Effect of Potentially Inappropriate Medication on Treatment Adherence in Elderly with Chronic Illness

Published by Oriental Scientific Publishing Company © 2018 This is an Open Access article licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (https://creativecommons.org/licenses/by-nc-sa/4.0/ ), which permits unrestricted Non Commercial use, distribution and reproduction in any medium, provided the original work is properly cited. Effect of Potentially Inappropriate Medication on Treatment Adherence in Elderly with Chronic Illness

Globally, the elderly population is expected to reach two billion and over 323 million in India by 2050 1 . The disease process in elderly is patho-physiologically complex predisposing them to recurrent, chronic illness and multiple co-morbidities, exposing them to poly-pharmacy (prescribed number of medications), increased risk of drug related problems including ADRs (adverse drug reactions) and non-adherence to treatment 2 . Non-adherence has been associated with poor clinical outcomes, an increased cost of healthcare, a lower quality of life, and a higher rate of mortality 3 . Various studies have hypothesized, that about 200 factors that influence adherence, some of them are intentional (rational decision-making, secondary effects) and some non-intentional (treatment complexity, inappropriate prescribing) 4 . In general, only 50% of general population has been estimated to adhere to their medications, and this may range from 47 to 100% in elderly 5 . Therefore, medication-using behaviour is extremely complex in elderly and it requires a multifactorial strategy to improve adherence. Identification of avoidable risk factors that reduce adherence and institution of appropriate interventions is essential for improving medication-taking behaviour and economic outcomes.
Use of potentially inappropriate medications (PIM) in elderly has received substantial research attention in recent years, with researches reporting high prevalence of PIM in developed as well as developing countries 6 . Various studies done in India have reported the prevalence of PIMs in elderly to be in the range of 20 to 25% 7,8 . Prescribing PIM is associated with increased risk of ADRs, morbidity, mortality and healthcare mis-utilization. Studies have reported an 83% increased risk of ADRs in elderly patients receiving a PIM 9 . Since, ADR is a recognized determinant of adherence; it is interesting to study how PIM affects adherence in elderly. Though, numerous studies have been performed to assess the prevalence and predictors of poor compliance, few studies have been done in the Indian population and elsewhere to assess the effect of PIM on medication adherence in elderly. Hence, the present study was taken up with the primary objective of exploring the influence of potentially inappropriate medication on adherence to drug treatment in elderly patients with chronic illnesses.
Researchers interested in assessing the adherence, often have used the eight item Morisky Medication Adherence Scale (MMAS-8), which measures adherence on a 3 level ordinal scale (low adherence d" 6, medium adherence = 6-8 and high adherence = 8) 10 . However, in most of the studies, the outcome variable, in spite of having three ordinal categories, have been considered as dichotomous (adherent vs. non-adherent) and binary/multivariate logistic regression model has been applied resulting in loss of data as well as a substantial loss of statistical power due to the sinking of some groups of the outcome variable 11,12 . The uniqueness of the present study, is to study the effects of independent variables on all the levels of ordered categorical dependent variable, so, special multivariate analysis for ordinal data, like ordinal logistic regression model was used as an alternative and Polytomous Universal Model (PLUM) or the SPSS Ordinal Regression Procedure which is an extension of the general linear model to ordinal categorical data, was used to assess the association between the ordinal outcome (i.e., different levels of adherence to medication) and the independent variables 13 . AGS Beers criteria 2015, the most widely preferred tool with a high reliability and reproducibility, was used to find out the PIM in elderly 14 .

Materials and Methods study design and setting
This was a hospital based cross-sectional study, carried out in M.K.C.G. Medical College and Hospital, a 1200 bedded tertiary care teaching hospital with 19 out-patients departments in Berhampur, a city in Eastern part of India from March 2017 to May 2017. The hospital caters to the tertiary health care needs of the population of Southern Odisha.

sample size calculation and sampling technique
The source population was elderly patients aged 65 years and more attending the out-patients departments (OPD). The sample size was calculated using nMaster 2.0, (Designed and Developed by Department of Biostatistics, Christian Medical College, Vellore, India). Assuming, the prevalence of adherence to be 60%, an absolute precision of 5, a desired confidence level of 95%, the sample size was calculated to be 369. Assuming a response rate of 85%, finally 425 elderly patients were included in the study. The OPDs of general medicine, surgery, ENT, ophthalmology, pulmonology, orthopaedics, endocrinology, cardiology, neurology, nephrology, psychiatry were selected for recruiting the study participants based on the assumption that any elderly patient coming to the hospital with one or more chronic illness shall have at least one disease condition that could be treated in the selected OPDs. Convenience sampling technique was adapted to include the participants and the study participants were identified from the different OPDs after reviewing their medical reports and history to confirm that the patient has at least one chronic condition. inclusion and exclusion criteria Any patients, new as well as repeat, of either gender who had completed 65 years of age as on 31 th January 2017 and being treated for one or more chronic illness like hypertension, diabetes mellitus, osteoarthritis etc. for at least six months were included in the study. Patients unable to communicate, seriously ill, mentally unstable, requiring ICU admission, on palliative care or unwilling to participate and those having incomplete information were excluded from the study.

ethical consideration
The study was commenced after obtaining ethical clearance from the Institutional Ethics Committee of MKCG Medical College, Berhampur, Odisha (Approval Number 517/2017). All the study participants were explained clearly about the purpose and nature of the study in the local language (odia) or in any other language they could understand. Written informed consent was obtained before including them in the study.

data collection
All the participants were interviewed once and their prescriptions were checked for the required information. The data was collected by the investigators in a structured case record form by the investigators. The case record form captured data on various independent variables like socioeconomic and demographic characteristics of the study participants like gender, age, educational status, family income, co-morbidity, duration of disease condition, poly-pharmacy, knowledge about their illness and the prescribed medications, experiencing any side effect or not, physically active or not, presence of care givers, any type of addiction, on-going drug therapy, number of drugs used, name of the drugs, dose, frequency and duration of administration etc. The data were checked for its completeness.

study tools
Medication adherence was measured using the Morisky Medication Adherence Scale (MMAS-8). It is a self-report questionnaire, with eight questions, simple to administer, reliable and economical tool. The MMAS-8 was initially designed to identify the barriers and behaviours associated with adher ence to chronic medication. The eight questions of MMAS-8 were translated into the local language and the translated version was tested in a group of 20 patients to check for understanding of the questions in accordance with its original meaning. The questions were understood identically by all, and subsequent alterations were not considered necessary. The degree of adherence was determined according to the score range from 0 to 8, calculated from the sum of all the correct answers: high adherence (eight points), Medium adherence (6 to < 8 points) and low adherence (< 6 points) 15 .
PIM was assessed using American Geriatric Society (AGS) Beers Criteria of 2015, which is a comprehensive set of explicit criteria that categorizes a drug as appropriate or inappropriate for the elderly aged 65 years and above in given conditions. The AGS Beers Criteria 2015 categorizes PIMs based on five criteria -according to organ system, therapeutic category and drug, according to disease or syndrome, according to drugs to be used with caution, according to drug interactions and according to renal function. In the present study a prescription was considered to be inappropriate if it contained one or more drugs included in any of the components of AGS Beers Criteria of 2015 (PIM vs. No PIM) 8,11 .

statistical analysis
Level of adherence, pattern of PIMs and other explanatory variables like gender, age, income of the family, education level of the patient, poly-pharmacy, co-morbidity, physical activity, presence of addiction, occurrence of ADR, presence of care giver and knowledge about the drug were presented using descriptive statistics.
Cross tabulation (Chi square test for linear trends) between adherence and various independent variables was followed by ordinal regression method using PLUM procedure to analyze the relationship between the dependent ordinal outcome variable, i.e. 3-level adherence score and the explanatory variables e.g. PIM and other sociodemographic factors. Initially all the variables were considered for inclusion in the ordinal regression model; however, only the variables with a 'p' value of less than 0.10, observed in univariate analyses were included in the in the final model. The variables were inserted at the same level of analysis using the method of forced entry 16,17 . The data were entered and analysed using SPSS version 16.0 (SPSS Inc. 2007).

results and discussion
In the present study out of 425 elderly patients, 258 (60.7%) were males and 167 (39.3%) were females. The average age of was 72.5±7.6 years (range 65 years to 95 years). The average There was no significant association between the patient's levels of income, occurrence of adverse effect, presence of caregiver with adherence therefore, these variables were not included in the ordinal regression model. The result of final ordinal logistic regression models, between the different levels of adherence and the explanatory variables is shown in Table 4. In the ordinal logistic regression models, high adherence had the least number of patients, so, it was taken as the reference. Similarly, the subgroups of the independent variables with the least number of patients were taken as reference in the ordinal logistic regression model. The independent variables for high adherence as predicted by the ordinal logistic regression model were an absence of PIM, female gender, age less than 75 years, educational level and number of drugs less than five. There was a significant association between the medication with an inappropriately prescribed drug and non-adherence. (OR: 2.089 with CI: 1.277-3.419, p = 0.003). Elderly patients with a potentially inappropriate medication as per AGS Beers Criteria of 2015 were twice likely be nonadherent to treatment than those without a PIM. However, this association was not observed when relating to number of PIMs in ordinal logistic regression.
In the present study it was observed that, female gender (OR = 3.438, CI: 2.216 -5.335; p = 0.000) and poly-pharmacy with less than five drugs use in elderly were predictors of high adherence (OR: 2.251, CI: 1.408-3.599; p = 0.001). Patient's age was a predictor of adherence (OR = 0.027, CI: 0.009-0.084; p = 0.000), however the negative parameter estimate indicates that these variables affects adherence negatively. Similarly, illiteracy in patients predicts adherence negatively (parameter estimate: -0.742, OR: 0.476, CI: 0.258-0.878; p = 0.018). The association between co-morbidity and low adherence could not be established in the ordinal logistic regression analysis, though in Chisquare analysis they showed a highly significant association. [Table 4] The present study has explored the association between potentially inappropriate medication as well as various socio-demographic variables and medication adherence in elderly population with chronic illness in an out-patient setup. Adherence was measured by three levelled MMAS-8, whereas PIM was evaluated by AGS Beers criteria 2015 and the association was analysed by ordinal logistic regression model.
It was observed that, 48.5% of the study participants were not adherent to the prescribed medications (MMAS-8 < 6), while 42.1% had a medium adherence (MMAS-8 6 to <8) and 9.4% were highly adherent to their medication with a MMAS-8 score of 8. These findings were consistent with results reported earlier studies 11,18 . High adherence levels to medication among patients is reported to be associated with higher treatment efficacy and overall reduction in healthcare wastage. Therefore, identifying the factors that lead to poor medication adherence is essential in any healthcare program. Our results show that PIM, gender, age, educational level, number of drugs are the significant predictors of adherence.
The association between potentially inappropriate medication and non-adherence was significant and elderly patients with a PIM were about twice likely to be non-adherent. It was observed in this study that, 23.8% of elderly patients in our study were prescribed at least one PIM. Similar observations were reported from other studies done in India and elsewhere 9,19,20 . The high prevalence of PIM in elderly may be due to the lack of awareness among physicians about the existence of Beers' criteria 21 . Further, the study participants had at least two chronic co-morbid conditions and were prescribed with an average of seven medications. Multiple medications are often necessary to treat multiple concomitant disease in elderly, however, unnecessary drugs add to the number, complexity and cost of an older person's drug regimen leading to non-adherence. This study also demonstrates that elderly patients with polypharmacy (use of more than five drugs) were twice likely to be non-adherent. Therefore, it is important to ensure that necessary medications are not omitted in the elderly patient. Wherever possible non-pharmacological therapy like, physiotherapy and advice on weight loss for osteoarthritis; psychological / social support for depression due to social isolation or recent bereavement; relaxation exercises like yoga for insomnia should be considered in lieu of prescribing and using medicines.
For improving medication adherence, one of the studies has suggested a combination of educational and behavioural strategies, selfmanagement interventions using medication organisation device like pillboxes or multidrug punch cards which are the least expensive and most widely used adherence aid for patients of chronic diseases 22,23 . Increasing patient's knowledge and providing them with basic skills to manage their disease can results in health benefits and could reduce their dependence on health care services and associated costs 11,18 . It was found that adherence decreased with the increase in the patient's age, this is in agreement with previous study 4 . Forgetfulness and physical disabilities are common in elderly. This may be the reason for low medication adherence in higher age group. On the contrary, some studies have reported that elderly patients were more adherent. The reason being that, they are suffering chronically, had a better knowledge of their disease condition and that most of the elderly are taken care by some care giver or a family member 15,22 . Another significant predictor of medication adherence was the level of education of the patient. It was observed that there is a negative association between illiteracy and high adherence. This finding corroborates with the observations of other studies which have concluded that uneducated people may not understand the importance of taking medications as advised, and unable to remember and follow the instructions related to medication accurately 10,24 .
There are few limitations of the present study. The foremost being that the study setting being the out-patient of a tertiary care government hospital, it might have captured the adherence level from elderly patients of low socio-economic group who are more likely to come to a government set-up for treatment. So, the study sample may not be representative of patients from other socioeconomic backgrounds and from domiciliary care set-up. The present study has excluded patients unable to communicate, comprehend and the very sick, so the findings cannot be extrapolated to this group. MMAS-8 is a self-reported method used to assess medication adherence, where, bias of over estimation is a documented concern. Nevertheless, MMAS-8 is a popular, practical and economical method of data collection, enabling the collection of a large amount of data in a short period of time. Besides, the questions are phrased in a way to avoid the bias of saying 'yes' as it is a common habit of patients to provide healthcare providers with a positive response. In this study, due to the non-availability of Indian criteria to find out the potentially inappropriate medication in elderly, AGS Beers Criteria of 2015 has been used. As such, Beers' criteria are not the gold standard, as they do not identify all the aspects of potentially inappropriate prescribing and are designed for population-based screening. They are not intended to substitute for professional judgment regarding the individualized needs of older adults.

conclusion
Potentially inappropriate medication is an important predictor of adherence to medication in elderly patients with chronic illness. Low adherence to prescribed medication among elderly patients is a recognized public health concern. Therefore, it is necessary to evaluate factors influencing nonadherence to decrease medication wastage and cost of healthcare as well as to improve disease outcomes, The findings of this study imply that, prescribers should carefully assess the necessity and appropriateness of medications prescribed in elderly to improve their adherence to therapy.