PREVENT cardiovasc risk calculator increases disease!

 

The American Heart Association developed their new PREVENT Cardiovascular Risk calculator in 2023 to improve upon the currently commonly used 2013 ASCVD risk calculator, modified in 2018. A new evaluation of PREVENT found that its use would be associated with a very large increase in myocardial infarction or stroke over 10 years: see ASCVD PREVENT risk assoc inc CVD JAMA2024 in dropbox, or doi:10.1001/jama.2024.12537

 

Details:

-- these researchers estimated the number of US adults who would experience changes in cardiovascular risk categorization, treatment eligibility, or clinical outcomes when comparing the anticipated results from the PREVENT equations to the results from the ACC/AHA ASCVD (atherosclerotic cardiovascular disease) risk calculator, by using the NHANES database

    -- the National Health and Nutrition Examination Surveys (NHANES) is a nationally representative cross-sectional survey of 7765 US adults aged 30-79, accessed for this study from 2011 to 2020. Response rates to these surveys were from 47% to 70%

    -- NHANES is comprised of surveys of the civilian noninstitutionalized US population, through biennial questionnaires requesting information on age, gender, race and ethnicity, and medical history; physical examination, including blood pressure levels; lab results, including serum cholesterol levels; and prescription medication data

-- the goal of this study was to compare outcomes anticipated in this NHANES population from:

    -- the pooled cohort equations (PCEs) used to create the 10-year risk of ASCVD events promoted in the 2013 guidelines and modified in 2018 to include diastolic blood pressure, LDL cholesterol, if the person was taking statins, and if they were taking aspirin, along with those in the 2013 ASCVD calculator that included age, diabetes, sex, smoking, total cholesterol and HDL, systolic blood pressure, hypertension treatment, and race

    -- with the outcomes of the PREVENT model, which incorporated estimated kidney GFR (eGFR) along with optional inclusion of urinary albumin-to-creatinine ratio, hemoglobin A1c, and a “social deprivation index” which used the person’s zip code as a (quite rough) surrogate marker of 7 social characteristics related to education, household composition, home ownership, overcrowding, car ownership, and employment rates, in a sex-specific but race-free calculator of persons 30-79yo. The results also included anticipated heart failure events, deaths, and total cardiovascular and atherosclerotic events

    -- it was notable that the 3+ million people in the derivation cohort used to develop the mathematical equation for the PREVENT risk calculator had remarkably high correlation with the results from a 3+ million external validation cohort, with only small improvements when the urine-to-albumen, A1c, and social deprivation index were added to the base model (for my critique of the PREVENT risk calculator, see  https://gmodestmedblogs.blogspot.com/2024/07/prevent-new-cardiac-risk-factor.html ).

-- this current study focused on the primary prevention of cardiovascular disease in people who were eligible for statins but did not have prior ASCVD (including stroke, myocardial infarction, or heart failure) in the whole NHANES population:

    -- median age 53; 51% women; 25% Hispanic, 23% Black, 13% Asian; age breakdowns by deciles from 30-39 to 60- 69 were evenly distributed, with few people 70 or older

    -- Baseline ASCVD risk score: 5.4 based on PCEs model, and 2.2 based on the PREVENT model

-- exclusion criteria included patients with missing lab values, pregnant women, and people with triglyceride levels greater than 400

Main outcomes: differences in predicted 10-year ASCVD risk,  ACC/AHA risk categorization, eligibility for statin or antihypertensive therapy, and projected occurrences of myocardial infarction or stroke.

Results:

-- numbers of people eligible in the total NHANES sample for statins by the above criteria (no baseline ASCVD, diabetes, or statin use):
    -- eligible with PCEs (n=909); eligible with PREVENT (n=277), newly eligible (n=4), no longer eligible (n=636), and total sample (n=7765)

 

-- after survey adjustment:

    -- mean 10-year ASCVD risk calculated using PREVENT was 4.6%

    -- mean 10-year risk using the PCEs was 9.0%

        -- the PREVENT scores were lower across all subgroups of age, gender, and race and ethnicity

        -- further analysis found that the differences in these scores were greater among persons at higher risk of cardiovascular disease

        -- people who were Black, male, and aged 70-79 had larger decreases in risk with PREVENT as compared to Hispanic persons, women, and persons aged 40-49

        -- in sum, most US adults will be assigned as lower risk by PREVENT than by PCEs

        -- overall, US adults aged 70-79 with no history of MI, stroke, or heart failure: 53.0% (51.2%-54.8%) would be assigned to lower risk categories by PREVENT; 0.41% (0.25%-0.62%) would be assigned to higher risk categories

 

-- The number of US persons receiving or recommended for statin therapy:

    -- PCEs group with ASCVD risk of at least 7.5%: 81.8 million people

    -- PREVENT group: 67.5 million

        -- difference of 14.3 million (12.6 million to 15.9 million)

            -- most of this decline would occur among:

                -- men and adults aged 50-69yo

                -- Black adults versus white adults (-9.9% vs -8.0%)

 

-- the number of US persons receiving or recommended for high-intensity statin therapy:

    -- PCEs group: 55.9 million

    -- PREVENT group: 51.1 million

        -- difference of 4.77 million (3.99 million to 5.54 million) who would be on moderate-intensity statins instead of the high-intensity statins

        -- this difference encompasses 92.8% (90.0%-95.7%) of the instances where high-intensity statin initiation would have been recommended based on ASCVD risk as calculated using PCEs

 

--Eligibility for antihypertensive therapy:

    -- PCEs group, with ASCVD risk score of at least 10%: 75.3 million persons

    -- PREVENT group: 72.7 million persons

    -- difference 2.62 million (2.02 million to 3.21 million)

        -- the major differences were in men and adults aged 50-69

        -- and more often Black adults would become ineligible per the PREVENT assessment as compared with white adults (-2.0% to -1.4%)

 

--   the number of US adults who would lose recommended eligibility by PREVENT for either statin therapy or antihypertensive therapy was estimated at 15.8 million (14.2 million to 17.6 million)

 

-- New-onset ASCVD events and new-onset diabetes:

    -- the baseline adherence rate to statins in the population is quite low: an estimated 40.5% of US adults who should have been recommended for primary prevention with statin therapy by the guidelines were actually receiving statins; this number likely reflects some non-prescribing by clinicians and some by patient nonadherence to the meds (more general information in the commentary below):

        -- calculations reflecting lower statin adherence would translate to 14.3 million fewer recommendations for statin therapy associated with using PREVENT and would translate to 5.78 million fewer adults receiving statins and benefiting from the 25% relative risk reduction for ASCVD  

        -- over 10 years, this decrease in statin use would be projected to result in 77,000 additional ASCVD events and 57,800 fewer occurrences of new-onset diabetes

-- this all represents an additional 7.93 million individuals with eligibility changes for either statin or antihypertensives and 77,000 additional ASCVD events over 10 years and 57,800 fewer occurrences of diabetes

-- 2.62 million fewer persons would have recommendations for antihypertensive therapy, resulting in 1.74 million fewer (66.6%) receiving blood pressure medications and 31,600 additional ASCVD events

    -- based on these numbers, the combination of decreased eligibility changes in PREVENT for statins and antihypertensive therapies would be projected to result in an estimated 107,000 additional ASCVD events over 10 years, affecting men more than women (0.077% versus 0.039%), but similar proportions for Black and white adults

-- only 4 persons in the NHANES sample would be newly eligible for statin therapy in the PREVENT model, translating to 37,000 US adults when scaling the NHANES population to the overall US population. All 4 had reduced kidney function with a mean eGFR of 23 (CKD is on the PREVENT risk scale)

-- by contrast, 636 participants representing 14.3 million US adults would no longer be recommended for statin therapy through PREVENT due to the differences in predicted risk (5.3% for PREVENT and 10.4% for PCEs group)

Commentary:

-- the PREVENT evaluation for ASCVD includes younger adults (down from the 40-79 for the ASCVD risk calculator to 30-79 for PREDICT), eGFR values to assess kidney function, laudibly does not use race as an input, and relies on many larger databases derived both from research and electronic medical records. The PREVENT investigators did use a validation cohort, though questions have been raised in the medical literature specifically dealing with the accuracy of the electronic medical record information

    -- another concern is that the data on cardiovascular mortality is not constant over time, and in particular cardiovascular disease mortality rates increased in 2020 and remained high in 2022, with over 228,000 excess CVD deaths (ie using baseline cardiovascular events from older studies might not reflect the current situation): https://www.ajpmonline.org/article/S0749-3797(23)00465-8/fulltext

-- this current analysis did produce the rather profound finding that the proposed PREVENT risk factor approach could reduce the number of US adults recommended for statins by about 14.3 million, and by 2.62 million recommended for antihypertensive therapy. Through data from older studies, this should translate into 77,000 additional ASCVD events over 10 years from not being on statins and 31,600 for not being on antihypertensives

-- It is important to emphasize, as has been done many times in prior blogs, that documented ASCVD risk factors are much broader than the traditional blood pressure and LDL cholesterol levels, though these items are major components: see http://gmodestmedblogs.blogspot.com/2023/10/update-ascvd-risk-factor-critique.html; other documented risk factors include increased ASCVD risk attributable to many psychosocial variables (stress, depression, racial dissparities, any alcohol consumption, unhealthy diet, lack of exercise), environmental ones (air pollution, microplastics), and medical ones (pretty much anyone with systemic inflammation, including those with visceral obesity, even well-controlled HIV, chronic kidney disease including marginally increased creatinine or urinary albumen excretion, psoriasis/rheumatoid arthritis/lupus, migraine, peripheral artery disease, hyperuricemia/gout). the UK developed a more inclusive but cumbersome ASCVD risk calculator, the QRISK3: http://gmodestmedblogs.blogspot.com/2017/08/a-new-atherosclerosis-risk-calculator.html 

-- this article basically shows that using the new PREVENT calculator is likely to be not only leading to more cardiovascular events over a 10-year period, but it undercuts the critical understanding that atherosclerotic disease begins very early (in persons in their teens), is a progressive process, and does not lead to clinical atherosclerotic events such as myocardial infarction and stroke until much later in life. And, instead of decreasing the use of statins and other preventative strategies, it is more important to address cardiovascular risk factor modification at a very young age, principally through nonpharmacologic approaches, but also with medications as needed.

    -- the accretion of coronary artery lipid plaques is a cumulative process that begins early (in teens), that the cumulative LDL cholesterol exposure (area under the curve) matters, and, in particular, that the amount of time that the coronaries are exposed to high LDL levels is important (eg, those in the CARDIA study of young adults with high LDLs had more cardiovascular events after reaching 40 years old: see lipid teens predicts future JACC2020 in dropbox, or doi.org/10.1016/j.jacc.2020.07.059)

    -- And the utility of a 10-year risk predictor in younger people is pretty useless. Using even the ASCVD criteria of a threshold of at least 7.5%, very few younger persons would qualify for statins

-- A major issue with statins is that the adherence rate is pretty abysmal: a 2010 study of the New Jersey Medicaid persons found about a 50% adherence rate, pretty much independent of the underlying reason for statins (whether primary prevention and a recent MI). another more recent study with a large database confirmed these findings: https://pmc.ncbi.nlm.nih.gov/articles/PMC6405715/ , so following the ACC/AHA 2013 guideline recommendations on prescribing statins to all qualifying patients, it  makes sense to check lipids pretty regularly on patients to make sure that these pretty crucial meds are being taken (i check lipids annually on almost all of my patients and, anecdotally, do not find that i am particularly accurate in my assumptions on either medication adherence or nonadherence)

-- it should be noted that statin-related diabetes tends to happen in those with more advanced glucose intolerance, and:

    -- the general increased incidence of diabetes attributed to statins is pretty small: about 0.1% per year, or about 1% over 10 years

    -- people with glucose intolerance have a significantly increased cardiovascular risk if remaining in this "prediabetes" range: http://gmodestmedblogs.blogspot.com/2022/11/prediabetes-increases-risk-heart.html; so even if people stayed in the "prediabetes" range, they are still at increased risk of ASCVD but are not included in any of the ASCVD risk calculators

    -- several mathematical modeling studies have come out finding that the net benefit of statins outshines the potentially worse outcomes of becoming diabetic (and, of course, pretty much everyone with diabetes should be on a statin independent of their lipid levels or having known heart disease)

Limitations:

-- this study, though well-conducted, did not assess real ASCVD outcomes (as was done in the validation cohort in the PREVENT study), but instead used information from prior studies of the expected outcomes from not putting people on statins or antihypertensive medications given their baseline risk

-- there are now widely used medications that moderate ASCVD risk, specifically the GLP-1 receptor agonists and the SGLT-2 inhibitors. these were not included in the 2013 or 2018 ASCVD risk calculators, or in PREVENT (the SGLT-2 inhibitors were FDA-approved in 2013; the GLP-1's were approved in 2005, were used in the UK in the 1990s). These meds are now remarkably commonly prescribed (and for a variety of good reasons) and really should be incorporated into newer risk models for cardiovascular and renal outcomes

-- though PREVENT did increase the age range tor 30-79 instead of 40-79, those 30-39 year olds were not recommended for primary prevention:

    -- that is certainly the case for women who could become pregnant when on lipid-lowering therapy

    -- the main focus on younger people is nonpharmacologic therapies

    -- but, as noted above, lipid problems begin in youth and need to be assessed and treated early

        -- lipid streaks are frequently found in teenage coronary arteries

        -- the American Association of Pediatrics recommends universal lipid testing in kids at age 9-11yo and again at 17-21yo, with screening kids as young as 2yo if there is a family with history of heart disease, high cholesterol, or other cardiovascular risk factors such as diabetes, hypertension or severe obesity (ie, lots of younger kids)

        -- so, meds may be necessary in some youths with high risk factors despite nonpharmacologic approaches to stem to accumulation of lipid plaques

-- it should be emphasized that the PREVENT model is not in current use, it may well be modified, though the ACC/AHA goal is to include it in future guidelines

-- this study did not include some of the additions to PREVENT including the hemoglobin A1c, urine albumin-to-creatinine ration, and zip code; however, adding these to the PREVENT risk model did not add substantially to the risk scores

-- the NHANES database did not include medication indications or dosages

So,

-- the general approach to cardiovascular disease risk reduction has evolved substantially from a construct based on targets of blood pressure and cholesterol and instead using strategies based on absolute risk, taking into account the constellation of risk factors. The PREVENT risk calculator results was designed to take some of these into account (renal disease, a not-so-great marker of social issues, hemoglobin A1c). but:

    -- this current study questions the accuracy of the benefit of PREVENT as an indicator of risk and a guide to therapy

    -- so, if implemented, PREVENT could dramatically reverse the increases in people on statins engendered by the risk ASCVD calculator:

        -- it would decrease the number of people on statins by 12.8 million US adults and 4.77 million would be on lower intensity statins

        -- and it a reduction of 2.62 million in people receiving the recommended antihypertensive therapy
    -- these are major changes, especially if validated by other studies, in projected morbidity and mortality, from taking medications (as well as nonpharmacological therapies), mostly with very low rates of severe adverse effects

 

geoff

-----------------------------------

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@bidmc.harvard.edu

to get access to all of the blogs:  go to http://gmodestmedblogs.blogspot.com/ to see the blogs in reverse chronological order

or you can just click on the magnifying glass on top right, then type in a name in the search box and get all the blogs with that name in them