Syphilis-related mortality in the United States, 2021–2025: trends, demographic disparities, and clinical implications

Abstract

Background

Syphilis-related mortality in the United States remains incompletely characterized in the post-pandemic period. This study aimed to quantify trends, demographic disparities, and geographic distribution of syphilis-related mortality from 2021 to 2025 using multiple causes of death data, with attention to their clinical implications.

Methods

A retrospective cross-sectional analysis was conducted using the CDC WONDER Multiple Cause-of-Death database. Decedents with ICD-10 codes A50–A53, listed as contributing causes, were included. Age-adjusted mortality rates (AAMRs) were calculated per 100,000 population using the 2000 US Standard Population. Poisson regression was used to estimate the annual percent change (APC) for temporal trends and rate ratios (RRs) for demographic comparisons. Eight pre-specified sensitivity analyses were performed, including restriction to the underlying cause of death only and assessment of HIV and COVID-19 co-occurrence.

Results

A total of 1,148 syphilis-related deaths were identified, yielding an overall AAMR of 0.06 per 100,000 (95% CI: 0.06–0.07). No significant temporal trend was identified (APC: −1.24%, p = 0.550). Late-stage syphilis was recorded in 51.4% of deaths. The mortality rate among men was 2.5 times that of women (RR: 2.50; 95% CI: 2.20–2.83). The mortality among non-Hispanic Black individuals was 4.8 times that of non-Hispanic White individuals (RR: 4.81; 95% CI: 4.20–5.50). The South accounted for 50.3% of all deaths. HIV was co-listed in 27.5% of deaths, with a borderline increasing trend (APC: +8.25%; p = 0.047). More than half of the deaths occurred outside inpatient hospital settings.

Conclusions

Syphilis-related mortality remained stable but was concentrated in identifiable high-risk groups and geographical areas. The predominance of late-stage disease, high HIV co-occurrence, and substantial proportion of deaths in community and long-term care settings may reflect gaps in early detection and treatment retention, although death certificate data cannot directly establish these mechanisms. These findings have implications for clinical practice, public health screening programs, and targeted interventions in high-burden populations.

Clinical trial number

Not Applicable.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12879-026-13981-5.

Introduction

Syphilis has re-emerged as a major public health concern in the United States, with steadily increasing incidence over the past two decades. National surveillance data indicate that reported cases have risen steadily since 2000, and in 2022, the total syphilis cases reached their highest level since the 1950s [1, 2]. In 2023, primary and secondary syphilis cases declined by 10%, the first substantial decline in more than two decades, whereas the total syphilis cases rose by only 1% [2]. The burden is unevenly distributed across the globe. In 2023, the highest primary and secondary syphilis rates occurred among American Indian or Alaska Native persons, at several times the national rate, and among Black or African American persons, while the longer-term rise has increasingly involved women and heterosexual men, paralleling a sharp increase in congenital syphilis [2]. These disparities are widely attributed to structural and social determinants, including differential access to sexual health care, rather than to differences in sexual behavior alone [3–5]. Despite well-documented increases in incidence, contemporary patterns of syphilis-related mortality remain poorly understood.

Clinically, syphilis-related mortality is largely driven by late-stage disease, principally neurosyphilis and cardiovascular syphilis [4]. Genitourinary involvement is uncommon; neurogenic bladder from tabes dorsalis is the principal urological manifestation, and gummatous disease of the genitourinary tract is rare. Although most syphilis cases in the United States are diagnosed and managed in primary care, infectious diseases, sexual health, and obstetric settings rather than by urologists, late neurosyphilis can present with lower urinary tract dysfunction, and clinicians across specialties may encounter these patients. In addition, the HIV-syphilis syndemic contributes to accelerated disease progression and adverse outcomes, reinforcing the importance of integrated clinical management [6]. Therefore, understanding mortality patterns is critical for clinical prioritization and public health responses.

Prior studies on syphilis mortality have important limitations. Many have relied on underlying cause of death (UCOD) definitions, which underestimate the true burden by excluding deaths in which syphilis is a contributing factor. Multiple cause of death (MCOD) approaches provide a more comprehensive assessment, particularly given that syphilis frequently functions as a contributing cause of death rather than the primary cause. Furthermore, contemporary analyses incorporating post-2020 data are limited, despite the known disruptions in STI services during the COVID-19 pandemic [7].

This study aimed to characterize syphilis-related mortality in the United States from 2021 to 2025 using MCOD data stratified by sex, age, race/ethnicity, geography, and place of death, and to estimate temporal trends using Poisson regression analysis. The secondary aim was to assess the robustness of the findings across the alternative case definitions.

Materials and methods

Study design and data source

This retrospective cross-sectional study analyzed national mortality data from the CDC Wide-ranging Online Data for Epidemiologic Research (WONDER) Multiple Cause of Death (MCOD) database [8]. Data were extracted for 2021–2024 using final mortality files and supplemented by 2025 provisional data [9], yielding a five-year study window. The MCOD file records all conditions listed on the U.S. standard death certificate, including the underlying cause of death (UCOD) and up to 20 contributing causes, enabling the identification of deaths in which syphilis contributed to, but was not designated as the primary cause [8]. This study was exempted from institutional review board approval because it used de-identified, publicly available aggregate data.

Case identification

Decedents were included if any of the following ICD-10 codes appeared as either the underlying or contributing cause of death: A50 (congenital syphilis), A51 (early syphilis), A52 (late syphilis, including neurosyphilis and cardiovascular syphilis), or A53 (other and unspecified syphilis), constituting the any-mention MCOD case definition. This any-mention definition identifies death certificates on which syphilis was recorded as causing or contributing to death; it does not establish that syphilis caused or directly contributed to death, and some records may reflect previously treated or incidentally documented infections. Therefore, we refer to syphilis-related deaths (deaths with syphilis listed) rather than deaths caused by syphilis. This approach is consistent with prior U.S. syphilis mortality surveillance analyses using CDC WONDER MCOD data and ICD-10 codes A50–A53 as the underlying or contributing causes of death [10]. Cases were classified by diagnostic stage to characterize the clinical profile of the mortality. HIV co-occurrence was identified by the additional presence of ICD-10 codes B20-B24 or R75 on the same death certificate. COVID-19 co-occurrence was identified using the ICD-10 code U07.1. Cells with fewer than 10 deaths were suppressed according to the CDC WONDER data-use restrictions [11].

Outcome measures

The primary outcome was syphilis-related mortality, defined by the any-mention MCOD criteria. Age-adjusted mortality rates (AAMRs) were calculated per 100,000 population using the 2000 United States Standard Population, enabling valid comparisons across subgroups and time periods with different age structures [12]. Crude mortality rates were reported alongside the AAMRs for descriptive context.

Statistical analysis

Two separate Poisson regression models were used in this study. To estimate demographic disparities, a cross-sectional Poisson model was fitted with pooled death counts (2021–2025) as the outcome, ln(population) as an offset, and demographic groups as predictors, yielding rate ratios (RRs) with 95% confidence intervals (CIs). Both Poisson models used the population as an offset and therefore estimated rate ratios from crude (non-age-standardized) rates. Age-adjusted mortality rates (AAMRs) were reported separately as descriptive measures and were not regression outcomes; therefore, RRs and AAMRs answer different questions and are not directly interchangeable. NH White individuals served as the reference for race/ethnicity comparisons, and females served as the reference for sex comparisons. To estimate temporal trends, a separate longitudinal Poisson model was fitted for each stratum with the annual death count as the outcome, ln(annual population) as the offset, and calendar year (centered on the stratum mean to reduce collinearity) as a continuous predictor. The annual percent change (APC) was derived as (e^β − 1) × 100, where β is the year coefficient, and 95% CIs were computed on the log scale and back-transformed. A five-year observation window precludes reliable joinpoint segmentation because NCI/NCHS guidance recommends caution when estimating joinpoints in short time series with limited data around candidate segments and endpoints [13]. Poisson regression is an appropriate alternative for short time series with count outcomes. Overdispersion was assessed using the Pearson chi-squared statistic divided by the residual degrees of freedom; strata with dispersion > 2 (age 0–24 years and congenital syphilis) are flagged in the results. Statistical significance was set at p < 0.05 (two-tailed). All analyses were conducted in Python (version 3.13) using the Statsmodels library (version 0.14.6) [14]. To place the 2021–2025 window in a longer context, we additionally analyzed any-mention syphilis mortality over 1999–2025 (1999–2020 from final files and 2021–2025 from provisional data, extracted June 2026) using joinpoint regression (National Cancer Institute Joinpoint, version 5.4.0.0). Because CDC WONDER reports the age-adjusted rate and its standard error to two decimal places, at these low rates, the standard error rounds to zero, and weighted joinpoint models are not estimable; therefore, we used a homoscedastic error model and confirmed the trend with an independent Poisson regression on annual death counts (offset = ln[population]), which is insensitive to rate rounding.

Stratification

Mortality was stratified by sex (male, female), age group (0–24, 25–44, 45–64, and ≥ 65 years), race, and ethnicity using the CDC WONDER Non-Hispanic Single Race 6 classification (NH White, NH Black or African American, NH American Indian or Alaska Native [AI/AN], and NH Asian) and Hispanic or Latino ethnicity from a separate query; these classifications were not mutually exclusive. Mortality was stratified by census region (Northeast, Midwest, South, and West), state of residence, and place of death. Age-stratified analyses reported crude rates; age adjustment within age strata was methodologically inappropriate and not performed.

Sensitivity analyses

Eight sensitivity analyses were conducted. (1) Restriction to UCOD only, quantifying the proportion of deaths in which syphilis was the primary cause rather than a contributing factor. (2) Restriction to A53.0 and A53.9 only, assessing the contribution of diagnostically unspecified coding. (3–5) Restriction to individual ICD-10 stage categories: congenital syphilis (A50), early stage syphilis (A51), and late-stage syphilis (A52). (6) Co-occurrence of COVID-19 (U07.1) and syphilis to evaluate pandemic-related diagnostic displacement. (7) Co-occurrence of HIV (B20–B24) and syphilis, quantifying the HIV–syphilis co-mortality burden. (8) Restriction to 2021–2024 final data only, excluding provisional 2025 records. The complete results are presented in Supplementary Table 7.

Results

Overall mortality

Between 2021 and 2025, 1,148 syphilis-related deaths were identified using the MCOD criteria, yielding an overall AAMR of 0.06 per 100,000 population (95% CI: 0.06–0.07). The annual death count ranged from 209 in 2023 to 243 in 2022. Poisson regression identified no statistically significant temporal trends over the study period (APC: −1.24%; 95% CI: −5.20% to 2.88%; p = 0.550). Late syphilis (A52) was the most commonly listed stage, present in 51.4% of deaths (n = 590), followed by unspecified syphilis (A53.0/A53.9, n = 453, 39.5%), congenital syphilis (A50, n = 95, 8.3%), and early syphilis (A51, n = 12, 1.0%). When syphilis was the underlying cause of death, the stage distribution was more concentrated in advanced and congenital disease: late syphilis (A52) predominated (approximately 70%), followed by congenital syphilis (A50, approximately 17%) and unspecified syphilis (A53, approximately 12%), with early syphilis (A51) accounting for fewer than 10 deaths (< 2%). Thus, unspecified coding made up a larger share of contributing causes than underlying cause deaths. The annual counts and AAMRs are presented in Supplementary Tables 1 and Fig. 1, respectively. The distribution of the clinical stages is shown in Table 1.

Fig. 1.

Annual deaths and age-adjusted mortality rates for syphilis-related mortality in the United States, 2021–2025

Table 1.

Syphilis-related mortality by clinical stage, United States, 2021–2025

Clinical stage	Deaths (n)	% of total
Early syphilis (A51)	12	1.0%
Congenital syphilis (A50)	95	8.3%
Late syphilis (A52)	590	51.4%
Unspecified syphilis (A53.0/A53.9)	453	39.5%
Total	1,148	100%

Notes: ICD-10 codes: A50 = congenital syphilis; A51 = early syphilis; A52 = late syphilis (including neurosyphilis and cardiovascular syphilis); and A53 = other and unspecified syphilis. Deaths were identified using the any-mention multiple causes of death criterion. The stages reflect any-mention coding and are not mutually exclusive. Two decedents had qualifying codes in more than one stage category; therefore, stage-specific counts sum to 1,150, while the total reflects 1,148 unique decedents. Percentages were calculated from the 1,148 total

Sex

Males accounted for 71.0% of deaths (n = 815; AAMR 0.10 per 100,000, 95% CI: 0.09–0.11) compared with females (n = 333; AAMR 0.03, 95% CI: 0.03–0.03). Poisson regression indicated a significantly higher mortality risk in men (RR: 2.50; 95% CI: 2.20–2.83; p < 0.001). Temporal trends were stable in both sexes: male APC − 0.36% (95% CI: −5.08% to 4.60%; p = 0.886); female APC − 3.32% (95% CI: −10.39% to 4.31%; p = 0.384). Annual data by sex are presented in Supplementary Tables 1 and are summarized in Table 2; Fig. 2, respectively.

Table 2.

Syphilis-related mortality by sex, United States, 2021–2025

Sex	Deaths (n)	% of total	AAMR (95% CI)	RR (95% CI)	APC (95% CI)	p-value (APC)
Female	333	29.0%	0.03 (0.03–0.03)	1.00 (referent)	—	—
Male	815	71.0%	0.10 (0.09–0.11)	2.50 (2.20–2.83)*	−0.36% (− 5.08% to 4.60%)	0.886
Total	1,148	100%	0.06 (0.06–0.07)	—	—	—

Note: AAMR = age-adjusted mortality rate per 100,000 population (2000 U.S. standard population). RR, rate ratio; Poisson regression; female, reference. APC = annual percent change; Poisson regression. * p < 0.001 for RR. The annual data are presented in Supplementary Table 1

Fig. 2.

Sex-specific syphilis-related mortality in the United States, 2021–2025

Age

Mortality was concentrated in older adults. Those aged ≥ 65 years accounted for 45.5% of deaths (n = 522; AAMR 0.16 per 100,000, 95% CI: 0.14–0.18), and those aged 45–64 years accounted for 31.3% (n = 359; AAMR 0.10, 95% CI: 0.09–0.11). A secondary concentration was observed in the 0–24 age group (n = 90, 7.8%; AAMR 0.02, 95% CI: 0.01–0.02), largely due to congenital syphilis. No statistically significant temporal trends were identified in any age group (all p > 0.05). The 0–24 age group showed overdispersion (Pearson dispersion 4.21), consistent with high year-to-year variability in small counts; this APC estimate should be interpreted with caution. The annual data are presented in Supplementary Tables 2 and are summarized in Table 3; Fig. 3.

Table 3.

Syphilis-related mortality by age group, United States, 2021–2025

Age group (years)	Deaths (n)	% of total	Crude rate per 100,000† (95% CI)	APC (95% CI)	p-value
0–24	90	7.8%	0.02 (0.01–0.02)	−2.88% (− 16.08% to 12.39%)‡	0.695
25–44	177	15.4%	0.05 (0.05–0.06)	+ 3.09% (− 7.12% to 14.41%)	0.568
45–64	359	31.3%	0.10 (0.09–0.11)	−2.31% (− 9.19% to 5.10%)	0.531
≥ 65	522	45.5%	0.16 (0.14–0.18)	−2.72% (− 8.46% to 3.39%)	0.375
Total	1,148	100%	0.06 (0.06–0.07)	—	—

Notes: † Rates within age strata are crude (non-age-standardized) rates per 100,000; age adjustment within strata is methodologically inappropriate and was not performed. The Total row shows the overall age-adjusted rate (0.06 per 100,000). APC = annual percent change; Poisson regression. ‡ Pearson dispersion = 4.21 in the 0–24 group; APC estimate was unstable and should be interpreted with caution. The annual data are listed in Supplementary Table 2

Fig. 3.

Syphilis-related mortality by age group, United States, 2021–2025

Race and ethnicity

NH AI/AN individuals had the highest AAMR at 0.30 per 100,000 (95% CI: 0.21–0.42; RR: 7.00; 95% CI: 4.98–9.84; p < 0.001 vs. NH White), although annual counts were suppressed in four of the five study years. This estimate should be interpreted with caution, particularly given the known racial misclassification of American Indian or Alaska Native decedents on death certificates [15]. NH Black or African American individuals accounted for the largest share of deaths (n = 430; AAMR 0.20, 95% CI: 0.18–0.22; RR: 4.81; 95% CI: 4.20–5.50; p < 0.001). Hispanic or Latino individuals (n = 227; AAMR 0.09, 95% CI: 0.08–0.10) had an RR of 1.70 (95% CI: 1.44–2.00; p < 0.001). NH Asian individuals (n = 42; AAMR 0.03, 95% CI: 0.02–0.04) did not differ significantly from the NH White reference (RR: 0.92; 95% CI: 0.67–1.27; p = 0.614). Temporal trends were stable across all groups, with sufficient, non-suppressed annual data. Annual data are presented in Supplementary Tables 3 and summarized in Tables 4 and 5; Fig. 4.

Table 4.

Syphilis-related mortality by race/ethnicity, United States, 2021–2025

Race/ethnicity	Deaths (n)	AAMR (95% CI)	RR (95% CI)	p-value
NH White (referent)	413	0.03 (0.02–0.03)	1.00 (referent)	—
NH AI/AN†	36	0.30 (0.21–0.42)	7.00 (4.98–9.84)	< 0.001
NH Black or African American	430	0.20 (0.18–0.22)	4.81 (4.20–5.50)	< 0.001
Hispanic or Latino‡	227	0.09 (0.08–0.10)	1.70 (1.44–2.00)	< 0.001
NH Asian	42	0.03 (0.02–0.04)	0.92 (0.67–1.27)	0.614
Total	1,148	0.06 (0.06–0.07)	—	—

Note: AAMR = age-adjusted mortality rate per 100,000 (2000 U.S. standard population). RR: rate ratio; Poisson regression; NH White: reference. NH = non-Hispanic. † NH AI/AN AAMR and RR should be interpreted with caution: annual counts suppressed in four of five years; estimate based on 36 total deaths. ‡ Hispanic estimates derived from a separate CDC WONDER ethnicity query; not mutually exclusive of the NH race groups. The annual data are listed in Supplementary Table 3

Table 5.

Annual percent change in syphilis-related mortality by race/ethnicity, United States, 2021–2025

Race/ethnicity	APC (95% CI)	p-value
NH White	−2.57% (− 8.99% to 4.31%)	0.455
NH Black or African American	−1.30% (− 7.68% to 5.53%)	0.702
Hispanic or Latino	+ 3.50% (− 5.62% to 13.51%)	0.465

Note: APC = annual percent change; Poisson regression. NH AI/AN and NH Asian excluded: annual counts suppressed in ≥ 4 of 5 years, precluding reliable trend estimation. All APCs non-significant (p > 0.05). The annual data are presented in Supplementary Table 3

Fig. 4.

Syphilis-related mortality by race and ethnicity in the United States, 2021–2025

Geography

The South (Census Region 3) accounted for 50.3% of all deaths (n = 578; AAMR 0.08 per 100,000, 95% CI: 0.07–0.09; RR vs. national: 1.29, 95% CI: 1.17–1.43; p < 0.001). The West contributed 25.5% (n = 293; AAMR 0.07, 95% CI: 0.06–0.08; RR: 1.10, 95% CI: 0.97–1.25; p = 0.151, not significant). The Midwest showed a statistically significant declining trend over the study period (APC: −12.46%; 95% CI: −22.26% to − 1.42%; p = 0.028) and had the lowest mortality burden relative to the national average (RR: 0.59; 95% CI: 0.49–0.70; p < 0.001). The Northeast also had below-national mortality (RR 0.70 [95% CI 0.59–0.84]; p < 0.001). The regional data are summarized in Table 6 and presented annually in the Supplementary Table 5.

Table 6.

Syphilis-related mortality in the U.S. Census region, 2021–2025

Census region	Deaths (n)	% of total	AAMR (95% CI)	RR vs. national (95% CI)	p-value	APC (95% CI)	APC p-value
South	578	50.3%	0.08 (0.07–0.09)	1.29 (1.17–1.43)	< 0.001	+ 0.39% (− 5.24% to 6.35%)	0.895
West	293	25.5%	0.07 (0.06–0.08)	1.10 (0.97–1.25)	0.151	−2.96% (− 10.51% to 5.22%)	0.466
Midwest	139	12.1%	0.02 (0.02–0.03)	0.59 (0.49–0.70)	< 0.001	−12.46% (− 22.26% to − 1.42%)*	0.028
Northeast	138	12.0%	0.03 (0.02–0.04)	0.70 (0.59–0.84)	< 0.001	+ 7.18% (− 4.77% to 20.64%)	0.250
National	1,148	100%	0.06 (0.06–0.07)	1.00 (referent)	—	−1.24% (− 5.20% to 2.88%)	0.550

Note: AAMR = age-adjusted mortality rate per 100,000 (2000 U.S. standard population). RR, rate ratio; Poisson regression, national rate = referent. APC = annual percent change; Poisson regression. * Statistically significant declining trend (p = 0.028). The annual data are listed in Supplementary Table 5

At the state level, Mississippi (AAMR 0.18, RR: 2.64, 95% CI: 1.80–3.87), Oklahoma (AAMR 0.18, RR: 2.55, 95% CI: 1.82–3.59), Louisiana (AAMR 0.14, RR: 2.33, 95% CI: 1.68–3.23), South Carolina (AAMR 0.13, RR: 2.22, 95% CI: 1.63–3.03), and Maryland (AAMR 0.13, RR: 2.08, 95% CI: 1.54–2.81) had the highest mortality rates relative to the national rate (all p < 0.001), although several of these estimates rest on small numbers of deaths with wide confidence intervals (e.g., Mississippi, n = 27) and should be interpreted cautiously. California (n = 164, RR = 1.24, p = 0.010) and Florida (n = 106, RR = 1.36, p = 0.003) had significantly elevated rates. Texas (n = 118, RR = 1.12, 95% CI = 0.93–1.36, p = 0.231) did not show statistical significance. Full state-level data are presented in Supplementary Table 4, selected Poisson comparisons are presented in Table 7, and the regional trends are shown in Fig. 5.

Table 7.

State-level Poisson regression results for syphilis-related mortality, United States, 2021–2025 (selected states)

State	Deaths (n)	AAMR (95% CI)	RR vs. national (95% CI)	p-value
Mississippi	27	0.18 (0.12–0.27)	2.64 (1.80–3.87)†	< 0.001
Oklahoma	34	0.18 (0.13–0.25)	2.55 (1.82–3.59)	< 0.001
Louisiana	37	0.14 (0.10–0.20)	2.33 (1.68–3.23)	< 0.001
South Carolina	41	0.13 (0.10–0.19)	2.22 (1.63–3.03)	< 0.001
Maryland	44	0.13 (0.09–0.18)	2.08 (1.54–2.81)	< 0.001
Florida	106	0.05 (0.04–0.07)	1.36 (1.11–1.66)	0.003
California	164	0.08 (0.07–0.10)	1.24 (1.05–1.46)	0.010
Texas	118	0.09 (0.07–0.10)	1.12 (0.93–1.36)	0.231 (ns)
National	1,148	0.06 (0.06–0.07)	1.00 (referent)	—

Note: The table shows the states with statistically significant elevated RRs and the two highest-volume states. AAMR = age-adjusted mortality rate per 100,000 (2000 U.S. standard population). RR, rate ratio; Poisson regression, national rate = referent. † Mississippi, n = 27; interpreted with caution. The Texas RR was not statistically significant (p = 0.231). ns = not significant. DC and 18 other states had suppressed data; the full data are presented in Supplementary Table 4

Fig. 5.

Age-adjusted syphilis-related mortality rates in the U.S. Census region, United States, 2021–2025

Place of death

Among the 1,148 decedents, 46.1% (n = 529) died in inpatient medical facilities, 21.9% (n = 251) at home, 14.6% (n = 168) in nursing homes or long-term care facilities, and 7.7% (n = 88) in hospice facilities. In total, 53.9% of deaths occurred outside inpatient hospital settings. Because the database contains no information on care received before death, the place of death is reported descriptively and is not evidence of the presence or absence of prior medical care. Population denominators and rates were not available for place-of-death stratification in CDC WONDER. The full distribution is presented in Table 8 and in Supplementary Table 6.

Table 8.

Syphilis-related mortality by place of death, United States, 2021–2025

Place of death	Deaths (n)	% of total deaths
Medical facility – inpatient	529	46.1%
Decedent’s home	251	21.9%
Nursing home / long-term care	168	14.6%
Hospice facility	88	7.7%
Medical facility – outpatient or ER	58	5.1%
Other	54	4.7%
Medical facility – dead on arrival / unknown	0	0.0%
Total	1,148	100%

Notes: Population denominators and age-adjusted rates were not available for place-of-death stratification in CDC WONDER. The data represent the cumulative totals for 2021–2025

Sensitivity Analyses

When restricted to UCOD only, 245 deaths were identified (21.3% of the MCOD total), indicating that syphilis was recorded predominantly as a contributing rather than the underlying cause on U.S. death certificates. The any-mention definition identified 4.7 times more deaths than the underlying-cause definition. The UCOD-only APC was not significant (+ 1.41%; 95% CI: −7.18% to 10.80%; p = 0.756), consistent with the primary analyses.

Late syphilis (A52, n = 590) showed a statistically significant declining trend over the study period (APC: −7.55%; 95% CI: −12.70% to − 2.11%; p = 0.007), whereas deaths coded to unspecified syphilis (A53.0/A53.9, n = 453) increased significantly (APC: +9.05%; 95% CI: +2.14% to + 16.43%; p = 0.009). These opposing trends are compatible with a shift in death certificate coding toward less specific syphilis classifications over time, although the data cannot distinguish coding change from a genuine change in stage-specific disease burden, and both may contribute.

HIV (B20–B24) was co-listed in 316 deaths (27.5%), with a borderline significant increasing trend in HIV–syphilis co-mortality (APC: +8.25%; 95% CI: +0.10% to + 17.08%; p = 0.047). COVID-19 (U07.1) was co-listed in 44 deaths (3.8%), concentrated in 2021 (n = 19) and 2022 (n = 15), and the subsequent years were suppressed. Congenital syphilis (A50, n = 95) showed no significant trend (APC: −5.24%; p = 0.459), although overdispersion (Pearson dispersion = 5.88) limited the confidence in this estimate. Early syphilis (A51) contributed to only 12 deaths over five years, with most annual cells suppressed and no trend model estimable.

Restriction to 2021–2024 final data only (n = 919; AAMR 0.05, 95% CI: 0.05–0.06; APC: −1.78%, 95% CI: −7.30% to 4.06%; p = 0.541) was fully consistent with the primary five-year analysis, indicating that the inclusion of provisional 2025 data did not change the findings. The full sensitivity analysis data are presented in Supplementary Tables 7 and Fig. 6. In a longer-term context analysis spanning 1999–2025 (Supplementary Table 8, Fig. 7), age-adjusted syphilis-related mortality declined to a nadir around 2015 (APC − 3.0%, 95% CI − 5.3 to − 0.7) and rose thereafter (2015–2025 APC + 11.3%, 95% CI + 6.1 to + 16.8), yielding a modest net increase across the period (AAPC + 2.3%, 95% CI + 0.0 to + 4.6); this increase was confined largely to males (AAPC + 2.9%, 95% CI + 0.3 to + 5.5), whereas female mortality reached an earlier nadir (around 2009) and showed no net change (AAPC − 1.5%, 95% CI − 4.6 to + 1.7). An independent Poisson model of annual death counts reproduced this pattern.

Fig. 6.

Sensitivity analyses of syphilis-related mortality using alternative case definitions and restricted time windows in the United States, 2021–2025

Fig. 7.

Long-term trends in age-adjusted syphilis-related mortality overall and by sex in the United States from 1999 to 2025 (joinpoint regression; observed rates with fitted segments)

Discussion

Overall mortality and preventable burden

In this national analysis of syphilis-related mortality from 2021 to 2025, mortality remained stable, with no statistically significant temporal trend, against a background of steadily increasing syphilis incidence in the United States [3, 5]. This stability is best observed over a longer trajectory. Over 1999–2025, syphilis-related mortality fell through the 2000s and the early 2010s and then rose from approximately 2015, consistent with the increase in syphilis-related mortality reported by Barragan et al. for 2015–2020 [10]. The flat 2021–2025 segment is therefore better interpreted as a recent plateau atop a longer rise than as the long-term stability. Because of the prolonged latency between infection and fatal complications, current late-stage mortality largely reflects historical rather than contemporary incidence, and the mortality consequences of the recent rise in incidence may not be apparent for years. Stability also reflects a fixed pool of individuals with long-standing untreated infections, alongside improved treatment of incident cases. A parallel divergence is visible in surveillance: the 2023 CDC STI surveillance report documented over 209,000 syphilis cases nationally, with primary and secondary syphilis declining for the first time in more than two decades, while late and unknown-duration syphilis continued to increase [2]. This divergence has several possible explanations, including expanded screening that detects more asymptomatic infections, missed primary and secondary lesions, and workforce and funding constraints in sexually transmitted infection services, in addition to any changes in care access; the present mortality data cannot adjudicate among them. Stable mortality should not be read as reassuring: 1,148 deaths over five years had syphilis recorded as a cause or contributor, and even allowing that syphilis was not necessarily the proximate cause in every case, it represents a substantial and largely preventable burden for a treatable infection. That preventability is not unconditional: the 2023 shortage of benzathine penicillin G strained treatment capacity during the latter part of the study window [16]. Benzathine penicillin G is the only recommended treatment for syphilis in pregnancy and congenital syphilis; therefore, the shortage was most consequential for pregnant patients and their infants. Non-pregnant adults can generally be treated with oral doxycycline as an alternative [17].

Late-stage disease and coding specificity

The predominance of late syphilis (51.4%) is consistent with fatal syphilis in the United States, arising mainly from chronic, long-standing infections rather than acute disease, although death certificate data cannot confirm the timing or treatment history of the underlying infection. The observed decline in late-stage syphilis coding alongside a concurrent increase in unspecified syphilis coding may reflect a shift toward less specific classification; however, this is one of several possible explanations and cannot be confirmed with the present data alone. Similar patterns of less specific cause-of-death coding have been described in other mortality analyses using ICD-10 data [18]. If such a shift occurs, analyses restricted to specific ICD-10 categories may underestimate the true burden of advanced disease; the any-mention MCOD definition is less sensitive to this problem and is preferable for future surveillance.

Sex disparities and urological relevance

The marked male predominance observed in this study is consistent with national incidence patterns, in which syphilis disproportionately affects males, particularly MSM [3, 5]. From a urological perspective, late syphilis encompasses several conditions that are relevant to clinical practice. Neurogenic lower urinary tract dysfunction is a recognized complication of neurosyphilis, with urodynamic studies in contemporary cases demonstrating low-capacity bladders with detrusor overactivity, urge incontinence, and urinary retention [19, 20]. The spectrum of lower urinary tract dysfunction associated with infectious and inflammatory neurological conditions, including neurosyphilis, has been systematically characterized in a recent multinational neuro-urology study [21]. Tabes dorsalis, the dorsal column demyelinating form of late neurosyphilis, remains clinically active in the modern antibiotic era and continues to present with bladder dysfunction as a prominent feature, including in HIV-negative patients [22, 23]. Beyond the lower urinary tract, late syphilis encompasses gummatous involvement of the genitourinary tract and cardiovascular complications such as syphilitic aortitis [24]. The broader clinical and translational context of the syphilis resurgence, including emerging treatment strategies relevant to late-stage diseases, has been reviewed by Xiong et al. [25].

Despite increasing awareness of the resurgence of syphilis, the absence of a significant decline in male mortality suggests persistent gaps in the early detection and management of this disease. Primary, secondary, and early latent syphilis remain treatable with a single intramuscular dose of benzathine penicillin G, and late presentation, loss to follow-up, or undiagnosed infection are plausible contributors to these deaths, although the data cannot establish which, if any, applies to individual cases [17]. For MSM, the demographic carrying the greatest incidence burden, the 2024 CDC guidelines on doxycycline post-exposure prophylaxis represent a now-available prevention tool: three large randomized controlled trials demonstrated that 200 mg doxycycline taken within 72 h of sex reduces syphilis acquisition by more than 70% in this population [26]. Clinicians, including urologists, should consider syphilis (with reflex HIV testing) in patients presenting with otherwise unexplained neurogenic bladder, tabes-like sensory symptoms, or syphilitic aortitis, stage a reactive serology rather than assuming remote infection, and refer confirmed cases to infectious diseases or sexual health services for treatment and partner management. Eligible MSM should be informed about doxyPEP as part of a comprehensive sexual health strategy.

Racial and ethnic disparities

Substantial racial and ethnic disparities were evident, with NH Black individuals experiencing a nearly five-fold higher mortality risk and NH American Indian or Alaska Native individuals demonstrating the highest rates overall. These disparities are consistent with well-documented structural inequities in the STI burden, including differences in healthcare access, socioeconomic conditions, and continuity of care [3, 27]. The magnitude of these disparities suggests that excess mortality is unlikely to be explained by differences in incidence alone; differences in access to testing and treatment, retention in care, and social determinants of health are plausible contributors, although none were measured here and cannot be quantified from mortality data. Hispanic populations also demonstrated an elevated mortality risk, highlighting the need for targeted public health investments in demographically vulnerable communities.

Geographic variation

Geographic variation further highlighted the disproportionate burden of syphilis mortality in the Southern United States, which accounted for over half of all deaths and exhibited significantly higher mortality rates than the national average. This pattern aligns with longstanding regional disparities in the STI burden, healthcare infrastructure, and socioeconomic determinants of health [2, 28]. The observed decline in the Midwest is a new observation but should be interpreted cautiously, given the small number of cases and potential statistical instability. State-level variation, with Mississippi, Oklahoma, Louisiana, South Carolina, and Maryland each exceeding twice the national mortality risk, identified geographic priorities for further study.

Place of death and missed clinical opportunities

More than half of the syphilis-related deaths occurred outside inpatient hospital settings, with 21.9% occurring at home and 14.6% in nursing homes or long-term care facilities. We interpret this result descriptively. The database holds no information on diagnosis timing, treatment history, healthcare use, or screening; therefore, it cannot distinguish missed opportunities for syphilis care from the at least equally plausible explanation that the place of death reflects the age, comorbidity burden, frailty, and hospice or end-of-life patterns expected when 45.5% of decedents were aged ≥ 65 years, and syphilis was usually a contributing rather than the underlying cause. A death at home or in long-term care is not, in itself, evidence of disengagement from care; residents of long-term care facilities are, by definition, under ongoing medical supervision. Existing guidance recommends risk-based syphilis screening in relevant populations [29, 30]; however, whether gaps in screening contributed to these deaths cannot be determined from death certificate data and would require linked clinical information that this study does not contain.

HIV co-occurrence and pandemic context

The high prevalence of HIV co-occurrence (27.5%) and its borderline significant increasing trend (APC + 8.25%, p = 0.047) reinforce the importance of the HIV-syphilis syndemic in shaping mortality rates. HIV infection accelerates syphilis progression, elevates the viral load, and reduces CD4 counts in co-infected individuals, increasing the risk of severe and late-stage disease [6, 31]. The observed increase in co-mortality likely reflects multiple converging factors: an aging cohort of people living with HIV who acquired the infection before effective antiretroviral therapy was available and are now entering the age groups with the highest syphilis mortality; rising syphilis incidence among HIV-positive MSM; and potential underutilization of doxyPEP in HIV-positive individuals for whom the 2024 CDC guidelines are explicitly applicable [26, 32]. These findings support the use of integrated screening and co-management strategies that simultaneously address both types of infections.

The COVID-19 pandemic likely influenced syphilis-related mortality patterns during the early stages of this study. Disruptions in STI screening, reduced access to clinical services, and reallocation of public health resources during 2020–2022 have been well documented [7, 33], and delayed diagnosis during this period may have expanded the pool of individuals at risk for late-stage complications [34]. The low COVID-19 co-listing rate (3.8%, concentrated in 2021–2022) and the stability of the primary trend estimate when restricted to the 2021–2024 final data suggest that the pandemic period did not materially confound the overall findings, although its longer-term effects on the late-stage disease pipeline warrant ongoing surveillance.

Sensitivity analyses and methodological robustness

The sensitivity analyses reinforce the two interpretive points. First, syphilis was recorded predominantly as a contributing rather than the underlying cause: the underlying-cause definition captured approximately one-fifth of the any-mention deaths, consistent with multiple-cause analyses of other chronic infections and supporting any-mention coding as the appropriate standard for surveillance of conditions that contribute to, rather than directly cause, death [35].

Second, the opposing late- and unspecified-syphilis coding trends are the most methodologically salient findings. Their near-mirror-image movements are compatible with a drift toward less specific coding, although we cannot exclude a genuine change in classification or documentation and present it as a hypothesis rather than a demonstrated mechanism. The 2024 CDC laboratory recommendations revised serological testing terminology but did not change the testing algorithms or their interpretation [36]. Regardless of the cause, surveillance relying solely on A52 will increasingly underestimate the late-stage burden, and the any-mention, broad-code definition is less sensitive to this problem and is preferable for future analyses.

The low COVID-19 co-listing and the stability of the estimate when restricted to 2021–2024 final data indicate that pandemic-period coding and provisional 2025 data did not materially affect the findings.

Limitations

This study had several limitations. First, and most important for interpretation, there is a long and variable latency, often spanning decades, between syphilis infection and fatal, late-stage complications. Because 51.4% of decedents had late syphilis and 45.5% were aged ≥ 65 years, the mortality observed during 2021–2025 largely reflects infections acquired years earlier rather than the recent rise in incidence. The stable trend across this window is largely uninformative regarding the future mortality consequences of recent incidence increases, which, in the absence of improved detection and treatment, may not become apparent in late-stage mortality for years. In contrast, congenital syphilis mortality reflects contemporary transmission and is the stratum most likely to track the current incidence. Second, syphilis mortality captured on death certificates depends on the accuracy and completeness of clinical coding [8], and these data have further accuracy limitations specific to syphilis. Many clinicians receive limited training in death certificate completion, syphilis is serologically and clinically difficult to stage, clinicians often lack the prior testing history needed to distinguish remote treated infection from active disease, and assignment of ICD-10 codes from free-text certificates introduces variability. Attributing a causal or contributory role to a chronic or latent infection in a decedent with multiple comorbidities is particularly uncertain. Regional differences in these practices may contribute to the geographic variation observed, and the findings should therefore be read as reported causes or contributing conditions rather than the verified causes of death. Syphilis may also be under-documented in older decedents, where late-stage complications overlap with cardiovascular diseases and dementia. Third, the cross-sectional surveillance design precludes causal inference; rate ratios describe population-level associations and cannot be attributed to individual causal pathways.

Fourth, the five-year observation window limits the statistical power of trend analyses; Poisson regression CIs are wide in strata with small annual counts, and APC estimates for the 0–24 age group (Pearson dispersion 4.21) and congenital syphilis (Pearson dispersion 5.88) are unstable and should be treated as descriptive. We restricted the primary analysis to 2021–2025 to characterize the most recent post-pandemic period within a single data vintage. To offset the limited window, we provide a 1999–2025 context analysis (Supplementary Table 8, Fig. 7), although the pre-2021 final and 2021–2025 provisional segments rest on slightly different population vintages. Fifth, CDC WONDER suppression rules resulted in missing data for 19 states and the District of Columbia [11]; the NH AI/AN rate ratio of 7.00 rests on 36 total deaths, with four of five annual cells suppressed, and should be read as a directional signal, a caution reinforced by the known racial misclassification of AI/AN individuals on death certificates [15]. Sixth, race and Hispanic ethnicity data were extracted from separate CDC WONDER queries and were not mutually exclusive of each other. Seventh, the 2025 data are provisional and subject to revision; although a sensitivity analysis restricted to the 2021–2024 final data was consistent with the primary findings, the most recent estimates may change when the final data are released. Finally, the place-of-death data lacked population denominators, precluding rate estimation for that stratum.

Conclusions

Between 2021 and 2025, syphilis-related mortality in the United States was stable overall but concentrated in identifiable groups: older adults, males, NH Black and AI/AN individuals, and residents of the South, with more than half of deaths occurring outside of inpatient settings. Late-stage syphilis was listed as the cause of death in 51.4% of cases, and HIV in 27.5%. The opposing A52 and A53 coding trends may reflect a loss of diagnostic specificity on death certificates rather than a true reduction in the late-stage burden. These patterns are consistent with gaps in early detection and treatment retention, although incidence, case ascertainment, and competing comorbidities cannot be disentangled using these data points. Syphilis testing and timely referral are relevant across primary care, infectious diseases, sexual health, and urology, rather than any single specialty. The concentration of deaths in the South and the declining trend in the Midwest warrant further investigation. At-risk populations and geographical areas are identifiable, and effective prevention and treatment tools are available.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AAMR

Age-adjusted mortality rate

APC

Annual percent change

AAPC

Average annual percent change

CDC

Centers for Disease Control and Prevention

CI

Confidence interval

COVID-19

Coronavirus disease 2019

doxyPEP

Doxycycline post-exposure prophylaxis

HIV

Human immunodeficiency virus

ICD-10

International Classification of Diseases, 10th Revision

IRB

Institutional review board

LUTD

Lower urinary tract dysfunction

LUTS

Lower urinary tract symptoms

MSM

Men who have sex with men

MCOD

Multiple causes of death

NH

Mon-Hispanic

NH AI/AN

Non-Hispanic American Indian or Alaska Native

NH Black

Non-Hispanic Black or African American

NUPC

Neurourology and Pelvic Floor Committee

RR

Rate ratio

STI

Sexually transmitted infection

UCOD

Underlying cause of death

USPSTF

US Preventive Services Task Force

WONDER

Wide-ranging Online Data for Epidemiologic Research

Author contributions

KA contributed to conceptualization, literature review, data interpretation, and writing of the original draft. EA contributed to data curation, formal analysis, and writing of the original draft. VD contributed to methodology, statistical analysis, and writing — review and editing. AUF contributed to data curation, visualization, and writing — review and editing. SQ contributed to conceptualization, supervision, methodology, project administration, and writing — review and editing. MAM contributed to supervision, critical appraisal of intellectual content, and writing — review and editing. All authors read and approved the final manuscript.

Funding

The authors did not receive any specific funding for this study. The APC was covered by Alfaisal University.

Data availability

All data analyzed in this study are publicly available through the CDC WONDER Multiple Cause of Death database at https://wonder.cdc.gov/mcd.html and found in supplementary file.

Declarations

Ethics approval and consent to participate

This study used de-identified, publicly available aggregate mortality data obtained from the CDC WONDER Multiple Cause of Death database, which is publicly available, de-identified data and exempted from institutional review board review. No individual patient data were accessed, and patient consent was not required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.