SEARCH

SEARCH BY CITATION

Keywords:

  • compliance;
  • intentional;
  • medication adherence;
  • patient satisfaction;
  • relationship

Summary

  1. Top of page
  2. Summary
  3. What is known and objective
  4. Methods
  5. Results
  6. Discussion
  7. What is new and conclusion
  8. Acknowledgements
  9. Funding
  10. References

What is known and objective

Patients' poor adherence to medications is reported to be related to the individual patients' beliefs and cognitions and their trust of the medical staff. However, the causes of the two forms of non-adherence, intentional and unintentional behaviours, have yet to be clarified. This study compared psychological latent factors associated with intentional and unintentional non-adherence to chronic medication regimens, focusing on the potential effects of (i) patients' dissatisfaction with treatment and their relationships with the medical staff and (ii) patients' subliminal rational thinking processes, which weighed the positive values such as their expectations of benefits from treatment against negative values such as their dissatisfaction.

Methods

Two cross-sectional surveys were undertaken of patients given medications for chronic diseases, using a questionnaire developed and validated in this study. One survey was undertaken in three hospitals and the other survey, online throughout Japan. We scored the individual latent factors using the questionnaire and calculated the differential score between two negatively correlated latent factors to quantify patients' subliminal rational thinking process. We compared the adjusted odds ratio (OR) of latent factors between intentional and unintentional non-adherence to medication in both surveys.

Results and discussion

Of the eligible subjects, 149 hospitalized patients and 524 survey participants completed the questionnaire. Intentional non-adherence was associated with patient dissatisfaction with treatment including interpersonal relationships with medical staff in both hospitalized patients and online survey participants (95% confidence interval of adjusted OR for Dissatisfaction, 1·20–16·26 in the hospital-based survey and 1·33–3·45 in the online survey). In both surveys, intentional non-adherence was significantly associated with the differential score between two negatively correlated latent factors, Willingness and Dissatisfaction (= 0·02 in the hospital-based survey and < 0·001 in the online survey). However, these associations were not evident in unintentionally non-adherent patients.

What is new and conclusions

Patients' dissatisfaction and their resulting rational judgments are unique, consistent determinants of intentional non-adherence to medications, but not of unintentional non-adherence.


What is known and objective

  1. Top of page
  2. Summary
  3. What is known and objective
  4. Methods
  5. Results
  6. Discussion
  7. What is new and conclusion
  8. Acknowledgements
  9. Funding
  10. References

Approximately 50% of patients do not adhere to prescribed medication regimens.[1, 2] This so-called non-adherence has been shown to increase mortality risk,[3] hospitalization rates[4] and healthcare costs.[5] Recent reviews of the literature on interventions enhancing adherence have concluded that to date, existing methods are neither cost-effective nor substantially beneficial.[6]

Although non-adherent patient behaviours were historically viewed as a homogenous entity, recent studies indicate that non-adherence can be further divided into intentional and unintentional behaviours.[7-11] Intentional non-adherence is defined as discontinuing, skipping or altering the dose of medication, whereas unintentional non-adherence encompasses inadvertent forgetfulness, careless missing of doses or lacking a clear understanding of the prescribed instructions.

Patient's beliefs and cognitions underlying these behaviours must inevitably differ from each other. Clarifying differences in the psychological components of the behaviours should be useful in identifying and customizing interventional strategies. However, only a few studies have examined the differences in the psychological factors underlying the behaviours.[8, 9, 11] Even those exiguous studies had inconsistent findings and did not explore issues of patient satisfaction with interpersonal relationships, even though trust and satisfaction with the medical staff have recently been suggested to influence non-adherence.[12]

This study aimed to compare the individual latent factors associated with intentional and unintentional non-adherent behaviours in patients with chronic diseases focusing on the potential effects of patient satisfaction with treatment, including their relationships with the medical staff. The study also focused on patients' subliminal rational thinking processes, which weighed the positive value of treatment and medication against negative values,[13] as a few reports[9, 11] suggested that rational judgment may influence the development of intentional non-adherence.

Methods

  1. Top of page
  2. Summary
  3. What is known and objective
  4. Methods
  5. Results
  6. Discussion
  7. What is new and conclusion
  8. Acknowledgements
  9. Funding
  10. References

We conducted two separate cross-sectional surveys, one hospital-based and one online nationwide. In the hospital-based study, we identified psychological latent factors regarding medication and treatment through administration of a questionnaire, the Medication Acceptance, Preference and Adherence Scale (MAPAS), which assessed each patient's beliefs, values and ideas concerning their acceptance and preference for medications and treatments. We then estimated each individual score for the identified latent factors and analysed the association between latent factors and medication non-adherence. The online survey was performed for cross-validation of the hospital-based survey findings. The study was approved by the respective ethics committees at Tokushima Bunri University, Kagawa University Hospital, Takamatsu Red Cross Hospital and Sakaide City Hospital.

Statistical analyses were performed using pasw statistical software (ver.18.0; IBM Corporation, Armonk, NY, USA), unless otherwise specified. All P-values were two-tailed with statistical significance defined a priori as < 0·05. The Mann–Whitney U-test evaluated the difference in frequency distribution of responses in two surveys.

Hospital-based survey

Subjects

The hospital-based survey was administered at three hospitals located in Kagawa Prefecture in Japan from June 2009 to April 2010. Kagawa University Hospital is a 613-bed tertiary emergency care facility. Takamatsu Red Cross Hospital, a 589-bed hospital, and Sakaide City Hospital, a 216-bed facility, provide secondary emergency care. Eligible hospitalized patients with chronic diseases under the care of pharmacists were enrolled if they were at least 35 years of age, took oral medications for chronic diseases for at least 6 months or had undergone injectable systemic chemotherapy during their hospitalization. In Japan, pharmacists routinely explain to hospitalized patients how to use their medications. Patients undergoing chemotherapy were included in our development of the MAPAS, as it enabled the identification of patients who must potentially tolerate severe adverse drug reactions. Patients were excluded who had poor performance status in the activities of daily living, poor understanding of the instructions, difficulty conversing and those for whom discharge was planned within 3 days. Poor performance status was defined as a score of four according to the Eastern Cooperative Oncology Group criteria.[14]

After obtaining written informed consent, the pharmacists met face-to-face with the patients and assessed their medication non-adherence prior to hospitalization using the questions described below. The pharmacists also queried them about chronic oral medications, length of medication use, adverse drug reactions and prior hospitalizations. Their patients were handed the 30-item pilot questionnaire, which was retrieved later. Patient characteristics, including sex, age, diagnoses, marital status and parity, were obtained from medical electronic records.

Development of the MAPAS

For the development of the MAPAS, we first created a pilot questionnaire consisting of 30 survey items, each graded from 1 (strongly disagree) to 5 (strongly agree), referring to factors identified in previous studies relating non-adherence to medications and patient satisfaction.[2, 8, 11, 12, 15-21]

Second, we administered the 30-item pilot questionnaire to the hospitalized patients. From the responses, we identified psychological latent factors regarding medication and treatment using unweighted least squares factor analysis with promax rotation. If the factor loading of survey items was 0·35 or higher, we accepted the item as a construct of a latent factor.

To confirm the validity of the latent factors, we estimated the internal consistency of the latent factors using Cronbach's alpha.[22] A low alpha value would suggest that the items were not all measuring the same construct. Values above 0·60 are recommended in group comparisons.[23, 24] We also calculated interfactor correlation coefficients to identify any correlations between the latent factors. With this data, we completed the MAPAS, including only the items pertaining to the identified latent factors.

Medication non-adherence

For assessing intentional and unintentional non-adherence, we asked the following two questions[8]: (i) ‘Have you intentionally adjusted any of your regular medications at least once during the few months prior to your hospital admission, by intentionally skipping or stopping the medications or altering the prescribed doses without your physician's consent?' and (ii) ‘Have you forgotten to take at least one of your regular medications at least once during the few months prior to your hospital admission?'

Patients were also asked about their frequency of intentional and unintentional non-adherence (i.e. every day, about twice per week, about once per week, about once per month, about once per 2–3 months, about once per four or more months or never).

Latent factor and medication non-adherence

To assess the association between latent factors and non-adherence to medications, we first summated each respondent's score of the survey items constructing the respective latent factors and converted the total score into values ranging from 0 to 100 to generate a standardized factor score.[23] We also calculated the standardized differential score ranging from −100 to 100 by subtracting the standardized factor scores between pairs of negatively correlated factors (subtracting the score of the latent factor with negative value from the score of the latent factor with positive value)[9, 17] to quantify patients' subliminal weight of positive value against negative value of treatment and medication.

Second, we compared these standardized scores between the three groups, the adherent and the intentionally and unintentionally non-adherent. Originally, four groups were planned, but we had no alternatives to include patients with both intentional and unintentional non-adherence into the intentional non-adherent group due to the extremely small number of the patients with intentional non-adherence alone. The differences in the mean values between these groups were tested using Tukey's nonparametric multiple comparison, which was calculated manually.[25]

Third, using a logistic regression model, we calculated an adjusted odds ratio (OR) for non-adherence to medications. In this model, non-adherence was entered as a dichotomous variable (‘never' is defined as adherence and the remaining responses are defined as non-adherence) in intentional non-adherence and unintentional non-adherence groups, respectively. Latent factor score was entered as a categorized variable; two categories were defined by median values in the standardized factor score and four categories by quartile values in the standardized differential score because the latter differential score exhibited a relatively wider range.

The adjusted OR was controlled for potential confounding variables when P-values < 0·10 for sex, age (below 45 years, 45–54, 55–64, 65–74 and 75 or over), number of different oral medications (1–3 types, 4–6, 7–9 or ≥10), duration of medication use (<1 year or ≥1 year in the hospital-based study, <5 years or ≥5 years in the online survey), performance status (0 or ≥1), adverse drug reactions, the experience of hospitalization, marital status and parity. In addition to these essential variables, the hospital-based survey included the clinical diagnoses (diabetes mellitus, hyperlipidaemia, hypertension, hepatocellular carcinoma and renal failure).

Nationwide online survey

The nationwide cross-sectional online survey was conducted over a 3-day period in March 2011. By means of probability sampling apportioned by age (in 10-year intervals) and residential location (four regions) in accordance with demographic data in Japan, 20 300 individuals aged 35–74 years were randomly selected among registrants of Macromil, Ltd. (Tokyo, Japan), and were invited to join the study through mail correspondence. They were allowed to participate if they had taken one or more prescribed oral medications for a chronic disease for at least 6 months. We set an upper age limit, as the elderly generally did not have access to the Internet.

The participants were asked to respond to the MAPAS and were queried on their medication non-adherence behaviour, types and duration of chronic oral medication use, current illnesses, performance status, adverse drug reactions, hospitalizations, education and employment status. For the question of medication non-adherence, we asked whether they adjusted or forgot to take any medications in the few months prior. Demographic variables, such as sex, age, marital status, parity and region, were obtained from information collected previously by the survey company.

Cronbach's alpha, interfactor correlation coefficients and the association between latent factors and non-adherence were estimated in addition to the hospital-based survey. As potentially confounding variables to estimate adjusted OR, the types of medications taken (antihypertensive, antihyperlipidaemic, antidiabetic, anti-anxiety, anti-allergic, antacids or stomach medications and steroids), education (university or graduate school degree, some college, vocational school, high school graduate and grade 12 or below) and employment status (full time, part-time, homemaker, retired and not working) were considered in addition to the essential variables in the hospital-based survey.

Results

  1. Top of page
  2. Summary
  3. What is known and objective
  4. Methods
  5. Results
  6. Discussion
  7. What is new and conclusion
  8. Acknowledgements
  9. Funding
  10. References

Subject characteristics

For the hospital-based survey, 264 patients taking regular oral medications for chronic diseases were enrolled. Of these, 82 failed to meet the inclusion criteria (30 for poor performance, 25 for poor understanding, 20 for planned discharge and 7 for difficulty in conversing), 20 did not agree to participate in the survey and 13 did not complete the questionnaire. The remaining 149 subjects, including 110 patients with oral medications and 39 patients receiving injectable systemic chemotherapy, completed the questionnaire. Of the 110 patients with oral medications, 108 patients answered the medication-taking practices survey. Among these, 29 (26·9%) patients were adherent. Of the remaining 79 non-adherent cases, 7 (8·9%) involved intentional medication adjustments alone, 44 (55·7%) resulted from unintentional forgetfulness alone and 28 (35·4%) exhibited both intentional and unintentional behaviours. Of 20 300 registrants invited, 12 281 subjects responded (response rate 60·5%) and 524 eligible consenting participants completed the online survey. Adherence was observed in 168 (32·1%) cases. Intentional medication adjustment was the sole cause of non-adherence in 15 (4·2%) cases, whereas 196 (55·1%) non-adherent cases were attributed to unintentional forgetfulness alone. Both intentional non-adherence and unintentional non-adherence were observed in 145 (40·7%) cases. In summary, the number of patients included in the adherent, unintentional non-adherent and intentional non-adherent groups was 29, 44 and 35, respectively, in the hospital-based survey and 168, 196 and 160, respectively, in the online survey.

Table 1 shows the characteristics of both sets of survey participants. Compared with the hospitalized patients, the online survey respondents tended to be younger due to differences in age inclusion criteria, took fewer medication types and reported better performance status.

Table 1. Subject characteristics
CharacteristicHospital-based survey, n (%)Online survey, n (%) (= 524)
Patients with oral medication (= 110)Patients with chemotherapy (= 39)
  1. a

    Data on disease in the hospital-based survey and the online survey were obtained from medical records and self-reported information, respectively.

  2. b

    Hospitalization in the hospital-based survey did not include the current hospital admission.

Sex: male68 (61·8)20 (51·3)337 (64·3)
Age (years)
35–4413 (11·8)1 (2·6)104 (19·8)
45–5412 (10·9)7 (17·9)104 (19·8)
55–6428 (25·5)15 (38·5)148 (28·2)
65–7430 (27·3)16 (41·0)168 (32·1)
≥7527 (24·5)0 (0·0)
Oral medication taken chronically
1–3 types40 (36·4)337 (64·3)
4–6 types33 (30·0)124 (23·7)
7–9 types21 (19·1)35 (6·7)
≥10 types15 (13·6)28 (5·3)
Antihypertensive234 (44·7)
Antihyperlipidaemic120 (22·9)
Antidiabetic65 (12·4)
Anti-anxiety84 (16·0)
Antacids or stomach72 (13·7)
Anti-allergic60 (11·5)
Steroids14 (2·7)
<1 year use24 (21·8)
<5 years use308 (58·8)
Diseasea
Metabolic disease81 (73·6)8 (20·5)278 (53·1)
Diabetes mellitus60 (40·9)6 (15·4)69 (13·2)
Hyperlipidaemia19 (17·3)3 (7·7)120 (22·9)
Cardiovascular disease52 (47·3)8 (20·5)297 (56·7)
Hypertension46 (41·8)8 (20·5)234 (44·7)
Blood disease1 (0·9)12 (30·8)0 (0·0)
Gastrointestinal disease5 (4·5)3 (7·7)45 (8·6)
Liver/gallbladder/pancreatic disease37 (33·6)4 (10·3)11 (2·1)
Hepatocellular carcinoma25 (22·7)0 (0·0)0 (0·0)
Renal disease14 (12·7)7 (17·9)6 (1·1)
Renal failure14 (12·7)1 (2·6)1 (0·2)
Respiratory disease4 (3·6)3 (7·7)28 (5·3)
Endocrine disease2 (1·8)3 (7·7)31 (5·9)
Psychiatric disease1 (0·9)0 (0·0)70 (13·4)
Neuromuscular disease7 (6·4)0 (0·0)10 (1·9)
Gynaecologic disease0 (0·0)8 (20·5)14 (2·7)
Performance status = 060 (54·5)14 (35·9)434 (82·8)
Experienced adverse drug reaction37 (33·6)37 (94·9)212 (40·5)
Experienced hospitalizationb74 (67·3)33 (84·6)387 (73·9)
Married79 (71·8)31 (79·5)414 (79·0)
Children89 (80·9)35 (89·7)394 (75·2)
Education
University or graduate degree211 (40·3)
Some college58 (11·1)
Vocational school47 (9·0)
High school graduate188 (35·9)
<Grade 1219 (3·6)
Employment status
Working full time217 (41·4)
Working part-time58 (11·1)
Homemaker88 (16·8)
Retired96 (18·3)
Currently not working56 (10·7)

Identified latent factors

Factor analysis revealed four medication- and treatment-related latent factors, which were as follows: (i) Dissatisfaction, (ii) Barrier, (iii) Willingness and (iv) Endurance. Table 2 shows the survey items making up each latent factor, the factor loading of respective items and the frequency distribution for survey responses. The Dissatisfaction factor addressed items of patient dissatisfaction with treatment. For these items, the factor loading of items about interpersonal relationships was higher, meaning that the Dissatisfaction factor had large implications for relationships. The Barrier factor entailed items regarding medication harm, dependence and burden. The Willingness factor included variables reflecting positive willingness and attitude towards medications and treatment. The Endurance factor involved items regarding tolerance of adverse drug reactions.

Table 2. Identified latent factors and the items included in the various latent factors
Latent factor/itemFactor loadingaFrequency distribution, n (%)
SurveyStrongly disagreeDisagreeUncertainAgreeStrongly agreeP-valueb
  1. a

    Factor loading was estimated using unweighted least squares with promax rotation in the hospital-based survey.

  2. b

    Mann–Whitney U-test.

  3. c

    Items opposite to implication of latent factor.

  4. d

    Hospital-based survey, = 149.

  5. e

    Online survey, = 524.

Dissatisfaction
I feel comfortable with my relationship with the medical staffc−0·76Hospitald1 (0·7)2 (1·3)19 (12·8)36 (24·2)91 (61·1)<0·001
Onlinee12 (2·3)29 (5·5)148 (28·2)249 (47·5)86 (16·4)
The medical staff understands my feeling about the treatment (medication)c−0·72Hospitald4 (2·7)5 (3·4)45 (30·2)33 (22·1)62 (41·6)<0·001
Onlinee14 (2·7)36 (6·9)186 (35·5)208 (39·7)80 (15·3)
The medical staff respects my feeling about the treatment (medication)c−0·61Hospitald0 (0·0)1 (0·7)46 (30·9)33 (22·1)69 (46·3)<0·001
Onlinee7 (1·3)24 (4·6)215 (41·0)202 (38·5)76 (14·5)
The treatment (medication) schedule matches my life patternc−0·52Hospitald6 (4·0)6 (4·0)33 (22·1)33 (22·1)71 (47·7)0·17
Onlinee3 (0·6)16 (3·1)114 (21·8)220 (42·0)171 (32·6)
I feel effects of the treatment (medication)c−0·42Hospitald3 (2·0)3 (2·0)29 (19·5)43 (28·9)71 (47·7)<0·001
Onlinee5 (1·0)18 (3·4)119 (22·7)274 (52·3)108 (20·6)
Barrier
The treatment (medication) is toxic to my body0·56Hospitald60 (40·3)16 (10·7)53 (35·6)9 (6·0)11 (7·4)<0·001
Onlinee50 (9·5)109 (20·8)244 (46·6)84 (16·0)37 (7·1)
I am afraid I may become dependent on the treatment (medication)0·54Hospitald45 (30·2)14 (9·4)42 (28·2)19 (12·8)29 (19·5)0·02
Onlinee69 (13·2)83 (15·8)147 (28·1)165 (31·5)60 (11·5)
I feel worried about taking many kinds of medications0·47Hospitald40 (26·8)10 (6·7)31 (20·8)28 (18·8)40 (26·8)<0·001
Onlinee23 (4·4)27 (5·2)93 (17·7)187 (35·7)194 (37·0)
I feel secure about receiving the treatment (taking medication)c0·42Hospitald1 (0·7)0 (0·0)13 (8·7)31 (20·8)104 (69·8)<0·001
Onlinee7 (1·3)21 (4·0)131 (25·0)251 (47·9)114 (21·8)
The side effects of the treatment (medication) are troublesome0·40Hospitald12 (8·1)10 (6·7)47 (31·5)24 (16·1)56 (37·6)<0·001
Onlinee74 (14·1)98 (18·7)174 (33·2)104 (19·8)74 (14·1)
It is a burden to receive treatment (to take medication or injections)0·37Hospitald58 (38·9)22 (14·8)22 (14·8)20 (13·4)27 (18·1)<0·001
Onlinee62 (11·8)87 (16·6)156 (29·8)139 (26·5)80 (15·3)
Willingness
I feel positive about receiving the treatment (medication)0·61Hospitald0 (0·0)0 (0·0)22 (14·8)36 (24·2)91 (61·1)<0·001
Onlinee2 (0·4)16 (3·1)172 (32·8)227 (43·3)107 (20·4)
I expect to see the desired effects of the treatment (medication)0·57Hospitald0 (0·0)1 (0·7)7 (4·7)24 (16·1)117 (78·5)<0·001
Onlinee5 (1·0)12 (2·3)47 (9·0)245 (46·8)215 (41·0)
I have confidence in continuing the treatment (medication)0·49Hospitald0 (0·0)2 (1·3)15 (10·1)38 (25·5)94 (63·1)<0·001
Onlinee7 (1·3)16 (3·1)101 (19·3)259 (49·4)141 (26·9)
I have confidence in following the advice provided by the medical staff0·44Hospitald0 (0·0)0 (0·0)11 (7·4)35 (23·5)103 (69·1)<0·001
Onlinee3 (0·6)28 (5·3)124 (23·7)259 (49·4)110 (21·0)
I want to continue the treatment (medication)0·43Hospitald14 (9·4)3 (2·0)24 (16·1)21 (14·1)87 (58·4)<0·001
Onlinee48 (9·2)83 (15·8)153 (29·2)176 (33·6)64 (12·2)
Endurance
The side effects are acceptable if the treatment (medication) is effective0·98Hospitald36 (24·2)10 (6·7)49 (32·9)24 (16·1)30 (20·1)<0·001
Onlinee124 (23·7)152 (29·0)161 (30·7)72 (13·7)15 (2·9)
I have confidence that I will endure the side effects of the treatment (medication)0·53Hospitald21 (14·1)12 (8·1)52 (34·9)31 (20·8)33 (22·1)<0·001
Onlinee66 (12·6)110 (21·0)242 (46·2)86 (16·4)20 (3·8)

Cronbach's alpha values showed acceptable values for all factors in both surveys including the marginally lower value of Endurance in the online survey. The values were as follows: Dissatisfaction, 0·77; Barrier, 0·64; Willingness, 0·66; and Endurance, 0·72, in the hospital-based survey; and Dissatisfaction, 0·77; Barrier, 0·60; Willingness, 0·68; and Endurance, 0·61, in the online survey. Of the factors, negative interfactor correlation was observed for the following pairs in the surveys: Willingness and Dissatisfaction (= −0·32 and −0·41, respectively) and Willingness and Barrier (= −0·26 and −0·16, respectively). The nearly identical values of the alpha and the correlation in the surveys meant that the four latent factors were valid, indicating that the MAPAS consisting of the items shown in Table 2 was a valid questionnaire.

Latent factor score and medication non-adherence

Figure 1 presents a comparison of the standardized scores of latent factors between the three groups of adherence, unintentional non-adherence and intentional non-adherence. Among both hospitalized patients and online survey respondents, intentionally non-adherent patients demonstrated significantly higher mean Dissatisfaction scores than adherent patients. Compared with the adherent group of participants in the online survey, the mean differential scores for Willingness–Dissatisfaction and Willingness–Barrier were significantly lower in the intentional non-adherent group. A similar but statistically non-significant trend was also observed in the hospitalized patients.

image

Figure 1. Comparison of latent factor scores between adherent, unintentional non-adherent and intentional non-adherent groups. Each box depicts the interquartile range (IQR), and the horizontal line within the box indicates the median. The upper whisker represents the maximum value within 1·5 times the IQR value added to the 75th percentile. The lower whisker signifies the minimum value within 1·5 times the IQR subtracted from the 25th percentile. Circles depict outliers, and notched circles represent extreme values. The * and ** symbols indicate < 0·05 and < 0·01, respectively, in the Tukey's tests for nonparametric multiple comparisons of the differences in the mean values. The numbers of patients comprising the adherent, unintentionally non-adherent and intentionally non-adherent groups were 29, 44 and 35, respectively, in the hospital-based survey and 168, 196 and 160, respectively, in the online survey.

Download figure to PowerPoint

Table 3 shows the adjusted ORs quantifying the association between medication non-adherence and the latent factors. Higher scores in Dissatisfaction were significantly associated with a greater risk of intentional non-adherence in both hospitalized patients (95% CI of adjusted OR 1·20–16·26, = 0·03) and online survey respondents (95% CI of adjusted OR 1·33–3·45, = 0·002). Higher scores for Willingness–Dissatisfaction and Willingness–Barrier were inversely associated with a reduced odds of intentional non-adherence (hospital-based survey, = 0·02 for both differential scores; online survey, < 0·001 for both differential scores). On the other hand, unintentional non-adherence was not associated with any factor or differential with the exception of the Endurance factor in the online survey.

Table 3. Association between latent factors and medication non-adherence
Latent factorHospital-based surveyOnline survey
Adherence, n (%)Non-adherence, n (%)Adjusted OR (95% CI)P-valueAdherence, n (%)Non-adherence, n (%)Adjusted OR (95% CI)P-value
  1. a

    Classification was categorized according to the median.

  2. b

    Adjusted for sex, number of different oral medications, age and the latent factor or differential score.

  3. c

    Adjusted for use of steroid agents, duration of medication use, adverse drug reactions, university or graduate school degree and the latent factor or differential score.

  4. d

    Adjusted for sex, number of different oral medications, age, Dissatisfaction score and the latent factor score.

  5. e

    Adjusted for use of steroid agents, duration of medication use, adverse drug reactions, university or graduate school degree, Dissatisfaction score and the latent factor score.

  6. f

    Classification was categorized according to the quartile.

  7. g

    P-value for a trend was estimated by substituting a continuum of ordinal values for each of the four categories (i.e. 1, 2, 3 and 4).

  8. h

    Adjusted for sex, duration of medication use and the latent factor or differential score in 72 cases (one case had a missing value).

  9. i

    Adjusted for use of antihypertensive agents, duration of medication use, university or graduate school degree, Endurance score and the latent factor or differential score.

  10. j

    Adjusted for use of antihypertensive agents, duration of medication use, university or graduate school degree and the latent factor score.

  11. OR, odds ratio; CI, confidence interval.

Intentional non-adherence = 29= 35  = 168= 160  
Dissatisfactiona
1 (low)18 (62·1)14 (40·0)1·00 (Reference)b0·0379 (47·0)54 (33·8)1·00 (Reference)c0·002
2 (high)11 (37·9)21 (60·0)4·43 (1·20–16·26)89 (53·0)106 (66·3)2·14 (1·33–3·45)
Barriera
1 (low)16 (55·2)13 (37·1)1·00 (Reference)d0·1170 (41·7)55 (34·4)1·00 (Reference)e0·54
2 (high)13 (44·8)22 (62·9)2·88 (0·78–10·61)98 (58·3)105 (65·6)1·16 (0·72–1·89)
Willingnessa
1 (low)11 (37·9)15 (42·9)1·00 (Reference)d0·6065 (38·7)81 (50·6)1·00 (Reference)e0·12
2 (high)18 (62·1)20 (57·1)0·69 (0·17–2·82)103 (61·3)79 (49·4)0·67 (0·40–1·11)
Endurancea
1 (low)19 (65·5)22 (62·9)1·00 (Reference)d0·4087 (51·8)83 (51·9)1·00 (Reference)e0·88
2 (high)10 (34·5)13 (37·1)0·55 (0·14–2·17)81 (48·2)77 (48·1)0·96 (0·61–1·53)
Willingness–Dissatisfactionf
1 (low)5 (17·2)15 (42·9)1·00 (Reference)b0·02g36 (21·4)57 (35·6)1·00 (Reference)c<0·001g
28 (27·6)6 (17·1)0·34 (0·06–1·85)47 (28·0)42 (26·3)0·60 (0·33–1·12)
36 (20·7)7 (20·0)0·32 (0·05–2·15)39 (23·2)38 (23·8)0·54 (0·28–1·02)
4 (high)10 (34·5)7 (20·0)0·12 (0·02–0·70)46 (27·4)23 (14·4)0·23 (0·11–0·47)
Willingness–Barrierf
1 (low)4 (13·8)12 (34·3)1·00 (Reference)b0·02g35 (20·8)61 (38·1)1·00 (Reference)c<0·001g
26 (20·7)7 (20·0)0·53 (0·08–3·48)49 (29·2)42 (26·3)0·51 (0·28–0·94)
311 (37·9)9 (25·7)0·18 (0·03–1·05)40 (23·8)34 (21·3)0·45 (0·23–0·86)
4 (high)8 (27·6)7 (20·0)0·13 (0·02–0·83)44 (26·2)23 (14·4)0·25 (0·13–0·50)
Unintentional non-adherence = 29= 44 = 168= 196 
Dissatisfactiona
1 (low)18 (62·1)27 (61·4)1·00 (Reference)h0·4779 (47·0)95 (48·5)1·00 (Reference)i0·67
2 (high)11 (37·9)17 (38·6)0·63 (0·19–2·16)89 (53·0)101 (51·5)0·91 (0·59–1·40)
Barriera
1 (low)16 (55·2)17 (38·6)1·00 (Reference)h0·2270 (41·7)92 (46·9)1·00 (Reference)i0·48
2 (high)13 (44·8)27 (61·4)2·05 (0·65–6·49)98 (58·3)104 (53·1)0·85 (0·56–1·32)
Willingnessa
1 (low)11 (37·9)22 (50·0)1·00 (Reference)h0·6965 (38·7)65 (33·2)1·00 (Reference)i0·25
2 (high)18 (62·1)22 (50·0)0·79 (0·25–2·52)103 (61·3)131 (66·8)1·30 (0·83–2·05)
Endurancea
1 (low)19 (65·5)25 (56·8)1·00 (Reference)h0·8587 (51·8)123 (62·8)1·00 (Reference)j0·02
2 (high)10 (34·5)19 (43·2)1·12 (0·36–3·51)81 (48·2)73 (37·2)0·60 (0·39–0·93)
Willingness–Dissatisfactionf
1 (low)5 (17·2)9 (20·5)1·00 (Reference)h0·69g36 (21·4)36 (18·4)1·00 (Reference)i0·58g
28 (27·6)12 (27·3)1·19 (0·21–6·62)47 (28·0)53 (27·0)1·24 (0·66–2·34)
36 (20·7)12 (27·3)2·39 (0·38–15·00)39 (23·2)57 (29·1)1·48 (0·78–2·79)
4 (high)10 (34·5)11 (25·0)1·28 (0·24–6·93)46 (27·4)50 (25·5)1·17 (0·62–2·22)
Willingness–Barrierf
1 (low)4 (13·8)14 (31·8)1·00 (Reference)h0·21g35 (20·8)37 (18·9)1·00 (Reference)i0·34g
26 (20·7)12 (27·3)1·19 (0·21–6·69)49 (29·2)50 (25·5)0·87 (0·46–1·64)
311 (37·9)7 (15·9)0·34 (0·07–1·82)40 (23·8)47 (24·0)0·97 (0·50–1·87)
4 (high)8 (27·6)11 (25·0)0·49 (0·10–2·32)44 (26·2)62 (31·6)1·27 (0·67–2·38)

Discussion

  1. Top of page
  2. Summary
  3. What is known and objective
  4. Methods
  5. Results
  6. Discussion
  7. What is new and conclusion
  8. Acknowledgements
  9. Funding
  10. References

This study demonstrated that patients' dissatisfaction with aspects of treatment including their relationships with their medical staff led them to intentionally adjust their medication regardless of whether they were treated in high-acuity or low-acuity settings. In contrast, the finding was not evident in patients with unintentional non-adherence.

Our findings are supported by a study by Sewitch et al.,[26] which evaluated a population with inflammatory bowel disease. In their study, intentional non-adherence tended to occur in cases of great discordance between the patient's and physician's perceptions regarding the patient's health status and their communication. In keeping with their findings, our study suggests that sincere efforts for partnership between patients and their medical staff may be particularly beneficial in improving intentional non-adherence. Medical staff should pay attention to potential causes for dissatisfaction among their patients.

Another noteworthy finding is that individuals who do not find subliminally sufficient levels of positive attitude (i.e. willingness for treatment) vs. negative thoughts (i.e. dissatisfaction and barrier) tended towards intentional non-adherence, but not unintentional non-adherence. These observations are consistent with the findings of two studies by Clifford et al.[9] and Wroe,[11] which suggested that such rational decision-making affects intentional behaviour. However, neither study examined the patient–clinician interpersonal relationships. Our results suggest that dissatisfaction with medical staff influences patient's rational judgments for intentional non-adherence. Further studies should clarify whether negative factors contribute more strongly to intentional non-adherence, patient dissatisfaction or barrier issues such as concerns over medication toxicity and dependence.

This study has several limitations. First, self-reported measures of non-adherence may be unreliable.[27] However, self-reports are still valuable in that they provide insight into the underlying causes for non-adherence[28] and assist in distinguishing between intentional and unintentional behaviours.

Secondly, the intentionally non-adherent group in this study included patients with both intentional and unintentional non-adherence. As such, our results may not show the characteristics of genuinely intentionally non-adherent patients. However, combining both types of non-adherent patients may weaken the unique characteristics of patients with intentional behaviour; that is, a group comprising only intentionally non-adherent patients may have generated a higher OR for dissatisfaction. On the other hand, given that most intentionally non-adherent patients often forget to take their medications,[8, 26] our study may be valuable for developing strategies to improve upon non-adherence in a real-world setting.

Thirdly, we assessed non-adherence to at least one medication, but not to a specific medication. Medication type may influence the frequency of non-adherence. Fourth, in the hospital-based study, we used hospitalized patients managed by pharmacists, which may not be representative of all patients with chronic disease. Accounting for this weakness, we conducted the nationwide online survey of only non-hospitalized individuals, recognizing that online survey respondents generally tend to be wealthier and more educated.[29] However, we ended up obtaining the same results in the two populations, suggesting that our findings are robust and highly generalizable. On the other hand, note that these subjects were a subset of patients who met the inclusion criteria, which may potentially affect the generalizability or external validity of our conclusions.

What is new and conclusion

  1. Top of page
  2. Summary
  3. What is known and objective
  4. Methods
  5. Results
  6. Discussion
  7. What is new and conclusion
  8. Acknowledgements
  9. Funding
  10. References

In conclusion, our study suggests that patients' dissatisfaction with treatment including their relationships with their medical staff as well as their subliminal rational judgment are unique, but critical and consistent determinants of intentional non-adherence to medications.

Acknowledgements

  1. Top of page
  2. Summary
  3. What is known and objective
  4. Methods
  5. Results
  6. Discussion
  7. What is new and conclusion
  8. Acknowledgements
  9. Funding
  10. References

We thank Tetsuko Goda, Hiroshi Muguruma, Kana Sumiyoshi, Eitoku Matsuoka, Katsuhiko Kamei, Kiyo Suzuki, Yoshie Isobe, Nami Mizukawa and Kumiko Kanzaki for their assistance with data collection.

Funding

  1. Top of page
  2. Summary
  3. What is known and objective
  4. Methods
  5. Results
  6. Discussion
  7. What is new and conclusion
  8. Acknowledgements
  9. Funding
  10. References

This work was supported by a research grant from the Pfizer Health Research Foundation. The funding source had no role in the collection or analysis of the study data.

References

  1. Top of page
  2. Summary
  3. What is known and objective
  4. Methods
  5. Results
  6. Discussion
  7. What is new and conclusion
  8. Acknowledgements
  9. Funding
  10. References
  • 1
    van Eijken M, Tsang S, Wensing M, de Smet PA, Grol RP. Interventions to improve medication compliance in older patients living in the community: a systematic review of the literature. Drugs Aging, 2003;20:229240.
  • 2
    Osterberg L, Blaschke T. Adherence to medication. N Engl J Med, 2005;353:487497.
  • 3
    Simpson SH, Eurich DT, Majumdar SR et al. A meta-analysis of the association between adherence to drug therapy and mortality. BMJ, 2006;333:1520.
  • 4
    Howard RL, Avery AJ, Howard PD, Partridge M. Investigation into the reasons for preventable drug related admissions to a medical admissions unit: observational study. Qual Saf Health Care, 2003;12:280285.
  • 5
    Sokol MC, McGuigan KA, Verbrugge RR, Epstein RS. Impact of medication adherence on hospitalization risk and healthcare cost. Med Care, 2005;43:521530.
  • 6
    Haynes RB, Ackloo E, Sahota N, McDonald HP, Yao X Interventions for enhancing medication adherence. Cochrane Database Syst Rev, 2008;CD000011.
  • 7
    Lindquist LA, Go L, Fleisher J, Jain N, Friesema E, Baker DW. Relationship of health literacy to intentional and unintentional non-adherence of hospital discharge medications. J Gen Intern Med, 2012;27:173178.
  • 8
    Iihara N, Kurosaki Y, Miyoshi C, Takabatake K, Morita S, Hori K. Comparison of individual perceptions of medication costs and benefits between intentional and unintentional medication non-adherence among Japanese patients. Patient Educ Couns, 2008;70:292299.
  • 9
    Clifford S, Barber N, Horne R. Understanding different beliefs held by adherers, unintentional nonadherers, and intentional nonadherers: application of the Necessity-Concerns Framework. J Psychosom Res, 2008;64:4146.
  • 10
    Atkins L, Fallowfield L. Intentional and non-intentional non-adherence to medication amongst breast cancer patients. Eur J Cancer, 2006;42:22712276.
  • 11
    Wroe AL. Intentional and unintentional nonadherence: a study of decision making. J Behav Med, 2002;25:355372.
  • 12
    George J, Mackinnon A, Kong DC, Stewart K. Development and validation of the Beliefs and Behaviour Questionnaire (BBQ). Patient Educ Couns, 2006;64:5060.
  • 13
    Atkinson MJ, Kumar R, Cappelleri JC, Hass SL. Hierarchical construct validity of the treatment satisfaction questionnaire for medication (TSQM version II) among outpatient pharmacy consumers. Value Health, 2005;8(Suppl):S9S24.
  • 14
    Oken M, Creech R, Tormey D et al. Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol, 1982;5:649655.
  • 15
    Carr AJ, Thompson PW, Cooper C. Factors associated with adherence and persistence to bisphosphonate therapy in osteoporosis: a cross-sectional survey. Osteoporos Int, 2006;17:16381644.
  • 16
    Brus H, van de Laar M, Taal E, Rasker J, Wiegman O. Determinants of compliance with medication in patients with rheumatoid arthritis: the importance of self-efficacy expectations. Patient Educ Couns, 1999;36:5764.
  • 17
    Horne R, Weinman J. patients' beliefs about prescribed medicines and their role in adherence to treatment in chronic physical illness. J Psychosom Res, 1999;47:555567.
  • 18
    Sewitch MJ, Abrahamowicz M, Dobkin PL, Tamblyn R. Measuring differences between patients' and physicians' health perceptions: the patient-physician discordance scale. J Behav Med, 2003;26:245264.
  • 19
    Horne R, Weinman J, Hankins M. The beliefs about medicines questionnaire: the development and evaluation of a new method for assessing the cognitive representation of medication. Psychol Health, 1999;14:124.
  • 20
    DiMatteo M, Hays R, Gritz E et al. Patient adherence to cancer control regimens: scale development and initial validation. Psychol Assess, 1993;5:102112.
  • 21
    Hogan TP, Awad AG, Eastwood R. A self-report scale predictive of drug compliance in schizophrenics: reliability and discriminative validity. Psychol Med, 1983;13:177183.
  • 22
    Cronbach L. Coefficient alpha and the internal structure of tests. Psychometrika, 1951;16:297334.
  • 23
    Fayers P, Machin D. Quality of life: assessment, analysis and interpretation. Chichester: John Wiley & Sons, 2000.
  • 24
    DeVellis R. Scale development: theory and applications. California: SAGE Publications, 1991.
  • 25
    Tango T. Statistics to medicine. Tokyo: Asakura Publishing, 2004.
  • 26
    Sewitch MJ, Abrahamowicz M, Barkun A et al. Patient nonadherence to medication in inflammatory bowel disease. Am J Gastroenterol, 2003;98:15351544.
    Direct Link:
  • 27
    Garber MC, Nau DP, Erickson SR, Aikens JE, Lawrence JB. The concordance of self-report with other measures of medication adherence: a summary of the literature. Med Care, 2004;42:649652.
  • 28
    George J, Kong DC, Stewart K. Adherence to disease management programs in patients with COPD. Int J Chron Obstruct Pulmon Dis, 2007;2:253262.
  • 29
    Tokuda Y, Okubo T, Yanai H, Doba N, Paasche-Orlow MK. Development and validation of a 15-item Japanese Health Knowledge Test. J Epidemiol, 2010;20:319328.