UCSF CME...7/27/2017 1 Michael G. Shlipak, MD MPH Advances in Primary Care: A Critical Review of the Year’s Most Important Papers WEDNESDAY WORKSHOP …

7/27/2017

1

Michael G. Shlipak, MD MPH

Advances in Primary Care: A Critical Review of the Year’s Most Important Papers

WEDNESDAY WORKSHOPAUGUST 9TH, 2017

Keeping up with the Literature• Impossible task for busy clinicians• Nearly impossible task for academician • My sources: Journal table of contents Email newsletters (Journal Watch, specialty newsletters, AMA, etc.) Popular press

• If the topic is interesting, then I go to the manuscript

Manuscript Review: Questions I Consider• Does this study address an important question?• Can the study design answer the question?• What were the results? Overall conclusion Strength of the findings Generalizability of population Ethical or cost considerations

• Does this change my practice?

Today’s Outline• 6 articles to discuss 10 minutes per article

• Audience participation is essential• Questions and comments can relate to: Overall topic Methods Clinical implications Practical experience

7/27/2017

2

Questions Addressed1. Is Celecoxib worse than Ibuprofen and Naproxen as a

Cardiovascular Risk promoter?2. How much lower are toxins and carcinogens in electronic

cigarettes and nicotine supplements vs. tobacco?3. Does CPAP Reduce Cardiovascular Risk in OSA?4. Does the BMI need to be interpreted differently across

different ethnic groups?5. In patients with moderate COPD, does oxygen therapy

prevent deaths and hospitalizations?6. Must Benzodiazepines be kept away from Opiate users?

Is Celecoxib worse than Ibuprofen and Naproxen as a Cardiovascular Risk promoter?

Clinical Question/Challenge• Situation: Your arthritic patient requires NSAIDS, but gets dyspepsia with most of them. You recall that Vioxx was removed from the market because of CVD harms.

Is celecoxib (celebrex) any worse for CVD risk than ibuprofen or naproxen?

7/27/2017

3

PRECISION TRIAL: Prospective Randomized Evaluation of Celecoxib Integrated Safety versus Ibuprofen or Naproxen• Design: RCT; 3 arms (celebrex, ibuprofen, naproxen)• Non-Inferiority Trial

• Co-intervention: PPI (esomeprazole 20-40mg) for all• N=24,222 (926 centers, 13 countries)• Inclusion: OA, RA; daily users of NSAIDS for pain• Outcomes: Primary- CVD events (MI, stroke, CVD death) Secondary- a) primary + revascularization, unstable angina, TIA

b) significant GI events Tertiary- renal events, anemia (GI origin), HF hospitalization, hypertension

hospitalization

• Funding: Pfizer

Nissen SE et al., N Eng J Med, 2016

What does “non-inferiority” mean?• Goal to prove that Celebrex is not worse than naproxen (one-tail

hypothesis); not that Celebrex is different () from naproxen (two-tails)

• This is like “proving” the null hypothesis of the treatments being equal.

• The funder is motivated to find no difference between medications• This is the complete opposite of an efficacy trial, where you want to

find differences between trial groups.• Crossover and dropout kill an efficacy trial, but actually help a non-

inferiority trial.• Statistically, in this trial, non-inferiority defined as HR ≤ 1.12

(comparing celecoxib with naproxen) with a CI≤ 1.33 at high end

Participants

Mean age 63Female 64%Race:

White 75%Black 14%Asian 2%Other 9%

Diagnosis:OA 90%RA 10%


Participants continued

ASA use 46%CVD 23%Diabetes 35%Hypertension 78%Current smoker 21%Statins 54%Creatinine 0.9 ± 0.2


7/27/2017

4

Initial Max Mean Dose in RCTCelecoxib 100mg BID 200mg BID 209Ibuprofen 600mg TID 800mg TID 2045Naproxen 375mg BID 500mg BID 852


• Compliance: Mean time on treatment: 20 months Follow-up: 34 months

• 69% stopped study drug

PRECISION TRIAL: Prospective Randomized Evaluation of Celecoxib Integrated Safety versus Ibuprofen or Naproxen


Results: Intention-to-Treat Analysis


Results: On-Treatment Analysis


Results: Celebrex Reduced GI Events

7/27/2017

5


Results: Celebrex Reduced Renal Events Ibuprofen vs. Naproxen

HR (95% CI)MI 1.39 (1.01-1.91)Primary outcome 1.08 (0.90-1.31)GI 1.08 (0.85-1.39)


Ibuprofen higher risk

Conclusions• Celecoxib/Celebrex appears to be no worse than naproxen or ibuprofen for cardiovascular risk.

• Celecoxib appears to offers lower GI and renal risk, even among PPI users.

• Naproxen may have lower MI risk than ibuprofen.How much lower are toxins and carcinogens in electronic cigarettes and nicotine supplements vs. tobacco?

7/27/2017

6

Clinical Question/Challenge• Scenario: Your patient is a long-term smoker, and has tried to quit many times. The patch and gum have not worked well, but he finds that e-cigarettes satisfy his cravings well enough. Are e-cigarettes almost as bad as tobacco cigarettes? Do nicotine levels differ between NRTs, e-cigarettes, and

“combustible cigarettes”, aka tobacco?

Design• Setting: London, UK 2014 (January-June)

• Design: cross-sectional Single session (30 minutes) 1 hour without food, water, or nicotine

• Participants: N= 180 Recruited from newspapers, online ads Ever smokers 6 months continuous tobacco use/non-use

• Funding: Cancer Research UK

Shahab L. et al. Ann Intern Med, 2017

Results• 5 groups (N= 36 per group)

Current Age Mean cigs/day Age started

Current tobacco 34 14 17

Tobacco + NRT 36 11 18

Tobacco + e-cig 39 12 17

NRT only 40 0 (15 before) 20

E-cig only 39 0 (16 before) 17


NRT= nicotine replacement therapy: patch, gum, lozenge, etc.E-cigarette= electronic cigarette

7/27/2017

7

Toxins Assessed from Urine and Saliva


Parent Compound Biomarker/MetaboliteTobacco-specific N-nitrosamines

4-(methylnitrosamino)-1-(3-pyridyl)-1-butane 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanolVolatile organic compounds

Acrolein N-acetyl-S-(3-hydroxypropyl)-L-cysteineAcrylamide N-acetyl-S-(2-carbamoylethyl)-L-cysteineAcrylonitrile N-acetyl-S-(2-cyanoethyl)-L-cysteine

1,3-butadiene N-acetyl-S-(4-hydroxy-2-buten-1-yl)-L-cysteineEthylene oxide N-acetyl-S-(2-hydroxyethyl)-L-cysteine

Nicotine Content Relative To Tobacco Only


ref

122% 127%

Urinary Metabolite Levels for Selected Toxins and Carcinogens, by Group


Tobacco Specific Nitrosamine

ref 12% 2.5%



Acrolein

ref 35% 33%

7/27/2017

8



Acrylamide

ref45% 43%



Acrylonitrile

ref11% 3%



1,3-butadiene

ref

20% 11%



Ethylene Oxide

ref

54% 43%

7/27/2017

9

Conclusions• Former smokers get equivalent nicotine from NRT or e-cigs as do current smokers

• Alternative nicotine products have lower levels of tobacco toxins

• No difference between NRTs and e-cigs observed• NRT and e-cigs both associated with quantifiable levels of carcinogens and toxins, so complete nicotine cessation remains the safest option.

Does CPAP Reduce Cardiovascular Risk in OSA?

Clinical Question/Challenge• OSA is believed to be a risk factor for stroke and CVD, perhaps by its

effects on blood pressure, sympathetic nervous system activation, and inflammation.

• RCTs show that CPAP lowers blood pressure, improves endothelial function, and increases insulin sensitivity.

• I have several patients with diagnosed CVD and OSA that I cannot get to use their CPAP.

• Would they have lower CVD risk if they used CPAP, and should I try harder to convince them?

7/27/2017

10

Design: The SAVE Trial• Sleep Apnea Cardiovascular Endpoints Study (SAVE Trial)• Design: RCT- parallel group, open-label 89 centers, 7 countries CPAP- 1346; Usual Care- 1341 REM Star equipment

• Inclusion: Prior CVD (excluding heart failure) Moderate-severe OSA- defined as 12 episodes/hour of ↓ SPO2 ≥ 4%

• Exclusion: Severe daytime sleepiness Severe hypoxemia (10% time) Cheyne-Stokes

• Run-in: sham CPAP ≥ 3 hours/night during 1-week

• Funding: National Health and Medical Research Council of Australia McEvoy RD et al., N Eng J Med, 2016

Characteristic CPAP Group (N=1346)

Usual-Care Group (N=1341)

Mean age 61 61Male sex 81% 81%Race

Asian 64% 63%White 25% 25%Other 11% 12%

BMI 29 28Apnea-hypopnea index 29 30

McEvoy RD et al., N Eng J Med, 2016

The SAVE Trial

Results• 3.5 hours/night average of CPAP use• In CPAP group, 42% ≥ 4 hours/night • API: 29 (pre) to 3 (CPAP)• Only 4% of usual care chose to use CPAP

McEvoy RD et al., N Eng J Med, 2016 McEvoy RD et al., N Eng J Med, 2016

7/27/2017

11

End PointCPAP Group

(N=1346)

Usual-Care

Group (N=1341)

Hazard Ratio (95% CI)

no. (%)Overall Primary outcome 229 (17.0) 207 (15.4) 1.10 (0.91-1.32)

CV death 25 (1.9) 20 (1.5) 1.22 (0.68-2.20)Myocardial infarction 42 (3.1) 39 (2.9) 1.06 (0.68-1.64)Stroke 67 (5.0) 68 (5.1) 0.97 (0.69-1.35)Heart failure 17 (1.3) 17 (1.3) 0.98 (0.50-1.92)Unstable angina 99 (7.4) 90 (6.7) 1.09 (0.82-1.45)

Death 40 (3.0) 43 (3.2) 0.91 (0.59-1.40)

McEvoy RD et al., N Eng J Med, 2016

Outcome CPAP Group (N=1346)

Usual-Care group (N=1341)

P-value

% change during study

% change during study

Epworth sleepiness scale score ↓42% ↓9%

7/27/2017

12

Clinical Question/Challenge• Situation: We are well-calibrated to the BMI categories, but they may not apply to our Asian patients. Within the “normal weight” population, BMI < 30, studies have show East Asians have elevated metabolic risk at lower BMI levels. My South Asian patient has a BMI of 24, but is this low enough to be ideal?

Design• Design: cross-sectional, community-based cohorts Multi-Ethnic Study of Atherosclerosis (MESA) Mediators of Atherosclerosis in South Asians Living in America (MASALA)

• Participants: MESA: 2622 Whites, 803 Chinese American, 1893 African American, and 1496

Hispanics MASALA: 803 South Asians

• Outcomes: HDL < 40 (men); 150 Glucose > 100 (or medication) Blood pressure > 130/85 (or medication)

• Metabolically Abnormal Normal Weight (MAN Phenotype) ≥ 2• Funding: NIH

Gujral UP, Annals Int Med, 2017

Prevalence (%)Race/Ethnicity

White 21%*Chinese 32%*African American 31%*Hispanic 39%*South Asian 44% (men 57%; women 26%)

Prevalence of Metabolically Abnormal Normal Weight (MAN phenotype)


* Similar in men and women

7/27/2017

13

Race/ethnic-specific BMI values associated with same prevalence of metabolic abnormalities as Whites with BMI 25

South AsianChinese Hispanic African AmericanWhite

BMI (kg/m2)2519.6 20.9 21.5 22.9

Pre

vale

nce

(%)

100

90

80

70

60

50

40

30

20

10

0


White 25 White 30Race/Ethnicity

African American 22.3 29.9

Hispanic 21.5 27.0Chinese 20.5 24.5South Asian 18.9 23.3

Ethnic-specific BMI Values for Similar Metabolic Profile


Prevalence Ratio (95% CI)

Multivariate Adjusted* Race/Ethnicity

White 1.00 (Reference)South Asian 2.53 (1.99, 3.22)Chinese 1.27 (1.02, 1.59)African American 1.66 (1.35, 2.04)Hispanic 1.56 (1.26, 1.92)

Prevalence Ratios of the Metabolically Abnormal Phenotype Among Normal Weight Individuals

* adjusted for age, sex, education, alcohol use, smoking status, physical activity, daily caloric intake, hepatic fat attenuation, and pericardial fat

volume


Conclusions• MAN phenotype (Metabolically Abnormal and Normal weight) is most common in Hispanics and South Asians.

• If diabetes screening is conducted by BMI category, then BMI thresholds should differ by race/ethnicity.

7/27/2017

14

In patients with moderate COPD, does oxygen therapy prevent deaths and hospitalizations?

Clinical Question/Challenge• Situation: My patient with advanced COPD presents with persistent dyspnea. At triage his oxygen saturation (SPO2) is 91%, but when I walk him around the office it drops to 87%. I feel pressure to get him home oxygen, because it will make him live longer, right? Won’t it at least keep him out of the hospital?

Design: Long Term Oxygen Treatment Trial• Background: In the 1970’s, oxygen therapy proven to decrease mortality in

COPD with severe resting hypoxemia Recommended for resting SPO2 < 89% Unclear benefit for SPO2 89-93% Medicare paid $2B for oxygen in 2011

The Long-Term Oxygen Treatment Trial Research Group. N Eng J Med, 2016

7/27/2017

15

Design: Long Term Oxygen Treatment Trial• Design: RCT- “parallel group”, meaning no blinding Supplemental oxygen, N= 368 No supplemental oxygen, N=370 14 centers

• Inclusion: Stable COPD and either former smoker or willing to quit Resting 89-93%, OR < 90% during walk test

• Oxygen Group: Portable and 2L at night Adjust to ≥ 90% while walking

• Control Group: No oxygen unless develop severe hypoxemia ≤ 88%

• Funding: NIH, CMMS


CharacteristicNo

Supplemental Oxygen (N=370)

Supplemental Oxygen(N=386)

Age – yr. 69 68White race 89% 85%Current smoker 25% 30%SpO2 at rest on room air 93% 93%Nadir SpO2 during 6-min walk

88% 35% 35%


The Long-Term Oxygen Treatment Trial Research Group. N Eng J Med, 2016 The Long-Term Oxygen Treatment Trial Research Group. N Eng J Med, 2016

Total Events: 498HR: 0.94 (0.79-1.12)

Primary Outcome

7/27/2017

16

Total Events: 139HR: 0.90 (0.64-1.25)

Deaths

Other Outcomes without Difference Between Groups

No Change in: COPD exacerbations Hospitalizations QOL Anxiety, depression Functional status 6 minute walk

• 51 adverse events from oxygen (23 trips on wires/hose; 6 fires)


Conclusions• In stable COPD, long-term supplemental oxygen has no benefit in

patients with only moderate de-saturation• No difference across multiple outcomes• Potential caveats: Participants less sick than general population? Effects of altitude?


Must Benzodiazepines be kept away from Opiate users?

7/27/2017

17

Clinical Question/Challenge• Opiate prescriptions: 3-fold rise over 15 years, and substantial rise in opioid related deaths

• 30% of fatal overdoses with opiates have concurrent BZDs.• The FDA has black-box caution for dual use of opiate/BZD.• Perhaps BZDs are a major contributor to the epidemic.For my patients that I inherit on combined opiate/BZD, how urgent is it to stop one or the other?

Design• Private Claims data: Marketscan 2001-2013 Included patients with at least one opiate treatment, AND Enrolled for entire 13 years, AND 18-64, without cancer during time period N= 315,428

• Dual use: overlap of treatment time period for opiate + BZD• Outcomes: ER visit or admission for opioid overdose Must be during the opiate treatment interval + 7 days

• Research Questions: Is there a rise of BZD use among opiate users over 13 years? What is BZD effect on overdose? What is attributable risk of BZDs on all opiate overdoses?

BZD No BZDMean Age 45 42Women 65% 57%Depression 17% 4%2001 9% --2013 17% --

Patient Characteristics

Sun EC et al. BMJ, 2017

Cardiometabolic Abnormalities Among Normal-Weight Persons FromFive Racial/Ethnic Groups in the United StatesA Cross-sectional Analysis of Two Cohort StudiesUnjali P. Gujral, PhD; Eric Vittinghoff, PhD; Morgana Mongraw-Chaffin, PhD; Dhananjay Vaidya, PhD;Namratha R. Kandula, MD, MPH; Matthew Allison, MD, MPH; Jeffrey Carr, MD; Kiang Liu, PhD; K.M. Venkat Narayan, MD; andAlka M. Kanaya, MD

Background: The relationship between body weight and car-diometabolic disease may vary substantially by race/ethnicity.

Objective: To determine the prevalence and correlates of thephenotype of metabolic abnormality but normal weight (MAN)for 5 racial/ethnic groups.

Design: Cross-sectional analysis.

Setting: 2 community-based cohorts.

Participants: 2622 white, 803 Chinese American, 1893 AfricanAmerican, and 1496 Hispanic persons from MESA (Multi-EthnicStudy of Atherosclerosis) and 803 South Asian participants in theMASALA (Mediators of Atherosclerosis in South Asians Living inAmerica) study.

Measurements: Prevalence of 2 or more cardiometabolic ab-normalities (high fasting glucose, low high-density lipoproteincholesterol, and high triglyceride levels and hypertension)among normal-weight participants was estimated. Correlates ofMAN were assessed by using log-binomial models.

Results: Among normal-weight participants (n = 846 whites,323 Chinese Americans, 334 African Americans, 252 Hispanics,and 195 South Asians), the prevalence of MAN was 21.0% (95%CI, 18.4% to 23.9%) in whites, 32.2% (CI, 27.3% to 37.4%) inChinese Americans, 31.1% (CI, 26.3% to 36.3%) in African Amer-

icans, 38.5% (CI, 32.6% to 44.6%) in Hispanics, and 43.6% (CI,36.8% to 50.6%) in South Asians. Adjustment for demographic,behavioral, and ectopic body fat measures did not explain racial/ethnic differences. After adjustment for age, sex, and race/eth-nicity–body mass index (BMI) interaction, for the equivalent MANprevalence at a BMI of 25.0 kg/m2 in whites, the correspondingBMI values were 22.9 kg/m2 (CI, 19.5 to 26.3 kg/m2) in AfricanAmericans, 21.5 kg/m2 (CI, 18.5 to 24.5 kg/m2) in Hispanics, 20.9kg/m2 (CI, 19.7 to 22.1 kg/m2) in Chinese Americans, and 19.6kg/m2 (CI, 17.2 to 22.0 kg/m2) in South Asians.

Limitation: Cross-sectional study design and lack of harmo-nized dietary data between studies.

Conclusion: Compared with whites, all racial/ethnic minoritygroups had a statistically significantly higher prevalence of MAN,which was not explained by demographic, behavioral, or ectopicfat measures. Using a BMI criterion for overweight to screen forcardiometabolic risk may result in a large proportion of racial/ethnic minority groups being overlooked.

Primary Funding Source: National Institutes of Health.

Ann Intern Med. 2017;166:628-636. doi:10.7326/M16-1895 Annals.orgFor author affiliations, see end of text.This article was published at Annals.org on 4 April 2017.

Overweight and obesity are well-known cardio-metabolic risk factors (1–3). However, some per-sons with normal weight have elevated cardiometa-bolic risk (4–7), and the relationship between excessadiposity and cardiometabolic abnormality may vary byrace/ethnicity (4–7). Although some information isavailable regarding the prevalence and correlates ofmetabolic abnormality but normal weight (MAN) innon-Hispanic whites, non-Hispanic African Americans,and Mexican Americans (4, 5), no direct comparisonshave been made among East or South Asians whoare at high risk for cardiometabolic abnormalities,even at relatively low levels of body mass index(BMI) (8–13).

We therefore compared the prevalence of MANamong members of 5 racial/ethnic groups, including 2Asian subgroups, by using data from 2 large, well-characterized community-based U.S. cohorts. We alsoexamined the correlates associated with MAN in the 4

racial/ethnic minority groups compared with whites.Lastly, we determined the BMI values in the racial/eth-nic minority participants that would yield a MAN prev-alence equal to that in whites with a BMI of 25 kg/m2.

METHODSWe conducted a cross-sectional analysis of pooled

data from MESA (Multi-Ethnic Study of Atherosclerosis)and the MASALA (Mediators of Atherosclerosis inSouth Asians Living in America) study. To maintain con-sistency with the lower age limit of MESA participants,we excluded 94 MASALA participants younger than 44years. Excluded participants differed from those whoremained in the study only by age-related clinical out-comes. We compared 803 South Asian participantsfrom MASALA with 2622 white, 803 Chinese American,1893 African American, and 1496 Hispanic participantsfrom MESA.

MESA StudyThe design and conduct of the MESA study have

been described elsewhere (14). In brief, study partici-pants included members of 4 racial/ethnic groups

See also:

Summary for Patients . . . . . . . . . . . . . . . . . . . . . . . I-20

ORIGINAL RESEARCH Annals of Internal Medicine

628 © 2017 American College of Physicians

Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/936204/ by a VA 94121 Medical Center User on 05/16/2017

http://www.annals.orghttp://www.annals.org

(white, Chinese American, African American, and His-panic) aged 45 to 84 years recruited from the greaterNew York, New York; Baltimore, Maryland; Chicago, Il-linois; Los Angeles, California; Minneapolis–Saint Paul,Minnesota; and Winston-Salem, North Carolina, areas.Baseline data collection and examinations wereconducted between July 2000 and July 2002. Ques-tionnaires were used to assess demographic and be-havioral characteristics, and seated blood pressurereadings, anthropometric measurements, and abdomi-nal and cardiac computed tomography (CT) scans wereobtained. Physical activity was assessed by using theTypical Week Physical Activity Questionnaire (15). Fast-ing serum glucose levels were evaluated by using theglucose oxidase method (Ortho Clinical Diagnostics).Insulin levels were determined by the Access system(Beckman Coulter) and harmonized with an Elecsys as-say (Roche Diagnostics). C-reactive protein values wereassessed by using a BN II nephelometer (N High-Sensitivity C-reactive protein test, Dade Behring). Totalcholesterol and high-density lipoprotein cholesterol(HDL-C) levels were determined by using the choles-terol oxidase method (Roche Diagnostics), and low-density lipoprotein cholesterol concentrations were cal-culated. Triglyceride levels were measured by usingTriglyceride GB reagent (Roche Diagnostics). Usual di-etary intake over the past year was assessed by using a120-item food-frequency questionnaire that was vali-dated in white, African American, and Hispanic popula-tions and modified to include Chinese foods (16).

MASALA StudyThe MASALA study involved measures and meth-

ods similar to those of MESA to allow for specific cross-racial/ethnic comparisons (17). Its design and objec-tives also have been described (17). In brief, MASALAstudied a community-based sample of South AsianAmericans who were aged 40 to 84 years, had no pre-viously known cardiovascular disease, and were livingin the greater San Francisco Bay and Chicago areas. Tobe eligible for the study, participants had to reportSouth Asian ethnicity (defined as having 3 or moregrandparents born in India, Pakistan, Nepal, Bangla-desh, or Sri Lanka) and be able to speak and read Eng-lish, Hindi, or Urdu. All other eligibility criteria wereidentical to those of MESA (17). Recruitment occurredbetween October 2010 and March 2013. All partici-pants were screened by telephone and invited to eitherthe University of California, San Francisco, or the North-western University field center for a baseline clinical ex-amination (17). Bilingual study staff assisted partici-pants in completing the questionnaires, which were thesame as those used in MESA. Because dietary intake isdistinct in South Asians, the MASALA investigators usedthe SHARE (Study of Health Assessment and Risk in Eth-nic groups) food-frequency questionnaire, which wasdeveloped for and validated in South Asians (17). Meancaloric intake was calculated by summing the productof the frequency of consumption, nutrient composition,and portion size of each item across all food items (18).

The protocols used in the MASALA study forseated blood pressure and anthropometry were thesame as those used in MESA. After resting in a seatedposition for 5 minutes, each participant had his or herblood pressure assessed with an automated bloodpressure machine (V100 Vital Signs Monitor, GE Health-care). Seated blood pressure was measured 3 times,and the last 2 readings were averaged to determinesystolic and diastolic blood pressure. Participant weightwas measured with a standing balance beam or digitalscale, height with a stadiometer. Body mass index wascalculated as weight in kilograms divided by height insquare meters. Waist circumference was determined byusing a flexible tape measure at the site of maximumcircumference, halfway between the lower ribs and theanterior superior iliac spine. The circumference wasmeasured twice, and the average was used for analysis.Blood samples were collected after a 12-hour over-night fast. Total cholesterol, triglyceride, and HDL-Clevels were analyzed by enzymatic methods, and low-density lipoprotein cholesterol concentrations were cal-culated. Fasting plasma glucose levels were analyzedby using the hexokinase method. Serum insulin wasmeasured by the sandwich immunoassay method (Elec-sys 2010, Roche Diagnostics) (19). As in MESA, Luminexadipokine panel A (EMD Millipore) was used to mea-sure adiponectin and resistin levels. The interassay co-efficient of variations was 2.34% to 4.12% for adiponec-tin and 3.25% to 5.03% for resistin (19). Computedtomography scans of the abdomen (Philips MedicalSystems, Toshiba Medical Systems, and Siemens Med-ical Solutions) were used to assess visceral, subcutane-ous, and intermuscular fat mass. Noncontrast cardiacCT images were obtained with a cardiac-gated CTscanner (Phillips 16D or Toshiba MSD Aquillion 64 atthe University of California, San Francisco, and SiemensSensation Cardiac 64 at Northwestern University) to as-sess pericardial fat volume and hepatic fat attenuation.Measurement methods and reading centers were simi-lar to those used in MESA (20).

Classification of Cardiometabolic AbnormalitiesWe used National Cholesterol Education Program–

Adult Treatment Panel III criteria to consider 4 cardio-metabolic abnormalities (21). Decreased HDL-C wasdefined as a level lower than 1.03 mmol/L (

BMI CategoriesFor white, African American, and Hispanic partici-

pants, BMI was classified according to World Health Or-ganization (WHO) standard cut points for normalweight (BMI, 18.5 to 24.9 kg/m2), overweight (BMI, 25.0to 29.9 kg/m2), and obesity (BMI, ≥30 kg/m2) (27). ForSouth Asian and Chinese American participants, BMIwas classified according to WHO Asian cut points fornormal weight (BMI, 18.5 to 22.9 kg/m2), overweight(BMI, 23.0 to 27.4 kg/m2), and obesity (BMI, ≥27.5 kg/m2) (28). We also conducted sensitivity analyses by us-ing the standard WHO BMI cut points for all racial/eth-nic groups.

Body size phenotypes were defined on the basis ofa combination of BMI category (normal weight) andcardiometabolic health. Combinations of BMI and car-diometabolic status yielded 2 distinct phenotypes(normal weight without cardiometabolic abnormalities

and normal weight with cardiometabolic abnormalities[MAN]). We focused our analysis on the discordantMAN phenotype.

Statistical AnalysisAnalyses were conducted by using pooled data

from the 2 cohorts. Participant characteristics were de-scribed as means, geometric means, and percentagesby race/ethnicity. Differences in these characteristicsacross race/ethnicity were assessed by using chi-squaretests or analysis of variance as appropriate. The preva-lence of metabolic abnormality was calculated by BMIstrata. Prevalence ratios of MAN in Chinese, AfricanAmerican, Hispanic, and South Asian participants com-pared with whites were estimated by using Poissonmodels with robust SEs (29). Multivariate models wereadjusted for age, sex, education, physical activity, dailycaloric intake, alcohol use, smoking status, hepatic fat

Figure 1. Prevalence of BMI categories and metabolic status, by race/ethnicity.

0

10

20

30

40

50

60

Prev

alen

ce, %

Prev

alen

ce, %

WhiteChinese AmericanAfrican AmericanHispanicSouth Asian

Normal Weight Overweight Obese

0

10

20

30

40

50

60

70

80

90

100

Metabolically abnormal

Metabolically normal

Normal Weight Overweight Obese

Whi

te

Sout

h As

ian

Hisp

anic

Chin

ese A

mer

ican

Afric

an A

mer

ican

Whi

te

Sout

h As

ian

Hisp

anic

Chin

ese A

mer

ican

Afric

an A

mer

ican

Whi

te

Sout

h As

ian

Hisp

anic

Chin

ese A

mer

ican

Afric

an A

mer

ican

Top. Prevalence of BMI category, by race/ethnicity. Error bars are 95% CIs. Bottom. Prevalence of metabolic normality, by BMI category andrace/ethnicity. Metabolically abnormal was defined as the presence of ≥2 of the following components: decreased high-density lipoprotein cho-lesterol levels (

attenuation, and pericardial fat volume. Additional, re-stricted models including only the subset of partici-pants who had measures of visceral fat, adiponectin,and resistin also were performed. To estimate the BMI

values for South Asian, African American, Hispanic, andChinese American participants that result in metabolicoutcomes equal to those in whites with a BMI of 25kg/m2 or 30 kg/m2, we first fit a proportional odds

Table 1. Characteristics of Participants With MAN Phenotype, by Race/Ethnicity*

Characteristic SouthAsian(n � 85)

White(n � 178)

PValue

ChineseAmerican(n � 104)

PValue

AfricanAmerican(n � 104)

PValue

Hispanic(n � 97)

PValue

Prevalence, % 43.6 21.0

model for the number of cardiometabolic abnormali-ties, with group-specific 4-knot restricted cubic splinesin BMI, adjusting for sex and a 4-knot restricted cubicspline in age, then calculated the marginal expectednumber of abnormalities among whites, evaluated at aBMI of 25 kg/m2 or 30 kg/m2. We then used a searchalgorithm to find the BMI values for each of the other 4groups that resulted in approximately the same mar-ginal expected number of abnormalities. We also usedthis procedure to obtain BMI values for South Asian,African American, Hispanic, and Chinese American par-ticipants, resulting in an expected MAN prevalence ap-proximately equal to that among whites with a BMI of25 kg/m2, based on a Poisson model for MAN, alsoadjusting for sex and a 4-knot spline in age. Confidenceintervals for the BMI estimates were obtained by usingbootstrap resampling. All analyses were performedwith SAS, version 9.3 (SAS Institute).

The MESA study protocol was approved by the in-stitutional review boards (IRBs) of the 6 field centersand the National Heart, Lung, and Blood Institute(NHLBI). The IRBs of Northwestern University and theUniversity of California, San Francisco, approved theMASALA study protocol. The analysis was approved bythe IRB at Emory University.

Role of the Funding SourceThis study was funded by grants from the National

Institutes of Health (NIH), National Center for ResearchResources (NCRR), and NHLBI. The funders had no rolein the design, conduct, or analysis of the study; collec-tion, analysis, or interpretation of the data; preparation,review, or approval of the manuscript; or decision tosubmit the manuscript for publication.

RESULTSThe total sample of 7617 participants comprised

2622 whites, 803 Chinese Americans, 1893 AfricanAmericans, and 1496 Hispanic Americans from MESAand 803 South Asians from MASALA. The sample alsowas made up of 1950 (25.6%) normal-weight, 3163(41.5%) overweight, and 2504 (32.9%) obese partici-pants. Overall, the South Asian participants were signif-icantly younger than members of all other racial/ethnicgroups, and this group had a significantly higher pro-portion of men compared with all other racial/ethnicgroups except for Chinese Americans (AppendixTable 1, available at Annals.org).

Prevalence of MANThe overall prevalence of normal weight and obe-

sity varied by race/ethnicity, with white and ChineseAmerican participants having the highest prevalence ofnormal weight (32.3% and 40.2%, respectively) and Af-rican American and Hispanic participants having thehighest prevalence of obesity (45.4% and 38.6%, re-spectively) (Figure 1, top). Overall, 29.1% of the partic-ipants with normal weight had the MAN phenotype,whereas 35.8% of those with obesity were metaboli-cally normal. The prevalence of MAN varied signifi-cantly by race/ethnicity: 21.0% (95% CI, 18.4% to

Table 2. Unadjusted and Multivariable-AdjustedPrevalence Ratios of the Metabolically AbnormalPhenotype Among Normal-Weight Persons*

Covariate Prevalence Ratio (95% CI)

Unadjusted Multivariable-Adjusted†

Race/ethnicityWhite 1.00 (reference) 1.00 (reference)South Asian 2.07 (1.69–2.55) 2.53 (1.99–3.22)Chinese American 1.53 (1.25–1.88) 1.27 (1.02–1.59)African American 1.48 (1.20–1.82) 1.66 (1.35–2.04)Hispanic 1.83 (1.49–2.24) 1.56 (1.26–1.92)

Age44–54 y 1.00 (reference) 1.00 (reference)55–64 y 1.71 (1.36–2.14) 1.37 (1.10–1.70)65–74 y 2.26 (1.82–2.79) 1.80 (1.45–2.22)75–84 y 2.52 (1.99–3.18) 1.94 (1.51–2.49)

SexMale 1.00 (reference) 1.00 (reference)Female 0.79 (0.69–0.90) 0.94 (0.81–1.10)

Highest education levelHigh school or less 1.00 (reference) 1.00 (reference)Less than a bachelor's degree 0.70 (0.58–0.83) 0.87 (0.73–1.05)Bachelor's degree 0.56 (0.45–0.69) 0.70 (0.56–0.87)Higher than a bachelor's

degree0.65 (0.54–0.78) 0.74 (0.60–0.92)

Alcohol use≥1 drink daily 1.00 (reference) 1.00 (reference)3000 metabolic equivalentmin/wk

0.78 (0.57–1.06) 1.04 (0.76–1.43)

Calories2010 kcal/d 0.76 (0.61–0.94) 0.75 (0.60–0.93)

Pericardial fat volume2.34 cm3 2.76 (2.20–3.46) 2.40 (1.88–3.05)

Hepatic fat attenuation5.47 Hounsfield units 0.50 (0.41–0.61) 0.54 (0.45–0.65)

* Metabolic abnormality was defined by the presence of ≥2 of thefollowing components: decreased high-density lipoprotein choles-terol levels (

23.9%) in whites, 32.2% (CI, 27.3% to 37.4%) in ChineseAmericans, 31.1% (CI, 26.3% to 36.3%) in African Amer-icans, 38.5% (CI, 32.6% to 44.6%) in Hispanics, and43.6% (CI, 36.8% to 50.6%) in South Asians (Figure 1,bottom). These patterns were consistent by sex in allracial/ethnic groups except for South Asians, in whomthe prevalence of MAN was greater in men (57.4%)than women (26.4%). In sensitivity analyses using thestandard BMI criterion for the 2 Asian American sub-groups, the prevalence of MAN was 40.4% in ChineseAmerican and 47.9% in South Asian participants.

Among participants with 2 or more cardiometa-bolic abnormalities, the most common risk factor com-bination in whites was hypertension and a low HDL-Clevel (40.0%). In all other racial/ethnic groups, the riskfactor combination of high glucose and low HDL-C lev-els was most common (48.7% in South Asians, 37.3% inChinese Americans, 36.4% in African Americans, and37.9% in Hispanics). Appendix Table 2 (available atAnnals.org) details the prevalence of risk factor combi-nations among all racial/ethnic groups.

Among participants with MAN, South Asians weresignificantly younger than members of all other racial/ethnic groups (Table 1). A significantly greater propor-tion of South Asians than whites or Hispanics had dia-betes. Mean daily caloric intake was significantly higherin South Asians than members of any other racial/eth-nic group except Hispanics. Levels of circulating adi-ponectin were significantly lower in South Asians thanmembers of all other racial/ethnic groups. South Asiansalso had less hepatic fat attenuation (more fat in theliver) than all other racial/ethnic groups and less peri-cardial fat volume than all other groups except AfricanAmericans. Appendix Table 3 (available at Annals.org)details the characteristics of participants who were nor-mal weight regardless of metabolic phenotype.

Correlates of the MAN PhenotypeCompared with whites, the prevalence of MAN was

approximately 100% greater in South Asians, 50% inChinese and African Americans, and 80% in Hispanics(Table 2). It was also higher in older participants andthose with greater pericardial fat volume and lower inthose with higher educational status and greaterhepatic fat attenuation (less fat in the liver). In amultivariable-adjusted model, South Asian, Chinese,African American, and Hispanic race/ethnicity re-mained independently associated with MAN, as didolder age, pericardial fat volume, educational status,and hepatic fat attenuation. Adjustment for age, sex,education, smoking status, alcohol use, physical activ-ity, daily caloric intake, hepatic fat attenuation, andpericardial fat volume did not explain the differences inMAN among the study groups.

In restricted models including only normal-weightpersons with measured visceral fat mass, adiponectin,and resistin (Appendix Table 4, available at Annals.org), MAN was more prevalent in South Asians, Chi-nese Americans, African Americans, and Hispanics thanwhites. In multivariable-adjusted models, the preva-lence of MAN remained greater in South Asians and

Hispanics, but not in Chinese and African Americanparticipants, compared with whites.

Ethnic-Specific BMI ValuesWe estimated the BMI values at which the ex-

pected numbers of metabolic abnormalities amongSouth Asians, Chinese Americans, African Americans,and Hispanics would equal those among whites with aBMI of 25.0 kg/m2 or 30.0 kg/m2. For the equivalentnumber of cardiometabolic abnormalities at a BMI of25.0 kg/m2 in white participants, the correspondingBMI values were 22.3 kg/m2 (CI, 19.7 to 24.9 kg/m2) inAfrican Americans, 21.5 kg/m2 (CI, 18.5 to 24.5 kg/m2)in Hispanics, 20.5 kg/m2 (CI, 19.6 to 21.4 kg/m2) in Chi-nese Americans, and 18.9 kg/m2 (CI, 16.7 to 21.1kg/m2) in South Asians. For the equivalent number at aBMI of 30.0 kg/m2 in whites, the corresponding BMIvalues were 29.9 kg/m2 (CI, 25.6 to 34.2 kg/m2) in Afri-can Americans, 27.0 kg/m2 (CI, 26.0 to 28.0 kg/m2) inHispanics, 24.5 kg/m2 (CI, 23.6 to 25.5 kg/m2) in Chi-nese Americans, and 23.3 kg/m2 (CI, 22.3 to 24.3kg/m2) in South Asians. Figure 2 displays the racial/eth-nic BMI values associated with MAN prevalence afteradjustment for age, sex, and race–BMI interaction. Forthe equivalent MAN prevalence at a BMI of 25.0 kg/m2

Figure 2. Race/ethnicity-specific BMI values associatedwith MAN compared with whites with a BMI of 25 kg/m2.

0

10

20

30

40

50

60

70

80

90

100

Prev

alen

ce, %

15 20 21.5 22.920.919.6 25

BMI, kg/m2

South AsiansChinese AmericansHispanicsAfrican AmericansWhites

MAN was defined as a BMI of 18.5 to 24.9 kg/m2 for white, AfricanAmerican, and Hispanic participants or a BMI of 18.5 to 22.9 kg/m2 forSouth Asian and Chinese American participants and ≥2 of the follow-ing components: decreased high-density lipoprotein cholesterol lev-els (

in whites, the corresponding BMI values were 22.9kg/m2 (CI, 19.5 to 26.3 kg/m2) in African Americans,21.5 kg/m2 (CI, 18.5 to 24.5 kg/m2) in Hispanics, 20.9kg/m2 (CI, 19.7 to 22.1 kg/m2) in Chinese Americans,and 19.6 kg/m2 (CI, 17.2 to 22.0 kg/m2) in South Asians.

DISCUSSIONIn this cross-sectional study of 2 large community-

based cohorts including participants from several ra-cial/ethnic groups in the United States, we foundthat nearly a third of those who were normal weighthad cardiometabolic abnormalities. Furthermore, MANprevalence varied by race/ethnicity, with a significantlyhigher proportion of South Asians and Hispanics, fol-lowed by Chinese and African Americans, having thisphenotype compared with whites. Adjustment for de-mographic, behavioral, and ectopic fat variables didnot explain these differences. For a MAN prevalenceequivalent to that in whites with a BMI of 25 kg/m2, thecorresponding BMI values were lower in all racial/eth-nic minority groups, suggesting that BMI alone is apoor indicator of cardiometabolic risk in most of thesepopulations. A recent, nationally representative studyassessing the prevalence and correlates of MAN inwhites, African Americans, and Mexican Americans re-ported that 23.5% of all normal-weight adults had met-abolic abnormalities (4). This percentage is lower thanour finding of 29%, which partly may be a result of theyounger mean age of the prior study's participants. An-other difference is that our study included South Asianand Chinese American participants as well as measuresof ectopic fat and adipokine levels; a previous studycomparing the MESA and MASALA populations foundsignificant differences in ectopic fat distribution andadipokine levels between South Asians and the 4 MESAracial/ethnic groups (30). Although these differencesmay partially account for the increased predispositionto insulin resistance and type 2 diabetes among SouthAsians, adjustment for ectopic fat measures and adipo-kine levels did not explain the difference in MANamong racial/ethnic groups in our study. Our findingsalso are consistent with those of a larger, longitudinalstudy, which found that a BMI cut point of 30 kg/m2 inwhites was equivalent to lower BMI cut points for SouthAsians, Chinese Americans, and African Americans interms of diabetes incidence (13). Finally, our resultsbuild on those of a study that found elevated glucoseand lipid levels at lower BMI values in non-European(South Asian, Chinese, and Aboriginal Canadian) versusEuropean populations (12). Taken together, these find-ings suggest that established BMI cut points may bepractical markers for detecting overweight but may notnecessarily correlate with overall cardiometabolichealth and that race/ethnicity alone may be a betterpredictor of cardiometabolic risk in racial/ethnic minor-ity populations.

Our study has several strengths. We investigatedcardiometabolic abnormalities in normal-weight per-sons from 5 U.S. racial/ethnic groups, including the rel-atively understudied South Asian and Chinese Ameri-

can populations, in whom previous studies showedcardiometabolic abnormalities developing at lowerBMI levels than in other racial/ethnic groups (11–13).Furthermore, our study used harmonized data from 2large cohorts that included several radiographic mea-sures of body composition to assess ectopic fat andadipokine levels.

However, our results also should be interpretedwithin the context of several limitations. The differencein timing of data collection between studies (2000 to2002 for MESA and 2010 to 2013 for MASALA) mayhave resulted in some differences in the prevalence ofoverweight and obesity between the 2 cohorts. Be-cause the initial enrollment of the MESA cohort began adecade and a half ago, secular changes may have oc-curred in the adoption of healthier behaviors, such as adecreased prevalence of smoking (31). However, theprevalence of obesity and diabetes has not decreasedsubstantially during the past 2 decades (32–34). Thus,we do not believe that the prevalence of metabolic ab-normalities observed in the MESA participants wouldbe much different from that observed in a current sam-ple of middle- to older-aged adults. Furthermore,MESA and MASALA used different food-frequencyquestionnaires, limiting our ability to assess whether di-etary patterns contribute to MAN prevalence. Of note,adjustment for daily caloric intake did not explaindifferences in MAN prevalence among racial/ethnicgroups. Although the MASALA and MESA cohorts arecommunity-based samples, neither is nationally repre-sentative; therefore, the results may not be generaliz-able to younger persons or South Asians and ChineseAmericans born in the United States.

In conclusion, our findings suggest a high preva-lence of cardiometabolic abnormality among normal-weight persons, particularly those in racial/ethnic mi-nority populations. This disparity cannot be explainedby differences in demographic, behavioral, or ectopicfat measures. Therefore, clinicians using overweightand obesity as the main criteria for cardiometabolicscreening, as currently recommended by the U.S. Pre-ventive Services Task Force for diabetes testing (35),may fail to identify cardiometabolic abnormalities inmany patients from racial/ethnic minority groups. Al-though the Task Force recommends earlier screeningin racial/ethnic minority populations, testing for cardio-metabolic abnormalities in normal-weight and under-weight members of these groups also may be an im-portant consideration. Future research is needed toidentify the prospective associations between MANand incident diabetes and cardiovascular disease invarious racial/ethnic groups.

From Emory University, Atlanta, Georgia; University of Califor-nia, San Francisco, San Francisco, and University of California,San Diego, San Diego, California; Wake Forest School ofMedicine, Winston-Salem, North Carolina; Johns Hopkins Uni-versity School of Medicine, Baltimore, Maryland; Northwest-ern University Feinberg School of Medicine, Chicago, andNorthwestern University, Evanston, Illinois; and VanderbiltUniversity, Nashville, Tennessee.

ORIGINAL RESEARCH Cardiometabolic Abnormalities Among Normal-Weight Persons

634 Annals of Internal Medicine • Vol. 166 No. 9 • 2 May 2017 Annals.org


http://www.annals.org

Acknowledgment: The authors thank the other investigators,the staff, and the MASALA and MESA participants for theirvaluable contributions. A full list of participating MESA inves-tigators and institutions may be found at www.mesa-nhlbi.org.

Grant Support: The MASALA study was supported by NIHgrants R01HL093009 and K24HL112827. Data collection atthe University of California, San Francisco, was supported byNIH/NCRR grant UL1 RR024131. The MESA study was sup-ported by contracts HHSN268201500003I, N01-HC-95159,N01-HC-95160, N01-HC-95161, N01-HC-95162, N01-HC-95163, N01-HC-95164, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168, and N01-HC-95169 from theNHLBI and by grants UL1-TR-000040 and UL1-TR-001079from the NCRR.

Disclosures: Dr. Vittinghoff reports grants from the NationalInstitute of Diabetes and Digestive and Kidney Diseases dur-ing the conduct of the study. Authors not named here havedisclosed no conflicts of interest. Disclosures can also beviewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M16-1895.

Reproducible Research Statement: Study protocol: MASALAprotocol available from Dr. Kanaya (e-mail, [email protected]); MESA protocol available at www.mesa-nhlbi.org. Statis-tical code: Available from Dr. Gujral (e-mail, [email protected]). Data set: Available with steering committee approvalfrom both MESA and MASALA.

Requests for Single Reprints: Unjali P. Gujral, PhD, Emory Uni-versity, Hubert Department of Global Health, Rollins School ofPublic Health, 1518 Clifton Road, CNR 7040-K, Atlanta, GA30322; e-mail, [email protected].

Current author addresses and author contributions are avail-able at Annals.org.

References1. Da Costa LA, Arora P, Garcia-Bailo B, Karmali M, El-Sohemy A,Badawi A. The association between obesity, cardiometabolic dis-ease biomarkers, and innate immunity-related inflammation in Cana-dian adults. Diabetes Metab Syndr Obes. 2012;5:347-55.2. Lastra G, Manrique C, Sowers JR. Obesity, cardiometabolic syn-drome, and chronic kidney disease: the weight of the evidence. AdvChronic Kidney Dis. 2006;13:365-73. [PMID: 17045222]3. Holmes MV, Lange LA, Palmer T, Lanktree MB, North KE, Almo-guera B, et al. Causal effects of body mass index on cardiometabolictraits and events: a Mendelian randomization analysis. Am J HumGenet. 2014;94:198-208. [PMID: 24462370] doi:10.1016/j.ajhg.2013.12.0144. Wildman RP, Muntner P, Reynolds K, McGinn AP, Rajpathak S,Wylie-Rosett J, et al. The obese without cardiometabolic risk factorclustering and the normal weight with cardiometabolic risk factorclustering: prevalence and correlates of 2 phenotypes among the USpopulation (NHANES 1999-2004). Arch Intern Med. 2008;168:1617-24. [PMID: 18695075] doi:10.1001/archinte.168.15.16175. St-Onge MP, Janssen I, Heymsfield SB. Metabolic syndrome innormal-weight Americans: new definition of the metabolically obese,normal-weight individual. Diabetes Care. 2004;27:2222-8. [PMID:15333488]6. Ruderman N, Chisholm D, Pi-Sunyer X, Schneider S. The metabol-ically obese, normal-weight individual revisited. Diabetes. 1998;47:699-713. [PMID: 9588440]

7. Conus F, Rabasa-Lhoret R, Péronnet F. Characteristics of metabol-ically obese normal-weight (MONW) subjects. Appl Physiol NutrMetab. 2007;32:4-12. [PMID: 17332780]8. Misra A, Khurana L. The metabolic syndrome in South Asians:epidemiology, determinants, and prevention. Metab Syndr RelatDisord. 2009;7:497-514. [PMID: 19900153] doi:10.1089/met.2009.00249. Misra A, Shrivastava U. Obesity and dyslipidemia in South Asians.Nutrients. 2013;5:2708-33. [PMID: 23863826] doi:10.3390/nu507270810. Ravikiran M, Bhansali A, Ravikumar P, Bhansali S, Dutta P, ThakurJS, et al. Prevalence and risk factors of metabolic syndrome amongAsian Indians: a community survey. Diabetes Res Clin Pract. 2010;89:181-8. [PMID: 20381187] doi:10.1016/j.diabres.2010.03.01011. Gordon-Larsen P, Adair LS, Meigs JB, Mayer-Davis E, Herring A,Yan SK, et al. Discordant risk: overweight and cardiometabolic risk inChinese adults. Obesity (Silver Spring). 2013;21:E166-74. [PMID:23505200] doi:10.1002/oby.2040912. Razak F, Anand SS, Shannon H, Vuksan V, Davis B, Jacobs R,et al. Defining obesity cut points in a multiethnic population. Circu-lation. 2007;115:2111-8. [PMID: 17420343]13. Chiu M, Austin PC, Manuel DG, Shah BR, Tu JV. Deriving ethnic-specific BMI cutoff points for assessing diabetes risk. Diabetes Care.2011;34:1741-8. [PMID: 21680722] doi:10.2337/dc10-230014. Bild DE, Bluemke DA, Burke GL, Detrano R, Diez Roux AV, Fol-som AR, et al. Multi-Ethnic Study of Atherosclerosis: objectives anddesign. Am J Epidemiol. 2002;156:871-81. [PMID: 12397006]15. Ainsworth BE, Irwin ML, Addy CL, Whitt MC, Stolarczyk LM. Mod-erate physical activity patterns of minority women: the Cross-CulturalActivity Participation Study. J Womens Health Gend Based Med.1999;8:805-13. [PMID: 10495261]16. Nettleton JA, Steffen LM, Mayer-Davis EJ, Jenny NS, Jiang R,Herrington DM, et al. Dietary patterns are associated with biochem-ical markers of inflammation and endothelial activation in the Multi-Ethnic Study of Atherosclerosis (MESA). Am J Clin Nutr. 2006;83:1369-79. [PMID: 16762949]17. Kanaya AM, Kandula N, Herrington D, Budoff MJ, Hulley S, Vit-tinghoff E, et al. Mediators of Atherosclerosis in South Asians Livingin America (MASALA) study: objectives, methods, and cohort de-scription. Clin Cardiol. 2013;36:713-20. [PMID: 24194499] doi:10.1002/clc.2221918. Wang ET, de Koning L, Kanaya AM. Higher protein intake isassociated with diabetes risk in South Asian Indians: the MetabolicSyndrome and Atherosclerosis in South Asians Living in America(MASALA) study. J Am Coll Nutr. 2010;29:130-5. [PMID: 20679148]19. Shah AD, Vittinghoff E, Kandula NR, Srivastava S, Kanaya AM.Correlates of prediabetes and type II diabetes in US South Asians:findings from the Mediators of Atherosclerosis in South Asians Livingin America (MASALA) study. Ann Epidemiol. 2015;25:77-83. [PMID:25459085] doi:10.1016/j.annepidem.2014.10.01320. Tota-Maharaj R, Blaha MJ, Zeb I, Katz R, Blankstein R, BlumenthalRS, et al. Ethnic and sex differences in fatty liver on cardiac computedtomography: the multi-ethnic study of atherosclerosis. Mayo ClinProc. 2014;89:493-503. [PMID: 24613289] doi:10.1016/j.mayocp.2013.12.01521. Grundy SM, Cleeman JI, Daniels SR, Donato KA, Eckel RH,Franklin BA, et al; American Heart Association. Diagnosis and man-agement of the metabolic syndrome: an American Heart Associa-tion/National Heart, Lung, and Blood Institute Scientific Statement.Circulation. 2005;112:2735-52. [PMID: 16157765]22. Alberti KG, Zimmet P, Shaw J. Metabolic syndrome—a newworld-wide definition. A consensus statement from the Inter-national Diabetes Federation. Diabet Med. 2006;23:469-80. [PMID:16681555]23. American Diabetes Association. Diagnosis and classification ofdiabetes mellitus. Diabetes Care. 2010;33 Suppl 1:S62-9. [PMID:20042775] doi:10.2337/dc10-S06224. Hinnouho GM, Czernichow S, Dugravot A, Nabi H, Brunner EJ,Kivimaki M, et al. Metabolically healthy obesity and the risk of car-diovascular disease and type 2 diabetes: the Whitehall II cohort

Cardiometabolic Abnormalities Among Normal-Weight Persons ORIGINAL RESEARCH

Annals.org Annals of Internal Medicine • Vol. 166 No. 9 • 2 May 2017 635


http://www.mesa-nhlbi.orghttp://www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M16-1895http://www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M16-1895mailto:[email protected]:[email protected]://www.mesa-nhlbi.orgmailto:[email protected]:[email protected]:[email protected]://www.annals.orghttp://www.annals.org

study. Eur Heart J. 2015;36:551-9. [PMID: 24670711] doi:10.1093/eurheartj/ehu12325. Aung K, Lorenzo C, Hinojosa MA, Haffner SM. Risk of developingdiabetes and cardiovascular disease in metabolically unhealthynormal-weight and metabolically healthy obese individuals. J ClinEndocrinol Metab. 2014;99:462-8. [PMID: 24257907] doi:10.1210/jc.2013-283226. Appleton SL, Seaborn CJ, Visvanathan R, Hill CL, Gill TK, TaylorAW, et al; North West Adelaide Health Study Team. Diabetes andcardiovascular disease outcomes in the metabolically healthy obesephenotype: a cohort study. Diabetes Care. 2013;36:2388-94. [PMID:23491523] doi:10.2337/dc12-197127. WHO Consultation on Obesity. Obesity: Preventing and Manag-ing the Global Epidemic: Report of a WHO Consultation on Obesity,Geneva, 3-5 June 1997. Geneva: World Health Organization; 1998.Accessed at http://apps.who.int/iris/handle/10665/63854 on 18March 2016.28. WHO Expert Consultation. Appropriate body-mass index forAsian populations and its implications for policy and interventionstrategies. Lancet. 2004;363:157-63. [PMID: 14726171]29. Zou G. A modified poisson regression approach to prospectivestudies with binary data. Am J Epidemiol. 2004;159:702-6. [PMID:15033648]30. Shah AD, Kandula NR, Lin F, Allison MA, Carr J, Herrington D,et al. Less favorable body composition and adipokines in South

Asians compared with other US ethnic groups: results from theMASALA and MESA studies. Int J Obes (Lond). 2016;40:639-45.[PMID: 26499444] doi:10.1038/ijo.2015.21931. Centers for Disease Control and Prevention (CDC). Vital signs:current cigarette smoking among adults aged ≥18 years with mentalillness—United States, 2009-2011. MMWR Morb Mortal Wkly Rep.2013;62:81-7. [PMID: 23388551]32. Flegal KM, Carroll MD, Kit BK, Ogden CL. Prevalence of obesityand trends in the distribution of body mass index among US adults,1999-2010. JAMA. 2012;307:491-7. [PMID: 22253363] doi:10.1001/jama.2012.3933. Cowie CC, Rust KF, Ford ES, Eberhardt MS, Byrd-Holt DD, Li C,et al. Full accounting of diabetes and pre-diabetes in the U.S. popu-lation in 1988-1994 and 2005-2006. Diabetes Care. 2009;32:287-94.[PMID: 19017771] doi:10.2337/dc08-129634. Bullard KM, Saydah SH, Imperatore G, Cowie CC, Gregg EW,Geiss LS, et al. Secular changes in U.S. prediabetes prevalence de-fined by hemoglobin A1c and fasting plasma glucose: NationalHealth and Nutrition Examination Surveys, 1999-2010. DiabetesCare. 2013;36:2286-93. [PMID: 23603918] doi:10.2337/dc12-256335. Siu AL; U.S. Preventive Services Task Force. Screening for abnor-mal blood glucose and type 2 diabetes mellitus: U.S. Preventive Ser-vices Task Force recommendation statement. Ann Intern Med. 2015;163:861-8. [PMID: 26501513] doi:10.7326/M15-2345

VITAL STATISTICS

More than 145 000 physicians and other health professionals receiveAnnals, and millions of people access it through institutional libaries, theWeb, or mobile devices. The most recent (2015) Impact Factor for Annalsof Internal Medicine is 16.593—the highest of any specialty journal in theThomson Reuters' General and Internal Medicine category. The ImpactFactor is a measurement of the frequency with which the “average article”has been cited. Annals is ranked 5th among 153 general medicine jour-nals. It is one of the most highly cited and influential journals in the world.

ORIGINAL RESEARCH Cardiometabolic Abnormalities Among Normal-Weight Persons

636 Annals of Internal Medicine • Vol. 166 No. 9 • 2 May 2017 Annals.org


http://apps.who.int/iris/handle/10665/63854http://www.annals.org

Current Author Addresses: Dr. Gujral: Emory University, Hu-bert Department of Global Health, Rollins School of PublicHealth, 1518 Clifton Road, CNR 7040-K, Atlanta, GA 30322.Dr. Vittinghoff: Department of Epidemiology and Biostatistics,School of Medicine, University of California, San Francisco,550 16th Street, 2nd Floor, San Francisco, CA 94158.Dr. Mongraw-Chaffin: Wake Forest School of Medicine, Med-ical Center Boulevard, Winston-Salem, NC 27157.Dr. Vaidya: Department of Medicine, Johns Hopkins Univer-sity School of Medicine, 1830 East Monument Street, Room8025, Baltimore, MD 21287.Dr. Kandula: Center for Community Health, 420 East Superior,6th Floor, Chicago, IL 60640.Dr. Allison: Department of Family and Preventive Medicine,University of California, San Diego, 8950 Villa La Jolla Drive,Suite B122, Mailcode 0811, La Jolla, CA 92037.Dr. Carr: Department of Radiology, Vanderbilt University,1161 21st Avenue South, Nashville, TN 37232.Dr. Liu: 680 North Lakeshore Drive, Suite 1200, Chicago, IL60640.Dr. Narayan: Rollins School of Public Health, Emory University,1518 Clifton Road Northeast, Room 7047, Atlanta, GA 30322.Dr. Kanaya: Division of General Internal Medicine, Universityof California, San Francisco, 1545 Divisadero Street, Suite 311,San Francisco, CA 94115.

Author Contributions: Conception and design: U.P. Gujral, M.Mongraw-Chaffin, K.M.V. Narayan, A.M. Kanaya.Analysis and interpretation of the data: U.P. Gujral, E. Vitting-hoff, M. Mongraw-Chaffin, D. Vaidya, N.R. Kandula, M. Allison,J. Carr, K.M.V. Narayan, A.M. Kanaya.Drafting of the article: U.P. Gujral.Critical revision for important intellectual content: U.P. Gujral,M. Mongraw-Chaffin, D. Vaidya, N.R. Kandula, M. Allison, J.Carr, K. Liu, K.M.V. Narayan, A.M. Kanaya.Final approval of the article: U.P. Gujral, E. Vittinghoff, M.Mongraw-Chaffin, D. Vaidya, N.R. Kandula, M. Allison, J. Carr,K. Liu, K.M.V. Narayan, A.M. Kanaya.Provision of study materials or patients: N.R. Kandula, A.M.Kanaya.Statistical expertise: U.P. Gujral, E. Vittinghoff, M. Mongraw-Chaffin, D. Vaidya.Obtaining of funding: N.R. Kandula, A.M. Kanaya.Administrative, technical, or logistic support: J. Carr, A.M.Kanaya.Collection and assembly of data: N.R. Kandula, J. Carr, A.M.Kanaya.

Annals.org Annals of Internal Medicine • Vol. 166 No. 9 • 2 May 2017


http://www.annals.org

Appendix Table 1. Characteristics of Study Participants, by Race/Ethnicity*


White(n � 2622)

PValue


PValue


PValue

Hispanic(n � 1496)

PValue

Missing,n


Appendix Table 3. Characteristics of Normal-Weight Participants, by Race/Ethnicity*


White(n � 846)

PValue


PValue


PValue

Hispanic(n � 252)

PValue

Missing,n


Appendix Table 4. Unadjusted and Multivariable-AdjustedPrevalence Ratios of the Metabolically AbnormalPhenotype Among Normal-Weight Persons*



Race/ethnicityWhite 1.00 (reference) 1.00 (reference)South Asian 2.10 (1.55–2.84) 1.49 (1.10–2.04)Chinese American 1.20 (0.75–1.91) 1.10 (0.79–1.53)African American 1.57 (1.05–2.37) 1.37 (0.99–1.88)Hispanic 1.97 (1.38–2.82) 1.30 (1.00–1.69)

Age44–54 y 1.00 (reference) 1.00 (reference)55–64 y 1.25 (0.90–1.72) 1.02 (0.82–1.27)65–74 y 1.65 (1.21–2.25) 1.16 (0.93–1.44)75–84 y 2.00 (1.40–2.86) 1.29 (0.93–1.78)

SexMen 1.00 (reference) 1.00 (reference)Women 0.61 (0.49–0.77) 1.07 (0.88–1.29)

EducationLess than a bachelor's

degree1.00 (reference) 1.00 (reference)

Bachelor's degree orhigher

0.82 (0.67–1.03) 0.87 (0.74–1.03)

Alcohol use≥1 drink daily 1.00 (reference) 1.00 (reference)3131 metabolicequivalent min/wk

0.91 (0.67–1.30) 1.14 (0.91–1.43)

Calories1912 kcal/d 1.13 (0.81–1.56) 0.91 (0.71–1.16)

Pericardial fat volume1.48 cm3 2.47 (1.73–3.52) 1.35 (0.99–1.82)

Hepatic fat attenuation5.60 Hounsfield units 0.38 (0.27–0.54) 0.88 (0.75–1.01)

Adiponectin22.93 ng/mL 0.33 (0.23–0.48) 0.80 (0.66–0.98)

Appendix Table 4—Continued



Resistin21.08 ng/mL 2.17 (1.52–3.11) 1.06 (0.98–1.17)

Visceral fat area184.68 cm2 3.67 (2.45–5.51) 1.35 (1.01–1.81)

* Metabolic abnormality was defined by the presence of ≥2 of thefollowing components: decreased high-density lipoprotein choles-terol (

The new england journal of medicine

n engl j med 375;26 nejm.org December 29, 2016 2519

established in 1812 December 29, 2016 vol. 375 no. 26

From the Cleveland Clinic, Cleveland (S.E.N., M.E.H., L.M.W., K.E.W., Q.W., V.M., A.M.L.); Western Sydney University, Campbelltown, NSW, Australia (N.D.Y.); Brigham and Women’s Hospital, Harvard Medical School, Boston (D.H.S., P.L.); University Hospital Zurich, Zurich, Swit-zerland (T.F.L., F.R.); Baylor College of Medicine, Veterans Affairs Medical Cen-ter, Houston (D.Y.G.); and State Univer-sity of New York, Downstate Health Sci-ences Center ( J.S.B.) and Pfizer (M.G., B.B., M.F.B., W.B.), New York. Address reprint requests to Dr. Nissen at Cleve-land Clinic J2-230, 9500 Euclid Ave., Cleve-land, OH 44195, or at nissens@ ccf . org.

* A complete list of the committees, study centers, and investigators participating in the Prospective Randomized Evalua-tion of Celecoxib Integrated Safety versus Ibuprofen or Naproxen (PRECISION) trial is provided in the Supplementary Appen-dix, available at NEJM.org.

This article was published on November 13, 2016, and last updated on December 2, 2016, at NEJM.org.

N Engl J Med 2016;375:2519-29.DOI: 10.1056/NEJMoa1611593Copyright © 2016 Massachusetts Medical Society.

BACKGROUNDThe cardiovascular safety of celecoxib, as compared with nonselective nonsteroidal anti-inflammatory drugs (NSAIDs), remains uncertain.

METHODSPatients who required NSAIDs for osteoarthritis or rheumatoid arthritis and were at in-creased cardiovascular risk were randomly assigned to receive celecoxib, ibuprofen, or naproxen. The goal of the trial was to assess the noninferiority of celecoxib with regard to the primary composite outcome of cardiovascular death (including hemorrhagic death), nonfatal myocardial infarction, or nonfatal stroke. Noninferiority required a haz-ard ratio of 1.12 or lower, as well as an upper 97.5% confidence limit of 1.33 or lower in the intention-to-treat population and of 1.40 or lower in the on-treatment population. Gastrointestinal and renal outcomes were also adjudicated.

RESULTSA total of 24,081 patients were randomly assigned to the celecoxib group (mean [±SD] daily dose, 209±37 mg), the naproxen group (852±103 mg), or the ibuprofen group (2045±246 mg) for a mean treatment duration of 20.3±16.0 months and a mean follow-up period of 34.1±13.4 months. During the trial, 68.8% of the patients stopped taking the study drug, and 27.4% of the patients discontinued follow-up. In the intention-to-treat analyses, a primary outcome event occurred in 188 patients in the celecoxib group (2.3%), 201 patients in the naproxen group (2.5%), and 218 patients in the ibuprofen group (2.7%) (hazard ratio for celecoxib vs. naproxen, 0.93; 95% confidence interval [CI], 0.76 to 1.13; hazard ratio for celecoxib vs. ibuprofen, 0.85; 95% CI, 0.70 to 1.04; P

n engl j med 375;26 nejm.org December 29, 20162520

T h e n e w e ngl a nd j o u r na l o f m e dic i n e

Nonsteroidal antiinflammatory drugs (NSAIDs) were introduced in the 1960s and became the most widely pre-scribed class of drugs in the world, with more than 100 million prescriptions issued annually in the United States alone.1 NSAIDs inhibit cyclo-oxygenase (COX), which reduces pain and inflam-mation through the inhibition of prostaglandins. However, the COX enzyme is also present in gastric mucosa, where it stimulates gastropro-tective prostaglandins. The identification of two isoforms, COX-1 and COX-2, and the recognition that antiinflammatory and analgesic effects are mediated through COX-2 inhibition — whereas the gastrointestinal toxic effects are linked to COX-1 inhibition — resulted in the development of selective COX-2 inhibitors that offered the po-tential to retain efficacy while reducing gastro-intestinal adverse effects.2

Evidence of adverse cardiovascular outcomes in a placebo-controlled trial resulted in the with-drawal of the selective COX-2 inhibitor rofecoxib in 2004.3 On the basis of a small number of events, the results of another trial suggested that cardio-vascular harm may result from the use of higher-than-approved doses of celecoxib.4 Subsequently, the Food and Drug Administration (FDA) allowed continued marketing of celecoxib, the sole re-maining selective COX-2 inhibitor, but mandated a cardiovascular safety trial. In the Prospective Randomized Evaluation of Celecoxib Integrated Safety versus Ibuprofen or Naproxen (PRECISION) trial, we sought to assess cardiovascular, gastro-intestinal, renal, and other outcomes with cele-coxib as compared with two nonselective NSAIDs.

Me thods

Trial Design and Oversight

PRECISION was a randomized, multicenter, dou-ble-blind, noninferiority trial involving patients who were at increased cardiovascular risk and had rheumatoid arthritis or osteoarthritis. Ran-domization was stratified according to the pri-mary diagnosis (osteoarthritis or rheumatoid arthritis), aspirin use, and geographic region. Detailed methods for the trial have been pub-lished previously,5 and both the protocol and the statistical analysis plan are available with the full text of this article at NEJM.org. At each center, either a central institutional review board (Schul-man IRB) or the local institutional review board

approved the trial, and the patients provided written informed consent. A multidisciplinary executive committee supervised the trial, and an independent data and safety monitoring com-mittee reviewed unblinded data to assess safety. The members of the committees are listed in Supplementary Appendix, available at NEJM.org. The members of the executive committee agreed not to accept any financial payments from any maker of NSAIDs for the duration of the trial. The trial sponsor (Pfizer) participated in the de-sign of the trial and in the writing of the proto-col in collaboration with the executive committee and in consultation with the FDA; the sponsor also assisted with data collection and maintained the trial database. The sponsor shared opera-tional roles with the Cleveland Clinic Coordinat-ing Center for Clinical Research (C5Research) and several contract research organizations. Af-ter the conclusion of the trial, the database was transferred to C5Research for statistical analy-ses. The academic authors wrote the manuscript. The sponsor was allowed to review and comment on the manuscript, but the decision to publish and the final contents were determined by the academic authors, with no contractual limits on the right to publish. All the authors had access to the final results, approved the manuscript, and assume responsibility for its accuracy and completeness and for the adherence of the trial and this report to the protocol.

Inclusion and Exclusion Criteria

We enrolled patients who were 18 years of age or older and who, as determined by the patient and physician, required daily treatment with NSAIDs for arthritis pain; patients whose arthritis pain was managed adequately with acetaminophen were not eligible. A key inclusion criterion was established cardiovascular disease or an increased risk of the development of cardiovascular dis-ease (defined in the Supplementary Appendix). Other inclusion criteria and the exclusion criteria are provided in the protocol and in a previous publication.5

Treatment

Patients were randomly assigned, in a 1:1:1 ratio, to receive celecoxib (100 mg twice a day), ibuprofen (600 mg three times a day), or naproxen (375 mg twice a day) with matching placebo. At subse-quent visits, for patients with rheumatoid arthri-

A Quick Take is available at

NEJM.org

The New England Journal of Medicine Downloaded from nejm.org at VA LIBRARY NETWORK on May 16, 2017. For personal use only. No other uses without permission.

Copyright © 2016 Massachusetts Medical Society. All rights reserved.


Cardiovascular Safety of Celecoxib, Naproxen, or Ibuprofen

tis, investigators could increase the dose of cele-coxib to 200 mg twice a day, the dose of ibuprofen to 800 mg three times a day, or the dose of naproxen to 500 mg twice a day for the treatment of symptoms. For patients with osteo-arthritis, increases in the doses of ibuprofen and naproxen were permitted; however, regulatory dosing restrictions precluded dose escalation for celecoxib in these patients. Esomeprazole (20 to 40 mg) was provided to all patients for gastric protection. Investigators were encouraged to pro-vide cardiovascular preventive management in ac-cordance with local standards and guidelines. Pa-tients who were taking low-dose aspirin (≤325 mg daily) were permitted to continue this therapy.

Adjudicated and Other Outcomes

The primary composite outcome, in a time-to-event analysis, was the first occurrence of an adverse event that met Antiplatelet Trialists Col-laboration (APTC) criteria (i.e., death from car-diovascular causes, including hemorrhagic death; nonfatal myocardial infarction; or nonfatal stroke).6 A secondary composite outcome, major adverse cardiovascular events, included the components of the primary outcome plus coronary revascu-larization or hospitalization for unstable angina or transient ischemic attack. Secondary outcomes also included clinically significant gastrointesti-nal events. Tertiary outcomes included clinically significant renal events, iron deficiency anemia of gastrointestinal origin, and hospitalization for heart failure or hypertension. (Definitions are provided in the Supplementary Appendix.) Al-though it is not described in the protocol, the composite outcome of clinically significant gas-trointestinal events or iron deficiency anemia of gastrointestinal origin was designated as the key gastrointestinal safety outcome before the trial data were unblinded. An independent committee of multidisciplinary specialists at C5Research who were unaware of the treatment assignments re-viewed and adjudicated events. An assessment of the intensity of arthritis pain with the use of the Visual Analogue Scale for Pain (VAS) (scores range from 0 to 100 mm, with higher scores indicating worse pain) was a nonadjudicated secondary out-come; differences greater than 13.7 mm are con-sidered to be clinically meaningful.7 The incidence of death from any cause was a prespecified tertiary outcome. Other prespecified outcomes are listed in the protocol and statistical analysis plan.

Statistical Analysis

Naproxen was designated as the primary com-parator for the assessment of the noninferiority of celecoxib. Noninferiority comparisons of cele-coxib versus ibuprofen and of ibuprofen versus naproxen were also prespecified. Noninferiority required four criteria to be met; in the original design, a hazard ratio not exceeding 1.12 was required, with an upper limit of the one-sided 97.5% confidence interval of less than 1.33 in both the intention-to-treat population and the on-treatment population. The assessment of the on-treatment population included events that occurred while patients were taking the study drug and during the 30 days after discontinuation. The trial was event-driven, requiring 762 events to provide 90% power to determine noninferiority. Under the assumption of an annual event rate of 2% and a treatment discontinuation rate of 40%, the required sample size was estimated to be 20,000 patients. The observed event rate was lower, the discontinuation rate higher, and the enrollment rate slower than anticipated. At the recommendation of the data and safety monitor-ing committee and after consultation with the FDA, the protocol was amended to have the study provide 80% power, and the upper 97.5% confi-dence limit for noninferiority in the on-treatment population was modified to 1.40, which required 580 events in the intention-to-treat population and 420 events in the on-treatment population. The protocol prespecified a minimum follow-up time of 18 months, with censoring of data from event-free patients after 30 months in the inten-tion-to-treat population and after 43 months in the on-treatment population.

A Cox proportional-hazards model with adjust-ment for stratification factors was used to calcu-late the hazard ratios and confidence intervals. A one-sided noninferiority P value of less than 0.025 was considered to indicate statistical sig-nificance for the primary end point, with no ad-justment for multiple comparisons. P values for secondary analyses in the intention-to-treat popu-lation are reported for descriptive purposes; a two-sided P value of less than 0.05 was considered to indicate statistical significance, with no ad-justment for multiple comparisons. For the on-treatment analyses, P values for noninferiority are reported for the primary APTC outcome, but P values are not reported for superiority compari-sons. Additional details regarding the statistical





analyses are provided in the Supplementary Ap-pendix.

R esult s

Patient Population

We screened 31,857 patients; a total of 24,222 patients underwent randomization at 926 centers in 13 countries between October 23, 2006, and June 30, 2014, and 141 were excluded from the analysis (106 were determined to be fraudulently enrolled, and 35 enrolled more than once), leav-ing 24,081 participants who could be included in the analysis. There were 8072 patients assigned to the celecoxib group (mean [±SD] daily dose, 209±37 mg), 7969 assigned to the naproxen group (852±103 mg), and 8040 assigned to the ibupro-fen group (2045±246 mg). The characteristics of the patients at baseline were similar among the treatment groups (Table 1). The mean durations of treatment and follow-up, respectively, were 20.3±16.0 and 34.1±13.4 months for all patients: 20.8±16.0 and 34.2±13.4 months in the celecoxib group, 20.5±15.9 and 34.2±13.3 months in the naproxen group, and 19.6±16.0 and 33.8±13.6 months in the ibuprofen group. During this 10-year trial, 68.8% of patients stopped taking the study drug, and 27.4% of patients discontinued follow-up; 2.5% of patients died, 8.3% withdrew consent in writing, 7.4% verbally expressed un-willingness to continue participation, and 7.2% were lost to follow-up before a final follow-up visit. Details regarding patient disposition, time to study-drug discontinuation, and time to non-retention in the trial are provided in Figures S1, S2, and S3 in the Supplementary Appendix.

Primary APTC Outcome

In the intention-to-treat population (Table 2 and Fig. 1), the primary APTC outcome occurred in 188 patients in the celecoxib group (2.3%), 201 in the naproxen group (2.5%), and 218 in the ibuprofen group (2.7%). The hazard ratio for this outcome in the celecoxib group, as compared with the naproxen group, was 0.93 (95% confi-dence interval [CI], 0.76 to 1.13; P



hazard ratio for gastrointestinal events in the ibuprofen group versus the naproxen group was 1.08 (95% CI, 0.85 to 1.39; P = 0.53). Serious renal events occurred at a significantly lower rate in the celecoxib group than in the ibuprofen group (haz-ard ratio, 0.61; 95% CI, 0.44 to 0.85; P = 0.004), but the difference in the rate of this outcome in the celecoxib group versus the naproxen group was not significant (hazard ratio, 0.79; 95% CI, 0.56 to 1.12; P = 0.19).

Other Outcomes

The rate of hospitalization for hypertension was significantly lower in the celecoxib group than in the ibuprofen group (hazard ratio, 0.60; 95% CI, 0.36 to 0.99; P = 0.04) but was not signifi-cantly lower in the celecoxib group than in the naproxen group (Table 2). The results of analy-ses of quality of life and efficacy for the relief of arthritis symptoms are reported in Table S3 in the Supplementary Appendix. In the assessment

CharacteristicCelecoxib Group

(N = 8072)Naproxen Group

(N = 7969)Ibuprofen Group

(N = 8040)

Age — yr 63.0±9.5 63.3±9.4 63.2±9.4

Female sex — no. (%) 5175 (64.1) 5096 (63.9) 5174 (64.4)

Race — no. (%)†

White 6058 (75.0) 5926 (74.4) 5991 (74.5)

Black 1090 (13.5) 1134 (14.2) 1108 (13.8)

Asian 164 (2.0) 172 (2.2) 173 (2.2)

Unspecified or other 760 (9.4) 737 (9.2) 768 (9.6)

Body-mass index‡ 32.7±7.3 32.6±7.3 32.5±7.4

Primary arthritis diagnosis — no. (%)

Osteoarthritis 7259 (89.9) 7178 (90.1) 7208 (89.7)

Rheumatoid arthritis 813 (10.1) 791 (9.9) 832 (10.3)

Current aspirin use — no. (%) 3701 (45.8) 3652 (45.8) 3712 (46.2)

Cardiovascular risk category — no. (%)

Primary prevention 6209 (76.9) 6186 (77.6) 6206 (77.2)

Secondary prevention 1863 (23.1) 1783 (22.4) 1834 (22.8)

History of diabetes — no. (%) 2843 (35.2) 2768 (34.7) 2885 (35.9)

History of hypertension — no. (%) 6296 (78.0) 6145 (77.1) 6303 (78.4)

History of dyslipidemia — no. (%) 5080 (62.9) 4966 (62.3) 5002 (62.2)

Current smoker — no. (%) 1689 (20.9) 1631 (20.5) 1680 (20.9)

Current statin use — no. (%) 4367 (54.1) 4304 (54.0) 4307 (53.6)

Current DMARD use — no. (%) 572 (7.1) 602 (7.6) 584 (7.3)

Systolic blood pressure — mm Hg§ 125.3±10.5 125.0±10.6 125.4±10.4

Diastolic blood pressure — mm Hg 75.5±8.0 75.4±8.0 75.5±7.9

Creatinine level — mg/dl 0.9±0.23 0.9±0.22 0.9±0.22

HAQ disability index¶ 1.1±0.61 1.1±0.61 1.1±0.61

VAS score — mm‖ 54.0±23.5 54.1±24.0 54.1±23.6

* Plus–minus values are means ±SD. Percentages may not total to 100 because of rounding. DMARD denotes disease-modifying antirheumatic drug.

† Race was self-reported.‡ The body-mass index is the weight in kilograms divided by the square of the height in meters.§ P = 0.044 for the comparison among the three treatment groups.¶ The Health Assessment Questionnaire (HAQ) disability index is based on 20 questions in eight categories regarding

daily functioning; overall scores range from 0 to 3, with 0 indicating no disability and 3 indicating complete disability.‖ Visual Analogue Scale of Pain (VAS) scores range from 0 to 100 mm, with higher scores indicating worse pain; differ-

ences greater than 13.7 mm are considered to be clinically significant.

Table 1. Baseline Characteristics of Patients in the Intention-to-Treat Population.*





Tabl

e 2.

Adj

udic

ated

Out

com

es in

the

Inte

ntio

n-to

-Tre

at P

opul

atio

n.

Out

com

eC

elec

oxib

Gro

up

(N =

807

2)N

apro

xen

Gro

up

(N =

796

9)Ib

upro

fen

Gro

up

(N =

804

0)C

elec

oxib

vs.

Nap

roxe

n*C

elec

oxib

vs.

Ibu

prof

en*

Adj

uste

d H

azar

d R

atio

(9

5% C

I)P

Val

ueA

djus

ted

Haz

ard

Rat

io

(95%

CI)

P V

alue

num

ber o

f pat

ient

s (p

erce

nt)

Prim

ary

APT

C e

nd p

oint

†18

8 (2

.3)

201

(2.5

)21

8 (2

.7)

0.93

(0.

76–1

.13)

0.45

0.85

(0.

70–1

.04)

0.12

Maj

or a

dver

se c

ardi

ovas

cula

r ev

ents

‡33

7 (4

.2)

346

(4.3

)38

4 (4

.8)

0.97

(0.

83–1

.12)

0.64

0.87

(0.

75–1

.01)

0.06

Com

posi

te o

f ser

ious

gas

troi

ntes

tinal

eve

nts

86 (

1.1)

119

(1.5

)13

0 (1

.6)

0.71

(0.

54–0

.93)

0.01

0.65

(0.

50–0

.85)

0.00

2

Clin

ical

ly s

igni

fican

t gas

troi

ntes

tinal

ev

ents

§55

(0.

7)56

(0.

7)72

(0.

9)0.

97 (

0.67

–1.4

0)0.

860.

76 (

0.53

–1.0

8)0.

12

Iron

-def

icie

ncy

anem

ia o

f gas

troi

ntes

tinal

or

igin

§33

(0.

4)69

(0.

9)64

(0.

8)0.

47 (

0.31

–0.7

1)



of pain with the use of the VAS scale, a sig-nificant but small benefit was found for naproxen relative to celecoxib or ibuprofen; the change in VAS score from baseline was −9.3±0.26 mm for celecoxib, −9.5±0.26 for ibuprofen, and −10.2±0.26 for naproxen (P



Patients with Event (%)10

0 80 60 40 20 00

612

1824

30

3.0

2.0

4.0

1.0

0.0

06

1218

2430

4236

3030

Mon

ths

sinc

e R

ando

miz

atio

n

DD

eath

from

Any

Cau

se

APr

imar

y A

PTC

Out

com

e: In

tent

ion-

to-T

reat

Pop

ulat

ion

Cel

ecox

ib v

s. ib

upro

fen,

haz

ard

ratio

, 0.8

5 (9

5% C

I, 0.

70–1

.04)

; non

infe

rior

ity P

<0.

001

Cel

ecox

ib v

s. n

apro

xen,

haz

ard

ratio

, 0.9

3(9

5% C

I, 0.

76–1

.13)

; non

infe

rior

ity P

<0.

001

Ibup

rofe

n vs

. nap

roxe

n, h

azar

d ra

tio, 1

.08

(95%

CI,

0.90

–1.3

1); n

onin

feri

ority

P=

0.02

Ibup

rofe

nN

apro

xen

Cel

ecox

ib

8040

7969

8072

No.

at R

isk

7445

7428

7545

7103

7215

7198

6794

6817

6863

6080

6115

6203

5516

5515

5645

Patients with Event (%)

100 80 60 40 20 0

06

1218

2430

3.0

2.0

1.5

2.5

1.0

0.5

0.0

06

1218

2430

3030

Mon

ths

sinc

e R

ando

miz

atio

n

Cel

ecox

ib v

s. ib

upro

fen,

haz

ard

ratio

, 0.9

2 (9

5% C

I, 0.

73–1

.17)

; P=

0.49

Cel

ecox

ib v

s. n

apro

xen,

haz

ard

ratio

, 0.8

0(9

5% C

I, 0.

63–1

.00)

; P=

0.05

2Ib

upro

fen

vs. n

apro

xen,

haz

ard

ratio

, 0.8

7(9

5% C

I, 0.

70–1

.09)

; P=

0.22

Ibup

rofe

nN

apro

xen

Cel

ecox

ib

8040

7969

8072

No.

at R

isk

7476

7450

7568

7160

7169

7253

6871

6883

6939

6164

6189

6289

5509

5602

5741

Patients with Event (%)

100 80 60 40 20 0

06

1218

2442

3.0

2.0

4.0

1.0

0.0

06

1218

24

Mon

ths

sinc

e R

ando

miz

atio

n

ESe

riou

s G

astr

oint

estin

al E

vent

s

BPr

imar

y A

PTC

Out

com

e: O

n-Tr

eatm

ent P

opul

Documents

UCSF CME...7/27/2017 1 Michael G. Shlipak, MD MPH Advances in Primary Care: A Critical Review of the Year’s Most Important Papers WEDNESDAY WORKSHOP …