Case-fatality From Orally-transmitted Acute Chagas Disease: A Systematic Review and Meta-analysis

Abstract

Orally-transmitted acute Chagas disease (CD) is emerging as an important public health problem. The prognosis of acute infection following oral transmission is unknown. The aim of this study was to analyze and summarize data on orally-transmitted acute CD. We searched for publications from 1968 to 31 January 2018. We included studies and unpublished data from government sources that reported patients with acute orally-transmitted CD. We identified 41 papers and we added 932 unpublished cases. In all, our study covered 2470 cases and occurrence of 97 deaths. Our meta-analysis estimated that the case-fatality rate was 1.0% (95% CI 0.0–4.0%). Lethality rates have declined over time (P = .02). In conclusion, orally-transmitted acute CD has considerable lethality in the first year after infection. The lethality in symptomatic cases is similar to that from other routes of infection. The lethality rate of orally-acquired disease has declined over the years.

Chagas disease, orally-transmitted Chagas disease, acute Chagas disease, case-fatality on acute Chagas disease

Chagas disease (CD), an infectious disease caused by the parasite Trypanosoma cruzi, affects 6 million people worldwide and costs more than 7 billion dollars per year for health care systems [1]. Although it has long been recognized as an endemic disease restricted to developing countries, CD has spread to other regions in the world in the last 2 decades, affecting many immigrants in Europe and in the US [2]. CD causes about 10 000 deaths per year in Latin America [3], mainly because of chronic heart disease. Because no specific treatment has been proved effective to prevent heart disease in chronic Chagas disease, early detection and treatment of CD in its acute phase is critical to reduce the burden of this disease.

Acute CD may result from the inoculation of feces from an infected triatomine into skin or mucus membrane after its bite or from consuming food or drink contaminated with Trypanosoma cruzi in triatomine feces or infected marsupial secretions. Although vector-borne infection had been the most common source, currently food-borne transmission has been reported as responsible for up to 70% of cases in Brazil [4], changing the epidemiological profile of the disease, and occurring mainly in outbreaks in the Amazon Basin [5]. While vector-borne infection may be associated with inflammation and swelling at the site of inoculation, food-borne infection does not present with localized signs [4], although some patients may develop a transient nonpruritic morbilliform rash called schizotrypanides. When detected and appropriately treated with a trypanocidal drug, acute CD can resolve in 50%–80% of cases [6]. Nevertheless, acute CD is often unrecognized, which results in high lethality rates as reported in some outbreaks in Brazil [5]. Previous studies have described acute CD due to vector-borne infection, reporting an estimated 0.5% mortality rate due to myocarditis and encephalitis [4–9].

However, data on disease progression and lethality after orally-transmitted acute CD rely on a few small studies and often lack information on source of infection and on subgroups that are more susceptible. Therefore, we performed a systematic review and meta-analysis of proportion to summarize all published and unpublished data on reported orally-transmitted acute CD to estimate its case-fatality ratio in 1 year of follow up.

METHODS

Study Design and Eligibility Criteria

We performed a systematic review, studies selection, proportions meta-analysis, and reported our findings in accordance with the PRISMA statement for meta-analysis [10]. We included all observational studies and case reports describing orally-acquired acute CD, published in English, Portuguese, and Spanish languages, from 1968 to 31 January 2018. We also included unpublished governmental data of reported cases of acute CD from Brazil. Studies without data on morbidity or mortality on orally-acquired acute Chagas disease were excluded.

Information Sources and Search

We systematically searched in PUBMED, EMBASE, LILACS, and Google scholar. Our search used the terms “heart,” “cardiovascular,” “mortality,” “morbidity,” “death,” “heart failure,” “Chagas cardiomyopathy,” “Chagas disease,” “Chagas,” “acute” and “acute Chagas disease” as descriptors (Medical Subject Headings—MeSH) and as free text in title and abstract to increase sensitivity. For Google Scholar, we searched for “Chagas disease” and “Chagas cardiomyopathy,” and used the first 50 results from each searched term, considering these as the most important from this source.

Because acute CD is a notifiable disease in Brazil, we also obtained unpublished cases of orally-transmitted CD from the State Departments of Health from the Amazon region and from the Brazilian Information System on Notifiable Diseases (Sistema de Informação de Agravos de Notificação—SINAN) [11]. Data provided by SINAN from 2007 to 2017 were used only in those locations where no cases had been reported in the literature.

For comparison among different routes of infection, we obtained data from SINAN for the vector-borne and other pathways (vertical, transfusion, accidental, ignored, and other). Data extracted from this platform were: number of confirmed cases, and number of deaths from acute Chagas disease between 2007 and 2017. These data represent all notified and confirmed cases of acute CD in Brazil in the period.

Study Selection and Data Extraction

After exclusion of duplicate studies using EndNote software, three authors (E. G. B.; M.M.F-S.; O. M. S.) independently reviewed the selected studies. Discrepancies were resolved by consensus. The following data were extracted from studies: author’s name, year of publication, date of occurrence, source of infection, country and state, duration of follow-up, number of cases, and number of deaths. In the studies with no clear information regarding source of transmission, we contacted the authors to attribute the source of infection for all cases.

We included only studies with patients with confirmed acute CD through either direct parasitological detection (fresh blood, thick drop, or Quantitative Buffy Coat), serological IgM anti-Trypanosoma cruzi tests or xenodiagnosis. We excluded from this meta-analysis studies that did not contain data regarding the morbidity and mortality. All data were entered in a table using Excel software.

Hypothesized Sources of Heterogeneity

We postulated a priori that reporting periods and country of origin may have differed between studies and led to heterogeneous reports. We used I² statistics to identify heterogeneity and applied meta-regression using “period” as potential modifier of lethality.

Quality Assessment

We used the Newcastle–Ottawa Scale for assessing the quality of these observational studies in the meta-analyses [12]. Funnel plots were used to assess for publication bias, as small studies with high lethality might be more likely to be published and could unduly influence lethality rates. In the absence of bias, the plot should form a symmetrical funnel, with less accurate studies composing the broad base of the funnel and more accurate studies forming the narrow top of the funnel. In the presence of bias, the distribution of studies in the graph is asymmetric.

Statistical Analysis

We performed a meta-analysis to estimate the proportion of deaths in the orally-acquired acute Chagas Disease (case fatality rates) during 1 year of follow-up. To assess whether the year of occurrence influenced the death rate, a meta-regression using either year of publication or year of occurrence (for data from government agencies) was performed. We used bubble plots to display the effect of the year in the lethality. In the meta-regression “bubble plot”, each circle represents an individual study and is positioned according to the year it was published (x-axis) and the estimated case-fatality rate (y-axis). The size of the circle varies according to the weight given to estimate, with larger circles having greater precision.

Due to variability in the selected studies, we performed a random-effects meta-analysis with Freeman-Tukey double arc-sine transformation to account for any violation of the assumption of normality in this variable. Heterogeneity and bias were assessed with the I² static and by examination of funnel plots. The meta-analysis was performed using Stata (StataCorp.) version 15.0. Case-fatality rate from vector-borne and other routes of infection (accidental, transfusion, vertical, and others) were calculated from the number of deaths reported among the total number of cases due to each route of transmission.

RESULTS

We screened 4150 studies of Chagas disease and found 41 studies reporting 1538 cases of orally-transmitted acute Chagas disease. All studies were descriptive with a case report or series of cases and lethality rates ranged from 0% to 100%. In addition, another 932 unpublished cases were obtained from government agencies. Overall, our meta-analysis summarized 2470 orally-transmitted acute CD cases that resulted in 97 deaths (Table 1 and Figure 1). The source of contamination was described for 525 (21%) patients.

Table 1.

Studies and Datasets Included in the Meta-Analysis Through Systematic Review

Author, year of publication	Cases	Death	Period of occurence	Food
Brazil
Silva, 1968 [13]	17	6	1965	ignored
Shaw, 1969 [14]	3	1	1968	ignored
Rodrigues, 1988 [15]	8	0	1984	ignored
Shikanai-Yasuda, 1991 [16]	26	1	1986	sugar cane juice
Viana, 1994 [17]	3	3	1993	açaí
Valente, 2009 [18]	17	0	1996	açaí
Pinto, 2001 [19]	13	0	1998	ignored
Pinto, 2004 [20]	3	2	1999–2000	ignored
Pinto, 2007 [21]	20	3	1990–2003	ignored
Medeiros, 2008 [22]	1	0	2004	açaí wine
Pinto, 2008 [23]	183	13	1988–2005	ignored
Borborema, 2005 [24]	9	0	2005	açaí
Steindel, 2008 [25]	24	3	2005	sugar cane juice
Dias, 2008 [26]	7	2	2006	water
de Barros, 2009 [27]	25	0	2007	ignored
Nobrega, 2009 [28]	11	0	2006	açaí
Cavalcanti, 2009 [29]	8	0	2006	soup
Bastos, 2010 [30]	13	2	2006	sugar cane juice
Barbosa-Ferreira, 2010 [31]	4	0	2006–2007	açaí
Borba, 2010 [32]	1	0	2010	açaí
Monteiro, 2010 [33]	25	0	2007	açaí
Pinto, 2011 [34]	4	0	2007	açaí
UNPUBLISHED (SINAN—Maranhão, 2011)	11	1	2011	ignored
UNPUBLISHED (Barbosa, 2011) [35]	12	0	2011	açaí
Souza-Lima, 2013 [36]	17	0	2009–2010	açaí
Souza, 2016 [37]	1	1	2016	açaí
Costa, 2017 [4]	668	20	1997–2012	ignored
UNPUBLISHED (Acre Health Secretary)	43	4	2009–2017	açaí
UNPUBLISHED (Tocantins Health Secretary)	22	1	2008–2014	açai, bacaba or babaçú
UNPUBLISHED (Amazonas Health Secretary)	24	0	2015–2017	açaí
Vargas, 2018 [38]	18	3	2016	sugar cane juice
UNPUBLISHED (Pará Health Secretary)	118	1	2014	ignored
UNPUBLISHED (Pará Health Secretary)	199	2	2015	ignored
UNPUBLISHED (Pará Health Secretary)	273	6	2016	ignored
UNPUBLISHED (Pará Health Secretary)	211	1	2017	ignored
Venezuela
UNPUBLISHED (Noya, 2015) [39]	3	0	2008	ignored
Martín, 2009 [40]	1	0	2007	ignored
Noya, 2009 [41]	88	4	2009	ignored
Noya, 2010 [42]	103	1	2007	guava juice
Añez, 2013 [43]	5	1	2013	ignored
Benítez, 2013 [44]	6	1	2010	ignored
UNPUBLISHED (Noya, 2015) [39]	8	1	2013	mango juice
UNPUBLISHED (Noya, 2015) [39]	3	0	2014	pumarosa juice
UNPUBLISHED (Noya, 2015) [39]	5	0	2014	ignored
Noya, 2016 [45]	89	5	2009	guava juice
Noya, 2017 [46]	3	0	2015	ignored
Bolivia
Vargas, 2011 [47]	14	0	2010	fruta majo
Colombia
Hernandez, 2008 [48]	10	2	2008	orange juice
Zambrano, 2010 [49]	20	3	2008–2009	ignored
Rios, 2011 [50]	11	1	2010	ignored
Dueñas, 2014 [51]	40	2	2014	ignored
Soto, 2014 [52]	11	0	2010	ignored
French Guiana
Blanchet, 2014 [53]	8	0	2005	palm juice
OVERALL	2470	97

Figure 1.

Flow chart of studies selection for the meta-analysis.

All reported cases of orally-transmitted acute CD occurred in Latin America, specifically in Venezuela (n = 314), Colombia (n = 92), Bolivia (n = 14), French Guiana (n = 8), and Brazil (n = 2042) (Figure 2). The annual number of cases remained stable until mid-2000, followed by 2 significant incidence peaks, approximately in 2008 and 2015 (Figure 3). Although the food source was unreported in many studies, açaí was the most common, followed by sugar cane juice, palm fruit, majo fruit, mango juice, orange juice, and others (Table 1). Even though we did not include other sources of infection in the meta-analysis, we found from government data a total of 731 cases that were notified as vector-borne transmission (n = 207) and other pathways (n = 524) between 2007 and 2017 in Brazil.

Figure 2.

Geographical regions with cases of orally-transmitted acute Chagas disease.

Figure 3.

Cases of Chagas disease and deaths since 1968 until January 2018 by year of publication. The dashed line represents cases of orally-acquired acute Chagas disease. Continuous line represents the number of deaths.

In the meta-analysis, the pooled estimated case-fatality rate from acute CD due to oral transmission was 1.0% at one-year follow-up [0.0–4.0% CI 95%] (Figure 4). Lethality due to vector borne transmission was 2.4% and to other routes of infection 1.4%. For all routes of transmission, the case-fatality rate of acute CD was estimated at 1.0% [0.1–3.0% CI 95%], but there was a moderate heterogeneity among the individual studies. When we split the cases in 2 eras, “until 2008” and “after 2008” (comparing studies until 2008 vs studies between 2009 and 2018), we found a case-fatality rate of 3% (0% – 8%) for studies until 2008 and 0% (0 %– 1%) after 2008 (Supplemental Figure). Through a meta-regression (Figure 5), we found that the lethality from orally acquired CD changed over time and displayed a statistically significant reduction over the years (P = .02). To assess publication bias we built a funnel plot (Figure 6). The effect measure logarithm was plotted on the x-axis against the standard error logarithm for each effect size.

Figure 4.

Forest plot for case-fatality rate on orally-transmitted acute Chagas disease. Horizontal lines represent the 95% confidence intervals around the point estimates for each study. The square areas are proportional to the weight given to each study and the diamonds represent the estimated lethality grouped by country and for all studies. Abbreviation: CI, confidence interval.

Figure 5.

Bubble plot for meta-regression of the case-fatality rate of acute Chagas disease orally-transmitted according to the year of publication. The x-axis represents the year of publication of the studies. The y-axis corresponds to the log of the proportion of deaths. The continuous line shows the decrease in lethality over the years. Each of the circles represents a study and its diameter is proportional to the precision of the study.

Figure 6.

Funnel plot for studies on acute Chagas disease orally-transmitted. Each point in the graph area represents a study. Less accurate studies are in the broad base of the funnel. The most accurate studies are plotted at the top. In the absence of bias, the graph shows symmetry between the distribution of studies in its quadrants.

DISCUSSION

In this systematic review and meta-analysis, we summarized, for the first time to our knowledge, the geographic distribution, source of infection, and lethality from orally-transmitted acute CD. We found that the case-fatality rate is roughly 1% during 1-year follow-up, which is similar to the disease acquired by other routes of transmission. Although the number of deaths due to orally-transmitted CD increased in the last 2 decades because of recent outbreaks, lethality appears to have decreased over time, mainly due to government policies implemented to increase awareness and faster recognition of cases with early therapeutic institution. All reported cases occurred in Latin America and were most commonly caused by the ingestion of açaí, a typical tropical fruit widely consumed in Brazil (Figure 2). Our findings provide insights on public policies that could help reduce the burden of CD. Preventing the main source of infection is a key for reducing the morbidity and mortality of chronic CD over the long term.

Our meta-analysis included 2470 cases of orally-transmitted acute CD. Studies compounding the current systematic review were mainly published as case reports or case series generally reported from outbreaks in Latin America. In the last 2 decades, cases of acute CD related to contaminated food occurred most frequently in the Amazon Basin in areas with a high production and consumption of açaí. Almost half of cases were diagnosed in the State of Pará, in the Amazon Basin, which is the main producer of açaí in the World [54]. Around 10% of the production is exported, corresponding to 6000 tons of açaí pulp [54]. Approximately 90% of this amount is destined to the US and Japan [54]. To our knowledge, no reports of orally-acquired CD have been reported outside Latin America to date. Another population under risk of contamination are travelers visiting Latin America. As we may see in Figure 2, several areas in red are tourist destinations where fresh juice are consumed with potential of contamination with Trypanosoma cruzi. It is important that health consultants on travel advise visitors to avoid foods that have a potential contaminant or make sure of their origin and safety [55]. Reported cases of orally acquired CD from Venezuela, Colombia, Bolivia, and French Guiana were not related to açaí consumption but from popular products such as majo, guava, and mango [41, 48].

Chagas disease is considered a neglected disease, and strategies to mitigate its spreading have focused on vector control [3]. Governmental efforts in this direction were taken in Brazil and culminated with the PAHO granting a certificate of interruption of vector-borne transmission by the T. infestans in the country in 2006 [56]. However, vector-borne transmission continues to occur in the country [56], either by persistence of residual foci of T. infestans or by other competent vectors as Triatoma brasiliensis [4]. Vector-borne transmission does not seem to be restricted to rural areas, as suggested by the presence of infected triatomines in urban areas in the state of Acre [57].

The main causes of death among those with acute CD are myocarditis and encephalitis. In the analysis that included only orally-acquired acute CD, 97 patients died in 1 year-follow-up. Notably, only 1 patient died of meningoencephalitis. Most studies included in this analysis did not show the cause of death. In chronic CD, 20%–30% of infected person develop cardiac complications, sudden death accounts for 55%–60% of deaths, heart failure for 25%–30%, and thromboembolic phenomena for 10%–15% [6].

Some conditions could explain the higher lethality presented in some studies. Among them, the T. cruzi strain [9], the patient’s age, the parasitic load contained in the contaminated food, and the time elapsed between infection and the therapeutic institution. Despite the apparent difference in lethality between countries in our study, it was not statistically significant (P = .46). Acute vector-borne CD is typically more severe in children, the elderly, and immunocompromised [3], populations with greatest rates of diagnosed acute disease the highest mortality.

The decrease in mortality over time may reflect increased awareness, more rapid diagnosis, and prompt institution of appropriate therapy. Orally-transmitted CD can be prevented by better control of food processing. Generally, açaí juice is prepared manually during the night, when triatomines are attracted by light and contaminate the juice. Contamination can occur through the maceration of triatomines in fruit processing or by the deposition of their feces; contamination of water used in pulp processing by secretions of infected marsupials has been proposed as an alternative source [16]. The Brazilian government has issued a set of regulations to prevent contamination of food with T. cruzi. For example, “branqueamento” consists of submitting açaí and bacaba fruits to a temperature of 80 °C for 10 s and then cooling them to room temperature [57, 58]. Another is pasteurization of the processed beverage and subsequent cooling [57]. Although these measures are mandatory in some states, such as Amapá and Pará, no surveillance program by governmental agencies has been implemented to control açaí production [59].

We found a substantial heterogeneity across studies, which remained significant after splitting between different time periods. Reasons for this heterogeneity are likely multifactorial and may result from differences in baseline patient characteristics, such as age and immunosuppression, or treatment. Although clinical-demographic characteristics were missing, the decline in case-fatality rates were observed in tandem with government policies implemented to increase awareness. For instance, health care workers were trained to identify the Trypanosoma during the blood smears analysis for malaria (thick blood film), which increased the likelihood of early diagnosis and specific treatment, potentially improving the case-fatality rates.

This study has several limitations. First, around 20% of patients with acute CD have missing data on the form of transmission in Brazil [60], and the number of cases of orally-transmitted acute CD may have been underestimated. Even when reported, the source of infection may be challenging to establish. The diagnosis of orally transmitted CD relies upon the exclusion of vector-borne, transfusional, vertical, and transplant-associated transmission. Although the oral acquisition cannot be determined with complete certainty, it is the most likely form of transmission in these reports originated in outbreaks from regions without vector colonization and no registered domiciliary vectorial transmission [61]. Despite this limitation, our data represent the best available information, including unpublished data from governmental agencies in Brazil, where acute CD is a disease of compulsory notification. In Brazil, an improvement of data quality in the mandatory disease notification systems could enable a better understanding of the disease’s epidemiology, avoiding under-reporting and facilitating the decision-making for coping with and preventing the disease. Second, the form of infection (vectorial versus oral) may be subject to misclassification bias. Because the incubation period of T. cruzi is 7 to 14 days, collecting food to search for parasite is often unfeasible, and most studies rely on retrospective data regarding food intake using standardized questionnaires. Third, our data represent mostly symptomatic cases of orally-transmitted acute CD. It is conceivable that symptoms occur only when the oral infection is more severe, overestimating the lethality rates from this route of infection. Nevertheless, screening asymptomatic individuals for orally-transmitted acute CD would be difficult, and our data indicate that, among symptomatic cases, lethality is likely similar between different infection pathways. Additionally, most studies did not report the time since infection to death and the period of follow-up varied up to 1 year. Because of that, we performed a sensitivity analysis removing studies with more than 60 days of follow up and the results were roughly the same (data not shown).

In conclusion, the findings of our systematic review and meta-analysis suggest that acute orally acquired Chagas disease has a considerable lethality in the first year after infection, which is similar to symptomatic vector-borne or other infection pathways. Improving the reporting tools and acute CD awareness in the Latin America would enable an early detection of the disease improving prognosis.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Financial support. This paper received no financial support.

Potential conflicts of interest. The authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.