Directly observed therapy (DOT) for human immunodeficiency virus (HIV) infection is commonly used in correctional settings; however, the efficacy of DOT for treating HIV infection has not been determined. We prospectively assessed adherence to antiretroviral therapy regimens among 31 HIV-infected prison inmates who were receiving ≤1 antiretrovirals via DOT. Adherence was measured by self-report, pill count, electronic monitoring caps, and, for DOT only, medication administration records. Overall, median adherence was 90%, as measured by pill count; 86%, by electronic monitoring caps; and 100%, by self-report. Adherence, as measured by electronic monitoring caps, was >90% in 32% of the subjects. In 91% of cases, adherence, as measured by medication administration records, was greater than that recorded by electronic monitoring caps for the same medications administered by DOT. Objective methods of measurement revealed that adherence to antiretroviral regimens administered wholly or in part by DOT was ≥90% in more than one-half of the patients. Different methods used to measure adherence revealed significantly different levels of adherence. These findings suggest that use of DOT does not ensure adherence to antiretroviral therapy.
Sustaining adherence to complicated antiretroviral regimens is a major challenge faced by HIV-infected persons for maintaining suppression of viral replication. Given the levels of adherence (>90%–95% adherence to the prescribed regimen) required to maintain a therapeutic response [1–8], several groups of investigators have proposed that antiretovirals be administered by directly observed therapy (DOT) to promote adherence [9–12]. DOT has been widely used to treat tuberculosis and is credited with preventing the emergence of drug resistance [13, 14]. However, use of DOT to treat HIV infection has not been well studied and has been largely limited to correctional settings [11, 12]. To evaluate adherence to antiretroviral therapy regimens in a state prison system where some or all of the antiretroviral agents are provided via DOT, we examined a cohort of HIV-infected prison inmates.
Subjects and Methods
Study site. This study was conducted at the North Carolina Department of Corrections (NCDOC). The NCDOC houses adult inmates at 74 separate prison facilities across the state. NCDOC policy dictates that all HIV protease inhibitors be administered by DOT. Other antiretroviral agents can be prescribed to be administered by DOT or by self-administered therapy (SAT), according to the preference of the clinician. To receive medications via DOT, patients must visit a medication window at specified times and swallow the medication in the sight of a nurse or a correctional officer. All medications are dispensed either to the patient, in the case of SAT, or to the prison medical staff, in the case of DOT, in 30-day allotments. Pill bottles are returned at the end of the 30-day period, and the prescription is refilled.
Subjects. From September 1999 through April 2000, 51 consecutive HIV-infected inmates housed at the 6 NCDOC prison facilities with the largest populations of HIV-infected inmates were asked during a routine appointment at the prison's infectious diseases clinic to participate in this study. Of the 6 prison facilities from which subjects were enrolled, 2 had full-time nursing staff available, whereas, at the other 4, nurses were available only on a part-time basis. Inmates were eligible if they had documented HIV infection, were not expected to be released during the 4 months following study entry, and were receiving ≤3 antiretrovirals, at least 1 of which was being administered via DOT. All subjects had been receiving their baseline antiretroviral regimen for at least 3 months prior to study entry.
Adherence monitoring. For each subject, adherence to their antiretroviral therapy regimen was assessed by electronic monitoring caps (eDEM; Aardex), pill counts of returned medication, and self-report . In addition, medication administration records completed by prison staff who dispensed medications for DOT were examined. For each method of measurement, adherence was calculated as the proportion of prescribed doses taken. At the time of study entry, the study pharmacist was notified of the subjects' enrollment and, at the next prescription refill, fitted the subject's SAT and DOT antiretroviral medication bottles with an electronic monitoring cap, which records the time and the date the medication bottle is opened. Subjects and prison staff were informed of the purpose of the caps at study entry and were instructed to open the bottles only when taking or dispensing the medication. The electronic monitoring caps were used for 1 cycle (typical duration, 30 days) of dispensed and returned antiretrovirals. Pill counts of returned antiretroviral medication bottles were conducted at the time the electronic monitoring caps were recovered. Medication administration records completed during the time in which the electronic monitoring caps were in place were collected and analyzed.
Because of security concerns, transportation of inmates from their facilities to a central location for research purposes was not possible. Therefore, to obtain self-reported information regarding adherence to antiretroviral therapy, subjects were interviewed at the first clinical appointment after study entry (i.e., ∼3–4 months later). Self-reported adherence was evaluated on the basis of the response to the question, “Many people don't take their medication perfectly all the time. Over the past 7 days, how many times did you miss a dose of [this medication]?” Attitudes toward DOT were assessed on the basis of the response to the question, “Would you rather keep your HIV meds [medications] in your locker or have prison staff give them to you every day?” In addition, subjects were asked, “If you took your HIV medicine yourself rather than having them handed to you every day by a correction officer or a facility nurse, do you think you would miss more doses, take more doses, or take the same doses of your HIV medications?”
Informed consent. All subjects provided informed consent to participate in this study. Consent was obtained in a private area of the clinic by study personnel who were not employees of the NCDOC. Subjects were informed that participation in this study would not impact their health care, the length of their sentence, or their eligibility for parole. The consent document and the study protocol were approved by the institutional review committees of the University of North Carolina School of Medicine (Chapel Hill, NC) and of the NCDOC.
Statistical methods. The distribution (means, medians, and percentages) of patient demographic characteristics, clinical factors, and medication regimens were evaluated. Univariate statistics of the full distribution for each method used to measure adherence (i.e., self-report, pill count, electronic monitoring caps, and medication administration records) were calculated. These marginal distributions were analyzed overall together and stratified by DOT and by SAT. In consideration of the size of the study population, bivariate comparisons between the different methods used to measure adherence were conducted with a 2-group nonparametric test (the Kruskal-Wallis test). Paired data of 2 × 2 tables were analyzed with McNemar's test, and paired data with continuous measurements were analyzed with a paired t test. Statistical analyses were performed with SAS software, version 6.12 (SAS Institute).
A total of 51 consecutive patients were screened, of which 41 (80%) agreed to participate. Demographic characteristics of the patients who declined to participate in the study were not different from those who provided consent (data not shown). Of the 41 subjects enrolled, 8 subjects discontinued DOT with antiretrovirals soon after study entry, and 2 transferred from prison during the study. These 10 subjects were excluded from the main analysis because their level of adherence was not measurable. The baseline characteristics of the remaining 31 subjects are listed in table 1. For 18 (58%) of the subjects, all antiretroviral agents were prescribed to be administered by DOT; for 12 subjects (39%), only the protease inhibitor was prescribed to be administered by DOT; and for 1 subject (3%), a nonnucleoside reverse-transcriptase inhibitor alone was prescribed to be administered by DOT.
Adherence monitoring. Median adherence to antiretroviral regimens for all subjects was 86%, as recorded by electronic monitoring caps; 90%, by pill count; and 100%, by self-report (table 2). Adherence to antiretrovirals, as measured by electronic monitoring caps, was >90% in only 32% of the subjects.
The extent of adherence to prescribed administrations of individual agents in a regimen did not appear to depend on the method of administration (SAT vs. DOT), although there was variability between the different measurement techniques (table 2). Median adherence to SAT-administered antiretrovirals was 92%, as measured by electronic monitoring caps; 90%, by pill count; and 100%, by self -report. Median adherence to DOT was 82%, as measured by electronic monitoring caps; 89%, by pill count; and 100%, by self-report. According to the medication administration records, median adherence to DOT-based regimens was 97%. For 91% of the monitored antiretroviral regimens, adherence as measured by medication administration records was greater than adherence as measured by electronic monitoring caps for the same medications administered by DOT.
Thirteen subjects (40%) received only 1 of their medications by DOT, and they received the remaining medications by SAT. For these subjects, there were no differences between adherence to DOT- and SAT-provided antiretrovirals, as measured by electronic monitoring caps (P = .7), pill count (P = .8), or self-report (P = 1.0). The level of adherence to DOT- and SAT-provided antiretrovirals was not significantly different, and, on the majority (74%) of the days that doses of any antiretroviral were missed, both DOT-based and SAT-provided medications were not taken.
Attitudes regarding DOT. Sixty-eight percent of the subjects responded that they would prefer to take medications on their own, rather than having them provided via DOT. When asked to envision how replacement of their DOT-provided antiretrovirals with SAT-provided antiretrovirals would impact adherence to their own medication regimen, only 8% of subjects expected SAT to lead to an increase in the number of missed doses.
To our knowledge, this is the first attempt to assess adherence to DOT-provided medications with electronic monitoring and the first electronic monitoring of therapy adherence in a prison. Compared with results of studies of adherence in nonincarcerated populations, adherence in our study group was relatively high. Although these findings are encouraging, the median level of adherence recorded in this study with the most objective methods of measurement is less than the range of levels (i.e., >90%–95%) required to maintain virological suppression during the long term .
These results raise questions about the effectiveness of DOT for treating HIV infection in a correctional setting. The median level of adherence to antiretroviral regimens in which ≤1 medication was administered by DOT was suboptimal, as recorded by the most objective adherence measurements; more than one-half of the subjects had adherence rates that would not be expected to provide lasting viral suppression.
In addition, the majority of subjects reported they would rather not receive HIV-related medications by DOT. Subjects must frequently stand in line to receive DOT, which thereby compromises confidentiality. In addition, when medical staff is unavailable, correctional officers administer medications to inmates, although in study interviews, inmates indicated a preference to receive medications only from medical personnel (data not shown).
Examination of data on adherence to SAT-based and DOT-based administration of antiretrovirals in the same subjects revealed no significant differences. Among the patients who received some of their antiretrovirals by DOT and the remainder by SAT, most of the occasions when doses were missed, both DOT- and SAT-based medications were not taken. This may reflect a conscientious effort on the part of the patient to minimize the risk of antiretroviral resistance by not taking a medication administered by SAT after missing one administered by DOT, or, more likely, it suggests that factors other than the method of administration of antiretrovirals impacted overall adherence levels.
It was unexpected that medication administration records would show a significantly higher level of adherence than did electronic monitoring caps and pill counts. This discrepancy may reflect the well-described limitations of electronic monitoring devices. However, prison staff were instructed to open bottles fitted with electronic monitoring caps only when dispensing medications and to dispense medications only from the patient's medication bottle. Furthermore, the pill counts also suggest that medication administration records overreport adherence. This observation reinforces data suggesting that, regardless of the method of drug administration, adherence is most accurately assessed by multiple methods of measurement .
There are several limitations to this study. The foremost limitation is that the method of drug administration is largely determined by the antiretroviral class, because administration of protease inhibitors by DOT is policy of the NCDOC. Therefore, because protease inhibitors are more likely to be administered via DOT, we cannot readily determine whether the comparable levels of adherence we observed between administration of antimicrobials by DOT and by SAT was due to difficulties associated with adherence to protease inhibitor therapy. However, the finding that, on the majority of days a dose of antiretrovirals was missed, both medications administered by SAT and those administered by DOT were not taken argues against a significant class effect. Accurate assessment of adherence is difficult, and each measurement technique has limitations. We employed several methods for measuring adherence to reduce the bias particular to any individual method. Obtaining self-reported adherence data was difficult because of the conditions under which the study was conducted. Access to subjects between clinical visits was not possible. Therefore, the survey of self-reported adherence required subjects to report adherence to their antiretroviral regimen during the time immediately prior to the final study visit, which may have been several weeks later than the time period during which adherence was recorded by the other methods used to measure adherence. Finally, adherence monitoring in this study was limited to one 30-day period. Larger or smaller differences in results of the adherence measurements may emerge over longer periods.
Although the efficacy of DOT-based administration of antiretroviral therapy has not been established, the past several years have seen the widespread introduction of this expensive and difficult-to-maintain intervention in many prison systems. Overall, our findings suggest that adherence to antiretroviral therapy among inmates receiving a portion or all of their medications via DOT is less than the range of levels that has been associated with durable suppression of HIV replication. Given the concern that the application of DOT is resource intensive and may lead to a loss of confidentiality for the infected inmate, careful evaluations of this adherence intervention are urgently needed.