Chronic illnesses are usually prolonged illnesses that are managed throughout the course of a patient’s life and are not often completely cured (Dowrick et al., 2005). They may require either life-long or long-term medication therapy and the success of long-term treatments depends on proper adherence to the medication therapy regimen (World Health Organization, 2003; Dowrick et al., 2005; Naqvi et al., 2019). According to the Institute for Health Metrics and Evaluation (IHME), chronic illnesses remain the leading cause of death alone, as well as death and disability combined. Seven chronic diseases are listed among the top 10 causes of death among the Jordanian population (Institute of Health Metrics and Evaluation IHME, 2020).

One of the most common reasons for failing to achieve treatment outcomes by patients with chronic illnesses is non-adherence to medication therapy (Naqvi et al., 2020a). The World Health Organization (WHO) defines adherence as the as the degree to which a person’s medicine taking behaviour, lifestyle changes, and dietary habits align with the recommendations given by a healthcare professional (World Health Organization, 2003). Non-adherence to pharmacotherapy may result in disease complications, prolonged disease condition, increased risk for hospitalization, risk of disability and death; all of these may further exacerbate the healthcare costs (Naqvi et al., 2020a; Al-Qasem et al., 2011; Iuga and McGuire, 2014). Non-adherence continues to prevail in developing countries as evidence highlights that only 50% of patients with chronic illnesses adhere to the treatment (World Health Organization, 2003). A review by Al-Qasem et al. highlighted that the practice of non-adherence was prevalent among patients in the Middle Eastern countries. The rate of non-adherence to medications was reported to be between 1.4% and 88% (Al-Qasem et al., 2011).

In Jordan, medication non-adherence is a serious healthcare issue where most patients are non-adherent (Al-Sahouri et al., 2019). For example, in a study that evaluated the effectiveness of home medication management review on medication adherence in Jordanian patients, most patients had medium adherence at baseline (Al‐Qudah et al., 2018). Moreover, in another study among patients with chronic ailments in Jordan, it was reported that 46.1% were non-adherent (Basheti et al., 2016). Focusing on type 2 diabetes (T2DM), it was reported that slightly less than half of the surveyed T2DM patients (46.5%) had moderate adherence while 12.2% had low adherence to their anti-diabetic medications (Al-Qerem et al., 2021).

In the past, adherence to medication among patients in Jordan has been documented using the Arabic versions of the 8–item Morisky’s Medication Adherence Scale (MMAS—8), Beliefs about Medication Questionnaire (BMQ), and Medication Adherence Report Scale (MARS) (Awwad et al., 2015; Basheti et al., 2016; Alsous et al., 2017; Al‐Qudah et al., 2018; Al-Qerem et al., 2021; Al-Qerem et al., 2022). Recently, the Arabic version of the General Medication Adherence Scale (GMAS) was validated in patients with chronic diseases in Saudi Arabia and in patients with diabetes in Sudan and Morocco (Al-Qasem et al., 2011; Mahmoud et al., 2021; Maryem et al., 2023). This scale considers non-adherence due to the cost of medications which adds a financial aspect in reporting medication adherence. The concurrent validity of the Arabic version has been evaluated by correlating the adherence scores obtained by GMAS—AR with the ones obtained by the Arabic versions of Adherence to Refills and Medications Scale (ARMS) and MARS in Saudi Arabia (Islam et al., 2021). Thus, validation of this scale is a prerequisite to enable clinicians to use the scale in Jordan.

The aim of this study was to validate the Arabic version of the General Medication Adherence Scale (GMAS—AR) in Jordanian patients with type 2 diabetes mellitus (T2DM).

A total of 119 completed and useable surveys were received.

A study by Awwad and others highlighted that most Jordanian patients had low adherence. In addition, it was highlighted in the study that education, monthly income, and knowledge of patients impact their adherence to medications (Awwad et al., 2015). Moreover, Basheti and others reported non-adherence in slightly less than half of their study population. Further, they establish a relationship between adherence score and comorbidity, number of medicines, etc. (Basheti et al., 2016). The current study reported a large proportion of patients to be adherent. However, due to the limited number of responses collected, it cannot be definitively confirmed if this outcome occurred by chance or represents a true pattern.

The validation process included several statistical approaches, such as CFA with SEM and evaluation of internal consistency through Cronbach’s (α), ITC and α based on item deletion. The values for fit indices excluding RMSEA obtained in this study were >0.95 for TLI and CFI, >0.9 for GFI, and <0.9 for AGFI. Usually, a value >0.95 is considered excellent, while a value >0.9 is considered acceptable (Sarmento and Costa, 2019). In addition, the value for RMSEA obtained in this study was <0.08. Though a smaller value, such as < 0.06, is appreciable for RMSEA, studies have provided 0.08 as a cut-off criterion as well (Sarmento and Costa, 2019). The GFI obtained in this study was 0.913 while AGFI was 0.860. Available evidence report that GFI and AGFI are considered as good if the value is close to 1.0. Also, the AGFI is always less than or equal to the GFI. Thus, based on the current literature the values for both were considered acceptable (Hayashi et al., 2011). In addition, the χ²/df value obtained for this study was between one to two which is considered as good (Sarmento and Costa, 2019).

In comparison, the Mahmood and others in their validation study involving GMAS-AR in patients with diabetes in Sudan reported a χ²/df value >3, and CFI, TLI, GFI and AGFI >0.9, while RMSEA was >0.073. The value for Cronbach’s α was 0.834 (Mahmoud et al., 2021). In addition, another study was conducted in Saudi Arabia in which Arabic version of GMAS was validated in patients with chronic ailments, reported a value >0.95 for all fit indices and <0.06 for RMSEA. A Cronbach’s α value of 0.865 was reported (Naqvi et al., 2020a). In this study, the value Cronbach’s α was reported at 0.907. Therefore, despite being small-scale compared to its predecessors, the results aligned with the previous two studies (Naqvi et al., 2020a; Mahmoud et al., 2021).

The CFA results provide support for the validity of the Arabic version of the GMAS in the studied population. The good fit of the model suggests that the scale adequately measures adherence based on the underlying constructs it is expected to measure. The factor loadings also substantiate that ability of the scale to measure adherence based on the expected factor structure. In addition, the item-total correlation values were positive and ranged from 0.275 to 0.807 with majority of items having corrected ITC >0.5. This highlighted that each item in the questionnaire had a contributed positively towards measuring medication adherence and the scale had construct validity.

The availability of this scale for use in Jordanian patients would help clinicians to evaluate and monitor the pharmacotherapy of patients and would set an example for further investigation pertaining to the reliability and validity of GMAS in other disease populations in Jordan. The element of costing distinguishes this scale from the available instruments. It was observed in this study that slightly more than a third of patients paid for the medications out-of-pocket. Hence, this scale would enable the clinicians to evaluate the non-adherence due to the cost of medications as well.

The study was based on convenience sampling and was carried out in a single hospital. Besides, the number of surveys obtained was low. Based on these factors, some limitations of this study are identified. These aspects would affect the generalizability of results and may not represent the entire population. The risk of sampling error cannot be ruled out. This study may have low statistical power and may not be able to detect small and meaningful association between participants’ characteristics and their adherence score. As a result, sub-group analyses and multivariate regression analyses highlighting determinants of adherence could not be conducted on this dataset. Nevertheless, it satisfied the sampling requirements based on the statistical parameters for CFA analysis. However, the results pertaining to medication adherence and clinical information should not be generalized.

The results of this study highlight that the Arabic version of the GMAS is a reliable and valid tool to assess the medication adherence of patients with T2DM in Jordan. A new scale is now available for documenting adherence to medications in this population. This study also serves as a model for further validation the scale in other patient populations.