How the Standardized Mortality Ratio Works

The standardized mortality ratio answers a narrow but useful question: did this population experience more deaths than a reference population of the same age composition would have? It is the ratio of observed deaths to expected deaths, where expected deaths come from applying the reference population's stratum-specific rates to the person-years your cohort actually accumulated. This is indirect standardization. A ratio of 1.00 means the two populations match, and the figure is conventionally multiplied by 100 so that 116 reads as 16% excess mortality.

Why Person-Years Rather Than Headcount

Cohort members rarely contribute equal observation time. People are recruited on different dates, some are lost to follow-up, and some die partway through the study. Person-years credit each individual for exactly the interval they were at risk, so a participant followed for six months contributes half a person-year. Substituting headcount inflates exposure and systematically pulls the ratio downward whenever follow-up is incomplete, which is why the denominator is always expressed in person-time.

Confidence Limits and Significance

Death counts are Poisson-distributed, so the interval around an SMR is asymmetric — the upper limit sits further from the point estimate than the lower one, especially when counts are small. This calculator applies Byar's approximation to the exact Poisson limits on the observed count, then divides by expected deaths, which is treated as fixed because it derives from a large reference population. If the resulting interval excludes 1.00, the difference is significant at your chosen level. The accompanying z-statistic and two-sided p-value provide the same test in a different form. With fewer than about five observed deaths, treat all of these figures cautiously; the approximation degrades and the interval becomes very wide.

Age Stratification Changes the Answer

A single crude reference rate is only adequate when your population's age structure closely resembles the reference. Enabling age stratification computes expected deaths band by band and sums them, which is what standardization is actually for. The same 145 observed deaths can produce an SMR above or below 1.00 depending purely on whether the cohort's age distribution is accounted for, so the stratified figure should be preferred whenever band-level person-years are available.

Interpretation Cautions

Two SMRs standardized indirectly against the same reference are not strictly comparable with each other, because each is weighted by its own population's age structure; direct standardization against a common standard is the tool for that comparison. Occupational cohorts also routinely show all-cause SMRs below 1.00 through the healthy worker effect, since employment itself selects for baseline health. Finally, an elevated ratio establishes association, not causation — confounding by smoking, socioeconomic status and referral patterns can produce identical signals. This calculator is intended for epidemiological and educational analysis and is not a clinical or regulatory decision tool.