan- cussions in a consistent format and location. We believe that

1652 JAMA Internal Medicine September 23, 2013 Volume 173, Number 17 jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/29/2021