This pooled analysis from seven geographically dispersed surveillance sites in Thailand demonstrates that parasite clearance time after AM has increased between 1997 and 2007. In particular, Trat, near the Cambodian border has had a dramatic increase: the prevalence of parasitaemia on day 2 was 0% in 1997 and steadily increased to 43% in 2007. In comparison, in Tak, near the Burma border, the prevalence of delayed parasite clearance has mostly been under 10%. Of note, both initial parasite density and age changed by region and time, which may be due to differences in endemicity, environmental factors, demographics or possibly due to changing migration patterns. After controlling for these factors, day-2 parasitaemia was still increasing over time and remained higher in the east.
A previous study of a refugees’ camp in Tak in western Thailand demonstrated an increase in the prevalence of day-2 parasitaemia after AM from 3.6 to 15% from 1995–2007, which is smaller than reported here
. The difference is likely due to site-to-site variability and alterations in how the study drugs were given. However, this is still much lower than the prevalence of day-2 parasitaemia reported here for eastern Thailand. A small study in Pailin, Cambodia found that artesunate monotherapy had a longer parasite clearance time and a higher recrudescence rate compared in Wang Pha, in north-west Thailand
. However, the small sample size, the differences in baseline characteristics between the groups and technical issues in outcome determination make this study difficult to interpret
. Lastly, Phyo et al. recently reported an increase in the parasite clearance half-lives along the northwestern border of Thailand, with the largest increase occurring between 2008 and 2010
While it is clear that the prevalence of parasitaemia on days 2 and 3 are increasing and are higher in the east, the significance of this finding is more difficult to interpret. The presence of parasitaemia on days 2 and 3 are predictive of AM treatment failure in individuals but they have poor sensitivity and positive predictive value. However, parasitaemia on days 2 and 3 do have high negative predictive value. Therefore, the lack of parasitaemia on days 2 and 3 implies the person has a good chance of achieving adequate clinical and parasitological response to the treatment. However, the presence of parasitaemia is less informative.
In addition, the prevalence of parasitaemia on days 2 or 3 was not related to the overall treatment failure rate. While parasitaemia on day 2 is increasing over time, treatment failure rates are relatively constant. This lack of an association could be explained by delayed parasite clearance being not the best indicator of treatment failure. Another possibility is that delayed parasite clearance time is an early indicator of the emergence of artesunate resistance, which is not yet affecting the treatment failure rates.
The disconnect between day-2 and day-3 parasitaemia and treatment failure may also be due to mefloquine resistance being the main factor influencing treatment failure but having only a minor impact on parasite clearance rates. Using molecular markers of mefloquine resistance (such as pfmdr1 genotype and copy number) may be a way to distinguish between mefloquine and artesunate decreased susceptibility and also would improve predictions of treatment failure.
When comparing day-2 and day-3 parasitaemia prevalence, there is little difference in their ability to predict AM treatment outcome. Day 2 and 3 had similar negative predictive ability but day 3 had slightly higher positive predictive power, which is consistent with previous studies
. ROC analysis suggests that neither are adequate surveillance tools for AM treatment failure.
The outcome in this analysis (28-day treatment failure) is suboptimal, as it is prone to misclassification as treatment failure can occur after this time period. However, a sensitivity analysis demonstrated that misclassification of this outcome has little influence of the results. In addition, the sensitivity and specificity reported here are similar to those calculated with a more stringent outcome measurement
Also of note, no PCR-correction was done for the patients with recurrent parasitaemia. However, malaria incidence is generally low in most of the Mekong Sub-region, especially in Thailand. The majority of cases are occupationally exposed during a jungle trip, which typically lasts over a week. Therefore, patients who were successfully followed-up weekly in these studies were unlikely to get re-exposed to malaria in the jungles. The effect of PCR correction on the magnitude of the treatment outcomes was shown to be negligible in a similar, occupationally-exposed population of south-western Cambodia
The main strength of this analysis is that this is the most extensive description of parasite clearance time following ACT reported to date, encompassing multiple sites throughout Thailand over 10 years. The main limitations are that because these data were generated from the routine drug resistance monitoring system of a national malaria control programme, metabolites for artesunate were not measured. Therefore suboptimal drug levels causing the increasing prevalence of parasites on days 2 and 3 could not be ruled out. In addition, haemoglobin E strongly influences parasite clearance time
 and therefore geographic variations in genotype could cause the regional differences in day-2 and −3 parasitaemia seen in this study. However, haemoglobin E genotypes cannot explain the increasing day-2 and −3 parasitaemia over time in the absence of massive population migration.
The presence of parasites on day 2 is more readily measured than on day 3, as often patients return any way on day 2 for directly observed therapy. Returning again on day 3 increases the cost of in vivo studies (due to incentives) without much benefit. Both day 2 and day 3 have poor sensitivity for treatment failure in individuals, as has been seen previously
. The positive predictive value is low when the prevalence of treatment failure rate is less than 20%. Since the WHO recommends a treatment regimen be changed prior to it reaching that level of treatment failure, day-2 and −3 parasitaemia may not be useful on its own. In addition, this study implies that day-2 parasitaemia cannot be used as a way to detect sites with high treatment failure rates. However, day-2 parasitaemia may be a way to increase the efficiency of surveillance for treatment failures in resource-limited areas. For example, patients that remain parasitaemic on day 2 should be closely followed for treatment failure. Patients who cleared their parasites early have a much lower risk of treatment failure, and therefore could be followed more passively. Similar data evaluating day-2 and day-3 parasitaemia prevalence and its association with subsequent treatment data may be available from other sites of routine TES monitoring and could be mined to further evaluate the usefulness of and limitations of these proxy measures.