Treatment-seeking for febrile illness in north-east India: an epidemiological study in the malaria endemic zone

Malaria is one of the major public health problems of many countries of world [1‐3]. According to WHO report, almost 300 million cases of malaria occur worldwide and more than one million people die due to malaria every year [4]. In India, malaria is reported from many states, but several deaths of epidemic proportion reported every year from different parts of north-east India [5‐7]. Assam, being the most populated (27.85 million) and geographically second largest State (78,523 km²) of north-east India, is contributing more than 5% of the total number of cases of malaria reported in India. Most districts of the State are malaria-endemic and many pockets in forest, forest fringes, and foothill villages along the inter-country/inter-state border are vulnerable to malaria outbreak as reported [8]. Chloroquine-resistant malaria is widespread in the State, and decreased sensitivity to other anti-malarial drugs has been documented [9]. The malaria cases are detected throughout the year [10], but higher incidence is reported during May to October [11]. There is often delay in reporting illness to health care providers and during these periods certain unorthodox practices are adopted as cost-saving measures.

Although investment in health sectors has been gradually increasing and many health programmes to control malaria, such as spraying DDT in high-risk areas, use of impregnated-bed nets and mass collection of blood slides form the community, have been launched by the government, but there is no appreciable change in malaria incidence. A clear understanding about the health-seeking behaviour of households is crucial for an effective control programme of malaria. For instance, the implementation of current strategy for the control of malaria is mostly based on early diagnosis. It is important to know the factors that determine the utilization of the services of a health care provider for suspected cases of malaria so that an effective strategy can be developed.

Utilization patterns of health service mainly depend on quality of service offered by health care providers and whether it is affordable by the individual user. Moreover, impact of quality differences on the use of health care facilities is to be assessed simultaneously with household and individual characteristics affecting the utilization. The preference of one health care provider over another will be affected not only by facility characteristics (quality of service and cost of service), but also by the individual characteristics such as preferences, affordability, convenient, education level, and household income.

The present study is an attempt to assess the various options of health care patients normally follow for seeking treatment of febrile illness or malaria and also estimate the factors associated with the choice of health services. In the endemic areas, fever is the most common symptom for suspected cases of malaria and is, therefore, being used interchangeably [12].

The general characteristics of household respondents such as age, household income, education, occupation, area of residence, health care services used, distance of nearest health center (government/private) from residence, are presented district-wise for comparison (Table 1). The mean household income was Rs. 2,534 ± 1,816 (INR) per month, but it was significantly lower in Tinsukia district. Variability in monthly household income was high as is reflected by the standard deviation. About 47% of respondents were illiterate and 10% educated up to secondary and above. A majority of the respondents were farmers (38.6%), followed by service class (22%) and self-employed (12.6%) people. Most of the respondents were Hindu (93.6%) and ethically belong to different tribal and non-tribal communities of Assam. The residential area of households was categorized as town (12%), forest (23.0%), flood (22.0%), tea garden (26.3%) and other (16.7%) plain area. The health care services attended initially for seeking treatment of febrile illness were traditional system (39.2%), government health services (29.3%), private health services (13.7%) and self-medication (17.8%) (Figure 1). Except self-medication, the use of health services differs significantly between the tow districts. The average distance to the nearest health center (Government/Private) was 4.2 ± 5.6 km, which was slightly lower for Golaghat district.

Health care services used by the households and its differences across socio-demographic characteristics of households, residential area and patient's characteristics are presented with statistical analysis (Table 2). The association of health care services was significant with all the variables, except gender of patients. The traditional system was the first choice of treatment among majority of people residence in tea garden (60%), flood (43.7%) and forest (33.8%) areas. However, self-medication was popular in forest (25.8%) and town (25.5%) areas. A decreasing pattern in the use of the traditional system was observed with increasing level of education of household heads. It was also recorded to be high among farmers (42.5%), tea tribes (57.1%) and Nepali (50.5%) communities. As an initial choice, self-medication was high among higher educational groups, i.e. secondary+ (20.7%); farmers (24.3%); Nepali (26.9%) and Assamese (20.2%) communities. It was also recorded high among higher household income groups (21.4%). The increasing pattern of self-medication was observed with increase age group of patients.

Government health service, as initial choice of seeking treatment, was high in all the residential areas (32% to 42%), except in tea garden areas (14.2%). It was also shown to be low among tea tribes of Assam (14.8%). Similarly, the private health services were commonly used in town (29.3%) and tea gardens (17.8%). Utilization of private health services was also high among tea tribes (20.3%) and higher household income group (17.7%).

Analysis of final choice of seeking treatment indicated the significant association with location, area of residence, occupation, ethnicity, household income and distance of nearest health center. However, the association of education and gender of patients was not significant. The utilization of government health services was significantly higher in Golaghat district (78.5%) compared to Tinsukia (52.5%). It was also recorded high in forest (80.6%) and flood area (78.9%). The private health services were used commonly in tea garden areas, especially by the tea tribes (Table 2).

The results of multinomial logistics regression (MLR) analysis with initial and final treatment-seeking choice are presented in Table 3. The estimated coefficients (β) of the variables in the model indicate how changes in each of the variable affect the household choice of health service providers for seeking treatment of febrile illness. Except for gender of patients, a significant effect of location, residential area, education, occupation, ethnicity, age of patients and household income was recorded with the dependent variables, i.e. health care services. Multivariate rate ratios (MRR) for traditional system as initial choice was significantly higher for residential area: tea garden (5.6), flood (3.2), forest (1.5); level of education: literate (1.7); ethnic group: Assamese (1.5) and tea tribes (3.4); and distance of health center: ≥ 5 km (1.4) compare with respective reference. It clearly indicates that the use of traditional system of medicine was much higher in tea garden, forest and flood affected area with compare to town and others. The MRR of government service as initial choice was significantly higher for education: literate (1.4); occupation: service and self-employed (2.4); others (1.9); and low household income (1.8). The household income was one of the important determinants for utilization of government services. However, there was no significant effect of residential area and ethnic background of people. The private services as initial choice shows the significantly higher value of MRR (odd) with location: Golaghat (1.5); education: literate (1.4); occupation: service and self-employed (4.1) and others (3.4); ethnic group: tea tribes (2.6); and age of patients: lower age group (1.9) compare to respective reference. However, there was no significant effect of residential area, household income and gender of patients on utilization of private health services.

As a final choice of seeking treatment, the MLR analysis shows the significant effect of location, education, occupation, ethnicity, household income and distance of nearest health center with government health services utilization with reference to private services. The estimated values of β and MRR shows the significant negative or decreasing effect with location: Tinsukia (-1.16; 0.3); education: literate (-0.47 & 0.6); occupation: service & self-employed (-1.15 & 0.32) and others (-0.93 & 0.2); and ethnic group: tea tribes (-1.59 & 0.2) with their respective reference. However, it was almost three times higher in favour of government health services for lower income (≤ 2000 INR) group (2.9). If the distance of health center increases (≥ 5 km), the MRR was significantly higher (1.6) in favour of government health services. The area of residence was not showing any effect with the final choice of two health services i.e. government or private.

A better understanding of health-seeking behaviour of people, especially when suffering from fever, a common symptom of malaria, is important for effective management and control of malaria. This is related with early treatment, accessibility and proper use of drugs. The most common treatment options reported by the households for seeking treatment of fever was use of traditional drugs, self-medication, government and private health services, which was also reported by others in many studies [16‐19].

The use of traditional medicines was prevalent in remote areas especially in tea gardens, forest and flood affected areas. It seems that internal factors of the specific geographical region like availability and accessibility of health services; traditional belief and practices was motivating the people for seeking medical assistance from traditional system (Vaidya). Although most of the people received treatment finally from government or private health services, but majority of them initially visited to traditional Vaidya for seeking treatment of febrile illness. The strong belief in traditional system was reflected even among literate people. A study in Tanzania also reported that traditional healers are an important factor of delay for malaria treatment [18].

Self-medication and consultations at medical stores was a common practice in town and forest areas. Such practices created a situation where people wait till the patient become serious enough before reporting to health centers. Obviously, some of these patients develop complications and even die by the time they reach a hospital. No doubt, the 'wait-and-watch' attitude is not only harmful to the patients, but also contributing to increased malaria. These practices are very common not only among un-educated and poor people, but across most of the variables included in the analysis. This was also reported and discussed by others [20‐22].

Over all, response of availing government health services was high by the people residing in forest and flood affected area. Moreover, the people from town and tea garden areas were seeking treatment mainly from private health services, because private health services were mainly confined in the town and private tea gardens located outside the town nearby the tea gardens. Despite longer distances and poor connectivity to motorable roads, the accessibility to government health services was recorded high among the people living in forest and flood areas. The increasing level of education of household head has decreasing effect with accessibility of government services. The MRR was also significantly low for literates with reference to illiterates. Household income is one of the important determinants for health service utilization. Similarly, increasing level of household income has decreasing effect with accessibility of government services. It was possibly due to the expensive treatment in private health centers. The socio-economic impact on choice of health services was also discussed by others [23‐26].

Early diagnosis of febrile illness or malaria can yield rich dividends for effective management and control of malaria [18, 27‐29]. Though the distance to a health center is one of the important factors of delay, its effect may be confounded with the place of residence. Residential areas such as forest, flood and tea garden, show the highest preference for traditional system and self-medication, which could be the reason of delay in proper diagnosis. Interestingly, there is no gender bias in the choice of seeking treatment, which is often reported as an important indicator to address this issue [29, 30]. However, the decision-making power on treatment-seeking behaviour mainly depends on male family elders [30].

Although treatment expenditure is one of the important components, which influence the household choice of health service providers [21], information could not be collected in this study due to recall bias and lack of records during survey. Moreover, household income is recorded as one of the important determinants influencing the utilization of health services. The identification of various other indicators in this study provides useful information to understand the interlinked behavioural practices related to malaria. As early diagnosis is the main theme of prevention and control of malaria, the effective strategies to provide health services at nearest periphery of village/community and parallel health awareness programme with people participation are needed to circumvent the situation.