The Most Reassuring Appointment in Medicine
Every year, tens of millions of healthy adults keep the same appointment: the annual physical, the executive medical, the wellness visit, a gown, a blood draw, a cholesterol lecture, and reassurance from a qualified professional. Britain's NHS offers free checks to everyone 40 to 74, American employers buy executive physicals by the thousand, and Medicare pays for an Annual Wellness Visit. The logic feels airtight: find disease early, live longer.
In 2012, a Danish team tested whether the logic holds. Lasse Krogsbøll's Nordic Cochrane Centre team gathered every randomized trial of systematic health checks in unselected adults: fourteen trials, 182,880 people, follow-up to 22 years. The BMJ result was a shrug: no mortality reduction of any kind, though seven years later the team updated the review and made the shrug more precise.
17 Trials, 251,891 People, One Number
The 2019 update pooled 17 randomized trials, 15 with outcome data, covering 251,891 adults across Europe and North America, and eleven of those trials tracked total mortality, which meant 233,298 participants, 21,535 deaths, and a median follow-up near nine years, the kind of sample that leaves nowhere for a real effect to hide. The pooled risk ratio was 1.00, with a 95% confidence interval of 0.97 to 1.03: not approximately one, but exactly 1.00, graded as high-certainty evidence with zero heterogeneity between trials. Cancer mortality came in at 1.01 with high certainty, cardiovascular mortality at 1.05 with moderate certainty, heart attacks at 0.98, strokes at 1.05, and every estimate pointed at nothing.
The checks did one thing reliably: in one trial, invited adults accumulated 20 percent more new diagnoses per person over six years than controls, with more hypertension, cholesterol, and prescriptions. The machine found more disease and treated more of it, and that is the whole story in a paragraph. Mortality did not move.
The Arithmetic Nobody Did
The arithmetic nobody did starts here: this null is not the underpowered kind, and with 21,535 deaths pooled, the data rule out any true mortality benefit larger than about 3 percent in relative terms. Roughly 9.2 percent of participants died during follow-up, so a 3 percent relative cut equals 0.28 percentage points in absolute terms, which means the most optimistic reading the data allow is one death averted for every 361 adults invited to health checks and followed for nine years, a number-needed-to-invite that would embarrass any screening program. The point estimate says the number is infinite, yet for scale, statins cut major vascular events roughly 21 percent per 1 mmol/L of LDL lowered in 27 trials and 175,000 people, so these checkup trials were easily large enough to detect a benefit one-seventh the size of a standard statin effect. They found nothing. Precisely nothing.
Now consider the diagnosis arithmetic, 20 percent more labels per person and zero fewer deaths, because had the extra diagnoses caught killers early, the mortality curve would have bent. It did not bend. The marginal diagnoses were therefore overdiagnosis: conditions found and treated without changing anyone's fate.
The Strongest Counterargument
The fiercest defense of the physical writes itself: most of these trials are old, some reaching back to the 1960s, testing pre-statin checkups without modern blood-pressure drugs or organized cancer screening, so a null from 1985 may not describe 2026. The comparison has a deeper flaw, because the control groups were never abandoned but received usual care from primary-care clinicians in Denmark, Britain, and Sweden who were already measuring blood pressure and prescribing preventively during visits for other reasons. The trials measured the marginal value of a systematic invitation layered on top of decent usual care, not checkups versus nothing, though Krogsbøll himself cautioned that the results should not be read as evidence against prevention itself.
What counts as a benefit is the next objection, and a 2021 JAMA review from Northwestern re-ran the literature through March 2021, confirmed the mortality null, yet found checks improved detection of depression and hypertension, risk-factor control, screening uptake, and self-rated health. In a Danish trial of 1,104 adults, invitees were more than twice as likely to start an antidepressant within a year. No lives were extended in the data, though some lives were plausibly improved, because mortality is not the only currency that matters.
The third objection is population, since the review excluded over-65s and anyone with known disease or risk factors, testing checkups on the healthy and unselected, precisely the group least likely to benefit. Point the same instrument at the already ill and the picture changes, as a recent cohort of 625,279 elderly Chinese adults with hypertension showed when it linked regular checkups to substantially lower cardiovascular and all-cause mortality, though the study was observational, so healthier habits among attendees could explain part of the gap, and even so it draws the boundary honestly, because the null belongs to healthy adults, not to everyone.
Two objections remain, starting with attendance: many invitees never attended, which drags any real effect toward zero in intention-to-treat analysis. And the review's senior author, Peter Gøtzsche, was expelled from Cochrane in 2018 after a bitter governance dispute, a fact critics cite endlessly, yet the finding survives both objections: the 2019 update carried high-certainty GRADE ratings, and an independent team confirmed the verdict two years later. You cannot expel a risk ratio of 1.00.
What We Didn’t Prove
The review never tested the modern executive physical, the multi-thousand-dollar morning of full-body scans and forty vials of blood, because almost no randomized trials of that product exist, which is its own scandal. It excluded over-65s, compared invitation against usual care, and followed people a median nine years, so lifetime effects are unknown, and it barely measured reassurance, a doctor who knows you, or the conversation that ends a smoking habit. One caution points the other way, since a smaller 2014 meta-analysis found cardiovascular mortality odds 30 percent higher in checked groups, a signal from six trials the large Cochrane pool does not confirm.
The Bottom Line
Seventeen randomized trials and a quarter-million adults say the systematic health check does not extend life, and the risk ratio sits at exactly 1.00 with high certainty while diagnoses climb. The checkup is a superb detection device wired to an unproven treatment theory. Keep the components that carry individual evidence, and stop treating the bundle as medicine's most important appointment.
What You Can Do
Do not fire your doctor; re-target the visit toward screenings with randomized evidence behind them, blood pressure, cholesterol, diabetes testing if at risk, colorectal screening from 45, because the components work but the bundle does not. If your employer offers an executive physical, take it for what it is, reassurance, convenience, a few genuinely useful tests, not extra years. Track your own numbers between visits, because a home blood-pressure cuff and scale capture most of what checks improve. If you have hypertension, diabetes, or heart disease, set this article aside, since follow-up for the already ill is a different intervention with different evidence. Spend the marginal effort where trials show payoff: blood-pressure control, statins if your risk warrants them, never smoking or quitting, because prevention works, even if the annual ritual is simply not where its working parts live.