Therapy satisfaction and health literacy are key factors to improve medication adherence in systemic sclerosis

Objective Although medication adherence (MA) contributes to therapeutic outcome in systemic sclerosis (SSc), research data are scarce. Factors influencing MA in SSc are hardly known. Method We conducted a monocentric, cross-sectional study on 85 patients with SSc at the University of Lübeck, Germany, using the Compliance Questionnaire of Rheumatology as the main measurement tool of MA. We also used the Scleroderma Health Assessment Questionnaire, Illness Perception Questionnaire – Revised, Health Literacy Questionnaire, Lübeck Medication Satisfaction Questionnaire (a novel instrument created for this study), and patients’ demographic and clinical data, to find factors contributing to MA. Results Good MA was seen in 51.8% of patients. MA was positively associated with therapy satisfaction (p < 0.001), modified Rodnan Skin Score (p = 0.032), age (p = 0.025), intake of micronutrients (p = 0.033), number of prescribed drugs (p = 0.014), and some dimensions of health literacy. Negative associations were found for patients with weight loss attributed to SSc (p = 0.009) and the perception that the disease is caused by the patient’s personality (p = 0.011) or emotional state (p = 0.037). Conclusion Although most SSc patients display good MA, non-adherence remains a major problem. Patients should be assessed for non-adherence. The factors affecting MA identified herein could help to improve therapeutic outcomes.

Systemic sclerosis (SSc) is a severe multi-systemic autoimmune disease, having the highest disease-related mortality among all rheumatic diseases (1). Interstitial lung disease (ILD), pulmonary arterial hypertension (PAH), and cardiac problems are the main causes of death (2). The disease is not only rare, but also very heterogeneous in its clinical manifestations and, therefore, often misdiagnosed and mistreated (3). Although therapy schemes have improved in the past years, clinical outcome still remains disappointing for many patients (4). Patients often find themselves restricted in their personal and work lives, resulting in occupational difficulties and loss of social status (5,6). A decline in the patients' quality of life is not uncommon (7). It is well known that better medication adherence (MA) leads not only to better clinical outcomes and increased quality of life for patients, but also to lower healthcare costs in any disease (8). Therefore, improving MA should be one of the primary goals in the management of SSc. However, the problem has not yet been sufficiently examined. Only a few studies have been published on this topic. Those studies that address the problem vary in their methods and differ in terms of results (9)(10)(11). Factors influencing MA have not yet been identified. Our goal in this study was to measure the degree of MA of patients with SSc, as well as to pinpoint factors influencing MA.

Patients
We conducted a cross-sectional study on consecutive SSc patients at the Department of Rheumatology and Clinical Immunology of the University Hospital Lübeck, Germany. Data were acquired between July 2020 and February 2021. Patients were recruited on the ward as well as during the weekly SSc outpatient clinic. They were included if they fulfilled the European League Against Rheumatism (EULAR)/American College of Rheumatology (ACR) diagnostic criteria (12) and were under medical treatment for SSc. Patients who legally could not consent to the study and patients unable to complete the questionnaire because of physical impairment, illiteracy, or language barriers were excluded. The patients were informed about the purpose and procedures of the study, as well as their legal rights. Written consent was duly obtained from all participants. Patients were also asked to provide information about their age, gender, native language, migration background, marital status, other members of their household, religion, highest educational degree, and current employment status. In addition, clinical data were extracted from patients' medical records. Specifically, these included: time since the onset of SSc, type of SSc, the presence of anti-Scl-70 and anti-centromere antibodies, presence of PAH, prescribed drugs, forced vital capacity (FVC), and modified Rodnan Skin Score (mRSS). This study was approved by the University of Lübeck ethics board on 27 May 2020 (reference number 20-163).

Patient participation
Initially, 88 patients were asked to participate in this study. Two patients declined to participate for reasons unrelated to the study, resulting in a response rate of 97.7%. Nine patients did not fully complete their questionnaires. Since the various sections of the questionnaires could be evaluated independently, it was decided to include all participating SSc patients, irrespective of the completeness for all clinical data. Descriptive statistics of demographic and clinical data are presented in Table 1.

Applied questionnaires
Patients were asked to independently complete several questionnaires. The Compliance Questionnaire of Rheumatology (CQR) is a self-administered questionnaire consisting of 19 statements rated by the patients on a four-point Likert scale. It is used to assess MA in rheumatic diseases. It was developed and published by de Klerk et al in 1999 (13). Four years later they published an adjusted, weighted version to discriminate more precisely between well-adherent and non-adherent patients (14). That version was validated against Medication Event Monitoring Systems (MEMS) -a wellaccepted gold standard in MA assessment. Using these two versions of evaluation, the score can be used either as a continuous measurement (non-weighted CQR score) or as a dichotomous measurement (weighted CQR score) of MA. The dichotomous score shows a sensitivity of 95% and a specificity of 62% for good adherence (14). Good adherence in this case is defined as taking > 80% of the prescribed doses, as proposed by Sackett et al (15). The CQR is a well-accepted tool in the field of adherence research in rheumatology (9,11,16).
Since the questionnaire had not been available in Germany beforehand, it was translated into German in preparation for this study. The translation was led by the guidelines proposed by Beaton et al (17). In the first step, two native translators independently translated the original English version to German. One of the translators had medical experience whereas the other one did not. Afterwards, the two translators merged their translations into one combined translation upon which they both agreed. That version was then given to a backtranslator, who was a native English speaker with experience in the medical field. He prepared a backtranslation that was later compared to the English original.
The Scleroderma Health Assessment Questionnaire (SHAQ) was used to assess the degree of impairment in the patients' daily lives. In this self-administered questionnaire, patients are asked to quantify their impairment in certain situations of their daily lives on a four-point Likert scale. The 20 situations are divided into eight subcategories whose average score equates to the total SHAQ score. That part of the questionnaire is based on the commonly used Health Assessment Questionnaire (HAQ) (18). The SHAQ also features visual analogue scales to quantify the severity of symptoms specific to SSc (19).
The Lübeck Medication Satisfaction Questionnaire (LMSQ) was developed for this study. It features 18 statements concerning the patients' medication satisfaction. The patients are asked to quantify their agreement with the statements on a four-point Likert scale. The statements can be divided into six categories whose mean score amounts to the total LMSQ score. The six categories are: satisfaction with the side effects, perceived effectiveness of the therapy, practicability of the therapy, satisfaction with the freedom and independence gained through the therapy, satisfaction with the healthcare professionals, and overall therapy satisfaction. A more detailed description of the questionnaire can be found in the supplementary material. The Illness Perception Questionnaire -Revised (IPQ-R) was used to assess patients' assumptions about their disease. The questionnaire consists of two parts (20,21). In the first part, patients are asked to quantify their agreement with 32 statements about their disease on a five-point Likert scale. These statements are assigned to seven categories that reflect the dimensions of the assumptions. In the second part of the questionnaire, patients are given 18 different causes and are asked to rate the likelihood of the specific causes for their SSc on a five-point Likert scale. In this study, the German translation by Galli et al was used (22).
The Health Literacy Questionnaire (HLQ) was used to assess patients' health literacy. Health literacy is defined as 'the cognitive and social skills which determine the motivation and ability of individuals to gain access to, understand and use information in ways which promote and maintain good health' (23). The HLQ consists of 44 statements that are to be answered by the patients on four-point and fivepoint Likert scales. The HLQ can be used to differentiate between nine dimensions of health literacy (24). In this study, the German translation of Nolte et al was used (25).

Statistical analyses
Statistical analysis was performed using Jamovi 1.2.27.0. MA was defined as the dependent variable. The other factors were defined as the independent variables. To take the various scale levels into account, chisquared tests were used to examine the relationship between qualitative variables and MA (presented as weighted CQR score) and Kendall's correlation was used to examine the relationship between quantitative variables and MA (presented as non-weighted CQR score). The results are presented in Tables 2 and 3, respectively. Significance was tested for a two-sided hypothesis and the significance level was set to α = 0.05. In addition, multiple linear regression was performed between CQR score as the dependent variable and the LMSQ subscales. The results are presented in Table 4 Results MA was assessed using the weighted CQR score. It was determined for 83 patients. A cut-off value of −0.58489 was used to distinguish between good and unsatisfactory adherence, as proposed by de Klerk et al (14). We found that 56 patients (67.5%) showed good adherence while 27 patients (32.5%) did not. Taking into consideration the sensitivity (95%) and specificity (62%) of the weighted CQR score, an adjusted adherence rate of 51.8% was calculated using the formula 67.5 = 95 * x + (100 − 62) * (1 − x), x is the adjusted adherence rate.
A weak positive correlation (τ = 0.173, p = 0.025) was found between patients' age and their MA. Patients' gender, native language, migration background, family status and household members, religion, educational status, and employment status yielded no significant correlations with  Impairment due to SSc was assessed using the SHAQ. Sufficient data were available for 85 patients. The median SHAQ score was 1.5. Kendall's correlation showed no significant statistical associations between the total SHAQ score or its subscales and the CQR score.
Satisfaction with therapy was assessed using the LMSQ. In total, 82 full data sets were available for analysis. Highly significant correlations between MA and perceived treatment effectiveness (τ = 0.341, p < 0.001), treatment practicability (τ = 0.301, p < 0.001), improvement in daily life (τ = 0.403, p < 0.001), satisfaction with healthcare workers (τ = 0.32, p < 0.001), general treatment satisfaction (τ = 0.474, p < 0.001), and total LMSQ (τ = 0.432, p < 0.001) became apparent. The perceived side effects did not show any correlation with the CQR. The multiple linear regression model between the subscales and the CQR score showed a significant result (adjusted R 2 = 0.388, F = 9.45, p < 0.001). Out of all the LMSQ subscales, general satisfaction with the treatment (t = 3.189, p = 0.002) and satisfaction with the degree of participation in the patients' daily life (t = 2.427, p = 0.018) were most strongly associated with MA.
Patients' assumptions about their disease were assessed using the IPQ-R. Among the symptoms that were tested, only weight loss attributed to SSc showed a significant association (13/27 vs 43/56, p = 0.009). These patients exhibited lower levels of MA. SSc patients who attributed the disease to their own personality (τ = −0.219, p = 0.011) or their own emotional state (τ = −0.177, p = 0.037) also showed significantly lower levels of MA. Other causes were not significantly associated with altered MA levels. Other patients' assumptions about their disease did not show any significant association with the level of MA either.

Discussion
To the best of our knowledge, this is the most extensive study to assess and identify factors influencing MA in SSc. Other studies on MA were not able to pinpoint factors that could explain differences in MA in patients with SSc (9)(10)(11).
In 2012, Hromadkova et al published the first study on MA of patients with SSc (9). They found 17 out of 41 patients (41.5%) to be well adherent, using the weighted CQR score as a measurement tool. In their study, MA was not associated with demographic characteristics, health status, pharmacotherapy, or quality of life. Iudici et al used the four-item Morisky Medication Adherence Scale (MMAS-4) to assess MA to glucocorticoid therapy in SSc patients (10), and found that 59 out of 92 patients (64.1%) showed good adherence. They found patients who considered the therapy to be necessary and had only a few concerns about it, as well as older patients, to be the most adherent. Clinical data/ health status, demographic data, and physical impairment were not associated with MA. In 2019, Brijs et al determined good MA, measured by the weighted CQR score, in 19 out of 66 (28.8%) of their patients (11). The psychological, clinical, and demographic factors assessed in that study were not associated with MA. The methods used by these three working groups to assess MA, and especially the MMAS-4, overestimate the real MA rate (14,26). Therefore, it is fair to assume that out of the four studies, our patient cohort showed the highest degree of MA.
Our results demonstrate an association between satisfaction with therapy and MA, as well as between health literacy and MA. Being able to participate in activities of daily life and general treatment satisfaction play the biggest roles in this result. This does not come as a surprise, given the findings in other diseases (27,28). Side effects did not have any effect, however. This novel finding contradicts the common idea of side effects playing an important role in non-adherence. Neither satisfaction with therapy nor health literacy has been tested in previous studies on MA in SSc. The association between health literacy and better MA is no surprise either. Therapies in SSc often are highly complex. The ability to understand their underlying rationale by actively searching for and dealing with health information plays an important role here. The commonly expressed call for more patient empowerment and shared decision making in medicine is backed up by the data gathered in this study. Healthcare workers should see this as motivation to educate their patients and, more importantly, to empower them to educate themselves.
Clinical data are of lesser relevance for MA. Most of the symptoms experienced, as well as the impairment in daily life, play no role in MA. The involvement of the skin, measured by the mRSS, is an exception in this case, as it was positively associated with MA. One could speculate about skin involvement being very important to the patients and therefore giving them more motivation to adhere to their therapy plan. However, that finding has not yet been backed up by other studies.
As shown here, attributing the disease to one's emotions and personality is associated with unsatisfactory MA. This indicates an association between neuroticism and MA. Although this association has not been tested in this study, it can be seen in numerous other diseases (29,30). Therefore, it could be of relevance to search for and address those emotional and personal behaviours, which could help to improve MA.
We found older patients to be more adherent than younger ones. This could be explained by more routine in the life of elderly people. In addition, age has been demonstrated to lead to changes in personality (31). However, since the oldest patient in the group was 82 years old, the findings cannot necessarily be transferred to even older patients.
The data do not show any clear relationships between MA and therapy. Therefore, the correlation between the number of drugs and MA should not be overinterpreted. We did not check for forms of administration, dosing, or frequency of intake. Therefore, this study does not contradict the consensus in the mainstream literature, which is a harmful effect of polypharmacy on MA (32). It is also important to point out the low sample size of our study. Some of the results may be flawed because of this.
Whether the determined degree of MA is satisfactory or not is up for debate. The assessed degree of medication is higher than that assessed in any other SSc patient cohort (9)(10)(11). However, it is roughly at the same level as in other connective tissue diseases (33)(34)(35)(36). Nonetheless, we find almost half of the patients presenting as non-adherent to be unsatisfactory.
Our study has several limitations. First, it is important to point out that this was a monocentric study including only a limited number of patients. Validation using a second independent cohort would be valuable. However, there is evidence to suggest that these findings are transferable to other fields, since they correspond with other findings in studies dealing with MA in various diseases (27,28).
All of the data were gathered during the coronavirus disease 2019 (COVID-19) pandemic. This could have shaped patients' views of disease and therapy and therefore the results of this study. There could also be a bias in patient recruitment because patients were recruited in the hospital only. Especially during the pandemic, some patients were afraid to enter hospitals, which could be reflected in the patient collection and therefore altered results (37). In addition, owing to the exclusion criteria some patients with special characteristics had to be excluded. Nevertheless, we still consider the patient cohort to be representative.
Moreover, there could theoretically be a problem with our translation of the CQR. Although we made sure to follow international standards in our translation process, the resulting questionnaire is different to the original one, which may have altered our results.
Owing to the study design, there is no way of making a statement about any causal relationship between MA and the factors tested. The results found in the study are therefore only useful to shed light on associations, and any statements about causal relationships are therefore purely speculative. More research with a different study design would be required to further elucidate the field.
Compared to the three previous studies, our SSc patient cohort showed the highest rate of MA. Owing to the lack of comparable studies assessing factors associated with MA, we cannot determine the underlying reasons with certainty. However, given the findings presented here, we have reason to assume that the improvement in SSc therapies and the special care that we provide for SSc patients at our clinic positively influence the MA. SSc patients treated in our clinic express high levels of satisfaction with their therapy. As a tertiary referral centre for SSc, we have not only developed special expertise in the treatment of SSc patients but also implemented structures to improve patient care. Thus, in our experience, training physicians and nurses in specialized SSc patient care provides great benefits for patient education and empowerment.
The development of structures to improve health literacy, and especially therapy satisfaction, has only played a minor role so far. The findings presented here should be interpreted as a call to improve on this situation.

Conclusion
Although our data indicate that a majority of SSc patients adhere to therapy, non-adherence is still highly prevalent in patients with SSc. So far, interventions targeting MA have not shown much success in other diseases (38). The factors discovered in this study may be of use to identify at-risk patients and also provide indications for improvements in MA. More knowledge about the influence of the structures of patient care, as well as patients' personality traits, on MA in SSc patients could be crucial for improving therapeutic strategies. Further research addressing these questions should be conducted. In any case, medication non-adherence should be kept at the forefront of every physician's mind when determining possible causes for a lack of clinical improvement.