Malaria is considered as the main differential diagnosis of acute febrile illness in the tropics, and alteration of various hematological parameters has been observed in patients with malaria. To ascertain if certain hematological parameters increase the probability of malaria in patients with acute febrile illnesses. Hospital based, prospective cohort study. All consecutive in patients with fever of less than seven days in duration were included in the study. Patients where localizing cause for fever could be determined were excluded. Hematological parameters (Hemoglobin, Red cell distribution width (RDW), Leukocyte count, and platelet counts) were determined by using automated counter, and peripheral smear examination for malarial parasite was taken as gold standard for the diagnosis of malaria. Diagnostic accuracy was measured by computing sensitivity, specificity, predictive values and likelihood ratios. The precision of these estimates was evaluated using 95% confidence intervals. A total of 184 patients were included in the study and 70 (38%) had a positive peripheral smear for malarial parasite. Thrombocytopenia alone (platelet countless than 150,000/mm3) was a predictor for malaria (Sn 60%, Sp 88%, LR+ 5.04) and in combination with anemia (Hb < 10 g/dl) it was next best parameter (Sn 69%, Sp 74%, LR+ 2.77). RDW and leukocyte count were not predictive. The conclusion of this study is that the presence of thrombocytopenia in a patient with acute febrile illness increases the probability of malarial infection.

BACKGROUND

Malaria is a major health problem in the tropicswith 300-500 million cases of malaria occurringannually, and an estimated 1.1-2.7 milliondeaths each year as a result of severemalaria.1 In malaria patients a prompt andaccurate diagnosis is the key to effectivedisease management. Clinical diagnosis, the Department of Medicine, Mahatma Gandhi Institute ofMedical Sciences, Sevagram, Maharashtra-442102,India.Correspondence:Department of Medicine, Mahatma Gandhi Institute of Medical Sciences, Sevagram, Maharashtra-442102, India. E-mail: rjoshi_mgims@rediffmail.com most widely used approach for diagnosis ofmalaria in the tropics, is unreliable because theclinical presentation of malaria is diverse, andin a tropical country it may difficult todistinguish it from viral fever, arboviralinfections, enteric fever or even leptospirosis.2Microscopic diagnosis, the established methodfor laboratory confirmation of malaria requirestechnical expertise and repeated smearexaminations. It is a valuable technique whenperformed correctly but unreliable and wastefulwhen poorly executed.2 The drug policy formalaria treatment under the National AntiMalaria Program in India is that any feverwithout any other obvious cause may beconsidered as malaria,investig considered as malaria,investigated and treatedaccordingly, to prevent death and reducemalarial morbidity.3 Especially in a chloroquineresistant era, early identification and treatmentof malaria is imperative, hence the search forother diagnostic indicators.A variety of hematological alterations likeprogressively increasing anemia,thrombocytopenia,4-5 leukocytosis orleucopenia and rarely DIC have been reportedin Plasmodium falciparum malaria.6 Thepresent study aims to detect if certain hematological indices would increase theprobability of malaria in patients with acutefebrile illness. Presence of such an indicatormay heighten the suspicion for malaria,prompting a more diligent search for themalarial parasite, and prompt institution ofspecific therapy.

SETTING AND DESIGN

Mahatma Gandhi Institute of Medical Sciencesis a rural based teaching and referral hospitalin Central India. Patients with fever, who areadmitted to the medicine wards, are usually severe cases, referred from primary healthcare providers. We included all such patientsin the study cohort.

METHOD AND MATERIALS

All consenting consecutive patients who wereadmitted to the medicine wards with fever ofacute onset (less than 7 days), between Julyand December 2003 were included in thestudy. Patient who had no fever recorded in thehospital stay, and those where a localizingcause (such as pneumonia, meningitis, skin or239 subcutaneous infections etc.) could bedetermined were excluded.All patients with fever were investigated withcomplete blood counts, serial peripheralsmears on admission and at spike of fever,chest film, serum chemistry (electrolytes,creatinine and liver enzymes) blood culture andserology for salmonella. In addition cerebrospinal fluid examination, urinemicroscopy and culture, abdominal and brainimaging was done where indicated. Thecomplete blood counts were done with anautomated counter (Coulter), and peripheralsmears were examined by a qualifiedpathologists. Pathologists were blinded to theautomated coulter results. Four hematologicalparameters were taken as index tests(Hemoglobin, Red cell distribution width(RDW), total leukocyte count, and plateletcount). Cut off value for low hemoglobin(Anemia) was taken as 10gm/dl,7 and highRDW was defined as a value more than 15%.8Total leukocyte count less than 4000/cu mm,9and platelet count of less than 150,000/cumm10 were used to define leucopenia, andthrombocytopenia respectively. Peripheralsmear positive for malarial parasite was takenas a gold standard for the diagnosis of malaria.

STATISTICAL ANALYSIS

W e used the t test for continuous variablesand the x2 test for categorical variables todetermine the univariate association of age,gender and hematological indices predictive ofmalana. Diagnostic accuracy was measured bycomputing sensitivity, specificity, predictivevalues and likelihood ratios. The precision ofthese estimates was evaluated using 95% confidence intervals. We also usedhematological parameters in combination as apost hoc analysis. All analyses were performedusing Stata software Version 8.0 and PEPI

RESULTS

A total of 184 patients of acute febrile illnesswithout localizing signs were included in thestudy, 70 (38%) of them were positive formalarial parasite on peripheral smearexamination and remaining114 patients werenegative. Plasmodium falciparum wasidentified in 64 (34%) patients, and the resthad vivax malaria. Amongst patients withfalciparum malaria, 21(32%) had cerebra malaria, three (4.6%) had acute renal failure,and other 40 (62%) had uncomplicatedmalaria. All six patients with vivax malaria hadno complications. Of the 114 patients whowere slide negative for malaria, 6 (5.2%)patients had enteric fever (positive bloodcultures for salmonella typhi), 17 (14.9%) hadprobable enteric fever (Widal titers greater than1:160), 36 (31%) patients were chloroquineresponsive and another 23 (20.1%) hadantibiotic responsive acute febrile illness.Presumptive diagnosis of viral encephalitis waskept in 32 (28%) patients (altered behavior,normal cerebrospinal fluid examination, andclinically non-responsive to antibiotics and anti-malarials). There was no difference in age, andsex profile in those who were slide positive formalaria vs those who were not Table1.Hemoglobin level and platelet counts weresignificantly lower in the patients with slidepositive malaria in comparison to other fevers.We obtained sensitivity, specificity, predictivevalues, and likelihood ratio for diagnosis ofmalaria with hematological parameters asindex test Table 2. Presence of anemia, lowleukocyte count, or high RDW had a poorsensitivity and specificity in diagnosis ofmalaria. Low platelet count (<150,000) was 60% sensitive and 88% specific for thediagnosis of malaria. This was the onlydiscriminator parameter, with positive andnegative likelihood ratio of 5.04, and 0.45respectively. We used hematologicalparameters in combination to see if itincreased the diagnostic yield. Combination ofanemia and thrombocytopenia had highersensitivity (69%), and a positive likelihood ratioof 2.77. The negative predictive value and LR-of this combination (19% and 0.40respectively) argues that malaria may be ruledout if this combination is absent. Othercombinations (High RDW &and low platelets,and low hemoglobin, low WBC count, & lowplatelets) also did not increase the diagnosticyield Table 3.

DISCUSSION

The present study demonstrates that lowhemoglobin and low platelet count are the twohematological variables that increase theprobability of malaria, by factor of 1.95 and5.04 respectively. These two variables alsoemerge useful when used in combination(Likelihood ratio 2.77). The 95% confidenceinterval for RDW however crosses one, which implies measurement of this parameter to beless precise.The low platelet count emerged as thestrongest predictor of malaria, a previousobservation which we confirm. In a study onover two thousand patients with fever, Erhartet al11 reported platelet count of less than150,000 increases the likelihood of malaria by12-15 times. Various other studies have alsofound thrombocytopenia to be commonlyassociated with malaria12,13 which resolves aftertherapy.14 The suggested mechanisms forthrombocytopenia include disseminatedintravascular coagulation, or excessive removalof platelets by reticulo-endothelial system.15Anti-Platelet IgG has also been implicated inthe pathogenesis of thrombocytopenia.16Thrombocytopenic malaria, in contrast to thenon-thrombocytopenic variety correlates with ahigher degree of parasitemia and increasedcytokine production.17The pathogenesis of anemia in malaria ismultifactorial. A complex chain of pathogeneticprocesses involving mechanical destruction ofparasitized RBC’s, marrow suppression,ineffective erythropoiesis and accelerated mmune destruction of non-parasitized RBC’shave been implicated.18 Thrombocytopenia isa common observation in falciparum malariawith spontaneous recovery on treatment.. Bothleukopenia9and leukocytosis19 have beendescribed in malaria. Increased red cellpopulation dispersions or red cell distributionwidth (RDW) has been observed in malaria,and has been attributed to the red cellresponse to malarial parasite, and correlatedwith the degree of macrocytosis.20We included all consecutive cases of acutefebrile illness in the study, and index test andgold standard were performed in all studypatients in a blinded fashion. There have beentwo important limitations, firstly the spectrumbias of a referral hospital could havecontributed to more severe cases beingincluded in the study. This could have resultedin a greater proportion of patients withthrombocytopema in our study, as comparedto the patients who would present to theprimary care physicians. Secondly mostpatients receive empirical anti-malanaltreatment prior to hospitalization, which couldbe responsible for a greater proportion of falsenegatives in our study. Clearance ofparasitemia, as well as resolution ofthrombocytopenia could have occurred aftertherapy in them. Other confounding factors forthrombocytopenia in acute fevers are sepsisand viral fevers, but a strong association ofplatelet count less than 150,000 in patientswith slide positive malaria, as compared to theothers argues for recognition ofthrombocytopenia as an important parameterwhich may be associated with malaria, itscomplications an even response to therapy

Larger studies should be undertaken toestablish its role as an indicator of responseto threpy.

CONCLUSION

Presence of thrombocytopenia in a patient withacute febrile illness in the tropics increases theprobability of malaria. This may be used inaddition to the clinical and microscopicparameters to heighten the suspicion of thisdisease, and prompt initiation of the therapy.

REFERENCES

1. WHO. WHO Expert Committee on malaria;Twentieth report 1998. Geneva, Switzerland2000.

2. WHO. New Perspectives, malaria diagnosis.Geneva 2000.

3. National Anti Malaria Program. New Delhi:Directorate of NAMP, Ministry of Health andFamily welfare 2002.

4. Kakar A, Bhoi S, Prakash V, Kakar S. Profoundthrombocytopenia in Plasmodium vivax malaria.Diagn Microbiol Infect Dis 1999;35:243-4.

5. Krishnan A, Karnad DR. Severe falciparummalaria: An important cause of multiple organfailure in Indian intensive care unit patients. CritCare Med 2003;31:2278-84.

6. Fleming AF. Hematological manifestations ofmalaria and other parasitic diseases. ClinHematol 1981;10:983-1011.

7. Warrell DA, Molyneux ME, Beales PP. Severe andComplicated malaria. 2nd (Ed). Geneva: WHOdivision of control of tropical diseases 1990.

8. Williams JW, Beutler E, Erslevm AJ, LichtmanMA. Hematology. 4th (Ed). USA: McGraw-Hill1990.

9. Lee GR, Foerster J, Leukens J, Paraskeras F,Greer JP, Rodgers GM. Wintrobe’s clinica hematology. 10th (Ed). Bethseda, Maryland:Lippincort Williams and Wilkins 1999.

10. Brittin GM. Automated optical counting of bloodplatelets. Blood 1971;38:422.

11. Erhart LM, Yingyuen K, Chuanak N, BuathongN, Laoboonchai A, Miller RS et al. Hematologicand clinical indices of malaria in a semi-immunepopulation of western Thailand. Am J Trop MedHyg 2004;70:8-14.

12. Oh MD, Shin H, Shin, Kim U, Lee S, Kim N et al.Clinical features of vivax malaria. Am J Trop MedHyg 2001;65:143-6.

13. Biswas R, Sengupta G, Mundle M. A controlledstudy on haemograms of malaria patients inCalcutta. Indian J Malariol 1999;36:42-8.

14. Song HH, 0 SO, Kim SH, Moon SH, Kim JB, YoonJW et al. Clinical features of Plasmodium vivaxmalaria. Korean J Intern Med 2003;18:220-4.

15. Beale PJ, Cormack JD, Oldrey TB.Thrombocytopenia in malaria with IgM changes.BMJ 1972;1:345-9

16. Yamaguchi S, Kubota T, Yamagishi T, OkamotoK, Izumi T, Takada M et al. Severethrombocytopenia suggesting immunologicalmechanisms in two cases of vivax malaria. Am JHematol 1997;56:183-6.

17. Park JW, Park SH, Yeom JS, et al. Serum cytokineprofiles in patients with Plasmodium vivaxmalaria: A comparison between those whopresented with and without thrombocytopenia.Ann Trop Med Parasitol 2003;97:339-44.

18. Sen R, Tewari AD, Sehgal PK, Singh U, Sikka R,Sen J. Clinico-hematological profile in acute andchronic plasmodium falciparum malaria inchildren. J Coin Dis 1994;26:31-8.

19. Sitalakshmi S, SrikrishnaA, Devi S, Damodar P,Matthew T, Varghese J. Changing trends inmalaria: A decade’s experience at a referralhospital. Ind J pathol Microbiol 2003;46:399-401.

20. Bunyaratvej A, Butthep P, Bunyaratvej P.Cytometry analysis of blood cells from malaria-infected patients and in vitro infected blood.Cytometry 1993;14:81-5