Analysis of the spatio-temporal pattern and evolutionary trends of syphilis at township level in Xining City, Qinghai Province, China from 2008 to 2024

Abstract

From 2008 to 2024, a total of 15,465 syphilis cases were reported in Xining City, Qinghai Province, yielding an average annual incidence of 4.44 per 10,000 population and a male-to-female ratio of 1.14. Temporal analysis revealed a steadily increasing trend over the 17‑year period, with a consistent seasonal peak in March. Spatial autocorrelation analysis showed significant clustering (global Moran’s I = 0.217, P < 0.05), with high‑high clusters concentrated in eastern urban districts and low‑low clusters predominantly in northern areas. The standard deviational ellipse indicated a dominant southeast–northwest directional trend. Spatio‑temporal scan statistics identified four statistically significant high‑incidence clusters, and Kriging interpolation produced a smoothed surface suggesting elevated transmission risk in the eastern and southern townships. These findings demonstrate that syphilis incidence in Xining increased steadily and expanded geographically from urban centers to peripheral areas over the study period, with pronounced spatial and spatio‑temporal heterogeneity. As a hypothesis‑generating descriptive spatial analysis, this study supports enhanced dynamic surveillance, targeted interventions in high‑risk regions, and locally adapted public health measures to strengthen syphilis control.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-53877-7.

Introduction

Syphilis, a chronic systemic infection caused by Treponema pallidum, is primarily transmitted through sexual contact or vertically from mother to child¹. The disease progresses through distinct clinical stages: an initial highly infectious phase characterized by chancres and rash, which can advance to latent and tertiary stages involving severe, often irreversible cardiovascular and neurological damage^2,3. Despite available treatments such as intramuscular benzathine penicillin (the first-line regimen recommended by WHO)⁴, global and national incidence rates continue to rise⁵. This trend underscores persistent challenges in disease control, reflecting underlying social and spatial inequities in healthcare access^6,7. Beyond its direct clinical burden, syphilis infection synergistically increases the risk of HIV acquisition, further amplifying its public health impact⁸. This sustained transmission highlights the critical need for enhanced surveillance and spatially informed intervention strategies.

Sexually transmitted infections (STIs) represent a major global public health challenge, with syphilis being particularly consequential due to its severe clinical outcomes and societal impact⁹. Epidemiological evidence indicates elevated risk among both younger and older groups: higher sexual activity and risky behaviors drive transmission in younger populations¹⁰, whereas declining immune function and delays in healthcare seeking increase vulnerability among older adults¹¹. In China, syphilis is classified as a Category B notifiable infectious disease under the Law on the Prevention and Treatment of Infectious Diseases, indicating its high transmissibility and substantial public health burden, which mandate stringent control measures¹², the reporting and management requirements for Category B diseases are second only to those for Category A diseases, such as plague. Globally, reported incidence remains high, with 7.1 million new cases in 2020, up from 6.3 million in 2016¹³. In China, syphilis is a Category B notifiable disease, and its incidence rose steadily for decades, peaking in 2019 before a recent gradual decline^14,15. Despite this trend, it remains a major threat requiring sustained control efforts.

Spatio-temporal analytical methods—such as spatial autocorrelation (Moran’s I)¹⁶, trend surface analysis¹⁷, Spatio-Temporal Scan Analysis¹⁸, and Kriging interpolation¹⁹—are established tools in infectious disease epidemiology for characterizing risk distribution, detecting transmission clusters, and identifying contributing factors²⁰. These approaches have been applied successfully to model the spatial dynamics of various communicable diseases, generating evidence that supports targeted public health actions. For syphilis, which is highly transmissible in its early stages, analyzing fine scale spatio-temporal patterns is critical for designing effective interventions²¹. Delineating high‑incidence areas through such methods enables efficient resource allocation and strengthens early prevention and treatment services²².

However, significant gaps persist in the spatial epidemiological understanding of syphilis. While temporal trends at national or provincial scales are documented, the micro-scale spatial clustering, long-term evolutionary patterns at the township level, and drivers of localized persistent hotspots remain unclear. Most existing studies focus on static, aggregated scales, lacking comprehensive long-term township-level evaluations^13,23–25. Furthermore, conventional time-series analyses often neglect spatial heterogeneity, limiting insight into spatio-temporal interactions²⁶. Although methodological advances in spatial analysis and environmental-health modeling offer powerful tools to uncover hidden transmission dynamics, their application to syphilis at a granular scale remains limited, and an analytical framework is yet to be established. This study aims to address these gaps by employing an integrated spatio-temporal approach to elucidate the micro-scale and long-term patterns of syphilis transmission.

This study examines the fine-scale, long-term spatio-temporal dynamics of syphilis by analyzing its 17-year incidence at the township level in Xining, a multi-ethnic region with socioeconomic disparities that may influence transmission patterns. Specifically, we investigate the spatial clustering and temporal stability of syphilis incidence, track the evolution of the epidemic’s geographic centroid and directional distribution, and identify the most significant and persistent spatio-temporal clusters. To address these objectives, we integrate spatial autocorrelation, trend surface analysis, standard deviational ellipse, spatio-temporal scan statistics, and ordinary kriging interpolation within a unified analytical framework. This approach locates transmission hotspots, quantifies directional expansion, and produces a continuous risk surface. To our knowledge, no previous spatial or spatio-temporal analysis of syphilis has been conducted in Xining or even in the broader Qinghai Province. This study thus provides the first such evidence at the township level, leveraging a 17-year observation period from 2008 to 2024 that spans multiple phases of China’s syphilis prevention policies. By combining five complementary spatial methods in a single framework and focusing on Xining—a multi-ethnic, rapidly urbanizing city on the eastern Qinghai-Tibet Plateau—this work offers a replicable approach for fine-scale infectious disease mapping. The findings are intended to support spatially targeted resource allocation and intervention strategies, and the analytical framework can be adapted to other urban settings and infectious diseases.

Methods

Study area

Administrative boundary data were sourced from the National Geospatial Information Public Service Platform (https://cloudcenter.tianditu.gov.cn/administrativeDivision/), certified under map approval number GS(2024)0650. Xining is situated at 36°02′N–37°28′N and 100°52′E–101°54′E, administering 76 townships across five districts (Chengdong, Chengxi, Chengzhong, Chengbei, Huangzhong) and two counties (Datong, Huangyuan), with a total area of 7,660 km². By the end of 2024, the permanent resident population of Xining had reached 5.39 million. The specific information can be found in Appendix 1 (Table S1).

Data sources

Syphilis case data for Xining from 2008 to 2024 were sourced from the China National Notifiable Disease Reporting System, with all cases clinically and laboratory confirmed according to national diagnostic guidelines. Diagnoses adhered to consistent national criteria (GB15974-1995 before 2007, WS 273–2007 from 2007 onward), ensuring alignment with surveillance practices in mainland China²⁷. The township was selected as the spatial unit because it represents the smallest administrative level at which stable, longitudinal population and case data are systematically available, allowing for fine-scale spatial analysis that can directly inform local intervention planning. The study period (2008–2024) captures long-term trends after the major revision of national diagnostic guidelines and spans distinct phases of syphilis prevention policy in China. Incidence was calculated as confirmed monthly cases per 100,000 population, using mid-year registered population figures from the Xining Statistical Yearbook as the denominator (http://tjj.qinghai.gov.cn/). These figures were compared with national census data to verify consistency. Case records were anonymized, geocoded to township boundaries obtained from the National Geomatics Center of China, and spatially joined for analysis.

Descriptive analysis

We conducted a descriptive analysis of syphilis epidemiological characteristics in Xining from 2008 to 2024 using Microsoft Excel2020. This analysis summarized annual case counts, temporal trends, and distributions by gender, age, occupation, and geographic region. To enable valid comparisons of disease risk across township-level units while accounting for age and sex differences, we computed the Standardized Morbidity Ratio (SMR) for each township annually²⁸. The crude incidence rate for township i in year t was calculated as the number of reported cases divided by the mid-year population estimate . Expected case countswere derived via indirect standardization using county-level, age- and sex-specific incidence rates as the reference. The expected count for each township represents the number of cases that would be expected if the township had the same age- and sex-specific incidence rates as the overall county. Since township-level age–sex population data were unavailable, we approximated each township’s demographic structure using the corresponding county-level distribution.This approach adjusts for potential confounding bypopulation structure while allowing meaningful spatial comparisons of syphilis burden. Specifically, the SMR for township i in year t is defined as:

where is the number of syphilis cases at the township level i(1 ≤ i ≤ 76) during year t(2008 ≤ t ≤ 2024) and is the expected cases in township i during year t, which can be calculated by multiplying the annual population composition and the reported incidence of syphilis in the county for a particular year t.

Spatial analysis

Spatial autocorrelation statistics assess the spatial distribution of disease cases by measuring whether the attribute values in one geographic unit are systematically related to those in neighboring units, thereby revealing patterns of clustering or dispersion. This approach, widely applied in epidemiological studies^29–31, comprises global and local forms. The global Moran’s I index measures the overall clustering tendency across the study area, whereas local spatial autocorrelation identifies specific locations where values are either similar (clusters) or dissimilar (outliers) relative to their surroundings³². In this study, spatial autocorrelation analysis was conducted using ArcGIS 10.8.0 to evaluate whether syphilis cases in Xining exhibited a non‑random spatial pattern. The Moran’s I statistic ranges from − 1 to 1; values significantly greater than zero (P < 0.05) indicate positive spatial autocorrelation, i.e.,significant geographical clustering of cases, while values near zero suggest a random distribution. To visualize local patterns, a Moran scatter plot was employed, which plots incidence values against their spatially lagged counterparts and classifies units into four quadrants: high-high (HH), high-low (HL), low-low (LL), and low-high (LH). In addition, it is crucial to note that LISA clusters are defined not by the absolute value of incidence in a single unit, but by the similarity of its value with its neighboring units. Identifying such spatial aggregation is essential for designing targeted control strategies and for informing subsequent spatial regression modeling³³. The global Moran’s I and local Moran’s I indices were computed according to the following formula³⁴:

In the formula, n denotes the number of spatial units in the study area, is the mean of the attribute values, represents the spatial weight between locations i and j, and W is the adjacency matrix defining their spatial relationships. Moran’s I ranges from − 1 to 1. A statistically significant positive value indicates spatial clustering of the attribute, meaning that similar values tend to be located near each other. Conversely, a significant negative value reflects spatial dispersion, where neighboring units exhibit dissimilar values.

Trend surface analysis

This study used the least squares method to fit a two-dimensional nonlinear regression function and developed a three-dimensional trend surface model to visualize the spatial variation patterns of syphilis across different geographic locations, thereby revealing the overall distribution trend and spatial heterogeneity within the study area³⁵. The analysis decomposes the observed values, including syphilis incidence, into three components: local anomalies, regional trends, and random errors. Using the SMR of syphilis across 76 townships in Xining from 2008 to 2024, we calculated the average SMR for each township, which serves as the dependent variable (Z-axis). The geographical coordinates (latitude and longitude) of each township were treated as independent variables (X-axis and Y-axis)³⁶. A polynomial was then fitted to the scatter plot, generating an optimal fitting curve that models the spatial trend of the disease. In this investigation, trend surface analysis was performed using ArcGIS 10.8.0 software.

Standard deviational ellipse

The Standard Deviational Ellipse (SDE) method was applied to quantitatively analyze the spatial distribution characteristics of the disease³⁷. By calculating the center, major and minor axes, orientation angle, and ellipticity of the SDE, the spatial distribution pattern and dominant expansion direction of the syphilis incidence can be quantitatively characterized³⁸. The center of the ellipse represents the centroid of case distribution, reflecting its average position in the two-dimensional space. In this study, each township’s syphilis incidence rate was assigned to the geometric centroid of its administrative boundary rather than to the exact locations of individual cases. The orientation angle, defined as the angle between the major axis and true north, indicates the primary direction of the epidemic. Ellipticity, defined as the ratio of the minor axis to the major axis, measures the degree of spatial concentration; a smaller ellipticity indicates a more concentrated case distribution and lower spatial dispersion. In this investigation, SDE analysis was performed using ArcGIS 10.8.0 software.

Spatio-temporal scan analysis

Spatio-temporal clusters of reported syphilis incidence in Xining were identified at the township level using the Poisson model in SaTScan software (version 9.5). The analysis employed a dynamic cylindrical scanning window, where the base radius defines the spatial extent of a potential cluster and its height corresponds to the temporal duration. The maximum spatial scanning radius was set to 50% of the population at risk, a threshold shown in prior studies to yield stable results³⁹. To evaluate the statistical significance of candidate clusters, Monte Carlo simulation with 999 random replications was conducted under the null hypothesis of purely random spatio-temporal distribution. This simulation generates empirical p-values for the observed clusters. To define the temporal dimension, the scanning window moved continuously across the entire study period (2008–2024) to detect intervals with statistically elevated incidence. The significance and intensity of each identified cluster were evaluated using the Log-Likelihood Ratio and the Relative Risk, which quantify the likelihood and magnitude of excess risk within the cluster relative to its spatio-temporal context. The formula is as follows⁴⁰:

where n is the observed number of cases within the scan window, N is the total number of cases, E(n) is the expected number of cases within the scan window, and I is an indicator.

Kriging interpolation method

Kriging is a geostatistical method that leverages spatial autocorrelation, the principle by which nearby locations tend to yield more similar values than those farther apart, to generate a continuous surface from discrete observations^16,41. Although typically applied to point data, this method can also be employed for areal data to reveal spatial trends and produce smoothed visualizations, provided the underlying spatial autocorrelation structure meets the necessary assumptions⁴². In this study, a spatial scale transformation was applied in which each township’s syphilis incidence rate, originally aggregated as areal data, was represented by its geometric centroid to enable the application of kriging. The presence of significant spatial autocorrelation was confirmed using the Global Moran’s I statistic (p < 0.05), thereby satisfying the prerequisite for ordinary kriging. The fitted variogram yielded a major range of approximately 0.191 decimal degrees (around 21 km), beyond which spatial autocorrelation became negligible; accordingly, the spatial search neighborhood for ordinary kriging was constrained to this distance threshold. Ordinary kriging was then implemented in ArcGIS 10.8.0 as a spatial smoothing and visualization tool to improve the interpretability of spatial patterns and to mitigate visual discontinuities arising from administrative boundaries. The variogram function serves as a key component in kriging estimation, and its equation is presented as follows:

The application of kriging interpolation is conducted through a sequential process, which comprises data organization, data exploration, model fitting, model diagnostics, model comparison, and ultimately, the production of a predictive distribution map⁴³.

Statistical analysis

We employed an integrated spatial analytical framework to characterize the spatio-temporal patterns of syphilis in Xining, progressing from exploratory analysis to predictive modeling. Initial exploratory analyses delineated the epidemic’s macro-scale structure: trend surface analysis revealed the dominant spatial gradient of incidence, while standard deviational ellipse quantified the geographic centroid, orientation, and dispersion of cases. These results informed subsequent inferential modeling, where Local Moran’s I and spatial scan statistics were used to detect and cross-validate statistically significant clusters. The observed spatial trends were then incorporated into an Ordinary Kriging model to generate a continuous risk surface for predicting incidence in unsampled areas and supporting cluster detection (Fig. 1).

Fig. 1.

Methodological flow diagram for the spatial and spatio-temporal analysis of syphilis incidence in Xining City, 2008–2024.

Results

Demographic characteristics and temporal pattern

From 2008 to 2024, Xining reported a total of 15,466 syphilis cases, corresponding to an average annual incidence of 44.44 per 10,000 population. The male-to-female case ratio was 0.90, based on 7,330 male and 8,136 female cases. During the study period, the overall incidence of syphilis exhibited an upward but fluctuating trend. It increased from 2008 to a peak in 2010, then decreased slightly in 2011, followed by another increase from 2012 to 2014. A marked decline occurred in 2015, after which incidence continued to rise from 2016 to 2021. It showed a significant decrease in 2022, followed by an increase from 2023 to 2024 (Fig. 2a). Seasonally, two consistent annual peaks were identified: a primary peak in March and a secondary peak between July and August (Fig. 2b). Regarding demographic distribution, individuals aged 60 years or older constituted the largest proportion of cases at 22.72%, followed by those aged 20–29 years at 21.98% and 30–39 years at 19.48% (Fig. 2c). The age group of 0–9 years accounted for the smallest proportion, at 5.05%. By occupation, farmers and herdsmen represented the largest group, accounting for 26.92% of cases, while homemakers or unemployed individuals comprised 21.51% (Fig. 2d). A summary of these epidemiological characteristics is presented in Fig. 2.

Fig. 2.

Epidemiological characteristics of syphilis in Xining, 2008–2024. (a) Annual incidence by gender; (b) Monthly distribution of reported cases; (c) Case distribution by age group; (d) Case distribution by occupation.

Spatial distribution and trend surface analysis

Figure 3 presents the spatio-temporal distribution of syphilis SMRs at the township level in Xining from 2008 to 2024, showing a clear expansion of the high-risk pattern across the municipality. Cases first emerged in Chengdong District and subsequently spread to adjacent central, western, and northern urban districts. The primary high-risk areas included Chengdong, Chengxi, Chengbei, Chengzhong, and Huangzhong districts, whereas Datong and Huangyuan counties reported comparatively fewer cases. Throughout the study period, Chengdong District consistently recorded the highest SMRs, remaining elevated relative to other regions.

Fig. 3.

Temporal-spatial distribution map of SMRs in Xining City from 2008 to 2024 (Map generated with ArcGIS 10.8. Software available at https://www.esri.com/en-us/arcgis/products/arcgis-desktop/resources).

Figure 4 illustrates the relationship between longitude (X-axis), latitude (Y-axis), and syphilis incidence rates (Z-axis). Analysis of the annual average incidence at the township level reveals significant spatial heterogeneity, characterized by a pronounced east-west gradient with consistently higher rates in the eastern regions. Along the north-south axis, the distribution follows an inverted U-shape, where incidence peaks in the central latitudes and declines toward both the northern and southern extremes, exhibiting less variation than observed along the east-west gradient. Ordinary kriging was employed, as the data met the assumption of normality.

Fig. 4.

Spatial three-dimensional trend surface analysis of syphilis.

Spatial autocorrelation analysis and spatial evolutionary trends in morbidity of syphilis

The spatial autocorrelation of syphilis incidence in Xining from 2008 to 2024 was assessed using the global Moran’s I index. The results, presented in Table 1, indicate a significant positive spatial autocorrelation in the average annual incidence across the study period, confirming a non-random, clustered distribution of cases. Specifically, significant positive spatial clustering was identified during two distinct intervals: from 2008 to 2016 and again in 2020. This pattern suggests that the spatial structure of the syphilis epidemic was not constant but exhibited phases of intensified geographic aggregation.

Table 1.

Global spatial autocorrelation analysis of syphilis in Xining City, China in 2008–2024.

Year	Moran’I	S.D.	Z-score	P-value
2008	0.169	0.016	2.589	0.009
2009	0.158	0.017	2.474	0.013
2010	0.157	0.018	2.424	0.015
2011	0.149	0.018	2.329	0.020
2012	0.102	0.018	1.995	0.046
2013	0.125	0.018	2.175	0.030
2014	0.150	0.018	2.376	0.018
2015	0.137	0.018	2.285	0.022
2016	0.120	0.018	2.126	0.034
2017	0.117	0.024	1.828	0.068
2018	-0.017	0.012	1.384	0.166
2019	-0.015	0.020	1.073	0.283
2020	0.037	0.009	2.130	0.033
2021	-0.095	0.022	0.484	0.628
2022	-0.003	0.008	1.756	0.079
2023	-0.344	0.008	-1.911	0.056
2024	-0.102	0.020	0.464	0.164

Spatial autocorrelation analysis identified distinct local clustering patterns of syphilis incidence in Xining from 2008 to 2024. Local Moran’s I statistics revealed persistent high-high clusters concentrated in the urban districts of Chengdong, Chengxi, Chengbei, and Chengzhong, with the northern urban area showing the most sustained clustering. Townships including Renmin, Mafang, and Xiaoqiao exhibited high-high clustering throughout the entire 17-year period. The local spatial autocorrelation clusters of syphilis incidence in Xining City are presented in Table S2. Conversely, low-low clusters were primarily found in the agricultural counties of Datong and Huangyuan. While all low-low clusters were located in Datong County before 2020, similar clusters emerged in Dazhua Township and Bayan Township of Huangyuan County from 2021 onward. These results delineate a fine-grained spatial dependence structure where high-high areas represent zones of aggregated elevated incidence, and low-low areas indicate spatially consistent low incidence—patterns that complement the incidence shown in Fig. 5.

Fig. 5.

Yearly local spatial autocorrelation of syphilis incidences at the township level in Xining city, China from 2008 to 2024 (Map generated with ArcGIS 10.8. Software available at https://www.esri.com/en-us/arcgis/products/arcgis-desktop/resources).

From 2008 to 2024, the spatial distribution of syphilis in Xining was characterized using a standard deviation ellipse set to one standard deviation (approximately 68% of cases) in ArcGIS. The ellipse azimuth ranged from 94.50° to 128.92°, increasing from 103.42° to 116.39° over the period. Axis lengths remained nearly constant, with the short axis varying between 36.63 km and 36.72 km and the long axis decreasing marginally from 101.72 km to 101.67 km. These stable dimensions indicate that the core geographic extent of syphilis incidence changed little across the 17 years. The epidemic consistently exhibited a northwest-southeast elliptical pattern, and its center of gravity shifted gradually northwestward from Pengjiazhai Town in Chengxi District toward Mafang Street in Chengbei District, remaining near the junction of Chengxi, Chengbei, and Huangzhong districts. An alternative visualization of the ellipse is provided in Appendix 1 (Figure S1) (Table 2; Fig. 6).

Table 2.

Shape parameters of the ellipse for the standard deviation of syphilis in Xining city, China between 2008 and 2024.

Year	Azimuth(°)	Short Axis(km)	Long Axis (km)	Aspect Ratio(Major/Minor)
2008	103.42	36.64	101.72	2.78
2009	94.77	36.63	101.71	2.78
2010	97.45	36.63	101.72	2.78
2011	128.92	36.72	101.60	2.77
2012	106.18	36.64	101.70	2.78
2013	99.21	36.63	101.70	2.78
2014	104.30	36.63	101.70	2.78
2015	94.50	36.63	101.70	2.78
2016	101.52	36.65	101.69	2.77
2017	99.73	36.66	101.68	2.77
2018	101.62	36.66	101.68	2.77
2019	99.24	36.66	101.67	2.77
2020	105.73	36.66	101.68	2.77
2021	128.92	36.72	101.60	2.77
2022	112.00	36.66	101.68	2.77
2023	118.34	36.67	101.69	2.77
2024	116.39	36.67	101.67	2.77

Fig. 6.

Standard deviation ellipse analysis of syphilis in Xining city, China between 2008 and 2024 (Map generated with ArcGIS 10.8. Software available at https://www.esri.com/en-us/arcgis/products/arcgis-desktop/resources).

Spatio-temporal clustering analysis

We identified four significant spatio-temporal clusters of syphilis incidence in Xining from 2008 to 2024 using a Poisson-based scan statistic, with the maximum spatial scanning window set to 50% of the total population and the maximum temporal window set to 50% of the study period. These clusters, detailed in Table 3; Fig. 7, are ranked from Level 1 to Level 4 in descending order of statistical significance as measured by the log-likelihood ratio. The primary, or Level 1, cluster was centered along Dongguangdajie Road in Chengdong District from 2017 to 2024 and exhibited a relative risk of 6.72. The Level 2 cluster, located on Bayi Road within the same district between 2009 and 2016, showed a higher relative risk of 14.00. A broader Level 3 cluster covered parts of Chengxi, Chengbei, Chengzhong and Huangzhong Districts from 2014 to 2021, encompassing 14 townships within a radius of 13.92 km with a relative risk of 1.73. The Level 4 cluster spanned Huangzhong District and Huangyuan County from 2017 to 2020, included 11 townships within a 40.51-kilometer radius, and had a relative risk of 1.32. Figure 7 delineates the geographic extent of these clusters while accounting for potential inter-regional spillover effects.

Table 3.

Spatio-temporal clusters of syphilis in Xining city, China in 2008–2024.

Most likely cluster	Cluster time frame	Coordinates / radius	Observed cases	Expected cases	RR	LLR	P-value
Level 1 cluster	2017–2024	(36.621060 N, 101.797320 E) / 0 km	968	152.14	6.72	997.49	< 0.001
Level 2 cluster	2009–2016	(36.575580 N, 101.800890 E) / 0 km	329	23.97	14.00	559.74	< 0.001
Level 3 cluster	2014–2021	(36.672610 N, 101.611180 E) / 13.92 km	3047	1925.75	1.73	324.61	< 0.001
Level 4 cluster	2017–2020	(36.756730 N, 100.988980 E) / 40.51 km	368	280.74	1.32	12.59	0.015

Fig. 7.

Spatio-temporal clusters of syphilis incidences at the township level in Xining city, China between 2008 and 2024 (Map generated with ArcGIS 10.8. Software available at https://www.esri.com/en-us/arcgis/products/arcgis-desktop/resources).

The results of kriging interpolation

Figure 8A presents the spatial distribution of syphilis cases in Xining from 2008 to 2024. Kriging was performed using the geometric centroids of township administrative units as input points. The semivariogram model parameters were as follows: Nugget = 0.0988, Range = 0.1908, Partial Sill = 0.4929, and Sill = 0.5917, with the Partial Sill to Sill ratio of 83.3%, suggesting strong spatial autocorrelation. The distribution map, standard error map, standardized error map, and normal Q-Q plot derived from the kriging interpolation for the period 2008 to 2024 are provided in Appendix 1 (Figure S10). High-incidence areas were primarily located in the eastern part of Xining, particularly within the central urban districts of Chengdong, Chengxi, Chengbei, and Chengzhong. Additionally, relatively high incidence was observed in Huangzhong District, the southern part of Huangyuan County, and the southern part of Datong County. Figure 8B further indicates that prediction errors were more pronounced in peripheral regions compared to the central areas of Xining. In addition, kriging interpolation was performed for each spatio-temporal cluster identified by SaTScan, as shown in Supplementary Figures S2 to S5. The corresponding information, standard error, standardized error, and normal Q-Q plots for each interpolation are provided in Appendix 1 (Figures S6–S9).

Fig. 8.

Spatial distribution of syphilis incidence in Xining city, China from 2008 to 2024 (A) Distribution map. (B) Error map (Map generated with ArcGIS 10.8. Software available at https://www.esri.com/en-us/arcgis/products/arcgis-desktop/resources).

Discussion

Syphilis remains a major global public health challenge. In China, persistent difficulties in controlling sexually transmitted infections are compounded by geographic and demographic variations in rising syphilis incidence^44–46. This study analyzed the spatio-temporal evolution of syphilis incidence in Xining from 2008 to 2024, focusing on transmission dynamics and geographic clustering. In this study, the average annual incidence rate was 4.44 per 10,000 population, which is higher than the national average over the same period (2.88/10,000)⁴⁷ and also higher than that of Nantong City (0.66/10,000)⁴⁸, reflecting regional differences in screening intensity and urbanization. From 2008 to 2024, syphilis incidence in Xining showed an overall upward trend, which may be associated with the implementation of the National Syphilis Prevention and Control Plan (2010–2020)⁴⁹. The reduction in late 2022 coincides with a nationwide adjustment in COVID-19 control measures, which may have affected healthcare-seeking behavior and case reporting⁵⁰. In response to the epidemic of syphilis, health authorities have enacted targeted interventions addressing key drivers such as high-risk sexual behaviors, insufficient prevention awareness among vulnerable groups, gaps in service coverage, and variability in diagnostic and treatment protocols^47,51,52.

Syphilis incidence in Xining varied across demographic groups. Reported cases were predominantly among females, individuals aged 20–29 years or ≥ 60 years, and those engaged in farming, herding, homemaking, or unemployment. This is consistent with the findings of a study in Beijing: in terms of age distribution, patients aged 20–29 years accounted for 22.60% (689/3048) in the latter; and in terms of occupational distribution, homemakers or unemployed individuals also constituted the highest proportion, at 30.61% (933/3048)⁵³. The somewhat higher incidence observed among women is likely attributable to biological susceptibility combined with socioeconomic constraints that reduce healthcare access^54,55. In the northwest region of China, where the present study was conducted, such barriers are especially marked⁵⁶. Additional contributing factors may include insufficient sexual health education and uneven coverage of screening services⁵⁷. The concentration of cases in young adults (20–29 years) is associated with higher sexual activity and risk behaviors^58–60, whereas the proportion among older adults (≥ 60 years) likely reflects more frequent health encounters and age-related immune decline^11,61,62. Occupation-based analysis showed that farmers, herdsmen, homemakers, and the unemployed accounted for the largest share of cases. These groups often experience geographic and social marginalization, resulting in limited sexual healthcare access and health literacy^63,64. Persistent economic strain and insufficient social support may further increase engagement in high-risk sexual practices, facilitating transmission in these populations^65,66.

Spatial autocorrelation analysis indicates that high-risk areas for syphilis in Xining remained geographically stable over the 17-year study period. Persistent high-high clusters were consistently identified in the central urban districts of Chengdong, Chengbei, Chengzhong, Chengxi, and Huangzhong, whereas low-low clusters were primarily concentrated in Datong County, showing a gradual northward shift. Notably, certain areas classified as high-high clusters maintained this status even after their absolute incidence declined from peak levels. This reflects the ability of local spatial statistics to identify zones of persistently elevated relative risk, which remain priority targets for intervention. Conversely, emerging high-low outliers may signal early transmission into previously low-risk areas, providing an early warning not apparent in aggregate incidence maps. These findings align with the results of spatio-temporal scan analysis and mirror patterns observed in other Chinese regions, where transmission is closely linked to local socioeconomic development levels^40,48. The spatial pattern is likely driven by socio-ecological factors: urban centers have both dense, mobile populations and frequent social interaction that facilitate transmission, as well as concentrated healthcare resources and active screening that increase case detection^67–69.

Three-dimensional trend surface analysis shows that the syphilis epidemic in Xining has remained spatially concentrated in the eastern and central urban districts over the 17-year study period. Although the outbreak has gradually extended to all townships with an expanding geographic range, its core exhibited stable, elliptically bounded growth with minimal change in axial length, indicating limited directional diffusion. The epidemic’s centroid shifted only slightly northwestward, aligning with the city’s urban development axis, which suggests transmission may continue to rise in tandem with ongoing population and economic growth in this corridor⁷⁰.While the causes underlying these spatial patterns cannot be determined from ecological data alone, the observed distribution is likely influenced by socioeconomic and urban development factors^38,71. Accelerated urbanization, increased population mobility, and evolving social networks have created more complex transmission pathways^48,72.

The interpolated model consistently showed higher incidence rates in Chengdong, Chengzhong, Chengxi, and Chengbei districts, while Datong and Huangyuan counties exhibited steadily declining rates, partly associated with industrial land use. Overall, these findings indicate that syphilis incidence in Xining has not yet peaked. The observed spatial stability, low variance, and limited spillover effects highlight the need for continued surveillance of regional shifts and potential emergence of new high-risk areas. The city’s socioeconomic core has shifted northwestward due to comparatively faster infrastructure and economic growth in western and northern districts, aligning with the observed movement of the epidemic’s focus. Furthermore, a reservoir of untreated historical infections in eastern and central districts, compounded by ongoing transmission, sustains overlapping cycles that facilitate spatial propagation⁷³. Concurrently, improved public awareness and screening accessibility have enhanced case detection and reporting, which may influence the observed spatial prevalence in surveillance data^74–76.

The spatio-temporal patterns of syphilis in Xining arise from a complex interaction of biological susceptibility, behavioral factors, healthcare access, and social determinants. This is consistent with findings from both domestic and international studies^72,77,78. Consequently, targeted prevention strategies should be prioritized. First, interventions must focus on key populations, including women, young adults, older adults, and individuals in low-income occupations. Strengthening health education, proactive screening, and integrated support for these groups can enhance disease control and management outcomes⁷³. Second, surveillance and early warning systems should be reinforced to monitor the expansion of existing hotspots and the emergence of new high-risk areas⁷⁹. Optimizing resource allocation using spatio-temporal epidemiological tools, alongside improved clinical training in primary care, will enable more precise and effective interventions⁸⁰. Third, service integration should be advanced by incorporating routine syphilis testing into existing sexual health services, voluntary HIV counseling and testing, and community-based treatment programs^81–83. This integrated approach can promote early detection, timely treatment, and reduction of the burden of advanced disease.

This study has several limitations. The findings rely on passive surveillance data, which are subject to underreporting and changes in diagnostic criteria. Registered population data may not fully capture temporary migrants, affecting incidence estimates. Crude rate comparisons across regions may reflect demographic differences rather than true disease risk. Key behavioral, socioeconomic, and mobility variables were not included. The township-level analysis is also subject to the modifiable areal unit problem. Nevertheless, core spatio-temporal trends remained consistent across analytical methods and the 17-year study period, supporting the robustness of our findings for identifying geographic intervention priorities. Recent studies have combined GIS analysis with machine learning-based random forest regression to integrate multi-source datasets for more accurate spatial health analyses, which aligns with SDG 3 (Good Health and Well-being)^84,85. Future work should adopt such advanced methods to address the aforementioned limitations.

Future research should integrate multi-source datasets, including socioeconomic, mobility, and behavioral factors, into predictive modeling frameworks to better elucidate transmission drivers. The analytical framework developed here is designed to be replicable and could be applied to examine the spatio-temporal dynamics of other infectious diseases in similar urban or regional settings.

Conclusion

This study presents the first GIS based spatial analysis of syphilis epidemiology in Xining. We identified a persistent northwest to southeast transmission corridor with stable hotspots and a shifting geographic center. The township level resolution allows for targeted surveillance and intervention strategies focusing on high risk clusters and emerging urban areas. Our findings directly support SDG target 3.3, which aims to end the epidemic of neglected tropical diseases, including syphilis, by 2030. By identifying high burden townships and vulnerable populations, this study provides spatial decision support for local health authorities to achieve syphilis control goals. This spatio-temporal framework offers a replicable model for analyzing infectious disease dynamics in similar urban settings, demonstrating how fine scale spatial analysis can translate epidemiological patterns into actionable prevention.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Wei Li and Yongkai Shi designed the study. Data collection was carried out by Ying Zhao. Data analysis was conducted by Yitaoren and Jinxiong Lin. The manuscript was written by Yitao Ren, with all authors contributing to its revision. All authors approved the final manuscript.

Funding

This work was supported by the leading academic team established by the Xining Center for Disease Control and Prevention (Document No. Ningjikyeye [2025] 21).

Data availability

Data on patients cannot be published because they contain a great deal of personal information about the patients and their families. Please contact the corresponding author directly if you need anything else.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

This study was approved by the Ethics Review Committee of the Xining Center for Disease Control and Prevention (Approval No. qhxncdcllsc—2025020) and followed relevant guidelines. Patient data, summarized at the city level, were retrospective and involved low-risk studies. All personal information involved in this article was conducted under the supervision of the relevant researchers at the Xining Municipal Center for Disease Control and Prevention, and only secondary aggregated data were used in the analysis, which did not involve participants’ names, identifying information, telephone numbers, or residential addresses; therefore, the Ethics Review Committee of the Xining Center for Disease Control and Prevention waived written informed consent.