Intervention portfolios analysis of Plasmodium vivax control in central China

Abstract

Background

The effects of a diverse spectrum of malaria interventions were evaluated through a deterministic Plasmodium vivax transmission model. This approach aimed to provide theoretical evidence of the performance of these interventions once implemented for achieving malaria elimination.

Methods

An integrated intervention portfolio, including mass drug administration, insecticide treatment, and untreated bed nets, was analyzed through modeling. Additionally, data-driven calibration was implemented to infer coverages that effectively reproduced historical malaria patterns in China from 1971 to 1983.

Results

MDA utilizing primaquine emerged as the most effective single intervention, achieving a 70% reduction in malaria incidence when implemented at full coverage. Furthermore, a strategic combination of MDA with primaquine, chloroquine, untreated bed nets, and seasonal insecticide treatments effectively eradicated malaria, attaining elimination at a coverage level of 70%. It was conclusively demonstrated that an integrated approach combining MDA and vector control measures is essential for the successful elimination of malaria.

Conclusion

High coverage of mass drug administration with primaquine and chloroquine before transmission was the key driver of the malaria decline in China from 1971 to 1983. The best-fit intervention coverage combinations derived from calibration are provided as a reference for malaria control in other countries.

Background

Malaria remains one of the most prevalent and deadly diseases worldwide, caused by the parasitic protozoa of the genus Plasmodium, transmitted to human hosts through the bite of infected female Anopheles mosquitoes [1]. Among the different forms of malaria, Plasmodium falciparum, responsible for malignant malaria, and Plasmodium vivax, causing vivax malaria, are the most common. Malignant malaria is characterized by rapid parasite multiplication and destruction of host red blood cells, often leading to severe complications, such as cerebral malaria, which is fatal without prompt treatment. In contrast, vivax malaria, though generally less severe, causes periodic fevers and anaemia that can lead to significant long-term health issues. The parasites can form dormant stages in the liver, leading to relapses of the disease. Although global morbidity and related deaths have declined substantially due to enormous efforts to control malaria since 2000 [2], the disease still posed a serious public health problem in 2020, with 241 million cases and 627,000 deaths reported worldwide [3].

The People's Republic of China reached a milestone in malaria elimination on June 30, 2021. Through more than 70 years of unremitting prevention and control, China reduced its malaria burden from 30 million annual cases in the 1940s to zero local cases in 2017, followed by four consecutive years to achieve WHO malaria-free certification [4]. This success sparked global enthusiasm and interest in the domestic intervention strategies used.

In central China, the transitional regions between subtropical and temperate zones experienced several historical malaria outbreaks. Henan Province was once a malaria-heavy endemic region because of favourable climatic conditions. In 1970, the number of malaria cases in the province exceeded 13 million, with P. vivax accounting for more than 90% of the cases. The incidence rate for the entire province dropped from 16.94% in 1970 to 0.14% in 1983 according to reported data [5], demonstrating the early success. Now, local malaria cases have not been reported since 2012, three years ahead of the target established by the National Malaria Elimination Action Plan (2010–2020) [6].

However, there is no silver bullet for malaria intervention. The Chinese government used a variety of simultaneous intervention measures to control and eliminate malaria (see Table 1). Although the outcome of the overall programme has clearly been demonstrated, its constituent interventions were neither thoroughly documented nor quantitatively evaluated. It remains unclear which intervention strategy played a major role in malaria control. Therefore, in the present study, a deterministic malaria transmission model was introduced to quantitatively analyze the effects of various interventions.

Table 1 Malaria control measures and types [7,8,9]

To date, scholars have established many mathematical models of malaria transmission to evaluate the performance of malaria control measures. For instance, Hsiang et al. assessed the important role of MDA in the process of eliminating malaria in Jiangsu, P. R. China [10]. Kassam et al. discussed the effect of untreated bed nets on the transmission of P. falciparum between mosquitoes and human hosts [11]. Dereje et al. proposed a mathematical model to analyse malaria transmission dynamics under interventions [12]. Beretta et al. introduced an age-structured mathematical model of malaria transmission including asymptomatic carriers [13]. The limitation of these studies is that they only analyse single interventions. Existing literature contains few quantitative studies on the simultaneous application of multiple interventions. Furthermore, previous studies have paid less attention to the prevention and control of P. vivax, which is an important challenge for many countries, including P. R. China, in the final malaria elimination stage because of dormant hypnozoites of P. vivax in the human liver that recur after a long interval of months or even years, while lower vivax parasite loads result in milder symptoms make it harder to diagnose.

In the present study, the temperate P. vivax transmission model presented by White et al. [14] was adapted to evaluate a diverse spectrum of interventions based on the epidemiological data of Guantang County of Henan Province from 1971 to 1983, where P. vivax accounted for over 90% of the total malaria cases. Model calibration was then performed by fitting the model to infer the possible interventions used in China from 1971 to 1983. The findings of this study will be useful for malaria elimination efforts in countries with similar ecological settings.

Methods

This section details the approach to modelling the historically implemented interventions. A diverse spectrum of interventions was investigated, including mass drug administration, population-engaged seasonal insecticide treatment, and untreated bed nets. It is crucial to note that while these interventions were indeed utilized historically, specific data on the extent of their implementation remain sparse. Accordingly, the model explores various hypothetical implementation ranges to estimate the most likely scenarios that align with historical trends observed from 1971 to 1983. All analyses were run in R (version 4.2.0) [15].

Study site and malaria historical data (1971–1983)

This study was based on historical data from Guantang County (latitude 33°51ʹ and longitude 115°21ʹ), which is located in the river network region of Henan Province. During the study period, Guantang had 114 villages with approximately 40,000 people [16]. Plasmodium vivax malaria was prevalent in this county with Anopheles sinensis as the only vector. Because of the malaria outbreak in central China (including Henan Province) in 1970, a series of pilot studies, comprising Guantang, were set up to comprehensively understand the transmission pattern and the effects of interventions. Therefore, the historical data on malaria in Guantang from 1971 to 1983 is well documented. Figure 1 shows the study site in Henan, P. R. China. The green curve inset in Fig. 1 presents the monthly reported number of malaria cases at the study site between 1971 and 1983.

Fig. 1

Study site and malaria monthly incidence in Guantang, Henan, P. R. China, between 1971 and 1983

Plasmodium vivax temperate transmission model modified with seasonality modifications

The present study utilized an extended compartmental Ross-MacDonald model, incorporating different characteristics of P. vivax, such as the latency period and relapse rate. The transmission dynamics of malaria can be divided into two sections: human and mosquito. To simulate the periodic variation of the adult mosquito population, seasonality was added into the model. In the model, seasonality is reflected through the parameter C, which is defined as:

$$C=1+\text{cos}\left(\frac{2\pi t}{\omega }-\varphi \right).$$

where t represents time (typically measured in days), ω denotes the total number of days in an annual cycle, and ϕ is the phase shift indicating the onset of seasonality. This formulation simulates the impact of environmental conditions (such as temperature and humidity) on mosquito population dynamics. These conditions vary across seasons, directly affecting mosquito reproduction and, consequently, the transmission rate of malaria. Diagrams of P. vivax transmission model in mosquitoes and humans are shown in Supplement Fig. S1. Parameters. The calibrated values used in the model were listed in Table 2.

Table 2 Parameters and their calibrated values used in the model

Intervention profiles

During the 1970s and 1980s in Guantang, chloroquine and primaquine therapy was administered in a five or eight-day course to passively detect cases as part of baseline case management. These cases typically involved individuals who sought treatment without active surveillance efforts. This therapeutic strategy was crucial for rapid disease identification and effective parasite eradication. Given the medical infrastructure and resource availability at the time, it was assessed that a moderate level of case management was achievable. It was hypothesized that approximately half of all infected individuals received treatment. Baseline treatments were consistently applied throughout the year, leveraging the anti-gametocytic properties of chloroquine and primaquine, which reduce the transmission potential from humans to mosquitoes. After consulting with malaria control experts, the reduction was quantified at 20%, reflecting the expected decrease in the human-to-mosquito transmission capacity under typical treatment conditions. In addition to case management, a diverse spectrum of interventions for malaria control was actually implemented in Guantang. These interventions included mass drug administration, population-engaged seasonal insecticide treatment, and the distribution of untreated bed nets. While these interventions were indeed used historically, the exact coverage and effectiveness varied over time and across different areas. The approach selected in this study models the link between intervention tool(s) and resulting health impact (incidence rate reduction) and defines the requirements of these interventions in terms of coverage and impact duration to reach the desired health goals, contingent on operational constraints. This approach allows for modeling the historical control of malaria in China, reflecting the actual interventions used while accounting for uncertainties in their precise implementation details.

Indoor residual spraying (IRS) is a control measure involving spraying persistent insecticides on indoor surfaces where mosquitoes rest. Anopheles sinensis, a species with zoophilic and outdoor blood-feeding behaviour, is the main malaria vector in Henan Province. Although IRS was advocated by World Health Organization (WHO), it was unsuitable for Henan. Hence, IRS was not considered in this study.

The specific interventions to be considered in this study are as follows.

Mass drug administration

MDA was conducted once every year during the study period. Before the malaria transmission season, each round of MDA began in March and lasted for two weeks. Both primaquine and chloroquine can be used as inhibitory drugs when conducting MDA. Primaquine increases the rate of hypnozoite clearance at the latent and dormant stages, ${\gamma }_{L}$ and ${\gamma }_{D}$ whereas chloroquine increases the blood-stage infection recovery rate ${r}_{h}$. The functional comparison of primaquine and chloroquine is listed in Table 3.

Table 3 Modelled intervention profiles

Models of MDA with primaquine and MDA with chloroquine were used to represent two drugs that could be employed as single interventions in the MDA process. In the single intervention portfolio analysis of the result section, these interventions are evaluated individually.

Since there is no single anti-malarial drug that efficiently kills P. vivax at both liver and erythrocytic stages, primaquine and chloroquine are usually used in combination in the MDA process. In the intervention combination portfolios analysis and model calibration of result section, the combination use of the two drugs is considered.

Although MDA is potentially very effective in practical application, literature shows that coverage must reach 80% for MDA to be successful [20]. The most important success factor for MDA is to achieve a coverage of at least 80%-90% of the target population.

Population-engaged seasonal insecticide treatment

Population-engaged seasonal insecticide treatment (PSIT) activities were carried out during the entire study period, referring to adult mosquito control using dichlorvos, benzene hexachloride, and fogging. These activities were performed several or dozens of times annually. According to the literature [21], PSIT could effectively kill adult mosquitoes with a population reduction rate of up to 90%, thus reducing the average mosquito population size ${M}_{0}$ in the model. However, details of this activity, such as the deployment time and duration, were not well documented. It was assumed that PSIT was conducted twice per week from May to September (an interval of approximately 150 days) based on historical data of mosquito density and feeding behaviour.

Untreated bed nets

Untreated bed nets were distributed to residents, who were encouraged to sleep under the untreated bed nets during the study period. Untreated bed nets reduce the mosquito biting rate $a$ in the model. However, residents in many areas of Henan were accustomed to resting or sleeping overnight outdoors during the summer, and An. sinensis prefers outdoor biting. A previous study [16] suggested that the indoor/outdoor An. sinensis population ratio in Henan is 1:7.73, so even if untreated bed nets were 100% deployed indoors, the biting rate could be reduced by no more than 12%.

The efficacy parameters for the intervention measures in this section were established based on the data presented in reference [5, 16, 21]. Below, the modelling methods for the efficacy of various intervention measures are discussed.

Untreated bed nets implemented result in the adjustment of the relevant parameters in the model. The way of adjustment is to multiply the original parameters by (1—coefficient adjustment * coverage), and the coefficient adjustment is related to the characteristics of the intervention itself. In the model, untreated bed nets were set to reduce 12% mosquito bites, then mosquito biting rate a is multiplied by (1–12%*coverage bed net) in the model, so the ${\lambda }_{1}$, ${\lambda }_{m}$ in the model changes accordingly. PESI affects the total number of mosquitoes ${{\varvec{M}}}_{0}$. During the implementation of PESI, the total number of mosquitoes is reduced to 10% of its original value. In practical applications, primaquine enhanced the clearance rate of hypnozoites at both the latent and dormant stages, represented by ${\gamma }_{L}$ and ${\gamma }_{D}$, respectively, while chloroquine increased the recovery rate from blood-stage infections, denoted by ${r}_{h}$. Given that the implementation of MDA requires a defined period, a cycle of 12 days was assumed during which chemotherapy is administered to 1/12th of the population daily. Consequently, the total number of infected individuals decreases by 1/12th each day, resulting in a non-infected state across the entire population after 12 days. Thus, in this context, the role of MDA directly influences the quantity of individuals in the infected compartment (I), rather than providing an enhancement to ${\gamma }_{L}$, ${\gamma }_{D}$ and ${r}_{h}$.

Intervention analysis, calibration, and timing

Intervention portfolios analysis

To evaluate the performance of various intervention measures, a two-step approach is adopted: intervention effect (incidence reduction rate) comparison between 1) every single intervention and 2) multiple interventions at various coverage rates.

A burn-in period of 30 years is included, which is sufficient to achieve natural equilibrium before intervention. The interventions were introduced into the model in the result section at the beginning of 1971 (year 31), according to each intervention scenario. The minimum and maximum coverage rates of interventions were all set to 0% and 100%, respectively. Malaria incidence in 1971 serves as the reference rate for evaluating intervention performance relative to the targeted incidence reduction rate from 1971 to 1983. The formula is

$$\text{incidence\, reduction\, rate}=\frac{{\text{incidence\, rate}}_{1971}-{\text{incidence\, rate}}_{1983}}{{\text{incidence\, rate}}_{1971}}.$$

Model calibration based on historical data

Acknowledging the absence of recorded coverage data for the interventions at the time, a rigorous model calibration process was undertaken to align the model outputs with historical incidence data, exploring the potential effects of various intervention strategies for eliminating P. vivax malaria. A grid search technique is utilized to identify the most effective coverage levels of each intervention by minimizing the mean square error (MSE) between model outputs and observed data from 1971 to 1983.

During this calibration phase, the intervention coverages were constrained to ensure realistic applicability based on documented usage rates and logistical feasibilities. For instance, the possible range for mass drug administration (MDA) was set between 60 and 90%, reflecting the high compliance rates observed. The upper coverage limit for untreated bed nets, widely used during the study period, was set at 80%, while the maximum feasible coverage for population-engaged seasonal insecticide treatment (PSIT), due to its intensive resource demands, was capped at 60%. The minimum coverage for both untreated bed nets and PSIT was maintained at 0% (see Table 2).

The target criterion for successful malaria control, as per the People's Republic of China's guidelines, is an annual incidence rate below 0.1%. This is adopted as a crucial threshold for the model's calibration process.

Optimal MDA timing simulation

The purpose of this simulation is to examine the effect of MDA deployment time on the incidence rate of the final year of simulation. Here, the coverage setting is set as the top 1 result of intervention combinations from model calibration. MDA implement dates are systematically varied throughout the year, starting from day one and continuing through to day 365. Each simulation evaluates the impact of consistently implementing MDA on the same calendar date across multiple years. This approach allows to assess how the specific timing of MDA commencement influences the incidence rate of the final year.

Results

Intervention portfolios analysis

The dynamic model is coupled with various interventions and simulated incidence across a range of coverages for certain combinations. Figure 2 depicts the variation in the incidence reduction rate caused by the interventions at different coverages over 13 years.

Fig. 2

Control effect of single and multiple interventions. The incidence reduction effect at coverage 0 indicates the effect of case management (baseline). The X-axis represents the coverage of interventions, and Y-axis represents the incidence reduction rate

Figure 2A demonstrates that MDA with primaquine was the most effective single intervention, particularly at a high coverage rate, followed by PSIT, untreated bed nets, and MDA with chloroquine. MDA with primaquine alone could reduce malaria transmission, with the incidence reduction rate increasing nonlinearly with the coverage rate. PSIT exhibited the second-best performance, resulting in a linear relationship between the incidence reduction rate change and the coverage rate. It was also noted that MDA with chloroquine exhibited little effect on the incidence rate, and untreated bed nets performed better than MDA with chloroquine. At 100% coverage, MDA with primaquine could reduce the incidence rate by more than 70%, while other interventions could only reduce the incidence rate by less than 20%; therefore, MDA with primaquine was the most effective single intervention.

Figure 2B shows the comparisons of single-drug interventions and their combinations. The results suggest that administering simultaneous primaquine and chloroquine further improves their performance for incidence reduction.

Figure 2C shows the comparisons between single vector control interventions and their combinations. Compared with single vector control interventions, combined vector control interventions exhibited greater effects on the incidence reduction rate.

Figure 2D shows the performance of three intervention combinations. The incidence reduction rate of all intervention combinations increased nonlinearly with the increase in coverage rate, and combined MDA performed better than untreated bed nets + PSIT. Combined MDA came close to eliminating malaria at 100% coverage, while untreated bed nets + PSIT could only reduce the incidence rate by around 50%. As the maximal combination of drug and vector interventions, combined MDA + untreated bed nets + PSIT could achieve elimination at 70% coverage.

Therefore, an integrated malaria intervention strategy with both drug and vector control is recommended. Notably, the drug intervention combination performed better than the vector control combination. This finding is attributed to the use of primaquine to clear the hypnozoite reservoir.

The preceding sections demonstrated that the combination of interventions can achieve malaria elimination. However, it is essential to exceed a certain coverage level in practice. Thus, optimization is performed in the following section according to integrated malaria intervention to infer coverage levels that are consistent with historical data. These optimal coverages represented the intervention implementation details in China from 1971 to 1983.

Model calibration based on historical data

Combined MDA, untreated bed nets, and PSIT were considered in the process. The top 20 coverage combinations of interventions calculated through fitting are listed in Table 4 and sorted by increasing MSE. These best-fitting models consistently had high MDA coverage (between 70 and 90%). They also exhibited high untreated bed nets coverage (55–80%). PSIT was absent in the results (0–5%). Figure 3 depicts the monthly malaria cases derived from the model simulation. The black curve represents the simulation result of the best-fit coverage combination, while the green curve represents the real data. The simulation result was a good approximation to the real data, especially towards the wave crest in the first five years. However, the simulation result was not able to reproduce two anomalous, irregular spikes in the data between 1980 and 1982.

Table 4 Results of intervention combinations from model calibration (top 20)

Fig. 3

Monthly malaria cases comparison between simulation under intervention with MDA coverage = 0.9 and untreated bednets coverage = 0.8 versus historical data

Figure 4 shows that the deployment time for MDA applications in the field is crucial to the success of the activity. In the first 2 months, the incidence rate is close to zero; it then shows a rapid increase and reaches its peak in June, followed by a consistent decline until December, when it returns to near zero.

Fig. 4

Control effect of MDA_combined when implemented in a different month. The X-axis represents the start day of intervention in each year, and the Y-axis represents the incidence rate of the final year of simulation. The coverage rate of MDA_combined was 90%, and the coverage rate of untreated bednets was 80%

In Fig. 5, the annual variation in the value of C is observed, with peaks typically occurring during the hot and rainy season, which provides optimal conditions for mosquito survival and reproduction, thereby potentially increasing the risk of malaria transmission. Conversely, during periods of low C values, usually in colder or drier seasons, mosquito activity and populations are suppressed, reducing the risk of malaria transmission. The comparison between Figs. 4 and 5 illustrates how seasonal variations can indirectly affect malaria incidence by influencing mosquito population sizes. Deploying MDA during months with low C values can maximize the effectiveness of interventions, thereby effectively reducing the transmission of malaria.

Fig. 5

The variation of the C value in a year

Discussion

With the WHO’s malaria-free certification of the P. R. China on June 30, 2021, there has been renewed interest in uncovering the details of malaria elimination. The P. R. China's successful experience can be summarized by the principle of “prevention first (MDA chemotherapy every spring), joint prevention and control (One Health approach) and actively developing anti-malarial drugs (such as artemisinin)”. Few previous studies have analysed the effects of different interventions on P. vivax elimination through the mathematical model based on pilot historical data in central China. This study has significant implications for public health and may inspire further scientific studies into intervention measures.

This study demonstrated that MDA using primaquine is more effective than MDA using chloroquine. Primaquine targets both liver-stage hypnozoites and partial gametocytes, thereby reducing the recurrence of P. vivax cases and further preventing human-to-mosquito transmission. Thus, primaquine played a pivotal role in the process of eliminating P. vivax in China. Chloroquine exhibits little efficacy on dormant liver-stage hypnozoites but can still clear blood-stage parasites and, therefore, reduces subsequent gametocyte production. Due to the drug mechanisms, the synergistic effects of the combined medication with primaquine and chloroquine were significant. The timing was also crucial to the success of MDA. The best time to conduct MDA is spring, during the fallow period preceding the transmission season (see Supplement Fig. S2), which eliminates the latent malaria parasites in residents while most of the target population is asymptomatic.

The success of MDA in the P. R. China was dependent on the high compliance rate of the Chinese people. Various intervention measures could be conducted effectively, and MDA coverage rates could exceed 60%. Governmental engagement and health education should be bolstered in other malaria-endemic countries where achieving high drug coverage is challenging.

However, in practice, MDA is constrained by the contradiction between treatment time, compliance, and severe haemolysis caused by glucose-6-phosphate dehydrogenase (G6PD) deficiency. The longer the treatment time, the lower the compliance rate, which diminishes the treatment's effect. Thus, the development of single-dose drugs has become appealing and may help to eliminate P. vivax malaria permanently. In areas such as the China-Myanmar border, where the prevalence of G6PD deficiency is close to 30% [22], high MDA coverage is unachievable, and many malaria-endemic countries lack reliable G6PD testing [23]. Therefore, it is not feasible to achieve malaria elimination through the mass administration of primaquine and chloroquine.

Vector control, such as PSIT, reduces the mosquito population and human-mosquito contacts, thereby reducing the transmission of P. vivax, although it cannot prevent recurrent infections. However, the high requisite coverage level is difficult to achieve in practice and results in a lesser contribution to malaria elimination, which is consistent with the results of the model calibration. Nevertheless, in China, such a health campaign enhanced the environmental hygiene of both urban and rural areas, thereby enhancing the population’s ability to resist other diseases. These efforts were instrumental in controlling epidemics, such as encephalomyelitis, malaria, and typhoid fever in rural areas from the mid-1960s to the late 1970s [24]. Untreated bed nets, another important vector control tool that specifically targets mosquito-borne diseases, can play a key role in reducing contact between vectors and susceptible populations. However, untreated bed nets alone are not the most desirable control measure in the regions where An. sinensis is the dominant malaria vector. The zoophilic and outdoor blooding feeding behaviour of An. sinensis limits the efficacy of untreated bed nets. In the late 1970s, coupled with the popularity of electric fans and air conditioners, Chinese people gradually shifted away from sleeping outdoors and reduced their exposure to mosquitoes.

Based on historical data from the 1970s in China, drug interventions appeared to be more impactful than vector control interventions during that period. However, this does not discount the significant contributions of vector control measures. Overall, while mosquito vector control has been demonstrated to be an effective malaria control measure [25, 26], its effectiveness and the difficulty of implementation vary between regions. In central China (temperate or warm tropics), there are few species of Anopheles mosquitoes, and the ecological habits of mosquitoes are well understood. However, in Africa, there are many species of Anopheles mosquitoes, and the ecological environment is complex. There, vector control methods, such as untreated bed nets, which are not applicable in Henan, are very effective against mosquitoes that prefer indoor biting.

To improve the effectiveness of mosquito control, it is necessary to develop measures based on the blood-feeding and activity habits of different Anopheles mosquito species. In recent years, the use of eave ribbons treated with transfluthrin has emerged as a promising strategy. These eave ribbons, by releasing a spatial repellent, can effectively reduce mosquito entry into houses and decrease outdoor mosquito contact with humans. This method is particularly valuable in areas where Anopheles mosquitoes exhibit significant outdoor and early evening biting behaviours, providing an alternative to the traditional net-based strategies, which may not cover all aspects of mosquito behaviour.

In addition to traditional vector control measures, recent advancements have introduced more sophisticated technologies. For instance, the Olyset Duo net not only serves as a barrier but also impairs the reproductive capabilities of mosquitoes, thereby reducing their prevalence. This added mechanism addresses one of the traditional limitations of untreated bed nets by actively decreasing mosquito populations over time, further enhancing the potential for sustained disease control. Results suggest that MDA chemotherapy with a high coverage rate played a significant role in the elimination of malaria in the P. R. China. In addition to the widespread use of untreated bed nets, the best fit PSIT coverage was low, suggesting that the intervention was difficult to implement in practice. Therefore, the most rational intervention combination was MDA and untreated bed nets with a combined 70–80% coverage.

This study corroborates previous research, demonstrating that the combined implementation of mass drug administration (MDA) and conventional bed nets has significantly contributed to malaria control efforts. Blackburn et al. [27] showed that in central Nigeria, the combination of bed net distribution and MDA significantly increased bed net ownership and usage, thereby improving malaria control outcomes. Gitaka et al. [28] conducted a study on low-endemic islands in Kenya, combining MDA with untreated bed nets, which resulted in a significant reduction in malaria prevalence. Yukich et al. [29] found in their Zambian study that the combination of MDA and untreated bed nets was cost-effective and significantly reduced malaria infection rates.

Several other studies have emphasized the synergistic effects of combined intervention measures. For instance, Griffin et al. [30] pointed out that in low to moderate transmission settings, the combination of Long-Lasting Insecticidal Nets and Indoor Residual Spraying significantly reduced malaria transmission. Kiware et al. [31] used model simulations to assess the impact of various intervention combinations on malaria transmission, demonstrating that combinations of multiple interventions were more effective than single measures.

The model reveals the effectiveness of intervention measures in the specific context of Henan Province, China, which may differ from other regions such as Southeast Asia or Africa. Roy et al. [32] studied the differences in epidemiological backgrounds and intervention measures across various regions, noting that these disparities may stem from geographical and socioeconomic factors, as well as differences in malaria transmission dynamics.

This study also has deficiencies. The model simplifies the distinction between blood-stage infections and clinical disease, which may influence the interpretation of intervention outcomes. This simplification was primarily implemented to maintain computational feasibility while capturing the overall dynamics of malaria transmission under various intervention scenarios. This simplification may affect the model's ability to accurately reproduce observed epidemiological patterns, particularly in areas where asymptomatic infections are prevalent. By treating all blood-stage infections as potential clinical cases, the model may overestimate the number of clinical cases, especially in regions with high immunity where asymptomatic infections are more common. Despite these limitations, the model remains capable of reproducing key patterns and intervention effects observed in malaria epidemiology, primarily through the following mechanisms: First, by introducing the seasonal variable C, which modulates transmission dynamics according to the time of year, the model can capture the cyclical patterns of incidence observed in many malaria-endemic areas. Second, by including compartments for latent and dormant periods, the model accounts for the unique relapse patterns characteristic of P. vivax. Third, the impact of various interventions is simulated by adjusting relevant parameters, such as reducing mosquito biting rates for untreated bed nets, thus reproducing observed intervention effects. Lastly, although model asymptomatic infections are not explicitly modeled, the approach of adjusting overall transmission rates based on interventions and seasonality allows to approximate the complex dynamics of malaria transmission, including the role of asymptomatic carriers in maintaining transmission.

While the intervention combinations proposed could effectively control and eliminate the temperate P. vivax strain in the P. R. China, these intervention programmes may not be ideal for other regions. If cost information were available for these interventions, it could further improve the model and intervention combinations. To achieve the goal of eliminating malaria in the Asia–Pacific region by 2030, research must be conducted on tropical P. vivax transmission control in countries with high malaria incidence, such as India, Indonesia, Pakistan, and Papua New Guinea. Because P. vivax was the dominant species in the P. R. China, this paper focused on the elimination of P. vivax malaria alone. Nevertheless, the study of the mixed transmission of P. vivax and P. falciparum is of great significance to countries where both falciparum malaria and vivax malaria are common and is a potential domain for future work.

Conclusion

By using a mathematical model, this study quantitatively analysed and summarized, for the first time, the successful experience of eliminating malaria in the P. R. China. An adapted model combining interventions was introduced to perform intervention portfolio analysis and simulate the reduction of P. vivax transmission in Guantang, Henan Province due to various interventions conducted from 1971 to 1983.

• MDA with primaquine was much more effective than with chloroquine.

• MDA strategy with both primaquine and chloroquine was a cornerstone of P. vivax transmission control.

• Drug interventions were much more effective than vector control interventions in the 1970s in the P. R. China.

• An integrated malaria intervention strategy with high coverage of MDA and vector control was the key to achieving malaria elimination in the P. R. China.