The introduction of antipsychotic treatment significantly improved the course and prognosis of patients with schizophrenia. While in the era before antipsychotics and at the time when the disease was recognized by Kraepelin as dementia praecox, three-quarters of patients ended up with severe impairment of function, only a quarter of patients end up with severe impairment of function nowadays (but this impairment is still significantly less severe than in the past) (19). Use, or rather non-use, of antipsychotics has also short-term serious consequences in the form of relapses, rehospitalization, time to achieve remission and suicide attempts (20).
The patient population participating in our study (N = 55, mean age 39.2 years, 72.7% male, mean CGI-S score 3.8, PANSS 58.9 and PSP 60.3) is comparable to other studies on adherence in schizophrenia patients using EAM (6,21–25).
38 patients (69.1%) met the combined adherence criteria (EAM + TDM); this “double-check” implies that these patients indeed adhered well to their medication regimen. After a week-long EAM, an increase in blood concentrations by more than 30% was observed in nine patients (16.4%) who opened the dispenser regularly during the study; in 1 patient (1.8%), an increase occurred despite insufficient dispenser opening. This indicates a high likelihood of non-adherence of these ten patients before the initiation of the EAM. Three patients (5.5%) showed stable blood concentrations despite insufficient use; this suggests that they used the drugs as insufficiently during the study period as before it. Finally, in 4 patients (7.3%), a decrease in blood concentrations by more than 30% was observed despite regular use. These patients may have taken the medication excessively before the EAM or in a different daily schedule than they stated, which could have affected the blood sampling times before and after the EAM and, thus, the initial measured blood concentrations.
The observed full adherence in 69.1% (95% CI 55.2-80.9) of patients falls within the wide ranges reported by Lacro et al. (2002) (5), Sendt et al. (2015) (26), or Misdrahi et al. (2018) (6) and is also in accordance with the expert-estimated interval of 51-70% (4). In one of their early works, Velligan et al. (2007) (21) found full adherence in 63.5% of outpatients based on a 12-week EAM limit of 80% of daily doses. A lower adherence rate in outpatients was found by Remington et al. (2007) (22), where over 4 weeks, only 48% of patients took daily medication in a sufficient dose on 80% of the days; it is, however, necessary to note that the patients in their study were characterized by a higher degree of severity of the condition according to the PANSS. In a study by Yang et al. (2012) (23), 58.8% of patients used more than 80% of daily doses within ± 3 hours of scheduled times during the 8-week EAM. Acosta et al. (2013) (24) found during a 3-month EAM that 76% of patients took their daily dose on more than 80% of the days, but when considering also the proper time of use (± 2 hours), only 35% of patients took the medication properly on more than 80% of the days. A similar rate of adherence when taking into account only the daily dose was reported by Brain et al. (2014) (25), who found during a year-long EAM that 73% of patients took their daily dose on more than 80% of days. On the other hand, Misdrahi et al. (2018) (6) monitored schizophrenic patients for 6 months after discharge from hospitalization and found that only 37.3% could be considered adherent, while some of the patients did not start taking medication at all after their discharge from the hospital and the proportion of adherent patients continued to decrease over time. In our case, 69.1% of patients were considered adherent based on the combined criterion of opening the dispenser properly within the specified interval (± 2 hours) and maintaining relatively stable blood concentrations of the monitored drug. This relatively high rate corresponds rather to the results of studies that did not condition adherence by use within a certain time frame. The difference can be explained by the fact that the dispensers we used repeatedly reminded the patients of the need to take the medication (sound signal), thereby increasing compliance with the regimen. This may also be related to the relatively short study period (7 days), so it was easier to follow the correct regimen during this time.
In 4 patients (7.3%), a decrease in blood concentrations during regular use was observed, which does not exclude the possibility that before the introduction of electronic monitoring, they used higher doses of drugs than prescribed. In a study by Acosta et al. (2013) (24) who performed electronic monitoring of patients with schizophrenia using the MEMS system (the Medication Event Monitoring System), at least one event of overdose with medication was recorded in 74.3% of monitored patients within the 3 months study period; overall, the use of an excessive dose occurred on average in as much as 4.7% of all monitoring days. Yang et al. (2007) (27) evaluated the Medication Possession Rate (MPR), which indicates the proportion of pills picked up from the pharmacy for a given period compared to the number corresponding to the recommended doses, 7.6% of patients with schizophrenia were found to pick up an excessive amount of medication from the pharmacy (MPR > 1.2). These patients did not differ demographically from the adherent group, but they used olanzapine, quetiapine, and risperidone more often than the adherent patients. Valenstein et al. (2002) (28) found that the odds ratio (OR) of hospital admission in patients with MPR > 1.1 compared to the adherent patients (0.8 ≤ MPR < 1.1) was 3.0, which was even more than patients collecting medication less than prescribed (MPR < 0.8) whose OR for hospital admission was 2.4 compared to the adherent group. At the same time, even a small increase in MPR was similarly risky for hospital admission as a relatively significant decrease in MPR (1.1 ≤ MPR <1.2 corresponded to 0.5 ≤ MPR < 0.6; 1.2 ≤ MPR < 1.3 corresponded to 0.2 ≤ MPR < 0.3). This excessive collection of drugs from the pharmacy could be caused by the deterioration of the condition requiring an increase in medication doses, but also by the disorganization in the use of medication on the side of the patient (28) or less strict monitoring of second-generation antipsychotics by the physicians due to fewer side effects (27). In any case, the results of our work confirm that a group of patients overusing medication probably exists, and it is necessary to pay attention to it because it can signal the risk of adverse clinical development of the patient’s condition. It should be noted that this medication overuse can be also observed in other groups of patients with mental illnesses, as we have reported in patients treated with antidepressants in the past (29).
No statistically significant difference was found between the group of adherent and non-adherent patients in any scale of adherence or functional status. This is, however, consistent with most of the aforementioned studies (6,21,24,25).
All three scales for the assessment of adherence (CRS, VAS, DAI-10) correlated to a certain degree, i.e. the subjective estimates of adherence by physicians and patients showed a certain degree of similarity. When these scores were used alone to assess adherence, they yielded very similar adherence estimates of around the unrealistic 90%. At the same time, however, these subjective evaluations did not differ significantly between the groups of adherent and non-adherent patients. It is, therefore, possible that the physician accepts to a certain degree the way the patient presents his cooperation and relationship to the medication, regardless of whether it corresponds to reality or not. However, since the mutual correlation was not high, the results rather reflect the inaccuracy of these scales for estimating adherence and the tendency to overestimate it (4). On the other hand, Brain et al. (2014) (25), who studied patients treated in a community psychiatry setting, reported that adherence according to the subjective assessment by the patient, informant, and staff showed a high degree of agreement with the results of EAM; they explained this by high-quality education of the patients by the staff. Furthermore, the adherence rate determined according to EAM correlated with that determined from the pill count and physician estimate (21), and with a DAI-10 score (23).
Unlike the most widely used “single-chamber” system MEMS® (https://aardexgroup.com/medication-adherence-packaging/), the “multi-chamber” electronic dispenser used by us (SimpleMed+®) supported the dispensing of all medications prescribed to the patient. This is potentially beneficial from the perspective of a stable intake of a complex mix of medications compared to the monitoring of usage of a single drug, as the latter might fail to capture changes in the blood concentrations due to the pharmacokinetic interactions.
A concentration change of 30% was chosen to determine non-adherence. According to our previous experience, this eliminates the variability of the method as well as partial fluctuations in the individual concentration on the part of the patient, including the regimen of the day, times of use and sampling, diet, etc. (17,29). The same rate of concentration change was also used for assessing non-adherence under controlled use in the study by Velligan et al. 2007 (21).
The omission of more than 20% of the planned doses of medication was the other criterion for determining non-adherence. It was, therefore, possible to completely miss a full day of taking the drug over the 7 days of follow-up with the patient still meeting the criteria for adherence (85.7% of correctly used doses). Such an omission would have likely been detected (based on the blood concentration) if it occurred towards the end of the study period but not if it occurred at its beginning.
Electronic monitoring of dispenser opening remains an indirect way of measuring adherence, as it is not possible to consider it the use of medication. On the other hand, the determination of blood concentrations is clearly related to the use of drugs, but the individual measurement is complicated by individual metabolic deviations as well as the regimen of medication use in the short interval before the measurement. The combination of EAM and TDM was able to reduce to a large extent the Hawthorne effect, i.e. the change of behavior within the scope of the research project (30). The research participants were not aware of the detailed research protocol until the moment of consent that was given immediately before the first examination of blood concentrations, and any change in the patients' behavior during the research influenced by the awareness of EAM was subsequently detectable by a change in blood concentration.
For the sake of completeness, it should be added that in the Czech Republic, by law, all patients are treated free of charge and their treatment is covered by compulsory health insurance, including all antipsychotics in at least one generic form, so access to care is not limited by any financial or other barriers on the part of the patient.
The patients who participated in the study evaluated the usefulness of the electronic dispenser positively, the median was the second-best grade (on a five-grade scale used in the Czech education system). The patients themselves did not find it difficult to use, but on the other hand, only 20% of them reported that they would be interested in long-term use of the dispenser. The causes of this relative reluctance lay in the care of the dispenser (filling it with pills, charging, connecting to the Internet, etc.) as well as in its poorer compactness, which did not allow it to be easily carried away from home.
The small set of patients can be considered a limitation of the study; the number of participants is, however, comparable to previously published studies. It is also necessary to assume that the research methodology including, among other things, the need to train patients to use the dispenser, could have influenced the selection of patients; in effect, patients who were more inclined to cooperate with physicians (and, therefore, more adherent) participated in the research; this “volunteer bias” is, however, an inherent problem associated with this type of studies. Finally, including the determination of metabolites in the adherence assessment would be also beneficial for further improving the accuracy of our approach.
On the other hand, the combined criterion of two parameters that were previously used independently (TDM and EAM), to determine adherence is, to the best of our knowledge, probably the most complex and reliable method employed so far, as it allows assessment of the adherence not only during the study, but also shortly before the study, which partly reduces the “volunteer bias”.