Risk factors for relapse of visceral leishmaniasis in Georgia


Corresponding Author Maia Kajaia, Maternal and Child Care Union, 10a Bakhtrioni st. Tbilisi 0177, Georgia. Tel.: +995 32 367878; Fax: +995 32 367501; E-mail: maiakajaia@yahoo.com


The number of relapses in patients treated for visceral leishmaniasis (VL) has increased, thus identifying prognostic factors may aid decisions on treatment. Demographic and clinical information was abstracted from medical records of patients diagnosed and treated in Georgia from 2002 to 2004. The 300 persons with VL were primarily children <5 years (73.3%), and ∼44% had delays in diagnosis of more than 30 days from symptom onset. All patients received standard therapy with pentavalent antimony (20 mg/kg/day), most for 20–25 days. Factors significantly associated with VL relapse were delay in diagnosis for >90 days (RR = 4.21, 95% CI: 1.58, 11.16), haemoglobin level <60 g/l (RR = 11.96, 95% CI: 4.12, 34.76) and age <1 year (RR = 2.36, 95% CI: 0.96, 5.80). Physician and public education is needed to reduce delays in diagnosis. Prolonging treatment for 30 days (e.g. WHO recommendation) or implementing new regimens may reduce the number of relapses.


Le nombre de rechutes chez les patients traités pour la leishmaniose viscérale (LV) a augmenté. Dès lors identifier les facteurs de pronostiques pourrait faciliter les décisions sur le traitement. Les informations démographiques et cliniques ont été extraites des dossiers médicaux de patients diagnostiqués et traités en Géorgie de 2002 à 2004. Les 300 personnes avec LV étaient principalement des enfants < 5 ans (73,3%) et environ 44% ont connu des retards de plus de 30 jours dans le diagnostic depuis l’apparition des symptômes. Tous les patients ont reçu un traitement standard à l’antimoine pentavalent (20 mg/kg/jour), la plupart durant 20 à 25 jours. Les facteurs significativement associés à la rechute de la LV étaient le retard dans le diagnostic > 90 jours (RR = 4,21; IC95%: 1,58–11,16), le taux d’hémoglobine < 60 g/L (RR = 11,96; IC95%: 4,12–34,76) et l’âge < 1 an (RR = 2,36; IC95%: 0,96–5,80). La formation des médecins et du public est nécessaire pour réduire les retards dans le diagnostic. Le prolongement du traitement à 30 jours (e.g. recommandation de l’OMS) ou l’implémentation de nouveaux schémas thérapeutiques pourraient réduire le nombre de rechutes.


Ha aumentado el número de recaídas en pacientes tratados para Leishmaniasis Visceral (LV), por lo cual identificar los factores pronósticos podría ayudar en la toma de decisiones sobre el tratamiento. Se obtuvo información demográfica y clínica de las historias clínicas de pacientes diagnosticados y tratados en Georgia entre el 2002 y el 2004. Las 300 personas con LV eran principalmente niños <5 años (73.3%), y 44% tenían retardos en el diagnóstico de más de 30 días a partir de la aparición de los síntomas. Todos los pacientes recibieron una terapia estándar con antimonio pentavalente (20 mg/kg/day), la mayoría durante 20 a 25 días. Los factores asociados de manera significativa con la recaída de LV eran el retardo en el diagnóstico de >90 días (RR = 4.21, 95% IC: 1.58, 11.16), niveles de hemoglobina <60 g/L (RR = 11.96, 95% IC: 4.12, 34.76), y una edad <1 año (RR = 2.36, IC 95%: 0.96, 5.80). La educación/información recibida del médico así como la pública son necesarias para reducir los retrasos en el diagnóstico. Prolongar el tratamiento a 30 días (e.j., recomendación OMS) o implementar nuevos regimenes antigénicos puede reducir el número de recaídas.


In the Republic of Georgia, visceral leishmaniasis (VL) historically occurred primarily as single sporadic cases. Since 1996, VL incidence has risen sharply, especially in the capital, Tbilisi (Figure 1). The increase had two causes: (i) Disease control measures were reduced or stopped because of the socio-economic crisis in Georgia after the collapse of Soviet Union. Thus, environmental conditions favourable for vectors developed, and the number of reservoirs, such as stray dogs, rose. (ii) The slow geographical spread of VL moved into a highly populated area (i.e. Tbilisi, population approximately 1.2 million). Thus, the number of people at risk of infection grew substantially. The disease is endemic in the Eastern part of Georgia, from the Azerbaijan boundary to Tbilisi and its surroundings, as a result of circulation of parasite reservoirs and vectors in natural foci (Chubabria & Zenaishvili 2002; Ponirovski et al. 2006) (Figure 2).

Figure 1.

 Number of visceral leishmaniasis cases in Georgia and Tbilisi 1965–2007.

Figure 2.

 Map of Georgia. Bold outline indicates geographical distribution of visceral leishmaniasis in Georgia.

Visceral leishmaniasis is a pure zoonosis in Georgia. Foxes (Vulpes vulpes alpherakyi Satunin) and jackals (Canis aureus Linne) are natural reservoirs (Maruashvili & Bardzhadze 1980), but dogs are the main source of disease for humans (Bardzhadze & Safianova 1979; Chubabria & Zenaishvili 2002; Ponirovski et al. 2006). Among 14 species of sandfly vectors in Georgia, Ph. Kandelaki and Ph. balanicus are the main transmitters of disease (Maruashvili & Bardzhadze 1966). No Georgian studies have been conducted to serotype Leishmania species using modern molecular methods, but based on geographical location and clinical presentation, Leishmania infantum (Sundar & Rai 2002) is considered to be the causative agent. The disease affects mainly children under 5 years of age, but rarely, adult cases are also registered (Chubabria & Zenaishvili 2002; Ponirovski et al. 2006). The incidence is highest from December through May. The seasonal differences of VL manifestation correspond exactly with the activity of vectors and incubation period of the disease.

Since 1996, the number of relapses in patients treated for VL has markedly increased in Georgia. A possible strategy to reduce the number of relapses that usually occur 3–4 months after treatment is to identify prognostic factors that could be used in clinical decision-making. Our objective was to identify risk factors for relapse of VL in Georgia.

Materials and methods

The study population included all residents of Georgia who were diagnosed with VL and treated at the Institute of Parasitology and Tropical Medicine for the first time in 2002, 2003 and 2004. All known VL cases in the country are referred to the Institute for treatment.

Patients with confirmed diagnoses of VL and without previous treatment were included in the study. Confirmation of a VL diagnosis was based on microscopically observing parasites on a Giemsa-stained smear of bone marrow (Sundar & Rai 2002; Da Silva et al. 2005) obtained by sternal (in adults) or iliac crest (in children) aspiration (Sundar & Rai 2002; Da Silva et al. 2005). VL relapse was defined as recurrent clinical symptoms confirmed by the presence of bone marrow parasites in patients previously cured for VL. Cure was defined as regression of splenomegaly and improvement of general condition. Repeated bone marrow aspiration was carried out in patients with persistent splenomegaly and those without demonstrable parasites in smears were considered cured.

Data were abstracted from medical records at the Institute of Medical Parasitology and Tropical Medicine. The abstracted data included patients’ demographic information, signs and symptoms, clinical data including laboratory test results and outcome (relapse status). Population estimates were obtained from the Georgian government Internet website (http://www.statistics.ge).

Demographic information included date of birth, gender and residence. For the analyses, age was defined by categories and residence by city or region where a patient lived.

Clinical data comprised measurement of spleen and liver enlargement in centimetres obtained by physical examination, body temperature at admission, duration of high temperature (38.1 °C or higher) as well as laboratory results such as haemoglobin (Hb) level, number of erythrocytes, leucocytes and lymphocytes in blood and erythrocyte sedimentation rate.

Horizontal and vertical sizes of spleen were determined by the S. Kandelaki splenometric method that has been used in clinical practice since 1935 at the Institute of Medical Parasitology and Tropical Medicine of Georgia (Melia & Zenaishvili 2006). Splenomegaly was categorized as: mild enlargement – less than 12, moderate enlargement – 12 to 24 and marked enlargement – more than 25. Liver enlargement was measured in centimetres from costal rib.

Other clinical data included date of symptom onset, date of VL diagnosis confirmation and duration of treatment. All variables in the study were categorized for analysis.

Data were entered into an SPSS database by Institute of Parasitology and Tropical Medicine staff and anonymized for analysis. Statistical analyses were conducted using SAS version 9.1. Descriptive statistics (frequencies and percentages) were computed for both univariate and bivariate analyses. Bivariate analyses included estimating the prevalence ratios, with 95% confidence intervals, between predictive factors and relapse of VL.

Poisson regression with robust variance estimates was used for multivariate analysis to identify the strongest predictive factors for relapse of VL (McNutt et al. 2003). All potential predictors were included in the full model, and the factors were dropped one at a time based on the largest P-values. The most parsimonious predictive model was developed where the variables were retained if the likelihood ratio test P-value was less than 0.05. To determine confounding, the reduced model and the full model were compared. Those dropped variables that substantially changed the parameters in the reduced model were considered to be confounders and added back into the model. The final model was used to interpret the prevalence ratios between factors and relapse adjusting for confounding. No interactions were included because of the small number of the outcome (n = 21); thus, this study could not address effect modification.

Institutional Review Board approvals were obtained from State University of New York (protocol number: 08-115) and the Georgian Rehabilitation Center.


Three hundred individuals were diagnosed and treated for VL at the Institute of Medical Parasitology and Tropical Medicine of Georgia during 2002, 2003 and 2004. The number of cases per year varied from 93 (2002) to 108 (2004) with an average annual incidence rate of 6.9 cases per 100 000 population. Consistent with historical trends, most cases occurred from December to June.

The majority of VL cases were diagnosed in children <5 years of age (73.3%), with about half being female (55.3%). All patients were residents of Georgia; most (65.7%) lived in Tbilisi, followed by rural areas (28%) and other urban areas (6%).

Delays in diagnosis were common; approximately 44% of infected individuals were diagnosed after 30 days of symptom onset, and delays were similar in rural and urban areas. At the time of diagnosis and admission for VL treatment, most patients had fever; 49.7% had a temperature >39.1 °C. Two-thirds of patients with a fever of ≥38.1 °C on admission achieved a normal temperature within 7 days.

Clinical signs at admission included moderate spleen enlargement in 61% and hepatomegaly in all but five cases (98%) with liver enlargement >3 cm in 46.7%. Almost all patients (96%) were anaemic on admission; 13.7% of patients had profound anaemia (<60 g/l). The majority had low red blood cell counts, leucopenia, lymphocytosis and an elevated sedimentation rate.

All patients received standard therapy with meglumine antimoniate (pentavalent antimony) at a dosage of 20 mg/kg/day. About half (54%) the patients received treatment for 20 days according to the Institute’s clinical protocol for VL at the time; however, some variations on treatment delivery occurred with 15–25 days of duration.

Seven percentage of patients relapsed within 4–18 months after treatment (mean = 9.3 months; median = 7 months). Based on bivariate analyses, age, time from symptom onset to diagnosis, clinical and laboratory findings on admission (e.g. haemoglobin level, red blood cell count, lymphocytosis, spleen size) were associated with VL relapse. Infants were more likely to relapse (16.7%) than older children (7.1%) and adults (2.5%) (Table 1).

Table 1.   Association between demographic factors and relapse of VL, Georgia, 2002–2004
Risk factorsRelapseNo relapsePR95%CIP-value
  1. VL, visceral leishmaniasis.

 <1 year616.73083.32.360.96–5.800.02
 1–4 years137.117192.91 
 5 years or older22.57897.53.500.08–1.53
 Tbilisi105.118794.91 0.22
 Other urban210.51794.41.100.14–8.07
Time from symptom onset to diagnosis
 <3 months93.723796.31 <0.001
 3–5 months615.43384.64.211.58–11.16
 More than 5 months640.0960.010.934.48–26.67

Delays in diagnosis and treatment were strongly associated with both profound anaemia on admission, suggesting disease progression related to diagnostic delay, and risk of relapse. Patients whose diagnosis was delayed by 3 months or more were more likely to have profound anaemia (HB level <60 g/l) on admission than those with an earlier diagnosis (29.6%vs. 10.0%; P = .001). Patients who were diagnosed 5 months after the onset of symptoms had a substantially higher risk of relapse (40%) (RR = 10.9, 05% CI 4.5–26.7) than those diagnosed within 1 month (Table 1).

Individuals with haemoglobin levels <60 g/l were approximately 12 times more likely to develop relapse than individuals with higher Hb levels (Table 2). Patients with red blood cell counts <2.0 × 1012/l were more likely to relapse than those with RBC 3.0 × 1012/l and more. Relapse occurred more frequently in individuals who had levels of lymphocytes exceeding 80% than in those with levels below 55% (Table 2). While not statistically significant, a longer duration of treatment appeared to be suggestive of a protective effect against relapse (Table 3).

Table 2.   Association between laboratory findings and relapse of VL, Georgia, 2002–2004
Risk factorsRelapseNo relapsePR95%CIP-value
  1. VL, visceral leishmaniasis.

  2. †Group includes three patients with Hb ≥ 120 g/l and 10 patients with Hb between 110 and 119 g/l. No patients with Hb > 110 g/l developed relapse.

Haemoglobin level (g/l)
 80 and more†42.61471  
Red blood cell count (×1012/l)
 3.0 or more33.38796.71 
Sedimentation rate of erythrocytes (mm/h)
 More than 6024.44395.60.660.15–2.970.66
 40 or less96.712693.31 
White blood cell count (×109/l)
 4.0 or more107.412592.61 
Lymphocytes (%)
 <5534.17095.91 0.03
 More than 80617.12982.94.171.12–15.71
Table 3.   Association between clinical findings and relapse of VL, Georgia, 2002–2004
Risk factorsRelapseNo relapsePR95%CIP-value
  1. VL, visceral leishmaniasis.

  2. †Splenomegaly grades according to Kandelaki splenometric method (Melia & Zenaishvili 2006).

  3. ‡Hepatomegaly in centimetres measured during physical examination.

Temperature on admission
 38 °C or less12.34397.710.46–26.850.41
 38.1–39 °C98.010492.23.500.45–25.49
 39.1 °C or more117.713292.33.38 
Length of fever 38.1 °C and more after admission
 More than 7 days712.05187.91.840.77–4.410.14
 7 days and less136.518693.51 
 <377.68592.41 0.13
 More than 3811.86088.21.550.59–4.06
Duration of treatment
 <20 days28.32291.72.150.42–11.040.29
 20 days159.314790.72.380.81–6.98
 21–24 days0011100.0  
 25 days and more43.99996.11 

Multivariate analysis identified the best predictive model for relapse of VL (Table 4). Adjusted for other factors, factors associated with relapse were age <1 year, diagnosis delay >90 days and haemoglobin level <60 g/l. The associations between risk factors and relapse of VL adjusted for multiple potential confounders were similar to the unadjusted associations.

Table 4.   Multivariate analysis of association between multiple potential risk factors and relapse of VL, Georgia, 2002–2004
FactorsRR95% CIP-value
  1. VL, visceral leishmaniasis.

  2. *Group includes three patients with Hb ≥ 120 g/l and 10 patients with Hb between 110 and 119 g/l. No patients with Hb>110 g/l developed relapse.

Age group
 <1 year2.551.22–5.360.01
 1 years or older (referent)  
Time from symptoms onset to diagnosis
 More than 90 days3.851.75–8.49<0.008
 <90 days (referent)  
Haemoglobin level (g/l)
 60 and more (referent)  


This study evaluated risk factors for relapse of VL in 300 patients diagnosed and treated at the Institute of Medical Parasitology and Tropical Medicine of Georgia. Relapse was observed in 21 patients (7.0%), among whom 19 (90%) were under 5 years of age. A similar relapse rate of VL (8.2%) was observed in a study conducted in Iran which evaluated epidemiological, clinical and therapeutic features of paediatric VL (Rahim & Ashkan 2007). Much lower relapse rates of VL (3.9% and 2.6%) were found in two studies in the Sudan (Abdelmoula et al. 2003; Collin et al. 2004). Georgia, Iran and Sudan treat patients with pentavalent antimony 20 mg/kg/day; Sudan follows the 30-day treatment recommendation of WHO. Georgia and Iran have historically treated for 20–25 days (PAHO/WHO/IDA 1996). Length of treatment appears to be the primary difference between these studies findings, strongly suggesting that the WHO guideline be followed more closely.

Three factors were important predictors of relapse of VL: delay in diagnosis for more than 90 days, haemoglobin level below 60 g/l and age <1 year. One possible explanation for these findings is that the delay in diagnosis of VL leads to disease progression (e.g. parasite replication increased) indicated by a low haemoglobin level. The shorter duration of treatment may be insufficient to kill the parasites that replicated for an extended length of time in some infected patients, particularly infants. Previous studies either did not identify predictors of relapse (Behe et al. 1994; Abdelmoula et al. 2003; Collin et al. 2004) or found similar results to this study, but the results did not reach statistical significance (Seaman et al. 1996). We also found that neither gender nor spleen size was important predictors of relapse. Clearly, interventions focused on more rapid diagnosis and effective treatment regimens are needed.

While VL has spread across southeastern Georgia, it has only become endemic in urban areas recently. Thus, it was a surprise to see that a delay in diagnosis occurred similarly in both rural and urban areas. Hence, education for physicians, particularly paediatricians and general practitioners, is indicated. Education about parasitic and tropical diseases has historically been a low priority in Georgia, but given the rising incidence of VL, widespread education efforts are needed. Based on these findings, and consistent with the clinical concerns of physicians specializing in VL treatment, improved educational efforts are being planned. While a need for professional education is indicated, education on diagnosis alone will be insufficient. The confirmatory test, bone marrow aspiration, presents another obstacle to diagnosis as many parents are reluctant to allow such a procedure on their children. Confirming the diagnosis is important given the toxicity of the treatment, thus this important barrier to timely treatment needs to be addressed. Developing communication skills that aid physicians in discussing invasive procedures with parents will be key to shortened diagnostic times and treatment delays.

Treatment alternatives that can be used for patients at risk of relapses of VL based on predictive factors by this study are also worth considering. Miltefosine, a new medication on the market, has several advantages for consideration as a first-line treatment: it is effective, orally administered, safe and cheap (Jha et al. 1999; Ganguly 2002; Sundar et al. 2002; Sundar & Chatterjee 2006). Another new antileishmanial drug, Paromomycin, an aminoglycoside which is administered parenterally, has low adverse reaction rates and is an effective treatment for VL in several countries (Guerin et al. 2002; Olliaro et al. 2005; Sundar et al. 2007a). Another option is Amphotericin B, the first-line parenteral therapy for VL in India, which arose because of large-scale resistance to pentavalent antimony therapy (Sundar et al. 2000, 2007b; Sundar & Rai 2005). Amphotericin B has a high cure rate, but it is toxic, treatment duration is long and patients need monitoring. Liposomal Amphotericin B is widely used in Europe to treat VL. Several clinical trials demonstrated that it is safe and has the highest therapeutic effect of antileishmanial drugs (Davidson et al. 1994; Bern et al. 2006). The major disadvantage of Liposomal Amphotericin B is its high cost, which makes it unaffordable for developing countries.

As in all medical chart review studies, limited availability of data has an impact on the analysis. While a strength of the study was the availability of medical records for all patients treated for VL in Georgia, some information in the records was clinically estimated (e.g. time from symptom onset to diagnosis) and other information was not consistently available (e.g. parasite load and treatment susceptibility). The sample size was insufficient to assess effect modification. Another limitation of the study was very small number of relapse in certain groups which was the reason of merging some categories in the regression analyses. While not available for this study, anecdotal conversations with physicians at the Institute suggested that physicians have recently increased the treatment regimen to 30 days, but the standing protocol has yet to be officially changed.

In summary, this study identified indicators for determining which patients are at particular risk for relapse of VL. The study might be useful in generating suggestions for more careful monitoring of such patients and making changes in VL treatment protocols, such as prolonging the treatment with pentavalent antimonials for 30 days (as recommended by WHO), implementing new drugs such as Miltefosine as first-line treatment, or treating patients with risk factors for relapse with alternative antileishmanial drugs. As VL continues to spread through the region, expanding both physician and public education is needed to reduce the time from symptom onset to diagnosis.


This research was supported in part by the New York State International Training and Research Program grants, 1D43TW007384 and 2D43TW000233, NIH – Fogarty International Center. We thank S.Virsaladze Research Institute of Medical Parasitology and Tropical Medicine for the kind help in data collection, and the Maternal and Child Care Union for providing assistance in study design and data analysis.