Scholarlink

Background <p>Alzheimer’s disease (AD) mortality has increased in the United States, with persistent sociodemographic and geographic disparities. We paired national surveillance with segmented trend modeling and conditional long-horizon projections.</p> Methods <p>Using CDC WONDER (NVSS) multiple-cause-of-death data, we analyzed U.S. deaths from 1999 to 2023 with ICD-10 code G30 listed anywhere on the death certificate. The primary outcome was the age-adjusted mortality rate (AAMR) per 100,000 (2000 U.S. standard), stratified by sex, race/ethnicity, state, and place of death. Trend inflection points were identified using joinpoint log-linear regression (WBIC selection), estimating annual percent change (APC). Projections for 2024–2043 were presented primarily as scenario-based trajectories anchored to 2023 (flat, continuation of pre-2020 growth, continuation of post-2020 decline). A univariate LSTM forecast with Monte Carlo dropout intervals was reported as a sensitivity analysis and benchmarked via back-testing (train 1999–2010; test 2011–2023) against classical models.</p> Results <p>From 1999 to 2023, 2,970,206 AD-associated deaths were recorded; AAMR increased from 45.01 to 52.72 per 100,000. Joinpoint analysis identified inflections in 2005, 2012, and 2020, with APCs of + 3.50%/year (1999–2005), − 2.46%/year (2005–2012), + 3.61%/year (2012–2020), and − 5.62%/year (2020–2023). Women had higher rates than men, and White individuals had the highest race-specific AAMR. Most deaths occurred in nursing homes/long-term care (55.68%). State crude rates ranged 22.70–103.39 per 100,000. Scenario projections for 2043 ranged from 12.51 (continued post-2020 decline) to 132.64 (continued pre-2020 growth), while the LSTM projected modest fluctuations in the low-to-mid 50s (e.g., 2043: 53.14; 95% PI 43.91–62.54).</p> Conclusion <p>U.S. Alzheimer’s mortality shows a multi-phase trend with persistent sex, racial/ethnic, and geographic disparities and most deaths occurring in long-term care. The post-2020 decline should be interpreted cautiously due to competing mortality and certification/coding changes during COVID-19. Long-horizon projections are best viewed as conditional scenarios shaped by aging, diagnosis, and future treatments, supporting equity-focused prevention and strengthened long-term care and home support.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

National trends and disparities in Alzheimer disease’s mortality (1999–2023) with scenario projections and a data-driven LSTM sensitivity analysis to 2043

  • 期刊论文

  • 发表日期：

    2026年4月24日

查看全文
  • Mohammad Bdair,
  • Anwar Zahran,
  • Hammam Jallad,
  • Ibrahim Barham,
  • Ather Qubbaj,
  • Mohammad AbuRabi’e,
  • Bara Jarrar,
  • Joud Aburajab,
  • Zaina Alsafareeni,
  • Amal Mansor,
  • Shahd Ghannam,
  • Adnan Daor,
  • Ahmad Daralbarmeel,
  • Ahmad R. Awwad,
  • Mohammad Khader,
  • Abdelrahman Awawdeh

摘要

Background

Alzheimer’s disease (AD) mortality has increased in the United States, with persistent sociodemographic and geographic disparities. We paired national surveillance with segmented trend modeling and conditional long-horizon projections.

Methods

Using CDC WONDER (NVSS) multiple-cause-of-death data, we analyzed U.S. deaths from 1999 to 2023 with ICD-10 code G30 listed anywhere on the death certificate. The primary outcome was the age-adjusted mortality rate (AAMR) per 100,000 (2000 U.S. standard), stratified by sex, race/ethnicity, state, and place of death. Trend inflection points were identified using joinpoint log-linear regression (WBIC selection), estimating annual percent change (APC). Projections for 2024–2043 were presented primarily as scenario-based trajectories anchored to 2023 (flat, continuation of pre-2020 growth, continuation of post-2020 decline). A univariate LSTM forecast with Monte Carlo dropout intervals was reported as a sensitivity analysis and benchmarked via back-testing (train 1999–2010; test 2011–2023) against classical models.

Results

From 1999 to 2023, 2,970,206 AD-associated deaths were recorded; AAMR increased from 45.01 to 52.72 per 100,000. Joinpoint analysis identified inflections in 2005, 2012, and 2020, with APCs of + 3.50%/year (1999–2005), − 2.46%/year (2005–2012), + 3.61%/year (2012–2020), and − 5.62%/year (2020–2023). Women had higher rates than men, and White individuals had the highest race-specific AAMR. Most deaths occurred in nursing homes/long-term care (55.68%). State crude rates ranged 22.70–103.39 per 100,000. Scenario projections for 2043 ranged from 12.51 (continued post-2020 decline) to 132.64 (continued pre-2020 growth), while the LSTM projected modest fluctuations in the low-to-mid 50s (e.g., 2043: 53.14; 95% PI 43.91–62.54).

Conclusion

U.S. Alzheimer’s mortality shows a multi-phase trend with persistent sex, racial/ethnic, and geographic disparities and most deaths occurring in long-term care. The post-2020 decline should be interpreted cautiously due to competing mortality and certification/coding changes during COVID-19. Long-horizon projections are best viewed as conditional scenarios shaped by aging, diagnosis, and future treatments, supporting equity-focused prevention and strengthened long-term care and home support.