Theories about the etiology of EMF have tried to explain the condition's unusual geography and pathology. The apparent concentration of EMF in the tropics has led to a search for infectious or nutritional causes. In particular, the similarity of EMF lesions to those in Löffler endocarditis and carcinoid heart disease has suggested a connection with serotonin or eosinophil toxicity.
Unfortunately, research on EMF peaked prior to the diffusion of echocardiography in the much of the tropics 
. The lack of non-invasive imaging restricted studies to small autopsy or angiocardiographic series 
. Descriptions of the clinical progression of the disease suffered from lack of diagnostic confirmation as well 
. Expansion of echocardiographic referral centers and the development of a sub-Saharan heart failure registry will do much to clarify the epidemiology of EMF in this region 
At present, only a few investigators have tested the proposed causes of EMF. Early enthusiasm for the role of serotonin in a plantain-based diet waned by the early 1970s 
. Encouraged at first by the demonstration of high 5-hydroxyindole-acetic acid (5-HIAA) levels in the urine of West and East Africans, this work culminated when McKinney and Crawford fed plantains to guinea pigs, rats, and Patus monkeys 
. They could not reproduce typical EMF lesions. With the finding that serum 5-hydroxytryptamine (5-HT) levels failed to rise in EMF patients fed a diet of plantains in Nigeria, investigation on this hypothesis ceased 
The eosinophil hypothesis gained prominence in the 1960s with reports of eosinophilic endomyocardial disease among European visitors to tropical regions 
. At the same time, Ive and Brockington in Nigeria found filariasis (onchocerciasis or loiasis) rates approaching 100% among 42 patients with angiographic EMF compared with 44% of 115 controls (p
. The suggestion that helminth-induced eosinophilia precipitated a tropical variant of the eosinophilic heart disease known as Löffler's found support in later accounts of helminth-associated EMF in natives and visitors to sub-Saharan Africa 
The case for the equivalence of end-stage Löffler's and EMF rests on two formal evaluations 
. The first study, published by Brockington and Olsen in 1975, compared the histology of 30 cases of Löffler's with 32 cases of EMF drawn from Uganda, Nigeria, and Brazil 
. On the basis of this work, Olsen proposed three stages of Löffler's 
. In patients with 1 to 2 mo of symptoms prior to autopsy, an eosinophilic myocarditis marked the necrotic stage. Those who died after 10 mo of symptoms had endocardial thickening and thrombosis rather than myocarditis. Among those 16 Löffler patients with more than 2 y of symptoms prior to autopsy, Olsen described a final fibrotic stage that he found identical to EMF. In a clinical and echocardiographic study published in 1983, Davies and colleagues confirmed these findings 
. In 11 patients from the United Kingdom on the one hand, and 47 patients from India and Brazil on the other, they found no significant differences between fibrotic stage Löffler's and EMF.
Endomyocardial biopsies have failed, however, to demonstrate an eosinophilic myocarditis in EMF. In a series of 49 patients with EMF who underwent biopsies in Uganda, none had tissue eosinophilia despite early presentation in several cases 
. Attempts to reproduce the Nigerian filariasis findings in other small studies have also failed to show a difference in prevalence of parasite exposure or eosinophilia between EMF cases and controls 
. In Uganda, one study found that 60% of echocardiographic cases had at least mild eosinophilia compared with 10% of controls (odds ratio 4.6) 
. Another small study from this country has not shown a difference in rates of eosinophilia 
More recently, Andy and colleagues in Nigeria have argued in favor of the helminth-driven eosinophilia hypothesis. In a fascinating study weakened somewhat by lack of diagnostic confirmation, the investigators found an inverse relationship between eosinophil levels and duration of EMF disease 
. In 89 cases of EMF, only 20% of patients who presented within 6 mo of symptom onset had normal eosinophil concentrations.
Aside from inconsistencies between the pathology of Löffler's and EMF, the mismatch between the geography of EMF and the ubiquity of parasite-induced eosinophilia calls into question the relationship between these entities 
. Despite the burden of tropical pulmonary eosinophilia on the basis of lymphatic filariasis in Southeast Asia, for example, these countries have not reported much EMF. Rural Haiti has not reported any EMF cases, despite an active echocardiography service at Deschapelles in the Artibonite Valley 
. Ecological research of disease causation leaves much room for confounding.
Some have sought a more direct connection between EMF and malarial infection. EMF cases in Uganda have disproportionately come from Rwanda-Burundi immigrant families 
. While the poverty of these migrants confounds association, others have suggested that movement from zones of lower to higher malaria prevalence might hold the key. Following van der Geld, during the late 1960s Shaper and colleagues found increased levels of anti-malarial and anti-heart antibodies among these migrants and EMF cases in particular 
. Ziegler, Patel, and colleagues identified a series of familial cases of EMF among Rwanda-Burundi migrants who had massive splenomegaly, a condition associated with malaria-induced immune hyper-reactivity 
. While the prevalence of plasmodial species does not match the geographic distribution of EMF, these findings point to changes in immunity as a possible pathway from malaria to endocardial disease.
In a separate line of inquiry in Kerala State, India, the high prevalence of EMF along a coastal zone free of filariasis has led investigators to pursue a geochemical hypothesis. Valiathan and Kartha have speculated that cerium or thorium present in monazite deposits may explain regional variation in EMF prevalence in this region 
. No empirical studies have yet come forward to support this theory.
Investigations into nutritional factors in EMF have focused on a possible connection with cassava toxicity. A case-control study from Uganda has shown an association between EMF and markers of poverty such as farming, lack of shoes, and cassava-based diets with little animal protein 
. Some have suggested that cerium-mediated cassava toxicity in the setting of protein deficiency may play a role in the pathogenesis of EMF 
. Despite the known role of cyanogens from improperly processed cassava in konzo, an upper-motor neuron disease reported from Central and East Africa, cardiac manifestations have not had a part in these outbreaks