Factors determining noncompliance to mass drug administration for lymphatic filariasis elimination.

Mass Drug Administration (MDA) for Lymphatic Filariasis (LF) elimination has been implemented worldwide and in India with a goal of eliminating the disease by 2020 and 2015 respectively. Compliance to MDA is less than adequate to achieve the goal in the desired time. This study aims to identify the factors related to awareness, acceptability and attitude and the role of certain theoretical constructs of health belief model in determining the compliant behavior to MDA. Within a cross-sectional study done in Thiruvananthapuram district of Kerala, India, undertaken to determine coverage, a comparison was done between compliant and noncompliant individuals. 300 households were selected using cluster sampling technique, for estimation of coverage of MDA. From these households, 99 noncompliant and 70 compliant individuals were selected as cases and controls. The independent factors determining noncompliance were client attitude of not perceiving the need with an adjusted odds ratio (OR) of 2.52 (1.29-4.92), an unfavorable provider attitude with an adjusted OR of 2.14 (1.05-4.35) and low drug administrator acceptability with an adjusted OR of 2.01 (1.01-3.99). In MDA, the person giving the drug to the beneficiary is the most important person, whose attitude and acceptability determines compliance. More rigorous selection and training for capacity building of drug administrators are essential to enhance the compliance level. Alternate drug delivery strategies, besides house to house campaign by voluntary drug administers also needs to be implemented.

INTRODUCTION

Lymphatic Filariasis (LF) is the second most common cause of physical disability worldwide. India alone contributes to 40% of the global filariasis burden and harbors 50% of the global population at risk of infection.[1] Filariasis is a major public health problem in the state of Kerala. About 6 million people are exposed to the risk of filariasis.[2] The International task force for disease eradication has identified LF as one of the potentially eradicable diseases.[3] Based on WHO resolution in 1997, LF is targeted for elimination globally by 2020. It is based on a dual approach (Twin pillars): Mass Drug Administration (MDA) and Morbidity management.[4] The principle behind MDA is that single dose Diethylcarbamazine (DEC) administered annually is able to reduce parasitemia to the same level, at the end of one year as 12 day chemotherapy. If this is continued for 4-5 years, which is the average life span of an adult filarial worm, the transmission of filariasis can be interrupted.[5] MDA for LF elimination has been implemented worldwide and in India since 1997 with a goal of eliminating the disease by 2020 and 2015, respectively. One of the major challenges in the transmission interruption using MDA is that a very high coverage of 85% is required to achieve the interruption of transmission in 4-6 years time. Current approaches to drug delivery do not achieve this.[6-8] Compliance with medical recommendations, especially with drug therapy, has been recognized to represent a complex challenge since its first mentioning by Hippocrates about 2400 years ago. [9] In general, the term compliance describes the extent to which a person's behavior coincides with medical advice.[10] One of the oldest and most widely used model to explain compliance is the Health Belief Model (HBM), by Rosenstock. This model used the concepts of perceived threat, perceived benefits and perceived barriers to explain the compliance with preventive behaviors like checkups and immunizations. It was further developed by Becker (1974) and applied to several areas of health psychology including sick role behavior, adherence to medical regimens and health promotion behaviors.[11] A concept of self efficacy has been introduced into this model by Bandura.[12] The concept of self efficacy is also part of the social cognitive theory and protection motivation theory.[1314] One of the limitations of the health belief model is that the model does not incorporate the influence of attitude and peer influences on people's behavior regarding their health beliefs. The theory of reasoned action provides a construct that links individual beliefs, attitudes, intentions and behavior. Attitude is a person's positive or negative feelings towards performing a defined behavior.[15] All these theories and models are relevant for fear appeals and attitude change. Compliance with health services is also determined by awareness, acceptability, availability, accessibility and affordability.[16]

Despite the availability of such good models to explain preventive behaviors, studies are lacking, regarding the applicability of the concepts in these models in MDA. Most works have identified reasons for noncompliance from the client and provider perspective. This study aims to identify the factors related to awareness, acceptability and attitude and the role of theoretical constructs of health belief model in determining the compliant behavior to MDA which may be beneficial in deigning health education to alter the compliance behavior favorably.

MATERIALS AND METHODS

A population-based case control comparison study was done within a cross sectional survey in Thiruvananthapuram district of Kerala, India. Households in Thiruvananthapuram district, in the area during the period of MDA 2007 were selected for the estimation of coverage. From each household, one adult member (>18years), who had received the drug and gave consent, was randomly selected for the study of factors determining non-compliance.

Dependent variable

A Case (Noncompliant) was defined as an adult member from the households selected, who had received the drug but reported to have not consumed the drug. A Control (Compliant) was defined as an adult member from the selected households who had received the drug and reported to have consumed the drug. Self reports have been validated as a reasonably accurate means of determining compliance.[1718]

Independent variables

Awareness related variables

This was assessed as 3 variables namely awareness of filariasis, experience with filariasis and awareness of MDA. Each of these variables was scored using 4 items. For analysis the median score was used as cut off and the 3 above mentioned awareness related variables awareness were classified as poor and good.

Health belief model variables

All the variables from the health belief model were scored using a 5-point Likert scale from strongly agree to strongly disagree. Perceived threat was assessed with items on perceived susceptibility and severity. Perceived susceptibility in turn was measured as the combined score of perceived risk and severity of threat. Perceived susceptibility is one's subjective perception of risk of contracting filariasis. Perceived severity is one's opinion of how serious a condition and its consequences are. Perceived benefits are one's opinion of the benefits of MDA to the individual and the society in reducing the threat. This was measured using 2 items. Self Efficacy is the confidence in one's ability to take action. This was assessed using 3 items. Perceived barriers are the believed negative consequences of taking the drug. In MDA no expenditure is incurred by the individual, therefore side effects have been considered a barrier. If they agreed with the statement that the drug would cause side effects then they were classified as having a high perceived barrier and if they disagreed then they were classified as having a low perceived barrier.

Acceptability related variables

Acceptability of system was assessed with 2 items on the acceptability of modern medicine and government system and scored from 1 to 4 each. Acceptability of drug distributor or volunteer was assessed with 4 items on the knowledge of the person, satisfaction with the information given, information given and ability to clear doubts. Drug acceptability was assessed in terms of acceptability by sight, cover, number and size. Using the median, all the acceptability related variables were dichotomized as low and high acceptability.

Attitude related variables

Provider or drug administrator attitude was a single item, with 3 options of 3 attitudes. If the attitude was ‘must take’ then the provider attitude was considered favorable. Otherwise it was taken as unfavorable. Client attitude of need was also a single item question to determine whether the person had a felt need of the program. A statement was put forth, as since I have no disease, I need not take the drug and the respondents asked to agree or disagree on a Likert 5 point scale. Those who agreed to the statement were classified as having an unfavorable attitude. Socio demographic variables assessed, included age, sex, education, occupation, income and place of residence.

Data collection

Self administered questionnaire was developed based on an in depth understanding of factors determining utilization behavior and compliance to a preventive intervention. Items on each variable were generated using our own understanding, focus group discussion (FGD), interview with key informants and expert consensus. The questionnaire was pretested. The investigator was present in the field during data collections and verified the understandability before administration and checked the completeness of the filled questionnaire before leaving the site. Sample size was determined for the cross sectional study on coverage.[8] Power calculations were done to establish the reliability of the study findings on factors determining compliance which are given in Table 1.

Table 1

Comparison of baseline socio demographic variables in noncompliant and compliant group

Ethics statement

The protocol was presented before the faculty of Clinical Epidemiology Research and Training Centre (CERTC) for approval. Then clearance was obtained from the institutional research and ethical committees (C26/ECO3/08/MCT dated 17-04-2008). Permission was obtained from the Director of Health Services (DHS) (No. PHI-29162/08/DHS dated 17-04-2008). Informed written consent from participants was also taken. No photographs or personal details of participants are published in the study.

Data entry and analysis

SPSS 11 was used for data entry and analysis. Initially Chi square test was done to identify the significant variables. A stratified analysis was done and the stratum specific OR and Mantel Haenszel OR were found to look for effect of confounding and interaction. All the variables which were significant in bivariable analysis were put into logistic regression for multivariable analysis. Then the factors which were not significant were removed in order of their significance, the least significant one being removed first.

RESULTS

There were 1232 people in the surveyed households of which 1185 were eligible for receiving the drug in MDA. There were 42 children under 2 years and 5 pregnant women who were not eligible for drug receipt. 625 individuals received the drug. Among them 247 individuals reported to have consumed the drug. There were 298 eligible cases and 196 controls. 99 cases and 70 controls were selected for study, including 1 individual, giving consent randomly from each house.

Factors determining noncompliance

The socio-demographic variables namely age, sex, education, occupation and income were not significantly different in the compliant and non-compliant group and hence controls and cases were comparable, as shown in Table 2. There was no significant association of compliance with awareness of filariasis, experience with filariasis or MDA awareness [Table 3]. Health workers were the most common source of knowledge regarding MDA (48.5%), followed by Television (20.7%) and newspapers (10.1%). Perceived benefits and self efficacy were significantly associated with compliance. Low perceived threat and perceived barrier in the form of side effects was not a significant factor determining noncompliance [Table 3]. Non-compliant status was significantly associated with the low acceptability of the system, low drug administrator acceptability and low drug acceptability [Table 4]. 42% of the drug administrators were of the attitude that the drugs must be taken. This was considered a favorable attitude. This provider attitude was significantly associated with compliant. The client attitude towards the necessity of drug in spite of not having the disease was found to be significantly associated with compliance [Table 5].

Table 2

Compliance status in relation to the awareness and health belief model variables

Table 3

Compliance status in relation to the acceptability related factors

Table 4

Compliance status in relation to attitude related variables

Table 5

Summary table of bivariate and multivariate analysis

All the variables which were significant in bivariableanalysis, were put into logistic regression for multivariable analysis. A stratified analysis was done to look for confounding and interaction. It was found that the effect of low drug administrator acceptability had been confounded, by the effect of perceived benefits. The stratum specific OR were 2.35 and 2.29, which were similar and lower than the crude OR of 2.6 (1.39-4.89). The Mantel-Haenszel OR was 2.33 (1.21-4.44). There was an interaction of self efficacy in the relation between client attitude of no disease and noncompliance. The stratum specific OR were 1.07 and 7.2. The crude OR was 2.67 (1.42-5.014) [Table 6].

Table 6

Final model on factors determining noncompliance

After multivariable analysis, three factors namely an unfavorable provider attitude, an unfavorable attitude towards the need of the program and low drug administrator acceptability were found to be significant independent factors determining noncompliance to the drug [Table 7].

Table 7

Power calculation

DISCUSSION

In this study attitude related variables were found to be most influential on compliant status. The attitude and norm variables often exert different degrees of influence over a person's intention. In one study, attitudinal beliefs exerted greater influences,[19] whereas in another peer norms appeared to be the best predictor.[20] Among the many reasons, for the unfavorable attitude towards the need of the program, the first could be that people are not seeing the morbidity of the disease in their immediate surroundings. The general prevalence of filariasis much less than in the past and the more florid manifestations such as elephantiasis and massive hydroceles are rarely seen. This means that filariasis not a major concern in the in the minds of the public any more. In this situation a health education focusing on the long term benefits to the society through eliminating the disease will be helpful in raising the perceived benefits. Similarly it has to be widely publicized that a huge burden of disease lies hidden in the community and the florid manifestations are only the tip of the iceberg. Secondly, people attribute the use of drugs only to treatment and not for prophylaxis. The role of immunization for prevention is well understood since it is being implemented for a long time. In spite of low prevalence and visible morbidity of vaccine preventable diseases, people don’t say “We have no disease, so immunization is not needed”. Immunization has come to stay in the minds of people as a preventive measure, but people have not conceived the idea of drugs for interrupting transmission and disease elimination. Health education should therefore focus on the importance of the disease and the need of the program to save the coming generation from this dreaded disease. Among the health belief model variables, the most significant in this study were the perceived benefits and self efficacy. Self efficacy as a concept of Health Belief Model (HBM) has been less studied. In a literature review of HBM studies from 1974-1984, it was identified that perceived barriers was the most influential variable for predicting and explaining health behaviors. Other significant dimensions were perceived benefits and perceived susceptibility, with perceived severity identified as least influential.[21] Behaviors like taking pain medication are likely to be perceived as very effective, since they have a noticeable immediate effect. Another research based on health belief model yielded mixed results.[22] The benefits of the program are not being perceived rightly. This has been reported as a factor for noncompliance in another study in the same state[23] and the country.[24] An individual who feels the need of taking the drug would perhaps find the time and overcome his difficulties with health condition and easily take the drug. This could be the reason for the interaction of self efficacy with unfavorable client attitude. Low acceptability of system, Low Acceptability of drug administrator, Low Acceptability of drug and Unfavorable Provider attitude were factors determining noncompliance from the perspective of provider. Of these, all the acceptability related factors perhaps reflected in the low drug administrator acceptability, since he/she was the final channel through which the program reached the community. The effect of low drug administrator acceptability was found to have been confounded by perceived benefits probably because the benefits of the program were perceived partly in relation to the explanations by the drug administrator. The final interplay is between client and drug administrator in determining compliance. The outcome depends on how competent each drug administrator is, which is influenced by several others as shown in the model in Figure 1. If real need has to be converted to felt need, the drug administrator/volunteer has to be equipped to make the people realize the real need, for which first the providers of the program have to realize this need, which in turn requires training. The role of organizational factors have also been demonstrated elsewhere.[7]

Figure 1

Model to explain the compliance behavior. The final interplay is between client and drug administrator

Since this study was a population based study, cases and controls, were taken from the same setting and base population. This ensured their comparability. In this study sample size was not fixed a priori. Therefore the power for detecting a significance was calculated for each of the 12 variables, using epitable of epiinfo 6. Table 7 gives the calculation of power for detecting significance.

MDA program is different from other preventive programs targeting health behaviors in many ways. MDA is being delivered at the door steps; a program operating in a campaign mode, the intervention is a drug and involving a huge target population (many times more than that targeted in a country's election). Hence models should be repeatedly validated in such programs as they could work differently in various setting. This forms a scope for future research so that these concepts can be used for creating appropriate messages. More rigorous selection and training for capacity building of drug administrators are essential to enhance the compliance level. Alternate drug delivery strategies, besides house to house campaign by voluntary drug administers, also needs to be implemented to achieve the goal of elimination.