Weekend versus weekday adherence: Do we, or do we not, thank God it’s Friday?

Adherence monitoring and intervention studies do not show improvements in clinical outcomes: why, and what can be done about it? Boucquemont et al’s article is on page 274. Adherence monitoring and intervention studies do not show improvements in clinical outcomes: why, and what can be done about it? Boucquemont et al’s article is on page 274. “Not everything that counts can be counted; and not everything that can be counted, counts.”Albert Einstein. In “Association Between Day of the Week and Medication Adherence Among Adolescent and Young Adult Kidney Transplant Recipients,” Boucquemont et al1Boucquemont J, Pai AL, Dharnidharka VR, et al. Association between day of the week and medication adherence among adolescent and young adult kidney transplant recipients [published online ahead of print 2019]. Am J Transplant. 1–8. https://doi.org/10.1111/ajt.15590.Google Scholar describe an interesting and important finding from a post-hoc analysis of the TAKE-IT (ClinicalTrial.gov NCT01356277) study. The authors report that adherence to immunosuppressant medications, as measured by electronic monitoring, was consistently worse on weekends compared with weekdays, during both the run-in and the study periods, in both the intervention and control groups. It is seldom possible to measure adherence in such detail, and therefore those results, from a rigorous study that evaluated adherence over time, are not just welcome but unique. What are the implications to practice? What are the implications for adherence research? The authors conclude that clinicians should counsel patients about the importance of consistent medication-taking routines, so that adherence would be maintained in the same way during the entire week. This seems like a reasonable conclusion. But in our view, the findings point to a more important insight that becomes apparent when one looks at the results of the present analysis in the context of the primary results from the TAKE-IT study.2Foster BJ Pai A Zelikovsky N et al.A randomized trial of a multicomponent intervention to promote medication adherence: the teen adherence in kidney transplant effectiveness of intervention trial (TAKE-IT).Am J Kidney Dis. 2018; 72: 30-41Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar Electronic monitoring can detect day-by-day and even hour-by hour variations, but the authors explain that they are unable to tell whether such variations had any clinical implications. They note that many of the patients who were nonadherent over the weekend were also nonadherent during weekdays, making it practically impossible to “disentangle” the clinical effects, if any, of weekend-vs-weekday nonadherence. In the primary TAKE-IT report, the authors, similarly, were unable to detect any effects of the intervention on the examined clinical outcomes, even though it seemed to improve adherence as measured by electronic monitoring. TAKE-IT was a multisite randomized controlled trial of an adherence intervention (involving coaching) offered to adolescent and young adult kidney transplant recipients in 8 transplant centers in the United States and Canada. The primary adherence analysis was done – by necessity – only on patients who were using the electronic monitors as intended; although 81 subjects were originally allocated to the intervention, only 64 had usable electronic monitoring data. When those subjects were compared with the 74 control subjects who used the monitors, the authors found that the intervention significantly improved adherence. Although adherence was improved as shown by the electronic monitoring, the authors report that the study was not powered to examine any clinical outcomes, and indeed, they found no significant intervention effect on the health and health utilization outcomes that were examined, which included rejection rates, graft failure rates, estimated glomerular filtration rate, infection rate (with the exception of cytomegalovirus infection, which was significantly more prevalent in the intervention group), and rates of hospitalization. They also report no significant effect on the degree of fluctuation of medication blood levels. How is it possible to improve adherence while not improving medication levels or any clinical outcomes? One explanation, of course, could be that it does not matter whether patients do or do not take their immunosuppressants posttransplant. This, however, is very unlikely, given that existing literature identifies nonadherence as one of the most important risks for organ rejection and loss3Ettenger R Albrecht R Alloway R et al.Meeting report: FDA public meeting on patient-focused drug development and medication adherence in solid organ transplant patients.Am J Transplant. 2018; 18: 564-573Abstract Full Text Full Text PDF PubMed Scopus (29) Google Scholar and given that complete immunosuppressant withdrawal, even if done within the confines of a very selective population, is rarely feasible.4Shaked A DesMarais MR Kopetskie H et al.Outcomes of immunosuppression minimization and withdrawal early after liver transplantation.Am J Transplant. 2019; 19: 1397-1409Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar Therefore, immunosuppression is, in fact, essential, in the vast majority of cases, posttransplant. Another explanation, which the authors invoke, is that the reason for a lack of observed effect on clinical outcomes is because the study was not powered to detect such outcomes. We note that even if not powered to show an effect on any single outcome, such as rejection, a truly effective intervention could still show an impact on at least one of many outcomes that were examined and would be expected to show an effect on blood levels. Most importantly, although it is impossible to tell whether a lack of effect on clinical outcomes was due to a lack of power to detect such effects in any single study, the picture becomes clearer when it is put in the context of findings from other trials. Indeed, a recent literature review, which was written before the results of TAKE-IT were available but in a sense anticipated them,5Duncan S Annunziato RA Dunphy C LaPointe RD Shneider BL Shemesh E. A systematic review of immunosuppressant adherence interventions in transplant recipients: decoding the streetlight effect.Pediatr Transplant. 2018; 22: e13086Crossref Scopus (36) Google Scholar showed that of 21 randomized clinical trials of adherence-improving interventions in transplant medicine, none were able to show an effect on clinical outcomes, although almost all claimed that the intervention was effective in improving adherence. Investigators in any one study could reasonably conclude that a particular single trial was not powered to look at clinical outcomes, and therefore when those were examined there was no observable effect. But when put together, the consistent and persistent lack of effect on clinical outcomes across randomized controlled trials examining adherence interventions (which almost always report an improvement in adherence, as measured in those trials) is very unlikely to be because of a lack of power. It is expected that if adherence is truly improved across multiple trials and many subjects, by chance alone some effect on a few outcomes could be demonstrated, even if infrequently. It is telling that a lack of effect on clinical outcomes seemed to have been specifically related to the use of electronic monitors to measure adherence.5Duncan S Annunziato RA Dunphy C LaPointe RD Shneider BL Shemesh E. A systematic review of immunosuppressant adherence interventions in transplant recipients: decoding the streetlight effect.Pediatr Transplant. 2018; 22: e13086Crossref Scopus (36) Google Scholar Is it possible that the monitors look at adherence in a too-detailed way and that an “improvement” sometimes means very little? Stated differently: does it really matter if a patient takes the medication an hour later, or even misses a day, every once in a while? To answer this question, one must first consider the issue of selection bias. Selection bias is particularly important in adherence intervention research, because nonadherent patients, almost by definition, are less likely to participate in such research.6Shemesh E Mitchell J Neighbors K et al.Recruiting a representative sample in adherence research-The MALT multisite prospective cohort study experience.Pediatr Transplant. 2017; 21: e13067Crossref Scopus (15) Google Scholar Patients who do not take their immunosuppressants as prescribed to the point that their health is threatened (severe nonadherence) may well also suffer from substantial psychosocial adversity and morbidity, which makes it harder for them to come to clinic in the first place and makes them less likely to consent to research.6Shemesh E Mitchell J Neighbors K et al.Recruiting a representative sample in adherence research-The MALT multisite prospective cohort study experience.Pediatr Transplant. 2017; 21: e13067Crossref Scopus (15) Google Scholar Even if they do consent, they will be less likely to follow study procedures. In other words: patients who are nonadherent to their prescribed regimen are also likely to be nonadherent to the study requirements. Studies that use “convenience sampling” are likely to recruit a preponderance of patients who are quite adherent and, therefore, do not stand to benefit from the intervention. The use of electronic monitoring potentially worsens this bias, as patients who use those monitors are the most adherent in the group (patients who cannot take their medications consistently are also likely to find it hard to use those monitors consistently). This has been demonstrated in TAKE-IT: the standard deviation of medication blood levels in intervention group patients who did not use the monitors was higher than that of the patients who did use them (1.2 to 1.6 in those who used the monitors vs 2.8 in those who did not); hence, patients who used the monitors were more adherent to their immunosuppressant regimen.2Foster BJ Pai A Zelikovsky N et al.A randomized trial of a multicomponent intervention to promote medication adherence: the teen adherence in kidney transplant effectiveness of intervention trial (TAKE-IT).Am J Kidney Dis. 2018; 72: 30-41Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar Such selection bias influences the interpretation of study results in 2 important ways.5Duncan S Annunziato RA Dunphy C LaPointe RD Shneider BL Shemesh E. A systematic review of immunosuppressant adherence interventions in transplant recipients: decoding the streetlight effect.Pediatr Transplant. 2018; 22: e13086Crossref Scopus (36) Google Scholar First, for patients who are sufficiently adherent at baseline, further improvement in adherence, even if statistically significant, would not improve clinical outcomes, because their baseline adherence was good enough to begin with. Second, interventions that look like they might work with adherent patients may be misguided when applied to nonadherent patients. For example, an intervention that involves multiple in-person visits with a therapist may seem like a good idea with patients who are able to come to those scheduled visits but is not likely to be of much use to patients who rarely come to their follow-up visits (or come late, etc.). Thus, selection bias in adherence research leads not just to “loss of power” to show an effect of an intervention. It also could lead to the testing, and maybe even endorsement, of interventions that are inappropriate for the target population of severely nonadherent patients.5Duncan S Annunziato RA Dunphy C LaPointe RD Shneider BL Shemesh E. A systematic review of immunosuppressant adherence interventions in transplant recipients: decoding the streetlight effect.Pediatr Transplant. 2018; 22: e13086Crossref Scopus (36) Google Scholar Therefore, interventions that improve adherence in substantially adherent patients may never lead to any clinical improvements even if a “well-powered” study is conducted. Improving adherence is, in and of itself, not necessarily clinically meaningful: it depends on the extent to which real improvement happens in the target population (not in “the clinic as a whole”). A clinical improvement has to be proven in a clinical trial, not just assumed. Targeting adherence intervention efforts to adherent patients may even impart a real danger. The TAKE-IT study reported a significant increase in the incidence of cytomegalovirus infection in the intervention group compared with the controls. In those who are sufficiently adherent to avoid immune consequences like rejection, enhanced adherence (which would lead to even more immunosuppression) has the potential to tip the balance of immunosuppression toward increased infectious or other complications of immunosuppressant medications. Clinical trials of adherence interventions need to be monitored for this potential adverse effect. It is also conceivable that continuous “counseling” about adherence, when indiscriminately offered, could alienate some patients, especially those who do not really need this additional “education.” This issue was noted by a recent panel meeting, convened by the US Food and Drug Administration,3Ettenger R Albrecht R Alloway R et al.Meeting report: FDA public meeting on patient-focused drug development and medication adherence in solid organ transplant patients.Am J Transplant. 2018; 18: 564-573Abstract Full Text Full Text PDF PubMed Scopus (29) Google Scholar which was focused on patients’ perspectives and reported that “Several participants felt that healthcare providers were too overbearing about immunosuppression therapy adherence … often leading to poor communication between the medical team and the recipient.” Where does this leave us? The present analysis1Boucquemont J, Pai AL, Dharnidharka VR, et al. Association between day of the week and medication adherence among adolescent and young adult kidney transplant recipients [published online ahead of print 2019]. Am J Transplant. 1–8. https://doi.org/10.1111/ajt.15590.Google Scholar was unable to examine, let alone observe, any differences in clinical outcomes between patients who did versus who did not adhere to their regimen on weekends and found that most of the patients who were nonadherent over the weekend were also nonadherent on weekdays. This suggests, to us, that it is unclear whether the nuance of a day-to-day variation in adherence has a clinical implication. Furthermore, because the parent TAKE-IT study and other studies report that it was not clear whether improving adherence as verified by electronic monitoring improves any clinical outcomes,5Duncan S Annunziato RA Dunphy C LaPointe RD Shneider BL Shemesh E. A systematic review of immunosuppressant adherence interventions in transplant recipients: decoding the streetlight effect.Pediatr Transplant. 2018; 22: e13086Crossref Scopus (36) Google Scholar it appears that such monitors may sometimes represent a less than ideal way to measure adherence. In TAKE-IT, less adherent patients were less likely to use those monitors, even in a context in which a “run-in” phase ensured training. This is not specific to TAKE-IT—our group’s previous results were worse: in a pediatric population, we reported that less than half of the patients were able to use those monitors as directed.7Kerkar N Annunziato RA Foley L et al.Prospective analysis of nonadherence in autoimmune hepatitis: a common problem.J Pediatr Gastroenterol Nutr. 2006; 43: 629-634Crossref PubMed Scopus (85) Google Scholar In addition, when electronic monitors are used, their ability to provide “real-time” information and detect minor and detailed variations in adherence, by day or by the hour, could in fact be a shortcoming. In our view, an alternative interpretation of the TAKE-IT analyses could be that some deviations in adherence, as detected by such detailed monitoring, may at times be clinically unimportant. Focusing on minor deviations—in research or in practice—could lead to a misguided use of resources, could alienate patients and families, and perhaps could even lead to negative consequences of “overadherence.” Clinicians, in our view, would do well to ask whether any recommendation related to the monitoring and management of nonadherence (whether a proposed intervention to improve adherence or the suggestion to address “weekend nonadherence”) is associated with an effect on clinical outcomes, before the community endorses such ideas. It is incorrect to assume that improving adherence behavior to any extent, in any patient, is a worthy clinical goal. In research efforts, similarly, showing improvement in adherence should not be assumed to be the same as demonstrating a clinical benefit. Adherence intervention studies may be underpowered because of a small sample size, may have selection bias favoring a more adherent group of patients (who do not need the intervention), may be too short to show a clinical effect, or a combination of those. And, obviously, some interventions may be clinically ineffective. It is time to conduct robust, sufficiently long, and well-targeted studies that could (and, hopefully, would) show clinical improvement or prove that the intervention has no clinical merit. Such studies would presumably use strategies that are appropriate for seriously nonadherent patients rather than methods that are applicable primarily to highly adherent subjects. When results from the present analysis and from the main TAKE-IT study are contemplated in the aggregate, they move the field forward, as would be expected from such rigorous and cutting-edge research, in a non–too-obvious direction. We believe that those results suggest that in adherence research, if we focus too much on the details, we might lose sight of the entire picture. Nonadherence can be a severe, life-threatening condition; we believe that it should be detected using robust measures that look at patients’ pattern of behavior over time. Looking at too many details, or at “real-time” snapshots of that behavior, could at times be counterproductive—in some circumstances, it may “flag” the wrong set of patients and could lead clinicians to focus on trivial issues rather than on the few serious threats. We do not advocate ignoring weekend adherence. But some slack in weekend adherence may be acceptable if not quite desired for patients who are sufficiently adherent in general. Perhaps “thank God it’s Friday” is sometimes, under the right circumstances, the right sentiment, after all. The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.