Article Text
PDF
Abstract
Introduction Thailand’s malaria surveillance system complements passive case detection with active case detection (ACD), comprising proactive ACD (PACD) methods and reactive ACD (RACD) methods that target community members near index cases. However, it is unclear if these resource-intensive surveillance strategies continue to provide useful yield. This study aimed to document the evolution of the ACD programme and to assess the potential to optimise PACD and RACD.
Methods This study used routine data from all 6 292 302 patients tested for malaria from fiscal year 2015 (FY15) to FY21. To assess trends over time and geography, ACD yield was defined as the proportion of cases detected among total screenings. To investigate geographical variation in yield from FY17 to FY21, we used intercept-only generalised linear regression models (binomial distribution), allowing random intercepts at different geographical levels. A costing analysis gathered the incremental financial costs for one instance of ACD per focus.
Results Test positivity for ACD was low (0.08%) and declined over time (from 0.14% to 0.03%), compared with 3.81% for passive case detection (5.62%–1.93%). Whereas PACD and RACD contributed nearly equal proportions of confirmed cases in FY15, by FY21 PACD represented just 32.37% of ACD cases, with 0.01% test positivity. Each geography showed different yields. We provide a calculator for PACD costs, which vary widely. RACD costs an expected US$226 per case investigation survey (US$1.62 per person tested) or US$461 per mass blood survey (US$1.10 per person tested).
Conclusion ACD yield, particularly for PACD, is waning alongside incidence, offering an opportunity to optimise. PACD may remain useful only in specific microcontexts with sharper targeting and implementation. RACD could be narrowed by defining demographic-based screening criteria rather than geographical based. Ultimately, ACD can continue to contribute to Thailand’s malaria elimination programme but with more deliberate targeting to balance operational costs.
- Elimination
- reactive case detection
- proactive case detection
Data availability statement
Data are available in a public, open access repository. The data supporting the conclusions of this article are available in the Malaria Online repository, http://malaria.ddc.moph.go.th/.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Active case detection (ACD) has been a standard ‘last mile’ component of malaria elimination programmes.
The WHO recommends a varying mix of proactive ACD (PACD) and reactive ACD (RACD) along the elimination spectrum; however, there is no known threshold correlated with effectiveness of RACD and PACD strategies for malaria surveillance.
WHAT THIS STUDY ADDS
In Thailand, ACD yields are declining alongside decreasing malaria incidence, with a noticeably higher yield for RACD compared with PACD and significant geographical differences.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This documentation of ACD implementation and results helps identify opportunities to simplify the complexity of Thailand’s various ACD methods and to sharpen targeting and implementation of interventions.
In particular, a PACD protocol could outline specific microcontexts where PACD could meaningfully curb transmission in an elimination setting.
Introduction
The six countries in the Greater Mekong Subregion (GMS)—Cambodia, China (Yunnan province), Lao People’s Democratic Republic, Myanmar, Thailand and Vietnam—collectively strive to eliminate malaria by 2030. Between 2000 and 2020, the number of Plasmodium falciparum malaria cases in the GMS fell by 93% and all-species malaria cases fell by 78%, resulting in just 82 000 malaria cases across the region in 2020.1
Thailand’s progress has acceleratedsince the adoption of its National Malaria Elimination Strategy (NMES) and the flagship 1-3-7 surveillance and response approach in FY16 (1 October 2015 to 30 September 2016).2 The country reported just 2898 cases in FY21, representing a 91% reduction since the year prior to NMES adoption3 and earning Thailand a place as one of eight new countries announced by the WHO as having the potential to eliminate malaria by 2025.4 Malaria cases have become increasingly clustered into three distinct transmission zones in Thailand: the east (along the border with Cambodia), south (adjacent to Malaysia) and west (along the border with Myanmar). The drivers of this success are thought to be a combination of human and demographic factors rather than environmental factors alone.5
As GMS countries advance towards the malaria elimination goal, passive case detection is complemented by more intensive and targeted active case detection (ACD) methods to interrupt and prevent onward malaria transmission.6 7 ACD strategies require health workers to actively identify and test populations that are at increased risk of malaria infection due to geography or behaviour.8 By seeking malaria cases among people who do not present to health facilities for various reasons, including lack of symptoms or difficulty accessing health services, ACD supports early case detection and management.9 It is a useful elimination strategy that is complementary to passive case detection, as passively detected cases and resulting case investigations can help identify areas or populations in which ACD could be most useful. ACD also provides an indication of the strength of a surveillance system to capture all cases, which is particularly important in a context with high Plasmodium vivax prevalence, and it supports credible documentation of malaria-free status.
In 1975, Thailand expanded its network of malaria clinics and concurrently launched ACD, with support from the United States Agency for International Development,10 to complement passive case detection through both vertical malaria-specific facilities and in general health facilities.11 ACD is usually planned and conducted by malaria officers or by village health volunteers (VHVs), who play an essential role to expand access to malaria services in rural and underserved communities. ACD comprises proactive ACD (PACD), which seeks new cases among areas and populations deemed high risk and reactive ACD (RACD), which targets people with known risk due to proximity with passively detected index cases. ACD is implemented in endemic areas with suitable vectors, comprising a broader geography than many malaria elimination activities focused on active foci areas only.
Thailand uses three PACD approaches: special case detection (SCD), mobile malaria clinics (MMCs) and fixed-schedule malaria clinics (FSMCs). SCD responds to unusual events, such as changes in population movement or unexpected increases in incidence. MMCs are ad hoc events reserved for remote, high-endemic areas; these events were initially designed under The Containment Project as a strategy to curb the spread of drug-resistant parasites by engaging mobile populations.12 FSMCs are planned events to extend case detection services from facilities into communities (table 1).
Description of Thailand’s various case detection methods
There is no protocol for PACD in Thailand, so determination of both methods and screening criteria is decentralised. Malaria officers are expected to use routine surveillance data to identify high-risk populations before and during the peak transmission season. Furthermore, PACD is usually included in annual planning and limited to two administrations per year, limiting flexibility in its application.
As part of Thailand’s 1-3-7 surveillance and response strategy, which is detailed in national protocols and the scientific literature,2 3 RACD is conducted within 7 days of identifying a confirmed index case to prevent onward transmission. Depending on the focus classification where the index case was found, RACD will either consist of a case investigation survey (CIS), which traces relatives, neighbours and community members within 1 km of an index case, or a mass blood survey (MBS), which screens all community members in a focus where an indigenous case was confirmed but unexpected, such as due to the absence of a suitable vector or long time without indigenous cases (table 1).
ACD has been a standard ‘last mile’ component of malaria elimination programmes where passive case detection is not sensitive enough to serve all surveillance needs.13 RACD specifically has been implemented in 13 of Asia Pacific region’s 14 countries, including China, which reached malaria-free status in 2021.14 15 Despite its wide implementation and acknowledged contributions to malaria elimination, ACD requires substantial labour and resource inputs, since methods rely on screening—often large—numbers of people. RACD yield in GMS countries ranges from just 0.13% to 1.65%, varying by study site and testing method.7
A meta-analysis of ACD studies showed that the average costs were US$38.63 per person tested and US$32.07 per case detected under RACD and US$4.79 per person tested and US$37.80 per case detected under PACD.16 A study from a low-transmission setting in Indonesia estimated microscopy-based RACD cost US$1178 per index case17; this is relevant for Thailand, where nearly all cases are confirmed by microscopy. ACD typically costs more than passive surveillance; however, the relative cost-effectiveness varies by context, including differences in vector distribution, community activities and malaria incidence rates.6 Thus, accurate identification and targeting of high-risk populations and areas where malaria cases are likely to be found, plus high-quality deployment of ACD methods including appropriate timing and duration, is crucial for efficient and effective use of resources.
ACD has diminishing gains as malaria burden decreases, and WHO recommends a varying mix of PACD and RACD along the elimination spectrum.18 Since FY20, Thailand’s incidence has been very low, at <0.1 per 1000 population; however, there is no known threshold correlated with effectiveness of RACD and PACD strategies for malaria surveillance.6 7 It is unclear if these resource-intensive surveillance strategies are yielding the same output as in the past and whether the ACD programme could be further optimised in response to Thailand’s changing epidemiology.
This study aimed to document the evolution of the ACD programme as Thailand aims for malaria elimination by 2024 and to assess the potential to optimise both PACD and RACD. The study examined ACD implementation, temporal and geospatial differences in ACD yield, and a summary of estimated costs per ACD instance. As ACD requires significant investment in staff, time and funds, these findings could support improved targeting and implementation of ACD, thereby increasing efficiency and releasing resources for other emerging priorities. This is particularly important with the anticipated fiscal contraction of public health expenditures in the aftermath of the COVID-19 pandemic.
Methods
Data sources
This analysis used two epidemiological datasets (case level and foci level) from Thailand’s Division of Vector-Borne Diseases (DVBD), in the Department of Disease Control of the Ministry of Public Health. The case-level dataset included all malaria cases confirmed by microscopy, rapid diagnostic test (RDT) or PCR reported in the national Malaria Information System from October 2014 to September 2021, representing FY15–FY21. The study used fiscal years because Thailand’s malaria programme and database are based on fiscal year targets.
The costing analysis collated the direct incremental financial costs from the perspective of the Thai malaria programme for one instance of ACD per focus. The time frame and analytical horizon for the analysis both equal the time spent in the field for each instance of the activity, which is up to 2 days. Based on the availability of unit count and unit cost data versus top-line budget data, we used a combination of a bottom-up ingredients approach—our preferred method—and a top-down activity-based approach.19 We divided costs according to the unit by which they accrued to the project: costs per sample gathered, costs per person-day of activity or costs per person-day of travel. We gathered most cost data from the budgets prepared for the Global Fund to Fight AIDS, Tuberculosis and Malaria (Global Fund) and supplemented these through discussions with DVBD staff that had experience conducting ACD.
Data analysis
ACD trends in Thailand
To assess trends over time, ACD yield was defined as the proportion of cases detected among total screenings. Variation in the yield of PACD and RACD was also documented by case finding strategies and screening criteria. We conducted parametric t-tests for mean difference in yield between foci with recent malaria transmission (A1 and A2 foci classifications, representing the presence of an indigenous case within the last 3 years) and foci without recent transmission (B1 and B2 classifications) but with the presence of indigenous cases in the current fiscal year.20 To ensure high-quality data and meaningful results, we verified, cross-checked and cleaned data using IBM SPSS Statistics V.25 (IBM SPSS Statistics for Windows, V.25.0., IBM). The final dataset comprising all methods of case finding over the study period excluded 3588 malaria cases (4.64%) with unknown parasite identification. An additional 818 cases (1.06%) did not have a case finding method available in the database and, therefore, were also dropped.
Spatial variation in ACD implementation
To compare the geographical variation of PACD from FY17 to FY21 (ie, after the 1-3-7 strategy was launched), we used an intercept-only generalised linear regression model (binomial distribution) with yield as the dependent variable.21 The model allowed random intercepts for tambons (subdistricts), nested within districts, nested within provinces. We calculated the overall intraclass correlation coefficient (ICC), representing the tendency for observations within groups to be similar.21 To quantify the degree of variation in yield at different geographical levels, we also compared ICCs at tambon, district and province levels. We then mapped total numbers screened through PACD and PACD yield by tambon in the three main transmission areas. To investigate the strength of geographical variation in the implementation of RACD, we ran four generalised linear regression models (binomial distribution), with dependent variables aligned to the 1-3-7 strategy: (1) the proportion of RACD activities conducted out of the total number required, (2) the proportion of RACD activities conducted on time (ie, within 7 days of case notification), (3) the proportion of cases investigated on time (ie, within 3 days) and (4) the proportion of cases reported within 1 day of detection. Models 1 and 2 included all foci where RACD activities were required (n=2344), and models 3 and 4, which serve as comparisons for other components of the 1-3-7 strategy, included all foci with reported cases (n=3340). As above, we calculated the overall and group-level ICCs for all models. Geographical variation in the implementation and yield of RACD activities was mapped by tambon. Analyses were conducted in R (V.4.2.1).
Cost estimates
We gathered all costs in their initial transaction currency and year, inflated them in their initial currency to their 2021 values using the consumer price index,22 23 and then converted them to 2021 USD at a conversion rate of THB31.98 to US$1.24 This order of operations allows us to account for different rates of inflation between USD and THB. To account for depreciation of capital goods, such as microscopes, we used a baseline discount rate of 0.45% based on an average 1.68% rate of return to a 10-year Thai Government bond in 202125 and a 1.23% average rate of inflation in Thailand for 2021.19 22 This approach follows the best practices for costing malaria programmes as outlined by Larson et al,19 but creates quite a low discount rate compared with the 3%–5% often suggested in health economics literature, so we provided for a variation in the discount rate of 0%–7% in our sensitivity analyses.26
For personnel costs, we considered staff salaries and benefits and divided them equally across the 240 workdays in Thailand in 2021.27 In addition, all personnel, except for military escorts, receive annual training on general malaria elimination activities from the DVBD; we similarly divided the costs of these annual trainings evenly across the working days. Together, these training and salary costs compose the cost per working day of each personnel member. Please see online supplemental table 1 for the full list of line items included in our calculations.
Supplemental material
Costs were gathered in Excel spreadsheets. Preliminary analyses were conducted in R (V.4.2.1), and final calculations were carried out in an Excel spreadsheet. To see the exact formulae and calculations used, please see online supplemental file 1.
Supplemental material
Patient and public involvement
This study highlights use of Thailand’s routine health information, with no primary data collection. As such, patients were not engaged nor involved in this work.
Results
ACD trends over time in Thailand
In Thailand from FY15 to FY21, among the 6 292 302 patients who received malaria blood testing, 66 502 (1.06%) were positive for malaria. ACD, compared with passive case detection, accounted for 73.75% of all blood tests and 5.53% of all confirmed cases (range 4.54%–6.31%). The test positivity rate (TPR) of ACD methods was 0.08% (range 0.14%–0.03%), compared with 3.81% for passive case detection (range 5.62%–1.93%) (figure 1).
Malaria cases and test positivity, FY15–FY21.
FY15 represents a baseline year before the implementation of RACD under the 1-3-7 strategy, which launched in FY16.2 TPR declined over time, with a spike in FY17 for ACD as the 1-3-7 surveillance strategy took full effect. Previously, RACD was conducted for 100–150 people (approximately 20–30 households) or within a radius of 1–2 km. In FY16, reduced malaria burden narrowed RACD screening to 50 people (approximately 10 households) within 1 km. The policy change was reflected in the number of blood samples screened, which notably decreased from 251 446 in FY15 to 169 363 in FY16. During FY15–FY17, RACD TPR was 0.24% (range 0.19%–0.26%); under the revised screening criteria launched in FY17, TPR for FY18–FY21 was 0.11% (range 0.07%–0.25%).
PACD and RACD yields have declined alongside malaria incidence in Thailand (figure 2).Whereas PACD and RACD contributed nearly equal proportions of confirmed cases at the start of the study period, by FY21 PACD represented just 32.41% of ACD cases, with 0.01% test positivity (range 0.09%–0.01%). Reciprocally, RACD increased from 55.14% to 67.63% of ACD cases (0.26%–0.07% test positivity).
Malaria incidence and PACD and RACD yields (%), FY15–FY21. PACD, proactive active case detection; RACD, reactive ACD.
Among all five ACD methods in Thailand, SCD made up the largest proportion of malaria tests conducted (51.46%), followed by CIS (24.77%) and MMC (17.78%) (figure 3A). Regarding the number of confirmed malaria cases identified, CIS (51.13%) and MMC (22.36%) were the highest yielding ACD methods, while FSMC (3.32%) was the lowest (figure 3B). PACD methods (SCD, MMC and FSMC) are shown in purple tones and constitute a higher proportion of tests, while RACD methods (CIS, MBS) are shown in blue tones and comprise a higher proportion of confirmed cases.
Contribution of malaria tests and confirmed cases, by ACD method, FY15–FY21. ACD, active case detection; CIS, case investigation survey; FSMC, fixed-schedule malaria clinic; MBS, mass blood survey; MMC, mobile malaria clinic; PACD, proactive ACD; RACD, reactive ACD; SCD, special case detection.
PACD test positivity varied by method across the study period, with decreasing yield over time: SCD (0.07%–0.01%), MMC (0.11%–0.02%) and FSMC (0.23%–0.00%). RACD test positivity also varied: CIS (0.24%–0.07%) and MBS (0.78%–0.05%) (table 2).
Malaria tests and test positivity by active case detection method, FY15–FY21
As expected, both PACD and RACD contributions were a significantly higher yield among active foci (A1 and A2) than other foci with a confirmed index case (B1 and B2) (p<0.05). There were also about triple the number of people screened in active foci than other foci (3 444 484 vs 1 152 809). PACD yield in active foci from FY15 to FY 21 was 0.06% (range: 0.09%–0.02%; total screened: 2 406 658), compared with 0.02% in other areas (range: 0.08%–0.01%; total screened: 937 137). RACD yield in the same time period among active foci was 0.17% (range: 0.22%–0.07%; total screened: 1 037 826) and among other foci was 0.15% (range: 0.09%–0.05%; total screened: 215 672) (table 3).
Malaria tests and test positivity by active case detection method and foci classification, FY15–FY21
Geographical variation in ACD implementation and yield
A hallmark of Thailand’s malaria elimination programme is decentralisation of funding and decision-making, allowing for flexibility in how the ACD programme is targeted and implemented. We analysed how PACD and RACD implementation and yield differed in these three areas during the era of 1-3-7 surveillance and response. From FY17 to FY21, PACD yield was 5.94 per 10 000 targeted (0.06%) and total RACD yield was 16 per 10 000 targeted (0.16%). RACD activities appear to be operating appropriately with 72.36% of cases reported within 1 day, 90.34% investigated within 3 days, 252 of 285 (97.72%) of required RACD implemented and 196 of the 252 (76.03%) having occurred within 7 days. These adherence rates are improving over time.2
Geographical variation in PACD implementation and yield
The model of PACD yield by tambon showed an overall ICC of 0.481, indicating that almost half of the variance in PACD yield was explained by the grouping structure of tambons, districts and provinces. Specific values of ICC were 0.196, 0.217 and 0.067 at tambon, district and province level, respectively. This indicates that foci with similar yield tended to cluster within tambons and districts.
Although relatively high numbers of people were screened through PACD in the eastern region, the yield was less than 0.01% in nearly all tambons. Yield exceeded 0.10% in only 6 tambons, 2 of which recorded fewer than 10 people screened (figure 4A). In seven tambons, over 1000 people were screened. PACD was implemented more aggressively in the south, with several tambons screening more than 10 000 patients from FY17 to FY21 (figure 4B). There was a cluster of several tambons where the yield was >0.10% during this period, while the yield appeared to be lower in eastern and northern tambons along the region’s periphery. PACD was also implemented thoroughly in Thailand’s western provinces, particularly along the western border, where malaria incidence is highest. The yield was generally lower than in the south but higher than in the east. Tambons in the western region that had higher yields showed a lower degree of spatial clustering compared with those in the south (figure 4C).
Total screened and yield of malaria cases detected through PACD, FY17–FY21. PACD, proactive active case detection.
Geographical variation in RACD implementation and yield
Table 4 shows the variation in the implementation of RACD at tambon, district and province levels. Variation was higher for the proportion of RACD activities conducted out of the total number required, and was lower for the proportion of RACD conducted within 7 days. The proportion of RACD activities conducted out of those required varied mostly within tambons, while the other model on timely RACD implementation had a higher proportion of variance at province level.
Strength of spatial variation in implementation of RACD activities
In model 1, the ICC of around 0.49 indicates that approximately half of the variance in the proportion of RACD conducted is explained by the structure of tambons nested in districts nested in provinces. The remaining 50% of variance reflects random variation or the influence of covariates which were not included. Higher values at tambon level compared with district and province indicate a higher share of this variation within tambons (eg, foci are similar within tambons), whereas higher values at higher levels indicate variation between foci within tambons, but similarity (or clustering) within higher levels. Models 3 and 4 offer comparisons for other components of Thailand’s 1-3-7 surveillance strategy.
The yield of case detection through RACD was less than 0.01% in most tambons in eastern Thailand from FY17 to FY21, although there were three tambons with a yield over 0.50% (figure 5A). In the deep south, tambons in which the RACD yield was above 0.10% tended to cluster together, while most tambons on the periphery of the ACD area had a yield of <0.01(figure 5B). In western Thailand, the yield was above 0.01% in most tambons where this activity was implemented, and tended to be higher along the Myanmar border (figure 5C).
RACD yield by tambon, FY17–FY21. RACD, reactive active case detection.
ACD costs
Costs per unit
The financial inputs, including unit costs and unit counts for each line item, appear in online supplemental table 1. These values are discounted where appropriate and summed together to generate the costs per sample, per slide, per day and per trip reported in table 5.
Summary unit costs for active case detection in Thailand (2021 USD)
Costs of PACD per instance
The costs of ACD vary as a function of the number of personnel required, the number of people who must travel for the activity, and the number of people screened. Due to the decentralised nature of Thailand’s malaria elimination programme, it is challenging to quantify an expected cost for an ‘average’ instance of each ACD method, particularly for PACD activities, which unlike RACD activities, vary widely in practice. Given this high degree of variability, we provide an Excel-based calculator (online supplemental file 1) into which readers can input their own values to calculate the expected costs of ACD.
As an example for PACD costs, if a health worker takes about 15 min to administer an RDT, 1 day of an MMC staffed by one clinic worker who travelled to the site plus two local VHVs could administer 252 RDTs at an expected maximum cost of US$213.39 per day per focus, or US$0.70 per person tested. If the clinic was staffed solely by two VHVs administering 168 tests per day, costs would change to US$119.89 or US$0.71 per person tested.
Costs of RACD per instance
In RACD, activities are generally well defined. The expected cost of a standard CIS, in which a team of two malaria clinic workers travel to a focus for 2 days and gather microscopy samples from 50 households with an average household size of 2.8 people per household,28 would be US$226.31 per instance per focus, or US$1.62 per person tested. For an MBS, which requires the same team and number of person-days visiting or gathering 150 households at a central location, the expected cost would increase to US$461.78 per instance per focus, but drop to US$1.10 per person tested. If these activities used RDTs only (instead of microscopy or a combination of testing), the expected costs per instance per focus would change to US$203.81 and US$394.28, respectively. For the microscopy scenario, varying the discount rate from 0% to 7% would vary the expected cost from US$226.10 (US$1.62 per person) to US$229.83 (US$1.64 per person) for CIS and from US$461.16 (US$1.10 per person) to US$472.23 (US$1.12 per person) for MBS. A six-person military escort in southern provinces for the 2 days of activity would increase costs by US$45.03 per instance per focus.
Discussion
Synthesis of the results of this study
ACD supports early detection and appropriate management of patients in communities at risk for malaria, which is essential for successful elimination of malaria transmission. In Thailand, ACD yields from FY15 to FY21 declined, mirroring the decline in malaria incidence as the DVBD accelerated elimination interventions. Whereas PACD contributed almost the same number of confirmed cases as RACD at the start of the study period, with a yield of 0.09%, by FY21 PACD represented just 32.37% of ACD cases with a yield of 0.01%. These results align with evidence from other settings, which suggests that ACD yields fall at very low incidence.7 29–33
Comparing the scale of screening with the scale of confirmed cases shows that some methods more efficiently identify malaria patients than others, with a noticeably higher yield for RACD compared with PACD. More specifically, SCD represented more than half of the ACD tests conducted, but with a total yield of 0.03%, represented only 17.46% of all confirmed cases identified by ACD. CIS, on the other hand, represented 25% of screening but more than 50% of confirmed cases, indicating that CIS efforts are well targeted. These analyses suggest that it could be the right time in Thailand’s malaria epidemiology to revise screening criteria, concentrating on the areas and populations in which cases are most likely to be found.
Each of Thailand’s three main malaria zones showed different yields, with eastern Thailand showing the lowest yield and southern Thailand the highest. In the eastern region bordering Cambodia, the malaria caseload has fallen dramatically; however, it remains an important area for the DVBD due to the need for close drug efficacy monitoring in this region.34 Our results show that yields for both PACD and RACD were less than 0.01% in nearly all tambons in this border area from FY17 to FY21. As the local teams are quite active, it is perhaps unsurprising that PACD screening counts were high. However, the waning cases and low yield could justify a less aggressive package of ACD interventions in this region.
In the deep south along the border with Malaysia, tambons with PACD yield over 0.1% during this period and tambons with RACD yield above 0.05% tended to cluster in the centre of the province. In Yala province in particular, local teams have implemented PACD and RACD intensively and with greater frequency than in other areas of Thailand. Yala province provides a good model of accurate targeting of population at high-risk for malaria, with strong collaboration between the malaria programme staff, border police and armed forces, which has enhanced malaria prevention and elimination activities. Since a successful 2016 pilot project implemented by the vectorborne disease unit to provide PACD to military members at a high risk of malaria, in collaboration with the Southern Medical Army Centre based in Patani province, PACD has been a core part of the southern provinces’ malaria elimination strategy. Between 2017 and 2020, ACD was expanded to include civilian populations.
The higher incidence western provinces saw higher yield tambons concentrated along the border with Myanmar. Most tambons had an RACD yield over 0.01%, which is consistent with the epidemiological reality of regular malaria transmission in these areas. As local malaria transmission continues to decline, imported parasites from neighbouring countries may represent a greater proportion of future malaria cases and become the main risk for onward transmission. As the region also has high population mobility, more frequent PACD that meets the right target groups at the right time and place in a safe way, could support Thailand’s elimination goals.
The costs of ACD vary as a function of the number of personnel required, the number of people who must travel for the activity, and the number of people screened. For PACD, implementation in Thailand varies so greatly by focus that we could not quantify the expected cost of implementation for an ‘average’ instance of PACD. We hope that the provided Excel-based calculator will provide readers the ability to understand the drivers of differences in PACD costs. This calculator may also be helpful for complex settings where supplemental costs, such as security for public health workers, may be needed to provide malaria services, or for locations that use a varying mix of microscopy and RDTs for diagnosis. For RACD, which is implemented following a reasonably set structure, the costs we calculate in this paper are lower than in the few other ACD studies. However, those studies included treatment costs in their analyses, which we excluded from consideration.16 19 35
Challenges with ACD implementation
A consistent success factor in Thailand’s malaria elimination programming is decentralisation of financing and decision-making.3 This flexibility uplifts local experts, allows for diverse capacities and approaches, and gives the DVBD an opportunity to practice mutuality with subnational officers. This localisation, however, also makes it difficult to distinguish and assess how each ACD method is implemented and performing. Although there are strict protocols for RACD implementation as part of Thailand’s 1-3-7 strategy, there is not a specific protocol for PACD implementation that clearly differentiates the three methods.
Another challenge with PACD implementation is the limited flexibility in funding streams. Budgeting generally occurs in an annual cycle and can be difficult to reprogram quickly enough to respond to an epidemiological change identified by Thailand’s real-time surveillance system. The DVBD budgets for two rounds of PACD per province each year, and it can be difficult to develop new plans and distribute adequate per diems for overnight work in remote areas during special ACD deployment.
Next steps to optimising ACD for elimination
The results of this study suggest that the utility of current ACD strategies is diminishing as burden reduces in Thailand, warranting a new strategic plan to accelerate towards elimination. Further analyses could shed light on how to optimise the mix of PACD and RACD, accounting for the variation in methods, subnational epidemiological contexts and costs.
Several tambons recorded PACD yield under 0.01%, indicating poor return on investment. Waning PACD yield could be an indication that this strategy is no longer relevant for Thailand’s epidemiology, but it more likely indicates an opportunity for better targeting PACD methods and more responsive implementation. Local teams are tasked with planning their own PACD sites and implementation. A protocol containing specific guidance about which surveillance data to review to generate subsequent evidence-based and systematic decisions to sharpen the targeting, timing and implementation of PACD could renew utility in this elimination strategy. The protocol development process could also be an opportunity to streamline redundancies among the three PACD methods, which are distinct on paper but overlap in real-world settings.
PACD is most likely to remain useful only in specific microcontexts in Thailand. Rather than a national policy, PACD could be alternatively implemented to maximise yield and reduce wastage, with individuals targeted based on identifying the networks to which they are affiliated. For example, imported cases could trigger screening of fellow travellers from malaria-endemic areas using snowball sampling or a variation on time-location sampling that would be safe and appropriate for this context.5 36 Using Thailand’s extensive VHV network could support higher quality PACD implementation by offering more flexible times and locations to reach potential patients per their convenience.
Moving forward, RACD could be adjusted where resources are scarce by limiting inclusion criteria to more narrowly defined high-risk populations. In very low transmission settings such as eastern Thailand, for example, risk may be related to demographic or behavioural characteristics (ie, military members or forest goers). In such situations, deploying RACD using demographics-based screening criteria, rather than geographical criteria, could increase efficient use of resources while maintaining a very low risk of missed cases.37 This future risk analysis could also consider key vector control interventions coverage, as gaps would indicate higher potential for onward transmission. Reactive drug administration is unlikely to be part of Thailand’s future elimination programming due to the complexities of radical cure for P. vivax, which is responsible for nearly all infections.
Ultimately, ACD can continue to contribute to Thailand’s malaria elimination programme but with more deliberate targeting, guided by the country’s high-quality surveillance data to balance known operational costs.
Limitations of the study
This study used Thailand’s national routine surveillance data as the main data source. Data completeness is very high for this type of data source; however, as reported in the results section, completeness is less than 100%.
The costing analysis presented in this paper has two key limitations. First, we relied on budget data and expert opinion for costing data. Although these are less reliable sources of funding, the DVBD has been implementing ACD for several years, and so we find it reasonable to assume that their budgets and experience track with actual expenditures. Second, although travel and lodging costs vary by geography, the DVBD does not keep records of actual travel expenditures; instead, all we could gather were the flat rates budgeted for travel. We hope that the provision of the spreadsheet used to generate expected costs given different inputs will allow programme designers, both within and outside of Thailand, to adjust our example estimates to better reflect their own context.
Conclusions
The DVBD has successfully implemented a broad and detailed ACD portfolio as part of its comprehensive transition from a malaria control to a malaria elimination setting. ACD yields are declining alongside decreasing malaria incidence, with a noticeably higher yield for RACD compared with PACD and significant geographical differences by tambon. This documentation of ACD implementation helps identify opportunities to simplify the complexity of Thailand’s various ACD methods and to sharpen targeting and implementation. A PACD protocol could outline a method to systematically identify specific microcontexts where PACD could meaningfully curb transmission in lieu of the current national policy. Similarly, further analyses could offer details on how to shift from geographical-based screening criteria to demographic or behavioural criteria for RACD. This work will support optimising the mix of PACD and RACD, accounting for the variation in ACD methods, subnational epidemiological contexts and costs.
Data availability statement
Data are available in a public, open access repository. The data supporting the conclusions of this article are available in the Malaria Online repository, http://malaria.ddc.moph.go.th/.
Ethics statements
Patient consent for publication
Acknowledgments
The authors recognise all local public health teams and partners who have contributed to planning and implementing active case detectionunder the leadership of Thailand’s Division of Vector Borne Diseases, Department of Disease Control, Ministry of Public Health. A special thank you is reserved for the Sa Kaeo Provincial Health Office and Mukdahan Vector-Borne Disease Control Center for sharing specific experiences in their jurisdictions.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Alberto L Garcia-Basteiro
Contributors SK, SN, PS and JAS conceptualised the manuscript. SN, HS and RS analysed the quantitative results, and HS and DB analysed the spatial results. SN and JAS led the writing of the manuscript. All authors, including CS, SP, WC, DG, JK, RT, NP and DS provided input and review. All authors read and approved the final manuscript. JAS is responsible for the overall content as guarantor.
Funding This study was made possible by the generous support of the American people through the US President’s Malaria Initiative (PMI) and United States Agency for International Development (USAID), under the terms of Cooperative Agreement AID-486-LA-15-00002 for Inform Asia: USAID’s Health Research Program.
Disclaimer The contents of this article are the responsibility of the authors and do not necessarily reflect the views of USAID, PMI, or the U.S. Government. DG is a staff member of the WHO and is responsible for the views expressed in this publication, which do not necessarily reflect the decisions or policies of WHO.
Map disclaimer The inclusion of any map (including the depiction of any boundaries therein), or of any geographic or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied.
Competing interests The authors declare that they have no competing interests.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.