Prospective evaluation of onchocerciasis control in the OCP: the application of a transmission model

a ) WORLD HEALTH ORGANTZATION ONCHOCERCIASIS CONTROL PROGRAMME IN WEST AFRICA PRO SPECTTVE EVALUATION OF ONCHOCERCIASIS CONTROL IN THE OCP The application of a transmission model DONORS'SEMINAR Paris, l8-19 October 1989 Information document -1- I. INTRODUCTION Modelling was introduced in the OCP for the analysis of the epidemiological evaluation data and the prediction of epidemiological trends durirtg the vector control period. Using a host-parasite model it was in 1985 predicted that the prevalence of infection would fall to insignificant levels after l5 years of successful vector control[11. The latest observed trends from the well-protected central OCP area are still consistent with those predictions and have shown the expected collapse of the prevalence of infectionlzl. The focus in modelling has now shifted to more dynamic epidemiological questions such as the risk of recrudescence of onchocerciasis after the cessation of vector control and thi impact of ivermectin treatment. A comprehensive transmission model has been developed in collaboration with the Institute of Public Health of the University of Rotterdam. This model allows the prediction of long term epidemiological trends after intervention with the currently available control methods, i.e. vector control, large scale ivermectin treatment and combinations of these two methods. In 1988 some results of preliminary simulations were presented to the EAC, which gave high priority to the lurther development and application of epidemiological modelling in order to provide the best possible information on which to base operational decisions. Since then the model has been improved, systematically tested and sensitivity analyses have been undertaken. Early l9E9 a start has been made with the utilization of the model in the prospective evaluation of alternative control strategies in the OCP. The present report provides a summary of the main findings obtained to date. I.1 THE TRANSMISSION MODEL A detailed description of the transmission model, and the computer program ONCHOSIM, is given elsewhsls[3l. The model has been quantified using literature data, OCP research data, and by fittirrg to epidemiological and entomological evaluation data. A systemaric sensitivity analysis of the most important parameters for the epidemiological trends during the control period has indicated that the average longevity of O.volvulas is in the range of 8-ll years. The best fits were obtained with an averaee longevity of 9-10 years, and a variability which was such that 95% of all adults female worms had a lifespan of less than l3-14 years. In this best fit, a fully productive adult female worm contributes on average 4-5 mf to the mf load per skin snip. This implies that in a village with a CMFL of 60 mf/s a human adult harbours on average 25-35 productive adult female \ilorms. Around this best fit there is a wide range of other quantifications which also produce an acceptable fit to the observed data. l. Onchocerciasis Co^ntrol-Programme in the Volta River Basin Area: Progress report of theWorld Health Organizatibn for 19E5. Geneva, OCp/pR/85, t985 2. Remme J., De $le G. and van Oortmarssen G.J. The predicted and observed decline in onchocerciasis infection during l4 years of successf ul Simulium control in West Africa with reference^to the reproduclive iifespan of Onchocerca volvulus. Bulletin of theWorld Health Organizatio,n (in press) 3. Plaisiqr^{.!.,--v-an.oortmarssen G.J., Habbema J.D.F., Remme J. and Alley E.S.ONCHOSIM: A computer program for modelling ttre transmission ild'control of onchocerciasis. Submitted for publication -, - II. PROSPECTTVE EVALUATION OF ALTERNATIVE CONTROL STRATEGIES. Since January 1989 the transmission model has been used in the prospective evaluation of the long term impact of control strategies based on vector control, large scale ivermectin treatment and combinations of these two methods. This has involved the simulation of onchocerciasis transmission and infection in a human population in an endemic focus, and of the epidemiological impact of control, over a period of decades. A simulation begins hundred years before the start of control in order to generate a stable human population with respect to age, onchocerciasis infection and blindness. Subsequently, a period of 50-75 years after the start of control is simulated. In this report five indices are used for the evaluation of the impact of control, i.e. the prevalence of mature female worms, the prevalence of microfilariae in the skin snips (assuming two skin snips.per person), the Community Microfilarial Load (CMFL), the Annual Transmission potential(ATP) and the prevalence of blindness. The results are presented in graphs which show the relative changes in these indices, taking for each index the pre-control value as 1000/0. 2.1 CESSATION OF YECTOR CONTROL AND THE RISK AND SPEED OF RECRUDESCENCE 2.1.1 The impact of successful vector conlrol Fig.l shows the predicted impact of l5 years of successful vector control, i.e. a period of control as foreseen in the Long Term Strategy. Fig.1. Impact of 15 years vector control 120 110 100 90 80 70 60 50 40 30 20 l0 0 Vector control a L I q) L o IR - Prev. mf in skin snip - Prev. blindness ---." CMFL ----- Prev.adult fernale worms ATP I 970 r 990 2010 2030 2050 -3- The trends in the five indices are very similar to those observed in the central OCP area. Vector control results in the virtual interruption of transmission and the ATP drops to zero after the start of control. The CMFL shows a near linear decline after short delay equal to the pre-patent period and falls to values close to zero after l0-12 years of control. The prevalence of adult worms and the prevalence of mf in the skin snip decrease initially slower but show subsequently an accelerated decline towards insignificant levels after l5 years of control. The prevalence of blindness hardly changes during the first 5 years of control when there is still some incidence of blindness in cases with high mf loads. After this period the prevalence of blindness decreases slowly as a result of mortality among the blind. After l5 years of control the local reservoir of the parasite has fallen so low that there is no longer any incidence of infection when the vector is allowed to return but when there is no reintroduction of the parasite. 2.1.2 Recrudescence alter pre-malure interrupliort of veclor control Fig.2. Recrudescence after 12 years vector control o a 6 oI o o I oLA o n rzo tlo loo 90 80 ?o 60 50 40 30 20 10 0 1970 r990 - Prev. ml in skin snip - Prcv. blindnes6 2 010 2030 ---..'. CMI'L ATp ---' Prev adult femrle rorrns 2050 The objective in the central OCP area is to prevent recrudescence of onchocerciasis. Fig.2 shows the results of a simulation of recrudescence if vector control is interrupted too early. The fact that recrudescence is taking place will initially not be obvious. The prevalence of mf continues to decrease for several years. New infections occur immediately after the return of the vector, as is shown by the increase in the prevalence of adult female worms, but it will be difficult to detect the recrudescence of infection by entomological methods and cross-sectional skin snip surveys. However, after a period of several years of apparent calm, the prevalence of mf will show an accelerated increase and return to the pre-control value if no intervention is undertaken. The other indices will incroase in a similar fashion but with a delay of 5 to 30 years. The increase in prevalence will cause an accelerated rise in the ATP, and this will result in a higher incidence of (super)infection and an increase in the CMFL. Once the CMFL has reached the pre-control level the parasitological situation will have returned to an equilibrium situation. It will take several more years before there has been a complete recrudescence of onchocercal blindness. Vcctor control -1- 2.1.3 Factors determining the risk and speed of recrudescence The risk of recrudescence after a period of vector control depends in the first place on the size of the parasite reservoir which is a function of the duration and effectiveness of vector control (see 2.1.4) and of the pre-control endemicity level. Other important factors are the actual biting rate per person after the return of the vector and the distribution of the remaining parasites in the human population. Sensitivity analysis has shown that the more heterogeneous the human population is with respect to exposure and intensity of infection, the greater are the chances of recrudescence. Similarly, the significance of immigration by infected individuals from outside the OCP area is very much determined by their intensity of infection and their subsequent exposure to the vector. A few fishermen immigrating from a place like Asubende will be far more dangerous than a large group of Fulani's from endemic areas in Nigeria. The main parasitological determinants of the risk of recrudescence are the requirement and probability of mating, and the relationship between the number of potentially infective larvae per fly and the skin mf load of the human host (the so-called'Ll-uptake'function). The fly feeding and transmission experiments have enabled a satisfactory quantification of the 'L I - uptake' function but the quantitative aspects of mating of O .volvulu.s remain largely unknown. The speed of recrudescence depends mainly on the biting rate and the'Ll-uptake'function. 2.1.4 Required duration of vector control A large series of simulations has been run to determine the risk of recrudescence in relation to the durition of vector control after taking the uncertainty in the quantification of the most sensitive parameters into account. Simulations of successful vector control indicate that recrudescence is certain if larviciding is stopped after 12 years of control. The breakpoint appears to be a control period of l3 years. The parasite reservoir died out in a large number of simulations of l3 years of control but in most simulations with very high biting rates and unfavourable assumptions on the probability of mating and the initial slope of the Ll-uptake function, there was recrudescence of infection. In simulations with l4 years of successful vector control.there was only recrudescence in simulations with unlikely combinations of extreme assumptions. In case of incomplete vector control the required duration of larviciding is longer and depends on the effectiveness of control. A reduction in the ABR of some 980/o has nearly the same impact as complete interruption of transmission. But a reduction in the ABR of 900/o would require a continuation of vector control for 25 years before a situation without risk of recrudescence has been reached. 2.1.5 EAC questions concerning recrudescence During its 1988 session the EAC requested the Programme to provide it with preliminary answers to the following questions concerning recrudescence (JpC9.3 par. 77.2): (a) lVhat is understood by recrudescence? (re-emergence ol the disease or infection, how man1, cases and at what stage). The simulations of recrudescence clearly show that it is important to differentiate between recrudescence of disease and recrudescence of infection. By the time the prevalence of blindness begins to show a significant increase, the parasitological situation will already have deteriorated to the pre-control situation. At that moment the achievements of the OCp will have been no -6- more than a temporary relief of a few decades without any long term impact. Recrudescence of infection will start long before recrudescence of disease. It will initially build up slowly and this favours the chances of success by intervention at this stage. Surveillance and maintenance activities should therefore aim at the detection and control of recrudescence of infection rather than disease. Re-emergence of infection is not the same as recrudescence of infection which implies a progressive increase in prevalence and intensity of infection. Isolated infections are likely to occur throughout the central OCP area after the interruption of larviciding. The request of the EAC to be provided with quantitative guidelines of what constitutes the first evidence of recrudescence is therefore very appropriate even though it is very difficult to give such guidelines. Available simulation results suggest that arr annual incidence of new infections of more than l%-2% is evidence of recrudescence while an incidence of less than 0.50,6 would appear to represent isolated infections without long term epidemiological consequences. These figures should be taken as very preliminary guidelines until a systematic series of simulations is undertaken to investigate this question. (b) li'hat type ol epidemiological surveillance? (active, passive case detection). According to the above reasoning, the objective of surveillance should be to determine the incidence of new infections and to assess if there is a situation of recrudescence of infection. Passive surveillance may provide additional information but the estimation of the incidence of infection can obviously only come from active surveillance based on longitudinal surveys in selected indicator villages using skin snipping, or maybe in the future using a highly specific immuno-diagnostic test. 2.2 I.ARGE SCALE TVERMECTIN TREATMENT 2.2.1 Modelling and quantification o/ the ellecl o/ ivermectin treatment. Ivermectin is an effective microfilaricide which reduces mean microfilarial loads by 960/o-990/0. In has also been claimed to be a long lasting microfilaricide because of the slow microfilarial repopulation of the skin in some clinical trials. However, several community trials have shown much faster mf repopulation dynamics following ivermectin treatment. The reasons for these different trends are not yet understood but they are likely to be important for the prediction of the long term impact of ivermectin treatment. At present the effect of ivermectin has been quantified using the results form the community trial in Asubende which has yielded the most extensive longitudinal skin snip data after ivermectin treatment. The mf repopulation in this focus was among the fastest observed to date, and from this point of view the model quantification is somewhat pessimistic concerning the impact of ivermectin treatment. However, in the model predictions it is also assumed that treatment has a cumulative effect on the reproductive potential of the adult female worm which is reduced by l0o[ at every treatment of the human host. This assumption is probably too optimistic because the latest results from repeated ivermectin treatments do not seem to provide evidence of a cumulative effect on the adult female worm. The treatment coverage, compliance and exclusion criteria have also been quantified using the results of the Asubende trial. 62.2.2 Intpact ol large scale ivermectin treatmenl on lransmission and disease The predicted impact of 25 years of annual large scale ivermectin treatment is shown in Fig. 3. In this simulation it has been assumed, rather unrealistically, that the treatment coverage does not decrease during the full 25 year period. Fig. 3 shows that every mass treatment results in an immediate drop in prevalence of skin mf and in the CMFL. During the subsequent l2 months both indices increase again as a result of mf repopulation. After the first treatments the prevalence returns nearly to the pre-treatment level but the CMFL stays below 40% of its initial value. The treatment has reduced transmission during the first post-treatment year by more than 50%. All these predictions are very similar to those which have been observed in the community trial of ivermectin in Asubende. Fig.3. Impact of ?5 years annual ivermectin treatment 120 ll0 100 90 EO 70 60 50 40 30 20 l0 0 1985 2005 20?5 ?045 2065 - Prev. mf io rkio snip " -'-" CMFL -,,,.,, ATp ---- Prcv.adult female worrns - Prcv. bhDdncss The reduction in transmission and in the incidence of infection has a cumulative effect over time and results in a steady reduction in the intensity of onchocerciasis transmission. However, the Iong term impact on the parasite reservoir is not sufficient to prevent the immediate and rapid recrudescence of the intensity of infection and disease when treatment is stopped after 25 years. Many other simulations of scenarios based on annual or six-monthly ivermectin treatment in non-controlled areas gave the same type of results. It seems therefore that large scale ivermectin treatment cannot achieve the progressive reduction and final elimination of a parasite reservoir in an endemic area. In the absence of other methods of intervention or a major cumulative effect on the adult worm, ivermectin treatment will have to be continued indef initely. Though large scale ivermectin treatment does not appear to be appropriate for transmission control in endemic areas, it is likely to be very effective for disease control as long as a sufficient coverage can be maintained. Comparative simulations of the incidence of blindness during control by larviciding versus control by ivermectin treatment clearly show the greater impact on disease with ivermectin treatment during the first 5-8 years of control. After this period the two methods are comparable from the point of view of disease control. o 6 o! I o IR Ivermectin -7 - The incidence of blindness has been modelled in a simplistic way as a function of mf accumulation by the human host during his life time. This enabled a sufficiently accurate simulation of the impact of vector control. However, recent results from ophthalmological follow-up surveys during the community trials of ivermectin suggest that ivermectin treatment may result in regression of early stage ocular lesions. The current model may therefore even underestimate the potential of ivermectin treatment for disease control. 2.3 RECRUDESCENCE CONIROL WITH IVERMECTIN The long term control of onchocerciasis in the OCP area will depend largely on the potential of ivermectin treatment for recrudescence control. According to the conclusions from section 2.1.5, recrudescence control should aim at controlling the recrudescence of infection. The intensity of transmission during the first years of recrudescence will be low and the parasite reservoir will still be fairly small. Various simulations have indicated that in this type of epidemiological situation the recrudescence of infection can be interrupted by ivermectin mass treatment. An example of such a simulation is given in Figure 4. Even when annual ivermectin treatment is only started 7 years after the beginning of recrudescence, it may prevent the further increase in the parasite reservoir, and l0 years of ivermectin treatment may be enough to fully control the situation. Fig.4. Recrudescence control with ivermectin Start control after 7 years of recrudescence tzo I r.o roo 90 80 70 60 50 40 30 20 l0 0 r 970 I 990 2010 2030 2050 The time lag between the beginning of recrudescence and the initiation of large scale ivermectin treatment will be crucial. Initially the recrudescence will be slow but once a certain threshold level of the parasite reservoir has been reached, ivermectin treatment will no longer be able to achieve a time-limited solution. Fig. 5 shows that l5 years of recrudescence will have established a parasite reservoir which cannot be brought under control with l5 years of annual ivermectin treatment' The simulations clearly indicate that the earlier recrudescence control is started, the Sreater are the chances of success and the smaller will be the effort required. This underscores the importance of a systematic longitudinal surveillance after the cessation of larviciding. 7{ 3+ 't crn 2,oO1q) 6 L o I qJ ke )e ? vlo- -r{q - Prcv. mf rn skrn snrp - Prev. blindness ----" CMFL ----- Prev.adult female worms -'------ ATP rcctor coDtrol 12 ycur no coDtrol ? vcur ' l"crm.ctln l0 t"u! I l-E- Fig.5. Recrudescence control with ivermectin Start control after 15 years of recrudcscence a) a aa > o o a I o r o * 120 110 100 90 60 ?o 60 50 40 30 zo l0 n 1970 1990 - Prev. mf in skin snip - Prcv. bliDdDcss 2010 ?0J0 '.'"-' CMFL ...... ATp ----- Prev.adult femele wormr 2050 2.4 COMBINATIONS OF YECTOR CONTROL AI{D TVERMECTIN TREATMENT Many simulations have been done of combinations of vector control and ivermectin treatment. These included the optimal onchocerciasis control strategy which consists of interruption of transmission and elimination of the local parasite reservoir by l4 years of successful vector control together with disease control by ivermectin treatment during the first 5-8 years. The main objective of the simulations of combined strategies was to investigate if there are combiriations which would be less costly than the current control activities, and which would still enable the achievement of the objectives of the OCP. The answer was generally negative, and mainly because of the limited effect of ivermectin treatment on transmission, which is insufficient to enable cost-effective reductions to be made in the level of vector control. The only alternative, which would result in major savings, is to replace vector control by ivermectin based mass treatment for the purpose of disease control for an unlimited period of time, as described in section 2.2.2. Some combinations may be investigated further, such as the use of ivermectin treatment during the last few years of control in order to reduce the control period by one year and to take care of isolated problem foci. But the simulation of combinations tended basically to highlight the different strengths of the two methods, i.e. interruption of transmission by vector control and prevention of onchocercal pathology by large scale ivermectin treatment. no control 15 ycrr lwrmcctiu 15 ycua t2 It vecto control