In Sabah, which has an area of 73,600 km² and population of 3.7 million (16), reporting of all malaria cases and associated deaths to the Sabah Department of Health (DoH) is mandatory; species are reported according to microscopy results. We obtained details of reported malaria-associated deaths during 2012–2014 from the Sabah DoH and reviewed district hospital case notes for clinical details. The study was approved by the ethics committees of the Malaysian Ministry of Health and Menzies School of Health Research.

We reviewed the Sabah DoH malaria notification database for the total number of microscopy-based P. knowlesi/P. malariae, P. falciparum, and P. vivax malaria case notifications during 2010–2014. These data were used to determine case-fatality rates (CFRs) among notified cases for each species (hereafter referred to as notified CFRs, defined as number of PCR-confirmed P. knowlesi, P. falciparum, and P. vivax malaria–associated deaths per 1,000 microscopy-based P. malariae/P. knowlesi, P. falciparum, and P. vivax malaria notifications). PCR-confirmed P. malariae infection is rare in Sabah, accounting for <1% of clinical samples diagnosed by microscopy as P. malariae or P. knowlesi (1); thus, most notifications of P. malariae/P. knowlesi can be assumed to be P. knowlesi. During 2010–2014 in Sabah, only 1 fatal malaria case, a microscopy-diagnosed P. malariae infection, lacked PCR confirmation; an adjusted CFR was calculated following inclusion of this case.

Sixteen malaria-associated deaths were reported in Sabah during 2012–2014, of which 15 were confirmed by PCR to be caused by P. knowlesi (7 cases; cases 1–7, Technical Appendix); P. falciparum (7 cases); or P. vivax (1 case). Details of 1 P. knowlesi case (case 2) were previously reported (11). The remaining fatal case was microscopy diagnosed (without PCR confirmation) as P. malariae infection (Technical Appendix).

Despite ongoing microscopy-based misdiagnoses of P. knowlesi infections, management of severe malaria in Sabah appears to have improved; all patients recognized to have severe P. knowlesi malaria on admission received intravenous artesunate as initial therapy. Although our findings clearly demonstrate the ability of P. knowlesi to cause fatal malaria despite optimal therapy, P. knowlesi notified CFRs in Sabah have fallen over the past 5 years in association with the increased early use of artesunate documented in this and other reports (6). Death from P. knowlesi malaria remains unreported in children, and all but 1 of the P. knowlesi–associated deaths in this series occurred in adults >50 years of age.

The absence of P. knowlesi–associated deaths among children, despite 373 P. malariae/P. knowlesi notified cases in this age group during 2010–2014, contrasts with the well-recognized risk for childhood deaths from P. falciparum malaria (17) and extends the lack of previous reports of either severe or fatal outcomes in children with P. knowlesi malaria (5,8,9). Furthermore, the large number of notified cases in children in this series suggests that the lack of P. knowlesi–associated deaths among children may not be due solely to the relative underrepresentation of children in previous series of P. knowlesi malaria (6,7,18,19). A lower risk for severe and fatal P. knowlesi malaria in children may be due to the previously documented strong association between age and parasitemia (6); the level of parasitemia in children is generally insufficient to cause severe and fatal disease (20). In addition, younger age may be associated with physiologic protection from severe and fatal P. knowlesi malaria, as suggested by previous findings of a lower risk of severe malaria after primary exposure to P. falciparum in nonimmune children compared with nonimmune adults (21).

Clinical and demographic characteristics for adult P. knowlesi malaria patients in this study were consistent with those in previous reports (1,6,7,18). Patients had a median age of 61 years. Because of the strong correlation between age and parasitemia, older patients are known to be at increased risk for severe P. knowlesi malaria (6). Of the 7 patients who died, 4 were female; thus the notified CFR was significantly higher among female than male patients. Although this discrepancy appears to be primarily due to the older age of female patients with P. knowlesi malaria (1), a trend toward increased notified CFRs for female P. knowlesi patients remained even after adjusting for age. This finding is consistent with the increased risk for severe P. knowlesi malaria found for female patients in some (7,18), but not all (6), previous studies. Larger studies are needed to clarify the association between sex and risk for severe P. knowlesi malaria.

The complications experienced by the P. knowlesi malaria patients in this study were generally consistent with those in other reports; hyperparasitemia, respiratory distress, shock, jaundice, and acute kidney injury were common in this and previous reports (6,7,18,22–24). Metabolic acidosis occurred in 5 patients in this series. This complication of severe P. knowlesi malaria was uncommon in a previous tertiary referral hospital study that involved early, including prereferral, use of artesunate and in which no deaths occurred (6). However, metabolic acidosis has been reported in most fatal P. knowlesi malaria cases (7,10,11,18,25), and, as with P. falciparum malaria, is likely a late complication signifying poor outcome. Acute lung injury was present at admission in 2 patients and developed after treatment initiation in all remaining P. knowlesi patients; this finding is consistent with a posttreatment inflammatory response, as previously postulated (6).

Coma has not previously been reported in P. knowlesi malaria. Although decreased conscious state occurred in 1 patient in this study, blood cultures, lumber puncture, and computed tomography brain scan were not performed, and pupillary asymmetry, reported in this patient, is unusual in coma due to P. falciparum malaria. Therefore, while decreased consciousness directly associated with P. knowlesi remains possible, alternative causes are plausible.

Microscopy-based misdiagnosis of P. knowlesi infection occurred in 6 of 7 cases. Four of the 6 case-patients had misdiagnoses of P. malariae infection; all had high parasitemia (2 had >100,000 parasites/µL of blood), which is inconsistent with a diagnosis of P. malariae infection but highly suggestive of P. knowlesi infection. Malaysia’s malaria guidelines recommend that blood films with parasites resembling P. malariae be reported as P. knowlesi/P. malariae (12); however, high parasitemia, particularly in the context of a very low statewide prevalence of P. malariae (1), makes P. malariae infection unlikely. The frequent microscopy-based misdiagnosis of P. knowlesi malaria in this and other reports (26) in Sabah highlights the need for alternative rapid diagnostic methods.

Despite the frequent misdiagnoses in this series, all patients with fatal P. knowlesi infection who were recognized as having severe malaria at admission were appropriately treated with early intravenous artesunate. In contrast, in our 2010–2011 review of malaria deaths (10), only 2 of 5 patients with severe P. knowlesi malaria received parenteral treatment; the other 3, who had misdiagnoses of P. malariae or P. vivax malaria, received oral chloroquine or sulfadoxine/pyramethamine. In the current study, 2 patients were thought to have uncomplicated malaria and received oral therapy (case-patient 5 was given chloroquine after 1 dose of intravenous artesunate; case-patient 7 was given artesunate/mefloquine). Postmortem reexamination of these patients’ initial blood films showed a parasite count substantially higher than initially reported (55,111 parasites/μL vs. 9,866 parasites/μL for case-patient 5; 263,772 parasites/μL vs. 22,666 parasites/μL for case-patient 7). Parasitemia has been shown to be a major risk factor for severe P. knowlesi malaria: in a recent prospective study, severity criteria were present in >50% of patients with >20,000 parasites/μL of blood and >80% of patients with >100,000 parasites/μL of blood (6). WHO guidelines now recommend that intravenous artesunate be used for all patients with P. knowlesi malaria and >100,000 parasites/μL of blood or, if testing for laboratory criteria for severe malaria is not available, >20,000 parasites/μL blood (14,15). The failure of oral therapy in case-patient 7 (initial blood slide reported as 22,666 parasites/μL; bilirubin not available) highlights the value of this recommendation. Moreover, these 2 cases demonstrate that parasitemia must be accurately quantified in patients with P. knowlesi malaria.

The use of chloroquine in case-patient 5 may have contributed to the poor outcome. Compared with artesunate/mefloquine, chloroquine has been associated with reduced parasite clearance time in P. knowlesi malaria (27,28) and is no longer recommended as first-line treatment for P. knowlesi malaria in Malaysia (12). In Sabah, parasite clearance time for P. vivax malaria treated with chloroquine is reduced compared with that for cases treated with artesunate/mefloquine, and treatment failures are common (29). Moreover, P. knowlesi and P. vivax are frequently confused in microscopy examination (26); hence, a unified treatment approach should be considered in Sabah, using artemisinin for all malaria cases.

Although this case series highlights the ability of P. knowlesi malaria to cause fatal disease in adults even after prompt administration of intravenous artesunate, it must be noted that the number of deaths has not increased over recent years, despite a rise in P. knowlesi malaria notifications from 384 in 2010 to 1,325 in 2014. Thus, the adult notified CFR has declined from 9.2 deaths/1,000 notifications in 2010 to 1.6 deaths/1,000 notifications in 2014. This improvement likely resulted from increased recognition of the ability of P. knowlesi to cause severe disease and to increased use of intravenous artesunate. Although intravenous artesunate has been recommended in Sabah since December 2008, intravenous quinine was still in use until at least 2010 (10). In a retrospective study of severe P. knowlesi malaria at a tertiary referral hospital in Sabah during 2007–2009, a total of 5 (31%) of 16 patients treated with intravenous quinine died. In contrast, at the same hospital during 2010–2011, none of the severe P. knowlesi malaria patients treated with early intravenous artesunate died (6). In addition, in Sabah, the increasing use of artemisinin combination treatment instead of chloroquine for uncomplicated P. knowlesi malaria (6) may also have contributed to the decline in notified CFRs, particularly for cases in which severe disease is unrecognized.

We also reported a case of fatal P. vivax malaria with ARDS, an increasingly well-recognized complication of P. vivax malaria that has resulted in fatalities (30–34). The pathophysiologic mechanism likely involves soluble mediators and endothelial damage, exacerbated by shock and leading to diffuse damage to alveolar membranes (30). As with the case in our study, most ARDS cases occur after treatment initiation (32,35), possibly resulting from an exacerbated inflammatory response to parasite killing. Most of the initial cases of P. vivax–associated ARDS were in returned travelers with single organ dysfunction and nonfatal outcome; however, more recent series from countries where P. vivax is endemic have reported ARDS cases with multiorgan dysfunction and considerable mortality (35,36).

This study had several limitations. First, the retrospective design of the case series resulted in unavoidably incomplete laboratory and clinical data. In particular, alternative diagnoses cannot be excluded in the case of possible P. knowlesi–associated coma. Second, our calculation of the microscopy-based notified CFR represents only an estimate of the true P. knowlesi–associated CFR. The accuracy of this estimate will depend on the accuracy of microscopy-based identification of all Plasmodium species, the notification rate of malaria cases, and the proportion of persons with malaria who seek care at a health clinic. We do not have data on the proportion of malaria cases in Sabah that are notified; it is probable, however, that some are not notified, so the notified CFR likely overestimates the true CFR. In addition, we cannot exclude the possibility that the reduction in the P. knowlesi–associated notified CFRs during 2010–2014 is due to an increase in the proportion of malaria cases that are notified. However, notification of malaria cases in Sabah has been mandatory since 1992, and there is no reason to suspect that the notification rate would have changed substantially since 2010. It is similarly unlikely that the proportion of P. knowlesi malaria cases diagnosed as P. falciparum malaria, and vice versa, changed sufficiently over the 5-year period to account for the observed decline in notified CFRs (1). Nonetheless, larger prospective studies involving molecular diagnostic methods are needed to obtain a more accurate assessment of the true P. knowlesi malaria CFR, including changes over time. Although we report notified CFRs for P. knowlesi, P. falciparum, and P. vivax malaria, these data may not reflect the relative virulence of each species. In this series, non-Malaysian citizens accounted for a higher proportion (5/7) of patients with fatal P. falciparum malaria than fatal P. knowlesi malaria, and it is possible that a delay in seeking care at a healthcare facility may be a confounding factor in comparing CFRs for malaria caused by these Plasmodium spp.

In conclusion, our findings show that despite increasing notifications of P. knowlesi malaria cases in Sabah, the number of fatal cases has not increased. The reduction in notified CFRs may be associated with the increased recognition of the ability of P. knowlesi to cause severe and fatal malaria and improved use of intravenous artesunate for severe malaria caused by any Plasmodium spp, as per recent policy changes (6,12). Nonetheless, this study demonstrates the ability of P. knowlesi to cause fatal malarial disease in adults, despite optimal therapy, and that P. knowlesi remains the most common cause of fatal malaria in adults in Sabah. In contrast, the study shows a notable absence of deaths among children with P. knowlesi malaria.