Efficiency and productivity of county-level CDCs in Qingdao municipal, China: a retrospective study based on 6 years of panel data

Abstract

Background

County-level Centers for Disease Control and Prevention (CDCs) are the foundational units of China’s public health system. The COVID-19 pandemic has highlighted the shortcomings of the system, making it urgent to strengthen the functions of county-level CDCs. The efficiency and equity of resource allocation are crucial for enhancing public health outcomes. Therefore, this study aimed to evaluate six-year trends (2018–2023) in equity and efficiency of resource allocation among the ten county-level CDCs in Qingdao and to identify actionable measures for optimizing grassroots public-health capacity in the post-COVID-19 era.

Methods

We used Gini coefficient and Lorenz curve to assess the equity of resource allocation in the county-level CDCs in Qingdao. The efficiency of CDCs was evaluated using data envelopment analysis (DEA) and Bootstrap DEA. Efficiency change was analyzed by employing the Malmquist productivity index (MPI).

Results

The health resources at the Qingdao CDCs showed an overall upward trend from 2018 to 2023. During 2018–2023, CDC personnel allocation showed good equity in demographic and economic dimensions (Gini < 0.3), but suboptimal equity in geographic distribution (Gini 0.3–0.5). The average technical efficiency was 0.940, pure technical efficiency averaged 0.995, and scale efficiency was 0.944. Among 10 CDCs, 50% achieved DEA efficiency. According to the total factor productivity (TFP) reflected by the indicator system constructed in this study, TFP grew at a rate of 6.0% between 2022 and 2023. Changes in technical efficiency change (TEC) and scale efficiency change (SEC) generally exceeded those in technological progress (TC).

Conclusion

In order to enhance the equity and efficiency of resource allocation in county-level CDCs in Qingdao City. The government needs to adopt targeted policies, such as implementing flexible staffing adjustments, increasing the number of technical personnel and senior experts, and strengthening continuous training to enhance workforce capabilities. Also, fiscal efficiency should be enhanced by linking fund allocation with performance evaluation. Meanwhile, accelerating technological modernization and enhancing technological capabilities through digital tools and management innovation are crucial. Overall, these comprehensive measures aim to enhance resource utilization efficiency and strengthen the resilience of public health services.

Clinical trial number

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-025-13478-y.

Background

China’s disease prevention and control system serves as a crucial safeguard for protecting public health, ensuring national health security, and maintaining socioeconomic stability, forming a necessary component of the nation’s health initiatives. On December 25, 2023, China issued the “Guidance on Promoting High-Quality Development of Disease Prevention and Control” [1]. This directive outlined specific requirements for strengthening the core functions of Centers for Disease Control and Prevention (CDC) at all levels, comprehensively enhancing their professional capabilities, fostering talent development, and establishing stable investment mechanisms.

The current professional prevention and control system, comprising national, provincial, municipal, and county-level CDCs, plays a significant role in responding to major epidemics, preventing and controlling infectious and chronic diseases, managing health emergencies, and providing public health services. Within this framework, county-level CDCs are foundational, acting as the leading institution of the grassroots “county, township, village” prevention and control network. They function as a critical nexus for communication, both top-down and bottom-up, and bear the significant responsibility of guiding primary healthcare institutions in fulfilling their public health mandates. In comparison to provincial and municipal CDC centers, county-level CDCs are more focused on the executing and implementing specific tasks, while also undertaking certain supervisory duties.

“The Basic Responsibilities of Centers for Disease Control and Prevention at Various Levels (2008)” [2] defines 266 specific tasks grouped into 7 core responsibility categories. However, the distribution of these tasks is uneven across administrative levels: provincial-level CDCs are assigned 244 tasks, municipal-level CDCs are assigned 236, and county-level CDCs are assigned only 21. Furthermore, the practical implementation of these guidelines regarding the division of responsibilities and functional levels among CDCs has been suboptimal. Consequently, a disproportionate operational burden has fallen on county-level CDCs. This creates a significant imbalance, as these grassroots institutions typically have the smallest staff numbers, lowest technical capabilities, poorest equipment configurations, and least stable funding [3]. This mismatch places immense pressure on county-level CDCs, presenting substantial challenges to their development and operations.

Efficiency and equity are crucial elements in health resource allocation [4]. As the primary-level units in the public health domain, the resource allocation status of county-level CDCs directly influences the advancement and development of public health initiatives. Equitable resource allocation is a prerequisite for enhancing the utilization efficiency of health resources [5]. Thus, health resource distribution strategies must aim to improve efficiency while maintaining equity, striving for a balance between these two crucial elements [6]. Previous studies have evaluated the equity of human resource allocation in China’s provincial level CDCs through population-based and geography-based measures [7–9]. However, solely analyzing equity remains insufficient to guide rational health resource allocation. Based on established equity foundations, further structural adjustments to public health resource distribution are necessary to optimize service efficiency [10].

According to data released by the China’s National Health Commission, the total number of county-level CDCs nationwide exceeded 2800 by 2023. Nevertheless, these institutions exhibit varying degrees of deficiencies and imbalances in human, material, and financial resources [11]. The COVID-19 pandemic has highlighted the urgency of strengthening the CDC system. While the response to the pandemic demonstrated the dedication and significant institutional advantages of China’s CDC network, it also exposed numerous deficiencies. In the post-pandemic era, addressing the systemic gaps that constrain the functionality of the CDC system is an urgent priority [12]. Consequently, in-depth research on the equity and efficiency enhancement in county-level CDCs is imperative.

EVANS & ETIENNE indicate that globally, approximately 20%-40% of health systems experience operational inefficiencies, manifested as suboptimal resource allocation that impedes achieving universal health coverage [13]. Previous international studies have conducted cross-county comparisons of health system resource allocation efficiency [14, 15]. Other scholars have analyzed the health resource allocation efficiency in their respective countries and proposed targeted strategies based on their national contexts. Ngobeni and colleagues [16] evaluated the technical efficiency of nine provincial health departments in South Africa using Data Envelopment Analysis (DEA), suggesting that rational resource allocation significantly enhances public healthcare service delivery. However, specific efficiency evaluations for county-level CDCs remain scarce. In China, Liu Chunyan’s team [17] developed an indicator system comprising (3 input, 7 output indicators) and analyzed data from 13 prefecture-level CDCs in Heilongjiang Province. Chengyue Li’s team [18] following expert validation, established an input-output indicator set and evaluated the efficiency of several county-level CDCs using DEA. Scholars have extended health systems research to equity dimensions, with empirical investigations encompassing: (a) resource distribution patterns in maternal and child health facilities [19] (b) service accessibility in traditional Chinese medicine hospitals [20] and (c) regional equity assessments [21]. However, existing studies predominantly rely on pre-pandemic data, failing to capture contemporary efficiency dynamics within CDCs—particularly the transformative shifts triggered by the COVID-19 pandemic. Furthermore, current literature exhibits a critical gap in integrated equity-efficiency analyses of county-level CDCs, with most studies examining these dimensions in isolation.

Efficiency serves as a key indicator reflecting workload and resource allocation, and is one of the critical factors influencing both social and economic benefits in healthcare institutions. Efficiency is categorized into two types: technical efficiency and allocative efficiency. Technical efficiency is achieved when resource allocation maximizes output for a given input level or minimizes input for a given output level. Allocative efficiency is attained when resource allocation maximizes output for a given cost of inputs or minimizes input costs for a given output level [22]. Previous studies have employed the DEA model for efficiency measurement, which is a non-parametric method widely used to evaluate the relative efficiency of decision-making units that utilize multiple inputs to produce multiple outputs [23]. Compared to stochastic frontier models, the DEA model has lower data requirements and can be applied when data is scarce or parametric models are difficult to implement. It also effectively handles undesirable outputs. Furthermore, DEA is particularly suitable for evaluating the efficiency of complex systems with multiple inputs and outputs, without requiring the predefinition of a production function form [24]. Since the CDC provides multiple health services, the DEA would be more suitable for this study. However, the relative efficiency scores computed by conventional DEA models exhibit systematic upward bias relative to true efficiency levels and remain vulnerable to environmental heterogeneity and stochastic disturbances. To rectify these methodological deficiencies, the bootstrap method has been formally incorporated into the estimation framework [25].

This study focuses on Qingdao City in Shandong Province as the study area. Qingdao is a major municipal-level city with a GDP of approximately 1.57 trillion yuan in 2023 (ranking eighth among China’s 293 prefecture-level cities) and a permanent population of about 10.43 million (ranking 12th nationally), with an urbanization rate of 78% (ranking 3rd nationally). Its economic development and demographic profile make it a representative case for examining how economically advanced regions can enhance public health system efficiency while maintaining equity.

By systematically analyzing the equity of resource allocation and the efficiency of the county-level CDCs with DEA model in Qingdao from 2018 to 2023, this research aims to provide a pivotal case reference and an evidence-based foundation for optimizing resource allocation efficiency specifically at the county level within China’s CDC system. Its core objectives are to: (1) Quantitatively assess the dynamic interplay between equity (using Gini coefficient and Lorenz curve) and efficiency (using Bootstrap-DEA and Malmquist index) over a critical six-year span; (2) Identify key drivers and impediments to efficiency (3) Propose targeted strategies to enhance the performance and resilience of the foundational tier of China’s public health defense system. The study’s significant innovation lies in: (a) Its comprehensive integration of equity and efficiency analyses for county-level CDCs, addressing a critical gap in existing literature which often examines these dimensions in isolation; (b) The application of Bootstrap-DEA methodology to correct inherent biases in traditional DEA estimates, yielding more robust and reliable efficiency scores; (c) The temporal scope encompassing the pre-pandemic, peak-pandemic, and post-peak transition phases (2018–2023), uniquely capturing the profound influence of a major public health emergency on system performance – insights currently lacking in pre-dominantly pre-pandemic studies; (d) Focusing on a nationally representative, economically advanced urban setting (Qingdao), offering valuable lessons applicable to similar regions while highlighting context-specific challenges. Therefore, this research seeks to make an important complement to existing knowledge by exploring these dynamic changes, with the overarching goal of identifying key challenges and actionable opportunities for strengthening the performance and sustainability of China’s crucial grassroots public health institutions. The findings hold substantial practical significance for policymakers and CDC administrators in refining resource allocation structures, implementing evidence-based reforms, and ultimately enhancing the capacity of county-level CDCs to fulfill their vital public health functions effectively.

Method

Study data

Primary data were collected using researcher-designed structured questionnaires (Supplementary File S1) administered to 10 county-level CDCs in Qingdao, which captured three key dimensions: institutional characteristics, resource allocation metrics, and service delivery outcomes. This three-dimensional framework enables a comprehensive evaluation of CDC operational efficiency. The survey covered a six-year period, from 2018 to 2023, enabling longitudinal analysis of trends.

Selection of input and output indicators

Accurate measurement of CDC efficiency requires selecting of a suitable and comprehensive set of input and output variables. For this study, input variables were chosen to represent three critical dimensions of resource allocation: human, material, and financial resources, specifically including: Number of CDC staff per 10,000 population, per capita fiscal allocation for CDC, and number of Class A laboratory equipment units per 10,000 population. Based on the core functions of the CDC, output variables selected include childhood immunization program vaccination coverage rate, premature mortality rate from major chronic diseases, and population health literacy level.

Data envelopment analysis (DEA)

DEA is a non-parametric statistical method used to evaluate the relative efficiency of multiple Decision-Making Units (DMUs) that utilize multiple inputs and outputs [23]. The methodology primarily includes two classic radial models [26]: the Charnes-Cooper-Rhodes (CCR) model, which assumes constant returns to scale (CRS) and measures Technical Efficiency (TE) to assess the overall efficiency by evaluating the proportional relationship between input increases and output expansions, and the Banker-Charnes-Cooper (BCC) model, which relaxes the CRS assumption to variable returns to scale (VRS) and isolates Pure Technical Efficiency (PTE) by excluding scale effects. The relationship among these efficiencies is defined as TE = PTE × Scale Efficiency (SE), where TE reflects overall technical and scale performance, PTE captures managerial or technological effectiveness, and SE indicates whether a DMU operates at the optimal production scale. DEA efficiency evaluation thus encompasses three dimensions: technical efficiency (TE), pure technical efficiency (PTE), and scale efficiency (SE). In healthcare resource assessment, DEA identifies inefficiencies in staffing, funding, or equipment allocation by benchmarking DMUs against an empirically derived efficient frontier, thereby providing actionable insights for performance optimization. Given that the counties included in this study are purposively sampled for their relatively high economic development and established health equity systems – contexts where maximizing public health impact within fixed resource constraints is a key management objective – an output-oriented model was selected.

The Bootstrap-DEA method improves upon the traditional DEA approach by generating numerous pseudo-samples through repeated resampling. It computes DEA efficiency scores for these samples and constructs an empirical distribution via repeated sampling. This empirical distribution approximates the true sampling distribution of DEA efficiency values, thereby correcting bias inherent in conventional DEA estimates. Therefore, we applied the bootstrap method with 1,000 replications to derive bias-corrected results [27].

To analyze the efficiency changes of DMUs, this study additionally employs the DEA-Malmquist model [28]which integrates the DEA framework with the Malmquist index to assess efficiency variations between period t and t + 1, addressing the limitation of conventional DEA models that typically evaluate cross-sectional data. The Malmquist index measures Total Factor Productivity (TFP) changes over time, which decomposes into technical change (TC) and technical efficiency change (TEC). Specifically, a Malmquist productivity index greater than 1 indicates efficiency improvement during the observed period, equal to 1 signifies no change, and less than 1 reflects efficiency decline. Further decomposition of TEC reveals contributions from pure technical efficiency change (PTEC) and scale efficiency change (SEC), aligning with the BCC model’s framework: TFP =  TC×PTEC × SEC.

Gini coefficient and Lorenz curve

The Lorenz curve, proposed by Max Otto Lorenz in 1905, is a graphical representation of resource distribution. A greater curvature of the Lorenz curve indicates more unequal distribution, whereas a flatter curve reflects greater equality [29]. The Gini coefficient, developed by Corrado Gini in 1912 [30], is calculated based on the Lorenz curve, and provides a quantitative measure of inequality. It ranges from 0 to 1, a Gini coefficient of < 0.2 is highly equitable, of 0.2–0.3 is relatively equitable, of 0.3–0.4 is moderately equitable, and a Gini coefficient of > 0.4 is inequitable [19]. The calculation formula is as follows:

where n is the total number of units, Xi is the cumulative percentage of the population, economy and geographical area; Yi is the cumulative percentage of staff in CDCs.

Descriptive statistical analysis of the input and output variables was conducted using R version 4.4.0 and RStudio software. Equity analyses, including the calculation of Gini coefficients and the generation of Lorenz curves, were conducted using Microsoft Excel 2019. CDC efficiency metrics (TE, PTE, SE) and the Malmquist productivity indices (TFP, TC, TEC, PTEC, SEC) were calculated using the DEAP (Data Envelopment Analysis Program) analytical software, Version 2.1.

Result

Descriptive statistical analysis

Table 1 presents an overview of the ten county-level CDCs in Qingdao from 2018 to 2023, while Table 2 displays their average input-output data. From 2018 to 2023, the number of on-duty staff, number of personnel with mid-level and senior professional titles, number of health technicians, number of personnel with bachelor’s degrees or higher, and staffing utilization rate all increased. Per capita fiscal allocation for CDC increased substantially during the pandemic period, reaching 74.56 million yuan in 2022, followed by a slight decrease in 2023. Annual per capita expenditure for on-duty personnel showed an overall upward trend, reaching 221,000 yuan in 2023 -representing a 21.98% increase compared to the 2018 level. Annual per capita income experienced modest growth amid overall fluctuations, standing at 148,000 yuan in 2023. Additionally, the number of laboratory equipment valued and the completion rate of Class A laboratory testing items both demonstrated an upward trend overall.

Table 1.

Basic characteristics of the 10 CDCs in Qingdao (2018–2023)

Category	Variable	2018	2019	2020	2021	2022	2023	AAGR
Human resources	Number of on-duty staff	553	573	697	922	1023	1118	15.12
	Number of personnel with mid-level and senior professional titles	252	280	297	377	390	453	12.44
	Number of health technicians	391	397	460	662	738	805	15.54
	Number of personnel with bachelor’s degree or higher	358	393	523	774	892	959	21.78
	Staffing utilization rate (%)	55.09	55.92	60.19	79.75	87.66	91.18	10.60
Financial resources	Personnel expenditure (10,000 yuan)	10673.12	12228.96	13317.93	17168.6	23,300	25713.17	19.23
	Annual per capita expenditure for on-duty personnel (10,000 yuan)	18.12	20.26	18.11	17.74	21.76	22.1	4.05
	Annual per capita income (10,000 yuan)	10.54	11.46	12.82	12.37	15.24	14.8	7.02
Material resources	Number of laboratory equipment valued over 500,000 RMB (units)	40	41	53	56	57	65	10.20
	Implementation rate of Class A laboratory testing items (%)	59.56	59.21	61.24	61.27	67.58	75.77	4.93

Notes: AAGR = Average Annual Growth Rate

Table 2.

Average input-output data of CDCs in 10 areas of Qingdao (2018–2023)

Category	Variable	2018	2019	2020	2021	2022	2023	AAGR (%)
Input	Number of CDC staff per 10,000 population	0.56	0.58	0.69	0.9	1.00	1.08	14.04
	Per capita fiscal allocation for CDC	26.58	33.84	59.02	55.37	74.56	64.98	19.58
	Number of Class A laboratory equipment units per 10,000 population	0.73	0.77	0.77	0.79	0.80	0.80	1.85
Output	Childhood immunization program vaccination coverage rate	99.82	99.86	99.82	99.9	99.93	99.49	-0.07
	Premature mortality rate from major chronic diseases	12.95	12.69	12.48	12.18	11.96	11.71	-1.99
	Population health literacy level	18.98	21.84	24.38	27.8	32.17	35.29	13.21

Note: Per capita fiscal allocation for CDC is presented as reported in the source document; units for this variable in Table 2 are assumed to be yuan, consistent with typical fiscal reporting, though Table 1 reports personnel expenditure in 10,000 yuan. AAGR = Average Annual Growth Rate. The data of Premature mortality rate from major chronic diseases for 2022 and 2023 are the results of calculations based on previous data

Equity of the allocation of staff in CDCs in 2018–2023

The equity of CDC staff allocation in Qingdao from 2018 to 2023 was assessed using Gini coefficients, considering demographic, economic, and geographic dimensions. The results are presented in Table 3. Throughout the 2018–2023 period, the Gini coefficients for CDC staff allocation based on the demographic dimension consistently remained below 0.2 (ranging from 0.109 to 0.135). This indicates highly equitable distribution of staff relative to population size across the districts. Similarly, the Gini coefficients for staff allocation based on the economic dimension (likely reflecting GDP or similar economic indicators of the districts) were also consistently below 0.3 (ranging from 0.250 to 0.290), suggesting relative equity in relation to economic capacity. Conversely, Gini coefficients for geographically-based allocation were significantly higher. From 2018 to 2021, these exceeded 0.4 (peaking at 0.451 in 2020), reflecting substantial geographic inequity. Improvement occurred in later years, with the coefficient declining to 0.389 (2022) and 0.353 (2023), indicating moderate equity.

Table 3.

Gini coefficients for CDC staff allocation in Qingdao (2018–2023)

Year	Dimension of population	Dimension of economy	Dimension of geography
2018	0.110	0.259	0.414
2019	0.111	0.266	0.432
2020	0.135	0.250	0.451
2021	0.109	0.254	0.419
2022	0.117	0.265	0.389
2023	0.128	0.290	0.353

Notes: Gini coefficient: < 0.2 highly equitable; 0.2–0.3 relatively equitable; 0.3–0.4 moderately equitable/mildly inequitable; >0.4 inequitable

Overall, the Gini coefficient for the population dimension was the lowest, followed by that of the economy dimension, while the Gini coefficient for the geography dimension was consistently the highest. This pattern suggests that staffing levels align well with population numbers and economic capacity, but geographical disparities in staffing persist. The Lorenz curve for 2023 (Fig. 1) visually reinforces these differences.

Fig. 1.

Lorenz curve of staff in CDCs, Qingdao, 2023

Efficiency values of CDC

Based on the indicator system in this study, the efficiency of county-level CDCs in Qingdao in 2023 was calculated (Table 4), the crude average TE was 0.940, PTE averaged 0.995, and SE stood at 0.944. After bias-correction via bootstrapping, the adjusted averages were 0.892 (TE), 0.981 (PTE), and 0.947 (SE). Among these districts, 50% achieved DEA efficiency, including Z2, Z3, Z5, Z6, and Z9. Z1 was weakly DEA efficient (PTE = 1), while Z4, Z7, Z8, and Z10 were DEA inefficient (accounting for 40%).

Table 4.

Efficiency of county-level CDCs in Qingdao, 2023

DMU	Original efficiency			Bias-corrected efficiency
	TE	PTE	SE	TE	PTE	SE
Z1	0.947	1.000	0.947	0.914	0.988	0.948
Z2	1.000	1.000	1.000	0.924	0.988	1.000
Z3	1.000	1.000	1.000	0.925	0.988	1.000
Z4	0.816	0.980	0.833	0.792	0.955	0.847
Z5	1.000	1.000	1.000	0.938	0.988	1.000
Z6	1.000	1.000	1.000	0.961	0.989	1.003
Z7	0.860	0.988	0.870	0.836	0.973	0.874
Z8	0.863	0.984	0.877	0.841	0.964	0.887
Z9	1.000	1.000	1.000	0.922	0.988	1.000
Z10	0.910	0.999	0.911	0.877	0.993	0.906
Mean	0.940	0.995	0.944	0.892	0.981	0.946

a DMU = decision making unit, which refers to each CDC

b TE = technical efficiency, which is used to evaluate the total technical efficiency of CDC

c PTE = pure technical efficiency, which is used to evaluate the efficiency affected by the system and management level

d SE = scale efficiency, which is used to evaluate the efficiency affected by scale factors

Projection value analysis was conducted for districts with TE < 1 to quantify gaps between actual operations and optimal resource allocation, thereby guiding CDC resource reallocation. As shown in Table 5, based on the three output indicators incorporated in this study, the districts of Z4, Z7, Z8, and Z10 exhibited input slack in fiscal expenditures compared to other districts in 2023. In Z4, substantial per capita fiscal expenditure increases occurred primarily due to infrastructure investments such as new building construction. For example, the results for Z7 suggest that with more optimal utilization of fiscal inputs, potential improvements in outputs could include a 1.20% increase in childhood immunization coverage, a 16.10% reduction in premature mortality from major chronic diseases, and a 17.54% increase in public health literacy levels. Meanwhile, Z10 demonstrated higher CDC staff per 10,000 population than required; full utilization of existing human resources would significantly improve output indicators.

Table 5.

Slack variables of input and output indicators in noneffective CDC

		Z4	Z7	Z8	Z10
Input redundancy and proportion	Number of CDC staff per 10,000 population	0(0%)	0(0%)	0(0%)	-0.2(-15.35%)
	Per capita fiscal allocation for CDC	-431.28(-89.18%)	-13.73(-25.57%)	-43.46(-52.44%)	0(0%)
	Number of Class A laboratory equipment units per 10,000 population	0(0%)	0(0%)	0(0%)	0(0%)
Output insufficiency and proportion	Childhood immunization program vaccination coverage rate	1.98(2.04%)	1.18(1.20%)	1.58(1.62%)	0.12(0.12%)
	Premature mortality rate from major chronic diseases	-0.43(-4.55%)	-2.21(-16.10%)	-1.75(-13.48%)	-2.30(-15.86%)
	Population health literacy level	0.77(2.04%)	5.52(17.54%)	0.61(1.62%)	7.90(27.20%)

Changes of CDC efficiency

The 10 DMUs’ TEC, TC, PTEC, SEC and TFP during 2018–2023 were 1.012, 0.933, 0.999, 1.013, and 0.945, respectively. As shown in Table 6, TFP showed an improvement during 2022–2023, with an increase of 6.0%. In contrast, TFP derived from EPI (Expanded Program on Immunization), NCD (Non-communicable Disease) control, and health literacy surveillance demonstrated contraction during 2020–2021 when institutional resources were diverted to COVID-19 containment operations. Periods exhibiting technical efficiency growth included 2018–2019, 2020–2021, and 2022–2023, peaking at a 7.1% improvement in 2020–2021. TC scores persistently below 1 may reflect challenges in technological innovation within the disease control systems, particularly during 2020 when pandemic containment operations disrupted innovation cycles.

Table 6.

Aggregate Malmquist index results of county-level CDCs in Qingdao, 2018–2023

Year	Technical efficiency change (TEC)	Technical change (TC)	Pure technical efficiency Change (PTEC)	Scale efficiency change (SEC)	Total factor productivity (TFP)
2018–2019	1.069	0.897	1.000	1.069	0.959
2019–2020	0.931	0.967	0.996	0.935	0.900
2020–2021	1.021	0.842	0.999	1.022	0.860
2021–2022	0.976	0.981	0.994	0.982	0.957
2022–2023	1.071	0.989	1.009	1.062	1.060
Mean	1.012	0.933	0.999	1.013	0.945

a Technical efficiency change (TEC) measures the degree of change in technical efficiency, reflecting the quality of management methods and accuracy of management decision-making

b Technical change (TC) refers to the movement of the effective frontier, which reflects the rate of technological advancement or innovation

c Pure technical efficiency Change (PTEC) measures the efficiency of resource allocation and usage

d Scale efficiency change (SEC) measures the change in efficiency of the scale of input resources

e Total factor productivity (TFP) measures total productivity change

The input-output efficiency of district-level CDCs in Qingdao during 2018–2023 was calculated (Table 7). A scatter plot depicting the Malmquist index decomposition of TFP for these CDCs was constructed, using EC as the x-axis and TC as the y-axis (Fig. 2). Analysis of the scatter plot reveals that Qingdao’s ten districts can be broadly categorized into two groups: Group 1 (Z2, Z3, Z4, Z5, Z6, Z8, Z9) exhibited EC ≥ 1 but TC < 1, indicating insufficient technological innovation; Group 2 (Z1, Z7, Z10) demonstrated both EC < 1 and TC < 1, suggesting deficiencies in managerial efficiency and technological advancement. Among all districts, only Z3 achieved TFP change > 1. Moreover, this district had EC > 1 despite TC < 1. Thus, the improvement in technical efficiency appears to be the primary driver of TFP growth in Z3, which may reflect effective practices in resource allocation, utilization, and organizational management.

Table 7.

Malmquist indices for individual county-level CDCs in Qingdao, 2018–2023

DMU	Technical efficiency change	Technical change	Pure technical efficiency change	Scale efficiency change	Total factor productivity change
Z1	0.988	0.916	1.000	0.988	0.905
Z2	1.000	0.993	1.000	1.000	0.993
Z3	1.076	0.939	1.001	1.074	1.010
Z4	1.015	0.919	0.996	1.019	0.932
Z5	1.000	0.964	1.000	1.000	0.964
Z6	1.011	0.988	1.000	1.011	0.999
Z7	0.970	0.889	0.998	0.973	0.862
Z8	1.018	0.914	0.997	1.022	0.931
Z9	1.020	0.890	1.000	1.020	0.907
Z10	1.026	0.930	1.003	1.023	0.954
Mean	1.012	0.933	0.999	1.013	0.945

Fig. 2.

Decomposition results of the malmquist index for TFP in county-level centers for CDCs of Qingdao, 2018–2023

Discussion

The period from 2018 to 2023, particularly marked by the COVID-19 outbreak, presented unprecedented challenges and demands on public health systems globally and within China. Qingdao’s CDC system demonstrated a capacity for rapid resource mobilization in terms of staff expansion, funding allocation, and equipment deployment in response to the pandemic. As the system transitions into a post-pandemic era, the critical issues of establishing long-term and sustainable mechanisms for CDC resource investment have gained increased attention. Consequently, ensuring sustainable investment and optimizing resource utilization remain important challenges.

The descriptive data on health resource allocation in Qingdao’s county CDCs from 2018 to 2023 reveal positive trends in workforce development. The number of in-service staff, personnel with intermediate/senior professional titles, health technical personnel, staff with bachelor’s degrees or higher, and the overall staffing utilization rate all showed consistent growth. This expansion is generally consistent with findings from previous studies [31]. The COVID-19 pandemic is widely recognized to have substantially increased the operational burden on the CDCs, which likely necessitated both workforce expansion and higher qualification thresholds for personnel [32]. Indeed, Qingdao has reported implementing various measures to bolster its public health workforce, including increasing approved staffing quotas for professional technical personnel, introducing innovative salary incentive mechanisms with a tilt towards operational departments and frontline positions, raising the proportion of senior titles in core operational roles, establishing a chief expert program incorporating them into high-level talent management frameworks [33].

Financial investment is recognized as a fundamental guarantee for CDC institutions to fulfill their public health functions [34]. Financial investment in county-level institutions across Qingdao showed an overall increasing trend, with significant surges post-2020, which were likely linked to pandemic-driven emergency capacity building. A slight decline in district-level funding occurred in 2023, potentially attributable to local fiscal pressures and post-pandemic budgetary adjustments. This trend suggests a need for vigilance regarding its impact on sustainable grassroots CDC capacity [35]. Furthermore, increased laboratory equipment may signify enhanced infrastructure capabilities, providing material foundations for infectious disease control, laboratory testing, and public health emergency response. The rising implementation rate of Class A testing items further suggests technical capability advancements in CDCs.

From 2018 to 2023, the demographic allocation and economic allocation of staff in CDCs were assessed to be at a highly equitable level and relatively equitable, consistent with relevant studies [36]. However, the geographic allocation surpassed critical inequity levels. Chinese regional planning of health resources is often mainly based on the allocation of health resources per 10,000 population. Studies have indicated that the equity of staff in population allocation is higher than that in geographical area allocation in China [37]. In this study, this situation could potentially be due to the significant differences in geographical area among different districts in Qingdao City. In 2023, the geographical area of Z9 was 98.6 times that of Z1 [38].

Based on our conventional DEA model, in 2023, the mean TE of district-level CDCs was 0.940, the mean PTE was 0.995, and the mean SE was 0.944. The PTE of individual district CDCs was essentially (or close to) 1.000, indicating that under the existing conditions, the resource management and organizational capacity of the CDCs had reached a relatively high level. However, the bootstrap DEA results suggest that the conventional DEA TE scores are overestimated. The SE fluctuated more noticeably in both models, suggesting that county-level CDCs may have yet to reach an optimal scale configuration, necessitating further structural adjustments for improved efficiency. The efficiency rankings of the ten DMUs measured by both models are largely consistent, which supports the robustness of the findings. Previous studies showed that the mean TE of county-level CDCs in Eastern China was 0.422 (in 2010) [18]while the average TE of county-level CDCs in Shandong Province was 0.934 (in 2020) [39]. This suggests that the TE of Qingdao’s CDCs, calculated in this study, exceeds both the Eastern China regional average and the average TE of county-level CDCs in Shandong Province at those respective times. This superior performance may be partly attributed to Qingdao’s advanced economic development and robust health resource allocation system, among other factors. As Shandong’s leading economic hub, Qingdao’s high GDP level enables greater government investment in CDC operations. This economic advantage could facilitate substantial funding for infrastructure development, equipment upgrades, and professional training, which may ultimately contribute to enhancing public health service capacity. These results demonstrate the significant role of regional economic development in strengthening public health systems.

Using our input-output metrics, the DEA revealed relative slack in fiscal and human resources across four districts. Capital-intensive infrastructure projects (Z4’s and Z7’s relocation) contributed to input slack in fiscal resources. The resultant performance outcomes would benefit from longitudinal observation over extended timeframes. This pattern aligns with existing literature documenting input-biased allocation practices in the public health system [18, 40, 41]. Further analysis suggested systemic budgeting inefficiencies, where fiscal allocations exceeded operational needs due to inaccurate demand forecasting. This was potentially exacerbated by suboptimal fund utilization amid dynamic environmental changes. Additionally, while Qingdao expanded government-approved staffing quotas during the “14th Five-Year Plan” period [42] relative personnel surpluses emerged in some areas owing to contextual factors like population density and disease prevalence. A case in point is the Z10 CDC (423 persons/km²), where moderate population density paradoxically reduced service accessibility and consequently diminished relative efficiency.

The Malmquist index results indicate that TFP in Qingdao’s districts showed significant improvement during 2022–2023, potentially linked to the city’s strengthened public health system development. In 2021, the “Shandong Provincial Pilot Reform Implementation Plan for Deepening and Expanding the Three-Tier CDC System” proposed comprehensive reforms for provincial-municipal-county CDCs [43]. As a pilot region, Qingdao increased funding allocations for CDC infrastructure, personnel operations, and equipment procurement, while establishing a flexible incentive mechanism linking performance-based salary adjustments to operational evaluation outcomes—substantially enhancing talent attraction. Concurrently, a technical collaboration network between municipal and district CDCs reduced redundant investments through shared laboratory resources and specialist teams. Conversely, the sharpest TFP decline (14%) occurred in 2020–2021, coinciding with a 15.8% regression in TC. This may be related to the impact of the COVID-19 pandemic on the grassroots disease control system. Both 2019–2020 and 2021–2022 witnessed simultaneous declines in EC and TC, with EC reduction appearing to exert a greater impact on TFP. This pattern highlights inadequate investment in technological innovation and application within CDCs, while also potentially reflecting disruptions from the COVID-19 pandemic, which caused resource allocation and management difficulties [44]. During the pandemic, emergency response demands compelled suboptimal CDC resource distribution, which may have deviated from optimal configurations due to emergent public health demands, potentially undermining overall efficiency. Furthermore, major public health events can disrupt routine operations, preventing Qingdao’s resource inputs from translating into expected outputs. Critically, the pandemic likely impacted premature mortality rates from chronic diseases [45] thereby affecting the TE of CDCs. Furthermore, healthcare system strain significantly disrupted routine chronic disease management, which could compromise standard care delivery and exacerbate population-level long-term health risks [46]. Essentially, pandemic-induced challenges intensified the complexity of chronic disease management, further diminishing input-output efficiency at the district CDC level.

From the perspective of dynamic efficiency in resource allocation across district-level CDCs, only Z3 demonstrated marginal growth in TFP, while all other districts exhibited declines of varying magnitudes—most notably in Z7 with a substantial decrease of 13.8%. Although TE improved in 80% of CDCs, TC declined universally at the district level. These findings suggest that the TC was the primary driver of TFP reduction, suggesting potential issues such as delayed technological upgrades or insufficient innovation — a conclusion consistent with existing research [47, 48]. Particularly for county-level CDCs, which shoulder the most concrete and demanding disease prevention responsibilities, prioritizing the upgrade of detection technologies and equipment modernization, strengthening IT infrastructure development and technical training, is considered essential to enhance resource utilization efficacy and disease control capacity.

Conclusion

Based on a comprehensive analysis of six-year panel data from Qingdao’s county-level CDCs, this study indicates that resource allocation exhibits high equity in demographic and economic dimensions, but persists in geographical inequity, which is largely influenced by vast inter-district area disparities. The technical efficiency of CDCs is notably high, however, there exists a trend of diminishing scale efficiency. The TFP of evaluated CDCs showed an increase after the transition in COVID-19 prevention and control strategies (2022–2023). Priority should be given to improving TFP by implementing flexible staffing adjustments, increasing the number of technical personnel and senior experts, and strengthening continuous training to enhance workforce capabilities. Also, fiscal efficiency should be enhanced by linking fund allocation with performance evaluation. To enhance system resilience and resource utilization efficiency, we propose the following recommendations: (1) Optimize human capital through flexible staffing models that increase technical personnel and senior experts, coupled with continuous competency-based training; (2) Accelerate technological modernization via digital surveillance platforms, AI-driven outbreak forecasting, and regional laboratory equipment sharing networks to address TC deficits; (3) Implement geographically targeted resource redistribution to mitigate spatial inequities, particularly in large-area districts. Collectively, these evidence-based strategies are expected to contribute to strengthening the foundational capacity of China’s public health defense system in the post-pandemic era.

Study limitations

First, while this study established the final indicator system based on representativeness and data availability, CDC operations encompass complex and diverse functions—including prevention and control of major diseases (infectious/chronic/endemic/occupational illnesses), management of health risk exposures (environmental, dietary, occupational/radiation hazards), laboratory testing and surveillance, public health informatics, and health education—which may not be comprehensively captured by the selected indicators. Additionally, quantifying certain output indicators poses challenges, potentially introducing measurement errors that could compromise the accuracy of analytical results. And the use of self-developed structured questionnaires (Supplementary File S1) introduces methodological limitations. As a non-validated instrument, its construct validity remains unverified against established frameworks, potentially omitting latent variables influencing CDC performance.

Second, although the DEA model and Malmquist index are widely applied for efficiency analysis, they exhibit inherent limitations. The DEA model primarily focuses on internal factors of decision-making units (DMUs) with limited consideration of external influences; meanwhile, inter-regional variations in economic development levels, population density, and public health needs across sample areas may exist. Consequently, subsequent exploratory research on influencing factors remains necessary.

Third, while Qingdao’s economic prominence provides robust resources for CDC operations, our findings may be less transferable to resource-constrained regions. The high average TE (0.940) and fiscal allocation (¥649,800 per capita in 2023) likely reflect regional advantages. These strategies should be prioritized in economically developed regions, whereas underfunded CDCs may require baseline infrastructure investments.

Finally, the study employs a six-year panel data framework (2018–2023), which captures medium-term trends but may not fully account for long-term structural changes or the sustained impact of health system reforms. The DEA-Malmquist model measures productivity changes over discrete periods, yet it does not incorporate dynamic adjustments or lag effects in resource allocation. In addition, the study period encompasses the COVID-19 pandemic, which introduced extraordinary disruptions and demands that may distort typical efficiency patterns and limit the comparability of pre-, during-, and post-pandemic performance.

Future studies could extend the temporal and geographic scope of the research, encompassing regions with varying economic conditions, and incorporate a broader set of indicators, such as qualitative measures and process-oriented metrics. These efforts should be coupled with validation of the measurement instrument and the application of more advanced modeling techniques that account for exogenous factors. Such an integrated approach would yield a more holistic and contextually nuanced understanding of both efficiency and equity in public health delivery systems.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CDC

Centers for Disease Control and Prevention

DEA

Data Envelopment Analysis

DMUs

Decision-Making Units

CCR

Charnes-Cooper-Rhodes

CRS

Constant Returns to Scale

TE

Technical Efficiency

BCC

Banker-Charnes-Cooper

VRS

Variable Returns to Scale

PTE

Pure Technical Efficiency

SE

Scale Efficiency

TFP

Total Factor Productivity

TC

Technical Change

TEC

Technical Efficiency Change

PTEC

Pure Technical Efficiency Change

SEC

Scale Efficiency Change

AAGR

Average Annual Growth Rate

Author contributions

CW: Proposed research ideas, data processing and analysis, manuscript writing. KL: Analysis and refinement of research results, manuscript writing.DL: Guidance on the paper’s ideas and paper polishing. ML: Data extraction and analysis. YZ: Material search and data extraction. AM: Paper polishing and formatting revisionsCorresponding Author: Conceptualization, supervision, reviewing, and editing of the paper.

Funding

This research was funded by IMICAMS Youth Talent Development Fund (Grant Number 2024YT11), Qingdao Key Medical and Health Discipline Project (Grant Number 20250160), and Policy Research Project by the Qingdao Municipal Health Commission (Grant Number QDWJZCYJ2025-039). The funding bodies did not play any role in the study design, in the collection, analysis and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication.

Data availability

Data was provided by the Qingdao Centers for Disease Control and Prevention. And the data that support the findings of this study are not openly available due to reasons of sensitivity and are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was performed in line with the ethical principles of the Declaration of Helsinki. And this study utilized anonymized administrative data routinely collected by Qingdao CDC branches for operational reporting purposes. No individual-level human data, biological samples, or personal identifiers were involved. According to China’s “Ethical Review Measures for Biomedical Research Involving Humans” (National Health Commission, 2016), analyses of aggregated institutional performance data without personal identifiers are exempt from ethics committee review. Informed consent was obtained from all the participants.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.