Patient Characteristics Associated with Non-Adherence to Tuberculosis Treatment: A Systematic Review

Background: A high level of adherence to treatment is essential for cure and prevention of tuberculosis (TB) treatment resistance. Methods: A Systematic review of 53 studies addressing the patient characteristics associated with TB medication non-adherence was performed. The publications were identified by searching the PubMed, World Health Organization (WHO), and Centers for Disease Control and Prevention (CDC) database, EmBase, Scopus database Arts, Humanities, Social Science database and Google scholar. Only English language publications were eligible. Potentially eligible studies were retrieved and the full articles were assessed. The potentially eligible studies were included if they concerned patients treated for tuberculosis, reported non adherence and reported on potential risk factors associated with non-adherence. Results: Factors that were most frequently consistently and statistically significantly related to non-adherence to tuberculosis treatment were: family income, patient movement and changing address or giving wrong address, tuberculosis relapse or multidrug-resistant TB (MDRTB), during intensive phase of treatment, history of default, treatment regimen (long course), response to treatment, homeless, stigma, seeking traditional healers, staff receptiveness, lack of directly observed therapy short course (DOTS), poor knowledge or lack of health education, side effects of drugs, feeling better, alcohol intake and lack of family and social support. Conclusions: Non-adherence to tuberculosis treatment was influenced by several factors.

It is estimated that approximately one-third of the world's population (approximately two billion people) is infected with tuberculosis bacillus [5] [6] [7]. In 2015, globally 10.4 million new cases of TB and 1.8 million deaths from TB occurred, of which 1.4 million among HIV-negative people and 0.4 million deaths were in HIV-positive people and thus officially classified as HIV deaths in the International Statistical Classification of Diseases [8] [9] [10]. Approximately 80% of TB cases are found in 22 countries; the highest incidence rates being found in Africa and South-East Asia [11].
TB has a high morbidity and mortality rate despite its status as a treatable disease [12] [13]. A high level of adherence to treatment is essential for cure and to avoid development of resistance [14] [15]. Thus, completion of antituberculosis treatment is the foremost priority of tuberculosis (TB) control programs. Treatment that is taken irregular, interrupted for two months or longer or is incomplete increases the risk of treatment failure, relapse of disease, acquisition of drug-resistant TB, death, and prolonged infectiousness [16]- [21]. In turn, treatment failure and relapse can increase transmission of TB. Poor adherence to treatment is common despite various interventions aimed at improving treatment completion [22]. Non-adherence to TB medication is a major barrier to its local and global control and worsen the treatment outcome. In addition, defaulting increases the risk of drug resistance, relapse and death, and may prolong infectiousness. The objective of this systematic review was to identify the patient characteristics that are consistently associated with TB medication non-adherence. This knowledge could be helpful for health care providers and health policy makers to improve treatment adherence. The patient characteristics that we studied were: socio-demographic factors; disease related factors; treatment and services related factors; behavioral factors and social factors.

Methods
This systematic review was conducted to answer the question: which patient characteristics are consistently and statistically significantly associated with TB treatment non-adherence. The study reviewed publications found on risk factors associated with defaulting tuberculosis treatment that were published in English between 1990 to 2017. The publications were identified by searching the PubMed database, WHO database, and CDC database, EmBase, Scopus database, Arts, humanities, Social Science database and google scholar. Using the key words "tuberculosis", "mycobacterium tuberculosis", "adherence", "non adherence", "defaulting", "risk factors", "compliance", "determinant factors", "outcome of tuberculosis treatment", "predictors", "leading", "impact", "noncompliance", "Anti tuberculosis treatment", and "anti tuberculosis therapy". Because of resource limitations, papers published in other languages were not considered. Additional reports were identified by manually reviewing the references of the studies found. [67]- [76].
When possible we extracted odds ratios (ORs), relative risks (RRs), and their 95% confidence interval. Otherwise the direction of the association and statistical significance was used. The relationship was considered consistent if the variable was reported in 3 studies or more to be statistically significantly related with non-adherence to tuberculosis treatment and more often statistically significantly related to non-adherence than none related. Ethics approval was not required for this systematic review.

Results
The database searching resulted in a total of 2131 identified citations. About 981 records remained after removal of duplication. Out of these 894 were excluded because they did not focus on the TB treatment and risk factors. Of the 87 abstracts that were potentially eligible, 34 were excluded after the review of the full paper or abstract (see Figure 1). Hence 53 eligible articles were finally included.  2.14 to 3.23) and feeling better (OR: 5.28 to 21.0).
Factors frequently but not consistently related to tuberculosis treatment nonadherence were: male, residential locality, distance to tuberculosis treatment unit, HIV status, IVDU, prison, and smoking.
All studies reported the default rates. The reported default rates ranged between 3% and 55.7%. These default rates varied substantially among the studies as either cumulative or incremental percentages of all patients. Also it varied according to regimen of treatment, age, sex, ethnicity, and residential locality.

Discussion
The results of this systematic review show that family income, moving of patient or giving wrong address, tuberculosis relapse or MDR TB, intensive phase of treatment, history of default, long course treatment regimen, response to treatment, homeless, stigma, seeking traditional healers, staff receptiveness, DOTS, poor knowledge or lack of health education were consistently and statistically significantly related to tuberculosis treatment non-adherence. Surprisingly some frequently cited factors which were traditionally thought to be related to tuberculosis treatment non-adherence appeared to be not consistently associated with non-adherence, these include age, sex, marital status, occupation, employment status, level of education. The most important methodological aspect of this systematic review that requires explanation is the use of a qualitative approach, i.e. counting statistical results of the patient characteristics related to non-adherence to tuberculosis treatment. Ideally, numerical results of individual studies are combined. However, when reviewing eligible studies it was discovered that in most of these primary studies only numerators were given if a factor was statistically significant related to non-adherence. Hence, only including the results of a characteristic if full numerical data was given had the potential to lead to an extreme bias. Since studies dating back to 1990 were included we assumed that it was unlikely to get full responses of all authors. Hence, we resorted to, admittedly a less optimal qualitative approach. Some other methodological aspects of this systematic review require attention: First of all, we attempted to identify all relevant articles published up to date of this review and found a large number of relevant references. However, due to language restriction and lack of resources we only included articles in the English language. Secondly, data extraction was hampered by unclear methodology and definitions in many relevant studies. In addition, many relevant factors for non-adherence were not taken into account in many studies. Thirdly, the majority of articles included in this study were conducted in developing countries; the findings are therefore most applicable to countries with low resources which carry the greatest burden of TB disease and where urgent interventions are needed to improve adherence to tuberculosis treatment. However, our findings may also be applicable to more developed countries. Fourthly, our review took many types of risk factors (social and biomedical) for non-adherence to Tb treatment into account while many individual articles solely have focused on the factors related to health services provision.
Overall we trust that the factors identified as consistently related to TB treatment non-adherence are of truly associated with non-adherence. For the potential risk factors for non-adherence that were not consistently associated a numerical approach in large data set of new studies might still show an association. However, it is likely that these associations are of little importance.
Exploration of the factors associated with non-adherence to tuberculosis treatment deserves some attention. Some of these factors identified can be changed and The doctors and health worker staff receptiveness and their effort in counseling and health education provision increases awareness of patients and their families and the whole community about Tuberculosis. The social support positively affects the patients' knowledge to counteract the feeling of guilt and shame (stigma). This will be reflected on patients' compliance with tuberculosis medication instead of seeking traditional medicine. Patients' compliance assists the tuberculosis control program to achieve high cure rates and decrease the MDR prevalence in the community.
This synthesis suggests that governmental and nongovernmental organizations need to give more attention to support poor patients financially and offer accommodation for the homeless. Another cost effective suggestion is to consider even distribution of health services to improve its availability, accessibility and affordability to all tuberculosis patients. This, combined with good referral system from health unit to hospital, will result in reduction of the cost of travelling, waiting time at tuberculosis clinic, and patient movement. Besides, good counseling to patients would encourage them to give their correct address and to report any change of address to health worker staffs and therefore improve adherence to tuberculosis treatment.
The review findings are important for the policy makers, medical practitioners, health worker staff and researchers to study the patient context as a whole (social, behavioral factors, believes and knowledge) and not to ignore their experience and opinion in treatment course. This will make the health care consumer share the responsibility of the treatment process and will combat non-adherence.
Moreover, it helps in creating good relationship between the patients, health care providers and the whole community. This new approach in thinking will im- prove tuberculosis treatment adherence and will reinforce the surveillance system of tuberculosis control program at state, national and international levels.
In addition to the findings of this systematic review we believe that further studies are needed to be conducted to deeply understand different aspects of tuberculosis disease and its treatment (social and biomedical) aspects which is a complex process. These studies should include the experience of policy makers, the health personnel working in the field, nongovernmental organizations, patient support groups and the community leaders. Moreover, the patients' knowledge about tuberculosis for example, patient satisfaction with health services provision and staff receptiveness. This will substantially help the health care system planners in developing good future strategy to improve treatment adherence among tuberculosis patients.

Conclusions
Non-adherence to tuberculosis treatment is a complex dynamic process, which was influenced by multiple factors. Identifying and understanding the nature of these factors facilitates development of appropriate and effective intervention plans.
The relationship between the identifiable variables and TB treatment default is important for the knowledge of policy makers and clinicians dealing with TB patients. Despite the fact that association of these variables with increased risk of TB treatment default had been reported in many individual studies, it had not received attention in terms of planning, training, guidelines, and research. This systematic review attempts to alter the current situation with the aim to increase knowledge about treatment default and to promote adherence to TB treatment.
The results of our review can help to find out the consistently statistically related factors to TB default treatment and assist the health care system designers, health service providers including clinicians, nurses, counselors, patients, families, and the community leading to high adherence to tuberculosis treatment, reinforcing the surveillance system of tuberculosis control program hence achieving the goal of tuberculosis elimination at state, national and international levels.