How Much Does It Really Matter If You Take Your Pills on Time?

As 3- and even 4-drug combinations have become the standard of care in the treatment of HIV infection, the new regimens have brought both increased hope for effective control of the virus and a new and sometimes confusing set of issues. These regimens are complex, often involving 20 or more pills a day, with differing requirements regarding timing of doses and whether or not the drugs should be taken with food. For many taking these combinations, the need to organize one's life around a pill-taking schedule is a chore, and sticking to the regimen is a constant challenge.

But adherence to a treatment regimen is more than just a matter of convenience. Study data strongly suggest that failure to take antiretroviral medications on schedule can lead to treatment failure and the development of resistant strains of HIV. Unfortunately, the existing data are not nearly precise enough to tell a given person how much latitude he or she has. Nearly every person with HIV wonders: will I be in trouble if I take my pills an hour late? Two hours? If I miss one dose? Two? Three? Unfortunately, scientists lack definitive answers. In addition, individual variations in drug absorption, metabolism and body chemistry probably make it impossible to give precise answers that will apply to everyone.

Still, useful lessons can be drawn both from study data and from the real-world experiences of doctors and people with HIV using these combination treatments. Much can be learned from experience with other diseases as well. This article will examine not only the data and the conclusions that doctors and researchers have reached thus far, but also some of the methods that physicians and people with HIV are using to make it easier to adhere to these difficult regimens and reduce the chances of treatment failure.

A Short Course in Viral Resistance

Doctors have long known that many people do not take prescription drugs the way they are directed. In an article published in 1984, Stephen A. Eraker, MD, and colleagues noted, "noncompliance may be the most significant problem in clinical practice today." Reviewing hundreds of studies done over the years, they found that treatment noncompliance is common among people with chronic conditions. For example, half of high blood pressure patients in one study discontinued care within a year, and of those who continued treatment, only about two-thirds took enough of their medication to control their blood pressure.

Poor adherence can be a problem with any serious illness. Drug-dosing schedules are designed to maintain adequate blood levels of medication. When failure to take medicines on time leads to concentrations that are too low, much or all of the benefit can be lost. In the case of high blood pressure, this might increase the risk of cardiovascular problems. Ample evidence shows that among those who use trimethoprim-sulfamethoxazole (TMP-SMX; brand name Bactrim or Septra) for Pneumocystis carinii pneumonia (PCP) prophylaxis, "breakthrough" cases of PCP occur most often among those who fail to take their medication as prescribed. In treatment of HIV disease, failure to maintain adequate drug levels in the blood can lead to viral resistance that can render a once-potent drug useless, and perhaps eliminate other treatment options as well. To understand how this works, it is necessary to know the basics of how resistance develops.

The HIV in an infected person is constantly replicating or making copies of itself — hundreds of millions of times per day. When the virus replicates it sometimes makes mistakes called mutations. Mutations only occur when HIV replicates; they do not occur when the virus is inactive.

Some mutations do not make any meaningful difference and some may weaken the virus. But sometimes mutations occur in the specific enzymes (reverse transcriptase and protease) that are targeted by anti-HIV drugs, and some of these mutations allow the virus to resist the effect of a given drug. With certain drugs, a single mutation can create high-level resistance; with other drugs, several mutations must accumulate before the virus becomes highly resistant.

When no anti-HIV drugs are being used, viruses with these resistance mutations have no special advantage and rarely become an established part of an individual's virus population. At the other extreme, if anti-HIV drugs manage to stop viral replication completely, there is little chance for mutations to occur and thus little chance for resistance to develop.

The possibility of the development of resistance arises when HIV manages to replicate despite the presence of antiretroviral drugs. The drugs exert what scientists call "selection pressure" that gives resistant mutants an advantage. One way to think of this is to picture a partly effective anti-HIV regimen as a filter that screens out the most drug-susceptible strains, while viruses with resistance mutations are more likely to get through. The result is that, over time, normal wild-type virus dies out and only resistant strains remain, rendering those particular anti-HIV drugs ineffective. Complicating the situation further is what is known as cross-resistance. Mutations that cause resistance to one drug can sometimes cause resistance to other, similar drugs as well, limiting the patient's future treatment options.

How quickly resistance develops is partly a function of how much selective pressure drugs exert. A weak regimen, such as AZT monotherapy, exerts relatively little selection pressure, and may be compared to a filter full of holes that lets a lot of wild-type virus through. While a patient on AZT monotherapy is likely to develop resistance eventually, it can take a year or even longer. Conversely, stronger regimens, if they don't stop HIV replication completely, act as more effective filters, and resistance can develop more quickly.

As knowledge of resistance has increased, more and more HIV/AIDS researchers and clinicians have come to regard the complete elimination of viral replication as the goal of therapy. Yet even a completely effective regimen can fail if not taken properly. If the gap between doses is too long, or if a drug that must be taken on an empty stomach to be properly absorbed is mistakenly taken with food, the result can be drug levels that are not high enough to completely stop HIV from replicating. Because some drug will still be present, the "filter effect" of selection pressure kicks in, giving resistant mutants an advantage. If enough of those mutants get through, a population of resistant virus can take hold. Obviously, the more erratic the drug-taking schedule, the greater the likelihood of drug levels dropping into the danger zone of partial viral suppression.

Old Drugs and New

A few years ago, when the standard of care was limited to weaker regimens such as monotherapy or 2-drug regimens involving AZT, ddI or ddC, there was less worry that noncompliance with treatment regimens might provoke resistance. The treatments never stopped replication completely, and they generally exerted relatively weak selection pressure. Missing a dose or two made some difference, but not a dramatic one. Now that regimens exist which can keep viral load to below detectable levels for extended periods, any break in that line of defense is seen as a cause for concern.

Another new factor is the different chemical behavior of the newer classes of antiretroviral drugs. The drugs that give these new combinations their punch, protease inhibitors and non-nucleoside reverse transcriptase inhibitors (NNRTI), behave somewhat differently in the body than the nucleoside analogs (e.g., AZT, ddI).

When nucleoside analog drugs enter a cell, they undergo a chemical change known as phosphorylation. This has the effect of keeping the drug inside the cell for a considerable length of time. But the protease inhibitors do not undergo such a change, according to John Leonard, MD, who, as head of the antiviral program at Abbott Laboratories, spearheaded the development of the protease inhibitor ritonavir (Norvir). "They're not trapped inside these cells by being metabolized" like the nucleoside analogs are, he explains. "From a metabolic point of view, they behave differently."

This is important because it is inside cells — where the virus is hijacking the cell's machinery to make copies of itself — that antiretroviral drugs do their work. Once a nucleoside analog drug gets into a cell, it tends to stay there, even if a missed dose causes the level of drug in the surrounding blood to drop. Because protease inhibitors are not chemically locked into cells, a protease inhibitor will flow back out of the cell when the drug level in the surrounding blood drops, until the levels inside and outside the cell are the same. The likely result is that there will not be enough drug left in the infected cell to do the job. Thus, protease inhibitors are far more sensitive than nucleoside analogs to varying levels of drug in the bloodstream, and missed doses are more likely to impact the drugs' efficacy.

The situation with the NNRTI is similar, notes Maureen Myers, MD, who directs clinical research in virology for Boehringer Ingelheim, maker of nevirapine (Viramune). Nevirapine, she notes, "is a small molecule, smaller than the protease inhibitors, so it penetrates the cell very easily." That can be an advantage in that the drug quickly gets to where it must go to be effective. However, it too can flow out of the cell easily if the drug level in the surrounding blood drops.

There are important differences between the protease inhibitors and the NNRTI. On the positive side, some NNRTI tend to stay in the blood longer before being processed out by the liver. This means that nevirapine, for example, can be taken twice a day without drastic changes in blood drug levels, while all of the approved protease inhibitors except ritonavir must be taken 3 times a day. It also means that NNRTI tend to have a somewhat greater margin of error, with a late dose less likely to send blood levels down into the danger zone. In another sense, NNRTI have a smaller margin of error, since a single mutation alone may produce high-level resistance, whereas several mutations are required to produce high-level resistance to the protease inhibitors.

Another complicating factor is the natural person-to person variation in drug metabolism. Many of the new drugs are primarily metabolized in the liver through what is known as the CP450 system. The rates at which one component of this system, the CYP3A enzyme, clears a drug from the blood can vary dramatically from one individual to another. Leonard estimates that this person-to-person variation is about 30% for ritonavir. Laurent Fischer, MD, Hoffmann-La Roche's Medical Director for HIV-Related Products, estimates that patient-to-patient variation in clearance of saquinavir (Invirase) is in the range of 40%. Dean Winslow, MD, Director of Medical Affairs for Agouron Pharmaceuticals, says variation in clearance of his company's drug nelfinavir (Viracept) is also in the 30% range.

This variation impacts how much margin of error a person has in taking pills. For most people taking ritonavir, Leonard says, "I'd guess there's a couple of hours of play. But there's always one guy who needs to be more precise because of how he metabolizes." There is no readily available test to measure CYP3A metabolism, but tests do exist to measure the levels of drug in a person's blood. Some doctors are starting to look at using such tests as part of their normal clinical practice.

What the Data Show

A few studies of anti-HIV treatments have looked at the relationship between adherence and the efficacy of the regimens being studied. But researchers hasten to point out that all of this data has limitations, since there is no perfect method of measuring how study volunteers take their medications. Much research relies on people's own reports of what they took and when, with the obvious problems of misreporting due to imperfect memories or embarrassment. Other studies have asked participants to bring their pill bottles with them on follow-up visits so investigators can count what remains in order to determine how much they have taken. This seems more precise, but Myers notes that "just because people return a bottle with 1 pill in it doesn't mean they consumed the other 49."

A few studies have used pill bottles fitted with electronic meters that record each time the bottle is opened. But again, the fact that a person opened the bottle does not prove that he or she actually took the required dose at that moment. The meter tells researchers nothing about whether the person remembered whether he or she was supposed to take the pills with or without food.

The first warnings about adherence to protease inhibitor regimens came from studies that did not specifically examine adherence at all. In early dose-ranging studies of ritonavir and indinavir (Crixivan) monotherapy, investigators noted a disturbing pattern in participants who were taking doses that proved to be too low. People on the lower doses, Leonard explains, "tended to have declines in virus, but the time they stayed suppressed was lower." Worse, when the virus bounced back, resistant strains emerged.

These studies provided clear evidence that suboptimal dosing can lead to quick rebound of viral load and the development of resistance, but they did not directly show what would happen in a real-world situation involving combination therapy at an appropriate dosage with occasional missed or late doses. Still, most clinicians and treatment advocates took the results as a signal that taking these drugs haphazardly would be risky.

More recently, several studies have shown a correlation between adherence to antiretroviral regimens and their effectiveness. For example, both Fischer and Leonard point to an ongoing ritonavir/saquinavir combination trial, in which researchers are doing pill counts as well as asking participants about their pill-taking. "Noncompliant" patients were defined as those in the low-dose arm who had taken 85% or less of their medication and those in the high-dose arm who had taken 60% or less. Final results have not been compiled, but at week 24 there was a clear difference between the compliant and noncompliant groups. Of compliant patients, 90% had a viral load of 200 copies/mL or below. Among noncompliant patients, only 66% reached a level of 200 copies/mL or lower.

The INCAS (Italy-Netherlands-Canada-Australia-States) trial, which compared AZT/ddI to AZT/nevirapine to AZT/ddI/nevirapine, produced similar findings. Overall the 3-drug combination produced greater and more prolonged reduction of viral load than either 2-drug regimen. Using what Myers acknowledges is an "arbitrary definition" of noncompliance — having missed 1 month's worth of at least 1 study drug during 28 weeks of follow-up — the adherent patients in the 3-drug group did strikingly better in terms of viral load suppression than those classed as noncompliant. When the results for individual participants were graphed, she adds, the results were "incredible, because you'd see an interruption [in treatment] and the viral load would go up."

A somewhat similar finding emerged from a small group of participants in a high-dose saquinavir monotherapy study conducted at Stanford University. After an initial cohort showed wide variation in response to the drug, investigators decided to follow a group of 7 patients closely in an attempt to correlate pill-taking (measured by metered pill bottles) with changes in viral load (measured monthly) and the emergence of resistant mutants. Their results were published in the December 25, 1996 issue of The Journal of the American Medical Association.

Two patients who took clear "drug holidays" (defined as 3 or more days off treatment) showed a dramatic spike in viral load shortly thereafter. Their viral load dropped again when they resumed treatment, but resistant mutants soon appeared. In 2 other patients, an increase in viral load late in the study seemed to correlate with more haphazard pill-taking.

But the pattern did not always hold. The most adherent patient, who hardly missed any doses, still developed resistance and saw his viral load eventually rebound. One patient who took a drug holiday showed no increase in viral load and no emergence of resistance.

Overall, the researchers wrote, the results "suggest a relationship between brief periods of low compliance and increases in plasma RNA levels." Still, other factors were clearly at work. For example, it may have been that the adherent patient whose viral load rebounded simply was not absorbing the drug well. Like the ritonavir dose-ranging trials discussed above, this was a monotherapy study not reflective of currently accepted combination treatments.

Several other studies are underway in which patient adherence is being monitored, but for now, existing research paints only a fairly crude picture — what San Francisco AIDS specialist Virginia Cafaro, MD, refers to as "wisps of data." Clearly, missing a large number of doses or several doses in a row is undesirable, but just how many lapses a person can have without getting into trouble — and how much this varies from individual to individual — remains an open question.

Is Testing the Answer?

The key to any drug's efficacy is reaching and maintaining an adequate level of drug in the body. Given the natural variations in absorption and metabolism of anti-HIV drugs, some believe a solution may lie in testing patients' blood levels of the drugs they are taking. In the May issue of the Los Angeles-based newsletter Being Alive, Bernard Bihari, MD, describes a combination trial using indinavir and nevirapine in addition to nucleoside analog therapy. "The 3 patients who did not show dramatic drops in viral load all described no indinavir side effects, and 2 still have none on 1,600 mg every 8 hours," he notes. That dose is 60% higher than what Boehringer Ingelheim has recommended be used to compensate for nevirapine-induced reduction in indinavir blood levels. Interestingly, Bihari noted a spike in the viral load of one person who briefly failed to follow the requirement that indinavir be taken on an empty stomach, an issue that researchers have thus far barely examined.

Bihari speculates that these individuals are not reaching therapeutic levels at this higher dose because of variations in absorption and/or metabolism, and compares the situation with "drugs used to treat such diseases as epilepsy, congestive heart failure, bipolar disorder and cardiac arrhythmias." Doses of drugs for these conditions, he writes, need to be adjusted by as much as 300% in normal clinical practice, adjustments that are done with the aid of tests for blood levels of drugs. Similar adjustments may be needed with some antiretrovirals. As a result, "I am arranging for the availability of indinavir and nevirapine blood level testing, and expect to incorporate these as routine clinical tools in my practice."

So far there has not been a headlong rush in the direction of blood level testing among AIDS physicians, although many are intrigued by the idea. Meg Newman, MD, a clinician at the San Francisco General Hospital (SFGH) AIDS ward and director of SFGH's AIDS education program, says, "I think it's interesting. And I think it needs to be tested and clinically validated."

However, Abbott's Leonard is skeptical about the usefulness of blood level testing, at least with the current level of knowledge. "There are plusses and minuses to it," he observes. While it would "be nice for a patient and physician to know, whatever drug he's taking, that he's in the ballpark" in terms of maintaining adequate levels, such testing could be "potentially misleading." He explains, "I can give you a [blood level] number for ritonavir based on populations infected with a virus that's never been exposed to a protease inhibitor before," but with increasing use of these drugs more and more people will have virus that has at least some exposure to them. Resistance to protease inhibitors is "a gradient," with many intermediate steps between zero resistance and high-level resistance, "and we don't know the [therapeutically effective blood level] number for each of those intermediate steps." For testing of blood levels of drugs to be truly meaningful, Leonard argues, "you have to have the genotypic [relating to the virus' genetic structure] and phenotypic [relating to the virus' actual characteristics] information to put alongside it." Such testing is not impossible, he continues, "that's what we do all the time in people we give antibiotics to." Whether it will become the standard of care in treatment of HIV disease remains to be seen.

Treatment Decisions

Fear that haphazard use of combination therapy could speed emergence of resistant strains that might in turn be transmitted to other people, leading to an epidemic of multidrug-resistant HIV, has provoked considerable discussion among physicians and others. Important questions include: should some people with HIV be refused combination antiretroviral treatment if a doctor does not think they can adhere to the regimen? How does a physician decide who can or cannot handle these treatments? Is there a conflict between a patient's right to optimal treatment and the public health needs of society?

Unfortunately, many media accounts seem to be based on stereotypical assumptions about certain groups of people which suggest, for example, that homeless individuals or users of illegal drugs are unable to handle combination anti-HIV regimens. But research suggests that many of these stereotypes are inaccurate.

In a number of studies, factors such as poverty, homelessness and lack of education have correlated poorly or not at all with failure to adhere to treatment regimens. "Sociodemographic characteristics have never predicted adherence very well," observes Lois Eldred, MD, of the Johns Hopkins University School of Medicine. Eldred's work with a mostly poor and African American urban clinic population was presented at the XI International Conference on AIDS in Vancouver, BC, last July. "Class has never predicted it. Race has never predicted it. We would like providers to not be biased in who they prescribe drugs to, but to take a closer look at people's lifestyles."

While some studies do seem to show a correlation between injection drug use and lack of adherence to HIV/AIDS treatments, in other research drug injectors have done just as well as non-drug users at sticking to their regimens. Eldred found no correlation between adherence and use of marijuana, alcohol or smoked or snorted cocaine, but did find that those who injected drugs at least "several times a week" were less likely than others to adhere to their treatments. Eldred suspects that more frequent drug injecting may be associated with what some researchers call a "chaotic lifestyle" — instability in many of the basics of day-to-day living such as food and shelter — that makes it harder for such individuals to take their medicines as directed. But she emphasizes that it is dangerous to assume that because a person uses drugs their life is chaotic, noting, "There are people who inject drugs who work full-time jobs."

David Bangsberg, MD, of SFGH has noted a similar pattern in his ongoing work with homeless and marginally housed HIV patients in a study headed by principal investigator Andrew Moss, PhD. In this group of patients, many of whom have substance abuse and mental health issues in addition to HIV infection, Bangsberg says that so far the SFGH researchers have found that "most of the homeless access medical care and adhere to medication in very reasonable ways, not dissimilar from stably housed HIV-infected people… But there's a small portion that have fairly chaotic lives" and have trouble staying on their treatments, keeping clinic appointments, etc. The poor and homeless, of course, hardly have a monopoly on chaotic lives. An educated, affluent, self-employed businessperson or professional with a frenzied schedule, constantly rushing from one appointment to another, may have as much trouble adapting to a complex medication regimen as a homeless person.

Anticipating who will do well and who will not is "a daunting challenge" for physicians, Bangsberg adds, noting that even doctors in quiet suburban practices have trouble assessing which patients are correctly taking their medications for high blood pressure or other chronic conditions. Indeed, Eraker and colleagues noted, "Few medical schools teach their students to recognize the conditions under which the patient can be expected to follow advice (much less the methods for communicating effectively with patients), or the interview skills needed to assess what the patient knows, believes or is concerned about."

Some factors do seem to be consistently significant in predicting adherence. For example, patients do better on simpler regimens than on complex ones. The number of times per day that medicines must be taken, rather than the number of pills taken each time, is the main factor. Eldred, who was studying 1- and 2-drug regimens, notes, "In my study, compliance really dropped off when there was more than twice-a-day dosing."

Depression and other psychological problems may interfere with adherence and must be addressed. Numerous studies suggest that communication between doctor and patient is critical, not just in terms of transmitting knowledge, but in establishing trust as well. One study found that a patient education program greatly increased patients' knowledge of the drugs they were taking, but did not significantly affect their adherence. The person with HIV must feel secure about the treatment that has been prescribed, and the doctor must take the time to understand the attitudes, beliefs and lifestyle factors that will impact the person's willingness and ability to stick to a treatment plan.

Promoting Adherence

What all of this means is that in 1997 it is no longer adequate for a physician to simply write out a prescription for antiretroviral drugs and hand it to a patient with a couple of sentences worth of mumbled instructions. With regimens now so complex that people often have to literal-ly organize their lives around their pill-taking, physicians are finding that they need to spend some time getting to understand those lives before recommending treatment.

"I'd certainly say that your first shot at a protease inhibitor is likely to be your best shot," Newman says, "and everything else may be salvage therapy." This means that physicians and patients must choose that first shot carefully, paying attention to a regimen's ability to fit an individual's lifestyle as well as its effectiveness, and must avoid the temptation to move too fast. "I've learned to take a deep breath and really get to know somebody and work with them," Newman comments, "even if it means deferring therapy for 2 to 6 or 8 months until they really understand what the regimens are about and they really make a personal choice that they're ready for the type of focus and commitment that these regimens require."

Physicians and people with HIV also must take the time to go over the requirements of a particular drug's dosing schedule and potential side effects to make as sure as possible that the treatment chosen will be one the individual can live with. Indinavir, for example, must be taken at least 1 hour before and 2 hours after eating, 3 times a day, which can present a real problem for people with hectic and irregular schedules. (See page 17 for new, FDA-approved meals that can be eaten with indinavir.)

Some possible combinations present such a logistical obstacle that physicians tend to avoid them. One example many cite is indinavir plus ddI, both of which require an empty stomach, but which must be taken at least an hour apart. Cafaro calls this "the nightmare combination if you have any semblance of a normal life."

Both Newman and Cafaro say they work closely with their patients to develop strategies for keeping track of when to take medications. "You try to simplify as much as possible," Cafaro says. That involves things like making sure people know which drugs can be taken together so that they can, for example, get into a routine of taking a set batch of pills together with breakfast every day. Cafaro notes, "that way they can package them in little baggies or pill containers" for easy access and organization. She adds, "I've got people actually going to the hardware store or sporting goods store and getting big, compartmented things like fishing tackle boxes," to organize their drugs. Whatever method is chosen, the idea is to organize pill-taking into a pattern that synchronizes reasonably well with the rest of a person's life.

It's also important that a doctor establish an atmosphere in which their patients feel relaxed enough to talk about their concerns and to acknowledge if a regimen is giving them trouble. "I think it involves getting to know someone and making sure they feel very comfortable," Newman says. "The last thing I want someone to do is feel shame about [not taking their medicine as directed]."

This process requires more effort than many doctors are accustomed to putting into their relationships with patients, but Newman has no doubt that it is necessary, since patients are being asked to do a tremendous amount as well. "I think the medicines are terribly unforgiving," she observes, and living with them is "a full-time job."

Although many AIDS specialists are now in the habit of working closely with people with HIV in choosing and monitoring their therapies, some physicians — especially those who do not do much HIV/AIDS care — are not accustomed to this approach. When in doubt, specialists agree, people should ask plenty of questions and make sure they understand what their doctor is suggesting and what is expected of them. Newman says, "Make sure before you leave the office that you have a very clear understanding of how to take your medicines, what to do if you have side effects, how to reach your provider, what you might anticipate. And if they don't tell you [whether to take your medications] with or without food, that should be a red signal that there's a huge problem." As Cafaro observes, "There is no such thing as a stupid question. There really isn't."

The bottom line, then, is: when in doubt, ask. In HIV therapy, it is better to slow down the process a bit if that's what it takes to make sure the choices a person with HIV and his or her doctor make are ones the person will be able to live with.

Bruce Mirken is a freelance journalist.

References

Bangsberg D and others. Protease inhibitors in the homeless. Journal of the American Medical Association 287(1): 63-65. July 2, 1997.

Berman S and others. The effectiveness of an educational program to improve patient understanding of and compliance with drug therapies used to treat HIV disease. IX International Conference on AIDS. Berlin, Germany. June 1993. Abstract PO-B30-2206.

Bihari B. Treatment of HIV infection with indinavir and nevirapine: an observational study of 45 patients. Being Alive. May 1997.

Broers B and others. A cohort study of drug users' compliance with zidovudine treatment. Archives of Internal Medicine 154: 1121-1127. May 23, 1994.

Eldred L and others. Medication adherence to long-term therapy in HIV disease. XI International Conference on AIDS. Vancouver, BC. July 1996. Abstract Mo.B.1165.

Eraker S and others. Understanding and improving patient compliance. Annals of Internal Medicine 100: 258-268. February 1984.

Ferrando S and others. Psychiatric morbidity correlates with poor compliance with zidovudine. American Journal of Drug and Alcohol Abuse 22: 475-487. November 1996.

Huengsberg M and others. Failure of prophylaxis for Pneumocystis carinii pneumonia. VIII International Conference on AIDS. July 1992. Abstract PuB 7256.

Molla A and others. Ordered accumulation of mutations in HIV protease confers resistance to ritonavir. Nature Medicine 2: 760-766. July 1996.

Morse E and others. Determinants of subject compliance within an experimental anti-HIV drug protocol. Social Science and Medicine 32: 1161-1167. 1991.

Polomeni P and others. Patients' compliance with prescriptions: a healthcare quality indicator in the treatment of HIV+ patients. XI International Conference on AIDS. Amsterdam, The Netherlands. July 1996. Abstract Tu.B.2338.

Singh N and others. Determinants of compliance with antiretroviral therapy in patients with human immunodeficiency virus: prospective assessment with implications for enhancing compliance. AIDS Care 8: 262-269. 1996.

Vanhove G and others. Patient compliance and drug failure in protease inhibitor monotherapy (letter). Journal of the American Medical Association 276: 1955-1956. December 25, 1996.