Broad Panels or Targeted Testing: Which Serves You Better

The Arithmetic of False Findings

This is not a soft argument, it follows from how reference ranges are constructed.

A reference interval is typically set to include the central 95 per cent of a healthy reference population, which means 5 per cent of healthy people fall outside it for any given analyte. Those results are not errors, they are the definition working as intended.

With 20 analytes, the probability that at least one falls outside range in a healthy person is high. With 100, several flagged results are expected. Each then requires interpretation, and some prompt a repeat test, an additional test, imaging or a referral.

The downstream cost is real. Cascade investigation following an incidental finding is a well-documented harm, involving procedural risk, cost and anxiety. It is the reason clinical guidelines generally recommend testing to answer a question rather than testing broadly.

The asymmetry that makes it worse: a false positive generates action, while a false reassurance generates none. Broad panels produce both, and only one of them is visible.

What a Deliberate Panel Contains

That is roughly twenty analytes, most of them inexpensive and standard. Each earns its place by answering a question whose answer would change something.

The test for any addition: name the two possible results and what you would do differently for each. If the answer is the same, the analyte is information rather than a decision, and it carries a false-finding cost without a corresponding benefit.

What Broad Panels Typically Add

Understanding what fills the gap between twenty and a hundred explains the trade-off.

Broad hormone panels in asymptomatic people. More analytes, more incidental abnormalities, and results that are frequently uninterpretable without timing context such as cycle day or morning draw.

Multi-cytokine inflammatory panels. hs-CRP does the job at a fraction of the cost with better assay reliability, and interleukin-6 and TNF-alpha have short half-lives and higher variability.

Extensive lipid subfractionation. Rarely changes management beyond what apoB and triglycerides already indicate.

Novel cardiovascular markers such as oxidised LDL, myeloperoxidase and Lp-PLA2, with limited clinical validation for routine use.

Broad micronutrient panels, where several analytes have poor status validity. Serum magnesium reflects a small fraction of body stores, and several intracellular nutrient panels have limited validation.

Tumour markers in asymptomatic people, which are generally not appropriate for screening and generate substantial anxiety and investigation.

Heavy metals without an exposure history.

Some of these are legitimate in specific contexts. As routine additions to a screening panel they add cost, findings and interpretive burden without changing decisions.

When Broad Is Genuinely Better

The case for breadth is not empty, and it applies in specific situations.

An undiagnosed problem with real symptoms. Where something is wrong and the cause is unclear, casting wider is appropriate, and that is a clinical process rather than a consumer purchase.

A first comprehensive assessment where nothing has ever been measured, since establishing a broad baseline once has value even for analytes you will not track.

Where a specific concern justifies it, such as a family history pointing at a particular system.

Where cost of the panel is negligible relative to the cost of the appointment, which is sometimes true and shifts the calculus.

Research participation, where broad data contributes to something beyond your own decisions.

The distinguishing feature in each case is that a question exists. Breadth in service of a question is investigation; breadth without one is screening, and screening a healthy person broadly has a well-documented cost.

The Interpretation Problem Scales Badly

A larger panel is harder to interpret rather than easier, for reasons worth spelling out.

Patterns get lost in lists. Six mildly abnormal values that together indicate insulin resistance with hepatic fat get reported as six separate findings with six separate recommendations. A smaller panel makes the pattern visible.

Flag-based reporting hides within-range movement, which is where much of the information lives. GGT, apoB, creatinine and fasting insulin all carry information inside their reference ranges, and a hundred-item report emphasising flags buries it.

Timing errors multiply. A broad panel drawn at one appointment cannot satisfy the conditions each analyte requires: morning for testosterone, fasted for insulin, away from exercise for creatine kinase and hs-CRP, cycle-timed for female hormones. Some results will be uninterpretable by construction.

Serial comparison becomes impractical. Tracking twenty analytes annually is feasible; tracking a hundred is not, so most of the panel is measured once and never used.

Reports lengthen and get read less. A forty-page interpretive document is less likely to produce an action than a one-page one.

A Practical Approach

Choose analytes rather than a package where the provider allows it. Most do, and the price difference is substantial.

If buying a package, decide in advance which results you will act on and disregard the rest. Writing that list before the results arrive prevents the retrospective search.

Prioritise the four commonly omitted items: apolipoprotein B, lipoprotein(a) once, fasting insulin, and a urine albumin to creatinine ratio. These change decisions and are usually absent.

Get the conditions right, since a well-chosen panel drawn during illness or after hard exercise is wasted.

Read patterns, not flags, and record numbers rather than normal or abnormal.

Repeat a small set annually rather than a large set once, since the sequence is where the information is.

Take genuinely abnormal results to a clinician, and expect to repeat before acting.

The summary: panel size is not thoroughness, deliberate selection is. Twenty analytes chosen to answer questions outperform a hundred chosen to look comprehensive, and cost less.