| dc.description.abstract | Brucellosis is a common disease among pastoralists, who live in close
association with their animals. A retrospective study of medical records was
undertaken to investigate the extent of brucellosis in causing ill-health among the
pastoralists of Narok District if' Kenya. Morbidity data for the past seven years (1986-
92) covering over 1 million cases and detailed case records involving 2077 patients
over the past two years (1991-92), from reporting and testing health units in the
district were evaluated. Two main objectives were investigated; first, morbidity data
was used to describe the occurrence, seasonal pattern and age distribution of human
brucellosis and other diseases presenting with "flu-like" symptoms in Narok and
second, to use data from detailed case records to investigate associations between
diagnosis of brucellosis and malaria and potential clinical predictors. All brucellosis
diagnosis was based on a positive Rose-Bengal (RB) test but most malaria diagnosis
was based on clinical findings only.
Diseases with flu-like symptoms constituted the majority (52 %) of reported
cases. Of these, malaria was the most commonly diagnosed (79%). Brucellosis
accounted for 0.8%, pyrexia of unknown origin (PUO) 2.4% and rheumatism 7.1 %.
However, only a small fraction (4/60) of clinics diagnosed any brucellosis cases. If
only clinics regularly testing for brucellosis (Rose-Bengal test) were considered, the
proportional morbidity of brucellosis among the cases with flu-like symptoms increased
to 13.7%, while malaria, rheumatism and pua accounted for 69.3%, 16.1 % and 1%
respectively. In my opinion, the higher proportional morbidity of brucellosis in testing
dispensaries is a better estimate. Although testing dispensaries might be considered as
XI
"referral" centres, each time a new dispensary begins testing for brucellosis a large
number of "new" cases are uncovered without decreasing rases at dispensaries already
testing for brucellosis. Considering all attendances reported at the four testing
dispensaries, brucellosis accounted for 5.5 % of all illnesses, rheumatism 6.4 %, PUO
0.4% and malaria 27.7%, against 0.4%,3.7%, 1.3% and 41.1 % considering all health
facilities in Narok District. It appears that brucellosis is grossly under-reported in the
district due to lack of testing (diagnosis) by most health facilities.
Brucellosis and malaria were responsible for 21.2 % and 55 % of patients with
flu-like symptoms in the detailed study of records. For brucellosis, clinical diagnosis
was not relied on but was always supported by laboratory tests. In fact, patients
visiting health facilities with flu-like symptoms in Narok were invariably considered to
have malaria on the initial visit. Brucellosis was only suspected after malaria therapy
failed. This diagnostic pattern created the impression that brucellosis was mainly
associated with a long duration of illness; however, in logistic regression models of
clinical signs among patients tested for brucellosis, patients positive to the RB test had
shorter duration of illness than negative patients (p = 0.003). Statistically, in patients
tested for brucellosis, a positive RB test was significantly associated with joint pain
(OR = 4.3; P = 0.009), headache (OR = 19.8; P = 0.004), duration class (p =
0.003), and interactions between joint pain-headache (OR = 0.05; p = 0.004) and
lameness-headache (OR = 8.38; p = 0.074). The stepwise logistic regression model
with these clinical signs correctly predicted the RB test result 62.3 % of the time with a
sensitivity (Se) of 66.6% and specificity (Sp) of 52.2 % if a 0.290 outpoint was used.
Malaria was more common, easier to diagnose clinically and affected younger
people than brucellosis. For patients subjected to blood smear examination,
XII
identification of malarial parasites was statisticaily associated with age class (p =
0.041), headache (OR = 2.2; P = 0.070), joint pain (OR = 7.7; P < 0.0001) and
interactions between emesis and pale mucous membranes (OR = 12.0; P = 0.058),
pale mucous membranes and headache (OR = 0.02; p = 0.002) and headache and
joint pain (OR = 0.315; p = 0.018). The stepwise logistic regression model correctly
predicted the blood smear test result 67.2% of the time with a Se of 62.1 % and Sp of
77.4% if a 0.350 cutpoint was used.
For both diseases, the value of routine laboratory testing or standard clinical
symptoms in differential diagnosis of these and other flu-like diseases could not be
established. The patients tested by either the RB test for brucellosis or the blood smear
examination for malaria were likely unrepresentative of all potential patients. Given
the high levels of brucellosis uncovered, further prospective studies in both human and
animal populations are currently underway. For humans, clinical and laboratory
diagnosis will be evaluated in all patients presented with flu-like symptoms. Most
importantly the high rate of human brucellosis in Narok is due to brucellosis in the
cattle, sheep and goat reservoirs. To effectively protect humans, a study to better
estimate the incidence and economic effects of brucellosis in these species is being
undertaken. Finally, it was observed that many patients in Narok considered
brucellosis treatment too costly, too long and too painful. Because of poor acceptance
of current treatments, clinical trials to identify a less costly, shorter, and better
accepted treatment regimen should be considered. | en |
