Analysis of asymptomatic and clinical malaria in urban and suburban settings of southwestern Ethiopia in the context of sustaining malaria control and approaching elimination

The study was conducted in Jimma town (7°41′ N, 36°50′ E, 1710–1800 m above sea level) in a highland low-transmission epidemic-prone area in Oromia Region, southwestern Ethiopia (Fig. 1). The population of Jimma town was about 176,500 in 2015. Jimma town has a warm climate with a mean annual temperature of 23 °C, a mean annual maximum temperature of 27 °C, and a mean annual minimum temperature of 19 °C. The mean annual precipitation is about 1500 mm, with one rainy season from May to September. December and January are the driest months.

PCD was conducted prospectively in all public health centres and hospitals in Jimma town including Becho Bore Health Centre, Jimma Health Centre, Jimma Higher Two Health Centre, Mendera Kochi Health Centre, and Shenen Gibe Hospital, except Jimma University Special Hospital (Fig. 1). Jimma University Special Hospital usually accepts referred severe malaria cases, to avoid double count, malaria cases from this hospital were not included in data analysis. These are all the government-run malaria treatment centres and hospitals. A clinical malaria case is defined as an individual with malaria-related symptoms (fever, i.e., axillary temperature ≥37.5 °C, chills, severe malaise, headache or vomiting) at the time of examination or 1–2 days prior to the examination, with a Plasmodium positive blood smear [11, 12]. All febrile patients with malaria-like symptoms attending these health facilities were enrolled into this study upon the signing of consent or assent forms for minors. Standard case report forms were used to collect the following information from each patient: village of residency, demographic characteristics, occupation, education level, clinical symptoms, blood hemoglobin level, history of malaria in the preceding 30 days, history of travel in the 2 weeks preceding the clinic visit, history of fever, treatment drug prescribed, and ownership and use of malaria prevention measures. If ITN was used, the types and usage (used or not during the preceding night) of ITN were recorded. Thick and thin blood smears were prepared for each patient. Slides were stained with 10 % Giemsa and examined microscopically by experienced laboratory technologists at the Medical Parasitology Laboratory of Jimma University. Species were identified and parasites were counted against 200 leukocytes. All positive slides and 10 % of negative slides were reexamined blindly by another senior laboratory technologist. PCD was conducted from July 2014 to June 2015 (Fig. 2).

Cross-sectional parasite prevalence surveys were conducted three times in October 2014 (high-transmission season), March 2015 (end of high-transmission season), and August 2015 (end of low-transmission season), in eight kebeles (the smallest administrative unit in Ethiopia, equivalent to a district in a city) representing both urban and suburban settings (Figs. 1, 2). Households were randomly selected from each kebele to maximize coverage so that the surveyed population would represent the eight kebeles. To avoid sampling bias, one volunteer was arbitrarily selected from each selected household for participation of this study. Peripheral blood was collected by the standard finger-prick method using disposable lancets in October 2014, March and August 2015, respectively. Thick and thin smears were prepared. Slides were stained and examined at the Jimma University Specialized Hospital laboratory according to standard World Health Organization procedures. Smears were stained with Giemsa solution, and examined microscopically at a magnification of 100× by experienced laboratory technologists. Parasitaemia was determined, and quality control for slide reading was implemented as described above. Each study participant found positive for malaria was referred to a nearby health facility for treatment, according to the national malaria diagnosis and treatment guidelines.

A subset of the blood samples was subjected to further PCR analysis to confirm parasite infection and to identify the parasite species [13‐15]. The initially plan was to do PCR on all blood samples, due to the zero prevalence rate detected by microscopy, PCR was only done on randomly selected samples after the second round survey.

Sample size was estimated based on the binomial model. Because the prevalence of malaria was low in the area, an estimate of 10 % prevalence was used for the peak-transmission season and 5 % for the low-transmission season in sample size determination [13, 16]. The resulting sample sizes were 559 persons for the peak-transmission season and 386 for the low-transmission season for prevalence estimation at precision level of 5 %. For the PCR method, assuming ≤5 % prevalence based on microscopic results, with a Type I error of 5 % and population (blood samples) size of 850, 218 samples could detect a marginal error of 2.5 %. To allow for 10 % PCR failure rate, 240 samples is enough.

The demographic survey was conducted in the same eight kebeles as the parasitological surveys (Figs. 1, 2). Households were randomly selected from each kebele to maximize coverage so that the surveyed population would represent the eight kebeles. All individuals living in the selected households were included in the study. For all consenting individuals, demographic characteristics (age, sex, occupation, and education level) and information on preventive measures (ITN, IRS, repellent, and others) were collected through questionnaires. The survey was conducted from May to June 2014. Information collected on ITNs included ITN ownership, type, and usage. The total population at each kebeles in Jimma was obtained from the Central Statistical Agency of Ethiopia based on 2015 demographic projections. Household ITN ownership was calculated as percentage of households that owns at least one ITN based on demographic survey.

The density of parasitaemia was expressed as the number of asexual Plasmodium per microlitre of blood, assuming a leukocyte count of 8000 per microlitre. Annual malaria incidence rate was calculated as the number of cases per 1000 population at risk. Parasite prevalence rate was calculated as the number of infected individuals per 100 people. The incidence rate was stratified into three age groups: age < 5, 5 ≤ age < 15, and ≥15 years. Differences in sex-, age-, and occupation-specific incidence rates were analysed using χ² test and odds ratios (ORs) were calculated for different categories separately against their respective reference group. Risk factors of clinical malaria were analysed using multivariate nominal logistic regression analysis. Risk factors included age, sex, occupation, history of malaria in the preceding 30 days, history of travel in the preceding 2 weeks, and preventive measure usage. Risk factor analysis was done separately for P. falciparum and P. vivax. SPSS (IBM Corporation, Armonk, New York, USA) was used for data analysis.

Ethical clearance was obtained from the Institutional Review Boards (IRBs) of Jimma University, Ethiopia, and the University of California at Irvine, USA. Written consent (assent for children under 18 years) was obtained from heads of households and study participants. All cases with a history of fever in the preceding 3 days, as well as those with fever on examination and a positive test for malaria parasite in the blood film examination, were offered anti-malarial treatment by physicians from Jimma University as per national malaria diagnosis and treatment guidelines. Infants younger than 6 months and individuals who were unwilling to participate were excluded from the study.

A total of 1434 suspected cases were examined and clinical malaria was confirmed in 428 cases. Among the confirmed cases, 327 (76.4 %) were P. vivax, 97 (22.7 %) were P. falciparum, and four (0.9 %) were mixed infections of P. vivax and P. falciparum. No Plasmodium malariae or Plasmodium ovale was detected. Demographic information was obtained from 10,119 individuals comprising 5278 (52.2 %) females and 4841 (47.8 %) males, with an average age of 20.7 years (Additional file 1). Among the confirmed malaria cases, 299 resided within the eight villages where demographic information was collected and available. Among these cases, 224 (74.9 %) were P. vivax, 71 (23.7 %) were P. falciparum, and 4 (1.3 %) were mixed infections of P. vivax and P. falciparum. The annual malaria incidence rate was 1.7 cases per 1000 people at risk (Table 1).

Total number of case report forms and confirmed malaria cases decreased monotonously from July 2014 to June 2015 (Fig. 3). Plasmodium falciparum malaria was seasonal and was mainly observed from July to October 2014, but the seasonality in clinical P. vivax malaria was not obvious (Fig. 3). While males had significantly higher P. falciparum malaria incidence rates (OR = 1.68, 95 % CI [1.04, 2.71], P < 0.05) (Table 1) than females, there was no gender difference in P. vivax malaria. Incidence rates did not vary by age for either malaria species (Table 1). There was a significant difference in P. vivax incidence rates among different occupation groups (Table 1). Trader and travellers had the lowest P. vivax incidence rate (0.53 cases per 1000 people year), whereas office workers and teachers had the highest incidence rate compared to the other groups (3.92 cases per 1000 people year; OR = 3.39, 95 % CI [2.08, 5.50]). Although incidence rates of P. falciparum malaria varied substantially among occupation groups, the variation was not statistically significant due to small number of falciparum cases detected (Table 1).

Five categories of malaria preventive measures were included in the questionnaire: bed nets, indoor residual spraying (IRS), repellent, others, and none. The household ITN ownership rate (percentage of households that owns at least one ITN) was 47.3 % (1173/2479) and ITN usage rate (i.e., proportion of individuals who reported to use the net the night before the survey) was 37.5 %. All ITNs were LLINs, and no repellent use was found in the questionnaire survey. The use of LLIN was not significantly associated with clinical malaria risk (Table 2). Travel during the preceding 14 days did not affect P. vivax malaria incidence, but it increased the likelihood of P. falciparum malaria by fourfolds (OR = 4.14, 95 % CI [2.11, 8.14], P < 0.001) (Table 2). On the other hand, having malaria during the preceding 30 days increased the likelihood of P. vivax malaria by nearly threefolds (OR = 2.63, 95 % CI [1.51, 4.57], P < 0.001) (Table 2).

Multiple logistic regression analysis indicated that male had marginally higher risk than females, and that people who travelled during the preceding 14 days had a significantly higher risk of having P. falciparum malaria after adjusting for other risk factors (Additional file 2). For P. vivax malaria, being an office worker or teacher, or had malaria during the preceding 30 days were the significant risk factors (Additional file 3).

In the cross-sectional survey low parasite prevalence was found consistently among the three sampling time points (Table 3). Overall, parasite prevalence was 0.8 % by microscope and 2.4 % by PCR. Microscopically identified parasite prevalence was about 2.7 % (0.8 % P. falciparum and 1.9 % P. vivax) during the high-transmission season in October 2014. Unfortunately, PCR was not done for this survey due to the loss of blood samples during sample transportation. Microscopy did not identify any infection for the surveys conducted in March and August 2015, but PCR analysis detected a prevalence of 2.6 and 1.7 %, respectively (Table 3). Overall, there were 23 (56.1 %) P. vivax positive samples and 18 (43.9 %) P. falciparum positive samples in the three surveys. No P. malariae, P. ovale, or mixed infections were identified by either microscopy or PCR.