In recent decades, the prevalence of diabetes mellitus (DM) has increased significantly, mainly as a result of a continuous rise in the incidence of type 2 DM. According to World Health Organization statistics, >422 million adults globally suffered from DM in 2014, and a continuous rise in DM prevalence is expected (Lovic et al., 2020). China is the most populous country with patients with diabetes: the prevalence of adult diabetes is 11.6% and prediabetes is 50.1%, according to an epidemiological study (Wang et al., 2021). The comprehensive control rate of DM was 2.0% (Ma et al., 2020), and only 39.7% of the estimated 113.9 million Chinese adults with diabetes have HbA1c ≤7.0% (Xu et al., 2013). This not only represents a heavy burden but also results in serious complications, such as diabetic nephropathy, diabetic retinopathy, and increased cardiovascular mortality (Cloete, 2022).

Intensive insulin treatment can help delay the onset of diabetes-related complications, and many patients with type 2 diabetes require insulin therapy at some stage to achieve or maintain good glycemic control (Riddle, 2021). However, many patients are still reluctant to initiate or adhere to insulin therapy due to reasons which include the fear of addiction, injection, side effects such as hypoglycemia and weight gain, inconvenience, and social stigma (Lee and Yoon, 2021; Dabas et al., 2023; Skriver et al., 2023). A systematic review (Boonpattharatthiti et al., 2024) indicated that the overall prevalence of adherence to insulin therapy is remarkably low, with adherence for T1D being 52.63% for T2D being 52.55%.

“Medication literacy” is the degree to which individuals can obtain, comprehend, communicate, and process patient-specific information about their medications in order to make informed decisions safely and to effectively use their medications, regardless of the mode by which the content is delivered (e.g., written, oral, and visual) (Pouliot et al., 2018). On this basis, Neiva Pantuzza et al. (2022) proposed a conceptual model for medication literacy which consisted of four domains: functional literacy, communicative literacy, critical literacy, and numeracy, including subdomains of accessing, understanding, evaluating, calculating, and communicating medication-related information. Compared to health literacy, “medications literacy” involves specific skills that are not completely covered in health literacy, such as understanding dosage instructions or information about a drug’s indication or adverse reaction. Medication literacy is thus a specialization of health literacy in the field of medicine.

Oral medications and insulin are effective treatments for type 2 diabetes, but insulin therapy is eventually indicated for patients once maximal doses of oral medications are no longer sufficient to control blood glucose levels. In type 1 diabetes, there is an absolute deficiency of insulin; insulin treatment should start from diagnosis for optimal glucose control and maintaining HbA1c. Several studies had shown that reasons for low adherence to insulin therapy are public embarrassment (Farsaei et al., 2014; Peyrot et al., 2012), concern over hypoglycemia (Peyrot et al., 2012; Bermeo-Cabrera et al., 2018), negative beliefs and attitude about insulin therapy (Yavuz et al., 2015), difficulties in preparing injection, poor knowledge regarding DM, and insulin self-injection (Mariye et al., 2019).Therefore, the systematic identification of people with limited ability to take responsibility for their insulin therapy is critical. To the best of our knowledge, most of the instruments available regarding insulin do not cover essential aspects of insulin literacy such as correct thinking, favorable attitude, and the ability to access and information on insulin. A structured and regular assessment of patients’ insulin literacy should be conducted to ensure that they are managing their insulin appropriately. This is especially necessary for those with type 1 diabetes who must use insulin therapy as their mainstay and type 2 diabetes who cannot control or tolerate oral medications. Until now, there have been a variety of generic medication literacy scales for the general population and particular populations or diseases. There are also some medication literacy scales for particular drugs (Table 1). Nonetheless, there is no specific scale to assess insulin medication literacy for patients with DM.

This scale was developed based on the conceptual framework of the knowledge–attitude–practice model (KAP) (Cleland, 1973), with “skills” added in the context of medication use. The framework describes four main competencies for making informed insulin related decisions: knowledge, attitude, practice, and skill. The aim of this study was to develop a valid and reliable assessment scale for use by medical staff and evaluate its psychometric properties for patients with DM in China.

To best of our knowledge, this is the first study to report the validity and reliability of an assessment scale to comprehensively evaluate the level of insulin medication literacy for patients with diabetes mellitus (DM) in Zhengzhou. According to its results, the Ch-InMLS demonstrated good validity and reliability, indicating that this scale can be used to assess the insulin literacy of DM patients.

We compared our scale with existing scales from the perspective of scale application. Although there were several existing evaluation tools for insulin, they were primarily directed at evaluating psychological status rather than a comprehensive ability to safely and effectively use insulin. For example, the Insulin Treatment Appraisal Scale (ITAS) (Snoek et al., 2007), self-administered Barriers to Insulin Treatment (BIT) questionnaire (Petrak et al., 2007), distress measurement for insulin injections among patients with diabetes (Choi et al., 2023), and the Korean version of Psychological Insulin Resistance (K-PIR) (Song et al., 2016) are designed to evaluate negative attitude toward insulin treatment. In addition, there were also several scales aiming to assess both positive and negative attitudes, such as the Chinese Attitudes to Starting Insulin Questionnaire (Ch-ASIQ) (Fu et al., 2013), Insulin Pump Attitudes Questionnaire (IPA-Questionnaire) (Bergis et al., 2021), Decisional Balance for Insulin Injection (DBII) scale (Hsu et al., 2019), Insulin Treatment Experience Questionnaire (Moock et al., 2010), Insulin Treatment Satisfaction questionnaire (Anderson et al., 2004), Assessing Barriers and Adherence to Insulin Injection technique (Ehrmann et al., 2024), and patient satisfaction with insulin therapy (Cappelleri et al., 2000), which still emphasize subjective cognition and could not objectively evaluate attitude and ability to take insulin. Furthermore, there were several scales designed to evaluate self-management of or self-efficacy about insulin, such as the Insulin Treatment Self-Management Scale (Karahan Okuroglu et al., 2020), Turkish Insulin Treatment Self-management Scale (Çövener Özçelik et al., 2019), and Insulin Therapy Self-efficacy Scale (ITSS) (Nakaue et al., 2019), which are still insufficient to explain the implications of insulin medication literacy, since this includes not only attitude to insulin, but also comprehending insulin related knowledge, practice of routine insulin therapy in daily life, and the abilities and skills to use insulin effectively.

Moreover, we compared our scale with existing scales from the methodology and the results. We developed an insulin medication literacy scale (Ch-InMLS) containing the four domains of knowledge, attitude, practice, skill, based on the conceptual knowledge–attitude–practice model (KAP) framework with “skills” added. A strict, step-by-step approach consisting of item generation, semi-structured interviews, cognitive interviews, two rounds of expert consultation, item selection, construct validity, and reliability analysis was carried out. Petrak et al. (2007) only verified construct validity by EFA and CFA and verified reliability by internal consistency reliability. They lacked systematic research for scale development and pilot survey, which was a weakness of their study. Snoek et al. (2007) simply referred to two studies in the development of their scale, which was inadequate, and the two-factor solution explained only 45% of the total variance. Moreover, item-total correlations (0.34–0.53) for the positive appraisal domain were relatively low. In Karahan Okuroglu et al. (2020), before formal investigation, only face validity was performed with 15 individuals, and they had no pilot survey. Participants were recruited from only one hospital, all of which made the study less convincing.

In content validity analysis, some items that differed from the results of expert consultation were deleted or adjusted so that the scale could accurately and comprehensively reflect the concept under study. For instance, in the first round of Delphi, the item “The use of insulin is the natural progression of diabetes and does not imply the aggravation of the disease” was deleted because it repeated the item “The initiation of insulin therapy is determined by a combination of factors such as my blood sugar level, pancreatic function, complications, and other factors, and does not represent the severity of the disease”, and the expression of former item was less professional than the latter. The item “Injecting insulin indicates failure of pre-insulin treatment” was deleted since the experts mentioned that patients receiving insulin treatment under special conditions could still return to non-insulin therapies. The item “Insulin can prevent damage to liver and kidney function” was deleted because this item seemed too “professional” for patients. the item “Injecting insulin is painful” was deleted because of the possibility of negative guidance. The item “Before injecting medium acting insulin and premixed insulin, I will mix them well” was deleted since there is no need to mix long-acting insulin. The item “Insulin is a physiological hormone secreted by the body that directly lowers blood sugar” was adjusted to the more accurate “Insulin is the only physiological hormone secreted by the body that directly lowers blood sugar.” The item “I know that different insulin cannot be freely converted between them” was adjusted to “I know that there is no arbitrary conversion between medium/long-acting insulin and short/rapid acting insulin” to make it sound more scientific. In the second round of Delphi, the item “Insulin can improve pancreatic function” was deleted since it seemed a little esoteric for patients. The item “Insulin means I have to give up activities I enjoy” was deleted since it is well known that proper exercise is beneficial to disease. The item “The initiation of insulin therapy is determined by a combination of factors such as my blood sugar level, pancreatic function, complications, and other factors and does not represent the severity of the disease” was adjusted to the more accurate “The initiation of insulin therapy is determined by a combination of factors such as my blood sugar level, pancreatic function, complications, and other factors, and does not fully represent the severity of the disease.” The item “I believe that the insulin prescribed by the doctor can prevent or delay the occurrence of complications, such as diabetic kidney disease and diabetic eye disease, etc.” was adjusted to the more accurate “I believe that the insulin prescribed by the doctor can help me control my blood sugar, so as to prevent or delay the occurrence of complications, such as diabetic kidney disease and diabetic eye disease, etc.“

In the pilot survey, the item discrimination analysis, correlation coefficient method, and homogeneity test simultaneously indicated that most items had high relevance and could effectively measure the intended research variables. One exception is the item “Insulin will make others perceive greater sickness”, for which the difference between the high and low score group factors was not significant, Pearson’s correlation coefficients were smaller than the acceptable threshold, and Cronbach’s α coefficient increased after deletion, suggesting that this item had no practical significance; this item was removed. Except for statistical reasons, it could be interpreted as two factors. First, insulin was initially developed by foreigners a century ago and was then introduced to China, which had very scarce in healthcare settings. Thus, people thought that insulin meant greater sickness. But more recently, insulin has been readily available on the market owing to the policy of centralized drug procurement. Second, first- and second-generation insulins were chemically unstable, and side effects were serious; thus, the safe use of insulin was strongly emphasized, deepening the perception that insulin meant greater sickness. Nowadays, with the continuous progress of technology, third-generation insulin has become increasingly stable and safe. Given this background, the item “Insulin will make others perceive greater sickness” was deleted before formal investigation.

Additionally, in the EFA of the present study, the results showed that the KMO test (KMO = 0.944) and Bartlett’s test of sphericity were preferable for factor analysis.

The first domain was “knowledge,” which was consistent with the first step of successful diabetes self-management being patients’ knowledge about the disease (Jiang et al., 2023) and that a reason for low adherence to insulin therapy is poor knowledge of DM and insulin self-injection (Mariye et al., 2019). The knowledge domain accounted for 19.427% of the total variance and it consisted of ten items related to the recognition of DM (item K1), the role of insulin in the treatment of DM (K2, K3, K4, K5), basic information about the insulin being used (K6, K7, K10), and storage conditions for opened and unopened insulin (K8, K9). As is well known, insufficient understanding of the role and status of insulin in the treatment of diabetes is universal. Examples are the belief that glucose can be well controlled just by taking oral medications and insulin, that therapy represented aggravation in disease, and that insulin was dependent or addictive (Wen, et al., 2024; Hussein, et al., 2019). Therefore, an understanding of related knowledge about insulin is essential for establishing insulin medication literacy.

The second domain was “attitude,” accounting for 20.893% of the total variance and was the most influential component of the scale. This domain consisted of 11 items related to personal beliefs about the benefit of insulin (item A1), psychological resistance to insulin (A2, A3), adherence to insulin (A4, A5, A6, A7, A8), the relationship of insulin to diet and exercise (A10), internet insulin literacy (A9), and the pain of insulin injection (A11). “Attitude” meant favorable or unfavorable feelings toward performing insulin management and served as a significant mediator between knowledge and behavior. Several studies have demonstrated that, even with related knowledge, some patients still interrupt or reduce the dose of insulin at will, and some miss their insulin when they go out because of embarrassment and shame (Ellis et al., 2018). Furthermore, there are also patients who are more likely to hold misconceptions about insulin (de Lusignan et al., 2022), and some exaggerate the pain of insulin injections (Liu et al., 2022). All of the above perspectives have a negative effect on acceptance and adherence to insulin therapy. Hence, attitude to insulin treatment was also a crucial indicator for assessing insulin medication literacy.

The third domain was “practice,” accounting for 12.537% of the total variance; it consisted of seven items related to daily behavior under insulin treatment: blood sugar monitoring (P2), dose adjusting (P3, P7), and dealing with special circumstances (P1, P4, P5, P6). In the international consensus about medication literacy, one of the clusters identified was the outcomes and goals of medication literacy (Pouliot, et al., 2018). According to the characteristics of DM, optimal glucose control during insulin use was expected to be achieved. Several studies had demonstrated that self-management, especially the practice of insulin treatment in daily life, improved glycemic control for people with DM (Liang, et al., 2023; Fabrizi et al., 2020). Consequently, practice related to adherence to insulin treatment, the individual management of insulin dosage, communication with medical staff, and self-monitoring of blood glucose behavior was incorporated into the development of insulin medication literacy.

The fourth domain was “skill,” accounting for 14.699% of the total variance; it consisted of eight items related to the injection technique (S1, S3, S5, S6, S7, S8), injection time (S4), and the treatment of hypoglycemia (S2). Unlike injections for in-patient use administered by medical staff, insulin was a special dosage form that mostly needs to be injected by patient themselves. The abilities and skills to use insulin effectively and safely according to printed information and education by medical staff was one of the elements we considered in operationalizing the concept of insulin medication literacy for DM; specifically, skills such as the technique of injection, coping with side effects, and administration time were incorporated into the construction of insulin medication literacy.

CFA was conducted to test goodness-of-fit for the identified four-domain model; IFI was >0.9, the χ2/df value was less <3, RMSEA was <0.08, and PGFI and PNFI were >0.5, both meeting the thresholds recommended. AGFI and GFI were slightly <0.9, RMR was slightly >0.05 and did not exactly meet the thresholds recommended. These indices, however, are likely to be underestimated when the sample size is <300 (Kai, et al., 2018). Additionally, convergent validity was identified by calculating AVE. The results demonstrated that the AVE of the four domains was >0.50, indicating that the variance between domains and associated items exceeded that caused by measurement errors (Acosta-Prado et al., 2020), establishing acceptable convergent validity. Discriminant validity was also assessed by calculating the average variance and squared correlation coefficient among domains. The results demonstrated that the square root of AVE was greater than all possible two-factor correlation coefficients, illustrating that acceptable discriminant validity was established. Overall, the CFA model fit indices of the scale were acceptable, although less than perfect, indicating coherence between the information and the theoretical structure.

Furthermore, later internal consistency evaluations showed that Cronbach’s α exceeded 0.9 for the total scale and each of the domains, indicating that the scale had high internal consistency and that no additional adjustments to the Ch-InMLS were required. The retest reliability coefficient of the scale was 0.944, indicating that the measurement time had little influence on the reliability of this scale and that it had strong time flexibility and stability in assessing Ch-InMLS. Our findings indicated that the scale had good overall reliability, indicating that the scale could adequately measure the level of insulin medication literacy.

Several limitations to the study should be noted. First, owing to constraints of time and resources, participants were recruited from one city (Zhengzhou) by a convenience sampling approach, which was not representative of the entire population. Moreover, we did not assess the participants’ cultural background, which may affect aspects of insulin medication literacy. Second, the investigation of participants was conducted online, so bias in ability to understand may occur in the self-reporting of participants. As with all scales, there was also a risk of overestimating insulin medication literacy due to the shame and fear of low literacy in the current medical culture. Furthermore, we did not evaluate criterion validity due to the lack of a golden criterion to assess the insulin medication literacy of patients with DM.

Generally, the Ch-InMLS can be applied as a valid and reliable instrument to measure insulin medication literacy amongst patients with diabetes mellitus (DM). It provides policymakers and hospital administrators with an applicable and reliable tool for formulating policies regarding insulin education and training programs. It could be used by medical staff to simplify the insulin regimen, which could in turn improve insulin compliance and improve the health outcomes of patients with DM. In addition, the scale could be integrated into clinical practice as part of routine clinical assessment so that physicians can propose specific interventions without delay when they detect low medication literacy about insulin.