Obesity and the Transformation of Monoclonal Gammopathy of Undetermined Significance to Multiple Myeloma: A Population-Based Cohort Study

Abstract

Background: Multiple myeloma (MM) is one of the most common hematologic malignancies in the United States and is consistently preceded by monoclonal gammopathy of undetermined significance (MGUS). This study investigates the role of obesity in the progression of MGUS to MM.

Methods: A retrospective identified cohort of patients in the US Veterans Health Administration database diagnosed with MGUS between October 1, 1999, and December 31, 2009, was followed through August 6, 2013. Patient-level clinical data were reviewed to verify MM diagnosis, if any. Survival analyses utilizing interval-censored data were used to investigate the risk of progression of MGUS to MM. Statistical tests were two-sided.

Results: The analytic cohort consisted of 7878 MGUS patients with a median follow-up of 68 months. Within the cohort, 39.8% were overweight and 33.8% were obese; 64.1% were of white race. During follow-up, 329 MGUS patients (4.2%) progressed to MM: 72 (3.5%) normal-weight patients (median follow-up = 61.9 months), 144 (4.6%) overweight patients (median follow-up = 69.1 months), and 113 (4.3%) obese patients (median follow-up = 70.6 months). In the multivariable analysis, overweight (hazard ratio [HR] = 1.55, 95% confidence interval [CI] = 1.16 to 2.06) and obesity (HR = 1.98, 95% CI = 1.47 to 2.68) were associated with an increased risk of transformation of MGUS to MM. Moreover, black race was associated with a higher risk of MM (HR = 1.98, 95% CI = 1.55 to 2.54).

Conclusions: Obesity and black race are risk factors for transformation of MGUS to MM. Future clinical trials should examine whether weight loss is a way to prevent the progression to MM in MGUS patients.

Multiple myeloma (MM) is one of the most common hematologic malignancies in the United States (1). In 2014, MM was estimated to account for 11 090 deaths, and 24 050 new MM cases were expected in the United States (1). MM is consistently preceded by monoclonal gammopathy of undetermined significance (MGUS) (2,3), a premalignant disorder characterized by an immunoglobulin (Ig) spike in serum or urine without myeloma-related end-organ damage (4,5), or amyloidosis, which presents with an abnormal plasma cell burden similar to MGUS. The prevalence of MGUS in the population age 50 years and older is about 3% (6), with a 1% annual risk of progression to more advanced diseases, including MM (7). Patients with MGUS are asymptomatic, and a diagnosis of MGUS does not warrant treatment.

Clinical risk factors for developing MM include older age, male sex, black race, obesity, family history, and MGUS (8). Previous studies also reported that serum M-protein concentration of 1.5 g/dL or higher, Ig subtype other than IgG, an abnormal serum-free light-chain ratio, proportion of bone marrow–aberrant plasma cells within the bone marrow plasma cell compartment of 95% or higher as assessed by flow cytometry (7,9), and reduced levels of one or two noninvolved Ig isotypes (5) are associated with progression of MGUS to MM. Nonetheless, little is known about predictors of progression of MGUS to MM (10). To date, no studies have provided clear evidence of any modifiable risk factors that might be associated with progression of MGUS to MM.

Past studies have shown that obesity contributes to an increased incidence of and/or death from many cancers (11–14), including MM (8,14–18). Moreover, it is the only modifiable risk factor for MM. However, epidemiologic studies have not determined if obesity is associated with increased MGUS incidence, an increased risk of transformation of MGUS to MM, or both. Deeper understanding of the relationship between obesity and MM is clinically relevant because a relationship between obesity and the progression of MGUS to MM would potentially encourage intervention in patients diagnosed with MGUS. In contrast, an association of obesity with MGUS incidence would require intervention across the entire obese population in order to influence MM incidence.

The goal of this study is to investigate the risk factors of MM, in particular obesity, in patients diagnosed with MGUS, using a cohort of US veterans within the Veterans Health Administration (VHA) system. To our knowledge, this study is the first to examine this association in MGUS patients.

Methods

Study Population and Design

We used the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) code 273.1 to identify patients with MGUS diagnosis in all 21 regional VHA districts throughout the United States (http://www.va.gov/directory/guide/division.asp?dnum=1). Patients with two or more ICD-9-CM codes for MGUS diagnosis between October 1, 1999, and December 31, 2009, were assembled into a retrospective cohort. MGUS diagnosis was based on blood or urine tests, for example, serum/urine protein electrophoresis tests, which were done in routine clinical care when patients presented with signs or symptoms concerning for a plasma cell dyscrasia. No prospective screenings were performed on these patients.

Unique patient identifiers were used to obtain data on sex, race, height, weight, comorbidities, creatinine concentration, marital status, geographical region of residence, and annual income. The Romano adaptation of the Charlson comorbidity index was calculated based on comorbidities present before MGUS diagnosis (baseline) (19). Creatinine concentration at baseline was defined as serum creatinine concentration level three months or less before or after MGUS diagnosis and closest to the time of MGUS diagnosis.

Body mass index (BMI), weight in kilograms divided by the square of height in meters, was used to categorize patients as underweight (<18.5), normal weight (18.5–24.9), overweight (25–29.9), or obese (≥30) (20). We used weight measured one month before or after MGUS diagnosis and closest to the time of MGUS diagnosis to compute BMI at baseline. A sensitivity analysis (SA) was performed using the highest weight measured between the date of MGUS diagnosis and one year before MM diagnosis or censoring in BMI computation to avoid underestimation of the association between BMI and the progression resulting from illness-induced weight loss (21), including MM-induced weight loss (22).

Institutional Review Boards at both the Washington University School of Medicine and the Saint Louis VHA Healthcare System approved the study.

Analytic Cohort

The analytic cohort was formed by excluding 1) patients with codes for MM remission/relapse but without codes for active MM; 2) patients whose MM diagnosis preceded MGUS diagnosis; 3) patients with alternative diagnoses; 4) patients with a missing diagnosis date or a diagnosis date before 1999; 5) patients who had their first MGUS diagnosis one year or less before MM diagnosis, death, or censoring; 6) individuals with no weight data one month or less before or after the first MGUS diagnosis; 7) underweight patients because they were more likely to die or become lost to follow-up before progression; and 8) patients who had human immunodeficiency virus (HIV). We performed a SA on an alternative analytic cohort with a modification of 5) by excluding patients who had their first MGUS diagnosis two or less years before MM diagnosis, death, or censoring.

Primary Outcome Measure

The primary outcome was patient age at MM diagnosis. Two investigators (TST and KKO) reviewed patient-level clinical data for patients who had one or more ICD-9-CM codes for MM to verify MM diagnosis based on the criteria defined by the International Myeloma Working Group (23) and the actual date of diagnosis. Patients without date of death information were assumed to be alive at the time of the last death recorded within the cohort, August 6, 2013. This assumption is supported by previous studies showing that more than 97% of death events are captured in VHA data (24,25).

Statistical Analyses

To compare the BMI groups, we used chi-square tests for categorical variables. For continuous variables, we used analysis of variance tests to examine differences in means and Kruskal-Wallis tests to examine differences in medians. A cumulative incidence curve was plotted for each BMI group. The log-rank test was used to compare the curves between BMI groups.

We analyzed the data using interval censoring techniques because age at MM incidence is bounded by age at the last protein electrophoresis test one or more months before MM diagnosis and age at MM diagnosis. Interval-censoring survival analyses with age as the outcome measure, instead of time from MGUS diagnosis to MM diagnosis, were used to avoid biases resulting from early diagnosis of MGUS due to the presence of comorbidities, rather than the indicators of the disease itself, which leads to a longer duration between MGUS and MM diagnoses unrelated to MM progression. Survival models with the Weibull distribution assumption for the random disturbance term were fit to interval-censored data in the multivariable analyses (see Supplementary Figure 1, available online, for the hazard function) (26). This model specification was chosen because it yielded the best model fit statistics (Supplementary Table 1, available online) among candidate models with a baseline hazard that increases with age. We included the following covariates in the multivariable analysis: BMI group, sex, race (white, black, other), marital status, income quartile, Charlson comorbidity index excluding diabetes, presence of diabetes, and serum creatinine concentration (≥, <1.5 mg/dL) at baseline. Serum creatinine was included because it, along with comorbidities, could alter the timing of MM diagnoses (27). When categorical variables were used, an unknown category was created for individuals with missing data.

Tests of statistical significance were two-sided. Statistical significance was assessed at the level of 0.05. All statistical analyses were performed using SAS 9.2 (SAS Institute, Cary, NC, USA).

Results

We identified 11 568 patients diagnosed with MGUS between October 1, 1999 and December 31, 2009 in the VHA dataset (Figure 1). We excluded 134 patients who had codes for MM remission/relapse but no codes for active MM and 445 patients diagnosed with MM before MGUS. From chart review, we removed 796 patients who had an alternative diagnosis and 329 people whose MGUS diagnosis date was before the study period or missing. We further excluded 619 patients who had first MGUS diagnosis less than 1 year before MM diagnosis, death, or censoring; 1 182 individuals without weight data at baseline; 113 underweight patients; and 52 patients with HIV. The final analytic cohort included 7 878 patients with a median follow-up of 68 months. Among them, 329 (4.2%) progressed to MM (Table 1).

Figure 1.

Consort diagram. ICD-9-CM = International Classification of Diseases, 9th Revision, Clinical Modification; MGUS = monoclonal gammopathy of undetermined significance; MM = multiple myeloma.

Table 1.

Demographic and clinical characteristics stratified by BMI group among 7878 US veterans diagnosed with MGUS between October 1, 1999, and December 31, 2009

Characteristic	Overall	BMI group			P
	(n = 7878)	Normal weight	Overweight	Obese
		(n = 2077)	(n = 3138)	(n = 2663)
Median age at MGUS diagnosis	72.0	75.0	73.0	67.0	<.001*
(range), y	(30.0–95.0)	(35.0–94.0)	(30.0–95.0)	(33.0–93.0)
Male, %	96.9	96.6	97.3	96.8	.26†
Race, %					.002†
White	64.1	63.5	65.5	62.9
Black	24.5	25.0	22.6	26.5
Other	1.4	1.8	1.1	1.5
Unknown	10.0	9.7	10.9	9.1
Marital status, %					<.001†
Married	60.6	55.2	62.2	62.9
Single	38.7	44.1	37.1	36.4
Unknown	0.7	0.7	0.7	0.8
Region, %					.19†
Northeast	18.2	17.6	18.9	17.8
Southeast	24.3	25.0	23.7	24.5
Central	21.4	19.8	21.2	22.7
West	17.2	18	16.7	17.3
Mid-south	19.0	19.6	19.5	17.8
Unknown	0.0	0.0	0.0	0.0
Annual income, %					.001†
1st quartile (<$4956)	25	25.7	25.6	23.7
2nd quartile ($4956–<$14 364)	24.3	25.7	24.2	23.4
3rd quartile ($14 364–<$28 716)	24.5	26	23	25
4th quartile (≥$28 716)	26.2	22.6	27.2	27.8
Unknown	0.1	0.1	0.1	0.2
Comorbidities without diabetes (mean Charlson score)	3.0	3.1	2.9	3.0	.10‡
Diabetes before MGUS, %	26.5	14.1	23.8	39.3	<.001†
Creatinine, %					<.001†
≥1.5 mg/dL	26.0	23.1	27.5	26.3
<1.5 mg/dL	59.4	62.4	56.6	60.4
Unknown	14.7	14.5	15.9	13.2
Incidence of MM, %	4.2	3.5	4.6	4.3	.14†
Median age at MM diagnosis	72.0	74.5	74.5	70	.005*
(range), y	(46.0–93.0)	(56.0–93.0)	(55.0–93.0)	(46.0–85.0)
Median age at the end of follow-up	78.0	80.0	79.0	73.0	<.001*
(range), y	(36.0–100.0)	(44.0–100.0)	(36.0–99.0)	(39.0–96.0)
Median time from MGUS to MM	44.4	41.3	41.2	48.2	.03*
diagnoses (range), mo	(12.1–148.5)	(12.4–148.5)	(12.1–148.5)	(14.0–145.2)
Median length of follow-up (range), mo	67.7	61.9	69.1	70.6	<.001*
	(12.1–166.7)	(12.3–165.5)	(12.1–166.7)	(12.5–166.7)

^*Two-sided Kruskal-Wallis tests. BMI = body mass index; MGUS = monoclonal gammopathy of undetermined significance; MM = multiple myeloma.

^†Two-sided chi-square test.

^‡Two-sided analysis of variance test.

Within the analytic cohort, 64.1% were of white race. The cohort was predominantly male (96.9%) and comprised of 2 077 (26.4%) normal-weight patients, 3,138 (39.8%) overweight patients, and 2 663 (33.8%) obese patients (Table 1). Patients in different BMI groups were statistically significantly different in their age, race, marital status, income, diabetes status, and creatinine level. Median age at MGUS diagnosis was 75 years (range = 35–94 years) for the normal weight group, 73 years (range = 30–95 years) for the overweight group, and 67 years (range = 33–93 years) for the obese group (P < .001). Median age at MM diagnosis was 75 years (range = 56–93 years) for the normal weight group, 75 years (range = 55–93 years) for the overweight group, and 70 years (range = 46–85 years) for the obese group (P = .005). The obese group contained the highest percentage of black patients (26.5% vs 22.6% for the overweight group and 25.0% for the normal weight group), while the overweight group had the highest percentage of white patients (65.5% vs 62.9% for the obese group and 63.5% for the normal weight group; P = .002). Obese patients had the highest percentage of diabetics (39.3%), followed by overweight patients (23.8%) and normal weight patients (14.1%; P < .001). Overweight patients (27.5%) and obese patients (26.3%) had a higher proportion of people with elevated serum creatinine than normal weight patients (23.1%; P < .001).

The cumulative incidence curves (Figure 2) show a statistically significant increase in MM incidence among overweight and obese patients (P = .002). During follow-up (median = 67.7 months, range = 12.1–166.7 months), 329 MGUS patients (4.2%) progressed to MM: 72 (3.5%) normal weight patients progressed to MM, with a median of 61.9 months of follow-up (range = 12.3–165.5 months); 144 (4.6%) overweight patients progressed to MM, with a median of 69.1 months of follow-up (range = 12.1–166.7 months), and 113 (4.3%) obese patients progressed to MM with a median of 70.6 months of follow-up (range = 12.5–166.7 months).

Figure 2.

Cumulative incidence of multiple myeloma among the analytic cohort of 7878 US veterans diagnosed with monoclonal gammopathy of undetermined significance between October 1, 1999, and December 31, 2009, by body mass index groups normal weight (blue), overweight (red), and obese (green). Two-sided log-rank test: P = .002. MM = multiple myeloma.

Table 2 presents the hazard ratios from the univariate and multivariable analyses. Multivariable analysis demonstrated that overweight (hazard ratio [HR] = 1.55, 95% confidence interval [CI] = 1.16 to 2.06) and obese (HR = 1.98, 95% CI = 1.47 to 2.68) MGUS patients had a higher risk of progression. Black patients were more likely to progress (HR = 1.98, 95% CI =  1.55 to 2.54). Having more comorbidities (HR = 0.89, 95% CI = 0.85 to 0.93) and elevated serum creatinine (HR = 0.61, 95% CI = 0.46 to 0.81) at baseline was inversely associated with progression.

Table 2.

Hazard ratios for developing multiple myeloma among 7878 US veterans diagnosed with MGUS between October 1, 1999, and December 31, 2009

Parameter	Univariate analyses		Multivariable analysis
	HR (95% CI)	P*	HR (95% CI)	P†
BMI group
Normal weight	1.00 (ref.)		1.00 (ref.)
Overweight	1.45 (1.09 to 1.92)	.01	1.55 (1.16 to 2.06)	.003
Obese	1.90 (1.41 to 2.56)	<.001	1.98 (1.47 to 2.68)	<.001
Race
White	1.00 (ref.)		1.00 (ref.)
Black	1.84 (1.44 to 2.33)	<.001	1.98 (1.55 to 2.54)	<.001
Other race	1.00 (0.37 to 2.70)	.99	1.08 (0.40 to 2.90)	0.88
Sex
Female	1.00 (ref.)		1.00 (ref.)
Male	0.51 (0.29 to 0.89)	.02	0.55 (0.31 to 0.98)	.04
Marital status
Single	1.00 (ref.)		1.00 (ref.)
Married	0.93 (0.75 to 1.17)	.55	1.06 (0.84 to 1.34)	.62
Annual income
1st quartile (<$4956)	1.00 (ref.)		1.00 (ref.)
2nd quartile ($4956–<$14 364)	1.27 (0.95 to 1.70)	.10	1.27 (0.95 to 1.71)	.11
3rd quartile ($14 364–<$28 716)	0.81 (0.60 to 1.11)	.20	0.84 (0.61 to 1.15)	.27
4th quartile (≥$28 716)	0.79 (0.57 to 1.08)	.14	0.79 (0.57 to 1.08)	.14
Comorbidity score	0.88 (0.84 to 0.92)	<.001	0.89 (0.85 to 0.93)	<.001
Diabetes	1.01 (0.78 to 1.30)	.96	0.94 (0.72 to 1.22)	.65
Creatinine <1.5 mg/dL	1.00 (ref.)		1.00 (ref.)
Creatinine ≥1.5 mg/dL	0.53 (0.40 to 0.70)	<.001	0.61 (0.46 to 0.81)	.001

^*Two-sided chi-square test. CI = confidence interval; HR = hazard ratio; MGUS = monoclonal gammopathy of undetermined significance.

^†Two-sided chi-square test.

The SA using the highest weight measured between the date of MGUS diagnosis and one year before MM diagnosis or censoring shows similar results (Supplementary Table 2, available online). Overweight (HR =  1.69, 95% CI =  1.22 to 2.35), obesity (HR =  2.35, 95% CI =  1.70 to 3.24), and black race (HR =  2.05, 95% CI =  1.63 to 2.58) were associated with a higher risk of progression. The SA using the same exclusion criteria except (5) to form the analytic cohort (n = 7366) also shows similar results (Supplementary Table 3, available online).

In addition to the finding that obesity was associated with a higher risk of progression to MM in MGUS patients, black race was a risk factor. We, therefore, present the demographic and clinical characteristics of MGUS patients stratified by race in Supplementary Table 4 (available online) in a similar fashion to Table 1. The racial groups are statistically significantly different in their age at MGUS diagnosis, sex, BMI group, marital status, region of residence, income, comorbidities, diabetes status, creatinine level, and age at MM diagnosis. In particular, median age at MGUS diagnosis was 73 years for patients of white race, 67 years for patients of black race, and 71 years for patients of other race (P < .001). Median age at MM diagnosis was 74 years for whites, 69 years for blacks, and 83 years for other race (P = .004). Mean Charlson score was 3.5 for whites, 3.9 for blacks, and 3.9 for other (P = .001).

Discussion

This is the first study to examine the association between BMI and the timing of transformation to MM in MGUS patients. We observed that overweight and obese MGUS patients had a higher risk of transformation of MGUS to MM than normal weight patients. This finding is important in that it provides the first evidence of a potentially modifiable risk factor that may alter the progression of MGUS to MM. If this observation is confirmed by others, it would support study of weight loss as an intervention to prevent MM in MGUS patients. Positive results from such a study could ultimately stimulate broader changes in the detection and management of MGUS to reduce the burden of MM in the population.

Despite evidence from studies indicating that excess body weight is a risk factor for MM (16,28), no previous studies have provided evidence that obesity is a risk factor for MM in MGUS patients. Cohort studies showing that obesity increases MM incidence could be explained entirely by an association of obesity with higher incidence of MGUS (29). However, our finding of a positive association between obesity and younger age at progression to MM among MGUS patients has two potential explanations. First, obesity could drive MGUS incidence at a younger age. We found that MGUS was diagnosed at a younger age in obese patients, though because MGUS is an asymptomatic condition diagnosed only in a fraction of patients harboring this condition, it is difficult to draw firm conclusions from this observation. Second, our observation could result from a faster rate of transformation of MGUS to MM.

We also observed that a higher comorbidity score (excluding diabetes) and an elevated creatinine level had a lower risk of progression. The reduced risk for patients with these characteristics could result from a higher risk of death before progression to MM. Also, obese patients frequently have more comorbid conditions than their leaner counterparts. In our cohort, the obese group had a higher proportion of patients with diabetes and/or elevated serum creatinine. If these obese patients had a higher risk of death before progression to MM, the estimated association of obesity would be biased downward.

We also found that black race was associated with an increased risk of progression to MM. While black race is associated with higher incidence of MGUS (29) and MM (30), Landgren et al. (31) reported that among 1312 white and 734 black MGUS patients with one or more hospital admissions between 1980 and 1996 in the VHA, the 10-year risk of progression to MM was similar (P = .37) and the relative risk of MM for blacks compared with whites was 1.22 (95% CI =  0.91 to 1.65). They suggested that the higher risk of MM in blacks stems from an increased risk of MGUS incidence alone, rather than from more frequent progression of MGUS to MM (32). While our study cannot rule out this rationale, several concerns should be noted. First, their study used adjusted Cox regression with time from MGUS diagnosis to MM diagnosis as the outcome variable. This could bias the results because MGUS is asymptomatic and the diagnosis is largely driven by frequency of doctor visits, which in turn is strongly related to the presence of comorbidities. Thus, if we were to assume the time to progression of MGUS to MM is the same between blacks and whites, if blacks are systematically diagnosed with MGUS earlier than whites because of more comorbidities (as presented in the “Results,” Charlson score = 3.9 for blacks vs 3.5 for whites, P = .001) (Supplementary Table 4, available online), resulting in more frequent doctor visits and lab testing, the time to progression to MM, computed as time from MGUS diagnosis to time of MM diagnosis, would appear to be longer than for whites. Second, the date of MM diagnosis in Landgren et al. was recorded as the date of discharge for the first hospitalization listing a discharge diagnosis of MM from inpatient records, a method that completely relied on the ICD-9-CM codes in the administrative data. This may be inaccurate as the results of our chart review showed that 779 of the 1159 MGUS patients (67%) who had one or more ICD-9-CM codes for MM in the cohort were found to have no active MM.

Our study has several strengths. First, the VHA database has a large enough number of patients that despite the relatively uncommon outcome of MM in patients with MGUS, our study still had sufficient statistical power. Second, each patient in our cohort was identified by two or more ICD-9-CM codes of MGUS diagnosis. Additional medical chart review was performed to verify MM diagnosis and date of diagnosis, instead of relying solely on ICD-9-CM codes and diagnosis dates. Therefore, our study had more accurate outcome ascertainment than other studies using only VHA administrative data. Last, instead of conducting a Cox analysis using time from MGUS diagnosis to MM diagnosis as the outcome variable, we performed interval-censored survival analysis using age at MM incidence as the outcome. The advantage of this approach is that timing of MGUS diagnosis is not used in the computation of the outcome variable because MGUS is usually diagnosed incidentally as a result of unrelated doctor visits, and as such timing of MGUS diagnosis is not an accurate indicator of actual MGUS incidence. Patients with more frequent doctor visits, for example, those who are obese and/or have other comorbidities, are therefore likely to be diagnosed with MGUS earlier than those with less frequent doctor visits, artificially inflating the time between MGUS and MM incidence. This phenomenon cannot be reversed even if we control for potential confounders (eg, age at MGUS diagnosis, race, BMI, and comorbidities).

This study also has limitations. First, our findings cannot conclude a faster transformation rate of MGUS to MM in obese or black patients because MGUS is asymptomatic and the diagnosis is often driven by tests performed for the diagnosis or management of other conditions. Therefore, it remains unclear if incidence of MM at a younger age is due to a faster rate of progression, earlier incidence of MGUS, or a combination of both. Second, our results may not be generalizable to the US general population because the patients served by the VHA are frequently from lower socioeconomic strata, older, more likely to be unemployed, and consist of a higher proportion of men (33–35). Third, although we verified MM diagnosis via chart review, we did not verify MGUS diagnoses for every patient in the cohort or MM status for those patients who did not have any ICD-9-CM code for MM. We were also unable to abstract data on patient serum M-protein concentration or immunoglobulin subtype. Last, although we designed our study to minimize potential biases, some sources of bias from unmeasured confounders could remain.

Our results provide evidence that overweight and obesity after MGUS diagnosis both increase the risk of progression of MGUS to MM. Further research is required to establish if this association is due to a faster rate of transformation. For black patients, close monitoring of the disease progression could lead to more timely diagnosis and treatment of MM, which may improve survival (36).

Funding

This work was supported by the Foundation for Barnes-Jewish Hospital; the Siteman Cancer Center; the National Institutes of Health (grant P30-CA091842); the National Institutes of Health (grant U54-CA155496); the Agency for Healthcare Research and Quality (grant K01-HS022330 to SHC); the American Cancer Society Clinical Research Professorship (to GAC); and the American Cancer Society (grant MSRG-13-077-01-CPHPS to KRC).

Notes

The funders had no role in the design of the study; the collection, analysis, or interpretation of the study; the writing of the manuscript; or in the decision to submit the manuscript for publication.

S-H. Chang had full access to all of 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: Chang, Colditz, Carson; analysis and interpretation of data: Chang, Luo, Colditz, Carlsson, Carson; drafting of the manuscript: Chang; critical revision of the manuscript for important intellectual content: Chang, Luo, Thomas, O’Brian, Colditz, Carlsson, Carson; statistical expertise: Chang, Luo; obtained funding: Chang, Colditz, Carson; administrative, technical, or material support: Chang, Colditz, Carson; study supervision: Chang, Colditz, Carson.

The conclusions and opinions presented herein are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health, the Agency for Healthcare Research and Quality, the American Cancer Society, or the Foundation for Barnes-Jewish Hospital.