Severity of Plasmodium falciparum and Non-falciparum Malaria in Travelers and Migrants: A Nationwide Observational Study Over 2 Decades in Sweden

Abstract Background The aim was to assess factors affecting disease severity in imported P. falciparum and non-falciparum malaria. Methods We reviewed medical records from 2793/3260 (85.7%) of all episodes notified in Sweden between 1995 and 2015 and performed multivariable logistic regression. Results Severe malaria according to WHO 2015 criteria was found in P. falciparum (9.4%), P. vivax (7.7%), P. ovale (5.3%), P. malariae (3.3%), and mixed P. falciparum episodes (21.1%). Factors associated with severe P. falciparum malaria were age <5 years and >40 years, origin in nonendemic country, pregnancy, HIV, region of diagnosis, and health care delay. Moreover, oral treatment of P. falciparum episodes with parasitemia ≥2% without severe signs at presentation was associated with progress to severe malaria with selected criteria. In non-falciparum, age >60 years, health care delay and endemic origin were identified as risk factors for severe disease. Among patients originating in endemic countries, a higher risk for severe malaria, both P. falciparum and non-falciparum, was observed among newly arrived migrants. Conclusions Severe malaria was observed in P. falciparum and non-falciparum episodes. Current WHO criteria for severe malaria may need optimization to better guide the management of malaria of different species in travelers and migrants in nonendemic areas.

The aim of this study was to describe clinical aspects and outcome of imported malaria and to assess factors that affect disease severity in different parasite species using nationwide data in Sweden over 2 decades.

Study Population
Malaria is a notifiable disease in Sweden, according to the Communicable Diseases Act. All malaria episodes reported between 1 January 1995 and 31 December 2015 in the National Surveillance Database at the Public Health Agency of Sweden, were linked to hospital data through the unique national identity number carried by all residents in Sweden, or the temporary numbers given to visitors or newly arrived migrants upon presentation to hospital. All 28 reporting hospitals contributed with medical records. Additional unreported episodes were identified through microbiology departments and inquiries in the electronic medical records systems of the respective hospitals. Individuals with additional episodes related to new travel or different Plasmodium species were regarded as new episodes while relapses and recrudescences were not included in the analysis. All included episodes were microscopy confirmed, generally at the regional departments of microbiology but also by the attending infectious disease specialist.

Data Collection
Demographic, epidemiological, and clinical data were collected from medical records, including travel details and use of chemoprophylaxis, comorbidities, pregnancy, clinical presentation, health care parameters regarding length of hospitalization, treatments, intensive care unit (ICU), delay to diagnosis, outcome, as well as blood chemistry and microbiology analyses.

Definitions
Severe malaria was defined according to the World Health Organization (WHO) criteria from 2015 (Supplementary Table  1) [1], with exception of circulatory shock which was exclusively defined as systolic blood pressure <80 mmHg in adults and <70 mmHg in children under 12 years (compensated circulatory shock could not be assessed in the retrospective data). The WHO 2015 definition was extended to include P. ovale and P. malariae according to criteria defined for P. vivax. Severity was also assessed using criteria prognostic of unfavorable outcome in severe malaria, based on selected criteria including impaired consciousness, acidosis, renal impairment and shock, associated with mortality in Bruneel et al and multiple convulsions, pulmonary edema, and significant bleeding in WHO 2000 [14,15]. Health care delay was defined as number of days from first health care contact until malaria diagnosis.

Statistical Analysis
The analyses were performed with Stata version 14.2 (StataCorp, College Station, TX). In the descriptive analyses, the categorical variables were summarized by proportions, and the numerical variables were summarized by medians and interquartile ranges (IQR). In comparative analyses, categorical variables were analyzed using the Pearson Χ 2 test, or the Fisher exact test when the latter was computationally possible. The Mann-Whitney U test was used for comparing the distribution of numeric variables between 2 groups and the Kruskal-Wallis test among multiple groups. Univariable logistic regression was used to test potential risk factors for severe malaria, and variables with a P value less than .2 were included in the multivariable logistic regression model. The variables with a P values less than .05 were kept in the multivariable model. The Wald test was used to test the regression coefficients. As individual patients could appear more than once in the dataset, the standard errors were estimated with the cluster-robust sandwich estimator. P values less than .05 were considered statistically significant.

RESULTS
During 1995 to 2015, 3099 malaria cases were reported to the Public Health Agency of Sweden; additionally, 161 unreported episodes were identified through hospital data. Medical records were available for 2793 (85.7%) episodes, with missing records from all regions. After exclusion of relapse and recrudescence episodes (n = 116) and reported episodes without microbiology confirmed parasites or only gametocytes without symptoms after recent treatment abroad (n = 24), 2653 episodes were included in the analysis (Figure 1).
Characteristics of patients infected with different species are presented in Table 1. The annual number of malaria episodes increased markedly in 2014 and 2015, explained by high incidence of P. vivax in newly arrived Eritrean migrants [16]. Also, the number of severe episodes increased in 2014, due to a high proportion of severe P. vivax in this group (Supplementary Figure 1A-C).
Pregnancy was strongly associated with severe P. falciparum malaria (OR, 7.4 [95% CI, 3.0-18.2]; P < .001); however, including pregnancy in the multivariate model caused instability and a drift in ORs likely to be explained by the small sample size. Similar effect on model stability was observed for HIV positivity, therefore, to avoid confounding, all episodes with confirmed pregnancy (n = 39) or HIV infection (n = 38) were excluded from the multivariate analysis.
Origin in Sweden or other nonendemic countries was associated with severe P. falciparum (aOR, 2.  Table 3).
Among the severe non-falciparum episodes, 7/72 (9.7%) were admitted to ICU, and all 7 fulfilled criteria prognostic of unfavorable outcome. In the severe episodes not requiring ICU care, hyperbilirubinemia alone was the most common criterion in both P. falciparum (22/64, 34.4%) and non-falciparum (33/67, 49.3%). Management is further described in Table 4.

DISCUSSION
This study summarizes the clinical presentation, management, and outcome of imported malaria, using nearly complete nationwide data in Sweden over 20 years. Severe malaria caused by both P. falciparum and non-falciparum species was identified using the WHO criteria. The most severe cases fulfilling criteria prognostic of unfavorable outcome were more commonly reported in P. falciparum episodes but were also found among the non-falciparum episodes. Factors associated with severe P. falciparum malaria were age, pregnancy, HIV infection, origin in nonendemic country, health care delay, and presentation in a region where malaria is less often diagnosed, as previously described [7,8,10,13,[17][18][19]. In addition, we identified newly arrived migrants to be a risk group for severe malaria. Moreover, we identified older age and longer health care delay as risk factors for severe non-falciparum malaria.
In P. falciparum, these risk factors predicted also the most severe forms of malaria, reflected by presence of criteria prognostic of unfavorable outcome. Nonetheless, the current WHO criteria for severe malaria (2015) [1] did not always reflect disease severity considering the low proportion of ICU admission in severe episodes without criteria prognostic of unfavorable outcome compared to those fulfilling these criteria (29.0% vs 76.2%). Our findings suggest that the criteria for severe malaria need to be optimized to different parasite species to better guide the management of imported malaria.
Severe malaria is well recognized to be caused predominantly by P. falciparum, although severe forms are described in all species [3,4,20,21]. Severe episodes in P. ovale and P. malariae are more rare [4,22], and not even included in the WHO criteria for severe malaria [1]. Here, most severe and all fatal episodes were caused by P. falciparum. Nonetheless, the proportion of severe P. vivax fulfilling the WHO criteria but also criteria for poor prognosis was notable, especially among newly arrived Eritreans arriving 2014 and 2015. A smaller subset of the severe episodes was caused by P. ovale and P. malariae (using the severe P. vivax criteria), some also fulfilled criteria prognostic of unfavorable outcome and were admitted to ICU, of which there have been only scarce previous reports [4,22].
Furthermore, severe malaria was more common in mixed P. falciparum infections than in P. falciparum monoinfections. The effect of simultaneous presence of another species of Plasmodium is not fully elucidated [23][24][25]; here, our results support an increased risk of severe malaria in mixed infections, including 1 case of death, although the most common criteria were anemia and hyperbilirubinemia.
In severe P. vivax, both multiorgan failure and death have been reported in endemic areas [23,24]. Among the imported P. vivax episodes in Sweden, we could not see the same level of severity and most episodes fulfilled only 1 criterion for severe P. vivax malaria, dominated by severe anemia or hyperbilirubinemia. Severe anemia could be debilitating and might require blood transfusions. Hyperbilirubinemia has been described as a common feature of P. falciparum and P. vivax malaria, appearing as a sign of erythrocyte destruction or liver dysfunction [25]. However, hyperbilirubinemia as a single criterion for severe malaria does not seem to have a convincing impact on severity or mortality [15,17,26,27]. This is supported by the findings in our study where hyperbilirubinemia irrespective of species often appeared without other signs of complicated disease, thus it does not seem to be a suitable criterion for imported severe malaria. In contrast, the selected criteria prognostic of unfavorable outcome (impaired consciousness, multiple convulsions, acidosis, renal impairment, shock, pulmonary edema, and significant bleeding) [14,15] were strongly associated with ICU admission.
The case fatality rate in our study corresponds to previous studies in nonendemic settings [8,20], and was particularly high (7%) among nonimmune male travelers with severe P. falciparum malaria. The overall malaria mortality in Sweden is, however, not known and our data only reflect the diagnosed cases. A study in the United States suggested that 15% of the lethal malaria cases were diagnosed at autopsy [8,28].
Delayed diagnosis and antimalarial treatment are well-known risk factors for severe malaria and death [8,10]. Our analysis showed that diagnosis was delayed at least 1 day after health care presentation in nearly one-fifth of all malaria episodes, as previously observed in other nonendemic settings [29]. For P. falciparum, health care delay was associated with severe malaria, criteria prognostic of unfavorable outcome, and ICU admission. Even in non-falciparum episodes, health care delay of 3 days or more affected disease severity significantly. Health care delay was most common in young children and patients born in nonendemic countries. This further emphasizes the necessity to enquire about travel history and promptly investigate for malaria in febrile patients returning from malaria-endemic countries, which is a challenge in a country such as Sweden where malaria is a relatively rare disease.
Health care presentation outside the 2 main cities in Sweden, Stockholm and Gothenburg, was an independent risk factor for severe P. falciparum malaria. This effect remained after adjusting for age, patient origin, health care delay, and even patient delay. Because most severe episodes fulfilled WHO criteria already at hospital admission, differences in the management of malaria in hospitals is unlikely to explain this effect. Rather, some differences in health care delay not mentioned in the medical chart may have contributed to the higher morbidity.
Episodes of P. falciparum progressing to severe malaria during oral treatment had a higher parasitemia compared to episodes without deterioration. Even though the number of episodes progressing to severe malaria is small, our findings suggest that high parasitemia is a warning sign, also acknowledged in the previous WHO 2010 definition of severe malaria where parasitemia >2% was specified as a criterion for severe malaria in nonimmune individuals and >5% in the expected immune [30]. In the current WHO 2015 definition [1], hyperparasitemia >10% is defined as single criterion; hence, patients without other severe signs and parasitemia <10% are recommended oral treatment. Our findings support early consideration of intravenous treatment at parasitemia ≥2% even in absence of other severe signs, as reported in previous studies [14,31] and stated in the present United Kingdom treatment guidelines [32].
Although patients born in malaria-endemic countries account for a proportionally high number of the imported P. falciparum cases, severe malaria and death is generally recognized to be less frequent in this group [8,[33][34][35], influenced by immunity acquired from previous infections [36]. However, protective immunity appears to decline without continuous exposure [37,38]. Here, the odds of severe P. falciparum among African immigrants increased with time of residency in Sweden, confirming our previous findings within a subset of Stockholm patients [37]. Interestingly, the odds of severe P. falciparum and non-falciparum episodes in newly arrived migrants from sub-Saharan Africa were even higher. This was most pronounced in non-falciparum, where over 50% of the severe episodes occurred in newly arrived migrants, possibly caused by prolonged symptom duration or insufficient treatment for prevention of relapse infections as well as factors associated with migration, including poor health care, malnutrition, and stress [23,[39][40][41]. Within this group, the fulfilled severe criteria were mainly hyperbilirubinemia and severe anemia, but also the most severe signs were recorded. Hence, a reduced risk of severe malaria cannot be presumed in the management of imported malaria in patients of African origin, and both newly arrived migrants and immigrants with long residency should be considered as potential risk groups for more severe malaria.
The strength of this study is the large dataset and nationwide coverage, containing detailed clinical data from the majority of malaria episodes reported to the national surveillance system during 2 decades. The limitations are inherent in the retrospective design and data relying on medical records with varying level of detail. Moreover, neither single nor mixed infections were systematically confirmed by PCR, thus underdiagnosis of mixed infections with P. falciparum may have contributed to the high rates of severe malaria observed in non-falciparum infections [42].
Our study contributes to the understanding of severity of imported malaria in different species and patient categories. Further studies are needed to establish effective severity criteria, adjusted to different species, populations, and settings that in turn could contribute to an improved management of malaria. Moreover, the findings of increased risk of severe malaria irrespective of species in newly arrived migrants highlights the need for increased attention to malaria in migrants.

CONCLUSION
In this nationwide clinical assessment of imported malaria in Sweden over 2 decades, we observed severe malaria, also with the most severe signs, caused by both P. falciparum and nonfalciparum species. Moreover, we identified newly arrived migrants as a new risk group for severe malaria. The WHO criteria for severe P. falciparum and non-falciparum malaria did not effectively reflect disease severity in this nonendemic setting, and improved criteria are needed to better guide the management of imported malaria in travelers and migrants.