Conceived and designed the experiments: VEMdA MC. Performed the experiments: VEMdA MHFM. Analyzed the data: VEMdA IAR MC. Contributed reagents/materials/analysis tools: MC. Wrote the paper: VEMdA AR MC.
The authors have declared that no competing interests exist.
In Brazil, lethality from visceral leishmaniasis (VL) is high and few studies have addressed prognostic factors. This historical cohort study was designed to investigate the prognostic factors for death from VL in Belo Horizonte (Brazil).
The analysis was based on data of the Reportable Disease Information System-SINAN (Brazilian Ministry of Health) relating to the clinical manifestations of the disease. During the study period (2002–2009), the SINAN changed platform from a Windows to a Net-version that differed with respect to some of the parameters collected. Multivariate logistic regression models were performed to identify variables associated with death from VL, and these were included in prognostic score.
Model 1 (period 2002–2009; 111 deaths from VL and 777 cured patients) included the variables present in both SINAN versions, whereas Model 2 (period 2007–2009; 49 deaths from VL and 327 cured patients) included variables common to both SINAN versions plus the additional variables included in the Net version. In Model 1, the variables significantly associated with a greater risk of death from VL were weakness (OR 2.9; 95%CI 1.3–6.4),
Knowledge regarding the factors associated with death from VL may improve clinical management of patients and contribute to lower mortality.
The visceral leishmaniasis (VL) is a disease potentially fatal if not diagnosed and treated opportunely. This article presents the results of the study on the manifestations identified at the time of the clinical suspicion of the VL cases. This study was conducted in Belo Horizonte, the capital of the State of Minas Gerais, located in southeastern Brazil. This study is both timely and substantive because the Belo Horizonte is an area of transmission of VL, with one of the highest VL-death proportions of Brazil. The patients with higher risk of death had at least one of the following characteristics: ≥60 years, weakness, HIV co-infection, bleeding, jaundice and other associated infections. During the period 2002–2009, 8% to 22% of the patients with VL progressed to death in Belo Horizonte, whilst the proportion in the country was much lower and varied between 5% and 9%. This study has identified vulnerable patients who are at higher risk of death from VL and who would benefit from early predictive evaluation of the prognostic. Hence, the knowledge regarding the factors associated with death may contribute for clinical management and for reduction of deaths from VL.
The number of new cases of visceral leishmaniasis (VL) is estimated to be around 500,000 per year worldwide with over 50,000 deaths
In Brazil since the 1980's the geographical distribution of VL has expanded, partly due to increased urbanization
In fact, between 1990 and 2009, a total of 57,973 clinical cases of VL had been reported
According to the Visceral Leishmaniais Control and Surveillance Program (VLCP) of Brazil
The control of VL in urban areas of Brazil represents, however, a difficult and continuous challenge despite the measures adopted by the Brazilian Ministry of Health, and implemented through of the VLCP
Since reduction of case fatality rate is one of the goals of VLCP
The study was approved by the Ethical Review Board of the Universidade Federal de Minas Gerais (No.211/09) and of the Municipality Health Service of Belo Horizonte (No.075.2008). Data were analyzed anonymously.
The study was carried out in Belo Horizonte, the capital of the State of Minas Gerais, located in southeastern Brazil, an area comprising 2,375,444 inhabitants
The epidemiological surveillance system of the Brazil involved registration of the suspected VL case at SINAN using a form comprising the following entries: date of notification, health unit responsible for notification, address, age, sex, level of schooling, occupation of patient, date of the start of symptoms and clinical manifestations (signs and symptoms). Subsequently, further information was added to the records including the results of specific laboratory examinations, date of beginning of treatment, initial drug used for treatment, drug used following failure of the initial therapy, and evolution of the case.
The SINAN database changed platform during the study period from a Windows-based version (2002–2006) to a Net version (2007–2009). As shown in
Field | Version Windows | Version Net |
Fever, weakness, emaciation, cough and/or diarrhea, splenomegaly, hepatomegaly | Added variables: swelling, pallidness, other infections, bleeding, jaundice and others manifestations | |
HIV, tuberculosis, other types of infection | HIV co-infection only | |
|
Inexistent field | Existing field |
Inexistent field | Existing field | |
|
Cure, death, unknown | Cure, abandoned treatment, VL-death, death by other causes, transfer (treatment interrupted in the original municipality, but continued at some other location) |
Statistical analyses of the data were performed using STATA version 11.0 software (Stata Corp., College Station, TX, USA) considering 111 deaths from VL and 777 cured patients. Univariate logistic regression analysis was used to evaluate the demographic and clinical variables according to the occurrence of death from VL. Variables associated with death from VL at a significance level of
Two multivariate logistic regression models were subsequently analyzed. Model 1 covered the period 2002 to 2009 and included the variables present in both versions of SINAN (111 deaths and 777 cured patients), whereas Model 2 covered the period 2007 to 2009 and included the variables common to both versions of SINAN together with the new variables included in the Net version (49 deaths and 327 cured patients). With the aim of avoiding selection bias, and to allow better adjustment of the model to VL data, the category “unknown” was maintained for variables for which information was missing. Variables presenting collinearity were evaluated and those that better explained the occurrence of death from VL were retained in the model.
A step-by-step backward selection procedure was used to select the variables and to produce the final multivariate logistic regression models. Only adjusted variables showing a significant association (
The predictive factors relating to death from VL that were identified by Model 2 (period 2007–2009) were used to create a prognosis score. According to the methodology described by Barquet et al.
The incidence of VL in Belo Horizonte during the period 2002–2009 varied from 3.4 to 6.6/100,000 inhabitants and the case fatality rate was 13.1% (
Year | VL Cases |
Incidence |
Deaths | Case fatality rate |
2002 | 77 | 3.4 | 9 | 11.7 |
2003 | 103 | 4.5 | 10 | 9.7 |
2004 | 134 | 5.8 | 25 | 18.7 |
2005 | 110 | 4.6 | 10 | 9.1 |
2006 | 128 | 5.3 | 12 | 9.4 |
2007 | 110 | 4.5 | 9 | 8.2 |
2008 | 161 | 6.6 | 20 | 12.4 |
2009 | 141 | 5.8 | 31 | 22.0 |
Total | 964 | .. |
126 | 13.1 |
Cases obtained from the Municipality Health Service of Belo Horizonte/SINAN.
Population data from Brazilian Institute of Geography and Statistics/IBGE.
Numerical information not applicable.
Age range(years) | Sex | Total | ||||||||||
Female | Male | |||||||||||
Cases | Deaths | Case fatality rate (%) | Cases | Deaths | Case fatality rate (%) | Cases | Deaths | Case fatality rate (%) | ||||
n | % | n | % | n | % | |||||||
0–4 | 112 | 31.9 | 9 | 8.0 | 120 | 22.4 | 11 | 9.2 | 232 | 26.1 | 20 | 8.6 |
5–9 | 56 | 16.0 | 3 | 5.4 | 54 | 10.1 | 7 | 13.0 | 110 | 12.4 | 10 | 9.1 |
10–19 | 38 | 10.8 | 1 | 2.6 | 51 | 9.5 | 2 | 3.9 | 89 | 10.0 | 3 | 3.4 |
20–29 | 40 | 11.4 | 3 | 7.5 | 70 | 13.0 | 13 | 18.6 | 110 | 12.4 | 16 | 14.6 |
30–39 | 23 | 6.6 | 4 | 17.4 | 71 | 13.2 | 15 | 21.1 | 94 | 10.6 | 19 | 20.2 |
40–49 | 36 | 10.3 | 6 | 16.7 | 69 | 12.9 | 8 | 11.6 | 105 | 11.8 | 14 | 13.3 |
50–59 | 21 | 6.0 | 3 | 14.3 | 41 | 7.6 | 5 | 12.2 | 62 | 7.0 | 8 | 12.9 |
≥60 | 25 | 7.1 | 5 | 20.0 | 61 | 11.4 | 16 | 26.2 | 86 | 9.7 | 21 | 24.4 |
Total | 351 | 100.0 | 34 | 9.7 | 537 | 100.0 | 77 | 14.3 | 888 | 100.0 | 111 | 12.5 |
The results of univariate logistic regression analysis of the demographic and clinical variables with respect of deaths from VL are shown in
Variables (n) |
Deaths from VL | Odds Ratio | 95% IC | p values | |||
No | Yes | ||||||
n | % | n | % | ||||
Female | 317 | 40.8 | 34 | 30.6 | 1.0 | .. |
.. |
Male | 460 | 59.2 | 77 | 69.4 | 1.6 | 1.0–2.4 | 0.04 |
0–4 | 212 | 27.3 | 20 | 18.0 | 1.0 | .. | .. |
5–9 | 100 | 12.9 | 10 | 9.0 | 1.1 | 0.5–2.4 | 0.89 |
10–19 | 86 | 11.1 | 3 | 2.7 | 0.4 | 0.1–1.3 | 0.12 |
20–29 | 94 | 12.1 | 16 | 14.4 | 1.8 | 0.9–3.6 | 0.10 |
30–39 | 75 | 9.7 | 19 | 17.1 | 2.7 | 1.4–5.3 | 0.00 |
40–49 | 91 | 11.7 | 14 | 12.6 | 1.6 | 0.8–3.4 | 0.19 |
50–59 | 54 | 6.9 | 8 | 7.2 | 1.6 | 0.7–3.8 | 0.31 |
≥60 | 65 | 8.4 | 21 | 18.9 | 3.4 | 1.8–6.7 | 0.00 |
|
|||||||
Fever ( |
723 | 96.1 | 102 | 94.4 | 0.7 | 0.3–1.7 | 0.41 |
Splenomegaly ( |
642 | 89.0 | 92 | 88.5 | 0.9 | 0.5–1.8 | 0.86 |
Hepatomegaly (n = 827) | 607 | 84.0 | 88 | 84.6 | 1.1 | 0.6–1.9 | 0.86 |
Weakness ( |
568 | 81.3 | 95 | 93.1 | 3.1 | 1.4–6.9 | 0.01 |
Emaciation ( |
516 | 75.0 | 78 | 82.1 | 1.5 | 0.9–2.7 | 0.13 |
Cough ( |
319 | 48.9 | 50 | 55.6 | 1.3 | 0.8–2.0 | 0.24 |
Edema |
24 | 11.6 | 13 | 39.4 | 5.0 | 2.2–11.2 | 0.00 |
Pallidness |
149 | 66.2 | 28 | 82.4 | 2.4 | 0.9–6.0 | 0.07 |
Other infections(3) ( |
44 | 21.8 | 19 | 55.9 | 4.6 | 2.1–9.7 | 0.00 |
Bleeding |
16 | 7.8 | 11 | 34.4 | 6.2 | 2.6–15.2 | 0.00 |
Jaundice |
13 | 6.3 | 14 | 43.8 | 11.6 | 4.7–28.3 | 0.00 |
Other manifestations |
55 | 31.1 | 7 | 24.1 | 0.7 | 0.3–1.8 | 0.45 |
|
|||||||
HIV ( |
39 | 8.0 | 12 | 17.7 | 2.5 | 1.2–5.0 | 0.01 |
Tuberculosis |
9 | 2.9 | 4 | 10.5 | 3.9 | 1.1–13.3 | 0.03 |
The total number of individuals listed in SINAN and included in the study (Belo Horizonte, Brazil) were 888, of which 512 were registered in the Windows version and 376 in the Net version of the database.
Numerical information not applicable.
Variable registered for VL cases occurring in the period 2007–2009 (Net version of SINAN).
Variable registered for VL cases occurring in the period 2002–2006 (Windows version of SINAN).
Deaths from VL | Odds Ratio | 95%IC | p values | ||||
Variables (n) |
No | Yes | |||||
n | % | n | % | ||||
up 4 | 441 | 56.8 | 54 | 48.7 | 1.0 | .. |
.. |
5–8 | 144 | 18.5 | 21 | 18.9 | 1.2 | 0.7–2.0 | 0.52 |
9–12 | 84 | 10.8 | 16 | 14.4 | 1.6 | 0.9–2.9 | 0.15 |
>12 | 108 | 13.9 | 20 | 18.0 | 1.5 | 0.9–2.6 | 0.14 |
|
|||||||
Pentavalent antimonial ( |
570 | 84.8 | 42 | 51.9 | 1.0 | .. | .. |
Amphotericin ( |
96 | 14.3 | 39 | 48.1 | 5.5 | 3.4–9.0 | 0.00 |
Pentamidine ( |
6 | 0.9 | - |
- | - | - | - |
<20 (n = 40) | 32 | 9.8 | 8 | 30.8 | 1.0 | .. | .. |
20 (n = 55) | 48 | 14.6 | 7 | 26.9 | 0.6 | 0.2–1.8 | 0.34 |
21–40 ( |
240 | 73.2 | 11 | 42.3 | 0.2 | 0.1–0.5 | 0.00 |
>40 ( |
8 | 2.4 | - | - | - | - | - |
Pentavalent antimonial (n = 11) | 10 | 21.7 | 1 | 14.3 | 1.0 | .. | .. |
Amphotericin (n = 41) | 35 | 76.1 | 6 | 85.7 | 1.7 | 0.2–16.0 | 0.64 |
Pentamidine (n = 1) | 1 | 2.2 | - | - | - | - | - |
The total number of individuals listed in SINAN and included in the study were 888, of which 512 were registered in the Windows version and 376 in the Net version of the database.
Numerical information not applicable.
symbol equal to zero.
The results of the multivariate logistic regression analysis of those variables that were associated (
Variables | Odds Ratio (95% CI) | Adjusted Odds Ratio (95% CI) | p values |
|
|||
Weakness | 3.1 (1.4–6.9) | 2.9 (1.3–6.4) | 0.01 |
2.5 (1.2–5.0) | 2.4 (1.2–4.8) | 0.02 | |
Age (≥60 |
2.6 (1.5–4.4) | 2.5 (1.5–4.3) | 0.00 |
|
|||
Other infections | 4.6 (2.1–9.7) | 3.2 (1.3–7.8) | 0,01 |
Bleeding | 6.2 (2.6–15.2) | 3.5 (1.2–10.3) | 0.02 |
Jaudice | 11.6 (4.7–28.3) | 10.1 (3.7–27.2) | 0.00 |
Age (≥60 |
2.6 (1.5–4.4) | 3.1 (1.4–7.1) | 0.02 |
Covering the period 2002–2009 and including the variables present in both versions of SINAN (111 death from VL and 777 cures).
Covering the period 2007–2009 and including the variables common to both versions of SINAN together with the new variables included in the Net version (49 death from VL and 327 cures).
Prognosis factor | Regression Coeficient | Score |
Other infections | 1.17 | 1 |
Bleeding | 1.25 | 1 |
Jaundice | 2.31 | 2 |
Age (≥60 versus<60 years) | 1.13 | 1 |
Scoring system based on the variables listed in the SINAN Net version.
In the present study, the factors associated with death from VL were weakness,
The records analyzed in the present study were obtained from a database containing details of registered VL cases, each of which would normally have been notified on clinical suspicion of the disease. Our results reveal that it is possible to detect the factors associated with death from VL at first clinical suspicion of the disease and, hence, to identify the most vulnerable patients. Timely application of specific and effective measures to the patients would contribute greatly to a reduction in the lethality of the disease. Several authors have suggested that knowledge regarding the laboratory and clinical profiles of patients and their association with death from VL could assist in the clinical management and reduce lethality
The case fatality rate registered in Belo Horizonte is one of the highest in Brazil. Some hypotheses could explain the variation in the rates as accessibility to the health service, suspicion of VL and delayed diagnosis, treatment opportunity, clinical management of the patient, toxicity of drugs, comorbities and
Factors associated with lethality from VL have been reported in the literature
The time interval between the onset of symptoms and time of diagnosis was estimated and this interval was not significantly associated with death from VL (
The weakness, which is one of the earliest clinical symptoms of VL, was found to be significantly associated with death. The most likely explanation for this finding is that the study was based on clinical manifestations presented by VL-suspect patients at their first medical appointment. However, the early detection of weakness may help to identify those patients presenting a higher likelihood of an unsatisfactory evolution of their disease. A study conducted in central west Brazil involving 55 individuals that had died from VL described the occurrence of hyporexia (65.5%), asthenia (58.1%) and adynamia (29.0%)
A VL-suspected case is defined as a patient presenting fever and splenomegaly who originates from an
The occurrence of jaundice was registered for 27 patients, 14 (51.9%) of whom died from VL. The presence of jaundice increased the chances of death from VL by a factor of 10 (ORadjusted 10.1; 95%CI 3.7–27.2). The large amplitude of the confidence interval can be explained by the small number of patients presenting jaundice. Following a study carried out in the Brazilian State of Piauí involving 12 deaths from VL and 78 cured individuals, Werneck et al.
The occurrence of jaundice may indicate liver damage
In the present study, the ages of cured patients varied from 3 to 93 years old whereas the ages of those who had died from VL were between 5 months and 86 years. The lethality in older individuals (≥60 years) was the highest (24.4%), a value similar to that reported by Oliveira et al.
Multivariate regression analysis models 1 and 2 identified age of ≥60 years as a factor associated with death from VL (OR 2.5 and 3.1, respectively). It is worth noting that the present study involved 86 patients aged ≥60 years and of these 21 died from VL accounting for 19.0% of the total deaths analyzed. However, in 2009, the Brazilian Ministry of Health mentioned that VL patients aged <1 and >40 years were at greater risk
In this study, 63 subjects presented other associated infections with significantly difference between the cured patients (n = 44; 5.7%) and those that died from VL (n = 19; 17.1%) Multivariate logistic regression analysis showed that the occurrence of other infections increased the chances of death by 3-fold. A recent study involving 55 hospitalized VL patients who progressed to death found that 65.5% had been diagnosed with other infections, most commonly sepsis (66.7%) and pneumonia (63.9%), at the time of admission and during hospitalization
Twenty-seven individuals presented history of bleeding at the first clinical examination with frequencies that were significantly different between the 16 cured patients (2.1%) and the 11 deaths from VL (9.9%). The presence of bleeding increased the chances of death by 3.5-fold. Oliveira et al.
Following a study of Tunisian children affected from VL, it was reported that bleeding and a period of more than 56 days between the onset symptoms and the first clinical examination were among the seven most important factors associated with a negative prognosis
According to univariate analysis,
In a study conducted in north-eastern Brazil involving a chronological serie of hospital records during a period of 10 years (1996–2005) and relating to 396 VL-patients (76 deaths and 320 cured patients), nine
Regarding the initial drug used in treatment (
Measures that have been applied to control the canine reservoir and the insect vector have not been successful in preventing the spread of VL in Brazil, and early diagnose and treatment of human cases remain the main approaches for reducing lethality. Guidelines published by the Brazilian Ministry of Health stress the key factors associated with death from VL with the aim of ensuring that patients who require special care can be identified and classified according to severity of risk
The prognostic score generated in the present study is based on four clinical variables (age≥60 years, bleeding, other associated infections and jaundice). The predictive performance of this score was: sensitivity 71.4%, specificity 73.7%, positive and negative predictive value (28.9% and 94.5%) and area under the ROC curve (75.6%). The predictive performance could be improved with addition of laboratory variables according to other studies carried out in Brazil
The key issue, however, is to define a simple prognosis score that could be applied in basic health units and would allow the early detection of VL cases for redirection to specialized health service. For the purposes of comparison, patients attaining a predictive score in the range from 1 to 5 received an allocated score of 1, while those presenting none of the death prognosis factors received an allocated score of 0. The scoring system presented here can be used to identify patients running a higher risk of death from VL at the time of clinical suspicion. Those patients with a score between 1 and 5 should receive specialized clinical management during treatment.
Our study had limitations that deserve to be discussed. The multivariate analyses did not take into account all possible factors that could contribute to unfavorable evolution of the VL case,
Another limitation of this study is related to the validation of the prognostic scoring system which was proposed using the same patients whose data was used in its preparation (patients in the period 2007–2009). In fact, as this score included the variables of the newer version of SINAN, there is until now no distinct set of patients for validation. In the future, we intend to validate this prognostic scoring system using several random samples from the SINAN database.
In order to improve the quality of assistance, the Brazilian Ministry of Health
In Brazil, suspected VL cases must be notified to the health authorities using the appropriate SINAN epidemiological form. All fields in the form must be completed even in the absence of information (missing code). However, some of the variables (schooling, occupation, ethnicity, weight, date of the start of treatment and relapse) could not be evaluated in the present study because the corresponding fields had high proportion of missing data. It would have been helpful to know the date of the start of treatment, since early diagnosis and treatment of VL are important in reducing the lethality of the disease. Indeed, the absence of vital information in partially completed medical records is a chronic problem in Brazil
SINAN has a specific module for VL which includes the registration of the variables described in the literature associated with the disease. In Brazil, the Epidemiological Surveillance Service works in conjunction with other information systems such as Mortality Information System (SIM). So it is routine for this service to verify if deaths from reportable diseases such as VL are listed in SINAN system. Of course, there may be some underreporting due to the difficulty in defining the VL clinical case or identifying the VL as cause of death. Therefore, underreporting is minimized by comparing the two systems mentioned.
This study may provide ammunition to the debate on the usefulness and limitations of the Reportable Disease Information System (SINAN) database. According to the World Health Organization
This study has identified vulnerable patients who are at higher risk of death from VL and who would benefit from early predictive evaluation of the prognostic. In conclusion the knowledge regarding the factors associated with death may contribute for clinical management and reduction of lethality from VL.
(DOC)
The authors wish to thank the Municipality Health Service of Belo Horizonte for facilitating access to data and for valuable assistance in verifying the consistency of the SINAN databases.