Ó 2008 John Wiley & Sons A/S.
Pediatr Transplantation 2009: 13: 590–598
Pediatric Transplantation DOI: 10.1111/j.1399-3046.2008.01038.x
Multidimensional Adherence Classification System: Initial development with adolescent transplant recipients Simons LE, Gilleland J, Blount RL, Amaral S, Berg A, Mee LL. Multidimensional Adherence Classification System: Initial development with adolescent transplant recipients. Pediatr Transplantation 2009: 13: 590–598. Ó 2008 John Wiley & Sons A/S. Abstract: As transplantation has progressively become a more viable option for children with life-threatening illness, ensuring that adolescents do not lose their new organ secondary to medication non-adherence is paramount. The first step to addressing non-adherence is adequate assessment of this construct. In this investigation, we introduce the MACS. The MACS includes self-report and drug assay levels. Self-report is a subjective measure with a low false-positive rate, but is vulnerable to social desirability. Drug assays are an objective measure of drug ingestion, but values suggestive of non-adherence may be influenced by medical complications and timing. The MACS builds on the strengths of both methods and attempts to contain their weaknesses. The sample in this study consisted of 82 adolescent solid organ transplant recipients. The non-adherence rate using the MACS in this sample was 61%. Initial data to support this system are promising. The occurrence of rejection episodes and mortality were significantly related to membership in the Genuinely Non-adherent category. Beyond providing initial support for the MACS, we discuss the clinical implications of this adherence classification system.
Laura E. Simons1, Jordan Gilleland2, Ronald L. Blount2, Sandra Amaral3, Alexandra Berg4 and Laura L. Mee5 1
Department of Psychiatry, ChildrenÕs Hospital Boston & Harvard Medical School, Boston, MA, USA, 2 Department of Psychology, University of Georgia, Athens, GA, USA, 3Renal Transplant Program, Department of Pediatrics, ChildrenÕs Healthcare of Atlanta & Emory University School of Medicine, Atlanta, GA, USA, 4Cardiology, ChildrenÕs Healthcare of Atlanta, Atlanta, GA, USA, 5Transplant Program, Department of Pediatrics, ChildrenÕs Healthcare of Atlanta & Emory University School of Medicine, Atlanta, GA, USA Key words: pediatric transplant – medication adherence – clinical outcome – mortality Laura E. Simons, Pediatric Pain Rehabilitation Center, Children's Hospital Boston at Waltham, 9 Hope Avenue, Waltham, MA 02453, USA Tel.: +781 216 1650 Fax: +781 216 1652 E-mail:
[email protected] Accepted for publication 30 June 2008
For adolescents with solid organ transplants, non-adherence to immunosuppressant therapy is a life-threatening issue. Non-adherence often results in serious health consequences for teens post-transplant including rejection, infection, hospitalization, and mortality (1–8). Rates of non-adherence within the pediatric population fall between 5% and 50% (2, 9). The large discrepancy in the rates of non-adherence likely reflects a lack of standardization in the literature as to how to define this construct. To accurately examine the clinical consequences of non-adherent behavior, medical professionals and researchers must use reliable and valid measures to assess
Abbreviations: ESRD-SCL, End-Stage Renal Disease Symptom Checklist-Transplant Module; HSD, honestly significant differences; MACS, Multidimensional Adherence Classification System; MAM, Medication Module of the Medication Adherence Measure.
590
the extent to which an adolescent adheres to a prescribed medication regimen. There is no agreed upon gold standard for measuring adherence. Each method of assessment has unique strengths, as well as challenges which must be overcome to insure valid classification. The most common methods include: selfreport, pill counts, pharmacy records, electronic microprocessors, clinician assessment, clinical outcome, and drug or marker level (6, 10, 11). Self-report is easily obtained and it reflects patient behavior that occurs at home, school, or other non-medical settings. However, it relies on honesty from the respondent and often results in under-reported adherence difficulties. Selfreporting accuracy can be improved by keeping the recall period short and asking detailed objective questions (12). Pill counts and pharmacy refill records indicate that the medication is decreasing or being ordered, but do not provide information about when, or even if, the
Multidimensional Adherence Classification
adolescent is ingesting the medication. Electronic microprocessors indicate when a medication is retrieved from the bottle, but also fail to insure that medication is actually taken. Furthermore, approximately 30% of a recent sample of pediatric kidney transplant patients reported that using an electronic microprocessor was ‘‘a burden’’ and/or impeded medication transport (11). Clinician ratings of patient adherence are low in specificity and positive predictive power (13). Clinician ratings suffer from the same subjective vulnerabilities as self-report and have been found to be no more accurate than a coin toss (14). Adverse clinical outcomes are sometimes used to classify non-adherence, yet this method seems counterintuitive if one of the main purposes of measuring adherence is to identify those at-risk and prevent negative medical outcomes. With regards to immunosuppressant drug levels, the frequency of laboratory visits to assess immunosuppressant drug levels vary patient to patient, but typically occur every 1–3 months. To index adherence levels, a clinician obtains the standard deviations of consecutive blood levels of immunosuppressants, with a higher standard deviation indicating more immunosuppressant dose fluctuations over time and more erratic adherence (6). This method in isolation poses some challenge as some values are invalid (i.e., non-trough) and must be examined within the context of each patient before inclusion in this form of analysis. Further, the standard deviation of serum values taken over potentially long periods of time may be minimally sensitive to short-term changes in adherence behaviors. A recent study by SchaferKeller et al. (15) examined assessment of immunosuppressive therapy adherence in 249 adult kidney transplant recipients. The authors found that no single measure demonstrated both high sensitivity and specificity; however, combining measures increased diagnostic accuracy. This investigation will use a multidimensional approach that combines the strengths of subjective and objective methods for classifying adherence in pediatric patients who have received solid organ transplants. Patient self-reports and parentsÕ reports of the adolescentsÕ adherence behaviors, and serum immunosuppressant assays, will be combined to provide a four category classification of patient adherence. The support for this classification system will be assessed by examining patient clinical outcome data, including rejections, hospitalization, and mortality (16). We hypothesize that negative clinical outcomes will be associated with being classified as non-adherent. In addition, rates of adherence using the multidimensional classification system
and clinician ratings will be compared. We also assess perceived side effects with the hypothesis that those individuals who are non-adherent report more perceived side effects. Finally, we outline a method for implementing this classification system into routine clinical practice. Method Sample characteristics Eighty-two adolescents and young adults 11–21 yr of age (mean = 15.8, s.d. = 2.4) with solid organ transplants participated in this study. Forty-six were males and 36 were females. The majority of participants were Caucasian (61%) and African American (32%), with a small percentage of Asian-East Indian (1%) and those who did not disclose a race or ethnicity (6%). Median family income was $25 000– $50 000. Medical diagnoses, transplant type, donor status, number of transplants, and time since transplant are detailed in Tables 1 and 2. These data represent one component of a large investigation of health behaviors and quality of life in adolescent transplant recipients. Inclusion criteria for this study required that participants be English speaking, at least 11-yr-old, living at home with their parents, and have received their transplant at least four months prior to their participation in the study. Recruitment began following Human Subjects approval by the Institutional Review Board. Potential participants were contacted by the transplant coordinator at clinic or via telephone and given a brief description of the study. Participants interested in the study contacted the principal researcher directly, completed an interest form, or consented verbally. Participants were then Table 1. Medical diagnoses of transplant recipients in this sample (n = 82) Diagnosis Kidney Cystic/hereditary/congenital diseases Glomerulonephritis Miscellaneous Secondary GN/vasculitis Interstitial nephritis/pyelonephritis Hypertension/large vessel disease Liver Biliary atresia Cirrhosis (post-necrotic, congenital, unknown) Acute fulminant hepatic failure Chronic hepatitis Metabolic syndrome Neoplasm of histiocytic and mast cells WilsonÕs disease Heart Hypoplastic left heart syndrome Cardiomyopathy (dilated, familial, idiopathic) Aortic coarctation Arrhythmogenic RV dysplasia Congenital heart block Graft coronary disease Tricuspid atresia, pulmonary atresia Transposition of the great arteries Unknown Lung Pulmonary hypertension
n (%)
20 12 9 3 2 1
(24.4) (14.6) (11.0) (3.7) (2.4) (1.2)
7 7 1 1 1 1 1
(8.5) (4.9) (1.2) (1.2) (1.2) (1.2) (1.2)
4 3 1 1 1 1 1 1 1
(4.9) (1.2) (1.2) (1.2) (1.2) (1.2) (1.2) (1.2) (1.2)
1 (1.2)
591
Simons et al. Table 2. Description of transplant recipients by number, organ group, and donor type
Transplants received Transplanted organ Kidney Living donor Deceased donor Liver Living donor Deceased donor Heart Lung
First transplant, n (%)
Second transplant, n (%)
Third transplant, n (%)
82 (89.1)
8 (9.76)
2 (2.17)
47 19 28 20 4 16 14 1
2 0 2 1 0 1 5 0
1 1 0 0 N/A N/A 1 0
(40.4) (59.6) (20) (80)
Transplant recipients (n = 82); total transplants (n = 92); time since current transplant (years) M = 4.78, s.d. = 4.38; ranging from three months to 15.4 yr; median 3.3 yr.
approached and given a detailed description of the study. Written informed consent and assent were obtained at the clinic or via mail. Eleven adolescents did not participate in the interview after assent, seven as a result of significant developmental delay and four because of unavailability. Two parents were unable to be reached after consent and thus were not involved in the interview. The resultant sample of interviewees consisted of 78 parents and 71 adolescents, including 68 parent/adolescent dyads from the same family. For the purpose of these analyses, we examined parent reported and adolescent reported medication taking behavior and perceived medication side effects using participant interviews. Interviews were conducted over the phone by trained research assistants who were not involved in the patientÕs care. Participants were informed that all information shared was not part of their medical record and would be kept confidential. Medical diagnosis, transplant type, prescribed medication regimen, drug assay levels, and clinical outcome data were acquired through electronic chart review. AdolescentsÕ perceptions of side effects
The ESRD-SCL (17), validated for use with adults, was adapted for use with this sample. Adaptation involved review by pediatric transplant physicians across organ groups to include relevant side effects for pediatric solid organ recipients and removal of less relevant items. The adapted scale measures the frequency of 39 different side effects (e.g., weight gain, bruising) on 5-point Likert-like scales. The scales range from never to always. Adequate construct validity and internal consistency have been demonstrated for the ESRD-SCL. The frequencies of symptoms were summed to derive a total frequency score. Internal consistency estimates for frequency of side effects (parent, a = 0.88; adolescent, a = 0.91) were excellent in this sample.
be associated with low immunosuppressant drug levels and subsequent chronic rejection (18). Parent and self-reported medication adherence
The MAM (19–21) was used to assess adherence to medical regimens. Parents and adolescents individually reported how many doses of each medication the adolescent missed or took late in the previous seven days. The number of prescribed minus number of missed/late doses, divided by number prescribed, times 100 yielded a percentage of missed and late doses. Individuals who missed or took late 10% or more of their prescribed dose were considered non-adherent, based on cut-offs often used in the adherence literature (9, 22). Preliminary data on the MAM suggest adequate convergent validity with established measures of adherence. In a sample of patients with renal disease (n = 25), the percent of missed doses identified on the MAM was significantly correlated with the missed doses tracked by the Medication Event Monitoring System electronic technology (r = 0.40, p = 0.04). In another study of outcomes among renal transplant recipients (n = 29), percent of missed doses identified on the MAM was associated with the number of documented acute rejection episodes by year 2 post-transplant (r = 0.62, p < 0.001), suggesting good predictive validity of clinical outcomes in this population (19). Immunosuppressant drug assay levels
Measures of immunosuppressant blood levels were collected for one yr preceding the patientÕs interview, or since transplantation, if that occurred less than one yr earlier. Standard deviations for tacrolimus levels were calculated, with higher standard deviations indicating medication level instability and suggestive of more irregular medication taking. Higher standard deviations have been shown to be predictive of negative clinical outcomes, such as rejection, and are indicative of poor adherence (6, 23, 24). Out-oftherapeutic range blood levels of cyclosporine, sirolimus, and tacrolimus were also examined as potentially suggestive of poor adherence (23). This determination was made in consultation with the transplant coordinator responsible for each patient who took into account factors such as time since transplantation, recent medication changes, or recent aggressive medical treatments.
Transplant coordinator classification Transplant coordinators for each of the solid organ teams, kidney, liver, and heart/lung, classified patients as ‘‘adherent’’ or ‘‘non-adherent’’ based on their extensive personal knowledge of each patient and their family. The transplant coordinators reviewed the medical records of each patient. Factors that contributed to classification included: personal knowledge of the patient, time since transplant, a review of all immunosuppressant drug levels collected on an outpatient basis for a year prior to participation in the study, number of rejection episodes, and hospitalizations. Transplant coordinators had no knowledge of the adherence information provided by the adolescent or parent during individual interviews.
Clinical outcome
Data were obtained from the medical record on: (i) occurrence of a rejection episode in the past six months, (ii) occurrence of a hospitalization in the past six months, and (iii) mortality in the year following the interview date. Each of these outcomes was measured in a dichotomous fashion (i.e., presence/absence). Acute rejection has been found to
592
MACS Initial data to support the utility of the MACS (25) are presented in this investigation. The MACS places each patient into one of four categories based on parent/self report and immunosuppressant drug levels: (i) those who report
Multidimensional Adherence Classification Table 3. Medication Adherence Classification System Genuinely Adherent
Deniers/Medically Complicated
Disclosers/Medically Stable
Genuinely Non-adherent
All drug levels obtained are within range (no high or low levels noted) s.d. of tacrolimus is below 3
A high or low drug level is noted and/or standard deviation of tacrolimus is above 3 Patient and parent report missing or taking late* 10% of any medication in the last 7 days
Patient and parent report missing or taking late* 10% of any medication in the last 7 days
*Late doses must be taken at least one h beyond the scheduled dosing time.
excellent adherence and have acceptable drug levels (Genuinely Adherent), (ii) those who report excellent adherence and have concerning drug levels (Deniers/Medically Complicated), (iii) those who report non-adherence and have acceptable drug levels (Disclosers/Medically Stable), and (iv) those who report non-adherence and have concerning drug levels (Genuinely Non-adherent) (see Table 3 for detailed category descriptions). This classification system retains information provided by both selfreported data and immunosuppressant drug level data.
Statistics Data were analyzed with parametric and non-parametric tests using spss 14.0 for Windows (SPSS Inc., Chicago, IL, USA). Chi-squared analyses were conducted to examine the relationship between categorical demographic and medical variable with adherence classification categories. One-way anova analyses were conducted to examine the relationship between continuous demographic and medical variables (e.g., child age) and adherence categories. With SpearmanÕs Rho correlations, we examined the bivariate relationships between adherence measures and clinical outcomes. Finally, chi-squared analyses were conducted to examine the relationship between clinical outcomes and adherence classification categories.
Results Non-adherence and immunosuppressant assay data
The frequencies of non-adherence based on adolescent self-report and parent report, as well as the frequencies of concerning immunosuppressant drug levels are reported in Table 4. Based on these data for each individual patient, the MACS was used to classify adolescents into one of the four adherence groups: Genuinely Adherent (n = 22), Deniers/ Medically Complicated (n = 10), Disclosers/ Medically Stable (n = 28), and Genuinely Non-adherent (n = 22). The overall rate of non-adherence was 61% (see Table 5 for adherence classification across organ groups). Chisquared analyses to examine organ group differences across each of four adherence groups were non-significant, adherent v2 (3, n = 82) = 0.44, ns; Disclosers/Medically stable v2 (3,
Table 4. Frequencies and means of adherence across assessment methods
Method of assessment Adolescent self-report (n = 67) Total who missed >10% Total who were late >10% Parent report (n = 75) Total who missed >10% Total who were late >10% Immunosuppressant medication s.d. of tacrolimus (n = 63) Those with s.d. >3.0 Tacrolimus drug levels (n = 68) Tacrolimus levels >17 Tacrolimus levels < 5 Cyclosporin drug levels (n = 10) Cyclosporin levels >400 units Cyclosporin levels 10 Rapamune levels
