Secondary Logo

Journal Logo

Original Articles: Hepatology & Nutrition

Prospective Analysis of Nonadherence in Autoimmune Hepatitis: A Common Problem

Kerkar, Nanda*; Annunziato, Rachel A; Foley, Liberty*; Schmeidler, James; Rumbo, Carolina*; Emre, Sukru*; Shneider, Benjamin*; Shemesh, Eyal*,†

Author Information
Journal of Pediatric Gastroenterology and Nutrition: November 2006 - Volume 43 - Issue 5 - p 629-634
doi: 10.1097/
  • Free



A key to the successful management of autoimmune hepatitis (AIH), a chronic inflammatory disease of the liver with frequent exacerbations, is maintaining stable immunosuppression (1). To be able to achieve stable immunosuppression, a patient needs to adhere to the prescribed medical regimen. Consequently, adherence to medications appears to be an important factor to consider in managing patients with AIH. Nevertheless, adherence has not been systematically evaluated in patients with AIH, there being little available information regarding whether, when and how patients actually take their medications as compared with what they have been prescribed.

Nonadherence may not be given adequate attention by gastroenterologists/hepatologists partially because assessing adherence on a regular basis can be challenging and time-consuming. Also, there is as yet no universally accepted standard method of performing this assessment. Previous studies (2) have established that subjective methods of assessment, such as self-reported questionnaires answered by physicians, nurses and caregivers, are not adequate measures of adherence. It is therefore necessary to use objective measures to obtain a realistic assessment of adherence. Because nonadherence has been noted to be a considerable problem while managing other chronic disease processes such as asthma (3) and HIV (4) and in solid organ transplantation (5), it is also likely to be an important factor that determines outcome in AIH. If so, it would be important to create methods and routines to evaluate adherence in clinical practice.

In pediatric liver transplant recipients our group has shown that although subjective methods of assessment of adherence did not correlate with medical outcome, an objective assessment method (fluctuation of medication blood levels) did (2). We previously computed the standard deviation (SD) of a series of blood levels of tacrolimus, obtained for each patient over 1 year of follow-up. In the cohort that we examined, a higher SD reflected increased variability between individual blood levels and correspondingly lesser adherence. This in turn was associated with poor outcome, biopsy proven rejection being taken as a measure for medical outcome in that study (2). In patients with AIH who have not undergone transplantation, medication blood levels are not checked at each clinic visit. We therefore decided to use electronic monitoring devices to monitor medication intake in patients with AIH for the purposes of the present study, and in addition used the tacrolimus blood-level measure in the group receiving tacrolimus.

ALT is characteristically used by clinicians to monitor disease activity and adjust immunosuppression in children with AIH (1,6). Histology is the gold standard (1), but it is not practically possible to perform liver biopsies repeatedly. A high ALT is generally considered equivalent to poor disease control (7) and leads to increase in medication, addition of another medication, liver biopsies, and in the worst case scenario, listing for liver transplantation if the child develops decompensation of liver disease (6). We used maximal ALT (the maximal ALT value recorded during the year of follow-up) as medical outcome measure for the purposes of this study (higher maximal ALT = worse outcome).

Once nonadherence is identified, it is important to evaluate risk factors that are associated with it to make informed therapeutic choices. We have previously focused our attention on posttraumatic stress disorder (PTSD) as a risk factor for nonadherence in medically ill patients (8). PTSD is a dysfunctional response to severe stress that is characterized by 3 clusters of symptoms. Patients report flashbacks of traumatic events (“re-experiencing” dimension), avoidance of reminders of the stressor (“avoidance” dimension), and prominent anxiety and hypervigilance (“hyperarousal” dimension) (9). A variety of stressors, including the diagnosis of a life-threatening illness, can lead to PTSD (9). Patients experiencing posttraumatic symptoms because of their illness may not take their medications as prescribed because the medications may remind them of their illness, a memory they wish to avoid (“avoidance cluster” of PTSD symptoms). If this is the case, then reducing PTSD symptoms may improve adherence (8).

The present prospective study aimed to use objective methods to assess nonadherence to immunosuppressant medications in a cohort of children with AIH, to determine whether nonadherence is associated with poor disease control and to assess whether disease-related posttraumatic stress is associated with nonadherence to immunosuppressant medications in this cohort.


The study was approved by the Institutional Review Board and involved an informed consent/child assent procedure. It was conducted in English or Spanish. The patient and family were made aware that this was a prospective study of medication adherence. Optimal adherence for the purposes of this study was considered to be 100% (patients take all of the prescribed doses within 2 hours of the expected time of administration).


All patients between the ages of 2 and 25 years with a diagnosis of AIH who did or did not have a liver transplant and were being treated in the Pediatric Liver/Liver Transplant Program at Mount Sinai Medical Center were included. Diagnosis of AIH was based on the presence of raised transaminases, autoantibodies (antinuclear antibody, smooth muscle antibody, liver kidney microsomal antibody), raised immunoglobulin G and characteristic features of interface hepatitis on liver biopsy (1). Patients with a diagnosis of AIH after transplantation either caused by recurrent disease (10) or de novo AIH (11) were also included. Patients were treated with prednisone and/or azathioprine. Liver transplant recipients were taking tacrolimus in addition to other immunosuppressants. Inability to understand the study procedures because of cognitive or language barriers was an exclusion criterion. Patients who only used liquid preparations were excluded from electronic monitoring because liquids cannot be monitored effectively with the electronic device that we used (see below).


After patients had completed their routine medical evaluation, they had an initial meeting with the assessor, during which inclusion/exclusion criteria were evaluated. Consent/assent was obtained and the patient/parent was given an electronic monitoring device (EM). A doctoral-level psychology intern gave the self-report questionnaires to the patients and helped children answer the questionnaire if needed but did not instruct them to use a specific answer or choice. The intern was blinded to the medical status of the patients. She was told that the subjects had a diagnosis of AIH but was not aware of whether or not the patient was suspected of being nonadherent, and was also not aware of ALT levels, medication blood levels or EM readings.

Assessments of Adherence

Electronic Monitoring

The EM device (a product of AARDEX/APREX) is a pill bottle with a lid containing a computer chip embedded in it that records each opening of the bottle. The assessor instructed the family to dispense the immunosuppressant medication only from the EM device. Only 1 medication (azathioprine, prednisone, tacrolimus, sirolimus or mycophenolate mofetil) per patient was put in the device. The dosing was once per day for azathioprine, prednisone and sirolimus, and twice per day for tacrolimus and mycophenolate mofetil. Tacrolimus was chosen in liver transplant recipients because trough levels are monitored on a regular basis. The device was not suitable for use with liquid medication. It was pointed out that opening the bottle for reasons other than drug ingestion (eg, to insert refills) should be documented and that the device would record the date and time the bottle was opened. Patients were asked to return the bottle in 2 to 3 months. At this time the reading was downloaded from the lid. We evaluated the extent to which patients took their medications “on time.” This was defined as taking medications within 2 hours of the expected time of intake. The 2-hour “forgiveness” interval with regard to taking the dose of medication is consistent with other studies of electronic monitoring of adherence in children (12). We also reported the absolute number of times that the bottle was opened as a percentage of the expected number of times. Another variable, the number of times that the patient opened the medication bottle (regardless of timing). This variable illustrates whether or not the problem has been only in timing of taking the medication, or whether there were patients who did not take the prescribed medication doses at all.

Blood Tests

In addition to using EM, we used standard deviation (SD) of trough tacrolimus blood levels as a marker for adherence in the posttransplant group. A complete description of this method is available elsewhere (2). Maximal ALT recorded during the year, for each patient, was selected as the medical outcome measure. In our clinic, ALT is measured at each visit for all patients being followed up for AIH.

Psychiatric Assessment

To assess symptoms of PTSD that were caused by the illness, we used a validated self-report questionnaire, the UCLA posttraumatic stress reaction index (PTSRI) (13). A cumulative score is obtained using 17 items from the questionnaire, corresponding to the Diagnostic and Statistical Manual for Mental Disorders, Fourth Edition (DSM-IV) criteria for PTSD (9). The PTSRI has been used extensively in various populations, including medically ill children. Psychometric properties of this questionnaire, including adequate reliability and validity, have been described in a recent review and compared to the gold standard, psychiatric interview (14). The PTSRI can be used either as a diagnostic tool where scores are computed separately on the 3 PTSD domains of avoidance, intrusion and hypervigilance, or as a summary score. Only the summary score has been shown to be a valid measure of PTSD (14). Therefore, for the purpose of the present study, we computed the summary PTSRI score as a measure of symptom severity. The PTSRI questionnaire was modified to include only the medical illness as the identified traumatic experience, not any other experience. This is because the relationship between PTSD and nonadherence is presumably caused by avoidance of medications. Medications can be viewed as traumatic reminders only if the trauma was connected to the medical condition. Hence, in the introductory section of the questionnaire, we asked patients to respond only to symptoms that are related to the medical illness as the identified trauma. We did not, however, make any changes to the body or scoring of the questionnaire.

Statistical Analyses

Analyses were performed using SPSS 12.0 statistical package and supervised by J.S. Pearson correlations were performed between measures of adherence, medical outcome and measures of psychiatric (PTSD) symptoms. Statistical tests are 2-tailed whenever applicable, and a P value of ≤0.05 was chosen as the level of statistical significance.


Thirty-four of 37 pediatric patients with AIH (Table 1) were enrolled in the present study. Three patients were excluded because consent for the study could not be obtained. Four children were prescribed only liquid preparation of prednisone and could not be given the EM device; hence, 30 patients were eligible to participate in the EM. Fifteen of 34 enrolled subjects were liver transplant recipients. Of the 15 liver transplant recipients, 6 had preexisting AIH and 9 developed de novo AIH after transplantation. However, in spite of initially consenting to the protocol, eventually only 14 of 30 eligible children returned the device after using it as directed during the 1-year study period. Of the 14, 11 were female and 8 were liver transplant recipients (Table 1). The EM readings were obtained during a median period of 6 months (range, 1–9 months). Of the monitored patients, none took their medications exactly as prescribed (EM reading range, 28%–94%). Indeed, 30% of the patients did not take their medications as prescribed ≥50% of the monitored time (Fig. 1).

Demographics and results of adherence monitoring in study population
FIG. 1
FIG. 1:
Results of downloading the EM device. The number of patients whose readings were monitored is shown along the Y axis and the EM readings (expressed as % taken on time [within 2 h of the prescribed interval]) are along the X axis. Of the monitored patients, 0 took their medications exactly as prescribed (EM reading range, 28%–94%). One third of the patients did not take their medications as prescribed ≥50% of the monitored time.

The percentage of times medication was taken on time correlated with the absolute number of doses taken (Pearson's correlation, 0.76, P = 0.001). The range of the absolute number of doses taken was 43.5% to 104% of prescribed doses. The mean was 83.4%, with a median of 81.3%. Table 1 shows the mean EM reading for AIH patients with and without transplant. There was no significant difference between the 2 groups (P = 0.40). One fourth of the patients opened the bottle <70% of the expected openings.

Correlations Between the 2 Measures of Adherence

A subsample of the cohort used tacrolimus and also used the EM device (n = 8). There was an inverse correlation between the 2 objective measures of adherence. Increased fluctuation in tacrolimus levels was correlated with decreased adherence as detected by the monitor (r = −0.72, P = 0.04). Thus, the 2 measures of adherence that were used in this study generated similar adherence determinations when used in the same patients.

Adherence and Medical Outcome

Electronic monitoring readings inversely correlated with maximal ALT levels (Pearson's, r = −0.59, P = 0.03, n = 14). The maximal ALT that we observed was during or after the EM period in 12 of the 14 patients and before we started the EM in the remaining 2 patients. Figure 2 illustrates the relationship between nonadherence and maximal ALT. It is clear that maximal ALT levels may be either low or high for EM readings below 70%, but are consistently low for EM readings of 70% or higher. Table 1 lists the mean maximal ALT levels for nontransplant versus transplant patients. The difference between these means was not significant (P = 0.46).

FIG. 2
FIG. 2:
Association between EM device readings and maximal ALT. The EM readings along the X axis indicate the number of times the medication was taken on time, expressed as a percentage of all monitored doses. Low EM readings were associated with a high maximal ALT in some patients. EM readings showing correct intake ≥70% of the time were consistently associated with a low maximal ALT.

Adherence and PTSD

A higher score on the PTSRI correlated with lower adherence as measured by the EM (P = 0.04, r = −0.61). Similarly, the SD of tacrolimus levels correlated with PTSRI scores (P = 0.03, r = 0.62), establishing that a higher level of PTSD symptoms is correlated with erratic medication-taking practices. Table 1 displays the mean PTSRI scores for patients who were versus those did not undergo transplantation. As Table 1 demonstrates, PTSRI scores in patients who did not undergo transplantation were higher than the scores obtained in posttransplant patients (almost twice as high, on average), but this difference did not reach statistical significance (t = 1.91, P = 0.07). The mean PTSRI score for patients who returned the EM was 17.40 (SD = 14.1) versus 14.2 (SD = 10.9) for those who did not. There was no significant difference in PTSRI scores between the groups (P = 0.59).


To the best of our knowledge, this is the first study to examine medication adherence prospectively in patients with AIH using objective measures. The EM device allows objective assessment of adherence in patients in whom measurement of blood levels of medication is not routinely possible. We think that the extent of nonadherence in this cohort of patients with AIH who not only had consented to participate in the study but also knew that their medication intake was being monitored is surprisingly high. There was not a single patient among those monitored who took medications as prescribed all of the time. Furthermore, of 30 eligible children only 14 used and returned their EM for downloading of data, which means that <50% of enrolled patients were adherent to the use of the monitor. This strongly suggests that our findings are conservative, and that the true rate of nonadherence in clinical populations is more than even the high rate that we report.

The nontransplant and transplant patients with AIH were grouped together because liver transplant was thought to be part of the spectrum of disease for patients with AIH. Those with poor control may go on to have a liver transplant, and recurrent AIH after liver transplantation is well described (15,16). Emotional trauma symptoms have previously been reported in relationship to liver transplantation but not to AIH. There were no significant differences between the patients who did versus those who did not undergo transplant in maximal ALT, EM readings and PTSRI scores.

We have chosen maximal ALT levels as our medical outcome parameter because ALT is a readily available measure that has been shown to correlate with the adequacy of immunosuppression as well as prognosis of AIH (17). Other parameters such as quantitative immunoglobulin G, liver biopsy and hospital stay may also reflect poor disease control, although they may not be as sensitive a marker of nonadherence as ALT. As part of routine medical care in our clinic, ALT is measured at every visit, but immunoglobulin G is not. Our clinical practice does not include routine surveillance liver biopsies, and hospital stays may not always be necessary for nonadherent patients. The maximal ALT was before the period of EM monitoring in 2 cases. It is possible that the patient had been nonadherent before monitoring began, resulting in a high ALT before the monitoring could be started. Blood tests of all patients are monitored regularly as dictated by clinical need. When abnormalities are noted, patients are called in for a clinic visit for appropriate intervention. A concerted effort was made not to call patients for research visits, but to perform the research during visits necessitated by clinical need. This approach results in minimal disruption to the patient's schedule. Nonadherence correlated with higher maximal ALT and, therefore, poor disease control in this population.

The 104% reading in 1 patient probably represents excessive opening of the bottle. Patients with EM readings showing correct intake ≥70% of the time had consistently low maximal ALT. A larger cohort of patients is needed to determine a potential adherence threshold in children with AIH. It is difficult to evaluate the temporal relationship between nonadherence and maximal ALT. Adherence is a dynamic phenomenon, and in this study adherence is examined over time, not at only 1 time point, whereas maximal ALT is only 1 reading. Furthermore, it is possible—indeed likely—that there is a lag, or disconnected relationship, between nonadherence and its impact on liver function. For example, a patient with erratic adherence may display a high ALT toward the end of the follow-up because of the cumulative effect of variable nonadherence on the liver, but it is possible that at the time the ALT was maximal, the patient has become temporarily more adherent. Therefore, for medical outcome, we used data for a full year, as opposed to ALT level at a single time point that may or may not be correlated with the precise time at which we observed a low medication blood level.

In a previous series, we reported that nonadherence had led to death in 3 adolescents with AIH who underwent liver transplant in our program (18). Given that the results of nonadherence in this group of patients can be fatal, our data strongly suggest that before a dose increase or medication change is made in patients whose blood level of a medication is low or whose clinical assessment indicates increased disease activity, clinicians must assess whether the patient actually has been taking the medication as prescribed. Examination of medication blood levels or use of EM devices is recommended rather than direct questioning of the patient and/or caregiver.

It is important to evaluate risk factors for nonadherence on the way to developing treatment strategies to improve adherence. We focused on posttraumatic stress symptoms (symptoms of PTSD) because we previously found that they are related to nonadherence in children who have undergone a liver transplant (8) and in adults who have had a myocardial infarction (19,20). In patients with AIH, the emotional trauma could be either the impact of being given a diagnosis of AIH, or the events that are associated with it such as lifelong immunosuppression, hospital visits, blood tests, liver biopsies and finally liver transplantation, if medical treatment is unsuccessful. We previously observed an association between PTSD and nonadherence to medications in different cohorts of pediatric and adult patients who experienced a discrete life-threatening event (8,21). We hypothesized that patients who are emotionally traumatized by their illness are more likely to avoid taking their medications because the action of taking the medication may be a painful reminder of the traumatic experience, the illness. The finding that higher levels of PTSD symptoms are associated with nonadherence suggests that assessment of PTSD may be useful in clinical practice to detect a group of patients who are at increased risk for nonadherence. PTSD is only 1 potential risk factor for nonadherence; other psychiatric symptom domains may also contribute to nonadherence and should be studied. It is possible that child PTSD influences adherence even though the parent is responsible for administering the medication. This could be because the child is actively trying to avoid the medication and the parent gives into the child's desires, and it could also be because parents of children with PTSD may be more likely to have PTSD themselves and hence avoid dealing with their child's medication. Because we did not examine parental symptoms in this study, we cannot prove these assertions as they relate to the present cohort. However, in a previous study we did report that medically ill children's PTSD symptoms are correlated with their parents' PTSD symptoms, and that a child's medical illness is traumatic to a parent, more so than a parent's own illness may be (22).

The small sample size and single-center nature of this study are limitations. Even in this small cohort, however, results did reach statistical significance. This suggests that the phenomenon that we observed was robust. Our data strongly support the notion that children with AIH frequently do not take their medications as prescribed. It is possible that even our alarming results were actually an underestimation of the true degree of nonadherence in this sample. First, we enrolled only patients who consented (the most severely nonadherent patients may not have given consent in the first place), and second, nonadherent patients may have been less likely to use the EM. Mount Sinai is a quaternary care center that serves a heterogeneous group of patients from all financial and social strata. The percentage of different ethnic groups and degree of enrollment of financially disadvantaged people may well be an institutionally determined bias.

Nonadherence has been identified as a major cause of morbidity and mortality in patients with chronic illnesses such as asthma (3) and HIV infection (23) and organ transplant recipients (5,24). Our study extends these observations to patients with AIH. It suggests that the extent of nonadherence in this cohort is substantial and also points to a potential risk factor for it, namely, disease-related distress or PTSD. We believe that it is prudent to conclude that nonadherence emerges as a common and clinically significant risk factor for poor outcome in AIH. Unfortunately, nonadherence is not always addressed as a risk factor for poor outcome in AIH in routine clinical practice. In our view, it is possible to monitor adherence and provide some educational guidance and follow-up for suspected cases, even if a strong psychosocial team is not available. We have also pointed to a potential intervention, by either gastroenterologists or mental health consultants, that could be undertaken to improve adherence, namely, management of disease-related distress and posttraumatic stress. This study is only the beginning. Larger studies in children and adults are needed to examine the full extent of this problem and its correlates on the way to designing interventions to improve adherence, and hence, outcomes.


1. Mieli-Vergani G, Vergani D. Autoimmune hepatitis in children. Clin Liver Dis 2002; 6:623–634.
2. Shemesh E, Shneider BL, Savitzky JK, et al. Medication adherence in pediatric and adolescent liver transplant recipients. Pediatrics 2004; 113:825–832.
3. Bauman LJ, Wright E, Leickly FE, et al. Relationship of adherence to pediatric asthma morbidity among inner-city children. Pediatrics 2002; 110:e6.
4. Eldred LJ, Wu AW, Chaisson RE, et al. Adherence to antiretroviral and pneumocystis prophylaxis in HIV disease. J AIDS Hum Retrovirol 1998; 18:117–125.
5. Dew MA, Kormos RL, Roth LH, et al. Early post-transplant medical compliance and mental health predict physical morbidity and mortality one to three years after heart transplantation. J Heart Lung Transplant 1999; 18:549–562.
6. Vergani D, Mieli-Vergani G. Autoimmune hepatitis. Minerva Gastroenterol Dietol 2004; 50:113–123.
7. Czaja AJ. Autoimmune liver disease. Curr Opin Gastroenterol 2006; 22:234–240.
8. Shemesh E, Lurie S, Stuber ML, et al. A pilot study of posttraumatic stress and nonadherence in pediatric liver transplant recipients. Pediatrics 2000; 105:e29.
9. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 4th edition (Text Revision) (DSM-IV-TR). Washington, DC: American Psychiatric Publishing, 2000.
10. Neuberger J, Portmann B, Calne R, et al. Recurrence of autoimmune chronic active hepatitis following orthotopic liver grafting. Transplantation 1984; 37:363–365.
11. Kerkar N, Hadzic N, Davies ET, et al. De-novo autoimmune hepatitis after liver transplantation. Lancet 1998; 351:409–413.
12. Rapoff MA, Belmont JM, Lindsley CB, et al. Electronically monitored adherence to medications by newly diagnosed patients with juvenile rheumatoid arthritis. Arthritis Rheum 2005; 53:905–910.
13. Rodriguez NS, Steinberg AM, Pynoos RS. UCLA Post-traumatic Stress Disorder Reaction Index for DSM-IV, Child, Adolescent and Parent Versions. Los Angeles: University of California Press, 1998.
14. Steinberg AM, Brymer MJ, Decker KB, et al. The University of California at Los Angeles Post-traumatic Stress Disorder Reaction Index. Curr Psychiatry Rep 2004; 6:96–100.
15. Birnbaum AH, Benkov KJ, Pittman NS, et al. Recurrence of autoimmune hepatitis in children after liver transplantation. J Pediatr Gastroenterol Nutr 1997; 25:20–25.
16. Devlin J, Donaldson P, Portmann B, et al. Recurrence of autoimmune hepatitis following liver transplantation. Liver Transpl Surg 1995; 1:162–165.
17. Miyake Y, Iwasaki Y, Terada R, et al. Persistent normalization of serum alanine aminotransferase levels improves the prognosis of type 1 autoimmune hepatitis. J Hepatol 2005; 43:951–957.
18. Lurie S, Shemesh E, Sheiner PA, et al. Non-adherence in pediatric liver transplant recipients—an assessment of risk factors and natural history. Pediatr Transplant 2000; 4:200–206.
19. Shemesh E, Stuber ML. Posttraumatic stress disorder in medically ill patients: what is known, what needs to be determined, and why is it important? CNS Spectr 2006; 11:106–117.
20. Shemesh E. Emotional trauma and medical illness: investigating the associations. CNS Spectr 2006; 11:103–104.
21. Shemesh E, Yehuda R, Milo O, et al. Posttraumatic stress, nonadherence, and adverse outcome in survivors of a myocardial infarction. Psychosom Med 2004; 66:521–526.
22. Shemesh E, Newcorn JH, Rockmore L, et al. Comparison of parent and child reports of emotional trauma symptoms in pediatric outpatient settings. Pediatrics 2005; 115:e582–e589.
23. Park-Wyllie LY, Phillips EJ. Challenges of adherence management in human immunodeficiency virus pharmacotherapy. Can J Clin Pharmacol 2003; 10:189–195.
24. Bunzel B, Laederach-Hofmann K. Solid organ transplantation: are there predictors for posttransplant noncompliance? A literature overview. Transplantation 2000; 70:711–716.

Nonadherence; Autoimmune hepatitis; Electronic monitoring; Posttraumatic stress

© 2006 Lippincott Williams & Wilkins, Inc.