Beyond LDL Cholesterol: Why Your Lipid Testing Strategy Determines Whether Statin Therapy Is Worth the Risk
Beyond LDL Cholesterol: Why Your Lipid Testing Strategy Determines Whether Statin Therapy Is Worth the Risk
Yoon Hang Kim, MD, MPH
Board-Certified in Preventive Medicine | Integrative & Functional Medicine Physician
One of the most consequential—and underappreciated—decisions in preventive cardiology is not whether to recommend a statin, but which lipid markers to use when making that decision. The choice between a standard lipid panel, apolipoprotein B (apoB), and lipoprotein particle number (LDL-P) is not merely academic. It determines who gets treated, how aggressively, and whether the risk-benefit calculus actually holds in your specific clinical context.
This article examines why the widely cited "NNT 100 / NNH 40" heuristic for statins in primary prevention is not a fixed biological truth but a population-level average—one that shifts substantially depending on how atherogenic burden is defined and measured.
Disclaimer: This article is for educational purposes only and does not constitute individualized medical advice. All clinical decisions should be made in partnership with a qualified healthcare provider who can evaluate your personal history, medications, and risk factors.
What the "NNT 100 / NNH 40" Heuristic Actually Means
NNT—number needed to treat—and NNH—number needed to harm—are tools for communicating absolute risk in plain terms. They transform relative risk reductions into something clinically tangible: how many people must be treated to benefit or harm one person over a defined time period.
For statins in lower-risk primary prevention, evidence-based summaries have used figures roughly in this range:
- NNT to prevent one major cardiovascular event over approximately five years: often cited in the range of 50 to over 200, depending on baseline risk and the endpoints measured.
- NNH for statin-attributable adverse effects: myalgia and related muscle symptoms have been cited at an NNH of approximately 21 in some analyses; new-onset diabetes at an NNH of approximately 38 to 204 depending on the statin dose and population studied.
The TheNNT.com summary for statins in low-risk primary prevention found a 0.46% absolute benefit for nonfatal heart attacks (NNT = 217) and noted an NNH of 38 for new-onset diabetes specifically in patients treated with high-dose atorvastatin in the SPARCL trial—a figure that went initially undisclosed. (TheNNT, 2023)
The "100/40" framing is therefore a reasonable heuristic shorthand for intermediate-risk primary prevention contexts, but it carries important caveats: it is population-averaged, highly sensitive to baseline risk, and was derived using LDL-C-based entry criteria—not apoB or particle number.
A Critical Caveat: NNT and NNH Cannot Be Directly Netted Against Each Other
Before going further, one logical pitfall in the "100/40" shorthand deserves explicit attention, because it is easy to misread. The figures invite a tempting but flawed inference: "one benefit per 100 treated versus one harm per 40 treated" seems to imply that harms outnumber benefits by roughly two and a half to one. That inference is only valid if the harm counted at "40" is of comparable clinical severity to the benefit counted at "100." NNT and NNH are not interchangeable units that can be subtracted or divided unless the two outcomes are matched for severity and measured over the same time horizon.
This distinction is not academic. The "40" in the shorthand can refer to very different harms depending on the source:
- If it refers to new-onset diabetes (NNH ≈ 38 with high-dose atorvastatin in the SPARCL trial), the comparison against a prevented major cardiovascular event is defensible—both are serious, largely persistent conditions over a similar timeframe.
- If it is treated as a stand-in for statin harm in general, it collapses toward the myalgia NNH of approximately 21—and a prevented heart attack or stroke is now being weighed against a muscle symptom that is typically reversible on discontinuation. That is not a logically legitimate one-to-one comparison, even though both numbers are individually accurate.
This is precisely why TheNNT.com, the source most often associated with this framing, explicitly declined to perform the arithmetic. Their "red" (not recommended) rating for statins in low-risk primary prevention is described by the authors as a value judgment, not a computed net benefit. The honest reading of "100/40" is not that harm mathematically outweighs benefit, but that in a low-risk population the benefits are modest, the harms are non-trivial, and the decision genuinely belongs to an informed client rather than to a formula.
Understanding the Three Lipid Metrics
Standard Lipid Panel: Cholesterol Mass, Not Particle Count
The conventional lipid panel measures cholesterol mass within lipoprotein fractions—LDL-C, non-HDL-C, HDL-C, and triglycerides. LDL-C is typically calculated using the Friedewald equation rather than directly measured, and it reflects how much cholesterol is carried inside LDL particles, not how many particles exist.
This distinction matters enormously when particles vary in their cholesterol content. In individuals with insulin resistance, elevated triglycerides, obesity, or metabolic syndrome, LDL particles tend to be smaller and more cholesterol-depleted—meaning particle count can be substantially elevated even when LDL-C appears normal or only modestly raised.
Apolipoprotein B: A Direct Count of Atherogenic Particles
ApoB offers a fundamentally different measurement. Every atherogenic particle—VLDL, IDL, LDL, and Lp(a)—carries exactly one apoB-100 molecule. ApoB therefore provides a direct count of total circulating atherogenic particles, regardless of how much cholesterol each one contains.
Multiple post-hoc analyses, population studies, and meta-analyses have demonstrated that apoB is generally at least as accurate as—and in discordant populations more accurate than—LDL-C as a predictor of ASCVD events. (Future Cardiology, 2025) The evidence is worth stating precisely rather than in slogans. The Boekholdt et al. individual patient data meta-analysis (JAMA, 2012), pooling 62,154 statin-treated participants across 8 randomized trials, found that on-treatment LDL-C, non-HDL-C, and apoB were each associated with residual cardiovascular risk; the strongest association was for non-HDL-C, and the difference between apoB and LDL-C in that particular analysis did not reach statistical significance. (JAMA, 2012)
The more striking discordance signal comes from the Johannesen et al. analysis of 13,015 statin-treated patients in the Copenhagen General Population Study (JACC, 2021). In that cohort, high apoB and high non-HDL-C were associated with increased all-cause mortality and myocardial infarction, whereas high LDL-C, when discordant, was not—supporting apoB as the better marker of residual risk specifically in the situations where the markers disagree. (JACC, 2021)
LDL Particle Number (LDL-P): Conceptually Equivalent, Practically Similar
NMR-based lipoprotein profiling directly quantifies the number of LDL particles in circulation. Conceptually, LDL-P and apoB capture similar information—both are particle-based metrics rather than cholesterol-mass metrics. In most clinical settings, apoB has emerged as the more practical of the two: it is simpler, less expensive, more widely available, and better standardized. It is worth being precise about its guideline status, however: in the current ACC/AHA and ESC/EAS frameworks, LDL-C remains the primary target and apoB is positioned as a secondary or optional target, valued especially in patients with high triglycerides, diabetes, obesity, or very low LDL-C. The direction of travel in the literature favors apoB, but it is not yet the primary guideline target. When LDL-P and apoB disagree, the discordance is often explained by the inclusion of non-LDL atherogenic particles (particularly VLDL remnants) in the apoB count but not LDL-P.
The Mechanism Behind Discordance
Atherosclerosis is driven by the entry and retention of apoB-containing particles within the arterial intima—not by the quantity of cholesterol inside those particles. A small, cholesterol-depleted LDL particle is no less capable of traversing the endothelium and becoming trapped in the arterial wall than a large, cholesterol-rich one.
This mechanistic reality is why particle-number metrics outperform LDL-C for risk prediction in discordant populations. A 2025 systematic review in the Journal of Clinical Lipidology compiled 15 discordance studies involving 593,354 participants across diverse populations and confirmed that apoB and LDL-P consistently outperform LDL-C and non-HDL-C when the two categories diverge. (Journal of Clinical Lipidology, 2025)
The clinical populations most vulnerable to LDL-C underestimation are precisely those commonly encountered in integrative and functional medicine practice:
- Metabolic syndrome and insulin resistance
- Type 2 diabetes and prediabetes
- Hypertriglyceridemia
- Obesity, particularly visceral adiposity
- Chronic inflammatory states (including autoimmune disease, Long COVID, and mast cell activation syndrome)
How Lipid Metric Choice Reshapes the NNT/NNH Calculus
The published NNT and NNH figures for statins in primary prevention were derived from trials that enrolled participants based on LDL-C thresholds, not apoB or LDL-P. This has two important consequences.
Consequence 1: Risk Stratification Errors That Flow in Both Directions
If you stratify solely by LDL-C, you will systematically misclassify a predictable subset of clients:
- Normal or borderline LDL-C with elevated apoB/LDL-P: These clients carry high atherogenic particle burden that LDL-C underestimates. They appear lower-risk than they are and may be undertreated. Because absolute benefit scales with baseline absolute risk—the relative risk reduction from LDL-lowering therapy is roughly constant, so a higher true starting risk yields a larger absolute risk reduction—their personal NNT, if therapy were offered and actually lowered their particle number, would likely be more favorable than the LDL-C-based heuristic implies. The key qualifier is that the therapy must reduce apoB, not merely move the LDL-C figure.
- Elevated LDL-C with low apoB: These clients often carry large, buoyant, cholesterol-rich particles with low actual particle count. Their LDL-C appears alarming but their atherogenic burden is modest. The population-level NNT of ~100 likely overestimates their individual benefit, because their true absolute risk—and therefore the absolute risk reduction available—is lower than the LDL-C reading implies. The harm side, meanwhile, does not shrink correspondingly. When benefit is smaller but harm is not, the balance shifts (subject to the severity-matching caveat above), which is a reason for caution rather than reflexive treatment.
The 2025 ATTICA study 20-year follow-up, published in the European Journal of Clinical Investigation, found that in discordance analysis, elevated apoB independently predicted increased 20-year ASCVD risk regardless of non-HDL-C and Lp(a) status, and that incorporating apoB helped explain part of the previously residual risk. Importantly, the authors noted that this independent predictive effect was observed only in individuals with concomitantly elevated LDL-C—a nuance worth keeping in mind, since it means apoB refined risk most clearly within a specific subset rather than uniformly across every lipid profile. (European Journal of Clinical Investigation, 2025)
Consequence 2: Treatment Adequacy Cannot Be Judged by LDL-C Alone
Two clients with identical on-treatment LDL-C values can have vastly different residual apoB levels and vastly different residual risk. This is not a theoretical concern—it is a predictable consequence of variable particle cholesterol content, particularly in populations with metabolic dysfunction.
A 2022 retrospective cohort study in Frontiers in Endocrinology found that among statin-treated participants, elevated apoB was significantly associated with the severity of coronary atherosclerosis and residual coronary artery disease risk, while elevated LDL-C in the same population was not significantly associated with either outcome. This particular study was small (131 participants) and retrospective, so it should be read as consistent with the larger body of evidence rather than as definitive on its own; the broader meta-analytic and large-cohort data cited above carry the weight here. (Frontiers in Endocrinology, 2022)
The clinical implication: titrating statin therapy to an LDL-C target without checking apoB may leave a client with high residual particle burden—and residual risk—despite achieving apparent LDL-C "control." The full benefit assumed in trial-derived NNT calculations cannot be realized if atherogenic particle suppression is incomplete.
Translating This to Practice: An Integrative Medicine Perspective
Who Benefits Most from apoB Testing?
In a practice centered on metabolic health, chronic inflammation, and individualized risk, apoB is most actionable when:
- LDL-C and triglycerides are discordant (LDL-C normal, TG elevated).
- The client has insulin resistance, prediabetes, or metabolic syndrome.
- Standard risk calculators place the client in a borderline or intermediate 10-year risk category where the benefit-harm balance is genuinely uncertain.
- The client has autoimmune disease, MCAS, or chronic inflammatory burden—conditions that may independently accelerate atherosclerotic progression.
- The client has responded to statin therapy with LDL-C reduction but residual risk concerns persist.
What to Do With the Information
When apoB is elevated despite acceptable LDL-C:
- Recalibrate cardiovascular risk upward. The true atherogenic burden is greater than the lipid panel suggests.
- Consider whether lipid-lowering therapy—statin or non-statin—is warranted at a lower LDL-C threshold than guidelines specify.
- Use apoB, not LDL-C alone, to monitor treatment adequacy.
When apoB is low despite elevated LDL-C:
- Avoid reflexive statin initiation based solely on LDL-C numbers.
- Consider coronary artery calcium (CAC) scoring and Lp(a) to further define risk before prescribing.
- If a decision is made to defer statins, frame it explicitly using the NNH data—the harm side of the equation becomes relatively more prominent when true benefit is smaller.
ApoB Target Ranges to Consider
While guidelines vary, the secondary apoB goals from the 2019 ESC/EAS dyslipidemia guidelines (carried into the 2025 focused update) provide reasonable reference targets, each corresponding to an LDL-C target:
- Moderate risk: apoB < 100 mg/dL (corresponding LDL-C < 100 mg/dL)
- High risk: apoB < 80 mg/dL (corresponding LDL-C < 70 mg/dL)
- Very high risk or established ASCVD: apoB < 65 mg/dL (corresponding LDL-C < 55 mg/dL)
These are population-derived targets, not absolute thresholds, and should be interpreted in the context of the full clinical picture.
The NNT Is Not a Fixed Number—It Is a Function of Your Measurement Strategy
The "NNT 100 / NNH 40" heuristic is a useful starting point for conversations about statin benefit and risk in primary prevention. But it is population-averaged and was derived from LDL-C-defined trial populations. It is not a biological constant that applies uniformly to every client who walks into your practice.
The effective NNT in your own clinical context improves when you:
- Use apoB or LDL-P to correctly identify clients with high atherogenic particle burden who would be missed by LDL-C alone.
- Titrate therapy until particle number—not just cholesterol mass—is suppressed.
- Avoid treating low-apoB, low-risk clients based solely on modestly elevated LDL-C, where the effective NNT worsens and the NNH remains constant.
The harm side of the equation requires a more careful statement than is often made. It is true that the NNH for a given statin at a given dose does not change based on which lipid marker you use to monitor the client—diabetes and myalgia risk are driven by client susceptibility, the specific statin, its dose, and how rigorously adverse effects are sought and attributed, none of which depend on whether you happen to be tracking apoB or LDL-C. In that narrow sense, harm is independent of your choice of monitoring marker while benefit is not.
However, this symmetry should not be overstated. Titrating to an apoB goal, as recommended above, will in many discordant clients mean escalating to higher-intensity statin therapy than an LDL-C target alone would have prompted—and higher intensity carries a larger diabetes signal. So while the harm rate per unit of drug exposure is marker-agnostic, the total harm exposure can rise precisely because apoB-guided care pushes toward more aggressive treatment. The accurate framing is therefore: harm is independent of the marker you monitor, but not independent of the treatment intensity that marker drives you toward. This is not an argument against apoB-guided care; it is an argument for making the intensified benefit-and-harm tradeoff explicit with the client rather than assuming the harm side stays fixed.
Advanced lipid testing is therefore not a luxury or an upselling opportunity. In the populations most often seen in integrative and functional medicine practice—those with metabolic dysfunction, inflammatory burden, and complex comorbidities—it is a precision tool that determines whether the statin conversation you are having is based on accurate data. To learn more about my approach to individualized cardiovascular risk assessment, visit www.directintegrativecare.com.
References
1. TheNNT.com. Statins in Persons at Low Risk of Cardiovascular Disease. Updated 2023.
About Dr. Kim
Yoon Hang "John" Kim, MD, MPH, is a board-certified Preventive Medicine physician with more than 20 years of clinical experience in integrative and functional medicine. He completed fellowship training as an Osher Fellow under Dr. Andrew Weil at the University of Arizona, holds UCLA certification in medical acupuncture, and is board-certified in both Preventive Medicine and Integrative & Holistic Medicine through the American Board of Integrative & Holistic Medicine.
Dr. Kim specializes in low-dose naltrexone, autoimmune and inflammatory conditions, chronic pain, integrative oncology, fibromyalgia, ME/CFS, mast cell activation syndrome, and mold toxicity. He is the author of three books, including the LDN Primer and a clinical LDN textbook, and has published more than 20 peer-reviewed articles.
Professional site: www.yoonhangkim.com | Clinical practice: www.directintegrativecare.com
