I had the privilege of chatting with Bob Anderson on Friday. Anderson is the director of the Mortality Statistics Branch at the CDC’s National Center for Health Statistics, the group responsible for counting America’s dead.

The last time we had spoken was in 2009, and Anderson and his crew had just finished compiling the official tally of 2006’s fatalities. It had taken a full 30 months to gather records from all 50 states, the District of Columbia, the five island territories from Puerto Rico to Guam, and New York City—which, for odd reasons, maintained its own vital records—then plug the many holes (recoding the cause of death for some states entirely), clear up discrepancies, sort, cross-reference, tabulate the numbers in hundreds of ways, and summarize the lot into a single, cohesive report.

Back then, much of the processing of mortality data at the state level, almost unthinkably, was being done by hand. Now, with the exception of eleven of the 57 reporting jurisdictions above—the five territories and six holdout states (West Virginia, North Carolina, Mississippi, Tennessee, Rhode Island, and Connecticut)—the process has become mostly electronic. Currently, 47% of death records are sent to Anderson’s team within 10 days of the person’s passing. The entire national tally now takes about eleven months to complete.

It’s no mean feat.

There are roughly 8,000 ways to die—which is to say there are about 8,000 categories in the 10^th and latest edition of the International Classification of Diseases. And Anderson and his crew of statisticians, analysts, and nosologists on the fifth floor of a seven-floor office building in Hyattsville, Maryland, will soon publish their report detailing precisely which of those ways took the lives of 2,712,630 U.S. residents in 2015. (An interim data brief was released last December.)

The answers are not always as black-and-white as the death certificates they’re written on. The cause, in each case, is determined by an attending physician or coroner or medical examiner, who in turn often relies on medical histories provided by the families—or, say in the case of a car accident or gunshot wound, on the sheer apparentness of injury. Contrary to the impression left by television crime dramas, formal autopsies are seldom done. (Fewer than one in ten bodies these days are hauled to the lab and cut open by a medical examiner.) Nor, in the vast majority of cases, are so-called pronouncing doctors trained in forensic pathology. Mistakes are as inevitable as they are expected.

On the whole, though, they tend toward the minor—conflating the “immediate” and “underlying” causes of death, for instance, or confusing two related forms of disease. Such flaws notwithstanding, death certificates are the best documentary evidence there are when it comes to studying the health and well being of any population. It is almost impossible, for instance, to understand disease burden or do major epidemiological studies without them. No public or private record offers a more definitive account of the extent of illness and injury in the country. To health policy officials, certainly, no nationwide set of data is anywhere near as essential.

Once a physician has signed off on the cause of death, it’s the task of a funeral director to supply the rest of the information on the legal certificate—the who, when, where, how old, and more—and send the form to the local or state registrar. State workers then send a record of this assessment to Anderson’s team at the NCHS who code it with one of those 8,000 or so potential causes in the ICD—which over a century of use and revision has become the globe’s single compendium to mortality. (Death, ironically, has managed to unite the world where the commonality of life cannot: ICD’s expansive alphanumeric code is perhaps the only language shared by nearly every government on the planet.)

These are the blunt, industrial-scale mechanics of counting death in the U.S. The marvel is in how such painting-by-numbers can yield, year after year, an image of near-animate detail: a national portrait of death—and to some extent, its inverse, a portrait of life.

So what do we see right now in that self-portrait of America? A face that’s a little more weathered than it was in the previous year, and perhaps a fraction less vital, too. From 2014 to 2015, life expectancy for the U.S. population actually dipped one-tenth of a year, from 78.9 years to 78.8—its first drop in more than two decades. The age-adjusted death rate in America rose for eight of the 10 leading causes of mortality and increased 1.2% overall.

Indeed, for one of those causes—Alzheimer’s disease—the rate didn’t merely creep upward, it leaped, shooting up 15.7% year over year. This, after climbing 55% between 1999 and 2014, according to a CDC report. Those numbers are a shout of warning, or ought to be.

“The increasing rates of Alzheimer’s deaths are not only problematic because of their obvious direct health effects on persons with Alzheimer’s,” says the CDC, plainly. “The debilitating nature of Alzheimer’s” translates into mammoth financial costs that are “borne by patients and their families, and by states and counties that operate publicly funded long-term care facilities.”

More than two thirds of the $259 billion cost of caring for those with Alzheimer’s and other forms of dementia will be paid by Medicare, Medicaid, and other public sources. But “most care provided to older adults with Alzheimer’s who do not live in long-term care facilities is provided by family members or other unpaid caregivers,” the CDC points out. And the agency calculates that this care amounted to 18.2 billion hours of unpaid assistance last year. (Yes, that’s billion with a “b.”)

Anderson has seen this and other American storylines unfold in real-time in the data of death: the stark and scary rise in opioid-related fatalities, the happy decline in infant mortality, the still-unexplained rise in suicides in the country.

A soft-spoken fellow from Sugarland, Texas, Anderson fell into his role by accident. After changing his undergraduate major four times, he pursued a graduate degree in demography at Penn State—but even then he was more interested in studying statistics about marriage and family than death. But “the job market was poor for demographers,” he says, and an ad for a position at the mortality branch just happened to land in his lap. And maybe, just maybe, the job was in his blood anyway: “My grandfather was a funeral director, mortician, ambulance driver, and county coroner, in Madison County, Idaho,” he says.

When I ask him what he’s learned in 20 years of counting bodies, Anderson answers quickly: “Everybody dies,” he says.

Then he pauses for a moment and offers something else: “Death gives life perspective.”

That one stuck with me.

This essay appears in today’s edition of the Fortune Brainstorm Health Daily. Get it delivered straight to your inbox.