Background
Childhood rheumatic diseases have significant physical, mental, emotional, economic and social impact on patients and their families [
1]. Treatment regimens for these diseases are complex and require constant adherence for a long period of time in order for beneficial effects to be observed and unwanted side effects to be minimized. These factors increase the risk of poor adherence [
2]. In addition to pharmacological treatment, appropriate physiotherapy conditioning, nutritional support and psychosocial assistance are often required [
1].
Adherence is defined by the World Health Organization as the extent to which a person’s behavior in taking medication, following a diet, or making lifestyle changes corresponds to the recommendations of a health care provider [
3]. Adherence to treatment for chronic diseases is lower in children than in adults and is less extensively studied in children [
4].
Poor adherence in children is associated with multiple related factors, including social, demographic, cultural, and psychological factors, and depends on the triad relationship between family, patient and health care provider. Satisfaction with treatment, understanding of disease and treatment by patients and their caregivers, complexity of the disease and/or the treatment plan and the health care system adopted may also interfere with adherence [
5‐
8].
Regarding socio-economic factors, studies have shown that limited financial resources and low education of caregivers are associated with health problems and poor adherence to treatment regimens in children [
9,
10]. The authors point out that part of the difficulty in adapting to the prescribed regimen may be due to stress related to low socioeconomic status and poor social support [
11,
12].
Concerning clinical factors, disease severity at the time of diagnosis proves to be a decisive factor for good adherence. Moreover, medication regimens involving no more than two daily medications, followed by rapid beneficial effects and few side effects, facilitate adherence [
5,
13].
Regarding family factors, high rates of adherence are observed among patients who belong to cohesive families that include caregivers who are in a stable relationship [
13], have good social skills such as assertiveness and empathy when interacting with children [
14,
15] and support the socialization of children with friends [
16].
Remembering to take medications as instructed, despite distractions inherent to busy lifestyles, is a behavior dependent on the ability to plan and execute multiple tasks simultaneously and/or sequentially, which requires cognitive skills such as attention, memory and executive functions [
17,
18]. Few studies have been done to evaluate the association between treatment adherence and cognitive functioning. Most of the existing studies were conducted in adult patients with systemic lupus erythematosus (SLE) or acquired immune deficiency syndrome (AIDS) [
17,
19‐
22]. The number of studies on treatment adherence related to cognitive abilities of pediatric caregivers is even scarcer. In one study involving pediatric patients with AIDS, a relationship between low cognitive performance of caregivers and poor adherence was observed, although this relationship declined in multivariate analysis [
23].
Considering that socioeconomics, clinical care, family and emotional / behavioral factors are known to influence adherence to treatment of chronic disease, and that little or no knowledge is available about the influence of cognitive aspects of caregivers on adherence in pediatric rheumatology, the aim of this study is to perform a descriptive analysis of psycho-cognitive aspects of primary caregivers of pediatric patients with chronic rheumatic diseases, as well as socioeconomic and demographic status, and family functioning, and to evaluate relationships between these factors and adherence to treatment regimens. We also evaluated the impact of factors related to the disease, the treatment, and treatment satisfaction on treatment adherence.
Results
From June 2013 to April 2015, 146 pairs of caregivers and patients were screened with the medication adherence scale. Of these, five who would be classified in the poor adherence group did not agree to participate in the study and the last 51 to be evaluated that would be classified in the good adherence group were not included because the study required participants presenting poor adherence. Thus, among the 146 who participated in the pre-selection, 45 (30.8%) presented poor adherence, while 101 (69.2%) presented good adherence.
The final sample consisted of 90 primary caregivers and patients older than 10 years (n = 63) were also evaluated. Forty (44.5%) participants were classified in the poor adherence group and 50 (55.5%) in the good adherence group. In the poor adherence group, 19 (47.5%) presented unintentional failure, 3 (7.5%) presented intentional failure, and 18 (45%) presented the two forms of failure.
For patients, the mean age of those in the poor adherence group was higher than that of those in the good adherence group (14.1 vs 10.1 yrs.;
p < 0.0001; Mann Whitney Test). Time of follow-up, disease activity and quantity of prescribed medications were similar in poor and good adherence groups (Table
1).
Table 1
Distribution of sociodemographic and clinical characteristics and treatment satisfaction in relation to good and poor adherence
Patient | Age | 11.9 (±4.4) | 10.1 (±4.3) | 14.1 (±3.4) | < 0.0001*** |
Follow-up (years) | 3.7 (±3.7) | 3.1 (±3.0) | 4.4 (±4.4) | 0.150 |
Disease activity | Active | 52.8% (47) | 52% (26) | 53.8% (21) | 0.516 |
Inactive | 47.2% (42) | 48% (24) | 46.2% (18) |
Quantity of prescribed medications | 5.5 (±2.2) | 5.2 (±2.2) | 5.8 (±2.2) | 0.294 |
Caregiver | Age | 39.2 (±8.6) | 37.1 (±8.2) | 41.7 (±8.5) | 0.0113* |
Kinship | Mother | 82.2% (74) | 86% (43) | 77.5% (31) | 0.252 |
Father | 12.2% (11) | 12% (6) | 12.5% (5) |
Other a | 5.6% (5) | 2% (1) | 10% (4) |
Marital status | Married | 75.5% (68) | 82% (41) | 67.5% (27) | 0.090 |
Other b | 24.5% (22) | 18% (9) | 32.5% (13) |
Schooling | Stage 1 c | 47.8% (43) | 38% (19) | 60% (24) | 0.031* |
Stage 2 d | 52.2% (47) | 62% (31) | 40% (16) |
ABEP socioeconomic score | 20.9 (±5.2) | 21.1 (±4.6) | 20.6 (±5.9) | 0.713 |
Treatment satisfaction | 90.1 (±11.0) | 91.5 (±9.3) | 0.471 |
For caregivers, the mean age of those in the poor adherence group was also higher than mean age of those in the good adherence group (41.7 vs 37.1 yrs.;
p = 0.011; unpaired t test). A smaller proportion of caregivers with higher education level was observed in the poor adherence group compared to the good adherence group (40% vs 62%;
p = 0.031; Chi square test). Sociodemographic factors and treatment satisfaction did not differ significantly between the caregivers in the poor and good adherence groups (Table
1).
Regarding family functioning, the average scores on the Family APGAR scale were similar between good and poor adherence groups. For categorical analysis we classified the scores as “Highly Functional” or “Dysfunctional”, combining “Moderately Dysfunctional” and “Severely Dysfunctional” into the “Dysfunctional” category due to low representation of those classifications. The categorical analysis showed a higher proportion of “Dysfunctional” in the poor adherence group and higher proportion of “Highly Functional” in the good adherence group, although these differences did not reach statistical significance (Table
2).
Table 2
Distribution of family functioning, number of children, and responsibility for medication management in relation to good and poor adherence
Family Functioning a | Highly Functional | 71.4% (35) | 58.3% (21) | 0.152 |
Dysfunctional | 28.6% (14) | 41.7% (15) |
Average score | 14.9 (±4.6) | 13.2 (±5.3) | 0.118 |
Number of Children b | 2.1 (±0.8) | 2.9 (±1.8) | 0.004** |
Responsibility for Medication Management b | Caregiver | 66% (33) | 35% (14) | < 0.0015** |
Patient | 6% (3) | 47.5% (19) |
Patient and Caregiver | 28% (14) | 17.5% (7) |
We also analyzed the number of children and responsibility for medication management in relation to adherence. The average number of children per caregiver was higher in the poor adherence group compared to the good adherence group (2.9 vs 2.1;
p = 0.004; Mann Whitney Test). The proportion of families where the patient was solely responsible for medication management was higher in the poor adherence group compared to good adherence group (47.5% vs 6%;
p < 0.0015; Chi square test with Bonferroni adjustment), while the proportion of families where the caregiver or the patient together with the caregiver were responsible for medication management was higher in the good adherence group compared to the poor adherence group (66% vs 35% and 28% vs 17.5%, respectively; p < 0.0015; Chi square test with Bonferroni adjustment) (Table
2).
When caregivers’ behavioral profiles were analyzed we found that, compared to caregivers in the good adherence group, those in the poor adherence group scored significantly higher on the depressive problems (
p = 0.028; Mann Whitney test), attention problems (
p = 0.0003; Mann Whitney test), externalizing problems (
p = 0.013; unpaired t test), and total problems (
p = 0.0017; Mann Whitney test) scales. Those in the poor adherence group also scored higher on the internalizing problems scale, but the difference did not quite reach statistical significance (
p = 0.063; unpaired t test) (Table
3).
Table 3
Caregivers’ behavioral/emotional profile in relation to good and poor adherence
Depressive Problems | Good Adherence | 57.7 ± 8.8 | 0.028* | 74% (37) | 26% (13) | 0.301 |
Poor Adherence | 61.3 ± 9.1 | 66.7% (26) | 33.3% (13) |
Anxiety Problems | Good Adherence | 62.8 ± 6.9 | 0.384 | 62% (31) | 38% (19) | 0.212 |
Poor Adherence | 64.2 ± 7.6 | 51.3% (20) | 48.7% (19) |
Attention Problems | Good Adherence | 56.0 ± 5.8 | 0.0003*** | 88% (44) | 12% (6) | 0.008** |
Poor Adherence | 61.3 ± 7.0 | 64.1% (25) | 35.9% (14) |
Internalizing Problems | Good Adherence | 59.4 ± 10.6 | 0.063 | 52% (26) | 48% (24) | 0.096 |
Poor Adherence | 63.6 ± 10.5 | 35.9% (14) | 64.1% (25) |
Externalizing Problems | Good Adherence | 52.8 ± 8.2 | 0.013* | 80% (40) | 20% (10) | 0.008** |
Poor Adherence | 58.2 ± 12.0 | 53.8% (21) | 46.2% (18) |
Total Problems | Good Adherence | 54.8 ± 8.9 | 0.0017** | 76% (38) | 24% (12) | 0.007** |
Poor Adherence | 60.9 ± 10.2 | 48.7% (19) | 51.3% (20) |
For categorical data analysis, we grouped the borderline and clinical classifications into one category termed “clinical” due to low representation in these categories. When these categories were analyzed for each behavioral parameter, we found that, compared to caregivers in the good adherence group, a higher proportion of caregivers in the poor adherence group were classified as clinical with regard to attention problems (35.9% vs 12%;
p = 0.010), externalizing problems (46.2% vs 20%,;
p = 0.008) and total problems (51.3% vs 24%;
p = 0.014; Chi square test) scales, with no significant difference in the other scales. (Table
3). Internalizing problems are dysfunctional private behavior patterns, which are characterized by dysphoria and retreat. Externalizing problems are dysfunctional behaviors against others and the environment, which include opposition, aggression, impatience, mood fluctuations, teasing, and self-centeredness. Total problems are the plus of internalizing and externalizing problems. In summary, caregivers in the poor adherence group scored higher and were more likely to be classified as clinical on the attention problems, externalizing problems and total problems scales. They also scored higher on the depressive problem scale.
Regarding cognitive profiles of caregivers, no significant differences were observed in mean Total IQ, Verbal IQ or Executive IQ between caregivers in the poor and good adherence groups. Similarly, no significant differences were seen in the mean scores for factor indexes (Table
4).
Table 4
Caregivers’ cognitive profile in relation to good and poor adherence
Total IQ | Good Adherence | 95.6 ± 9.5 | 0.827 | 30% (15) | 54% (27) | 16% (8) | 0.274 |
Poor Adherence | 96.1 ± 13.6 | 39.5% (15) | 36.8% (14) | 23.7% (9) |
Verbal IQ | Good Adherence | 92.4 ± 8.5 | 0.708 | 40% (20) | 56% (28) | 4% (2) | 0.109 |
Poor Adherence | 93.2 ± 12.5 | 50% (19) | 36.8% (14) | 13.2% (5) |
Executive IQ | Good Adherence | 99.8 ± 12.6 | 0.936 | 24% (12) | 56% (28) | 20% (10) | 0.836 |
Poor Adherence | 99.8 ± 14.6 | 28.9% (11) | 50% (19) | 21.1% (8) |
Verbal Comprehension Index | Good Adherence | 90.1 ± 8.7 | 0.935 | 50% (25) | 48% (24) | 2% (1) | 0.403 |
Poor Adherence | 90.3 ± 12.8 | 50% (19) | 42.1% (16) | 7.9% (3) |
Perceptual Organization Index | Good Adherence | 96.9 ± 12.5 | 0.467 | 32% (16) | 52% (26) | 16% (8) | 0.639 |
Poor Adherence | 99.5 ± 14.5 | 31.6% (12) | 44.7% (17) | 23.7% (9) |
Working Memory Index | Good Adherence | 95.5 ± 10.0 | 0.687 | 22% (11) | 72% (36) | 6% (3) | 0.080 |
Poor Adherence | 96.4 ± 14.0 | 39.5% (15) | 44.7% (17) | 15.8% (6) |
Processing Speed Index | Good Adherence | 108.4 ± 11.4 | 0.716 | 4% (2) | 56% (28) | 40% (20) | 0.368 |
Poor Adherence | 107.4 ± 14.8 | 10.6% (4) | 44.7% (17) | 44.7% (17) |
For categorical data analysis, we grouped the IQ results into tertiles defined as below average (69–89), average (90–109) and above average (≥110). The proportion of scores in the below average, average and above average tertiles were determined for Total IQ, Verbal IQ and Executive IQ. No significant difference in tertile proportions between caregivers from the good and poor adherence groups was found for any of the IQ scores. There were trends, however, toward a higher proportion of caregivers scoring below average on the Working Memory Index (WMI) in the poor adherence group compared to the good adherence group (39.5% vs 22%), and a trend for higher proportion of caregivers scoring average on the WMI in the good adherence group compared to the poor adherence group (72% vs 44.7%;
p = 0.080; Chi Square Test with Bonferroni adjustment) (Table
4). Overall, this data indicates that caregivers cognitive functioning profile as assessed by the WAIS-III test is not associated with medication adherence.
To clarify the independent value of the psycho-cognitive and sociodemographic characteristics, a multivariate logistic regression analysis was performed (Table
5). In the final model, what remains significant is the patient’s age (
p = 0.026), the patient as solely responsible for managing medication (
p = 0.008), externalizing problems (
p = 0.012) and the WMI (
p = 0.041). Regarding the patient’s age, for each 1 year increase there was a 17.0% (OD = 0.83) reduction in the chance of adherence, adjusted for the presence of the other variables in the model. In addition, the patient’s chance of adherence was 91% lower (OD = 0.09) if he is solely responsible for managing medication than if the caregivers are responsible for administering the medication. There were no differences in chance of adherence among those in whom both the patient together with the caregiver were responsible for medication management and those whose caregivers were solely responsible (OD = 1.28;
p = 0.729). It was also noted that children whose caregivers were classified as clinical in externalizing problems have 81% (OD = 0.19) lower chance to present good treatment adherence than that of children whose caregivers were classified as normal. In the caregivers WMI, when the index was below average, there was a 66% (OD = 0.34) lower chance of good adherence, or when the index was above the average, curiously there was also an 88% (OD = 0.12) lower chance of good adherence. This result indicates that working memory index within the average favors good adherence to the treatment.
Table 5
Multivariate logistic regression analysis on poor adherence
Patient’s age | 0.78 (0.63–0.98) | 0.032 | 0.83 (0.71–0.98) | 0.026 |
Caregiver | | | | |
Age | 1.09 (0.98–1.21) | 0.112 | – | – |
Marital status (ref. = Married) | | | | |
Other | 0.41 (0.08–2.10) | 0.285 | – | – |
Schooling (ref. = Stage 1) | | | | |
Stage 2 | 3.39 (0.75–15.31) | 0.112 | – | – |
Number of Children | 0.70 (0.35–1.37) | 0.295 | | |
Responsibility for Medication Management (ref. = Caregiver) | | 0.008 | | 0.008 |
Patient | 0.042 (0.005–0.355) | 0.004 | 0.09 (0.02–0.49) | 0.006 |
Patient and Caregiver | 1.04 (0.19–5.76) | 0.966 | 1.28 (0.32–5.11) | 0.729 |
Depressive Problems - score | 1.03 (0.90–1.17) | 0.689 | – | – |
Clinical Attention Problems | 0.24 (0.03–1.82) | 0.169 | | |
Clinical Internalizing Problems | 0.91 (0.13–6.12) | 0.919 | – | – |
Clinical Externalizing Problems | 0.20 (0.03–1.29) | 0.091 | 0.19 (0.05–0.69) | 0.012 |
Clinical Total Problems | 1.61 (0.17–15.05) | 0.678 | – | – |
Working Memory Index (ref. = Average) | | 0.056 | | 0.041 |
≤ Below Average | 0.42 (0.08–2.06) | 0.283 | 0.34 (0.09–1.23) | 0.099 |
≥ Above Average | 0.07 (0.01–0.63) | 0.018 | 0.12 (0.02–0.76) | 0.024 |
Discussion
This study is unique in pediatric rheumatology because, in addition to considering socioeconomic, clinical and family factors, we also investigated the association between behavioral and cognitive profiles of primary caregivers in relation to medication adherence.
In the pre-selection phase, we obtained a rate of poor adherence to the medication of 30%. A previous study by our group showed a similar rate of poor adherence, 20.2% [
13]. Other studies in rheumatic diseases also observed similar rates of poor adherence, ranging from 15 to 48% [
5,
39,
40]. However, we must consider that these rates vary according to the method used. In the present study we used the Morisk, Green and Levine test that is useful to evaluate medication adherence in practically all type of diseases. The previous study used a scale developed by the researchers themselves [
41]. Usually more objective evaluation methods, such as electronic monitoring and counting of the number of tablets, detect higher rates of poor adherence in relation to self-reported scales [
5,
42,
43]. We chose to use the Morisk, Green and Levine test because it is a brief instrument, easy to apply and has been designed to underestimate good adherence and overestimate poor adherence.
Our study revealed a correlation between poor adherence and caregiver profiles indicative of behavioral problems, but no correlation between poor adherence and caregivers’ cognitive functioning profiles, except for WMI in the multivariate analysis. We also found that caregivers in the poor adherence group had higher mean age and lower education level compared to those in the good adherence group, although these observations do not remain significant in the multivariate analysis. The mean age of pediatric patients in the poor adherence group was also higher, and the multivariate analysis strongly confirmed this finding. This result probably explains why a higher proportion of older caregivers belonged to poor adherence group in the univariate analysis. More patients took the responsibility of managing medications in the poor adherence group than in the good adherence group, and the multivariate analysis also strongly confirmed this finding. In addition, families in the poor adherence group had a significantly higher mean number of children compared to those in the good adherence group; however this association was not observed in the multivariate analysis. Some sociodemographic characteristics such as caregivers with higher mean age, lower education level, and with higher mean number of children were related to poor adherence, but only in the univarietary analysis. Still, we suggest that special attention should be given to caregivers with these characteristics, in order to refer them to a psychological or/and psychiatric consultation when necessary.
The higher mean age of patients in the poor adherence group is consistent with previous studies that indicate an increase in poor adherence in children as they grow older and become adolescents [
4,
7,
23,
44]. Several characteristics of adolescents could be considered risk factors for poor adherence. One is a sense of omnipotence [
45,
46], which might lead adolescent patients to assume that dangerous situations have no consequences. Hence, they may stop treatment as a way to test whether they need to continue adhering, especially when they are asymptomatic, even if they had experienced serious morbidity previously. Furthermore, adolescents often engage in several extra-curricular activities and generally spend most of their time at school, which may hinder the incorporation of a medication regimen into their routines.
Another characteristic of adolescents that might impact medication adherence is the desire for quick results. Control of chronic rheumatic disease only occurs after a long period of treatment. In the short term the treatment may cause undesirable side effects such as change in appearance and malaise. Adolescents also prefer to make decisions for themselves, which can make it difficult to convince them to follow treatment regimens if they do not want to [
47].
We observed a higher prevalence of adherence failures among adolescents who took sole responsibility for administration of medications, as has been seen in previous studies [
9,
48]. In general, adolescents do not have fully developed autonomy, and thus lack skills required to assume sole responsibility for following treatment regimens. A study of patients with JIA showed that it is common for adolescents with this chronic disease to have difficulty with the management of the treatment during the transition to adulthood [
49]. In this same study it was observed that the most common reason for skipping medications was forgetfulness coupled with difficulty in taking medications as directed, keeping a calendar of appointments, and maintaining a personal medical file. Executive functioning, which includes planning and medications self-management, is among the last functions to be acquired during brain development [
50], which is not to say that good training cannot help the adolescent improve medication management. Therefore, adolescent patients who take on the responsibility for medication management might have better treatment adherence if they assume that responsibility gradually under the supervision of adult caregivers.
Furthermore, other reasons can delay the improvement of self-management in adolescent patients. The risk for parental overprotection of children with a chronic disease may delay the process of gaining independence from caregivers. Some transition programs have been running in pediatric rheumatology clinics to provide adolescent-oriented care in order to improve self-management. These programs also aim to promote successful transfer to adult rheumatology care. Some strategies within such programs include parenting orientation. The adolescent patients also have some small amount of time alone during consultation for confidential discussions regarding health issues including sexual health, risk-taking behaviors, family problems, and vocational issues [
51].
Caregivers in the poor adherence group showed more behavioral profile problems than those in the good adherence group, specifically in the areas of attention problems, depressive problems and externalizing problems. However, based on the multivariate analysis, we observed that only externalizing problems interfere effectively on medication adherence. Attention problems include problems with forgetfulness, concentration, planning, task completion, setting priorities, lack of energy, disorganization, and a tendency to lose things. These problems can be the result of cognitive impairment, but we did not observe significant differences in cognition between caregivers in the poor and good adherence groups through the univariate analysis. However, it called our attention that in the multivariate analysis, caregivers with WMI that were grouped below the average or above the average presented a lower chance of good adherence. Therefore, the attention problems observed in this study may reflect the difficulty in working memory ability. However, more specific neuropsychological instruments to investigate these aspects would be necessary to confirm this hypothesis. In addition, attention problems experienced by caregivers in the poor adherence group could be related to emotional factors and/or the complexity of life in a big city. Families with individuals inflicted by chronic disease are burdened due to the usual complex treatment regimen, especially when government support is not ideal. Health professionals could help lessen the impact of working memory difficulty and attention problems on treatment adherence by developing programs to help caregivers manage treatment and successfully incorporate treatment regimens into their busy lives.
Caregivers in the poor adherence group also exhibited more depressive problems than those in the good adherence group, according to univariate analysis. Although, multivariate analysis did not confirm depressive characteristics as effectively affecting adherence, special attention should be given to caregivers with such characteristics. Depressive problems include sadness, difficulty making decisions, lack of energy, and pessimism. Caregivers with depressive problems may lack motivation and vitality to accomplish daily activities, including maintaining a medication regimen. Due to lack of emotional resources, caregivers with depressive problems may shift the responsibility for managing medications to the patient, possibly without proper training. Depressive problems characteristics observed in this study may also be a reflection of externalizing problems, which were seen to interfere in adherence and were strongly associated to poor adherence in both univariate and multivariate analysis. Externalizing problems include opposition, aggression, impatience, mood fluctuations, teasing, and self-centeredness. Impulsivity is frequently found in depressed individuals and correlates positively with aggressive behavior [
52]. Individuals with major depressive disorder show higher impulsivity and more severe aggression than individuals without these conditions [
53].
Those in the poor adherence group also reported more internalizing problems in the univariate analysis, which are characterized by dysphoria and retreat, but the difference was not significant. Other studies have found that caregivers of pediatric patients with chronic diseases often have quality of life issues, including internalizing problems [
54‐
56]. Altogether, these findings underscore the importance of emotional and behavioral status of caregivers in adherence to treatment regimens for children with chronic diseases.
Although adherence to treatment regimens requires cognitive skills, in this study cognitive functioning problems in caregivers was not associated with poor adherence in the univariate analysis. However, we observed a lower chance of good adherence in caregivers with WMI grouped below average and above average. Caregivers with above average WMI presenting lower chance to adherence may result from a coincidence as the total number of caregivers with this condition was low (
n = 9). In any case, we observed that working memory index within the average favors good adherence to the treatment. We found only one report association between problems in cognitive functioning of caregivers and poor adherence to medication treatment, which was a study performed in children with AIDS [
23]. However, that association lost significance in multivariate analysis. Nevertheless, this AIDS study shows the importance of considering the emotional state of caregivers when evaluating cognitive functioning.
In the present study more caregivers in the poor adherence group scored below average on total, verbal and executive IQ, and on WMI and processing speed index compared to those in the good adherence group, but the differences did not reach statistical significance except for WMI in the multivariate analysis. It is known that the working memory is one of the essential cognitive functions that helps in organizing and planning and in facilitating the administration of medications [
17,
18]. Since a difficulty in working memory ability, and the presence of attention problems in caregivers in the poor adherence group was observed, more specific neuropsychological instruments to investigate these aspects would be necessary. Our study was not without limitations. Adherence was evaluated only through self-reporting, rather than through more direct and objective methods, such as counting the number of pills or measuring serum levels of medications. The absence of specific neuropsychological instruments to assess attention and working memory, the cross-sectional study design, the relatively small number of study participants, and the variety of illnesses studied were also a limiting factor.