The public health emergency management system in China: trends from 2002 to 2012

The survey methods have previously been published [14]. Briefly, we conducted two rounds of cross-sectional surveys in 2006 and 2013. The two surveys were retrospective and selected the same agencies in the two rounds. The survey of 2006 collected the data from 2002 to 2005, and the survey of 2013 collected data of 2012. We conducted a multistage sampling to select CDCs at different administration levels, selected all 32 provincial CDCs and used systematic random sampling to select municipal and county CDCs. As governmental funding is the most critical control point of public health emergency management for the CDCs [15],we used “governmental funding to CDCs per thousand people” as a basis to determine sample size [16]. A sample size of 123 municipal and 457 county CDCs was calculated based on the following formula [17].where n is the number of the minimal sample size; αis the probability of type I error, and β is the probability of type II error, here α = 0.05,β = 0.05; u_αand u_βare standard normal distribution values corresponding to α and β respectively;σis the population standard deviation, hereσ = 404.3 yuan; δ is the allowable error. For municipal CDCs, δ = 54.9yuan, σ = 210.0 yuan. For county-level CDCs, δ = 62.5yuan, σ = 404.3yuan (1 U.S. dollar = 6.6 yuan).

The municipal and county level CDCs were all selected through random sampling. The sampling process was conducted based on the national standard coding (GB coding, the corresponding administrative regional code which is unique for each city or county [14]). We used a computer-generated random number to identify the first institution, and then selected every third municipal CDC and every sixth county level CDC. Finally, we selected 32 provincial CDCs, 139 municipal CDCs, and 489 county CDCs.

The study was approved by the former Ministry of Health (MOH) in China and reviewed by the Medical Research Ethics Committee at the School of Public Health of Fudan University.

We selected twenty-one indicators associated with the PHEMS from the National Assessment Criteria for CDC Performance. Based on the crisis management theory which was commonly used in the field of public emergency management [18, 19], the whole process was divided into four stages including preparation, readiness, response and recovery [20]. According to the framework, we grouped the indicators into 4 stages and 13 capabilities. Table 1 showed the features, units and measurements of these indicators.

According to the National Regulations on Public Health Emergency Management [21], each sampled CDC graded five public health emergencies handled in the year before the survey with the full mark of 10 points for each indicator; at CDCs where the total numbers of handled public health emergencies were fewer than five, all public health emergencies were graded instead.

The Bureau of Disease Prevention and Control of the former MOH approved and organized two rounds of field surveys, and 32 provincial Health Departments coordinated data collection.

A pilot survey was conducted to ensure validity and reliability. After receiving uniform training from the MOH, the provincial quality supervisors trained investigators from sampled CDCs in their corresponding provinces. The investigators collected relevant data from sampled CDCs and submitted the completed questionnaires to their provincial quality supervisors via e-mail or CD-ROM. Simultaneously, paper copies with official stamps were submitted.

The second round of survey data were obtained from National Disease Control and Prevention Performance Evaluation Platform. The quality control process was set up and carried out by the platform with backend logic judgments and audit procedures.

As the final step of quality control in both surveys, research group rechecked data and contacted CDCs with abnormal or absent values via email or phone. Finally, the overall response rate was 95.8% in 2002 and 99.5% in 2012.

We established a dataset using Excel 2013(Microsoft Redmond WA). We only used the data of the year 2002 and 2012 for analysis. After data cleaning and sorting, descriptive analysis and statistical tests were performed using SPSS 21.0 (IBM SPSS, Chicago, IL, USA). We used McNemar’s test to test differences in proportions and paired sample t test to test differences in means between 2002 and 2012. Since noticeable differences existed between China’s regions, the division of regions was based on the 2003 Chinese Economics Yearbook and the First National Economic Census.

Establishing organization comprised building an emergency response office and forming a leadership group and an expert panel. The average percentage of CDCs with an emergency response office was 61.6% in 2002 and 95.0% in 2012. The average percentages with a leadership group and an expert panel were 47.9% and 78.6% in 2002 and 95.7% and 96.8% in 2012, respectively. Similar trends also occurred across different levels and regions (Table 2).

The capability for building mechanisms in terms of information sharing and on-site treatment increased by 93.5% and 89.4%, respectively. Increasing by 127.5%, response-material deployment mechanism gained the highest growth rate. Municipal CDCs had the highest percentages, followed by provincial and county CDCs. The central region not only had the highest percentages, but also experienced the highest growth rate.

Average number of emergency response personnel per CDC increased from 15 in 2002 to 31 in 2012, which was significant. In 2012, provincial CDCs had the highest number of personnel (n = 92), followed by municipal (n = 47) and county (n = 22) CDCs. Moreover, the average number decreased from eastern (n = 35) to western regions (n = 29) (Table 3).

The percentage of fully stockpiling emergency resources significantly increased from 16.7% in 2002 to 41.2% in 2012. Provincial CDCs had the highest percentage (74.2%) in 2012 and increased by 102.2%, whereas county CDCs had the lowest percentage (34.5%) in 2012 and increased by 141.3%. Nevertheless, the average percentage at each administrative level did not meet the corresponding performance assessment criteria. Average percentages of fully stockpiling emergency resources decreased from eastern (56.7%) to western (31.7%) regions.

The mean percentage of formulating emergency plan increased from 40.6% in 2002 to 89.9% in 2012, statistically significantly increasing by 121.4%. Provincial CDCs had the highest percentage (93.5%) in 2012, and the difference between municipal (89.1%) and county CDCs (89.9%) was not significant. CDCs in central region had the highest percentage (92.5%), followed by western (89.2%) and eastern (86.0%) regions (Table 2).

The average length of emergency response training increased from 9.7 days per person in 2002 to 14.6 days per person in 2012; however, this 50.5% increase was not statistically significant. Provincial CDCs had the highest average length of response training (44.3 days per person), followed by municipal and county CDCs (Table 3).

Comparing the statistics in 2002 and 2012, the average times of exercises did not change with statistical significance. In 2012, county CDCs had higher average times of exercises than did municipal (1.7) and provincial (1.5) CDCs; nevertheless, only provincial CDCs had increased average times of exercises during the past decade. From regional perspective, the average times of exercises decreased from western (2.7) to eastern (1.8) regions (Table 3).

There were 63.7% and 23.0% of disease surveillances conducted per month and per week in 2012, respectively. Compared with statistics in 2002, frequencies of daily, weekly, and monthly surveillance analysis increased, among which weekly surveillance analysis increased with statistical significance. Meanwhile, the frequencies of disease surveillance analysis per ten days, quarter, and year decreased with statistical significance (Table 2).

According to “contingency rules of paroxysmal public health events”, public health emergency events are classified into four levels (I, II, III and IV), with severity decreasing from Level I to Level IV. In 2012, there were 3092 public health emergencies directly reported via the Disease Surveillance Information Management System, which accounted for 98.8%.The percentage of timely reporting by county CDCs emergency levels in 2012 was presented in Table 4. Moreover, the average scores for indicators of coping capability were high in 2012 (Table 4).

The average scores for capabilities at recovery stage were lower than those for capabilities at response stage. The average score for data archiving was 8.33, then followed by those for data analyzing (5.83) and concluding (5.69) (Table 5).

The available assessment indicators are relatively narrower in comparison with those such as the Capability Assessment for Readiness and the Target Capabilities List of Homeland Security Exercise and Evaluation Program in the United States.

Nearly half the indicators were binary (“yes” or “no”), so the quality of policy implementation and accountability could not be judged.

Although logic judgments and audit procedures were conducted, recall bias may still exist. Despite these limitations, the main contribution of this paper are the findings based on the data from two rounds of national field surveys conducted in 2002 to 2012 in China. We believe that this contribution is theoretically and practically relevant because the lessons China’s government learned from the 2003 SARS outbreak provide an emergency response framework that can be employed by developing countries.