Measuring elimination of podoconiosis, endemicity classifications, case definition and targets: an international Delphi exercise

Background Podoconiosis is one of the major causes of lymphoedema in the tropics. Nonetheless, currently there are no endemicity classifications or elimination targets to monitor the effects of interventions. This study aimed at establishing case definitions and indicators that can be used to assess endemicity, elimination and clinical outcomes of podoconiosis. Methods This paper describes the result of a Delphi technique used among 28 experts. A questionnaire outlining possible case definitions, endemicity classifications, elimination targets and clinical outcomes was developed. The questionnaire was distributed to experts working on podoconiosis and other neglected tropical diseases in two rounds. The experts rated the importance of case definitions, endemic classifications, elimination targets and the clinical outcome measures. Median and mode were used to describe the central tendency of expert responses. The coefficient of variation was used to describe the dispersals of expert responses. Results Consensus on definitions and indicators for assessing endemicity, elimination and clinical outcomes of podoconiosis directed at policy makers and health workers was achieved following the two rounds of Delphi approach among the experts. Conclusions Based on the two Delphi rounds we discuss potential indicators and endemicity classification of this disabling disease, and the ongoing challenges to its elimination in countries with the highest prevalence. Consensus will help to increase effectiveness of podoconiosis elimination efforts and ensure comparability of outcome data.


Introduction
The possibility of eradicating and eliminating diseases has been discussed over several decades. 1,2 Historically, several diseases have been targeted for elimination or eradication, including yellow fever, yaws, malaria and leprosy. 3 Smallpox has been eradicated, while polio and guinea worm are in the end stages. 4 The London Declaration on neglected tropical diseases includes ambitious goals to eradicate guinea worm disease, and to eliminate lymphatic filariasis, leprosy, sleeping sickness (human African trypanosomiasis) and blinding trachoma by 2020. 5 Establishing elimination targets and endemicity classifications is vital for creating a standardized monitoring framework, allowing effective communication and ultimately ascertaining elimination of disease. Defining these targets and the associated metrics is fundamental to creating global consensus on disease control across endemic countries. Podoconiosis is a geochemical, non-filarial elephantiasis often occurring in barefoot subsistence farmers who are in long-term contact with irritant red clay soil of volcanic origin. 6 Podoconiosis was first identified in 1980 as an important cause of tropical lymphoedema, yet still continues to cause significant morbidity. 7,8 Globally, podoconiosis has been reported in 25 countries in tropical Africa, South East Asia and Latin America, with an estimated 4 million cases. 9,10 Recently, three countries (Ethiopia, Cameroon and Rwanda) have taken the initiative to map the distribution of podoconiosis. As plans for the mapping of podoconiosis continue to progress, it is imperative to have targets for elimination and a threshold by which to define endemicity. An endemicity threshold will help to identify areas considered endemic for podoconiosis and requiring interventions. This information will be useful for public health policy and planning.
In 2011 WHO identified podoconiosis as one of the neglected tropical diseases (NTDs). 11 Although this inclusion was an important step forward, to date there is no clear global strategy for the elimination or control of podoconiosis. Communities or countries require a pre-specified threshold to achieve the goal of elimination of podoconiosis as a public health problem. For other NTDs, elimination targets and endemicity classifications are often based on prevalence of the infective organism. For lymphatic filariasis (LF), the target is the reduction of the microfilaraemia rate to below 1% in previously endemic districts. 12,13 This is not possible with podoconiosis, since the agent is mineral rather than biological, so targets will rather be based on morbidity (lymphoedema) prevalence.
As global interest in podoconiosis increases it is important to have clear goals and indicators to measure progress towards control and elimination. Here we present the results of a Delphi assessment by experts working on podoconiosis and other NTDs. The objectives of this paper are 1) produce case definitions of podoconiosis for surveillance; 2) identify appropriate endemicity classes for podoconiosis (non-endemic, hypo-, meso-and hyperendemic); 3) identify targets for elimination of podoconiosis from a country; 4) define clinical outcomes to monitor progress.

Description of Delphi techniques
The Delphi survey technique is an iterative email or web-based survey to reach a consensus among a group of experts who are familiar with a subject area. 14 The approach uses well-designed sequential surveys in which each round depends on the response from the previous round. Each round uses feedback from the previous rounds from the same group of experts. Once the experts are selected, they are asked to provide judgement on the concepts of the subject under consideration using openended questions. 15 The next phase asks the same panelists to rank the items based on a Likert scale, measuring their importance. Following that, several rounds may be needed until a consensus on each item has been achieved based on the results of the quantitative analysis. 16

Design and participants
We used a Delphi method to define case definition, endemicity class, elimination targets and clinical outcomes of podoconiosis. We used a four-round Delphi method. In the first two rounds a small group of experts working on podoconiosis were involved in developing the tool. In the subsequent two rounds a wider group of experts working on podoconiosis and other NTDs was involved. These later rounds were conducted from August to November 2014 ( Figure 1).
The selection of experts for the Delphi process was done using a purposive sampling technique (expert sampling). For the first two rounds, experts with experience in podoconiosis research and programs were included. For the subsequent two rounds, the following eligibility criteria were used to select the experts: individuals working on NTDs at national or global level; individuals working on podoconiosis research or program management; senior podoconiosis professionals with in-depth knowledge and experience of control; experts involved in developing indicators and targets for other NTDs.
During recruitment, potential experts were approached (initially via e-mail or by telephone) and provided with a detailed explanation of the study and its objectives. A total of 28 national and international experts participated in the Delphi analysis process.

Questionnaire development
We developed the questionnaire after review of the literature and in consultation with podoconiosis experts. Experiences of developing elimination targets for other diseases were reviewed, and podoconiosis prevalence estimates from affected countries [17][18][19][20][21] were used to draft the questionnaire. The questionnaire included general information about the purpose of the study and the need for targets for podoconiosis control. A list of possible case definitions, endemicity classifications, elimination targets and monitoring indicators was provided, and respondents were asked to indicate their agreement or disagreement with each item and to provide alternative options for each (Supplementary information and Figure 1).

Data collection and analysis
The case definitions, endemicity classifications, elimination targets and monitoring indicators were developed between October 2012 and November 2014 in a series of four reiterative surveys. Feedback on each item was received and the questionnaire was modified. In the last two rounds, experts were asked to score items in terms of importance on a 9-point Likert scale for each indicator ranging from 1=not important to 9=extremely important.
For this study and in line with other related studies, 22,23 we set the consensus level as follows: 1) consensus of inclusion: >70% of participants scored the item ≥7; 2) consensus of exclusion: >70% of participants scored the item ≤5; 3) no consensus: item failed to meet either of the above criteria. The responses in the first-round survey were analyzed using descriptive statistics and the results were sent back to the experts for review and ratification. Items which were recommended for modification by the experts were revised and added to the second round questionnaire; new items suggested by the experts were also added to the second round questionnaire. In the second questionnaire, participants were asked to re-rank the items from the first round. Second round Delphi responses that reached 70% consensus were determined to be appropriate items for assessing different aspects of podoconiosis elimination. The final results were presented to experts for discussion and final consensus, leading to selection of the final targets and indicators.
Median and mode were used to describe the central tendency of expert responses. 15,24,25 The coefficient of variation (CV) was used to describe the variability of expert responses. 15,23 The CV is the ratio of the standard deviation of the responses of the experts on a specific item to its corresponding mean (average). Therefore, each survey item in each round yielded one CV value. 16 International Health

Characteristics of the experts
During the first round of the Delphi survey, questionnaires were sent to 35 experts, 28 of whom responded. These 28 experts were working at global or national level and were based in six countries. All of the experts had Masters level education or above. Twenty-five experts responded to the second round survey. Descriptive information on the experts is presented in Table 1.

Results of the first round Delphi survey
The results of the first round Delphi survey are shown in Table 2. The median score of items shown in Table 2 ranged from 6.5 to 9.0. Most of the items had CV score of less than 0.3 indicating good central tendencies for expert scores, and thus good consensus.
For case definitions, the median and mode scores were at least 7.5 and 8.0 respectively. All of the indicators in this category had CV scores of 0.3 or less. Apart from the 'probable case' definition, the rest reflected at least 70% agreement (≥7) among the experts. The experts suggested that the 'probable case' definition should be modified. In the endemicity category, all the items had a median score of at least 7.5 and a mode score of 9.0. In this category at least 70% of experts were in consensus over all the indicators, and apart from the 'non-endemic' category, all had CV scores of 0.3 or less.
In the 'elimination target' category, of the ten items, nine had median score ≥7.0 and mode score of 9.0. Apart from three items, consensus was found among 70% or more of the experts. Two of the items with less than 70% agreement were related to lymphoedema management. The experts commented that the term 'adequate treatment' should be clearly defined.
In the 'monitoring clinical outcome' category, all five items had median score of ≥8.0 and mode score of 9.0. Of the five items, three reached consensus among >70% of the experts. The two items with less than 70% agreement were 'treatment success' and 'measure of proper foot hygiene'. In the former, the experts recommended that the term 'incapacitated' should be defined. They also indicated that the measurement of foot hygiene is poorly defined, and essentially a subjective measurement.
None of the items fulfilled the criteria of consensus of exclusion, therefore none were deleted from the second round of the survey. In the 'case definition' category, experts suggested that 3 months duration of the symptoms was too short due to the chronic nature of podoconiosis lymphoedema, so the case definition was revised to reflect lymphoedema duration of at least 1 year. In addition, the experts recommended that long term visitors should not be included in the case definition since podoconiosis requires an extended period of exposure to red clay soil in order to occur. 9,26 Long term visits (defined as at least 6 months) would not be sufficient exposure for podoconiosis occurrence. For treatment of lymphoedema, the experts indicated that the 90% target was low and suggested it should be raised to 95% for round 2.

Results of the second round Delphi survey
All of the case definitions, endemicity classes, elimination targets and clinical outcomes reached >70% consensus among experts (Table 3). The median and mode scores ranged between 8.0 and 9.0, and the CV score was less than 0.3 for all of the indicators and case definitions, demonstrating a high degree of consensus among the experts.

Case definition of podoconiosis for surveillance
Three levels of case definition were arrived at: 1) suspected case: any lymphoedema of the lower limb of any duration; 2) probable case: any lymphoedema of the lower limb present for more than 1 year in a resident of an endemic area; 3) confirmed case: lymphoedema of the lower limb present for more than 1 year in a resident of an endemic area, for which other causes (e.g. onchocerciasis, LF, leprosy Milroy syndrome, heart or liver failure) have been excluded.

Elimination targets
Consensus was reached on the following elimination targets: Podoconiosis is defined as being eliminated from an endemic district or implementation unit if the prevalence of untreated podoconiosis is less than 1% (among individuals ≥15 years old), and more than 95% of lymphoedema cases are treated adequately after 10 years of program implementation. Podoconiosis is defined as being eliminated from a country when the following four targets are achieved: 1) Prevalence of untreated podoconiosis is maintained at less than 1%(among individuals >15 years old) in 100% of sample villages over a 10-year period. 2) Prevalence of early signs of podoconiosis among children 10 to 15 years after 10 years of control program implementation is less than 1 in 10 000. 3) Greater than 95% of the population in endemic districts wear protective shoes. 4) Greater than 95% of lymphoedema cases are treated adequately. The assumption here is that if the prevalence in an implementation unit is less than 1%, the formal health sector can manage patients reactively, provided the formal health sector is trained and equipped to do so, and prevention activities can be run by the health system, without the need for a control program.

Discussion
In 2011 WHO defined podoconiosis as one of the NTDs, including it under the 'other conditions' category. 11 While this was a step forward for the control of the disease, no clear targets or endemicity classifications exist. Such targets and definitions are important to consolidate efforts to control podoconiosis. Historically, podoconiosis appears to have been eliminated from northern Africa due to socioeconomic development and widespread shoewearing practices, 7 suggesting that podoconiosis elimination is within our reach.
Using a Delphi assessment process, we defined the endemicity threshold for podoconiosis to be ≥1%, the underlying assumption being that in the absence of point-of-care diagnosis International Health Section 1: case definitions Suspected case: any lymphoedema c of the lower limb of any duration (at this stage we do not expect to make a differential diagnosis but need to record the actual numbers of people with lymphoedema, even if a medical diagnosis has not been confirmed). Confirmed case: lymphoedema c of the lower limb present for more than 3 months in a resident of, or long term visitor to an endemic area, for which other causes have been excluded (onchocerciasis, LF, leprosy, Milroy syndrome, heart or liver failure, etc). Defaulter: a patient who has been on treatment and whose treatment was interrupted for 2 or more consecutive months. overestimation of cases may occur. In areas where there is no podoconiosis or LF, the underlying prevalence of lymphoedema is approximately 0.1-0.5%. 27 Hence a threshold of ≥1% will help take into account such residual lymphoedema cases. The second rationale for such a cut-off point is that even if the 1% of cases are confirmed to be due to podoconiosis, the local health system can respond without the need for establishing a control program. Nonetheless, even in such cases, the health providers working in the system should be given clear training and be equipped with the necessary supplies. We also generated endemicity classifications for podoconiosis. Evidence from national and small-scale surveys indicates that the prevalence of podoconiosis ranges from 1-10% with few prevalence estimates >10% per district. [17][18][19]21 Three endemicity classes will allow areas within the highest category to be prioritized and will enable monitoring of progress within these areas. These thresholds are intended more to prioritize endemic implementation units (administrative units used as the basis for making decisions about morbidity management), than to indicate any particular biological consequences of disease prevalence in contrast to infectious diseases. 28 We have also produced targets for monitoring elimination of podoconiosis. Since there is no biological cause of podoconiosis, we cannot aim for interruption of transmission; the focus will be on access to preventive and treatment services. The combination of these elimination targets is intended to avoid the debates such as those surrounding the leprosy elimination targets. 29 Our targets address access to morbidity management for the backlog of cases as well as prevention of new disease (monitored through shoe-wearing practice and identification of new cases among children aged between 10-15 years). We suggest a 10 year time frame for the evaluation of control and elimination programs, given the challenge of addressing the backlog of cases and addressing behavioral change with regard to preventive behaviors such as shoe wearing and regular foot hygiene practice. This will also enable progress in reducing incident cases to be monitored. Table 4 summarises the sources of data from which measures of key indicators are likely to be derived. These include routine sources (health management information systems and clinical records) and specific patient and community surveys.
For the first time, we have developed a case definition for podoconiosis which can be used within the routine surveillance system. The sensitivity and specificity of our definition must be evaluated in future studies. These broad definitions will capture individuals with lymphoedema who may require foot care services. With global movement towards integrated foot care services, 30 identifying the specific causes of lymphoedema may not be necessary. While the definition here is sufficient for surveillance purposes, if identification of the specific cause of lymphoedema is necessary, confirmatory tests to exclude other potential causes of lymphoedema will be required until podoconiosis-specific point-of-care diagnostic tools are developed. Other causes of lymphoedema are excluded on the basis of history and physical examination. 18,31 The diagnosis of podoconiosis is based on a history including details of where in the lower limb the swelling started from, the age at which swelling started and the presence of a family member with similar conditions. 31,32 Clinical examination for intact sensation in the lower leg, and the absence of swelling in the groin area, is also important. In addition, podoconiosis patients must be negative for all tests for LF. 31 Key indicators for monitoring progress: >95% of the population in endemic districts consistently wears protective shoes (measured for the last 1 year).  Section 1: case definitions Suspected case: any lymphoedema c of the lower limb of any duration (at this stage we do not expect to make a differential diagnosis but need to record the actual numbers of people with lymphoedema, even if a medical diagnosis has not been confirmed). Defaulter: a patient who has been on treatment and whose treatment was interrupted for 2 or more consecutive months. Measuring clinical outcomes of indivduals with podoconiosis is a complex phenomenon. Although simple lymphoedema management for people with podoconiosis exists (i.e. foot hygiene, foot and skin care, wound care, compression, exercises and elevation, treatment of acute attack and use of shoes and socks to reduce further exposure to the irritant soil), [33][34][35] it has not been properly evaluated or standardized. The duration of treatment and followup frequency are areas currently under research. Therefore, only three areas of clinical outcome (treatment completion, defaulter and treatment success) are considered in the 'monitoring clinical outcomes' definition (Table 3). Other areas of clinical outcome will be defined once the ongoing randomized controlled trial 36 is completed. Currently, a patient is regarded as having completed treatment when he/she has followed supervised morbidity management for at least 3 months. Within these 3 months, if a patient misses two consecutive monthly clinical appointments, they are considered a defaulter. These end points are important in monitoring the success of morbidity management programs.
The interventions for both prevention and treatment of podoconiosis are simple, 9,34 and short term goals such as restored function and improved quality of life can be achieved among lymphoedema cases after just 3 months of treatment. 33,34 Promotion of shoe wearing for podoconiosis prevention may have multiple health benefits. 37 Integrated morbidity management in the context of foot care may become an important approach to avoid duplication of efforts and to utilize the available resources efficiently. It is possible to integrate foot care services for podoconiosis with those for Buruli ulcer, LF and leprosy. 30 Currently the morbidity management services are provided by very few faith-based organizations and NGOs in endemic countries. 33,34 Provision of free-ofcharge or low-cost prevention and treatment through government programmes will be critical. People with podoconiosis are often poor 38 and marginalized due to their condition, and introduction of user fees might hinder their access to the services.
Prevention of podoconiosis is an important component of its elimination. Promotion of shoe wearing as a health intervention should be advocated for. 37 For those individuals who cannot afford shoes themselves, subsidized shoe distribution should be considered. Continued research should focus on identifying point-of-care diagnostic tools for podoconiosis to ascertain cases and verify elimination. At the start of elimination programs, definitive diagnosis may not be a priority, but as the 'end game' approaches, robust, sensitive and specific diagnostic tests will be required. In the earlier phases of elimination, a syndromic approach may suffice for the diagnosis of podoconiosis. 31 In the long term, identification of biomarkers of podoconiosis will be vital.
With all these important findings our study is not without limitations. First, research into and intervention against podoconiosis are currently present in only a few endemic countries, so there are few experts in podoconiosis. Nonetheless, we have included almost all the top experts in podoconiosis research and programing working in endemic counties globally. Second, there are many important questions unanswered so far about podoconiosis, such as treatment duration and outcomes. This means that there is a need to define these indicators once appropriate evidence is generated. Third, full operationalisation of several components of the case definitions and elimination targets is still in progress. Operational definition of 'adequate treatment' of lymphoedema is an issue being explored in relation to lymphatic filariasis as well as podoconiosis. Similarly 'practicing proper foot hygiene' and Key indicators for monitoring progress: >95% of the population in endemic districts consistently wears protective shoes (measured for the last 1 year). International Health 'wearing protective footwear' will both require robust operational definitions.

Conclusions
Untreated lymphoedema due to podoconiosis causes significant deformity, disability, social and productivity loss in endemic countries. [38][39][40] The disabilities associated with podoconiosis are preventable with simple preventive measures (consistent shoe wearing and foot hygiene) and early lymphoedema morbidity management. 18,34 These simple interventions can be integrated into the health system and delivered at low cost. 41 It is time to capitalize on the experience of some endemic countries and provide the intervention sustainably and at scale, with clear aims and targets. 33,35,42,43 Defining elimination targets and endemicity classification is critical for mobilizing communities and stakeholders and ensuring accountability in relation to elimination of podoconiosis. The indicators described here may serve as a starting point towards a global strategy for podoconiosis elimination.
Authors' contributions: KD and GD conceived the study and designed the study protocol; KD carried out data collection, analysis and interpretation. KD, SW, OS, EM and IU and GD drafted the manuscript. SW, OS, EM and IU critically revised the manuscript for intellectual content. All authors read and approved the final manuscript. GD is guarantor of the paper.