There is a growing body of research which evaluates how well adherence interventions work. Unfortunately, as there is no consensus about either the definition or measurement of adherence, evaluation studies have to set their own standards of success.

Typically, studies demonstrate better results for the intervention under investigation either against their own controls or against previously reported adherence levels. There is rarely any attempt to assess the relative efficacy of interventions or to attempt to evaluate the interventions on more systematic terms. For example, an intervention may be demonstrably effective, but this may be at the expense of practicality, economy or efficiency - issues which are likely to decide the long-term adoption and success of any recommended practice.

As knowledge grows in this area, health care providers are increasingly utilising a range of adherence interventions. These include the provision of information about HIV and treatments, skills building, practical adherence tools and aids, substance use interventions, and counselling about attitudes and expectations.

Interventions by health professionals

Most of the recent studies tackle the challenge presented by adherence not by identifying 'new' interventions but by developing or strengthening the roles of specific professionals, including pharmacists, social workers, nurses and doctors.

Two studies evaluated an enhanced role for pharmacists. In a randomised controlled trial (Knobel 1999), pharmacists discussed adapting therapy to patients' lifestyles, explained the clinical benefits of adherence and provided telephone support. The result was much higher levels of adherence compared to the control group. Similar outcomes were found for a pharmacist-lead programme of support, education and planning for everyday life (Graham 1998).

Professional advice for social workers is presented (without evaluation) in two papers: one describes a collaborative patient-team model of problem-solving within community social work (Wood 1998); the other describes the variety of treatment issues, including maintenance of adherence, that social workers have a role in supporting (Linsk 1997).

A paper on the role of the nursing profession makes clear that single interventions to tackle non-adherence are inadequate. It recommends supervised therapy, improving the nurse-patient relationship and patient education (Crespo-Fierro 1997). A similar paper draws attention to the important role that community nurses can play in developing home care plans for people with HIV, given that this is likely to be the setting for many problems in managing therapy (Ungvarski 1997). A pilot project known as Client Adherence Profiling-Intervention Tailoring (CAP-IT) conducted in a community nursing setting has identified knowledge and skills deficits in relation to adherence and the management of side-effects (Holzemer 2000). As discussed in more detail below, nurses have been involved in providing effective adherence education which combines element of education and counselling.

Given the centrality of doctors in patients' clinical experience, the paucity of papers discussing their professional practice is disappointing. A US study found that most doctors provided basic HIV adherence counselling. However, it was doctors who treated a greater number of HIV patients, and felt they had the skill, time and space to adequately counsel patients on adherence who provided extensive adherence counselling (Golin 2004). While the somewhat amorphous concept of a 'positive doctor-patient relationship' has been associated with improved adherence (Delgado 2003), it is doctors with adequate training and time who provide the most detailed adherence support.

Research has identified a number of barriers to adherence: doctors' incomplete knowledge of the demands of patients' lives; prejudice; insufficient appointment time; the lack of resources for patients who had doubts or questions; negative stereotypes about doctors, and insufficient multi-disciplinary communication (Vilas 1998; Bogart 2004).

Very high adherence rates have been reported by a clinical practice where a range of interventions were implemented: extended consultation time; tailoring the regimen to patient lifestyle; frequent follow-up; rapid viral load feedback and reminder calls and alarms (Workman 1998).

Spanish researchers have found that doctors need to have an individualised approach to adherence due to the variety of risk factors for non-adherence. They suggest that two questions aimed at identifying individuals who have problems taking medication as required should be asked at the end of every clinic visit: 'do you feel capable of taking medication as it has been prescribed?' and 'how much effort do you require to take the medication properly?' (Tuldra 1999).

Directly observed therapy (DOT) is also on the table as a strategy to improve adherence among HIV-infected people, particularly homeless people and drug users who may be socially marginalised (Mitty 2002; McCance-Katz 2002). Health outreach workers have a major role in the implementation of any such antiretroviral DOT program. An outreach program conducted among people who had poor adherence showed a significant reduction in non-adherence at 12 months and reduce viral load (Stenzel 2001). DOT has also been trialled in the prison setting (Babudiere 2000; Kirkland 2002). Results have been encouraging although these non-randomised, uncontrolled studies do not have a high level of scientific reliability.

Education, behaviour and support

Adherence interventions generally aim to inform people about HIV treatments and thus improve adherence, or to generate behaviour change through incentives, suggestions or emotional support. These strategies are often dubbed cognitive, behavioural and affective interventions:

• Cognitive interventions are designed to teach, clarify or instruct (eg. treatment information).

• Behavioural strategies are designed to shape, reinforce or influence behaviour (eg. pager system, individual assessment and counselling, drop-in programs).

• Affective strategies are designed to optimise social and emotional support (peer support, broadly targeted psychotherapy).

In practice, many adherence programs combine various elements of the three approaches.

Several studies suggest that interventions which provide patient education, counselling and support can improve adherence to antiretroviral therapy. However, the benefits of adherence education may be short-lived.

A randomised study which used a psychoeducative intervention at weeks 0, 4, 24 and 48 to improve long-term adherence to anti-HIV treatments found that 94% of the group who were exposed to the educative intervention and 69% of the control group achieved a viral load below 400 copies/ml after 48 weeks (Tuldra 2000). This study also reported the common finding that the effect of interventions may wane once they are removed, suggesting the need for ongoing support rather than one-off activities.

Group-based patient education produces better adherence than simply giving printed material but is not superior to social support, according to a randomised study of 168 people. Group-based patient education was led by a trained nurse and a peer educator and taught adherence and self-care skills over six two-hour sessions. Adherence was measured by self-report. Despite initial improvements in adherence and viral load, six-month follow-up showed there was no difference in adherence between the three groups. The authors concluded that benefits gained from adherence interventions may not persist over the longer term without reinforcement (Gifford 2001).

Goujard and colleagues randomised 365 HAART recipients to take part in a Treatment Education Programme (TEP) or to a control arm which received standard follow-up (Goujard 2003). The TEP included four face-to-face educational sessions of one hour each, conducted by doctor or a nurse, using a toolkit called Ciel Bleu, designed to teach patients about HIV pathogenesis, disease progression, the rationale for anti-HIV therapy, and the importance of adherence. TEP patients also received a beeper pillbox, and a number of devices to aid treatment scheduling. Every six months, study participants self-reported adherence over the previous week via a questionnaire. At entry, mean viral load was 2.42 log, mean duration of prior treatment was 4.0 years, 57% had viral load below 200 copies/ml, and adherence levels were comparable across the two arms, with 46% belonging to the upper level adherence stratum. Over six months follow-up, adherence levels improved in the TEP group, and fell in the control arm. There was no appreciable change in viral load and CD4 count in either group.

An 'adherence package' consisting of an education program, individualised planning of regimens, and the opportunity for participants to select from a range of adherence aids and reminder devices was tested in a randomised trial involving 43 people. Self-reported adherence improved in the week before assessment but did not change over a month, and viral load and CD4 markers did not change (Fairley 2003).

Three individual adherence sessions a with trained nurse produced some improvements in adherence in a randomised French study of 244 people on anti-HIV. In the intervention group, adherence rose from 58% at baseline to 75%, while in the non-intervention group adherence fell from 63% to 61%. Furthermore, average viral load fell significantly by 0.2 log in the intervention group (Pradier 2003). Another randomised study looked at the effect of three sessions with nurse counsellors. The technique used in this study was motivational counselling. Follow-up at two months indicated that people in the intervention group were more likely to report adherence than those in the control group (DiIorio 2003).

Motivational interviewing, cognitive behavioural therapy and problem-solving techniques were used as part of 'Life-Steps' - a single session counselling intervention for people reporting less than full adherence to anti-HIV treatments. Compared with self-monitoring using a pill diary and a questionnaire, 'Life Steps' produced faster improvements in adherence (Safren 2001).

A 'self-management program' which provided skills development, three monthly consultations, and monthly feedback on adherence using electronic monitoring of pill bottles improved adherence to medication in a randomised study from North Carolina (Smith 2003).

A study of 23 young people with HIV showed that a 12-week course and the support of a family member or treatment buddy boosted adherence in most participants. Reminder alarms were also used in this study (Lyon 2003).

Several non-randomised, pilot studies have also reported that behaviourally based programs can improve adherence (Molassiotis 2003; Dunbar 2003).

However, not all studies suggest programs and counselling are effective at improving adherence. For instance, a randomised study of ten sessions of cognitive-behavioural stress management and expressive supportive therapy among HIV-positive women in Miami, a largely non-white cohort many with histories of drug and/or alcohol addition, showed the intervention did not improve adherence although it did improve denial-based coping (Jones 2003). A health literacy program also failed to boost adherence despite successfully education 40 Latino men and women about HIV treatments (van Servellen 2003).

Reminders, trials and monitoring

While the studies described above tend to deploy multiple interventions, several studies have evaluated specific reminder interventions or the treatment 'dummy run'.

Collier and colleagues randomised 282 individuals enrolled in a comparative antiretroviral therapy trial (ACTG 388) to standard adherence care versus standard care plus scripted telephone calls (16 calls over 96 weeks). Adherence was measured at clinic visits by self-report over previous four days. Seventy-three percent of phone calls were completed successfully. Over 64% of subjects in each arm reported more than 95% adherence, and over 61% reported 100% adherence. Virological failure occurred in 34% of subjects although there was no difference in time to virological failure between arms. The authors concluded that within a clinical trial setting, telephone calls did not improve high levels of reported adherence, or virological outcome (Collier 2002).

A two-way pager system has also been trialled over seven months in an uncontrolled study of 25 people. Of the 19 people who used the pagers for at least three months, 58% showed perfect adherence on short-term assessment and 79% reported that the pager had improved their adherence (Dunbar 2003).

Safren and colleagues evaluated an online reminder system as an aid to HIV treatment adherence. After a two-week period where adherence to antiretroviral therapy was monitored using an electronic pill-cap, 71 people who recorded above 90% adherence were randomised to continued monitoring or to receive a pager (MediMom, an internet-based paging system). Adherence level at randomisation was 56%. The pager was associated with improved adherence at week 2 (70% versus 56% in the non-pager group), and at week 12 (64% versus 52%). It is noted that adherence in this population remained inadequate even following the intervention (Safren 2003).

More frequent viral load and CD4 cell monitoring may also contribute to adherence. A randomised study compared feedback on viral load every two months versus twice a year. At six months, adherence was not different between the two groups based on self-report. However, those who received more frequent feedback had a greater reduction in viral load (0.8 vs. 0.4 log₁₀; Haubrich 1999).

A study involving 20 men allowed the participants to commence combination antiretroviral therapy when they had achieved over 90% adherence in a pre-treatment one-week dummy run. Fifteen men subsequently started treatment; nine men needed only one dummy run but eight men required up to four weeks practice with placebo before they achieved full adherence. After one month taking active treatment, 13 of 15 men had over 90% adherence (Wagner 2002).

Interventions for people with drug and alcohol addictions

Given the association between drug and alcohol use and poor adherence, several studies have focused on improving the adherence of drug users. One study found abstinence was associated with improved adherence (Lucas 2002); another found that heroin substitution programs also improve adherence to anti-HIV regimens (Moatti 2000).

A recent study compared the use of directly observed therapy (DOT) in people on antiretroviral therapy in a methadone programme with either methadone users not receiving DOT, or non-drug users receiving HAART without adherence support. The found a significant improvement in viral load and CD4 counts, lasting up to one year, in those receiving DOT, compared with either of the control groups^[1]. However, DOT only seems to be beneficial for patients with documented adherence problems: in a similar study of community-delivered DOT in patients in Los Angeles County, adherence support counselling led to similar treatment responses to the DOT patients^[2].

In one recent study of drug users' treatment-taking behaviour, high levels of adherence were achieved by patients taking on a role as peer health advocates for each other (Broadhead 2002).

Rigsby and colleagues randomised 55 people (predominantly male, black and with a history of heroin or cocaine use) on stable antiretroviral therapy to receive either weekly nondirective inquiry about adherence (control); 'cue-dose training' designed to enable development of personalised cues to medication taking, with feedback from MEMS pill bottle cap; or cue-dose training plus cash payment for correctly timed pill bottle opening. Adherence was measured using MEMS and defined as opening the pill bottle two hours before or after a predetermined time. Over a four week period, adherence was significantly improved in recipients of cash payments, but not in those receiving cue-dose training alone, compared with the control arm. Eight weeks after the intervention, adherence returned to baseline levels in the cash payment group (Rigsby 2000).

Samet and colleagues randomised 151 people with a history of alcohol problems who were receiving antiretroviral therapy to usual follow-up or an intervention (4 meetings with a nurse trained in motivational interviewing: addressed alcohol problems; provided a programmable watch; enhanced perception of treatment efficacy). Adherence was measured by self reported 30 day recall. After 13 months follow-up, there were no differences in adherence between the two groups, or in other outcome measures (CD4 count, viral load, alcohol consumption (Samet 2002).

Conclusion

The literature on adherence to antiretroviral therapy presents a choice of perspectives for making sense of patient behaviour.

The analysis of population factors associated with adherence does not deliver a predictive model of the adherent individual. Arguably, it has heuristic value: drawing attention to areas of concern - such as drug use or psychiatric illness - which in any individual case may or may not be important, but which professionals ought to be sensitive to in their practice. However, it seems unlikely that doctors need to be reminded of such insights.

Ironically, the value of these studies may lie precisely in their failure to come up with consistent determinants of adherence or non-adherence. The surprise expressed by researchers when encountering findings contradictory to their expectations amply demonstrates the ease with which assumptions can be made about adherent or non-adherent individuals. The most interesting results are those which challenge common assumptions, such as the studies which suggest that peer support increases non-adherence; that those who doubt the therapy's efficacy are better at taking it; or that the more medications you take the more likely you are to adhere to them.

The lack of predictive findings indicates that specific groups of people (such as drug users) should not be excluded from therapy. However, the related conclusion is just as important: that professional assumptions about those who will be adherent are likely to be just as suspect as assumptions about those who will not. Clearly, the consequences of such mistaken assumptions may be considerable.

The studies which pay attention to patients' own reasons for non-adherence provide insight into why prejudging adherence on the part of any individual is both unhelpful and impossible. These reasons predominantly concern the idiosyncrasies of daily lives and the peculiar disturbances created by the imposition of a new routine on existing ones. They point the importance of adapting daily routines and minimising subsequent disruption to them.

From this perspective, adherence is about remembering and acting; about thinking and doing. Although this suggests an important role for cognitive-behavioural interventions, they do not feature prominently in the literature on interventions to improve adherence. Currently, this literature reflects the anxieties within all professions allied to medicine to be doing the 'right thing' to improve HIV treatment adherence, without any clear sense of exactly who should be doing what.