1. Epidemic situation of human brucellosis
(1) Time distribution
During the study period of 2007 to 2016, there were 399,578 reported cases of human brucellosis in mainland China. Annual case totals ranged from 19721 in 2007 to 57222 in 2014, with a mean number of approximately 34498 (95% CI: 25610, 43386) cases reported per year. The annual mean incidence of brucellosis was 2.97/100,000, with a low of 1.50/100,000 in 2007 and a high of 4.22/100,000 in 2014. As a whole, annual incidences follow a growth trend, with a linear relationship of y=0.0211x+1.7003 from 2007 to 2016 (Figure 1). Furthermore, annual and monthly incidences fluctuated during this 10-year period. Between 2007 and 2009, the annual average incidence of brucellosis in China demonstrated an upward trend. Following that, a short-term decline occurred in 2010, after which incidences of the disease began to rise again, continuing to rise until 2014, and then decline again in 2015 and 2016. A purely temporal analysis between 2007 and 2016 demonstrated that high-incidence seasons of human brucellosis were the time between late spring and early summer, while the incidence was highest in May.
(The histogram shows monthly incidences and the dots show annual incidences of brucellosis. Annual incidences have a y=0.0211x+1.7003 relationship. Results of a purely temporal analysis between 2007 and 2016 demonstrated that high-incidence seasons of human brucellosis occurred between late spring and early summer, with incidences highest in May.)
(2) Spatial distribution
Incidences of human brucellosis in each province vary significantly, with an interquartile range of 4.25/100,000 (0.045~4.297/100,000) and a range of 70.10/100,000 (0~70.10/100,000). We defined regions with an incidence greater than 10/100,000 as high incidence areas, while incidences lower than 0.999/100,000 were defined as low incidence areas. Over the total study period, high incidence areas increased from two to five, with the number of high incidence provinces and cities increasing year by year. In 2007, there were six provinces (6/31, 19.35%) with no cases of human brucellosis. These were mainly located in central and southern regions such as Guizhou, Hunan, Jiangxi and Hainan. As time passed, the epidemic area gradually expanded into all provinces and cities in China. The spatial and temporal distribution map (Figure 2) shows that 31 provinces (municipalities and autonomous regions) had incidences between 2007 and 2016. The epidemic mainly occurred in the north of China, such as the Inner Mongolia and Shanxi provinces. After 2009, the high incidence area grandly expanded in other northern areas, such as the Heilongjiang, Xinjiang and Ningxia provinces. In Inner Mongolia, disease incidence was the highest in history, exceeding the 50/100,000 population between 2009 and 2011. Between 2014 and 2016, incidences increased considerably in areas which had been historically low areas such as Gansu, Qinghai and Shandong. This was most marked in Ningxia, where incidences increased sharply from 0.4636/100,000 in 2007 to 43.6557/100,000 in 2015.
2.Three-dimensional trend analysis
The overall three-dimensional trend for 2007-2016 indicated an uneven distribution of human brucellosis in mainland China (Figure 3). The curve on the XZ plane represents the trend of East-West direction, while the curve on the YZ plane represents the change of North-South direction. Each point represents the projection of incidence. It can be seen that incidences of human brucellosis in the northern region were higher than in the southern region. However, in the east-west direction, incidences of brucellosis were evenly distributed in mainland China..
3.Gravity center migration
Most cases of human brucellosis were reported in northern China during 2007 to 2016, while most provinces in northern China experienced a serious epidemic (Figure 4). Between 2007 and 2012, the gravity-center of brucellosis was concentrated in the central area of Inner Mongolia. After 2012, there was a trend of gradual migration to the southwest. However, the distance and direction of annual migration were not obvious. Results showed that brucellosis was still prevalent in the northern part of China. During the study period, the gravity-center of brucellosis migrated 906.43 km southwest. The distance and direction of annual migration can be seen in Table 1.
Table 1. Distance and direction of center of gravity migration of brucellosis between 2007 and 2016.
Time
|
Distance
|
Direction
|
Time
|
Distance
|
Direction
|
2007-2008
|
69.89km
|
northeast
|
2008-2009
|
53.39 km
|
northeast
|
2009-2010
|
49.95km
|
northwest
|
2010-2011
|
140.28km
|
southwest
|
2011-2012
|
31.91 km
|
southwest
|
2012-2013
|
281.90km
|
southwest
|
2013-2014
|
344.76km
|
southwest
|
2014-2015
|
119.16km
|
southwest
|
2015-2016
|
80.72km
|
northwest
|
|
|
|
4. Spatial autocorrelation analysis
(1) Global spatial autocorrelation
Using provincial units to carry out the global autocorrelation analysis, we obtained Moran’s I value, variance, Z score and P value from 2007 to 2016, respectively (see Table 2). The values of Moran’s I were 0.1179 and 0.1181 respectively for 2013 and 2014, while Z values were greater than 1.96 (all P < 0.05), indicating that incidences of brucellosis in China between 2013 to 2014 had a non-random distribution, while there was no spatial autocorrelation in either 2007-2012 or 2015-2016 (Z < 1.96, P >0.05, meaning there was no clustering trend.
Table 2. Global autocorrelation of Moran’s I values of brucellosis in mainland China from 2007 to 2016.
Year
|
Moran’s I
|
Variance
|
Z Score
|
P Value
|
Aggregation
|
2007
|
0.041785
|
0.002682
|
1.450621
|
0.146886
|
NO
|
2008
|
0.051605
|
0.002669
|
1.644175
|
0.100140
|
NO
|
2009
|
0.043654
|
0.002241
|
1.626341
|
0.103877
|
NO
|
2010
|
0.040263
|
0.002109
|
1.602407
|
0.109066
|
NO
|
2011
|
0.043491
|
0.002197
|
1.638966
|
0.101220
|
NO
|
2012
|
0.076004
|
0.003579
|
1.827598
|
0.067610
|
NO
|
2013
|
0.117938
|
0.004963
|
2.147162
|
0.031780
|
YES
|
2014
|
0.118131
|
0.005379
|
2.065198
|
0.038904
|
YES
|
2015
|
0.093595
|
0.005191
|
1.761692
|
0.078121
|
NO
|
2016
|
0.092068
|
0.005165
|
1.744833
|
0.081014
|
NO
|
(2) Local spatial autocorrelation
While global clustering indices evaluate the distribution of events, they do not specify the location of hotspots. To identify areas characterized by a higher rate of human brucellosis incidences, a local spatial autocorrelation analysis was carried out for the years which had a statistical significance of Moran's I value, and a local spatial autocorrelation aggregation distribution map was generated (Figure 5). There were no High-High, High-Low or Low-High cluster areas in China in 2013, while Low-Low cluster areas tended to be concentrated in the southern provinces of China, including the provinces of Yunnan, Guizhou, Guangxi, Hunan, Guangdong, Hubei, Jiangxi, Fujian, Zhejiang and Anhui, with an average incidence of 0.08/100,000. The new High-Low cluster area in 2014 was Xinjiang, while the Low-Low cluster area decreased compared to 2013, now including Yunnan, Guizhou, Guangxi, Hunan, Guangdong, Jiangxi, Fujian, Zhejiang and Anhui provinces. The mean incidence of Low-Low clusters was 0.14/100,000 in 2014, meaning the incidence of Low-Low clusters in 2014 was higher than in 2013. Incidences in the Hubei Province increased rapidly from 0.08/100,000 in 2013 to 0.34/100,000 in 2014, thus reducing Hubei Province in Low-Low cluster areas in 2014. Cluster maps for 2013 and 2014 demonstrate that the cluster area in China was relatively stable, and that compared to the spatial distribution map of incidence, the Low-Low cluster area in each year tends to be the same as the low incidence area, all of which are located in the south of China.