Table 3.
Reported Associations with Pharmacy-Based Adherence Measures (PAMs) in High-Income Countries
| Study (year) | Design | Type of care | Region | ART naive (%) | ART regimen (%)a | PAM category | PAM descriptionin study | PAM duration, monthsb | Sample size | Key findingsc |
| Maher et al [42] (1999) | Retrospective cohort | VA, minor costs | USA | No (26) | PI (100) | PPU | Adherence' occurred if the patient consistently filled 4 prescriptions on time “nonadherent did not do this) | 4 (varied) | 205 | Adherent status predicted VL suppression (P < .01) and CD4 cell count increase (P < .01), whereas ”nonadherent” status predicted VL suppression (P < .05) but not CD4 increase over 5–9 months of follow-up |
| Singh et al [46] (1999) | Prospective cohort | VA, private | USA | No (7) | NR | PPU | Adherence occurred if refills picked-up/refills prescribed was >90% | 6 (varied) | 123 | Adherence predicted greater change in the CD4 cell count (P < .05) |
| Descamps et al [33] (2000) | Case-control study in an RCT | Free | France | Yes | PI, 2NRTIf | PC | (Pills prescribed – remnant pills)/pills to cover the interval | 6 (0–6) | 116 | Mean PAM for zidovudine and indinavir predicted VL rebounde (P < .05) |
| Low-Beer et al [41] (2000) | Retrospective cohort | Public | Canada | Yes | NNRTI, PI (NR) | MPR | Months ART prescribed/months follow-up in 1st year | 12 (0–12) | 886 | Increasing PAM strata (<70%,70%-80%, 80%-90%, 90%-95%, and 95%-100%) predicted VL suppression during follow-up (median duration of follow-up, 19 months; P < .01) |
| Liu et al [40] (2001) | Prospective cohort | Private | USA | Yes | NNRTI, PI (NR) | PC | 1 – [(actual pills – expected pills)/pills per dose/prescribed doses for the period ] | 2 (0–2) 6 (0–6) |
108 | 1. Increasing PAM strata predicted VL suppression at 2 and 6 months (P < .01) 2. no difference was shown between PC and MEMS at predicting VL at 2 and 6 months, but both were superior to self-reported adherence at 2 months (P < .01) |
| McNabb et al [43] (2001) | Prospective cohort | Private | USA | No | PI (63), NNRTI, 2NRTI | PC | (1) doses taken/doses prescribed, OR if return after 30 days, then(2) doses taken/doses required for interval | 3 (varied) | 40 | PAM and self-report changes were not associated with VL change, whereas increasing MEMS adherence was associated with decreasing VL (P < .05) and VL suppression (P < .01) |
| Hogg et al [38] (2002) | Retrospective cohort | Public | Canada | Yes | PI (73), NNRTI | MPR | Months ART prescribed/months follow-up in first year | 12 (0–12) | 1282 | PAM adherence of <75% predicted mortality, or mortality plus new AIDS diagnosis, over maximum follow-up of 50 months (P < .01) |
| Alcoba et al [30] (2003) | Retrospective cohort | NR | Spain | No | PI (100) | MPR | Patient was “ nonadherent” if (days in the interval – days dispensed)/days in the interval is >10% | 3 (varied) | 106 | “Nonadherent” status, self-report, and ARV plasma levels were not associated with VL |
| Wood et al [48] (2003) | Retrospective cohort | Public | Canada | Yes | NNRTI, PI (NR) | MPR | Months ART prescribed/months follow-up in 1st year | 12 (0–12) | 1422 | PAM adherence of >95% predicted time to VL suppression and time to VL rebounde over follow-up, which was NR but variable and maximum of 67 months (P < .01) |
| Grossberg et al [37] (2004) | Retrospective cohort | VA, minor costs | USA | No(35) | NNRTI, PI, 3NRTI (NR) | MPR | (Total pills/daily number of pills)/days between refills | 3 (varied) | 110 | Self-reported adherence and increasing PAM strata predicted VL reductions (P < .01), apart from self-report in ART naive |
| Kitahata et al [39] (2004) | Retrospective cohort | Free ART | USA | Yes | PI (78), NNRTI | MPR | Mean for all ARVs of [(1 – days without ARV)/days in the interval] | 6 (0–6) | 212 | 1. Increasing PAM strata (<70%, 70%-90%, and >90%) predicted viral reboundh (P < .01) and higher CD4 cell counts over 12–24 months (P < .05)2. PAM adherence <70% predicted new AIDS or death, compared with PAM adherence of >70%, over 24 months (P < .01) (but PAM adherence of 70%-90%, compared with >90%, did not) |
| Wood et al [47] (2004) | Retrospective cohort | Public | Canada | Yes | PI (69), NNRTI | MPR | Months ART prescribed/months follow-up in 1st year | 12 (0–12) | 1522 | PAM adherence <75% predicted a lower increase in the CD4 cell count over 24 months (P < .01), whereas PAM strata >75% increasingly predicted increases in the CD4 cell count over 24 months (P < .01) |
| Fairley et al [34] (2005) | Retrospective cohort | Public | Australia | No | NNRTI, PI (NR) | MPR | Days ART prescribed/days in the interval | Variable range, 12-44 | 752 | Increasing PAM and self-reported adherence predicted VL suppression (P < .01) |
| Fletcher et al [35] (2005) | RCT (Prior VF on PI regimen) | Free ART | USA | No | NNRTI + PI (100) | PC | (doses dispensed – doses returned)/doses dispensed | 1(0–1) | 220 | PAM did not predict VL changes at 4 months; self-reported adherence (P < .05) (n = 244) and ARV plasma levels (P < .05) (n = 180) predicted VL changes at 4 months, wheras MEMS did not (n = 62) |
| Harrigan et al [10] (2005) | Retrospective cohort | Public | Canada | Yes | PI (74), NNRTI | MPR | Months ART prescribed/months follow-up in 1st year | 12 (0–12) | 1191 | PAM adherence of 80%-90% is the highest predictor of single and multiple category HIVDR over 24 months, compared with PAM adherence of 0%-20% (P < .01) |
| King et al [28] (2005) | RCT | Free ART | Multi- continent | Yes | PI (100) | PC | Pills consumed/pills expected to be consumed | Variable range, 2 – 3 | 590 | Decreasing PAM strata increasingly predicted VFg (P < .01);the mean PAM adherence rate was lower in persons with detectable HIVDR to PI and/or 3TC (P < .01) |
| Inciardi et al [29] (2005) | Retrospective cohort | Private | USA | No | NNRTI (56), PI | MPR | Sum of (interval days – ARV days) for all ARVs/sum of interval days for all ARVsi | Variableh | 94 | Decreasing PAM adherence was associated with VL increase (P < .01) |
| Gross et al [36] (2006) | Retrospective cohort | Public | Canada | No | NNRTI, PI (NR) | MPR | (Days ART [any ARV] dispensed between 3 refills +30)/days between 3 refills | Variable range, 2-6 |
1634 | Decreasing PAM strata (<70%, 70%-95%, and >95%) in a 2-6 observed interval, predicted a higher proportion with VFd (P < .01) |
| Yes | NNRTI, PI (NR) | MPR | (Days ART [any ARV] dispensed +30)/fays between refills | Variable median, 29 |
1634 | PAM adherence <95% (treated as a time-varying variable, with or without a 30-day grace period) predicted viral rebounde over the period of observation (P < .05) | ||||
| Braithwaite et al [27] (2007) | Retrospective cohort | VA, minor costs | USA | Yes | PI (58), NNRTI, 3NRTI | MPR | Days ART prescribed/days in interval | 12 (0–12) | 6394 | Increasing PAM strata increasingly predicted VL change, VL suppression, or changes in the CD4 cell count at 12 months |
| Townsend et al [31] (2007) | Retrospective cohort | VA, minor costs | USA | No | PI(50), NNRTI, NRTI | MPR | Days ART prescribed/days in the interval | 6 (varied) | 58 | PAM was not associated with VL; PAM adherence <70% was associated with changes in the CD4 cell count (P < .05), but PAM adherence of 70%-90%, compared with >90%, was not |
| Saberi et al [45] (2008) | Retrospective cohort | Private | USA | No | NNRTI (100) | MPR | (Pills dispensed/pills prescribed per day)/days between refills | Variable range, 3-18 | 151 | PAM adherence >85% maintained VL suppression in 8 of 10 patients between 2 VL measurements |
| Lima et al [11] (2009) | Retrospective cohort | Public | Canada | Yes | PI (64), NNRTI | MPR | Days ART prescribed/days of follow-up | Variable maximum, 30 | 903 | PAM adherence <95% predicted mortality over follow-up period (maximum, 55 months) (P < .05) |
| Nellen et al [44] (2009) | Retro- and Prospective cohort | NR | Holland | No | NNRTI (58), PI, 3NRTI | PPU | ART dispensed/ART prescribed | 6 (varied) | 115 | PAM adherence <85% did not predict VFg (but did for an ART-naïve subgroup; P < .01) over 24 months; self-reported adherence and ARV plasma levels did not predict VFg over 24 months |
| Cambiano et al [32] (2010) | Retrospective cohort | Public | England | No | PI(47), NNRTI, NRTI | MPR | Days with ≥3 ARV prescriptions/study interval | 6 (varied) | 1632 | PAM strata <95% predicted (P < .01) viral rebound, but PAM adherence of 95%-99% did not, h over the subsequent 9 months, compared with PAM adherence of 100% |
NOTE. ART, antiretroviral therapy; ARV, antiretroviral; HIVDR, HIV drug resistance; MEMS medication event monitoring system; MPR, Medication possession ratio; NNRTI, nonnucleoside reverse-transcriptase inhibitor; NR, not reported; PC, pill count; PI, protease inhibitor; PPU, pill pick-up; RCT, randomized control trial; 3NRTI, triple nucleoside reverse-transcriptase inhibitor; 3TC, lamivudine; 2NRTI, double nucleoside reverse-transcriptase inhibitor; VA, veterans affairs hospital; VF, virological failure; VL, viral load.
a
Data are ART regimens for that study. The numbers in parentheses represent the percentages of subjects receiving the predominant regimen.
b
Duration of adherence assessment, with the months over which assessed in parentheses. If there was a variable duration of adherence assessment, than the median, mean, or range is listed.
c
The number after ”PAM” is the percentage adherence.
d
Two viral loads separated in time above threshold.
e
Two viral loads above threshold after previous VL suppression.
f
All patients received triple-drug PI regimens for 3 months and were then randomized to receive double-NRTI (50%) or 1 PI plus 1 NRTI (36%) or to continue the PI regimen (14%).
g
Single viral load above threshold.
h
Single viral load above threshold after previous VL suppression.
i
“Interval days” was the sum of multiple 3-month periods prior to VL tests performed over a 2-year period, and “ARV days” was the sum of ARVs prescribed over these same periods.