The Role of Health Systems in Reducing Tobacco Dependence

June 11, 2019

Addressing tobacco use is an important health system role. This process evaluation discusses facilitators and barriers to implementing systems changes to improve tobacco treatment delivery.

ABSTRACT

Objectives: Health systems play an important role in addressing tobacco use. Research indicates that implementing systems changes in clinical settings may lead to greater rates of tobacco treatment delivery and reductions in tobacco use prevalence compared with clinics and health systems that do not implement such changes. Few studies have described facilitators and barriers to implementing these changes.

Study Design: A process evaluation was conducted of 5 Minnesota health systems that implemented multiple systems changes to make tobacco treatment delivery a standard of care. Three large integrated health systems (1 in the Twin Cities metropolitan area, 1 in northern Minnesota, and 1 in central Minnesota), a federally recognized Minnesota Chippewa Tribe, and a safety net dental practice were evaluated.

Methods: An external evaluator conducted 3 waves of key informant interviews with each system. Purposive sampling was used to select key informants from each health system. A total of 49 interviews among 30 staff were conducted. Project documents were also reviewed. Evaluators used both deductive and inductive approaches to identify cross-cutting themes.

Results: Several facilitators were identified, including using a team-based approach to engage staff, implementing new protocols and training staff, and utilizing tools such as electronic health records and data to conduct quality improvement initiatives. Barriers included delays in electronic health record changes and keeping tobacco treatment prioritized in the organization.

Conclusions: Health systems change can provide a renewed sense of enthusiasm and ownership of tobacco treatment among providers and staff and can be an effective way to help prioritize addressing tobacco use.

The American Journal of Accountable Care. 2019;7(2):4-11Tobacco use screening and brief intervention is 1 of the top 3 preventive services in terms of cost savings and potential population health improvement.1,2 Health systems play an important role in addressing tobacco use. Seventy-five percent of current smokers report visiting a healthcare provider in the past year.3 The majority of smokers want their healthcare provider to address their smoking,4,5 and satisfaction with care is highest among smokers who receive cessation assistance or follow-up.4,6

The US Public Health Service (PHS) Clinical Practice Guideline, Treating Tobacco Use and Dependence, recommends implementing the 5 As (Ask, Advise, Assess, Assist, and Arrange) to systematically address tobacco use.7,8 Implementing the 5 As, including brief interventions by providers, is associated with greater tobacco cessation efforts among patients compared with no intervention9; research also demonstrates that brief advice from a physician increases successful quitting.10 Nationally, healthcare providers consistently assess for tobacco use (90%) and frequently advise patients to quit (71%), but far fewer assist patients to quit (49%).4 A similar gap is seen in Minnesota.11

The PHS Clinical Practice Guideline and the CDC also recommend that clinics and health systems implement health systems changes to improve tobacco treatment delivery (eg, establishing a process to identify tobacco users, educating staff on tobacco treatment, providing resources and feedback to promote interventions).7,12 Evidence suggests that health systems change can improve care delivery processes compared with clinical settings where such changes were not implemented.7,13 Although the evidence is mixed regarding whether systems change improves cessation outcomes,13 some studies have shown that systems change in clinic settings can reduce the prevalence of tobacco use.8,9 However, few studies have examined factors that influence systems change implementation.14

ClearWay Minnesota, an independent nonprofit organization, released a competitive request for proposals to fund Minnesota healthcare systems for the implementation of health systems changes to more successfully address tobacco use. Applicants applied for up to $200,000 to fund a 2-year project; they were instructed to propose evidence-based strategies that aligned with their organization’s goals to better assess and address tobacco use. ClearWay Minnesota identified 3 areas of interest for applicants to consider: incorporating best practices for systems change, such as those outlined in the Clinical Practice Guideline7; optimizing their electronic health record (EHR); and using quality improvement processes. An expert review panel evaluated proposals and made funding recommendations. Funding decisions were made by ClearWay Minnesota’s Board of Directors. Three integrated health systems (1 in the Twin Cities metropolitan area, 1 in northern Minnesota, and 1 in central Minnesota), a federally recognized Minnesota Chippewa Tribe, and a safety net dental practice were funded (see Table 1 for health system characteristics).

We conducted a process evaluation to better understand facilitators and barriers to systems change implementation experienced by these diverse health systems. The insights reported in this paper can inform other systems change efforts.

METHODS

Study Design

Professional Data Analysts, an independent external evaluation firm, conducted the process evaluation. A qualitative approach, informed by Yin’s case study methodology15 and Patton’s qualitative design principles,16 was used to capture the complexity of the systems change process, as well as to gain insight on the facilitators, barriers, lessons learned, and potential sustainability of these changes. Intervention approaches differed across sites; examples included training staff and providers on delivering the 5 As, optimizing EHRs for clinical decision support and documentation, and creating standard workflows and procedures for identifying and treating tobacco users. All 5 health systems conducted their systems change activities over a 2-year period; 3 sites conducted activities from 2014 to 2016 and 2 sites from 2015 to 2017. A contracted technical assistance provider supported grantees on an as-needed basis.

Document review. ClearWay Minnesota provided the evaluators with key documents for each health system (eg, grant application, progress reports, meeting notes). Throughout the grant period and before each round of interviews, 2 evaluators independently reviewed all documents to inform interview protocol development.

Key informant interviews. The evaluators conducted 3 waves of semistructured interviews with key informants at each health system at the beginning, midpoint, and end of each 2-year grant period. Interview protocols were based on document review, previous systems change studies,4,8,9,14 tobacco control best-practice guidelines,12 and input from ClearWay Minnesota staff. Although each health system’s interview protocol was tailored to its project, all interviews were used to gather information about facilitators, barriers, lessons learned, and potential sustainability. Table 2 lists example interview questions.

Participants

Interviewees were selected through purposive sampling.16 ClearWay Minnesota staff and health system staff identified key informants within each system who were knowledgeable about the project, and evaluators invited them to participate by email. No participants declined an interview. Table 3 describes key informant characteristics.

Evaluators interviewed a minimum of 2 key informants from each health system during each interview wave. Interviews lasted 30 to 90 minutes; almost all were conducted face to face by 2 evaluators (1 primary, 1 secondary), but 2 interviews were conducted by phone. The primary evaluator was involved in all interviews; 1 of 2 other evaluators served as a secondary interviewer. A total of 49 interviews were conducted with 30 individuals across waves and across the 5 health systems (Table 3). All interviewees consented to have their interviews recorded. The evaluators created a detailed summary of each interview and sent it to the interviewees to review for completeness and accuracy. Subsequent corrections or additions from interviewees were incorporated into final summaries.

Data Analysis

After each interview wave, evaluators conducted content analysis of each interview summary, organizing the data into 4 a priori categories based on key lines of interview questioning: facilitators, barriers, lessons learned, and potential sustainability. After the last interview wave, evaluators used the organized summaries from all 5 sites to conduct a cross-site analysis to identify common themes within each of the 4 categories. Evaluators used both deductive and inductive approaches to identify themes across sites.16,17 The health systems change literature4,8,9,14 provided initial guidance for themes that might be identified during analysis. The primary evaluator identified common themes from the data, comparing data across the 5 health systems. These themes were then reviewed by the second evaluator. The 2 evaluators discussed any new themes or differences in interpretation until they reached consensus. Quotations or excerpts from interview notes and recordings were deidentified to protect the confidentiality of the individual and the health system.

The Minnesota Department of Health Institutional Review Board determined this study to be exempt from further review.

RESULTS

Facilitators, barriers, and lessons learned, as well as opportunities and challenges to sustaining systems change, are reported here.

Facilitators

Six facilitators of change were identified. Because each health system is unique, strategies varied based on the health system’s goals. Additional strategies are listed in Table 4.

“Having that buy-in from a leadership level all the way up to the CEO [chief executive officer] of the organization to say, ‘This is a priority, and we’re investing in it.’ ” — Project manager

Each grantee recognized the importance of building system-level support to elevate tobacco use as a priority and to leverage internal resources. Engaging organizational leaders by including them on project teams, as well as identifying clinic champions, helped to send a powerful signal across the organization of the importance of this work and to increase staff buy-in and enthusiasm for systems change.

“The [grant] Steering Committee is a really robust group of providers, nurse managers, data team members, community health staff, TTS [Tobacco Treatment Specialist] counselors, and primary care leadership.” — Project manager

“This team was instrumental in communication between the cessation program and clinic providers. Our clinic champion was key.” — Tobacco team member

A second facilitator was taking a team approach to implementing systems change. Although each project team was structured differently, all grantees engaged multiple levels of staff and providers to foster buy-in, as well as to design and implement new workflows and standard operating procedures. Some grantees also used surveys and meetings to obtain feedback from leadership and clinic staff to inform the design and implementation of changes. It was motivating for staff and providers to know that their input was valued. This also helped to ensure that workflows and standard operating procedures were aligned with clinic practices.

A third facilitator was capitalizing on internal and external priorities. Some grantees leveraged concurrent internal systemwide process change efforts and incorporated tobacco interventions into primary care workflows. One grantee worked with its Screening, Brief Intervention, and Referral to Treatment (SBIRT)18 trainer to incorporate tobacco dependence treatment education into existing SBIRT training.

“Get on your health system’s agenda to look at population health and total cost of care. Tobacco treatment affects many other areas, and it can be prioritized once you see how it impacts overall health and healthcare savings.” — Executive champion

A key external factor driving change was EHR Meaningful Use requirements. In 2011, CMS established the Meaningful Use incentive payment program to encourage eligible providers and hospitals to meet specific EHR criteria and reporting requirements.19 Multiple grantees used Meaningful Use measures (eg, clinical quality measures for diabetes and vascular care) to justify the need for tobacco-related systems change. Project staff presented to leadership and clinic staff, highlighting how tobacco dependence treatment was tied to multiple chronic disease outcomes; inpatient readmission rates; and other clinic, departmental, and systemwide goals. This information helped demonstrate how addressing tobacco use could improve performance on critical system priorities.

“I think the project manager’s ability to bring a group of resources together has been the greatest accomplishment, and [getting] the resources we need.” — Primary care director

Moreover, 1 grantee successfully made the case that tobacco use was highly related to system-level priorities, resulting in other departments and the health system’s charitable foundation providing additional financial support.

Implementing new protocols and training staff also facilitated change. Grantees embedded tobacco treatment into routine care by either modifying existing workflows or creating new standard operating procedures. Staff roles for implementing these processes were also defined.

“We developed a comprehensive half-day initial training session that was effective, and succeeded in motivating our clinical staff members to embrace our tobacco control protocol.” — Clinic staff

“The training of TTSs at the provider level has been integral for improving utilization rates and the delivery of evidence-based treatment. Training sparks a personal commitment to drive change within the clinical setting.” — Clinic staff

Training and retraining staff on new protocols was crucial to improve performance. In addition, 2 grantees provided existing staff with Tobacco Treatment Specialist (TTS) training. This specialized training builds knowledge and skills to treat tobacco dependence and to integrate evidence-based treatments into health systems.20 The TTSs became additional systems change champions, served as resources for providers and staff, and increased treatment delivery capacity.

“The EHR customized modifications were appropriate for supporting our standard operating procedures.” — Clinic staff

All grantees modified the EHR to facilitate systems change efforts. For example, some grantees included tobacco-specific templates within their EHR to allow providers and staff to more easily document patient tobacco use and refer patients to cessation resources.

“Dissemination of the clinic experience has garnered the attention and support of clinic quality leadership and regional/departmental sites for replication, enhancement, and the renewed spirit that process change can and will produce positive outcomes.” — Clinic staff

A final facilitator of change was monitoring data and providing feedback to staff to improve compliance with new protocols. Data helped identify additional training needs to improve processes. Multiple grantees created reports using tobacco-related EHR data. Sharing these reports with staff and leadership helped to generate interest in and support for systems change processes beyond a single department or clinic.

Barriers and Lessons Learned

One barrier was implementing tobacco-specific EHR changes. Many grantees had difficulty implementing these changes due to competing demands for information technology (IT) resources. Others were challenged by the amount of time that it took to make modifications. Working closely with IT staff/departments from the beginning of the project and obtaining leadership and management support helped prioritize the initial implementation of these changes.

Another barrier was keeping tobacco systems changes prioritized due to competing initiatives within the health system, finite staff time and resources, and project staff turnover. Many grantees overcame these challenges by capitalizing on other internal and external priorities, sharing data to help prioritize the work, and training existing staff members as TTSs to serve as an internal tobacco cessation resource.

A third barrier was informing all staff about new tobacco protocols and procedures. Grantees that were implementing systems changes in several clinics or departments found it challenging to communicate with all staff. Therefore, they used multiple communication methods (eg, the intranet, staff newsletters) to reach staff. Tobacco team members attended regular clinic staff meetings to provide reminders about the new protocols and answer questions. Multiple grantees also used employee orientations to train new staff.

Sustainability

After grant funding ended, each system varied in the level of systems change activities that they were able to continue. The majority (54%-90%) of grant funds were used for personnel costs to implement systems changes. When grant funding ended, some grantees were unable to continue staffing the project, which limited or ended their ability to continue these activities. However, 1 grantee decided to create a permanent tobacco systems change position after the grant ended to continue and expand implementation of changes across their multistate system. Two other grantees continued some of the work by building it into existing staff responsibilities.

“The overall investment of the organization in making [tobacco] a priority is one of those big success factors from my perspective.” — Project manager

More than 1 grantee referred to their systems change grant as “seed money” that fostered their system’s ability to prioritize addressing tobacco use, allowing them to build on those efforts after the grant ended. Creating a permanent tobacco systems change position is one example of this. A second example is continuing to work with leadership to prioritize the systematic addressing of tobacco use within other areas (eg, behavioral health clinics, hospitals).

“The past 2 years have been transformative for the number of engaged clinicians, the level of tobacco impact understanding, and a commitment to continue to improve and better our processes.” — Clinic staff

“The standard operating procedures are now part of our universal system. So, the systems will continue, even though the grant period is ending.” — Chief operating officer

This work changed both clinical practice and social norms among providers and staff, which helped keep tobacco use prioritized. Many grantees developed new clinical workflows or rooming protocols, and all grantees modified their EHR to implement these new protocols. Integrating tobacco user identification and interventions into their standard of care fostered sustainability. DISCUSSION

Multiple factors influence how and to what extent health systems can implement systemic changes to improve identification and treatment of tobacco use. Our findings further describe both facilitators and barriers to implementing such changes and also align with the existing literature.

Engaging leadership at all levels of the organization facilitates project implementation and expansion, ensures that resources are available, and promotes sustainability.7,8,14,21 Cultivating clinic champions is also important to support system integration; in particular, physician champions can significantly improve clinic performance in the delivery of cessation interventions.22 The grantees in our study incorporated multiple levels of leadership across the organization into their projects and reported that this cultivated program support and garnered additional resources.

Implementing new protocols and building capacity through staff trainings are important facilitators of systems change13,14 and increases staff confidence in helping patients quit using tobacco.21 All of the grantees incorporated staff trainings into their projects. Some grantees also leveraged additional funds from other departments and grants to support trainings.

Additionally, effectively using data promotes action and facilitates sustainability.23 The EHR can support routine clinical smoking cessation protocols and documention9,24-26 and is a key component of systems change.14 Embedding clear workflows into the EHR and utilizing “smart forms” and reports to track and link tobacco use with other health conditions can facilitate improvements in patient care.21 All grantees modified their EHRs to collect data on how tobacco use was addressed during clinical encounters and to monitor performance. Data were shared with staff and leadership to create buy-in and improve processes.

Lastly, external influences, such as the Meaningful Use initiative, can affect program outcomes and sustainability. Capitalizing on environmental changes and existing initiatives can facilitate systems change,14 and many grantees leveraged these factors to make the case for addressing tobacco use.

Limitations

There are several limitations to this study. This was an observational study and we cannot conclude that changes were made solely as a result of grant funding. Although the evaluation was informed by the literature,4,8,9,14 it was not designed using a specific theory or framework. The primary goal of this evaluation was to identify key facilitators, barriers, and lessons learned from each grantee’s work. It was not designed to measure long-term sustainability of changes or to identify the impact on patients (ie, patient satisfaction or quitting success). On-site observation of systems change implementation was not feasible given available resources. In addition, generalizability of these findings is limited due to the small number of health systems and the fact that all were located in Minnesota. However, many of the themes identified in this evaluation align with the health systems change literature. Furthermore, although the health systems varied in their size, reach, and population served, common themes emerged. Lastly, these health systems responded to a competitive request for proposals. Therefore, these systems had already identified tobacco use as a priority, which may have further facilitated systems change implementation.

CONCLUSIONS

Implementing health systems change interventions is an effective way to make tobacco dependence treatment a routine part of patient care compared with clinics and health systems that have not implemented such changes.8,9 Systems change activities can be tailored to meet the needs of diverse health systems. Developing system-level support, taking a team approach, capitalizing on internal and external priorities, implementing new protocols and training staff, modifying EHRs, and monitoring data and providing feedback may contribute to successful implementation. Furthermore, such changes can provide a renewed sense of enthusiasm for, and ownership of, tobacco treatment among providers and staff and can help prioritize addressing tobacco use.

Acknowledgments

The authors would like to thank the systems change grantees and their teams for contributing to this work. Any conclusions from this project and the content of this publication are solely the responsibility of the authors and do not necessarily represent the official views of ClearWay Minnesota.Author Affiliations: ClearWay Minnesota (MNW, PAK), Minneapolis, MN; Professional Data Analysts (TRC, HGZ), Minneapolis, MN.

Source of Funding: This study was funded by ClearWay Minnesota, an independent nonprofit organization funded with 3% of Minnesota’s settlement with the tobacco industry.

Author Disclosures: Ms Whittet and Ms Keller are employed by ClearWay Minnesota, which funded the grants and evaluation reported in this paper. Ms Capesius and Ms Zook are employed by Professional Data Analysts, which has a consulting contract with ClearWay Minnesota and received financial compensation as part of that contract to assist with the writing and review of the manuscript.

Authorship Information: Concept and design (MNW, TRC, HGZ, PAK); acquisition of data (TRC, HGZ); analysis and interpretation of data (MNW, TRC, HGZ); drafting of the manuscript (MNW, TRC, HGZ, PAK); critical revision of the manuscript for important intellectual content (MNW, TRC, HGZ, PAK); and supervision (MNW, PAK).

Send Correspondence to: Paula A. Keller, MPH, ClearWay Minnesota, Two AppleTree Square, 8011 34th Ave S, Ste 400, Minneapolis, MN 55425. Email: pkeller@clearwaymn.org.REFERENCES

1. Maciosek MV, LaFrance AB, Dehmer SP, et al. Updated priorities among effective clinical preventive services. Ann Fam Med. 2017;15(1):14-22. doi: 10.1370/afm.2017.

2. Maciosek MV, LaFrance AB, Dehmer SP, et al. Health benefits and cost-effectiveness of brief clinician tobacco counseling for youth and adults. Ann Fam Med. 2017;15(1):37-47. doi: 10.1370/afm.2022.

3. King BA, Dube SR, Babb SD, McAfee TA. Patient-reported recall of smoking cessation interventions from a health professional. Prev Med. 2013;57(5):715-717. doi: 10.1016/j.ypmed. 2013.07.010.

4. Quinn VP, Stevens VJ, Hollis JF, et al. Tobacco-cessation services and patient satisfaction in nine nonprofit HMOs. Am J Prev Med. 2005;29(2):77-84. doi: 10.1016/j.amepre.2005.04.006.

5. Halladay JR, Vu M, Ripley-Moffitt C, Gupta SK, O’Meara C, Goldstein AO. Patient perspectives on tobacco use treatment in primary care. Prev Chronic Dis. 2015;12:E14. doi: 10.5888/pcd12.140408.

6. Conroy MB, Majchrzak NE, Regan S, Silverman CB, Schneider LI, Rigotti NA. The association between patient-reported receipt of tobacco intervention at a primary care visit and smokers’ satisfaction with their health care. Nicotine Tob Res. 2005;7(suppl 1):S29-S34. doi: 10.1080/14622200500078063.

7. Fiore M, Jaén C, Baker T, et al. Treating Tobacco Use and Dependence: 2008 Update. Rockville, MD: HHS; 2008.

8. Land TG, Rigotti NA, Levy DE, Schilling T, Warner D, Li W. The effect of systematic clinical interventions with cigarette smokers on quit status and the rates of smoking-related primary care office visits. PLoS One. 2012;7(7):e41649. doi: 10.1371/journal.pone.0041649.

9. Moody-Thomas S, Nasuti L, Yi Y, Celestin MD Jr, Horswell R, Land TG. Effect of systems change and use of electronic health records on quit rates among tobacco users in a public hospital system. Am J Public Health. 2015;105(suppl 2):e1-e7. doi: 10.2105/AJPH.2014.302274.

10. Stead LF, Buitrago D, Preciado N, Sanchez G, Hartmann-Boyce J, Lancaster T. Physician advice for smoking cessation. Cochrane Database Syst Rev. 2013;(5):CD000165. doi: 10.1002/14651858.CD000165.pub4.

11. ClearWay Minnesota; Minnesota Department of Health. Tobacco Use in Minnesota: 2018 Update. Minneapolis, MN: ClearWay Minnesota and Minnesota Department of Health; 2019. clearwaymn.org/wp-content/uploads/2019/03/MATS_2018_Full_Technical_Report_2-26-19_FINAL_clean.pdf. Accessed January 30, 2019.

12. CDC. Best Practices for Comprehensive Tobacco Control Programs: 2014. Atlanta, GA: HHS; 2014. cdc.gov/tobacco/stateandcommunity/best_practices/pdfs/2014/comprehensive.pdf. Accessed June 27, 2018.

13. Thomas D, Abramson MJ, Bonevski B, George J. System change interventions for smoking cessation. Cochrane Database Syst Rev. 2017;2:CD010742. doi: 10.1002/14651858.CD010742.pub2.

14. Jansen AL, Capesius TR, Lachter R, Greenseid LO, Keller PA. Facilitators of health systems change for tobacco dependence treatment: a qualitative study of stakeholders’ perceptions. BMC Health Serv Res. 2014;14:575. doi: 10.1186/s12913-014-0575-4.

15. Yin RK. Case Study Research: Design and Methods. 5th ed. Thousand Oaks, CA: SAGE Publications, Inc; 2014.

16. Patton MQ. Qualitative Research and Evaluation Methods: Integrating Theory and Practice. 4th ed. Thousand Oaks, CA: SAGE Publications, Inc; 2015.

17. Charmaz K. Constructing Grounded Theory: A Practical Guide Through Qualitative Analysis. London, United Kingdom: SAGE Publications, Inc; 2006.

18. Screening, Brief Intervention, and Referral to Treatment (SBIRT). Substance Abuse and Mental Health Services Administration website. samhsa.gov/sbirt. Updated September 15, 2017. Accessed February 16, 2018.

19. Promoting Interoperability (PI). CMS website. cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/index.html. Updated March 25, 2019. Accessed January 29, 2018.

20. Sheffer CE, Payne T, Ostroff JS, et al; Training Issues Committee of the Association for the Treatment of Tobacco Use and Dependence. Increasing the quality and availability of evidence-based treatment for tobacco dependence through united certification of tobacco treatment specialists. J Smok Cessat. 2016;11(4):229-235. doi: 10.1017/jsc.2014.30.

21. Rojewski AM, Bailey SR, Bernstein SL, et al; Comorbidities Workgroup of the Society for Research on Nicotine and Tobacco (SRNT) Treatment Network. Considering systemic barriers to treating tobacco use in clinical settings in the United States [published online June 15, 2018]. Nicotine Tob Res. doi: 10.1093/ntr/nty123.

22. Papadakis S, Cole AG, Reid RD, et al. Increasing rates of tobacco treatment delivery in primary care practice: evaluation of the Ottawa Model for Smoking Cessation. Ann Fam Med. 2016;14(3):235-243. doi: 10.1370/afm.1909.

23. Lavinghouze SR, Snyder K, Rieker PP. The component model of infrastructure: a practical approach to understanding public health program infrastructure. Am J Public Health. 2014;104(8):e14-e24. doi: 10.2105/AJPH.2014.302033.

24. Boyle R, Solberg L, Fiore M. Use of electronic health records to support smoking cessation. Cochrane Database Syst Rev. 2011;(12):CD008743. doi: 10.1002/14651858.CD008743.pub2.

25. Silfen SL, Farley SM, Shih SC, et al; CDC. Increases in smoking cessation interventions after a feedback and improvement initiative using electronic health records—19 community health centers, New York City, October 2010-March 2012. MMWR Morb Mortal Wkly Rep. 2014;63(41):921-924.

26. Linder JA, Rigotti NA, Schneider LI, Kelley JH, Brawarsky P, Haas JS. An electronic health record—based intervention to improve tobacco treatment in primary care: a cluster-randomized controlled trial. Arch Intern Med. 2009;169(8):781-787. doi: 10.1001/archinternmed.2009.53.