A simple method for rapid community assessment of tungiasis

Authors


Corresponding AuthorJörg Heukelbach, Departamento de Saúde Comunitária, Faculdade de Medicina, Universidade Federal do Ceará, Rua Prof. Costa Mendes 1608, 5. andar, Fortaleza CE 60430-140, Brazil. Tel.: +55 85 33668045; Fax: +55 85 33668050; E-mail: heukelbach@web.de

Summary

Objective  To evaluate a rapid assessment method to estimate the overall prevalence of tungiasis and severity of disease in endemic communities.

Methods  We analysed data from 10 population-based surveys on tungiasis, performed in five endemic communities in Brazil and Nigeria between 2001 and 2008. To assess the association between occurrence of tungiasis on six defined topographic areas of the feet and the true prevalence/prevalence of severe disease, linear regression analyses were performed. Estimated prevalences were calculated for each of the 10 surveys and compared to true prevalences. We then selected the most useful topographic localization to define a rapid assessment method, based on the strength of association and operational aspects.

Results  In total, 7121 individuals of the five communities were examined. Prevalence of tungiasis varied between 21.1% and 54.4%. The presence of periungual lesions on the toes was identified as the most useful rapid assessment to estimate the prevalence of tungiasis (absolute errors: −4% to +3.6%; R= 96%; P < 0.0001). Prevalence of severe tungiasis (>20 lesions) was also estimated by the method (absolute errors: −3.1% to +2.5%; R= 76%; P = 0.001).

Conclusion  Prevalence of tungiasis and prevalence of severe disease can be reliably estimated in communities with distinct cultural and geographical characteristics, by applying a simple and rapid epidemiological method. This approach will help to detect high-risk communities and to monitor control measures aimed at the reduction of tungiasis.

Abstract

Une méthode simple pour la rapide évaluation communautaire de la tungose

Objectif:  Evaluer une méthode rapide d’évaluation pour estimer la prévalence globale de la tungose et la sévérité de la maladie dans les communautés endémiques.

Méthodes:  Nous avons analysé les données de 10 enquêtes de population sur la tungose, effectuées dans 5 communautés endémiques au Brésil et au Nigéria entre 2001 et 2008. Afin d’évaluer l’association entre la survenue de la tungose dans 6 zones topographiques définies des pieds et la prévalence réelle/prévalence de maladie sévère, des analyses de régression linéaire ont été effectuées. Les prévalences estimées ont été calculées pour chacune des 10 enquêtes et par rapport à la prévalence réelle. Nous avons ensuite sélectionné la zone topographique la plus utile pour définir une méthode d’évaluation rapide, basée sur la force de l’association et les aspects opérationnels.

Résultats:  7121 individus dans les 5 communautés ont été examinés. La prévalence de la tungose variait de 21,1%à 54,4%. La présence de lésions péri-unguéales sur les orteils a été identifiée comme l’évaluation rapide la plus utile pour estimer la prévalence de la tungose (erreurs absolues: -4%à +3,6%; R2 = 96%; p <0,0001). La prévalence de la tungose sévère (> 20 lésions) a également été estimée de manière fiable par la méthode (erreurs absolues: -3,1%à +2,5%; R2 = 76%; p = 0,001).

Conclusion:  La prévalence de la tungose et la prévalence de la maladie sévère peuvent être estimées de manière fiable dans des collectivités avec des caractéristiques géographiques et culturelles distinctes, en appliquant une méthode épidémiologique rapide et simple. Cette approche permettra de détecter les communautés à haut risque et de surveiller les mesures de lutte visant à la réduction de la tungose.

Abstract

Un método simple para realizar una estimación rápida y comunitaria de la Tungiasis

Objetivo:  Evaluar un método rápido para calcular la prevalencia total de tungiasis y la severidad de la enfermedad en comunidades endémicas.

Métodos:  Hemos analizado datos de 10 estudios de tungiasis basados en la población, realizados entre el 2001 y el 2008 en 5 comunidades endémicas del Brasil y Nigeria. Mediante análisis de regresión linear se evaluó la asociación entre la incidencia de tungiasis en 6 áreas topográficas definidas de los pies y la prevalencia real / prevalencia de la enfermedad severa. Las prevalencias estimadas fueron calculadas para cada uno de los 10 estudios y comparadas con las prevalencias reales. Después seleccionamos la localización topográfica más útil para definir un método de estimación rápido, basado en la fortaleza de la asociación y los aspectos operativos.

Resultados:  Se examinaron 7121 individuos de las 5 comunidades. La prevalencia de tungiasis varió entre 21.1% y 54.4%. La presencia de lesiones periungueales en los dedos del pie se identificó como el método más rápido y útil para estimar la prevalencia de tungiasis (errores absolutos: -4% a +3.6%; R2=96%; P<0.0001). La prevalencia de tungiasis severa (>20 lesiones) también se estimó de manera fiable con este método (errores absolutos: -3.1% a +2.5%; R2 = 76%; P=0.001).

Conclusión:  La prevalencia de tungiasis y la prevalencia de la enfermedad severa pueden estimarse de forma fiable en comunidades con características culturales y geográficas distintivas, aplicando un método epidemiológico rápido y simple. Este método podría contribuír a detectar comunidades de algo riesgo y monitorizar medidas de control dirigidas a reducir la tungiasis.

Introduction

Tungiasis is a tropical parasitic skin disease caused by penetration of the jigger flea Tunga penetrans (Linnaeus 1758) into the skin of human or animal hosts (Heukelbach 2005). Hundreds of parasites may accumulate in heavily infested individuals (Feldmeier et al. 2003; Joseph et al. 2006; Ugbomoiko et al. 2007). The disease is a self-limited infestation (Eisele et al. 2003; Feldmeier & Heukelbach 2009), but complications such as bacterial super-infection and debilitating sequels are often seen in endemic areas (Bezerra 1994; Heukelbach et al. 2001; Feldmeier et al. 2002, 2003; Joseph et al. 2006; Ariza et al. 2007; Ugbomoiko et al. 2008). Septicaemia and tetanus are life-threatening complications of tungiasis (Tonge 1989; Litvoc et al. 1991; Greco et al. 2001; Feldmeier et al. 2002; Joseph et al. 2006).

Typically, the disease occurs in poor communities in Latin America, the Caribbean and sub–Saharan Africa (Heukelbach et al. 2001; Heukelbach 2005). In recent cross-sectional studies from endemic areas in Brazil, Cameroon, Madagascar, Nigeria and Trinidad & Tobago, point prevalences ranged between 16% and 54% (Chadee 1998; Njeumi et al. 2002; Wilcke et al. 2002; Carvalho et al. 2003; Muehlen et al. 2003; Joseph et al. 2006; Ugbomoiko et al. 2007; Ratovonjato et al. 2008). However, prevalence and distribution of the disease are not documented in most endemic areas.

In settings where financial and human resources are scarce, policy makers need cost-effective and simple methods to estimate prevalence and severity of disease in affected populations (Anker 1991; Vlassoff & Tanner 1992; Macintyre 1999; Macintyre et al. 1999). As a consequence, rapid assessment methods have been developed for a variety parasitic diseases and health conditions, mainly in low-income countries (Anker 1991; Vlassoff & Tanner 1992; Macintyre 1999; Macintyre et al. 1999). For example, the macroscopic presence of haematuria (Lengeler et al. 1991, 2002a,b; Red Urine Study Group 1995), the identification of palpable nodules in the skin (Ngoumou et al. 1994; Whitworth & Gemade 1999; Kipp & Bamhuhiiga 2002) and the presence of elephantiasis and hydrocele (Gyapong et al. 1996, 1998a,b; Weerasooriya et al. 2008) have been used to estimate the prevalence of urinary schistosomiasis, onchocerciasis and lymphatic filariasis, respectively. Rapid assessment methods are commonly used to plan and monitor mass interventions, but also to detect parasitized individuals. Control of tungiasis at the community level has rarely been attempted (Heukelbach et al. 2001; Pilger et al. 2008), and rapid assessments methods are not available.

Because in endemic areas 95–98% of sand flea lesions are restricted to the feet (Heukelbach et al. 2002, 2007a; Ugbomoiko et al. 2007), we assessed different topographic areas of the feet to be used as a rapid method for the presence of sand fleas. An area was identified that would give a reliable proxy to estimate prevalence of tungiasis. The occurrence of tungiasis on periungual sites of the toes was the most reliable and practical approach to estimate overall prevalence and severity of disease.

Materials and methods

Study areas

We included data from 10 cross-sectional population-based surveys conducted in five communities between 2001 and 2008. Three communities are located in Northeast Brazil, two in Southwest Nigeria. The communities studied in Brazil were Balbino, a fishing village in Ceará State (Northeast Brazil); Morro do Sandras, an urban slum in the city of Fortaleza (capital of Ceará State); and Feliz Deserto, a rural community in Alagoas State. In Nigeria, data were collected in Yovoyan and Okunilaje, two small fishing villages in Lagos State (Southwest Nigeria, Figure 1).

Figure 1.

 Location of study areas (a) in Brazil (Ceará and Alagoas State) and (b) Nigeria (Lagos State).

All five communities were characterized by low socio-economic status, but showed distinct cultural and geographical characteristics. The study areas had in common that streets were not paved, and families were extremely poor (mean monthly family income equivalent to € 45.00). Illiteracy rates ranged between 15% and 30%. Whereas in Brazil, more than 90% of households had electric power supply, no access to electricity existed in the Nigerian communities. In Nigeria, the vast majority of houses were built of palm stems; and in Brazil, most houses were made of bricks or adobe. The main characteristics of the communities and their populations are depicted in Table 1.

Table 1.   Characteristics of communities and surveys
CommunityCountrySurvey No.Date of surveyTarget populationMain characteristics
Month/yearSeasonType of communityLocalization
BalbinoBrazil 107/2001Dry (beginning)605 individualsFishing village60 km south of Fortaleza, capital of Ceará State (northeast Brazil)
 208/2001Dry (end)
 304/2002Rainy
Morro do Sandra’sBrazil 403/2001Rainy (beginning)1468 individualsUrban slumFortaleza, capital of Ceará State (northeast Brazil)
 506/2001Rainy (end)
 609/2001Dry
Feliz DesertoBrazil 706–07/2003Rainy1146 individualsRural community120 km south of Maceió, capital of Alagoas State (northeast Brazil)
 810–11/2003Dry1087 individuals
YovoyanNigeria 902/2008Dry260 individualsFishing village63 km east of Lagos, capital of Lagos State (south Nigeria).
OkunilajeNigeria1003/2008Dry160 individualsFishing village56 km east of Lagos (south Nigeria).

As tungiasis is known to show a particular seasonal variation (Heukelbach et al. 2005), data were collected in different periods of the year (dry and rainy season). In Nigeria, surveys were conducted during dry season only, as in the rainy season isolated communities are not accessible. In addition, according to key informants, tungiasis virtually does not occur in these communities during the heavy rain falls typical of this region in Nigeria.

Detailed prevalence data on the Brazilian communities have been published previously (Wilcke et al. 2002; Muehlen et al. 2003; Heukelbach et al. 2005, 2007a). Clinical examinations were performed by investigators trained in an endemic area in Brazil and monitored by the same team leader. Field investigators were monitored regularly, and cross-checks were performed to reduce observer bias.

Study design

In all surveys, data were collected according to identical procedures. The body of study participants was examined clinically for the presence of tungiasis, excluding the genital areas. In the first surveys conducted in Brazil, <1% of individuals had lesions at ectopic sites but not on the feet. Thus, we considered this small error acceptable and, for the sake of logistic simplicity, did not assess ectopic lesions in subsequent surveys.

Diagnosis of tungiasis was made clinically, and disease was staged according to the Fortaleza Classification (Eisele et al. 2003): presence of a red-brown itching spot with a diameter of 1–2 mm; presence of a yellow-white watch glass-like patch with a diameter of 3–10 mm with a central dark spot; or a brown-black crust with or without surrounding necrosis. Sand flea lesions with evidence of manipulation with needles or thorns by the patient or a caretaker were also documented. The exact location, stage and number of lesions were documented.

Data entry and statistical analysis

Data were recorded on pre-tested standardized forms, entered into separate databases using Epi Info software package (version 6.04d; Centers for Disease Control and Prevention, Atlanta, GA, USA) and checked for entry errors. Then, data of all surveys were merged into a single dataset and transferred to Stata® software package (version 9.0; Stata Corporation, College Station, USA) for statistical analysis. As the number of lesions per individual was not normally distributed, medians and interquartile ranges are given to indicate average and dispersion of data. Fisher’s exact test was applied to compare relative frequencies.

Six topographic areas were defined and analysed as possible approaches for the estimation of prevalence of tungiasis: (i) right foot (including toes, interdigital areas, heels, lateral rim, sole); (ii) periungual area of right foot; (iii) periungual area of any foot (right or left); (iv) first right toe; (v) periungual area of first right toe; (vi) periungual area of any first toe (right or left). Periungual areas were defined as the locations around toe nails (Figure 2). True prevalence of tungiasis was based on the number of individuals with tungiasis on the feet in a community.

Figure 2.

 Right foot of a patient with approximately 100 sand flea lesions with several periungual lesions. Periungual site is exemplified by circles around the toe nail.

To assess the association between occurrence of tungiasis at one of the six topographic areas and the true prevalence or prevalence of severe disease, correlations and linear regression analysis were performed. Based on the linear regression equations, prevalences were estimated for each of the 10 surveys, and the absolute errors (estimated prevalence–real prevalence) were calculated. The same procedure was performed to estimate the prevalence of severe tungiasis (defined as the presence of >20 lesions in an individual).

We then selected the most useful localization to define a rapid assessment method based on the strength of association (R2) and operational aspects, such as time needed to perform an examination, simplicity and disturbance to the individuals. R2 indicates the per cent of variation of one variable that can be explained by linear relationship with another variable.

Ethical aspects

Studies were approved by the respective Ethical Review Boards (Ethical Review Board of the Federal University of Ceará, Ethical Committee of the School of Medical Sciences of Alagoas, Brazil; ad hoc Ethical Committee of Cascavel Municipality, and Ethical Committee of the University of Ilorin, Nigeria). Meetings with community health workers and village representatives were held prior to the studies, in which objectives were explained in detail. Informed written consent was obtained from individuals or their caretakers. In case of illiteracy, the informed consent form was read out by one of the investigators, and approval was obtained by thumb prints. In Nigeria, the traditional chiefs of Badagry (Ankra) and of local communities (Baales) also approved the study.

Results

General characteristics

In total, 7121 individuals were included in the 10 surveys. This corresponded to 57.8–91.1% of the respective target populations (Table 2). Median age of the populations ranged between 13.5 and 20 years.

Table 2.   Characteristics of study populations and infestation status in 10 population-based surveys
Survey No.BrazilNigeria
12345678910
  1. †Interquartile range.

  2. ‡>20 lesions.

Individuals examined/total population (% of target population)548/605 (90.6%)505/605 (83.5%)535/605 (88.4%)1185/1468 (80.7%)1192/1468 (81.2%)849/1468 (57.8%)1015/1146 (88.6%)990/1087 (91.1%)186/260 (71.5%)116/160 (72.5%)
Prevalence of tungiasis (95% CI)51.3% (47.1–55.5)52.1% (47.7–56.5)31.2% (27.3–35.2)33.6% (30.9–36.3)23.7% (21.3–26.2)54.4% (51.1–57.8)21.1% (18.6–23.6)28.9% (26.1–31.7)51.1% (43.8–58.3)41.8% (28.4–55.3)
Male/female prevalence ratio1.11.01.31.71.61.31.21.21.31.1
Number of lesions:
 Median (IQR†)2 (1–6)3 (1–8)3 (1–8)3 (1–9)2 (1–5)3 (1–8)2 (1–3)2 (1–4)6 (2–15)3 (2–11)
 Maximum145199781585011542457540
Prevalence of severe tungiasis‡ (95% CI)6.0% (4.2–8.4)6.5% (4.5–9.1)2.8% (1.6–4.6)2.2% (1.4–3.2)1.3% (0.8–2.2)4.7% (3.4–6.4)0.1% (0.0–0.5)0.8% (0.3–1.6)10.2% (6.3–15.5)5.2% (1.9–10.9)

Prevalence and severity of tungiasis varied considerably between and within communities (Table 2), with point prevalences from 21.1% (214/1015; survey 7 – Brazilian rural community, rainy season) to 54.4% (462/849; survey 6 – Brazilian urban slum, dry season). In general, highest prevalences of tungiasis and of severe disease were found in the dry season (surveys 1, 2, 6, 9, 10). The highest prevalence (10.2%; 19/186) of severe disease (>20 lesions) was found in one of the fishing communities in Nigeria (survey 9). The maximal number of lesions per individual ranged between 40 and 199 (Table 2).

In most surveys, men were more frequently infested (prevalence 22.9%; 105/459–63.5%; 235/370) than women (18.5%; 119/644–51.8%; 143/276) with a male/female prevalence ratio between 1.7 and 1.0 (Table 2). However, the difference was only statistically significant in the urban slum in Brazil (P < 0.0001).

Rapid assessment of prevalence in the communities

The six topographic areas considered as putative sites for rapid assessment and their characteristics are summarized in Table 3. Considering a high R2 value, time and simplicity of the procedure, we identified the periungual areas of the toes as the most promising site. In fact, the estimated prevalence based on this localization very reliably predicted the overall true prevalence of tungiasis (R= 96%; P < 0.0001; Figure 3a).

Table 3.   Comparison of possible topographic areas used for rapid estimation of prevalence of tungiasis
Topographical areaTime needed to examine one person†CommentsStrength of association (R2)
  1. †Relative values (+, ++, +++) are given, as the time to examine one person varies considerably within and between communities (according to presence and number of lesions, age of individual, socio-cultural setting and experience of investigator).

Periungual areas of feet+Individual can be examined with minor disturbance; sandals or thongs can be kept.
Diagnosis easily made by lay personnel
96%
Right foot+++Individual has to take off thongs and needs to stand up to show heel and plantar side of foot. Areas between and beneath toes are difficult to access, and toes need to be spread.
Requires more skills of investigators, increased risk of investigator bias.
95%
Periungual areas of right foot+Individuals can be examined with minor disturbance without taking off thongs/sandals.93%
Periungual area of any 1st toe+Individual can be examined with minor disturbance without taking off thongs/sandals.
Increased risk for errors at data collection.
93%
Periungual area of right 1st toe+Individual can be examined with minor disturbance without taking off thongs/sandals.88%
Right 1st toe++Individual has to take off thong and needs to stand up to show plantar side of toe. Area between toes I and II is difficult to access, and toes need to be spread. Increased disturbance.
Requires more skills of investigators, increased risk of investigator bias.
70%
Figure 3.

 Linear regression analysis for estimating prevalence of tungiasis (a) and of severe tungiasis (>20 lesions; b).

After estimating the prevalence of tungiasis (expressed as [estimated prevalence] = 1.12 × [prevalence on periungual sites] + 5.0), absolute errors ranged between −4% (survey 1) and +3.6% (survey 4; Table 4). The mean absolute error was 1.9%.

Table 4.   Estimated and true prevalences of tungiasis and severe tungiasis (>20 lesions)
 BrazilNigeria
12345678910
  1. †According to the equation y = 1.12 (x) + 5.0 where x is the prevalence determined to the rapid assessment method and y the estimated prevalence.

  2. ‡According to the equation y = 0.24 (x) – 3.4 where x is the prevalence determined to the rapid assessment method and y the estimated prevalence.

Prevalence of tungiasis on periungual sites37.8%42.0%21.1%28.8%18.8%44.3%15.0%20.8%44.1%30.2%
Prevalence of tungiasis
 True prevalence51.3%52.1%31.2%33.6%23.7%54.4%21.1%28.9%51.1%40.5%
 Estimated prevalence† (95% CI)47.3% (43.0–51.5)52.0% (47.6–56.5)28.7% (24.8–32.6)37.2% (34.6–40.0)26.0% (23.5–28.6)54.6% (51.1–58.8)21.8% (21.8–24.4)28.3% (25.5–31.2)54.4% (46.9–61.6)38.8% (30.0–48.3)
 Absolute error−4.0%−0.1%−2.5%3.6%2.3%0.2%0.7%−0.6%3.3%−1.7%
Prevalence of severe tungiasis (>20 lesions)
 True prevalence6.0%6.5%2.8%2.2%1.3%4.7%0.1%0.8%10.2%5.2%
 Estimated prevalence‡ (95% CI)5.7% (3.9–7.9)6.7% (4.7–9.2)1.7% (0.7–3.1)3.5% (2.5–4.7)1.1% (0.6–2.0)7.2% (5.5–9.1)0.2% (0.0–0.7)1.6% (0.9–2.6)7.2% (3.8–11.7)3.8% (0.9–8.6)
 Absolute error−0.3%0.2%−1.1%1.3%−0.2%2.5%0.1%0.8%−3.0%−1.4%

Presence of embedded sand fleas at the periungual areas of the feet also reliably estimated prevalence of severe disease using the equation (R= 76%; P = 0.001; Figure 3b). Absolute errors of estimated prevalence of severe disease, when compared to true prevalence of severe disease, ranged between −3.1% (survey 9; Nigeria) and +2.5% (survey 6, Brazil; Table 4). The mean absolute error was 0.9.

Discussion

Similar to other parasitic skin diseases, tungiasis is underestimated and can be considered a Neglected Tropical Disease (Heukelbach et al. 2001; Franck et al. 2003; Heukelbach & Feldmeier 2004; Heukelbach & Ugbomoiko 2007b; Feldmeier & Heukelbach 2009). It occurs in many resource-poor communities in endemic countries and causes considerable morbidity and loss of quality of life, widely unnoticed by policy makers, the pharmaceutical industry and health professionals (Feldmeier et al. 2003; Heukelbach 2005; Joseph et al. 2006; Heukelbach & Ugbomoiko 2007b; Ugbomoiko et al. 2007). Despite its obvious importance as a public health problem, data on disease occurrence in endemic communities and reliable data on the geographical distribution of tungiasis are not available. As a consequence, control of tungiasis has rarely been attempted (Heukelbach et al. 2001; Pilger et al. 2008).

Our data show that identification of tungiasis on periungual areas of the feet can be used to estimate prevalence of tungiasis and of severe disease in culturally and geographically distinct communities in South America and West Africa. The method is cheap, reliable and can be rapidly applied, with minimal disturbance of affected individuals. The assessment can be performed by paramedical workers or community members, as diagnosis of tungiasis can easily be performed by lay personnel living in endemic areas (Heukelbach 2005). In fact, in endemic areas, locals commonly affected by the disease diagnose tungiasis usually with a higher degree of certainty than health professionals (Heukelbach 2004).

The method is an alternative to time-consuming and sophisticated analyses for the precise assessment of morbidity used in a previous study on tungiasis (Kehr et al. 2007). As the number of lesions and morbidity are closely co-related (Kehr et al. 2007), the estimation of prevalence of individuals with more than 20 lesions indicates the occurrence of severe morbidity in a community. However, in our study, the strength of association was lower for the estimation of prevalence of severe tungiasis than for the overall prevalence. The rapid estimation method for severe tungiasis cannot be applied in the case of low prevalence of tungiasis at periungual sites, as in this case severe disease is rarely observed in a community.

The delimitation of tungiasis-endemic areas based on valid data is essential to highlight the epidemiological situation in a country or region. It is also a prerequisite for disease control at the population level. Thus, the rapid epidemiologic assessment method proposed fills these gaps in endemic areas.

In fact, rapid assessment tools were developed for the diagnosis of lymphatic filariasis with similar objectives: to determine the distribution of disease, identify high risk communities, and raise the attention of health policy makers (Gyapong et al. 1996, 1998a, b). As a result, about a decade later the ‘Global Programme to Eliminate Lymphatic Filariasis’ was implemented in collaboration with WHO. On the other hand, rapid assessment methods established for schistosomiasis and onchocerciasis were required to plan and monitor mass interventions programmes (WHO 1993, 1995; Red Urine Study Group 1995). In general, all rapid approaches provide valid data quickly (Gyapong et al. 1996; Macintyre 1999). We believe that our rapid assessment method for tungiasis will be similarly helpful in launching, planning and monitoring community control measures against the disease.

Rapid methods for other parasitic diseases, for example, for schistosomiasis, onchocerciasis and lymphatic filariasis, rely on the presence of indirect clinical or laboratory markers (Lengeler et al. 1991; Ngoumou et al. 1994; Gyapong et al. 1996; Kipp & Bamhuhiiga 2002; French et al. 2007; Weerasooriya et al. 2008; Ugbomoiko et al. 2009). However, the objective of our study was not to use an indirect marker for prediction of disease in an individual. The proposed rapid assessment for tungiasis was based on direct identification of the parasite as diagnosis of tungiasis in an individual is easy to perform by clinical examination. Thus, rapid diagnosis on the individual level with calculation of accuracy markers, such as sensitivity, is not needed in this case. On the other hand, reliable estimation of the true prevalence of tungiasis and severity of disease in a given community is useful. Our study shows that the strength of association when applying the rapid estimation was very high.

In communities of different size and with different point prevalences, absolute errors of the estimated prevalences were low and R2 values were high. The reliability of the proposed rapid assessment method did not vary considerably between seasons or populations with distinct socio-cultural characteristics. The wide diversity of characteristics of the examined populations indicates that this rapid assessment method may also be applicable in other endemic regions in Latin America, the Caribbean or sub-Saharan Africa. However, its external validity still has to be determined for other settings with particular ecological, epidemiological, geographical and socio-cultural characteristics.

We cannot rule out that inter-observer as well as intra-observer bias may have influenced the results, as surveys were conducted by different members of the research group, and during a rather long period. We aimed to reduce this source of bias by training all investigators performing the clinical examinations in one study site in Brazil, and by cross-checking quality of clinical examinations.

Conclusion

Our rapid assessment method will be helpful in the identification of communities at risk, as well as in planning and monitoring control measures aimed at the reduction of tungiasis in affected communities. The assessment can be performed by community health workers and paramedical personnel.

Acknowledgements

We thank the leaders and the people of the communities involved. We also thank Fundação Mandacaru for supporting the studies and Valéria Santos, Vania Santos (Brazil) and ‘Nick’ (Nigeria) for skillful assistance. Travel grants were made available by Deutscher Akademischer Austauschdienst (DAAD/Germany) and by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES/Brazil), through the PROBRAL German-Brazilian Academic Exchange Program, and by a PROÁFRICA grant from the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq/Brazil). L.A. received a PhD scholarship from CAPES. J.H. is research fellow from CNPq.

Ancillary