HEALTH-RELATED QUALITY OF LIFE AS PREDICTOR OF ADHERENCE WITH ANTIRETROVIRAL MEDICATION

Title of the Study: Health-related quality of life (HRQoL) as predictor of medication adherence in patients infected with Human Immune-deficiency Virus (HIV). Summary: Quality of life (QoL) is a broad term which involves evaluation of all aspects of life including, health, education, family life, housing, friendship, marriage, standard of living, and work. Health is one of the domains that affect our quality of life. The measurement of HRQoL is becoming an increasingly common activity in healthcare systems around the world. Health-related quality of life or biological outcome of treatment might predict adherence to HIV medication. This research is aimed to study the effect of HRQoL on medication adherence in patients infected with HIV. The original study was funded by NIH and conducted by Dr. Cynthia Willey, at University of Rhode Island during the years 1995-98. The purpose of the original project was to assess the stages of changes for adherence with HIV-Related Medications. The sample consisted of 145 patients. The questionnaire was developed by AIDS Clinical Trial Group (AACTG). Questionnaires were distributed to the patients in Rhode Island at different sites affiliated with Brown University AIDS program. These sites included: 1. The Miriam Hospital Immunology Center: This center serves majority of the HIV positive women in Rhode Island. 2. Stanley Street Treatment and Resources: This center serves the Greater Fall River Massachusetts area and provides care to indigent and intravenous drug users. 3. Veteran Affairs Medical Center in Providence, Rl: This center treats approximately 60 HIV positive men. Methodology: The data was collected by administering a standardized selfreported questionnaire to the subjects to assess the compliance to HIV drugs. The questionnaire covered various aspects like, Demographic, Economic status, Coping, Quality of life, Medication, etc. Four domains of health-related quality of life were measured using 12 questions based on SF-36 included in the questionnaire. Medication adherence was assessed as self-reported adherence and also using Medication Adherence Scale. Univariate and bivariate tests were run to check for confounding variables in the data set. Logistic regression was used to determine any interaction between independent variables. The effect of Quality of life domains on medication adherence were assessed by running logistic regression model after controlling for potential confounding factors. Results: The results of this study indicate that "vitality/fatigue" is significantly associated with 95% self-reported adherence in patients taking protease inhibitors. This study thus confirms that patients with better mental health are more likely to adhere to their medication regimen. This is consistent with previous findings by other researchers. No other meaningful association was found between any other domain of QoL and medication adherence.


INTRODUCTION
Acquired Immunodeficiency Syndrome (AIDS) is a significant healthcare, social, and psychological problem facing the mankind. When AIDS emerged from the shadows two decades ago, few people could predict how the epidemic would evolve, and fewer still could describe the best ways of combating it. As we face the third decade of AIDS epidemic, the impact of this disease is enormous. Human immunodeficiency virus is a major public health problem in all parts of the world [1].
Worldwide, nearly 40 million people are living with human immunodeficiency virus (HIV), which causes AIDS [2]. These include 19 .6 million males, 17 .6 million females, and 2.7 million children (less than 15 years of age). Globally, AIDS is a fourth leading cause of death [3]. In United States alone, approximately 1 million people live with HIV infection [2]. HIV/AIDS has become one of the highest expenditure infections in terms of cost of treatment and care provided, lost productivity hours, and disability [ 1].
Mankind faces multiple challenges in fighting AIDS. Social stigma and discrimination are the major obstacles to effective HIV/AIDS prevention and care. Fear of discrimination may prevent people from seeking treatment for AIDS or from acknowledging their HIV status.
Until quite recently, the disease was considered to carry an almost certain debilitating, downward course leading to early death from opportunistic infections. Not long ago, zidovudine, a nucleoside analogue reverse transcriptase inhibitor, was the only drug used to treat HIV. This drug interferes with the actions of specific HIV enzyme involved in the replication of cycle of HIV.
But, the treatment of HIV virus has changed immensely. With advances in HIV treatment regimens, HIV has become a treatable chronic illness that requires extensive clinical management [ 4]. New potent drugs are prolonging the lives of thousands of patients infected with HIV. There are now dozens of medications available to attack different enzymes in HIV virus lifecycles. These drugs can be classified into three categories depending on the enzymes they target in HIV lifecycle. These are: 1. Nucleoside reverse transcriptase inhibitors (NRTI's) 2. Non-nucleoside reverse transcriptase inhibitors. (NNRTI's)

Protease inhibitors (PI's)
Since the potential for mutation is very high with HIV, drugs are more effective when used in combination. Convergent therapy uses drugs from same class to target same enzyme, while divergent therapy used drugs from different class to target different enzymes. The combination of both convergent and divergent therapy is called highly active anti-retroviral therapy (HAART). The first of protease inhibitors were introduced in 1995. Since the late 1997, when HAART was first introduced, the combination of protease and reverse transcriptase inhibitors had proven effective in driving HIV viral loads to very low or undetectable levels [5] . The most commonly used combination includes one potent protease inhibitor and two NRTI' s. The impact of protease inhibitor based combination therapy has resulted in astonishing improvements in survival [6]. The survival rates have increased with both longer AIDS free survival and lower mortality [7] . This shift to the use of HAART for treating HN has led to increasingly complex drug regimen [8] . Adherence has been often called the "Achilles' heel" of highly active anti-retroviral therapy.
HAART is highly effective but the drugs have short half-lives and are highly selective, leading to drug resistant strains if therapeutic levels are not maintained [9]. The longterm effectiveness of HAART is dependent upon achieving maximum and durable suppression of HN plasma viral load [9,10]. Even in successfully suppressed patients, HN replication will rapidly rebound if HAART is discontinued. One of the major challenges to good adherence to HAART has been complexity of regimens.
HIV medication regimens are complicated and require extensive time and effort from the patient [ 11]. Many drugs need to be taken three times a day and the pill burden is overwhelming i.e. 15 -20 pills daily. Some drugs need to be taken with food, some without food and still others with dietary supplements. The complexity of HAART regimen sometimes requires patients to change their eating and sleeping pattern. This level of lifestyle change and accommodation may result in frustration and treatment failure [12]. Non-adherence to HN treatment regimen is a primary cause of treatment failure now. As a result of the pivotal role that adherence plays in the success of HAART, a tremendous amount has been written emphasizing the importance of adherence [5].
Adherence, often used interchangeably with compliance, is "an act, action, or quality of being consistent [13) with administration of prescribed medication". The term adherence is preferred over compliance because it affirms patient's active participation in choosing and maintaining a treatment regimen. The concept of adherence additionally extends beyond medication management, to encompass a comprehensive treatment plan.
Adherence simply means how accurately patients take their medications. One hundred percent adherence to any medication regimen is not easy to achieve. One recent study on adherence behavior in HN has shown that only 55-62% of patients are highly adherent to their medication therapy [ 14]. Many factors have been shown to affect patient's adherence.
Patient Factors and Health Beliefs: Demographic characteristics (like, age, gender, race, etc.) have not been consistently found to be predictive of adherence [15,16).
However, other factors like heavy alcohol use, drug abuse, and depression have been found to be associated with adherence [17,18). Various aspects of patients' beliefs about the nature of their disease [11), perceived importance of medication used [12), the health care system, and cultural factors affect medication adherence. Patients who perceive that their disease is a "serious health problem" and believe that the prescribed medications are necessary are more likely to adhere to prescribed medication regimens [12). In general, increased knowledge about disease and purposes of therapies increase adherence [11]. However, concerns regarding safety of the medication on long term and the medication side effects could decrease medication adherence.
Disease Factors: Investigations have found little relationship between type of illness and level of adherence [ 16] . Patients experiencing extreme pain or who are symptomatic are more likely to be adherent to their medications than patients who are asymptomatic. Many patients interrupt or stop taking their medications when they are asymptomatic.
Provider Factors: General satisfaction with medical care appears to have no bearing on adherence. However, patient's dissatisfaction and unfulfilled expectations with the treatment and the doctor results in low adherence rates [ 11,19] . The quality of interaction between doctor and patient can have major influence on health outcomes.
Provider-patient relationship, especially communication about chronic nature of disease, the need for regular therapy, the role of medications, and discussion of side effects improves patient adherence. Poor provider-patient communication is associated with poor adherence. Affordability of medication, insurance coverage, and cost of therapy is a barrier to achieving adherence.
Treatment complexities: Adherence is poorer in patients treated with more complex regimens [11,20]. It has also been shown that adherence normally decreases over time and with greater number of pills that one is required to take [ 11]. One study measuring compliance with inhaled medication in asthma showed that as medication dosing became more frequent, the adherence decreased from 71 % twice a day to 18% four times a day [20]. Anticipatory fear of side effects and secondary effects of illness such as nausea and dizziness can also reduce patient adherence.
Psychosocial Factors: Given the complexity of human behavior, multiple determinants including patient characteristics may affect medication adherence. Behavior associated with chaotic lifestyle, depression, alcohol and elicit drug use often reduce adherence [11]. Patients with positive adaptive coping have been found to be more adherent to their medications [21]. Other factors like, presence of stress, lack of motivation, pessimism, and depression appears to result in non-adherence. Patients having sufficient levels of practical, emotional, and cognitive social support show higher level of adherence [ 11,22].
Adherence is an important factor to achieve best outcomes in RN disease management. Although it is not known how adherent patients have to be to achieve best results, but it is believed that more adherent the patients, the more likely he is to have best results. Strict adherence to medication regimen yields high success. As more emphasis is laid on maintaining low viral count in HIV-infected patients, adherence to medication regiment has become an important issue [23]. Sub-optimal patient adherence has been shown to be related to inadequate viral suppression [24], reduced exposure to anti-retroviral drugs [25], the emergence to viral resistance [26], and RN disease progression and mortality [26]. Since the virus has the ability to mutate rapidly in absence of drug, taking anti-retroviral medication exactly as prescribed is the key for success of therapy. HAAR T is very effective but drugs have short half-life and are highly selective favoring drug-resistant strains if therapeutic levels are not maintained.
The rate of virologic failure sharply increases when less than 95 % of prescribed dose of drug is actually taken [17] . In addition to taking adequate prescribed medication, anti-retroviral and protease inhibitors need to be taken according to correct timeschedule, and for several drugs, dietary prescription. For most HIV-infected individuals, the alternative to lifelong adherent therapy may be devastating complication, often resulting in death.

Measuring Adherence:
One problem of measuring adherence is the lack of a standard measurement [ 15]. The methods commonly used to measure adherence can be classified as direct or indirect methods. Direct methods include pill count, biological assay, and electronic monitoring method, whereas indirect methods use questionnaire, interviews, or diary to estimate self reported adherence. All these methods have their respective advantages and disadvantages as discussed below.
Self-Reported Medication Adherence: This is one of the simplest methods used to measure adherence. Three main types of self reports have been used including surveys, interviews and diaries. These involve asking patients how often they took (or missed) their medications by use of variety of surveys. However, reports of nonadherence could be more reliable than reports of adherence. For example, the AIDS Clinical Trials Groups survey asks patients how many medication doses they missed during previous day, 2 days or, 4 days. Despite the ease of administration, the self reported adherence measure has several limitations. Investigations have revealed that patients tend to overstate the actual adherence [27]. The patients want to present socially acceptable responses. Even if the patients are truthful, there is no data to prove how long the patients can remember what doses were missed several days ago [27].
Pill Count: Many investigators use pill count as method of measuring adherence. The pill count is generally done by study personnel like nurse, physician or, other health care practitioner. Although pill count avoids the subjective evaluation of adherence and has been demonstrate to correlate more highly with electronically measured adherence [28], the proportion of doses measured by pill count often exceeds actual number of doses taken. This method has several other limitations. Patients may forget to bring the bottles to the pharmacy. Patents also 'pill dump' and dispose extra medication doses to appear more adherent [29].
Biological Assays: Biological markers and tracer compounds indicate patient compliance over an extended period [30]. Plasma levels of antiretroviral drugs provide unequivocal evidence that the medication has been taken. To assess adherence using this method, the time and dose of the medication must be noted. Also, repeated plasma levels need to be withdrawn from the patient to improve the sensitivity of the test.
This measure also has some drawbacks. Firstly, plasma levels only measure adherence to a dose prior to the visit or sample drawn [31]. Also, studies in many disease states have demonstrated, patients who are aware that they may have a clinical sample like blood drawn to measure adherence will be more likely to adhere to the dose immediately prior to the visit than to other doses [32] . Besides, the pharmacokinetics of many drugs, especially protease inhibitors, varies significantly from person to person [32].
Electronic Monitors: MEMS (Medication event monitoring system) and eDEM are two commercially available monitors to measure adherence [15]. These devices are fitted with special pill bottle caps equipped with electronic chip that records each time a patient opens the bottle. This method has various advantages over subjective and other measures. Over estimation of self-reported adherence and pill dumping can be detected by this method. Additionally, this method also provides the time and date the pill bottle was opened each time. However, the cost of these high-tech devices is a problem for some investigators. Moreover, patient may open the bottle but not ingest the medicine. Some patients make a cache of medicine to be taken at office. This may not be recorded by this measure.

The Importance of Health-related Quality of Life:
The term "quality of life" was first used in 1943, in a novel about working in aircraft factory. However, it became popular with social scientists in the 1970' s, as US cities and states tried to rate the "quality of life" they offered. World Health Organization (WHO) Quality of Life group defines quality of life as "an individual' s perception of their positions in the life in context of the culture and value systems in which they live and in relation to their goals, expectations, standards and concerns. Quality of life is an evaluation of all aspects of life including, health, education, family life, community, housing, friendship, marriage, nation, neighborhood, standard of living, and work. Some authors also add spiritual security to the list of domains affecting quality of life.
Health is one of the domains that affect our quality of life. WHO defined health in its constitution as "The state of optimum, physical, mental, and social well-being, and not merely the absence of disease or infirmity." The term health-related quality of life (HRQoL) encompasses multiple dimensions, including physical functioning, psychological state, general health status, family situational interaction, social ability, and somatic sensation.
The measurement of HRQoL is becoming an increasingly common activity m healthcare systems around the world. These measurements are taken for variety of reasons including, as indicators of population health status, outcome measures in clinical trials, in economic evaluation of new technologies, and in some ,cases, for individual patient management. There are two basic approaches to measuring HRQoL.
The first involves use of generic instruments that measure broad aspects of HRQoL.
These instruments are not designed to assess HRQoL relative to particular medical condition, but rather to provide a general sense of the effects of an illness. Medical outcomes study  is an example of generic HRQoL instrument. It measures HRQoL along 8 different domains: physical functioning, role limitation due to physical problems, bodily pain, general health, vitality, social functioning, role limitation due to emotional problems, and mental health. It assesses both physical and mental health scores of the individual and measures both positive and negative aspects of Physical and Mental Health [33].
Generic measures can be administered to different populations to examine the impact of various healthcare/therapeutic programs on HRQoL. These measures allow for comparisons of HRQoL across a variety of medical conditions. The major limitation of generic instruments is that may not be sensitive enough to detect subtle treatment effects specific to a particular disease.
The second approach involves the use of instruments that are specific to a disease (e.g., osteoporosis), a population (e.g., the elderly), or clinical problem (e.g. , pain).
These measures are more sensitive towards specific disease or population, and therefore to have greater relevance to practicing clinician. The Arthritis Impact Measurement Scale (AIMS) is an example of disease-specific instrument that measures HRQoL specific to arthritis.
Information about Quality of life of patients gathered systematically and routinely directly before consultation could be integrated in complex medical decision-making process [34]. Health related quality of life has been shown to be related with patient satisfaction and the main determinants of the health service quality improvement [35].
Health-related quality of life or biologic outcome of treatment might predict adherence to HIV medication. The correlation between HRQoL and adherence is complex and merits careful study [36]. Many authors have reported the impact of HRQoL on patient's ability to adhere to treatment. Quality of life (QoL) may be an important consideration in maximizing treatment consideration in Hepatitis patient [37]. Suboptimal patient adherence was shown to be related to inadequate viral suppression [38]. Previous research suggests that patients with higher symptoms (poor physical health) and depression are more likely to be non-adherent to medications [21,39]. The patients with better quality of life, do feel better about themselves and world, and are more likely to be adherent. Research data also supports the belief that survival and biomedical outcomes increase when the patient's perception of the impact of treatment on quality oflife is taken into consideration [ 40].
Most Highly Active Anti-Retroviral Treatment (HAART) regnnens are far from convenient for patients. HAART may have negative impact on patient's quality of life [ 41]. Short-term and long-term toxicities frequently occur which may have negative impact on the patient's Health-related quality of life (HRQoL). In addition, need for strict adherence to substantial number of pills, rigid time schedule, and dietary prescription may interfere with patient's daily activity. In patients with symptomatic HIV infection or AIDS, opportunistic infection could significantly affect HRQoL. In addition, social stigma , associated with the disease and associated pain caused interferes with patient's ability to perform daily activities. It is therefore assumed that most people infected with HIV I AIDS do not feel well and that disease has an effect on their quality of life. Economic status: Family income, job, type of health insurance coverage, and cost of treatment.

METHODOLOGY
• Coping: ways of coping with HIV.
• Social support: Social, financial and emotional support provided by family and friends.
• Quality of Life: Physical and mental functioning, general health, number of days spend in hospital in past two weeks, social ability and psychological health (as measured by SF-36).
• Medication used: Adherence levels, number of doses missed in past one month and three months, side effects, etc.

Measure of Health-related Quality of Life:
HRQoL is generally measured with a collection items, scales and domains. An item is a single question in an instrument. A domain identifies a particular focus of attention, like bodily pain, vitality, etc. SF-36 is a reliable, precise, and validated method of measuring quality of life [33,42]. There is a strong basis for interpreting SF-36 scales as measure of health and health-related quality of life. A positive correlation (ranging from 0.43 to 0.69) has been found between SF-36 scales and general measure of quality of life [33]. The four domains of physical and mental health were measured for the purpose of this study. This enabled the production of scores with the same reliability and validity as those reported here and in other Medical Outcome Studies.
Following set of questions were used to measure four domains of physical and mental health.

Physical Health:
Two domains of physical health were assessed by using the responses to the following questions.   This recoding method helped to convert second question to a six-level item of roughly equal variance to first question.
Computing Raw Scores: After item recoding, raw scores were computed for each four scales i.e. General health, Bodily pain, vitality, and Mental Health individually. The raw score is the algebric sum of responses for all items in that scale. For example, the raw scale score for the mental health scale is the sum of the scores for all five responses pertaining to general mental health. As recommended, if any respondent has missed more than one question on any domain, the score for that domain was not calculated. Finally each raw score was transformed to a 0 to 100 scale using the formula shown below.
Transformed Score= [CActual raw score -lowest possible raw score)] *100 Possible raw score range This transformation converts the lowest and highest possible scores to zero and 100, respectively. This transformation enables comparison of scores with norms derived from national health surveys and other published and forthcoming results. The scores were calculated separately for anti-retroviral and protease inhibitors drugs for the ease in analysis.

Measure of Adherence:
1. Percentage Adherence: Percentage adherence was calculated separately for anti-retroviral drugs and protease inhibitors. Percentage adherence in past three months was calculated using two sets of questions from the questionnaire. The number of doses missed was determined by using the response to question, "During the past three months, about how many times did you miss a dose of this medication"? The response to question, "How often do you take this medication" was used to determine the total number of doses the patient is supposed to take in three month period. The questionnaire has separate questions for different class of drugs, thus percentage adherence was calculated separately for anti-retroviral and protease inhibitors drugs.
The average percentage adherence for the past three months was calculated for both anti-retroviral and protease inhibitors.
No measure of adherence is perfect, and self-reported measures like questionnaires, often present the disadvantage of over reporting [ 43]. Although most literature suggests a cut-off limit of 80% in chronic conditions, successful HN therapy requires higher level of adherence. The rate of virologic failure sharply increases when less than 95% of prescribed dose of drug is actually taken [17] . Thus, a cut-off limit of 95% adherence was selected to dichotomize respondents. All patients showing greater than or equal to 95% adherence were classified as "adherent" and those showing less than 95% adherence were classified as "non-adherent". The adherents were coded as "l" and non-adherents were coded as "O".
2. Medication adherence scale: This is a previously validated scale to measure compliance. It contains six questions that are answered yes or no. Positive response indicates less medication adherence. A "Yes" was recoded as "l" and a ''No" was recoded as "2". The aggregate score for each respondent was obtained by taking the sum of all six responses. The aggregate score ranged from 6 to 12 with higher score indicating poor adherence. This scale includes following questions: • During the last 3 months, have you ever stopped taking your protease inhibitor/antiretroviral medication because you felt worse?
• During the last 3 months, have you ever forgotten to take your protease inhibitor/ antiretroviral medication?
• During the last 3 months, have you at times been careless about taking your protease inhibitor/antiretroviral medication?
• During the last 3 months, have you ever taken less of your protease inhibitor/antiretroviral medication than your doctor prescribed because you felt better?
• During the last 3 months, have you ever taken less of your protease inhibitor/antiretroviral medication than your doctor prescribed because you felt worse?
• Since you began taking protease inhibitors/antiretroviral medications, have you ever purposely taken more/less of the medication than your physician prescribed or discontinued your medications?
The MAS scores were calculated separately for Anti-retroviral and protease inhibitor drugs. The MAS score was not calculated (coded as missing) if any respondent has missed more than one question in six item scale. Also, the average score of antiretroviral and protease inhibitors were individually calculated. Respondents with a score of 6 were classified as "adherent" and respondents with a score of more than 6 were classified as "non-adherent". Thus any respondent who has marked "yes" to at least one of the six questions was classified as non-adherent. This rather strict cut-off level of adherence was selected to offset over-reporting of adherence in self-reported measure.

Statistical Analysis:
Dependent Variable: • Adherence to anti-retroviral drugs in the past three months (dichotomous with 95% cut-off level).
• Adherence to protease inhibitors in the past three months (dichotomous with 95% cut-off level) • • Adherence as measured using MAS for anti-retroviral drugs Adherence as determined using MAS for protease inhibitors . variables, each variable was sequentially dropped with replacement and its effect on the odds ratio and confidence interval was studied. The variables which did not have any effect on the odds ratio were dropped from the model. Separate models were run for each primary independent variables due to high correlation between them.   Patients with adherence levels of more than 95% were classified as adherent, while patients with adherence level of less than 95% were classified as non-adherence.

RESULTS
Using this cut-off, 87.7% (N= 115) of the patients were found to be adherent, while 12.2% (N= 169) were found to be non-adherent.
Medication Adherence Scale reported 42.2% (N= 62) being adherent, while 54.7% (N=75) reported not being adherent to their medication Patients with adherence levels of more than 95% were classified as adherent, while patients with adherence level of less than 95% were classified as non-adherence.
Using this cut-off, 83.7% (N= 62) of the patients were found to be adherent, while 16.2% (N= 12) were found to be non-adherent.
Medication Adherence Scale showed that 48.0% (N= 37) have bee adherent, while 51.9% (N=40) reported not being adherent to their medication   Chi square test showed no significant association between the adherence and other independent variables. This is consistent with past research that has shown that demographic variables are not associated with HN adherence. Chi square test showed no significant association between the adherence and other independent variables. This is consistent with past research that has shown that demographic variables are not associated with HN adherence. Chi square test showed no significant difference between the adherence and other independent variable except for variable "gender". Males were found to be more adherent than females . The variable "annual family income", and "T-cell count" were found to be significantly associated with MAS adherence. Patients with annual income more than $15,000 were found to be more adherent than their poorer counterparts. Similarly, patients with T-cell count of less than 200 were found to be more adherent than patients with T-cell count of more than 200. This is found to be consistent with previous research. Table. 11: Multiple Chi-Square tests done on "General Health" (Categorical) and other Independent Variables (Categorical) for people on anti-retroviral medication.
The variable "ethnicity" and "annual family income" were found to be significantly associated with "general health". Whites reported better general health compared to non-whites. Similarly, patients with annual income greater than $15,000 had better mean score on general health compared to patients with annual income less than $15,000. Table. 12: Multiple Chi-Square tests done on "Bodily Pain" (Categorical) and other Independent Variables (Categorical) for people on anti-retroviral medication.
The variable "ethnicity" was found to be significantly associated with "bodily pain".
Whites reported better mean score on bodily pain compared to non-whites. This difference was significant at a p-value of 0.0327. No other variable of interest show any significant association with bodily pain. Table. 13: Multiple Chi-Square tests done on "Vitality/Fatigue" (Categorical) and other Independent Variables (Categorical) for people on anti-retroviral medication.
The variable "gender" was found to be significantly associated with "vitality" (p-value= 0.0047). Females had better score on vitality than their male counterparts. No other variable show any significant association with vitality. Table. 14: Multiple Chi-Square tests done on "Mental Health" (Categorical) and other Independent Variables (Categorical) for people on anti-retroviral medication.
No variable show any significant association with the variable "mental health". Table. 15: Multiple Chi-Square tests done on "General Health" (Categorical) and other Independent Variables (Categorical) for people on Protease Inhibitors medication.
The variable "years of education" was found to be significantly associated with "general health". Patients with more than 12 years of education had better mean score of general health than patients with less than 12 years of education (p-value= 0.026). The variable "annual family income" was found to be significantly associated with "bodily pain". Patents with annual family income more than $15,000 reported better mean score on bodily pain compared to patients with annual income less than $15,000.
No other variable of interest show any significant association with bodily pain. No variable show any significant association with the variable "vitality/fatigue". The variable "gender" was found to be significantly associated with "mental health".    This table summarizes the final logistic model run between dependent variable (MAS adherence) and each of primary independent variable controlling for the confounding variables. The variable "annual family income" was found to be significantly associated with both MAS adherence and bodily pain.

DISCUSSION
Adherence is an important factor to achieve the best outcomes in HN disease management [14]. Strict adherence to medication regimen in HN therapy is important for maintaining low viral loads. The purpose of this study was to assess HRQoL as a predictor of medication adherence. Medication adherence scale and percentage adherence were used as measures of adherence in this study. A rather strict 95% cutoff mark was selected to define adherence based on recent studies and literature review [17,45]. Given the complexities of HN therapy and viral response, the alternative to strict adherence for many patients could be death.
Various factors affect patient's ability and desire to adhere to medication regimens.
Two demographic factors, namely "gender" and "annual family income", were found to be associated with 95% adherence and MAS adherence respectively in patients taking protease inhibitors. Males reported more adherence compared to females. This could be because HN positive females might feel more depressed, anxious, and distressed than HIV positive males [ 46]. Similarly, patients with high family income reported more adherence compared to patients with low family income. This association is difficult to interpret, as the cost per prescription is not known. Although wealthy patients have more financial access to medications then low-income patients, this advantage is offset by prescription coverage available to poor patients through

Medicaid.
No other meaningful association was found between any other demographic variable (age, years of education, living status, ethnicity, duration of ( illness, etc) and medication adherence in anti-retroviral or protease inhibitor drugs. This is consistent with previous studies and research [39,47]. Any inconsistency could be explained by study limitations discussed later.
In recent years, interest has increased in the measurement of HRQoL, in relation to health-care. In a chronic disease like HN, the patients' physical and mental health is significantly affected. The opportunistic infections experienced by AIDS patients often have detrimental effect on their physical and mental health. Fatigue, emotional and psychological stress is often associated with HN infection. In this study, many demographic and clinical factors were found to be associated with HRQoL domains.

For patients on anti-retroviral medications:
Physical health domains like "general health" and "bodily pain" were significantly associated with "ethnicity". Whites reported better "general health" and lower "bodily pain" than non-whites. Patients with high annual family income reported better general health and vitality that patients with annual family income of less than $15,000. This is consistent with Center for Disease Control findings [48]. Also, males reported better vitality and low fatigue than females. Previous research also confirms similar difference in perceived quality of life between males and females [33,49].
For patients on protease inhibitors: "Years of education" and "annual family income" were significantly associated with both "general health" and "bodily pain".
Patients with higher level of education and annual family income of more than ( $15,000 reported better general health and low bodily pain than their counterparts.
This could be because educated people are less likely to adopt risk-taking behavior such as smoking, drugs, and unprotected sex [ 48]. Also, education and family income increases self-esteem and confidence, life opportunities and social support. It allows people to adopt healthier lifestyle and seek better treatment. As with patients on antiretroviral medications, gender was found to be significantly associated with "mental health" in patients on protease inhibitors.
The patients in this study scored significantly lower on all domains of HRQoL than general US population (Results: adherence. Patients with high vitality score were found to be more adherent and patients with low vitality score were found to be less adherent to their medication. This is consistent with the results of previous studies, which suggest association of poor mental health, particularly depression, with medication adherence [39].
Although, other domains of quality of life, especially physical health, were expected to be associated with medication adherence, the result of the study does not confirm the same. The limitations of the study could have affected results significantly.

Measurement of adherence:
There is no way to measure adherence in the outpatient setting with absolute precision and accuracy [50]. Although no measure of adherence is perfect, self-reporting method, often, tends to overestimate medication adherence.
Recall bias and patients subjectivity often influence patients responses to questions.
This could lead to inaccurate statistical analysis and inaccurate results. found to be skewed. Statistical techniques were used to rectify this shortcoming.
The data was collected in 96-97, when combination therapies were recently introduced. This study precedes many currently available anti-retroviral drugs and HAART. Thus, the results of this study may not be generalizable to today.

CONCLUSION
Though overestimation of self-reported medication adherence cannot be ruled out, patients in this study group reported good level of adherence. Approximately 85% of the patients were adherent based on 95% cut-off limit. This high level of adherence is necessary for the successful management of HIV. In-contrast, MAS showed that only 55% of the patients were adherent to their medication. One of the mental health domains i.e. vitality/fatigue was found to be significantly associated with 95% self reported adherence in patients taking protease inhibitors. This study confirms that patients with good mental health are more likely to adhere to their medication regimen, using 95% cut-off adherence in patients taking protease inhibitors.
All HIV patients scored considerably low on all health domains when compared with US general population. Interestingly, HIV patients scored lower on all domains when compared with medical conditions like diabetes, hypertension, arthritis, CHF, and clinical depression [33]. These scores suggest that HIV patients suffer from strong physical and mental impairments, and have low quality of life. Stable demographic factors, like sex, race, years of education, and annual income were found to predict HRQoL. These factors precede medication adherence. The society as well as public health professionals should take a note of this, designing new policies and interventions to improve patient's HRQoL.  GH=(GHr-1)/4*100;