Epidemiological analysis of wild mushroom poisoning in Zhejiang province, China, 2016–2018

Abstract Wild mushroom poisoning has been recognized as a global problem threatening human health. In this study, we aimed to explore characteristics of wild mushroom poisoning in Zhejiang province, China. From 2016 to 2018, 429 cases of wild mushroom poisoning were reported, and among them, there were 2 deaths and 84 hospitalizations, with the incidence of 0.2526 per 100,000 and the case fatality rate of 0.47%. Digestive symptoms were found in all cases. Systemic symptoms and signs, neurological symptoms, and urinary symptoms were found in 28.90% (124/429), 11.66% (50/429), and 4.90% (21/429) of the cases, respectively. The proportion of cases with incubation period <6 h was 85.78%, and those with ≥6 h accounted for 14.22%. The peak period of poisoning occurred from June to October annually. Quzhou (Moran’s I = 1.242, p < .05) and Lishui (Moran’s I = 0.759, p < .05) with mild climate, more mountains, and abundant rainfall were “hot spots” for the incidence of wild mushroom poisoning, showing a state of high‐incidence aggregation. Epidemiological analysis showed that there were seasonal, high‐incidence areas and high‐risk groups in wild mushroom poisoning. The government should give early warning to high‐incidence areas and strengthen publicity to high‐risk groups before wild mushrooms mature every summer and autumn. In addition, we recommend that ordinary people not pick wild mushrooms outdoors for consumption, because it is difficult to distinguish whether wild mushrooms are poisonous and do not buy wild mushrooms of unknown sources.


| Data sources
The data came from "Zhejiang Province Foodborne Disease Surveillance and Reporting System," which has collected sporadic cases of foodborne diseases (including infectious cases and poisoned cases) from 340 sentinel hospitals in 11 prefectural cities.
All data about the wild mushroom poisoning cases reported from January 2016 to December 2018 were included. The diagnosis of wild mushroom poisoning is based on epidemiological investigation, history of eating wild mushroom, and clinical manifestations. The address data of 429 cases were transformed to geographic spatial data based on the Baidu geocoding API (http://lbsyun.baidu.com/index.

| Descriptive analysis
The following parameters were extracted (if available) from the charts of included patients: age, sex, hospitalization, occupation, home address, time of onset, time of visit, time of death, main symptoms and signs, method of acquiring poisonous mushroom, type of dining place, and time of eating. Data were analyzed descriptively using the software of GraphPad Prism version 9.2.0 (GraphPad Software, 2021).

| Production of spatial distribution map
We used ArcGis10.2 software to connect case database (including geocoding, number of cases, incidence, etc.), geographical information database (including geocoding, dimension and longitude) with population information database (including geocoding, population) through geocoding connection, established cases geographic information database in Zhejiang province, and then made a space distribution.

| Spatial autocorrelation analysis
Spatial autocorrelation is a measure of the degree to which one thing or phenomenon is similar to other nearby things or phenomena, which includes global autocorrelation and local autocorrelation.
Global autocorrelation describes the degree of association between spatial objects in the whole research area to indicate whether there is a significant spatial distribution pattern between spatial objects. A well-known spatial statistic for measuring spatial autocorrelation is Moran's I (Moran, 1948). It was defined as follows: where N is the number of regions, x i is the value of attribute X (e.g., incidence of mushroom poisoning) at region i, μ is the mean value of X, and w ij is a matrix indicating the spatial weight between regions i and j. S 0 is the sum of the weight calculated by: Moran's I value is between −1 and +1, and the value is positive, indicating a positive correlation with the data. The closer the value is to 1, the stronger the positive spatial correlation is, and the case occurrence space presents an aggregation distribution. Moran's I was negative, and the data were negatively correlated. The closer to −1, the more scattered the cases were, and the greater the difference between samples was. Moran's I is 0, indicating that cases occur randomly in space. However, the global autocorrelation only reflects the spatial correlation of the research target. To understand the correlation (positive and negative, as well as the degree of correlation) between a district and surrounding counties, we need to use the local Moran's I coefficient. Local autocorrelation analysis is used to test the probability level of local spatial aggregation around each observation unit. When there is no global spatial autocorrelation, local autocorrelation analysis is used to find the location of local spatial autocorrelation that may be masked. When there is global spatial autocorrelation, local autocorrelation analysis is used to investigate whether there is spatial heterogeneity.
To explore the possible reasons behind the spatial patterns of mushroom poisoning, we calculated the global Moran's I and local Moran's I of the incidence of mushroom poisoning. ArcGIS software was used for the analysis. Weight matrix W was defined by the "contiguity" method, which only includes neighboring polygons for computation.

| Regional distribution
The highest number of cases of mushroom poisoning occurred in Jinhua, followed by Quzhou and Lishui. The highest incidence was reported in Quzhou and the lowest in Zhoushan ( Table 2). The counties with higher incidence were Songyang county (5.4234/100,000), Changshan county (5.3149/100,000), and Pujiang county (2.1422/ 100,000) ( Figure 1).

| Time distribution
The occurrence of mushroom poisoning had obvious seasonality. Mushroom poisoning occurred most frequently from June to October, during which 373 cases occurred, accounting for 86.9% of the total cases ( Figure 2).

| Population distribution
Poisoning cases ranged in age from 1 to 88 years, with a median age of 46 years. Cases aged 26-60 years accounted for the largest proportion (68.30%). The occupation with the highest incidence rate was farmers (48.72%), followed by migrant workers (13.52%) and workers (7.23%). We found that 37 (8.62%, 37/429) poisoning cases were from other provinces in China, and the rest were residents of Zhejiang province.

| Distribution of poisoning places
Most cases of mushroom poisoning occurred in households (91.61%), and the main sources of mushrooms were picking (93.94%) or buying (6.06%) wild mushrooms (Table 3). A small amount of mushroom poisoning occurred in collective canteens (5.59%), and the poisonous mushrooms mainly came from picking (62.5%) or buying (29.17%) wild mushrooms.

| Global spatial autocorrelation analysis
The global Moran's I of average incidence of mushroom poisoning was 0.337 (z = 2.37, p = .02 < .05), suggesting that the incidence distribution of wild mushroom poisoning in Zhejiang province during 2016-2018 was not random, but had a statistically significant spatial positive correlation, indicating a significant aggregation distribution (

| Local spatial autocorrelation analysis
There are 11 prefecture-level cities in Zhejiang province. The results of local autocorrelation analysis are shown in Figure 3 and Table 5.
The red area represents the high-high spatial correlation model, which is mainly distributed in Quzhou and Lishui. The area with high incidence of wild mushroom poisoning is surrounded by high incidence, showing a state of high value aggregation (p < .05). It was a "hot spot" region for the incidence of wild mushroom poisoning in Zhejiang province from 2016 to 2018. The blue area represented the low-low spatial correlation model, which is mainly distributed in Jiaxing and Ningbo. The area with low incidence of wild mushroom poisoning was surrounded by low incidence, showing a state of low value aggregation (p < .05). It was a "cold spot" region for the Moran's I showed that not only the high-incidence area of poisonous mushroom should arouse the concern, but also the neighborhood area of the high value area should be paid more attention as potential high-incidence areas.

| D ISCUSS I ON
The incidence of mushroom poisoning varies worldwide. The incidence of mushroom poisoning is higher in countries where wild mushrooms have traditionally played an important role in food, such as Russia, Eastern Europe, and China (Govorushko et al., 2019). Yunnan and Guizhou in China had the highest mortality rates (Sun, Li, et al., 2018). Asian countries such as Nepal (Adhikari et al., 2005), Thailand (Parnmen et al., 2016), and Iran (Soltaninejad, 2018) also have high mortality rates. By contrast, deaths induced by mushrooms are rarely reported in the United States (Kintziger et al., 2011), Canada (Fleury et al., 2008), and Australia (Roberts et al., 2013), probably due to better health care.
Although the incidence and mortality of mushroom poisoning in Zhejiang province were lower than those in some other provinces of China, such as Hunan (Liang et al., 2018), Yunnan (Wang et al., 2016), and Guangxi (Jiang et al., 2018), mushroom poisoning has become the main cause of death from foodborne disease outbreaks (Sun, Chen, et al., 2018), which does require more attention from relevant departments. One of the reasons for the high mortality rate of mushroom poisoning in China is that people in mountainous areas lack the ability to identify highly toxic mushrooms. On the other hand, many hospitals lack experience in the treatment of highly toxic mushrooms and often miss the best treatment period. Therefore, it is very important to strengthen the publicity of mushroom poisoning and the training of doctors. In addition, good communication between mushroom experts, epidemiologists, and doctors should be established to save cases of highly toxic mushroom poisoning.
Since the incubation period after consumption of wild mushrooms can provide useful information for clinicians to understand the severity and prognosis of wild mushroom poisoning, we collected the incubation period for all cases. In our study, a total of 85.78% of poisoning cases belonged to the early type (<6 h) and the prognosis was good. The incubation periods of the two deaths were 6 and 9 h, respectively, and their initial symptoms were gastrointestinal symptoms, but they later died of liver and kidney failure.
Findings showed that the incidence of mushroom poisoning in Zhejiang province was high from June to October, indicating that prevention and control should be strengthened in summer and autumn, and an early warning should be issued before poisonous mushrooms mature. The incidence of mushroom poisoning in Zhejiang Notes: 1, high-high; 2, low-low; 3, low-high.