Resource Use and Guideline Concordance in Evaluation of Pulmonary Nodules for Cancer: Too Much and Too Little Care

Importance  Pulmonary nodules are common, and more will be found with implementation of lung cancer screening. How potentially malignant pulmonary nodules are evaluated may affect patient outcomes, health care costs, and effectiveness of lung cancer screening programs. Guidelines for evaluating pulmonary nodules for cancer exist, but little is known about how nodules are evaluated in the usual care setting.

Objective  To characterize nodule evaluation and concordance with guidelines.

Design, Setting, and Participants  A retrospective cohort study was conducted including detailed review of medical records from pulmonary nodule detection through evaluation completion, cancer diagnosis, or study end (December 31, 2012). The participants included 300 adults with pulmonary nodules from 15 Veterans Affairs hospitals.

Main Outcomes and Measures  Resources used for evaluation at any Veterans Affairs facility and guideline-concordant evaluation served as the main outcomes.

Results  Twenty-seven of 300 patients (9.0%) with pulmonary nodules ultimately received a diagnosis of lung cancer: 1 of 57 (1.8%) with a nodule of 4 mm or less, 4 of 134 (3.0%) with a nodule of 5 to 8 mm, and 22 of 109 (20.2%) with a nodule larger than 8 mm. Nodule evaluation entailed 1044 imaging studies, 147 consultations, 76 biopsies, 13 resections, and 21 hospitalizations. Radiographic surveillance (n = 277) lasted a median of 13 months but ranged from less than 0.5 months to 8.5 years. Forty-six patients underwent invasive procedures (range per patient, 1-4): 41.3% (19 patients) did not have cancer and 17.4% (8) experienced complications, including 1 death. Notably, 15 of the 300 (5.0%) received no purposeful evaluation and had no obvious reason for deferral, seemingly “falling through the cracks.” Among 197 patients with a nodule detected after release of the Fleischner Society guidelines, 44.7% received care inconsistent with guidelines (17.8% overevaluation, 26.9% underevaluation). In multivariable analyses, the strongest predictor of guideline-inconsistent care was inappropriate radiologist recommendations (overevaluation relative risk, 4.6 [95% CI, 2.3-9.2]; underevaluation, 4.3 [2.7-6.8]). Other systems factors associated with underevaluation included receiving care at more than 1 facility (2.0 [1.5-2.7]) and nodule detection during an inpatient or preoperative visit (1.6 [1.1-2.5]).

Conclusions and Relevance  Pulmonary nodule evaluation is often inconsistent with guidelines, including cases with no workup and others with prolonged surveillance or unneeded procedures that may cause harm. Systems to improve quality (eg, aligning radiologist recommendations with guidelines and facilitating communication across providers) are needed before lung cancer screening is widely implemented.

It has been estimated that hundreds of thousands of pulmonary nodules are detected each year¹ and that more will be found now that the US Preventive Services Task Force² recommends annual computed tomographic (CT) lung cancer screening among high-risk individuals. Pulmonary nodules present a diagnostic challenge: some are cancers, but most are not. To rule out malignancy, evaluation typically entails radiographic surveillance; some patients also undergo invasive procedures (biopsy and/or surgical resection).³ Unfortunately, surveillance subjects patients to prolonged uncertainty, anxiety, and radiation, and invasive testing can cause physical complications.^4-10

Consequently, pulmonary nodule evaluation has the potential to create a tremendous burden on individual patients and the health care system. In 2003, the American College of Chest Physicians (ACCP)¹¹ guidelines recommended limiting surveillance to 2 years for most patients. The 2005 Fleischner Society guidelines¹² sought to further reduce the burden of evaluation by recommending that patients at lower risk of cancer (nonsmokers or those with smaller nodules) receive fewer tests. The ACCP guidelines were updated in 2007¹³ and in 2013³ to match the Fleischner Society algorithms. These guidelines apply to incidental and screening-detected nodules.

Despite the existence of guidelines, little is known about how pulmonary nodules are typically managed. The intensity of evaluation has important implications for patient health, costs, and effectiveness of lung cancer screening programs.^14,15 Although the management of screening-detected nodules in clinical trials has been reported,^16-20 it is unclear whether management in usual care settings will reflect the trials or guideline recommendations. We therefore addressed 3 questions in a representative sample of 300 veterans evaluated in 15 Veterans Affairs (VA) facilities. First, what resources are used to evaluate potentially malignant pulmonary nodules? Second, is evaluation consistent with guideline recommendations? Finally, what harms are associated with nodule evaluation?

We performed a retrospective cohort study based on review of medical records of patients with indeterminate (not known to be malignant or benign) pulmonary nodules. The institutional review boards of the White River Junction VA Medical Center, White River Junction, Vermont, and Edith Nourse Rogers Memorial Veterans Hospital, Bedford, Massachusetts, approved this study.

Our goal was to identify a representative cohort of 300 veterans with “typical” indeterminate pulmonary nodules for which nodule evaluation guidelines would apply (Figure 1). We included patients whose nodule was detected between January 1, 2003, and December 31, 2006, because (1) the start date coincides with the publication of the ACCP guidelines recommending surveillance be limited to 2 years,¹¹ providing an upper bound for expected duration of surveillance; (2) the period encompasses the release of the Fleischner Society guidelines,¹² allowing us to assess their influence on the intensity of evaluation; and (3) these dates allowed an extended period (6-10 years) from nodule detection through end of medical record review (December 31, 2012), allowing us to capture cases of prolonged surveillance.

Figure 1 illustrates the steps used to assemble our cohort. First, we created a VA FileMan algorithm to search for text strings in radiology reports of all chest radiographs and CT scans performed at 2 VA facilities in 2006. Our algorithm, designed to be more sensitive than specific, searched for nodul and mass, and then discarded reports containing the phrase no pulmonary nod. If the 2006 report (flagged study) was a follow-up test for a nodule detected previously, we worked backward in the VA’s integrated electronic medical record (VistaWeb) until we identified the study corresponding to the first detection of the nodule (index study), which may have been performed at another facility.

Steps 2 and 3 were performed iteratively until we reached our target sample of 300. We randomly selected patients from our initial cohort (N = 2366) for manual review of the radiology report to confirm the presence of a pulmonary nodule. We then conducted a limited medical record review to confirm eligibility. Exclusion criteria were designed to eliminate patients for whom guidelines for nodule evaluation would not apply (Figure 1).

We developed a standardized data abstraction form to capture baseline (ie, at time of index study) patient and nodule characteristics, events during evaluation (change in nodule size, appearance of new nodule, or transfer of care to another VA facility), resources used for evaluation at any facility, and patient outcomes (final diagnosis, complications of invasive procedures). Two trained individuals reviewed medical records in duplicate; the lead investigator (R.S.W.) resolved discrepancies.

We recorded all resources used for nodule evaluation: imaging studies (radiography, CT, and positron emission tomography [PET]), consultations, biopsy procedures (bronchoscopy, transthoracic needle biopsy, mediastinoscopy, and other [eg, peripheral lymph node biopsy]), preprocedure testing (cardiac stress test, pulmonary function tests), hospitalizations, and surgical resection. For each resource used, we graded how likely it was to be related to nodule evaluation: definitely (eg, indication for CT listed as “follow-up pulmonary nodule”), probably (eg, pulmonary consultation for “abnormal CT” in a patient with a newly detected nodule, and chest radiographs or CT scans performed during a hospitalization associated with nodule evaluation), or possibly (eg, follow-up CT ordered at an interval consistent with nodule evaluation but without mention of nodule in indication). Resources graded as “possibly related” were excluded from our analyses. For cases in which no purposeful nodule evaluation occurred, we recorded possible reasons for deferral (eg, severe comorbidity, patient refusal).

We set predetermined stopping rules for data abstraction: (1) nodule ruled out (eg, “nodule” seen on chest radiograph resolved as nipple shadow on subsequent CT); (2) cancer diagnosed (lung or other primary); and (3) patient died or was lost from the VA health care system (ie, no further visits at any VA site). If none of these criteria were met, we reviewed all records through December 31, 2012 (6-10 years after nodule detection), recording any resources used for nodule evaluation.

Our main outcomes were resources used for nodule evaluation and the proportion that received evaluation consistent with the Fleischner Society guidelines.¹² Concordance with guideline recommendations was determined by 2 experts³ in pulmonary nodule evaluation (R.S.W. reviewed every case to make the initial determination, and borderline cases were resolved through discussion with M.K.G.). We also categorized radiologist recommendations as being concordant with, more intensive than, or less intensive than guideline recommendations. More intensive evaluation (overevaluation) could include more frequent testing than recommended, more prolonged surveillance than recommended, or performance of tests that are not recommended (eg, PET or biopsy for nodules <8 mm); conversely, less intensive evaluation (underevaluation) encompassed delays or failures to obtain recommended tests. Because the 2005 Fleischner Society¹² and 2007 ACCP guidelines¹³ were ambiguous in the recommended duration of surveillance for subsolid (ie, ground glass or part-solid) nodules, we used nodule size to determine the appropriate duration of surveillance for all nodules regardless of attenuation. As secondary outcomes, we assessed the influence of the release of the Fleischner Society¹² guidelines on the intensity of nodule evaluation and factors associated with evaluation intensity.

All analyses were performed in Stata, version 10.1 (StataCorp). Because guidelines for nodule evaluation are based on nodule size, we reported baseline patient characteristics and outcomes stratified by nodule size at the time of the index study, comparing medians and proportions using the Kruskal-Wallis, χ², or Fisher exact test as appropriate. For all analyses, we used 2-tailed α = .05 as the threshold for statistical significance.

We first determined the number and proportion of patients who underwent each type of evaluation. We then calculated the median number and interquartile range (IQR) or full range of tests or services performed per patient, as well as the total number performed among all patients.

For our primary analysis, we calculated the proportions of patients who received guideline-concordant evaluation, overevaluation, and underevaluation among the subset of patients with a pulmonary nodule detected after release of the Fleischner Society guidelines¹² (index study on or after November 1, 2005). As a secondary analysis to examine the influence of guidelines on evaluation intensity, we compared the proportion that received overevaluation (using the Fleischner Society algorithm as our benchmark) among patients with a nodule detected before vs after release of these guidelines.

Using bivariate logistic regression, we explored factors associated with intensity of evaluation, including patient characteristics (age, tobacco use, chronic obstructive pulmonary disease, and symptoms suggestive of lung cancer), nodule characteristics (index size, ground-glass attenuation [ie, subsolid nodules], spiculation, and upper lobe location), and evaluation characteristics (index chest radiograph [vs CT], index study during a preoperative or inpatient visit, nodule evaluation at >1 facility, development of a new nodule during surveillance, radiologist recommendations for guideline-inconsistent evaluation [eg, recommendation for more intensive evaluation compared with guideline-consistent, less intensive evaluation, or no explicit recommendation], and patient refusal). Factors associated with evaluation intensity on bivariate analysis (based on P < .20) were included in multivariate models. Because outcomes were common (>10%), we estimated relative risks and 95% CIs using a Poisson regression model clustered by facility.

In this representative sample of 300 veterans with an indeterminate pulmonary nodule, 57 of the sample (19.0%) had a nodule of 4 mm or less in the index study, 134 (44.7%) had a 5- to 8-mm nodule, and 109 (36.3%) had a nodule larger than 8 mm (Table 1). The sample was typical of the VA population: mostly men who had smoked. Most nodules were incidentally detected and did not have features associated with malignancy. Ultimately, 27 (9.0%) patients received a diagnosis of lung cancer; the likelihood of cancer was significantly associated with nodule size (P < .001).

The 300 patients in our cohort underwent nodule evaluation at 15 VA facilities around the country (Arizona, California, Connecticut, Florida, Georgia, Kentucky, Maine, Massachusetts, New Hampshire, Rhode Island, Texas, and Vermont). More than 10% (11.3% [34]) underwent evaluation at more than 1 facility. Counting only faculty-level clinicians (not residents or fellows), more than 300 clinicians at 15 sites were involved in guiding nodule evaluation, including 87 radiologists, 114 primary care providers, 51 pulmonologists, and 54 other clinicians (eg, various consultants and inpatient clinicians who ordered follow-up testing).

Twenty-three patients (7.7%) received no apparent purposeful nodule evaluation. In 8 of these cases, a likely rationale was evident, such as severe comorbidities, patient refusal, or clinician notes dismissing the nodule as clinically insignificant (eg, 1-2 mm). However, the other 15 patients (5.0% of the cohort) had no obvious reason for lack of follow-up, seemingly “falling through the cracks” (ie, not receiving appropriate care). Fortunately, none of these 23 patients was determined to have lung cancer (at least through December 31, 2012).

Among the 277 patients who received at least 1 follow-up test, nodule evaluation entailed substantial resource use, including 1044 imaging studies (292 chest radiographs, 710 chest CT scans, and 42 PET scans), 147 consultations (101 pulmonary, 25 thoracic surgery, and 21 other), 22 preinvasive tests (17 pulmonary function tests, 5 cardiac stress tests), 76 biopsies (46 bronchoscopies, 11 transthoracic needle biopsies, 8 mediastinoscopies, and 11 other biopsies), 13 resections (6 wedge resections, 7 lobectomies), and 21 hospitalizations (Table 2 and Supplement [eTable]). The median number of tests for nodule evaluation was 2 (IQR, 1-5; range, 1-32) among patients with benign nodules and 8 (IQR, 4-14; range, 2-24) among patients with lung cancer (P < .001).

Most patients (277 of 300 [92.3%]) underwent at least 1 follow-up imaging study (Table 2). The median duration of surveillance was 13 months (IQR, 3-33 months; range, <0.5 months to 8.5 years). After exclusion of patients whose nodules were ruled out (ie, no longer present) in a subsequent imaging study, median duration of surveillance was 26 months (IQR, 10-40 months) and did not differ significantly by baseline nodule size (P = .22). The median duration of surveillance was 11 months (range, <1-51 months) among the 27 patients who ultimately received a diagnosis of lung cancer.

In our sample, 15.3% of the patients (46 of 300) underwent invasive testing (Table 2); 41.3% of those patients (19 of 46) did not have lung cancer. Among patients who underwent biopsy, the median number of biopsies was 1 (range, 1-4), but 19.6% (9 of 46) underwent 3 or more biopsy procedures before a diagnosis was established (Table 2 and Supplement [eTable]). Thirteen patients underwent surgical resection for presumptive malignant neoplasm; 4 of these (30.8%) had benign nodules. Eight patients (17.4%) who underwent invasive procedures experienced a total of 11 complications, including 7 pneumothoraces (5 serious enough to require hospitalization), 2 hemorrhages (1 led to hospitalization), and 2 pneumonias (both required hospitalization; 1 patient died). Among 19 patients who underwent invasive procedures for a benign nodule, 4 (21.1%) experienced complications.

Figure 2 depicts the evaluation process and outcomes of 197 patients whose nodules were detected after release of the Fleischner Society guidelines.¹² This figure plots the duration of surveillance, which extended well beyond the recommended 2-year period in many cases, and highlights the complexity of evaluation among patients who required more than simple surveillance; in many cases, these patients cycled back and forth between surveillance, PET scan, and invasive testing, including multiple biopsies.

Among the 197 patients with a nodule detected after release of the Fleischner Society guidelines,¹² 55.3% received guideline-concordant care and 44.7% received care inconsistent with guidelines (17.8% received overevaluation, and 26.9% received underevaluation) (Table 3). Overevaluation was inversely associated with baseline nodule size (44.4% for nodules ≤4 mm, 15.4% for nodules 5-8 mm, and 11.4% for nodules >8 mm; P = .001) (Supplement [eFigure]). The Fleischner Society guidelines appear to have achieved the stated purpose of reducing the burden of nodule evaluation: overevaluation was far more common among patients with a nodule detected before compared with after publication of the guidelines (57.3% vs 17.8%; P < .001). Radiologist recommendations were often consistent with guidelines (81.2%); when radiologist recommendations deviated from guidelines, they were far more likely to recommend more intensive (16.2%) than less intensive (2.0%) evaluation.

Regardless of whether radiologist recommendations were consistent with guidelines, the intensity of nodule evaluation reflected the intensity of radiologist recommendations in more than 60% of the cases. In multivariate analyses (Table 4), radiologist recommendations for overly intensive evaluation and nodule detection by CT rather than radiograph were significantly associated with overevaluation. Meanwhile, older age, nodule detection during an inpatient or preoperative visit, nodule evaluation at more than 1 facility, radiologist recommendations for less intensive evaluation, and patient refusal were all significantly associated with underevaluation.

To our knowledge, this is the first study to describe pulmonary nodule evaluation in the United States in the usual care setting (ie, not in the context of a lung cancer screening study^17-20 or dedicated pulmonary nodule clinic²¹). The strength of our study is the characterization of complete episodes of pulmonary nodule evaluation during a period of several years and their relationship to guideline recommendations. We found that evaluation of pulmonary nodules for cancer consumed substantial resources and was often inconsistent with guideline recommendations. Many patients (17.8%) received overevaluation, including cases of prolonged surveillance and multiple biopsies, exposing them to unneeded radiation (which confers a small but cumulative risk of radiation-induced cancers)⁶ and the potential for physical⁹ and emotional^5,8,10 harm. Meanwhile, other patients (26.9%) received less intensive evaluation than guidelines recommend—or no workup—exposing them to the possibility of delayed cancer diagnosis.

Our findings are similar to the results of the only other research assessing pulmonary nodule evaluation in the usual care setting: a French group found tremendous variation in nodule evaluation²² and associated resource use.²³ Similarly, in physician surveys, evaluation choices were highly variable.²⁴ Even in a dedicated pulmonary nodule clinic, 45% of patients did not complete the recommended duration of surveillance, highlighting the difficulty of achieving guideline-concordant care.²¹

When exploring reasons for nonconcordance with guidelines, we found radiologist recommendations to be the strongest predictor of evaluation intensity. In our sample, radiologist recommendations were inconsistent with guidelines in 17.8% of the cases (16.2% more intensive, 2.0% less intensive). Other studies have found an even higher rate of nonconcordance between guidelines and radiologist recommendations, which may reflect the fact that VA facilities are typically academic affiliates. In national surveys,^25,26 39% to 73% of radiologists recommended care that differed from guidelines in response to pulmonary nodule vignettes. Similarly, in audits of radiology reports,^27,28 21% to 66% recommended nodule evaluation inconsistent with guidelines. Mirroring our findings, these studies^25,26,28,29 found that radiologists erred on the side of recommending overevaluation. Standardizing radiology report templates to include the Fleischner Society algorithm¹² for nodule evaluation may help improve concordance with guideline recommendations.^30,31

The other modifiable systems factors associated with receipt of inappropriate evaluation (in this case, underevaluation) were initial nodule detection in the inpatient or preoperative setting—in other words, having a nodule identified by a provider who would not be the one to direct subsequent nodule evaluation—and undergoing evaluation at more than 1 facility. Both likely reflect a failure in communication between care teams, a factor cited by the Institute of Medicine³² as one of the largest barriers to high-quality medical care. To improve the quality of nodule evaluation and reduce delays in lung cancer diagnosis, systems should be implemented to notify not only the ordering provider but also the primary care provider when a pulmonary nodule is identified. Another successful model is to appoint a dedicated clinician (often a midlevel provider) who is notified whenever a new pulmonary nodule is detected, maintains a registry of these patients, and coordinates their evaluation.³³ Finally, during care transitions (eg, inpatient to outpatient or one facility to another), explicit summaries of ongoing health care issues and pending action items may help avoid delays or gaps in care.^34,35

The appropriateness of nodule evaluation will affect the cost-effectiveness and risk to benefit tradeoffs of CT lung cancer screening.^14,15 In the National Lung Screening Trial (NLST),³⁶ 42% of invasive procedures were performed in patients with benign nodules, nearly identical to the rate in our study. Similarly, in the NLST, 44% of the patients who underwent resection had a benign nodule, whereas in our study, 30.8% of the patients who underwent resection had a benign nodule. Thus, at least in the present sample, triage of patients for invasive testing in these VA facilities reflects similar decision making as described in studies of lung cancer screening. The inefficiencies we observed related primarily to imaging, including both overuse and underuse. Although the per-unit cost of imaging studies is lower than that associated with invasive testing, the aggregate cost of overuse of imaging may be substantial.³⁷

Both overevaluation and underevaluation may have important implications for patient outcomes. Many patients in our study underwent multiple biopsies, and complications following invasive procedures were more common in our study than in the NLST¹⁷: 17.4% vs 10%. Our 17.4% complication rate closely approximates the 16% rate of complications associated with transthoracic needle biopsy of pulmonary nodules reported in 4 states.⁹ The NLST’s 10% complication rate is likely unrealistically low, reflecting both the inclusion of highly skilled comprehensive cancer centers as study sites and the healthy participant bias common to clinical trials. Although our numbers were too small to reliably estimate effects on patient outcomes, 26.9% of our sample received underevaluation. Delays or outright failures to obtain appropriate tests may prolong the time to lung cancer diagnosis (which may or may not affect outcomes such as resectability and lung cancer mortality).

Our study has limitations. We evaluated 300 episodes of pulmonary nodule evaluation in the VA system, which may not represent care at other sites. Although we found substantial variation in nodule evaluation, with instances of both underevaluation and overevaluation, it is possible that there is less variation in the single-payer, academically affiliated VA system, which has an integrated electronic medical record system, than in the broader community. Although we were able to capture evaluation conducted at any VA site, we may have missed evaluation studies that were performed in the private sector and never documented in the VA medical record. Because we were limited to information in the record, we had to make inferences when notes were not explicit. This may have resulted in misattribution of resources, including both failure to count resources used for nodule evaluation and ascribing resources to nodule evaluation that were intended for another purpose. Any misattribution also may have affected whether episodes of nodule evaluation were classified as concordant with guidelines. In particular, there may have been reasons that were not documented that would explain why some patients received no apparent nodule evaluation (eg, verbal communication between the radiologist and treating clinician that a small nodule appeared to be an intrapulmonary lymph node); any such cases would have resulted in erroneous conclusions that the patient “fell through the cracks,” inadvertently receiving inappropriate care, when evaluation was purposely deferred. We found radiologists’ recommendations to be a very strong predictor of care received, but it should be noted that both the radiologists’ recommendations and the care received were categorized in relation to the guidelines. Finally, although we targeted nodules identified between 2003 and 2006 to capture a period during which guidelines were introduced and to allow a prolonged period for follow-up, care may differ for nodules detected in more recent years.

Overall, we observed a substantial burden of pulmonary nodule evaluation at the patient and health care system levels. This study raises questions about the expected cost-effectiveness and risk to benefit tradeoffs of lung cancer screening in the usual care setting. Systems to improve efficiency and safety of nodule evaluation are needed, especially before wide-scale adoption of lung cancer screening. Possible solutions that warrant further exploration are inclusion of the Fleischner Society algorithm in the radiologist reports describing a pulmonary nodule and improved systems to communicate findings of a new nodule and the current stage of evaluation between care teams.

Accepted for Publication: January 29, 2014.

Corresponding Author: Renda Soylemez Wiener, MD, MPH, Center for Healthcare Organization and Implementation Research, Edith Nourse Rogers Memorial Veterans Hospital, 200 Springs Rd, Bldg 70 (mail code 152), Bedford, MA 01730 (rwiener@bu.edu).

Published Online: April 7, 2014. doi:10.1001/jamainternmed.2014.561.

Author Contributions: Dr Wiener had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Wiener, Schwartz, Woloshin.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Wiener, Woloshin.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Wiener, Slatore.

Obtained funding: Wiener.

Study supervision: Wiener, Schwartz, Woloshin.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by the Veterans Affairs (VA) Health Services Research and Development (HSR&D) (PPO 08-401) and with resources from the White River Junction VA Medical Center, Edith Nourse Rogers Memorial Veterans Hospital, and Portland VA Medical Center. Dr Wiener is also supported by career development award K07 CA138772 from the National Cancer Institute, and Dr Slatore is supported by a VA HSR&D Career Development Award.

Role of the Sponsor: The Department of Veterans Affairs had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed herein do not necessarily represent the views of the funding agencies, the Department of Veterans Affairs, or the US government.

Previous Presentation: An earlier version of this work was presented at the VA Health Services Research and Development National Conference; July 18, 2012; National Harbor, Maryland.

Additional Contributions: We thank Rick Hines, Amanda Donaher, and Jim Rothlender, MD, who created the Veterans Affairs FileMan algorithms to assemble our study cohorts, as well as Terry Kneeland, MPH, and Elaine Hickey, RN, who were instrumental in data abstraction for the initial medical record review. Mr Hines received financial compensation for his contributions. We recognize the contribution of our colleagues in the VA Outcomes Group and the writers’ group of the Center for Healthcare Organization and Implementation Research, whose voluntary feedback enhanced both our thinking and the presentation of our results.