Rates and Standardization

Comparing how often deaths and diseases occur in populations of different sizes and ages. Counts, proportions, ratios and rates; person-time; incidence and prevalence; crude and age-specific rates and the Poisson model for their precision; then direct standardization with a standard population, its choice and its gamma confidence interval, and indirect standardization with the standardized mortality ratio. Worked on U.S. deaths from CDC WONDER and on Maine and Alaska, where the crude death rates and the age-adjusted rates rank the two states in opposite orders.

In 2021, 1,258 of every 100,000 residents of Maine died, against 847 of every 100,000 in Alaska. Read at face value, the figures say that living in Maine carried half again the risk of dying. But Maine had the oldest population of any state that year, with a fifth of its residents aged 65 or over. Alaska had one of the youngest. And death rates rise roughly a thousandfold between childhood and the oldest ages. So which state's residents were really more likely to die?

Compared at the same ages, Alaska's mortality is the higher of the two. The National Center for Health Statistics (NCHS) adjusted both states to a common age structure and put Maine at 889.4 deaths per 100,000 and Alaska at 931.0. This unit is about making such comparisons fairly.

A rate puts a count of events (deaths, new cases of a disease, hospital admissions) on a common scale: events per person per unit of time. A count says how large a problem is. A rate says how often the event occurs, so that populations of different sizes, watched for different lengths of time, can be compared. Standardization goes one step further. It removes the part of a difference between two rates that comes only from the populations' make-up by age (or sex, or another variable).

The unit covers the subject in two steps:

  1. first, the vocabulary and the statistics of a single rate, ending with a comparison of Maine and Alaska that is statistically precise and still wrong;
  2. then, the two classical adjustments that correct it, and what each one assumes.

The examples use real vital statistics throughout:

  • deaths of U.S. residents by age from CDC WONDER, the federal query system for vital statistics, for 2007, 2010, 2013 and 2019;
  • the final 2021 figures for the states, published by NCHS;
  • the Census Bureau's population estimates by single year of age.

The national data hold two surprises. Between 2007 and 2013, every one of the eleven age-specific death rates fell, while the crude death rate rose. Between 2010 and 2019, some age groups improved and others got worse, the young adults of the opioid epidemic among them. Whether mortality fell or rose then depends on which ages the summary weights most. Every number comes from a script in the site's repository, and each computed rate is checked against the figure NCHS published.

Articles

  • Rates in Public Health: what makes a number a rate, and how far can a single rate be trusted? Counts, proportions, ratios and rates, with their units. Person-time on an eight-person cohort, and the risk a rate implies. Incidence and prevalence, the steady-state relation $P/(1 - P) = I \times D$ derived and checked by simulation, and why prevalent cases have lasted longer than the average case. Crude, age-specific and cause-specific rates, and the crude rate as a population-weighted average of specific rates. The Poisson model, standard errors, normal and exact intervals with their exact coverage, and the NCHS rule of 20 deaths. Comparing two crude rates, and why the comparison of Maine and Alaska is precise and still wrong.
  • Direct and Indirect Standardization: how do we compare death rates at the same ages, and does the answer depend on how we do it? The directly standardized rate as a weighted average of specific rates, worked in full for 2007 and 2013. A proof that a population with lower rates at every age has the lower adjusted rate under every standard, while its crude rate can be higher. The 2000 U.S. standard and four others, and how the choice of standard turns the 2010 to 2019 trend from a fall of 4.2% under the 2000 standard into a rise of 4.6% when only ages under 55 are weighted. The variance of an adjusted rate and the gamma interval of Fay and Feuer, with coverage checked by simulation on small populations. Expected deaths, the standardized mortality ratio and its exact interval for four states, and when each method is the right one.

This unit builds on the previous one and borrows tools from later ones.

The next unit, Probability, sets out the rules for reasoning about uncertain events. One of them, the law of total probability, turns out to be direct standardization written in the language of probability.

References

  • Ahmad, O. B., Boschi-Pinto, C., Lopez, A. D., Murray, C. J. L., Lozano, R., and Inoue, M. (2001). Age Standardization of Rates: A New WHO Standard. GPE Discussion Paper Series No. 31. World Health Organization. record · 2 pages
  • Anderson, R. N., and Rosenberg, H. M. (1998). Age standardization of death rates: Implementation of the year 2000 standard. National Vital Statistics Reports, 47(3). National Center for Health Statistics. record · 3 pages
  • Breslow, N. E., and Day, N. E. (1987). Statistical Methods in Cancer Research, Volume II: The Design and Analysis of Cohort Studies, Chapter 2 (Rates and rate standardization). IARC Scientific Publications No. 82. International Agency for Research on Cancer. record · 3 pages
  • Fay, M. P., and Feuer, E. J. (1997). Confidence intervals for directly standardized rates: A method based on the gamma distribution. Statistics in Medicine, 16(7), 791–801. record · 2 pages
  • Klein, R. J., and Schoenborn, C. A. (2001). Age Adjustment Using the 2000 Projected U.S. Population. Healthy People Statistical Notes No. 20. National Center for Health Statistics. record · 2 pages
  • Murphy, S. L., Kochanek, K. D., Xu, J., and Arias, E. (2024). Deaths: Final data for 2021. National Vital Statistics Reports, 73(8). National Center for Health Statistics. record · 4 pages
  • Pagano, M., Gauvreau, K., and Mattie, H. (2022). Principles of Biostatistics (3rd ed.). CRC Press. The course text. record · 35 pages