The present analysis strongly suggests that only a small proportion of severe cases reach tertiary care (0.2-1.6%), since a mean of 177 malaria patients per year were admitted to CMCH, the only tertiary care facility in the area, whereas the estimated total numbers of severe cases nationally is around 11,000-80,000 per year of which 80% reside in the CHT. One of the explanations could be the financial cost of hospitalization, which is high in proportion to the average income in Bangladesh and can thus be a strong disincentive against admission of less severe patients. Many of the less severe cases we probably treated in peripheral Thana or District hospitals and many cases of severe malaria probably remained untreated in the community. Of the patients admitted to CMCH, 96% had P. falciparum and 4% of patients had P. vivax, whereas mortality was only associated with P. falciparum. Previous studies in uncomplicated malaria in this region show that 70% of malaria is caused by P. falciparum and 30% by P. vivax[1, 5, 8], which emphasizes the benign nature of P. vivax infections in this region. Malaria infection was commonest in young adult males in agreement with previous data from the south of Chittagong Division
. This may be related to greater occupational exposure of this group to forest malaria
. As the breadwinners, it could be postulated that they are also the family member most likely to be supported to attend a distant hospital for potentially expensive treatment
, hence their over-representation in those screened for malaria.
The number of hospitalized P. falciparum cases at CMCH decreased since 2007, despite an increase in the number of patients screened. This decrease was particularly dramatic from 2008–2010. During this period, there have been a number of changes in malaria programmes in the community and a greater than five-fold increase in funding for malaria control
[3, 10]. A large increase in the numbers of patients who receive early antimalarial treatment has occurred in the feeder hospitals and clinics referring to CMCH. In particular there has been a doubling of ACT usage from 2007 to 2008
[3, 10] and an increase in the availability of parenteral antimalarials. In addition, a new large-scale programme of free distribution of insecticide-treated bednets
[3, 10] and introduction of rapid diagnostic tests in the community began in 2008
[3, 10]. The present study adds to the evidence that these strategies are having a significant impact. However, potential pockets of high transmission, which are mostly in remote areas
 may not have been sufficiently covered by this study. Over the past two years there have been initiatives aiming at increased availability of early treatment in more remote areas, including early intravenous therapy in the District hospitals, and as a result it is possible a smaller proportion of cases are being referred to tertiary hospitals.
There was a strong and consistent seasonal pattern of P. falciparum incidence, with a large peak between May and September each year largely coinciding with the maximum rainfall during the monsoon season (June-August). This finding is in contrast with the pattern reported in two earlier reports
[3, 6] describing one transmission peak in March-May and one in September-November, with June-August being described as ‘off-peak months’
. However, in both these publications no monthly incidence data were presented. For the incidence of P. vivax an additional peak was observed in the months from February to March, before arrival of the rains and transmission of P. falciparum, which uses the same vector system. The same seasonality has been described in vivax malaria in Hooghly District in West Bengal in India, and suggests that these are long latency relapse cases of vivax malaria
By far the majority of individuals screened for malaria were from the south of Chittagong Division with relatively few from the CHT. All the malaria positive cases were from the south of Chittagong Division in the five endemic Districts: Chittagong, Cox’s Bazar, Kagrachari, Rangamati and Bandarban. Over 80% of cases in Bangladesh are thought to be resident in the CHT
[7–9], although in this study only 12% of those referred to CMCH lived there. This is despite it being the main referral hospital for those needing more advanced care and local policy being to refer the sickest cases from all other government hospitals in the area to CMCH.
As expected from existing epidemiological data
, the probability of having malaria was highest in screened patients coming from the CHT, particularly the southern part: Lama and Alikadam Thana in Bandarban District but also adjacent Lohagara in Chittagong District. The high rate in Lohagara was mainly caused by an apparent focal epidemic of P. falciparum in 2010.
Population density in the CHT is much lower than elsewhere in Chittagong Division. The most densely inhabited areas are Chittagong City and Cox’s Bazar and their surroundings and both these areas had very few malaria cases. The highest malaria positivity rates per population density in this study were mostly in those from areas of low population density in a band from north to south through the centre of Chittagong Division. These were Fatikchari and Rangonia in Chittagong District, Kawkhali and Rangamati Sadar in Rangamati District and Bandardan Sadar, Lama, Alikadam, and Thanchi Thana in Bandarban District plus Chakaria in Cox’s Bazar District. The high rates in Fatikchari, Rangonia and Chakaria were a particular surprise as these had not been previously identified as high risk areas, although they are near to the forest fringe. Few of the affected individuals in these three areas had visited the adjacent CHT. Even though very few cases of malaria seen at CMCH lived in the CHT, this study indicated almost three times this number, over a third of the total, are likely to have become infected there. Thus the CHT are an important source of malaria both for residents and travelers, although an important proportion of malaria transmission is outside this area, as has also been described in previous studies
. Albeit lower transmission outside the CHT, the much larger population in this area contributes significantly to the malaria case load. In this study two thirds of cases had not visited the CHT during the time in which they became infected. Malaria control efforts to date have been particularly focused on the CHT
 but these data suggest a broader area should be targeted.
There was a limited number of cases from Khagrachari District, the northern third of the CHT, previously found to be the area with the highest transmission in Bangladesh
. The small District hospital in Khagrachari town is similar to those in Bandarban and Rangamati and lacks facilities for mechanical ventilation or renal dialysis, which are often needed for patients with severe falciparum malaria. The long travel time to Chittagong might discourage families and physicians from referral to CMCH.
Another underrepresented area known to be highly malarious was Rangamati District, particularly in the east near the border with India, although this area is also an endemic zone with relatively high transmission
. There are a number of possible reasons for this. A previous study showed that there is a strong preference among indigenous people in this area for seeking treatment from alternative practitioners in the first instance, although this may be different for the severely ill
. Transport from this area to Chittagong is also difficult. Between much of this area and Chittagong city is a large man-made lake, Kaptai Lake, and this can only be crossed by a long journey by boat. Road links to this area are poor, particularly in the wet (malaria) season, and from many areas the travel is long and arduous. These difficult travel conditions might encourage people to seek treatment locally. However, patients from other remote areas with similar difficulties during the wet season did reach CMCH.
Although there are many pharmacies and health centres in the CHT
, few of these can provide intravenous treatment and are very limited in their ability to provide more extended supportive care. There are larger and better equipped District hospitals in Khagrachari, Rangamati and Bandarban towns, but these cannot offer mechanical ventilation or renal dialysis, for which referral to a tertiary center is necessary. The dearth of referrals from the CHT to CMCH thus indicates there is likely to be a large burden of patients receiving suboptimal medical treatment in the periphery, and the perceived and real risks of long transportation times to a tertiary treatment centre are likely part of the explanation. For very remote areas in the wet season, it may be that the risk of transport is just too high and the emphasis therefore has to be on improving care locally as much as possible. One recent example has been the introduction of pre-referral treatment with rectal artesunate. This has the advantage that patients begin effective treatment earlier and could potentially ‘buy time’ to allow them to transfer more safely to a better equipped facility
. Expanding the availability of effective early antimalarial treatment in general, as has been occurring over the past few years
, will also mean fewer patients progressing to severe disease.
Although overall numbers were large, this study had several limitations. All data were from a single tertiary referral hospital. Data were only collected on those patients who had a malaria test by the on-site malaria diagnostic laboratory. There are several private laboratories in Chittagong who also provide malaria tests, although they charge a fee for this service. A small proportion of patients still undergo testing by these private laboratories although the vast majority of these are retested by the hospital laboratory. The study relies on the assumption that the quality of malaria diagnosis did not change significantly from 1999–2011. This is likely to be the case, as the same highly experienced staff were employed throughout and used the same techniques. It does, however also rely on the medical staff referring the same group of patients for testing during this period but data on this were not collected. Numbers of cases from the CHT and total numbers of P. vivax were small. Conclusions regarding P. vivax epidemiology are thus limited in their scope.
Since it has been reported in the most recent government report that in 2008 3.8% of P. falciparum cases in Bangladesh occurred in Chittagong District
, and we found that, in 2008, 130/165 CMCH admitted severe malaria cases were from that area, the total national number of severe malaria cases in Bangladesh was at least 130/0.038 = 3460. This figure ignores any patients with severe malaria admitted to the many other hospitals in Chittagong District, as well as those that did not reach healthcare and the actual total is thus likely to be much larger. Official data report a total number of 3591 severe cases for the whole country from mid 2008 to mid 2009, which is thus a severe underestimation
 and there is a clear need for more accurate and complete reporting. Current systems for collating this data are incomplete and many confirmed cases are missed from the official totals
[1–4]. Of particular interest would be the trends in numbers of cases, severe malaria and malaria deaths in the CHT which is essential information to assess the efficacy of malaria control measures as well as for the allocation of resources for patient care.