The study, published in the open-access journal Nursing Open, drew on de-identified aggregate mortality records from the Centers for Disease Control and Prevention’s Wide-ranging Online Data for Epidemiologic Research platform, known as CDC WONDER. The researchers defined a case as any death among a U.S. resident aged 65 or older in which both Alzheimer’s disease, coded G30 in the International Classification of Diseases tenth revision, and a malignant neoplasm, coded C00 through C97, appeared anywhere on the same certificate, whether as the underlying cause or a contributing factor. Over the full 25-year window, 143,052 such deaths were identified. Because either condition could occupy any position on the certificate, the case definition captures deaths in which both diseases were recorded, not a cohort of people known to be living with both diagnoses, and the authors are careful to stress that the resulting rates describe population-level patterns rather than individual mortality risk or survival.

To translate raw counts into comparable trends, the team calculated age-adjusted mortality rates standardized to the year 2000 U.S. standard population and modeled temporal patterns with joinpoint regression, a statistical technique that fits a series of connected straight lines on a logarithmic scale and pinpoints the years where the trend changes direction. Annual percent changes describe each segment, while average annual percent changes summarize the whole period. Across the entire study span, the age-adjusted rate fell from 13.96 to 12.05 deaths per 100,000 older adults, a statistically significant average annual decline of 0.73 percent. Yet the joinpoint analysis revealed that this gentle long-term slope conceals a striking rollercoaster: rates rose through the mid-2000s, dropped sharply between roughly 2006 and 2014, climbed again from 2014 to 2021, and then turned downward in the final two years, a dip the authors treat cautiously because it overlaps with the COVID-19 pandemic.

That recurring arc, an early rise, a steep middle-period decline, and a rebound after 2014, appeared in nearly every subgroup examined, from men and women to every age band and census region. Among women, who accounted for 79,117 deaths, the rate rose 1.69 percent annually from 1999 to 2006, plunged 4.33 percent per year through 2014, then surged 3.11 percent annually through 2021. Men, with 63,935 deaths, followed an almost identical choreography, and only men showed a statistically significant full-period decline, at 1.17 percent per year, compared with a non-significant 0.55 percent among women. The consistency of the pattern across sexes and ages suggests that forces operating at the population level, such as shifts in diagnostic practice, death-certificate reporting habits, and the composition of the aging population, may have shaped the recorded rates as much as any change in the underlying biology.

Age left the deepest imprint on the data. Adults aged 85 and older accounted for 82,538 of the deaths, and their crude mortality rate, roughly 57 to 59 per 100,000, dwarfed the 12 to 16 per 100,000 seen among those aged 75 to 84 and the mere 1.5 to 2 per 100,000 among those aged 65 to 74. Only the 75-to-84 group showed a statistically significant long-term decline, at 1.26 percent per year. The authors note that frailty, multimorbidity, and diminished physiological reserve are plausible contributors to the extreme rates in the oldest group, and that biological pathways involving inflammation, DNA repair, and mitochondrial function have been investigated in relation to both neurodegeneration and cancer, but they emphasize that their ecological design cannot measure or confirm any of these mechanisms.

Geography told an equally layered story. The South recorded the most deaths, 48,389, followed by the West with 36,533, the Midwest with 35,260, and the Northeast with 22,870. Yet the Northeast posted the steepest significant decline in its age-adjusted rate, falling from 12.45 to 8.64 per 100,000, an average annual drop of 1.46 percent, while the Midwest also declined significantly. The South’s rate fell as well, though not significantly, and the West bucked the national direction entirely, its rate inching upward from 13.94 to 14.70 per 100,000 with a non-significant average annual change of 0.40 percent. At the state level, cumulative death counts were highest in California, Texas, Ohio, Florida, and Pennsylvania, while point estimates of trend were most negative in New Hampshire, Maryland, Illinois, and Massachusetts and most positive in Oklahoma and Oregon, findings the investigators label exploratory because no adjustment for multiple comparisons was made.

Race and ethnicity produced some of the study’s most consequential contrasts. Non-Hispanic White individuals accounted for 122,214 deaths, and their rate declined only modestly and non-significantly, from 14.21 to 12.85 per 100,000. Non-Hispanic Black individuals, who began the period with the highest rate at 16.35 per 100,000, experienced a significant decline of 1.69 percent per year, converging numerically with the White rate by the study’s end. Hispanic individuals moved in the opposite direction, with a significant average annual increase of 2.58 percent, from a very low baseline of 5.93 per 100,000 in 1999, when only 75 such deaths were recorded, to 9.77 per 100,000 in 2023. The authors caution that the enormous percentage growth in death counts among Hispanic and non-Hispanic Other populations partly reflects those tiny baselines, and that a change from bridged-race to single-race classification in 2021 complicates comparisons, though an overlap analysis for 2018 through 2020 found differences below 1.3 percent for the largest groups.

Urbanization added a final geographic dimension. Restricted to 1999 through 2020 by the available data, the analysis found that nonmetropolitan counties carried consistently higher rates than metropolitan ones, 15.52 versus 13.55 per 100,000 at the start and 14.77 versus 12.70 at the end, though neither area showed a statistically significant full-period trend and no formal between-group comparison was performed. A sensitivity analysis excluding the pandemic years confirmed the overall picture: the pre-pandemic decline of 0.78 percent per year closely matched the full-period estimate, and the significant rise among Hispanic populations persisted, slightly attenuated. The authors interpret these overlapping-period comparisons as consistency checks, not as measurements of COVID-19’s causal impact.

The study’s limitations are substantial and candidly acknowledged. Death certificates are known to underreport Alzheimer’s disease, particularly among older adults burdened with multiple competing conditions, so the true frequency of dual involvement is almost certainly higher than the recorded rates suggest. Without clinical data on cognition, cancer stage, treatment, or comorbidity, the analysis cannot distinguish changes in disease occurrence from changes in certification practice, and it cannot estimate survival among people living with both diagnoses. Socioeconomic conditions, healthcare access, and environmental exposures were not measured, making explanations for the demographic and geographic patterns hypotheses rather than conclusions. What the study does establish is a clear signal for clinicians and health systems: as populations age, a growing number of older adults will face cognitive decline and cancer simultaneously, and coordinated geriatric, oncological, and neurological care, with attention to cognitive assessment, treatment burden, medication management, caregiver support, and goals of care, will become increasingly essential. The authors argue that linking mortality records with individual-level clinical and social data is the necessary next step, so that the patterns now visible on death certificates can be translated into better care for the vulnerable population they represent.

Subject of Research: Population-level mortality trends for deaths with both Alzheimer's disease and cancer recorded on U.S. death certificates among adults aged 65 and older, 1999–2023

Article Title: Trends in Population‐Level Mortality Rates for Deaths With Both Alzheimer's Disease and Cancer Recorded on Death Certificates Among U.S. Adults Aged ≥ 65 Years, 1999–2023

Article References: Zhang, X., Xiong, Y.-J., Lv, T., & Chen, Y. (2026). Trends in Population‐Level Mortality Rates for Deaths With Both Alzheimer's Disease and Cancer Recorded on Death Certificates Among U.S. Adults Aged ≥ 65 Years, 1999–2023. Nursing Open, 13(10), Article e70897. https://doi.org/10.1002/nop2.70897

Image Credits: AI Generated

DOI: 10.1002/nop2.70897

Keywords: Alzheimer's disease, cancer, mortality trends, death certificates, CDC WONDER, older adults, joinpoint regression, health disparities, multimorbidity, epidemiology, geriatrics, United States

The study’s foundation is a simple but powerful epidemiological logic. If SARS-CoV-2 were still exerting a serious toll on the population, the age groups most vulnerable to the virus, principally those aged 60 and older, should show mortality rates above their pre-pandemic baselines. The researchers therefore calculated mortality rate ratios for every combination of sex and ten-year age band, dividing the 2024 death rate by the average rate across 2017 to 2019. A ratio above 1.00 would signal excess mortality; a ratio below 1.00 would suggest that deaths have fallen relative to before the pandemic. Confidence intervals were derived using the delta method on the log scale, allowing the team to distinguish signal from statistical noise.

The results were unambiguous for older Austrians. Despite 1,212 officially reported COVID-19 deaths in 2024, all-cause mortality rates in every stratum aged 40 and over were equal to or lower than before the pandemic, and in the 60-plus groups, the very populations at risk from the virus, rates were equal or lower in 2024 compared with 2019. In many of these strata the ratios fell well below 1.00. The team stress-tested this finding against several counterfactuals. An optimistic sensitivity analysis assumed that mortality would have continued improving after 2019 at the pace seen between 2012 and 2014, yielding an expected ratio of 0.929; five of the twelve age-sex strata aged 40 and over beat even that demanding benchmark. A trend-of-trends model trained on Austrian data from 2003 to 2019 produced an expected ratio of 0.976, and ten of twelve strata performed better than that. A Lee-Carter mortality forecasting model, fitted to age- and sex-specific rates from 2003 to 2019 and projected to 2024 via ARIMA with bootstrapped confidence intervals, showed observed 2024 life expectancy to be numerically higher than predicted in every age-sex stratum, significantly so for females across all age groups.

The picture for young Austrians was markedly different, and more troubling. Mortality rate ratios were elevated in the 10-to-19, 20-to-29 and 30-to-39 age bands, significantly so among men aged 30 to 39. Crucially, these increases cannot be attributed to COVID-19 itself: only a single individual across these young strata was recorded as having died of the disease in 2024. When the researchers decomposed the excess by cause, they found moderate increases spread across multiple high-level categories, with consistent rises in mental and behavioural disorders, external causes and diseases of the nervous system, though only external causes in the 10-to-19 group reached statistical significance. The authors suggest these patterns may reflect adverse consequences of the pandemic and the pandemic response, noting that escalating violence, alcohol problems, drug overdoses and suicides among young people have been documented in several countries and may constitute a lasting legacy that requires tracking and intervention.

The second half of the study tackles a subtler problem: how to count a COVID-19 death. Death certificates record an underlying cause of death, the condition that initiates the chain leading to death, along with other contributing conditions. In practice, anyone who dies with a positive SARS-CoV-2 test may be coded as a COVID-19 death even when other conditions are more relevant, a distinction that matters enormously in elderly and chronically ill populations. To address this, the team applied a weighted multiple-cause methodology to the full text of Austrian death certificates. In their main analysis, the underlying cause received a weight of 50 percent and all other mentioned causes shared the remaining 50 percent; when no other cause was listed, the underlying cause carried full weight. Ill-defined and duplicate causes were excluded. From more than 538,000 death certificates spanning 2019 to 2024, containing over 2.5 million non-duplicate ICD code mentions, the researchers constructed age-standardised rates for both unweighted underlying-cause counts and weighted counts.

The comparison between the two measures is the study’s most technically revealing result. If a cause is genuinely driving deaths, it should appear disproportionately as the underlying cause, and its weighted rate should approach its unweighted rate. For COVID-19 in the 60-plus age groups, the ratio of weighted to unweighted rates sat between 0.51 and 0.58, both in 2020 and in 2024, meaning the virus was far more likely to be listed as the underlying cause than other conditions were, and that this preference barely changed even as absolute death counts collapsed. Sensitivity analyses using alternative weighting schemes, equal weighting of all causes or double weight for the underlying cause, produced even lower weighted estimates, for example 4.15 versus a main-analysis 7.12 for 2024, and more stringent data-cleaning criteria left the COVID-19 findings intact. Taken together, the authors argue, these patterns point to systematic overcounting of COVID-19 deaths, potentially throughout the entire pandemic rather than only in its aftermath.

Supporting evidence comes from how COVID-19 behaved within the broader cause-of-death landscape. Respiratory diseases traditionally show a weighted-to-unweighted ratio well above 1.00, around 1.47 in 2019, because they are typically recorded as contributing rather than underlying conditions. During the pandemic years 2020 to 2023, once COVID-19 was stripped out of the respiratory category, other respiratory diseases fell markedly, and the category’s ratio rose to 1.64, as if COVID-19 had absorbed deaths that would otherwise have been attributed to pneumonia and related conditions. The causes most frequently co-mentioned with COVID-19 in 2024 were pneumonia, residual infections, hypertensive disease, ischaemic heart disease and renal failure, and even in 2024 COVID-19 was still listed as the underlying cause in 68 to 85 percent of those co-occurrences, down from 88 to 98 percent in 2020. Meanwhile, the apparent decline in ischaemic heart disease between 2019 and 2024 was, in the authors’ view, too large to reflect genuine improvement and may partly represent recoding of deaths with multiple comorbidities.

The Austrian findings align with a growing international literature based on clinical record audits. In Greece, only 64.9 percent of hospital deaths listed as COVID-19 were judged attributable or related to the virus upon audit. In Sweden, auditors found that 24 percent of certificate-listed COVID-19 deaths had absolutely no relationship to the infection, and that in most of the remainder COVID-19 was contributing rather than underlying; death certificates counted 799 underlying-cause COVID-19 deaths where clinical audit supported only 213. In Ireland, clinicians considered COVID-19 the primary cause in only 72.7 percent of audited deaths during the winter of 2021 to 2022, and Danish surveillance documented increasing overcounting with the Omicron variants. Notably, the feared undercounting in lower-income settings appears smaller than assumed: a Colombian audit found only 6 percent undercounting in 2021, although an Iranian audit of 339 certificates found major errors in 58 percent. The overall pattern suggests overcounting, not undercounting, has been the dominant error in well-resourced countries.

The study has limitations the authors acknowledge candidly. Without comorbidity data, the results cannot be generalised to specific high-risk populations such as immunocompromised patients or nursing home residents, among whom COVID-19 mortality may remain meaningful. Death certificate reporting itself is imperfect, with empirical studies suggesting major errors exceed 50 percent even in pre-pandemic years, and Austria’s automated IRIS coding algorithms may have favoured selecting COVID-19 as the underlying cause. There is also genuine uncertainty about the right counterfactual: whether mortality should have continued declining after 2019 at all, given that life expectancy gains in Austria and other high-income countries had already slowed dramatically in the decade before the pandemic, adding just 0.41 years between 2014 and 2019 compared with more than a year in each of the three preceding five-year periods. Some of the lower mortality among the very old in 2024 may reflect mortality displacement, the premature deaths during the pandemic of people who would otherwise have died a few years later.

Even so, the implications for policy are concrete. Accurate post-pandemic burden estimates directly shape vaccination recommendations, which currently diverge widely across Europe, from yearly boosters offered from age 12 in Austria to age 65 and older in Denmark. The authors’ earlier work estimated that, assuming correct COVID-19 death counts, the only Austrian group with a mortality rate implying a number needed to vaccinate below 1,000, residents of nursing homes aged 85 and over, had average life expectancies below one year. With an estimated 2.8 million SARS-CoV-2 infections in Austria in 2024, the reported 1,212 deaths already imply an infection fatality rate of just 0.043 percent, and the weighted analysis suggests even that figure is too high. The study’s most urgent message, however, may be the one about the young: a post-pandemic rise in mortality driven by mental health, external and neurological causes demands confirmation, explanation and intervention, long after the virus itself has faded from the mortality statistics.

Subject of Research: Post-pandemic all-cause mortality patterns and estimation of the residual COVID-19 death burden using weighted multiple-cause-of-death analysis in Austria

Article Title: Post-pandemic mortality patterns and COVID-19 burden considering multiple death causes

Article References: Riedmann, U., Levitt, M., Pilz, S., & Ioannidis, J. P. (2026). Post-pandemic mortality patterns and COVID-19 burden considering multiple death causes. European Journal of Epidemiology. https://doi.org/10.1007/s10654-026-01462-7

Image Credits: AI Generated

DOI: 10.1007/s10654-026-01462-7

Keywords: COVID-19, mortality, Austria, death certificates, multiple causes of death, excess mortality, epidemiology, vaccination policy, life expectancy, infection fatality rate, public health, youth mortality