Vital Surveillances: County-Level Hotspot Identification and Spatial Regression Analysis of Health Loss from Kashin–Beck Disease — China, 2019 and 2023

The data for this study were sourced from the National KBD Surveillance System and the National Case Surveys conducted in 2019 and 2023 (6). The system conducts epidemiological surveys in KBD-endemic regions using active case-finding methods, including household visits and physical examinations. It identifies individuals based on KBD diagnostic criteria (WS/T 207-2010), which include clinically diagnosed and imaging-confirmed cases. The dataset included the demographic information of the respondents, such as sex, age, occupation, place of residence, and disease severity. The per capita gross domestic product (GDP) was extracted from the China County Statistical Yearbook, and per capita income was derived from the same surveillance system. Data on the number of hospital beds and healthcare technicians were obtained from the National Bureau of Statistics of China. Missing values in county-level economic and healthcare indicators were addressed through multiple imputations by chained equations (MICE) (7). Specifically, this method preserves the inherent variability of the original data and, consequently, improves the accuracy and statistical efficiency of regression estimates.

We used YLDs to assess the loss of healthy life expectancy in patients with KBD due to the non-lethal nature of KBD. This measure effectively reflects the burden of non-fatal disability caused by different severities of KBD at both the individual and societal levels (8).

The YLD was calculated as follows:

where N represents the number of patients with KBD and DW denotes the disability weight. Because previous studies have not reported the disability weights for KBD and its clinical manifestations resemble those of rheumatoid arthritis (9), disability weights for rheumatoid arthritis from the Global Burden of Disease (GBD) 2019 study were used as proxies for KBD. These values were 0.117, 0.317, and 0.581 for degrees I, II, and III, respectively (10).

The ArcGIS software (version 10.8; ESRI, California, USA) was used to assess the overall spatial aggregation of YLDs among patients with KBD using global spatial autocorrelation, whereas the Getis-Ord Gi^* statistic was employed to identify local hotspots (11). The ordinary least squares (OLS) model was used to initially examine the relationships between per capita gross domestic product (GDP), per capita income, number of hospital beds, number of health technicians, and the YLD rate among patients with KBD. Multicollinearity was assessed using the variance inflation factor (VIF), with VIF values of <10 indicating acceptable collinearity. Subsequently, a geographically weighted regression (GWR) model was used to conduct localized estimations, revealing spatial non-stationarity in the relationships between each variable and the YLD rate across endemic areas and counties. These patterns were visualized using local regression coefficients. Key areas for disease management and priority interventions in adult patients with KBD have been identified (12–13).

KBD is endemic to 13 Chinese PLADs. Data during the first national surveillance in 2019 were collected from 325 endemic counties, reporting 164,914 prevalent cases, including 105,142 degree I, 46,702 degree II, and 13,070 degree III cases. By 2023, data collected from 379 endemic counties indicated 165,348 prevalent cases, including 103,969 degree I, 49,796 degree II, and 11,583 degree III cases.

Significant regional differences were observed in the YLD rates and the YLDs of KBD in 2019 and 2023. The regions with severe loss of healthy life expectancy in 2019 were the disease area counties of Yantang County (497.87 YLD/10,000, 2,074.54 YLDs) and Aba County (289.53 YLD/10,000, 1,604.27 YLDs), both are in Sichuan Province. By 2023, the regions with severe losses were Heshui County (86.02 YLD/10,000, 1,621.58 YLDs) and Ning County (44.68 YLD/10,000, 1,367.49) in Gansu Province.

Moran’s I indices for KBD-related YLDs and YLD rates were 0.11 and 0.07, respectively, in 2019, and 0.17 and 0.08, respectively, in 2023, with P<0.05 in both years. These results indicate significant spatial clustering of KBD-related health losses at the county level nationwide (Supplementary Figure S1).

The Getis-Ord-Gi^* statistic was applied to further identify hotspots. In 2019, the hotspot areas of YLD among patients with KBD were primarily concentrated in the central and western regions of China, covering 155 endemic counties across PLADs, including Henan, Shaanxi, Qinghai, Gansu, Sichuan, and Shanxi (Table 1). The hotspot areas had shifted by 2023, with concentrations in both the western and northern regions of China spanning PLADs such as Shaanxi, Qinghai, Gansu, Sichuan, Inner Mongolia, and Heilongjiang and comprising 90 endemic counties (Table 2).

The data were standardized to eliminate the effects of different units and magnitudes among the variables. An OLS model was used to assess the association between the YLD rate of patients with KBD and the different socioeconomic factors globally. All independent variables had VIF values less than 10, indicating acceptable levels of multicollinearity. In 2019, the YLD rates of patients with KBD were positively associated with per capita GDP, per capita income, and the number of hospital beds, although none of these associations were statistically significant. In 2023, the YLD rates were negatively associated with per capita GDP, the number of hospital beds, and positively associated with per capita income, with none of these associations reaching statistical significance. Furthermore, a significant negative association (P<0.05) was observed between the number of health technicians and YLD rates in both 2019 and 2023 (Table 3).

We used corrected Akaike information criterion (AICc) as the model selection metric to apply the GWR model that yielded AICc values of 769.14 and 845.17 in 2019 and 2023, respectively, whereas the OLS model produced values of 859.77 and 863.15, respectively. Thus, the GWR model provided a better fit for both years.

The GWR parameter estimation results revealed varying coefficients across different counties, further confirming the spatial heterogeneity in the determinants of YLD rates for KBD (Table 3). The OLS model results demonstrated that the number of health technicians was always negatively associated with YLD rates in both years. This negative correlation was observed in 230 counties across 10 PLADs, including Shanxi, Inner Mongolia, Jilin, Heilongjiang, Gansu, Qinghai, Sichuan, Henan, Shaanxi, and Xizang (Supplementary Table S1). Thus, there is a need to increase human resource allocation in primary healthcare to strengthen disease prevention and control systems.

In this study, we selected YLDs over single measures, such as prevalence, to more comprehensively reflect the population-level and societal burden of nonfatal disability associated with KBD. In addition, the use of YLDs provides a more in-depth evaluation of the disease burden among existing patients with KBD across the country.

The YLDs among patients with KBD exhibited significant spatial clustering at the county level. The extent of YLD hotspots declined in 2023 compared to that in 2019; however, persistent hotspots remained, primarily located in 81 endemic counties in Shaanxi, Gansu, and Sichuan in the western region and 9 counties in Heilongjiang and Inner Mongolia in the northern region. The formation of these hotspots may be associated with long-standing geographic and environmental characteristics at the district and county levels. Historically, regional geographic environments have shaped unique living and dietary patterns among local populations, influencing the incidence of KBD to varying extents over time. Consequently, substantial regional disparities in KBD prevalence have emerged, contributing to the current spatial heterogeneity in the YLD rates among patients with KBD (14).

Previous studies have reported that KBD occurs in areas with lagging economic development (15). However, we found no statistically significant correlation between YLD rates and local economic levels in both 2019 and 2023. The implementation of policies such as the National Twelfth Five-Year Plan for the Prevention and Control of Endemic Diseases and the Three-Year Special Action Plan for Tackling Endemic Diseases (2018–2020), which include relocation, targeted care, health poverty relief, and increased funding for disease control, may have significantly improved the economic conditions in KBD-endemic areas. Narrowing regional economic disparities, and increased investment in medical resources (e.g., hospital construction, medical equipment, and drug supply) in these regions have effectively alleviating uneven distribution of healthcare resources.

Despite this progress, our study identified a significant negative correlation between the number of health technicians and YLD rates for both 2019 and 2023. Thus, a shortage of specialized health personnel may limit patients’ access to adequate treatment and rehabilitation services, thereby negatively affecting their quality of life.

Therefore, targeted interventions should be introduced to identify areas with high YLD rates and insufficient healthcare personnel. These include the recruitment of professional health technicians, tailored training programs, and optimized medical resource allocation to enhance KBD prevention and treatment in underserved areas. In addition, efforts should be made to increase the supply of medications and medical equipment to primary healthcare institutions, ensure the availability of symptomatic treatments, and ultimately improve the quality of life of patients with KBD.

Future studies could include additional metrics such as functional impairment scores (e.g., WOMAC and HAQ) and quality of life measures (e.g., SF-36) among patients with KBD to enhance the framework for evaluating the health burden of KBD and provide a more comprehensive and systematic understanding of its impact.

This study had two main limitations. Firstly, no established studies have defined the disease-specific disability weights for patients with KBD. Therefore, disability weights for rheumatoid arthritis from the GBD 2019 study were used as proxies. Secondly, a small proportion (<5%) of cases were missing due to population movement, mortality, and urbanization and were therefore excluded from the spatial analysis. Although this proportion was minor, it was unlikely to affect the overall representativeness of the dataset substantially.

We extend our gratitude to the disease control and prevention centers of 13 provinces (autonomous regions and municipalities) for their assistance, and we thank the patients participating in the Kashin-Beck Disease (KBD) surveillance program for their trust and cooperation.

The study protocol was approved by the Ethics Committee of Harbin Medical University, with approval number (hrbmuecdc20221102). The research was conducted in accordance with the ethical guidelines and principles outlined in the Declaration of Helsinki.