The incidence of nonadherence after organ transplant: Ensuring that our efforts at counting really count
Year: 2006
Type: editorial
Abstract: Much of the potential for optimal long-term outcomes in adult organ transplantation lies in the patients' hands: adherence to the multifaceted posttransplantation medical regimen has repeatedly been shown to be an important determinant of morbidity and mortality.1-3 However, despite continued growth in research in this area, most reviews of this literature conclude that we have but a poor understanding of either the incidence of posttransplantation nonadherence or its risk factors. Why have we failed to make better progress? We suggest that it is due to a combination of factors. These include, first, a failure to thoroughly consider adherence across the full range of components of the posttransplantation medical regimen. Second, in most studies we have failed to report nonadherence data using useful metrics. Third, there have been few attempts at quantitative cumulation or synthesis of results across studies, and this omission has led—we believe—to an inability to take stock of what is known and what issues should be pursued next. We briefly consider each of these limitations to our progress, with particular reference to empirical studies of adherence to the medical regimen after liver transplantation in adults. Liver recipients must engage in a number of behaviors in order to maximize their health posttransplantation, including taking immunosuppressant and other medications, attending routine clinic appointments, having blood work completed, monitoring vital signs, maintaining a diet and a level of physical activity that reduce the risk of obesity, and refraining from alcohol use, smoking, and use of illicit substances. Nonadherence in each of these areas, defined as the degree to which behavior fails to correspond with agreed recommendations from a healthcare provider,4 is no less important after liver transplantation than it is in other types of solid organ transplantation. However, in contrast to other types of organ transplant, the bulk of studies in liver transplantation have focused on alcohol use. As shown in Figure 1, of the empirical studies published from 1981 (the advent of cyclosporine use) through 2005, 18 examined alcohol consumption, less than half that number considered immunosuppressant adherence, and few to none have considered a variety of other health behaviors. We identified these reports during an exhaustive literature search we undertook as part of a meta-analysis we are preparing on adherence to the medical regimen after all types of adult organ transplantation. We included all reports that provided information on the number of respondents who were considered nonadherent by the investigators' definition. Figure 1 is based on only those studies that consider the general population of liver recipients. Thus, for the present discussion we excluded 39 additional reports that present data only for special subgroups of liver recipients (either patients transplanted for alcoholic liver disease [n = 37] or for heroin abuse [n = 1]). (These excluded reports focus almost exclusively on adherence to alcohol and substance use proscriptions.) Figure 1 shows, for comparison purposes, the distribution of studies examining adherence after heart transplantation. These studies constitute an interesting comparison because heart transplantation became an accepted treatment during the same era as liver transplantation, and it has similar short- and long-term survival rates. Figure 1 shows that studies of heart recipients have considered adherence relatively more evenly across a wider range of areas of the posttransplantation regimen. Numbers of studies examining nonadherence to each component of the medical regimen after liver or heart transplantation. *Includes studies of unselected populations; additional studies of only subgroups of transplant recipients (e.g., those transplanted for alcoholic liver disease) are excluded. In noting the emphasis on alcohol use in the liver transplant literature on adherence, we do not mean to imply that alcohol use is not a uniquely important issue in liver recipients. In fact, we ourselves have devoted considerable attention to alcohol and other substance use in these patients.5, 6 However, the relative dearth of studies of other adherence outcomes in liver recipients is noteworthy. Thus, until the report by O'Carroll et al.7 (this issue), no studies of the general population of liver recipients had considered adherence to follow-up clinic appointments, despite the clear value of follow-up clinical care for monitoring patients' health status and for the early identification of major posttransplantation complications, including cytomegalovirus infection and malignancies.8, 9 Efforts to improve screening and detection of critical morbidities posttransplantation will ultimately fail if patients return to the clinic for follow-up care at insufficient rates to ensure that screening and detection programs can be consistently applied. O'Carroll et al.7 consider immunosuppressant nonadherence as well, and this represents another pivotal yet understudied area of liver recipients' behavior. They carefully consider the methodologic difficulties of defining and assessing nonadherence in this area. In general, their decision to focus on 2 relatively neglected areas of liver transplant recipients' medical regimen should open the door for continued work examining these and other understudied areas of patients' behavior and health outcomes. Until greater attention is directed to such areas, we will be unable to form a complete picture of the nature and distribution of patients' difficulties in adhering to the full range of behaviors and activities expected of them after receiving a liver transplant. In the absence of a complete picture of the problem, interventions mounted to improve adherence may suffer from being too piecemeal—they may serve only as Band-Aids instead of generalized strategies to promote optimal patient behavior and health-promoting activities. Investigator A reports that, among 250 liver recipients transplanted at his center, 73% are alive. Investigator B reports that in her cohort of 500, 93% are alive. Investigator C reports that 50% of his sample of 100 are alive. What should be concluded from these disparate rates? Is it that the survival rate at Investigator C's center is particularly poor? No, your likely response is that these rates are uninterpretable in the absence of information regarding the duration of follow-up since transplant in each cohort. In fact, the rates would immediately make sense (and conform to U.S. national data10) if we learned that the Investigator B is reporting actuarial survival for the first year posttransplantation (93%), A is reporting a 5-yr survival rate (73%), and C is reporting an 11-yr rate (50%). The data now have meaning and clinical utility. The majority of studies of nonadherence after liver and other types of transplantation report simple rates similar to those introduced in the example above, with no adjustment for follow-up duration. Thus, typically reported results note only that a certain percentage of a sample of n patients were nonadherent in a given area. It is equally common for reviews of this literature to state, as O'Carroll et al.7 note, that posttransplantation nonadherence rates vary from 20 to 50%.11 Both assertions may be technically accurate—the number of nonadherent cases was correctly counted but the resulting rates are clinically and practically uninterpretable. They cannot be interpreted because the period of observation during which nonadherent individuals were identified was not taken into account in the computation of the rates. And the observation period matters: if some individuals in a cohort have been followed for only a brief period, they have not—in a very real sense—had the same opportunity to become nonadherent as other individuals in the cohort who were followed for much longer. Unequal follow-up is the rule rather than the exception in studies of transplant cohorts, in which individuals typically are observed for differing durations because they either entered the study at different time points (e.g., different dates of transplant), or they were lost to follow-up at varying time points (e.g., due to death, transfer to a different center). This issue of unequal follow-up is equally relevant to comparisons between studies that themselves differ in total duration of observation and that therefore are difficult to directly compare. Fortunately, there are well-known, readily available methods to address the situation of unequal follow-up.12, 13 Unfortunately, these methods are applied in only a minority of investigations of nonadherence after transplantation. For example, of the investigations summarized in Figure 1, only 36% of liver studies and 26% of heart studies presented rates adjusted for duration of follow-up (or, alternatively, studied samples consisting of individuals who had all been followed for the same duration of time). We suggest that, in the same way that it is already unacceptable to publish other outcome data (e.g., survival rates, complication rates) without regard to duration of follow-up, the literature on posttransplantation adherence must adopt higher standards as well. If we are to interpret data on the magnitude of nonadherence after liver and other types of transplantation, it is imperative that we report meaningful metrics that explicitly incorporate information on duration of follow-up and that thereby facilitate comparison of findings across studies. Given the difficulties noted above, it is perhaps not surprising that there have been few attempts at quantitative cumulation or synthesis of the adherence literature in transplantation, and in liver transplantation in particular.14 Should such attempts utilize only the few studies that present nonadherence results adjusted for duration of follow-up? Can nothing be learned from the many remaining studies which, taken together, include over 5500 liver recipients and would encompass the work of O'Carroll et al.7? Although the results of any such quantitative review would be the most accurate if based on rates that were appropriately calculated by the original investigators, it is possible to retrospectively adjust the rates if sufficient descriptive data are included in the original reports in order to estimate the total person-years of observation. Person-years are then used as the denominator for the computation of the nonadherence incidence rate in each sample.12 Perhaps the greatest limitation of calculating risk for nonadherence based on a person-years approach is the necessary assumption that the risk per unit of time is constant during the period of observation for the individual transplant recipient. However, the overall effect of this limitation is modest when based upon studies that have average durations of follow-up of several years or more, and when the goal is to describe a total rate of nonadherence over the entire follow-up rather than consider rate differences during narrow time periods.12 And, fortunately, the average duration of follow-up in adherence studies of liver recipients that report this information is almost 3 yr, ranging up to almost 18 yr. Indeed, the person-years approach has been adopted as one of several strategies for the reporting of information on outcomes after transplantation in U.S. transplant recipients.10 We illustrate in Figure 2 the results of applying this approach to the data from published reports of nonadherence in liver recipients. For example, among the 16 studies of alcohol use that provided sufficient data (out of a total of 18 published reports), the rates of nonadherence range from 0 cases per 100 person years of observation to 17.2 cases, with a median of 5.3 cases. Although some adherence areas have been examined in very few studies, it appears that all types of substance use, as well as nonadherence to immunosuppressants show lower incidence rates posttransplantation than nonadherence to other types of medications and nonadherence to exercise regimens. (No studies provide usable data on adherence to clinic appointments, diet, or any other area.) Given the importance of immunosuppressants for graft function, it is encouraging that 5 of 7 studies in this area of the regimen find extremely low nonadherence rates. Rates of nonadherence to components of the medical regimen after liver transplantation, published reports from 1981 to 2005. *Rate = number of cases per 100 person-years of observation. Each point represents an independent study, each bar indicates the median across studies in a given area. For alcohol, tobacco, and illicit substance use, data were extracted from each report on presence of any (vs. no) use. For all other areas, data were extracted on number of respondents nonadherent by each study's definition of what constituted nonadherent behavior. In sum, the value of our approach is that it allows for almost all published studies to contribute toward the estimation of the rates of nonadherence, as long as they provided descriptive information about the duration of follow-up (and few studies fail to provide any such data). In addition, it yields rates that are all on the same scale and thus can be more easily compared. The findings in Figure 2 constitute a bare beginning for a full cumulation and synthesis of the nonadherence literature in liver transplantation, and in adult organ transplantation more generally. Thus, while we have emphasized the need for adjustment of rates by duration of follow-up in order to promote comparisons across investigations, there are clearly other issues that also require attention and can begin to be addressed through a quantitative review. For example, what role do differences in the methodologies used to assess adherence play in influencing the rates plotted in Figure 2? Do biologic measures lead to higher or lower nonadherence rates than self-report and other clinical assessments? In addition, what demographic and psychosocial factors are reliably associated with risk of nonadherence to each component of the regimen? We are beginning to examine these issues in our meta-analysis. But any synthetic review will be only as good as the data on which it is based. In the same way that we urge that higher standards be adopted in the reporting of nonadherence rates, we recommend that the range of correlates examined in future studies be expanded beyond standard demographic variables to more routinely encompass both intrapersonal (e.g., beliefs, coping strategies) and interpersonal factors (e.g., social supports) that are likely to affect one's ability to adhere to the complex posttransplantation regimen. O'Carroll et al.'s7 findings in their small pilot sample suggest that patients' beliefs about their health and about the regimen, as well as their views of the transplant experience were important correlates of immunosuppressant medication adherence. It will only be through examination of similarly rich arrays of potential correlates in additional studies, and the ultimate comparing and combining of these effects across studies, that we will arrive at a more complete understanding of the importance of psychosocial correlates of adherence. This, in turn, will provide the bedrock needed to develop interventions with the real potential to maximize patients' adherence and ultimately their health after transplant.
Source: Liver Transplantation
DOI: 10.1002/lt.20854
Authors: Mary Amanda Dew, Andrea F. DiMartini
Language: English
Subfield: Family Practice
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W1974077743