|Vertical Analysis of Human African Trypanosomiasis (Institut Tropical - Tropical Institute, Antwerp, Belgium, 1997)|
|III. The Kasongo study|
In 1977 Mercenier published a paper in which he analysed detection mode and presenting symptoms of a series of 384 trypanosomiasis cases in the Kasongo health zone (Maniema, then Zaire), where a number of functioning general health care structures were available, and where a specialised (mobile) trypanosomiasis control team was also active (Mercenier, 1977). This analysis showed that in the examined series 292 (76%) of detected cases were diagnosed after a spontaneous consultation, and that of the 68 who had not yet consulted and for whom the information was available, only 3 were symptom-free (as determined a posteriori). Thus the potential of passive case detection through permanent, decentralised general service points, in this series, appeared to be very important.
As this ran contrary to the prevailing (though mostly implicit) trypanosomiasis control doctrines, this statement was sufficiently controversial to make the idea grow to set up a carefully designed population based prospective study into early signs and symptoms that could be used to detect the disease. The Kasongo district provided a good setting for such a study, since it contained old and more or less "stable" trypanosomiasis foci. In one of these, in the area of the Kongolo health centre, a cluster of villages was selected, containing some 2500 people. Based on the experience of active and passive case detection in the past, the annual detection rate in this population was about 10 to 12 new cases per 1000 inhabitants. A prospective multi-round survey over two years was thus expected to yield some 50 cases of trypanosomiasis.
The study was designed around the following questions:
- What is the predictive power of indicators of
risk, detectable in a rural setting with a permanent and accessible first line
health service, and to what extent can they contribute to early detection of
trypanosomiasis with T.b. gambiense?
- What would be the marginal effectiveness of an active case finding strategy as compared to passive case finding in an adequately functioning basic health service?
"Indicators of risk" (more accurately, maybe, "risk markers") were understood as, on one hand, a number of variables about which no prior hypotheses are formulated, but for which the individual information would be available in the study (like sex, age, area of residence, etc.). On the other hand there are a number of variables about which prior hypotheses are formulated and tested. Examples are cervical gland status, serological status, history of previous treatment for trypanosomiasis, and declared symptoms or health related complaints, expected to possess (some) significant or relevant predictive or prognostic power. The choice of these variables is justified by the feasibility of their identification in a first line health service setting. With the hindsight we now have, we must add that the autoagglutination phenomenon was not, at the time of the study design, within our field of vision - regrettably.
In each of the survey rounds the entire study population was to be screened for four signs:
- a positive serological test result, as determined by the Indirect Fluorescence Antibody Test (IFAT), carried out on dried blood, collected on Whatman 4 paper, in the laboratories of Kasongo and of the Institute of Tropical Medicine in Antwerp. Results of this IFAT test were expressed in terms of a semi-quantitative scale with four possibilities ("negative", "weakly positive", "clearly positive", "strongly positive");
- the presence of "typical" neck glands (i.e., in practice, which can be punctured with a needle);
- declaration of one or more "suggestive" symptoms, according to a preestablished list, with a duration of at least 30 days;
- being an infant (less than 2 years of age) born to a mother who was detected as a parasite carrier.
Every individual who screened positive for one of these criteria was eligible for further parasitological investigation, in the same and possibly following rounds until found positive, with a combination of tests including gland puncture (if "typical" glands were present), microscopic examination of fresh blood, microscopic examination of a Giemsa coloured thick smear and of the "buffy coat" after centrifugation in heparinised 100 ml capillary tubes. If any of these were positive, cerebrospinal fluid was tapped and examined (white cell count and microscopic search for parasites).
Three months after a thorough census of the study population, the first round was started in January 1989.
According to the protocol, 4 rounds were foreseen in 1989 and 4 in 1990, covering a period of 24 months. In practice, only four rounds could be done, over a period of 18 months. Part of this can be explained by important problems of logistics; the second part of the explanation is the culmination of the political tensions between Zaire and Belgium in June 1990, resulting in a stop of all official technical co-operation and reducing the duration of the survey by 6 months.
A detailed account of the analysis and results has been provided elsewhere (Kegels, 1995). Here we will restrict the presentation of findings to a summary of the main results, divided over 4 areas of interest.
In terms of reliability of the IFAT screening results, quite a bit of random error misclassification occurs. Comparing two laboratories, they appear to disagree on the other lab's positive results in about 1 out of 3 cases, independently of the fluorescence threshold chosen (i.e. including "weakly positives" or not). Within one laboratory, in a blinded reproducibility study, this proportion is reduced to somewhere between 10 and 15%. However, comparison of the two laboratories does not reveal any systematic bias, since they arrive at the same overall serological prevalence.
When formerly or presently treated cases are eliminated from the cohort, observed sero-conversion rates between the rounds of the survey are not meaningfully different from what can be expected on the basis of the test's inherent lack of precision (at this level of transmission).
Lack of IFAT reproducibility is concentrated in a relatively small group of individuals. Changes in IFAT results (conversion or inversion) are far less meaningful in previously treated cases.
Over time, individual IFAT screening test results appear to fluctuate a lot under and over the positivity threshold, also among those who were never treated before. The majority of recent "conversions" end up negative again, without treatment.
The reliability of the declaration of symptoms in answer to open questions (in a population based study setting) is questionable. Possible explanations may be linked to the interview situation itself and to perception and motivation of the interviewees to declare health problems.
In this area and this population, the presence of "typical" neck glands, or of any type of neck glands, is not associated with serological IFAT positivity. However, the (rare) individuals who produce both signs (typical glands and IFAT positivity) are very likely to be detected as parasite carriers.
The prevalence of IFAT positivity appears to increase progressively and continuously with age, also among those who have never before been treated for trypanosomiasis. If we include "weak positivity", this serological prevalence (among those never treated before) increases gradually from 1.4% below the age of 10 years to more than 12% above the age of 50. Excluding "weak positivity", this gradient goes from 0.6% to some 5%.
The prevalence of "typical glands" (or any kind of neck glands) decreases gradually with age, from more than 25% with "typical glands" below the age of 20 years to some 5% above the age of 60.
IFAT positivity among previously treated individuals is much more frequent than among never treated individuals. The frequency of this serological positivity decreases with time elapsed since the last treatment, but this evolution is a matter of years rather than months.
Among the parasitologically confirmed patients, the distribution of cell counts in cerebrospinal fluid appears to be bimodal. Two sub-populations are separated by an empty "dip" between 40 and 80 cells/ml.
Parasitologically confirmed infection occurs at all ages, but is most frequent in the age group between 10 and 50 years in this population.
In this study setting, 15 out of 40 parasitologically positive individuals have been identified in a "passive" mode. Among these patients almost all (14/15) have cerebrospinal fluid cell counts of more than 80 cells/ml.
A positive IFAT screening test result is no conclusive evidence of patent infection.
As measured against parasitological confirmation, the IFAT screening test's sensitivity in this situation is of the order of 83%, for cases confirmed in the same round. Sensitivity increases to some 88% when parasitological positivity in the same or following rounds is accepted as definition of disease. These figures are independent of the IFAT positivity threshold used ("weakly" or "clearly" positive). Therefore, the higher positivity threshold appears to be as effective for screening purposes as the lower one, and much more efficient.
As measured against parasitological negativity in the same round, and using the higher positivity threshold for the IFAT test, the latter's specificity per round is between 97.4 and 98.6%. The highest specificity figure is reached when "absence of disease" is defined as parasitological negativity during the whole survey as well as before the survey (never treated), and would be 98.5% for the population of the first round.
The sensitivity of the sign presence of "typical neck glands", as measured against parasitologically confirmed cases of the same round, is some 41% in this situation. Sensitivity increases to 49% if calculated on the group of parasitological positives in the same or following rounds. The specificity of the sign "typical neck glands", as measured against parasitological negativity in the same round, ranges between 79.4 and 90%. If "absence of disease" is defined as parasitological negativity throughout the 4 rounds of the survey, the specificity of the sign in the first round would be 79.4%.
The odds ratio (OR) of the single sign "typical neck glands" (as measured against parasitological confirmation) is consistently very low, as well as its positive predictive value (PPV). However, when combined with IFAT positivity, this very rare combination has a very high OR as well as PPV, but unacceptably low sensitivity from the point of view of disease control effectiveness. Moreover, the majority of parasitologically confirmed infections with this combination of signs have high cell counts in the cerebrospinal fluid (the latter is true, for that matter, for all the screening signs).
The presence of the single screening sign "any kind of declared symptom of at least 7 days' duration" identifies a group in the first round population in which two thirds of its eventually parasitologically confirmed cases will be found.
If parasitological confirmation is maintained as the necessary condition for treatment, and if parasitological investigations, when negative, are continued after screening, the best combination of sensitivity and specificity (effectiveness and efficiency) is obtained with the single screening criterion "IFAT at least clearly positive".
Within an unchanging cohort population (in this epidemiological situation) the marginal effectiveness of repeated IFAT screening (at intervals of 5 to 6 months) is negligible as compared with pursued parasitological follow-up of the initial IFAT positives.