Timing, thresholds, and what can make a trend uninterpretable

Either a fasting or nonfasting specimen is acceptable for baseline and most follow-up lipid testing. The one specimen-driven trigger to know is that a nonfasting triglyceride result at or above 400 mg/dL calls for a repeat fasting specimen, because a Friedewald-calculated LDL-C is not considered valid once triglycerides reach that level. This matters directly for lipid-lowering therapy: a triglyceride-lowering drug (fenofibrate, icosapent ethyl, or a PCSK9-directed add-on in a patient with elevated triglycerides) can move a patient's triglyceride value across that 400 mg/dL line between baseline and follow-up. When that happens, one of the two calculated LDL-C values may not exist or may not be comparable to the other, independent of any true change in LDL particle burden.

A second source of an apparent trend that is not a treatment effect is a change in the LDL-C calculation method itself. Friedewald, Martin-Hopkins, and Sampson-NIH equations can produce different numeric results from the identical specimen. Laboratory accreditation standards require that the equation used to produce a calculated result stay identifiable to the ordering clinician, precisely because an equation change between two time points can look like a treatment-related change when it is not. Non-HDL-C, calculated as total cholesterol minus HDL-C, has no triglyceride-based estimation step and stays calculable and interpretable at any triglyceride level, which makes it a more stable comparator than calculated LDL-C when a patient's triglycerides are changing across a treatment course.

Acute illness is a third confounder, and it is not a drug effect: acute coronary syndrome, infection, surgery, or significant physiologic stress transiently lowers total cholesterol and LDL-C and can alter triglycerides and HDL-C independent of any medication change. A treatment-response lipid panel drawn during acute illness is either deferred or interpreted with that confounder stated explicitly, because it is not comparable to a stable-state baseline or follow-up value.

Before reading a lipid trend as a drug effect, confirm the two draws sit inside the locally applicable comparison protocol, on the same reported equation, and outside a period of acute illness; if any of those is not true, say so rather than letting the number speak for itself.