Development and psychometric assessment of self-reported patient medication safety scale (SR-PMSS)

Abstract

Objectives:

Patient medication safety can affect their clinical outcomes and plays an important role in patient safety management. However, few tools have been developed to assess patient medication safety. This study aimed to develop and validate the self-reported patient medication safety scale (SR-PMSS).

Methods:

We developed SR-PMSS guided by the Donabedian Structure-Process-Outcome framework and used psychometric methods to test its validity and reliability.

Results:

A total of 501 patients with an average age of 56.81 ± 14.47 were enrolled in this study. The SR-PMSS consisted of 21 items and 5 factors. The content validity was good with item-level content validity index (CVI) > 0.78, average scale-level CVI (S-CVI) > 0.9, and universal agreement S-CVI > 0.8. Exploratory factor analysis extracted a five-factor solution with eigenvalues > 0.1, explaining 67.766% of the variance. Confirmatory factor analysis showed good model fit, acceptable convergent validity, and discriminant validity. The Cronbach’s α coefficient for SR-PMSS was 0.929, the split-half reliability coefficient was 0.855, and the test–retest reliability coefficient was 0.978.

Conclusions:

The SR-PMSS was a valid and reliable instrument with good reliability and validity to evaluate the level of patient medication safety. The target users of the SR-PMSS are all people who are taking or have used prescription medications. The SR-PMSS can be used by healthcare providers in clinical practice and research to identify patients at risk for medication use and intervene with them to reduce adverse medication events and provide support for patient safety management.

Plain Language Summary

SR-PMSS – a self-reported tool to assess patient medication safety

Medication therapy was the most common and frequent treatment method to prevent and cure diseases. Medication safety issues may occur in the process of medication use. Patient medication safety can affect their clinical outcomes and plays an important role in patient safety management. However, there are few tools to assess patient medication safety currently, and most of them focused on medication safety related to hospitals or healthcare workers. We developed the self-reported patient medication safety scale (SR-PMSS) guided by the Donabedian Structure-Process-Outcome framework. Then, we conducted a two-round expert consultation, clarity verification, and item simplification to determine the final version of the scale. The SR-PMSS consisted of 21 items and 5 factors and it had good validity and reliability. The target users of the SR-PMSS are all people who are taking or have used prescription medications. Healthcare providers can use the SR-PMSS in clinical practice and research to identify patients at risk for medication use and intervene with them to reduce adverse medication events and provide support for patient safety management.

Keywords

medication safety patient reliability scale self-report validity

Introduction

Medication therapy was the most common and frequent treatment method to prevent and cure diseases. Medication safety issues may occur in the process of medication use. Medication safety was emphasized as an important component of patient safety by World Health Organization (WHO) and it was defined as ‘Freedom from accidental injury during the course of medication use; activities to avoid, prevent, or correct adverse drug events which may result from the use of medications’.¹ However, medication safety issues were still prevalent at present, threatening patient safety and reducing the quality of medical services. A scoping review showed medication was one common reason for medical adverse events in hospitals ranking among the top three.² A scoping review showed medication safety program characteristics included educational training, quality improvement tools, informatics, patient education, and feedback provision. And the incidence of medication errors and reported adverse events or drug-related problems were outcome indicators to assess the effectiveness of these programs in improving patient safety.³ Alqenae et al.⁴ found that the median rate of medication errors following discharge from hospital to community settings was 53% for adult patients. Gurwitz et al.⁵ found that adverse drug events were common among ambulatory geriatric patients. In the ambulatory setting, the percentage of adverse drug events that were deemed preventable more closely mirrored the hospital setting. Garzón González et al.⁶ revealed that 47% of medication errors occurred in primary care centers, and over 25% of them were patient medication errors. In England, 237 million medication errors may occur at some point annually during medication. It was found that 27.9% of medication errors were potential significant errors occurring in clinical practices which may prolong the length of stay, increase the economic burden and even cause death, even though 72.2% of them may have no or little harm to clinical health outcomes.⁷ A systematic review showed each medication error may cause additional medical expenses ranging from €2.6 to €111 727.1.⁸ Thus, medication safety needs to be improved. Although interventions such as medication review, barcoding systems, preprinted order sheets, and specialist pharmacist roles have been shown to be effective in reducing medication errors,⁹ medication error, as one of the direct medication safety outcomes, did not reveal the whole medication process and the specific causes of medication unsafe issues. Thus, there is a lack of assessment system reflecting details and practices during patients’ medication process. Identified problems and causes of unsafe medication practices could provide comprehensive evidence to guide patients to develop good habits of medication. Therefore, a reliable tool to assess the status of medication safety was indispensable.

Several tools to assess medication safety have been developed so far, and most of them focused on medication safety related to hospitals or healthcare workers. Winterstein et al.¹⁰ developed a tool to assess medication safety and quality for small rural hospitals. Yu et al.¹¹ developed a nurse medication unsafe behavior scale to assess nurses’ medication safety based on medication administration processes. Park and Seomun¹² developed a medication safety competence scale for nurses based on constitutive factor analysis. Hohl et al.^13,14 derived clinical decision rules that identified emergency department patients with adverse drug events with high sensitivity and also validated diagnostic accuracy of the rules was by a prospective study. However, medication safety not only involved hospital management and healthcare workers, but also patients themselves. Medication unsafe events may occur at any point for patients’ medication without instructions and supervision of healthcare workers. Thus, patients themselves should pay attention to medication safety and obey medication regulations, which was one key process for medication safety management. Wang et al.¹⁵ developed the patients’ involvement in medication safety scale (IIMSS) to assess inpatients’ behaviors to participate in clinical medication safety. IIMSS has good reliability and validity, but only focused on inpatients. There were some specific situations not applicable for patients’ medication process, such as poor doctor–patient communication, time limitation, patient introversion, patients discharged, and so on. Therefore, medication safety tools to assess patient medication safety should be further developed.

An important issue that cannot be ignored was that the tools previous studies have developed mainly based on literature review, and the adequacy and fit of the theoretical framework were unknown. WHO proposed five moments for medication safety in 2019 and created a checklist for patients to manage their medication behaviors.¹⁶ However, the checklist was not convenient for self-reports resulting in the difficulty of investigation and statistical analysis. Meanwhile, an appropriate theoretical framework can guide the development process of tools to improve model fit and content adequacy. Medication safety was one of the indicators of healthcare quality, the Donabedian Structure-Process-Outcome (SPO) framework was one available framework.¹⁷ The Donabedian SPO framework has been identified to be effective for guiding healthcare quality improvement projects.¹⁸ Overall, this study aimed to develop a self-reported tool to assess patient medication safety based on five moments for medication safety and the Donabedian SPO framework and to test its reliability and validity. It can provide a scientific and convenient tool for healthcare personnel to assess patients’ medication safety levels, helping healthcare workers establish related support and management measures.

Methods

SR-PMSS development

The first draft of self-reported patient medication safety scale (SR-PMSS) with 19 items was developed by literature review and group roundtable discussions. A literature review was conducted to identify the concept and operationalized indicators of patient medication safety. Safety has been defined as ‘the reduction of risk of unnecessary harm to an acceptable minimum’ and patient safety as ‘the reduction of risk of unnecessary harm associated with healthcare to an acceptable minimum’.¹⁹ Patient medication safety was defined as ‘safe actions in patients’ medication process can greatly reduce the risk of unnecessary harm associated with medication to an acceptable minimum’ in this study. Critically, ‘patients’ in this study was defined as ‘all individuals currently taking or had taken prescription medications’ instead of ‘inpatients or discharged patients’. Based on the Donabedian SPO framework and five moments for medication safety proposed by World Health Organization,¹⁶ the operationalized framework of patient medication safety was developed through group roundtable discussions and expert interviews. In this study, the ‘structure’ was defined as ‘medication environmental safety for patients’ and consisted of one factor (medication environmental safety) with four indicators (medication affordability, family support, accessibility of professional medication services, and medication accessibility). The ‘process’ was defined as ‘potential safety issues may occur in the actual medication practices’ and consisted of three factors. The factor of behavioral safety to acquire medication information included two indicators (Medication prescription reading and drug information query), the factor of behavioral safety to take medication included two indicators (Expiration date check and correct medication), the factor of behavioral safety to medication monitoring and feedback included two indicators (Medication effect monitoring and feedback, adverse drug reaction monitoring and feedback). The ‘outcome’ was defined as ‘medication safety outcomes’ and consisted of one factor with three indicators (medication adherence, medication-related adverse events, and medication error). The initial item pool with 19 items was developed based on the above framework.

SR-PMSS modifications

Expert consultation

A two-round expert consultation was performed to evaluate the plausibility, scientificalness, and suitability of the framework, indicators, and items. We invited experts to evaluate the relevance of the entries and clarity of presentation of the scale based on their professional knowledge and clinical experience, and the scale was revised for the first time by combining expert opinions. After informing the content of the adjustment, experts were invited again to evaluate the relevance of the entries. A consultation letter consisting of three sections of introduction, consultation form, and basic information of experts was delivered to experts by e-mail or face-to-face. Inclusion criteria of experts: (1) with over 10 years of work experience involving areas of safety management, pharmacy, public health education/management; (2) with a master’s degree or above; (3) with an associate professor title or above; (4) master in development and psychometric assessment of a scale; (5) showed interests in this survey and was glad to offer advice or suggestions. Seven experts completed the first round of consultation and six experts finished the second round. Six experts were professors and one was an associate professor. Two experts were clinical pharmacy specialists, two nursing experts specialized in safety management, two experts master in public health education (one expert with a professor title could not reply consultation letter within 2 weeks in the second round), and one expert specialized in public health management.

In the first round, seven experts proposed 24 comments. We discussed all expert opinions, adopted 20 comments, and made 6 modifications to items: (1) Considering the different routes of administration of prescription medications, such as oral administration, spray, external application, suppositories, and so on, the phrase ‘take medication’ was used in the scale instead of ‘take oral medicine’. (2) For the indicator of family support, in the item ‘My family members support me a lot to take medicine’, ‘medication support’ was too vague, so we revised it to ‘My family members can assist, supervise, and encourage me to obtain and take medications’ to make the meaning clearer. (3) For the indicator of medication accessibility, the item ‘I can obtain medicine easily’ was ambiguous. Medication acquisition easily not only could be helpful for disease treatment but also may cause drug abuse. Thus, we revised it as ‘I can get the required medications easily according to my doctor’s prescription’. (4) Due to the overlap between the item 6 ‘I will read drug instructions’ and item 7 ‘I will inquire about the correct storage methods for medicine’, and the correct storage methods, considering the close relationship between the drug indications and contraindications, usage and dosage methods, adverse reactions in the drug instructions and patient medication safety, so item 6 were replaced with three specific items – item ‘I will query the indications and contraindications for my medications’, item ‘I will query the usage and dosage for my medications’, and item ‘I will query the adverse drug reactions for my medication’. (5) Medication adherence was closely related to patient medication safety.²⁰ For the indicator of behavioral safety to take medication, and medication outcome safety, referring to the Morisky Medication Adherence Scale-8 and the description and definition of medication adherence,^21,22 we made corresponding revisions in the order of initiation, implementation, and discontinuation to manifest patient medication safety across the whole medication process. To avoid duplicate items, we deleted the item ‘I will strictly follow the doctor’s prescription and will not change the dose on my own’, added the item ‘I will make sure my route of each medication is right’ to assess the behavioral safety to take medication and the item ‘I have never forgotten taking my medications’ to assess the medication outcome safety. Meanwhile, we adjusted the item ‘I have been able to achieve the expected goals of the medication after taking it’ to assess the medication effect monitoring. (6) In the item ‘I often have adverse drug reactions’ and item ‘I often have medication errors’, the expression of the frequent word “often” was inappropriate to assess the frequency of occurrence for individuals, so we used “never” instead of “often” to make the expression clearer. In the second round, six experts proposed three comments, and we did not modify them after group discussion.

Clarity verification

A face-to-face structured interview was used to ask patients about their evaluation of the readability and understandability of the initially drafted scale. Ten patients of different age, gender, and education levels were invited to participate in the pre-test. First, informed consent was obtained from the patients. Then, an investigator invited them to complete the scale themselves. Once there was an item that is hard to read or understand, feedback can be given to the investigator. In addition, a visual analogue scale score was used to evaluate the readability and understandability of the scale, and each item can be scored ranging from 0 (not clear) to 10 (very clear).²³ When an item was evaluated with a score of fewer than 7 points, patients were asked to give comments and suggestions on the item. The results of the pre-test showed the readability and understandability of the primary scale were good with all items of ⩾7 points. Finally, we made a summary of the results in pre-test and formed the primary SR-PMSS with 22 items for item simplification.

Item simplification

A total of 120 patients were selected with convenience sampling from a tertiary hospital in Changsha city of China from October to November 2021. Socio-demographic information (gender, age, single or not, education, occupational situation, current residence, economic satisfaction, and chronic disease condition) and the primary SR-PMSS made up the questionnaire. The primary SR-PMSS had 22 items scoring on a 4-point Likert-type scale (0 point: disagree; 1 point: partly agree; 2 points: mostly agree; 3 points: totally agree), and negative items were scored reversely. Inclusion criteria of patients were as follows: (1) individuals above the age of 18 years old; (2) currently taking or had taken prescription medication; (3) have basic reading and writing skills; (4) agree to participate in the survey. In addition, patients with unclear consciousness or who have communication barriers were excluded. For item selection, item discrimination analysis, correlation coefficient method, and Cronbach’s α reliability analysis were used and described below.²⁴

We sorted the data according to the total score of primary SR-PMSS from high to low and compared the differences between the total scores of the top 27% (high score group) and of the last 27% (low score group) over each item using an independent sample t test. The results of item discrimination analysis showed that there was a significant difference statistically for items 1–22 (except item 14) between the high score group and the low score group (critical ration: 3.050–16.098, p < 0.05). For item 14, there was no statistical difference between the two groups (critical ration = 1.321, p = 0.192), which indicated weak discrimination. The results of the correlation coefficient method showed r > 0.3 with p < 0.001 between the score of each item and the total score of the scale for items 2–13 and items 15–22 and r < 0.3 for item 1 and item 14 (item 1: r = 0.297, p < 0.001; item 14: r = 0.119, p = 0.194). For the correlation between each factor and the scale, all the Pearson’s correlation coefficient (r = 0.644–0.874, p < 0.001) was greater than 0.3. The results of Cronbach’s α reliability analysis showed that the Cronbach’s α coefficient (0.928) for the total score increased after deleting item 1 (0.931), item 14 (0.935), and item 20 (0.929). Based on item selection analysis and specialty practicalities, we decided to delete item 14 because of the weak discrimination, poor correlation, and Cronbach’s α reliability after group discussion, and the remaining items were retained for the formal survey.

SR-PMSS validation

We recruited patients from a tertiary hospital and two community healthcare centers in Changsha city of China with convenience sampling methods from November to December 2021. The investigators informed patients one-on-one in detail about the purpose, content, and significance of the study on-site and conducted the survey anonymously after obtaining verbal informed consent. The questionnaire was collected and checked on the spot, promptly reminding the patient to add any incompleteness found and confirming possible contradictions with the patient. The questionnaire would be considered invalid if the patient filled in too short a time or selected the same entry. Inclusion and exclusion criteria were consistent with that in the item simplification section. Two methods were used to identify the sample size: (1) a previous study pointed out that the sample size in the formal survey for scale validation should be 5–10 times the number of items of the scale, so the minimum sample size was calculated as 110;²⁵ (2) previous studies showed that the minimum sample size for confirmatory factor analysis (CFA) was 200 and the minimum sample size for exploratory factor analysis (EFA) was 100.^26,27 Thus, we identified the minimum sample size of 300 to ensure the power of a test. Three trained investigators collected data with a face-to-face method after obtaining the informed consent of patients.

Instruments

The Instruments included a socio-demographic information questionnaire and SR-PMSS. Socio-demographic information consisted of gender, age, single or not, education, occupational situation, current residence, economic satisfaction, and chronic disease condition. SR-PMSS included five factors namely medication environmental safety with four items, behavioral safety to acquire medication information with five items, behavioral safety to take the medication with four items, behavioral safety to medication monitoring and feedback with five items, and medication outcome safety with three items. A 4-point Likert-type scale ( ‘Disagree’, ‘Partly agree’, ‘Mostly agree’ and ‘Totally agree’) was adopted in this scale, in which scores of 0, 1, 2, and 3 were applied accordingly. The total score was the sum of 21 items with a range of 0–63. The healthcare providers issue the scale to the eligible patients (target users), who are invited to self-report, and then make a judgment based on the patients’ responses. Higher scores indicated higher levels of patient medication safety (Supplementary materials). With the term healthcare providers, we refer to all occupations engaged in organizing and delivering healthcare, including unpaid caregivers, volunteers, and informal health workers. Depending on the respective setting, this can comprise general practitioners or family medicine specialists, nurses, auxiliary nurses, pharmacists and pharmacist assistants, public health nurses, community health workers, or social workers. The target users of SR-PMSS include all people who are taking or have used prescription medications. For those who are taking medications, we can determine the safety of their current medications, and for those who have taken prescription medications, we can use the scale to predict the possible risks of their future use and provide early health education.

Data analysis

Descriptive statistics

We used descriptive statistics to describe the general socio-demographic information of samples using means, standard deviations (SDs), frequencies, and percentages (%) with SPSS 26.0.

Validity test

Content validity, construct validity, convergent validity, and discriminant validity were used to test the validity of SR-PMSS.

Content validity referred to the degree to which the items of a scale are fairly representative of the entire concept being measured.²⁸ Content validity index (CVI) was used to assess content validity based on the second round of consultation in this study. CVI included item-level CVI (I-CVI) and scale-level CVI (S-CVI), and S-CVI also included average S-CVI (S-CVI/Ave) and universal agreement S-CVI (S-CVI/UA). When the number of experts was more than five, I-CVI ⩾ 0.78, S-CVI/Ave ⩾ 0.9, and S-CVI/UA ⩾ 0.8 indicated good content validity.

Construct validity referred to the extent to which the instruments test the hypothesis or theory they were measuring.²⁹ We used EFA and CFA to test construct validity. EFA using SPSS 26.0 was performed with principal component analysis with varimax orthogonal rotation to explore the factor structure. Kaiser–Meyer–Olkin (KMO) test and Bartlett’s test were conducted to identify data suitability for EFA. When the KMO value was > 0.7 and the Bartlett’s test was significant (p < 0.001), it indicated that the scale was suitable for factor analysis. We extracted factors of which the eigenvalue was greater than 1 criterion, when the factor loadings of items were equal to or greater than 0.4 and the cumulative variance contribution rate was greater than 60%, it showed suitable construct validity. CFA using AMOS 24.0 was performed to identify the factor structure of SR-PMSS. We used chi-square value/freedom degree (χ²/df), root mean square error of approximation (RMSEA), root mean square residual (RMR), goodness of fit index (GFI), adjusted goodness of fit index (AGFI), comparative fit index (CFI), incremental fit index (IFI), Tucker–Lewis index (TLI), parsimonious comparative fit index (PCFI), and parsimonious normed fit index (PNFI) to assess the model fit. When χ²/df < 3, RMSEA < 0.08, RMR < 0.05, GFI > 0.9, AGFI > 0.9, CFI > 0.9, IFI > 0.9, TLI > 0.9, PCFI > 0.5, PNFI > 0.5, it showed the model fit well.³⁰

Convergent validity was used to assess the consistency of the items and was measured by average variance extracted (AVE) and construct reliability (CR). When AVE > 0.5 and CR > 0.7, it showed good convergent validity of the scale.³¹

Discriminant validity was used to assess the independence of the items using the Fornell-Larcker criterion. When the square root of AVE for one factor was greater than all the correlation coefficients between this factor and other factors, it showed good discriminant validity of the scale.³²

Reliability test

We assessed reliability from two aspects of internal consistency and temporal stability with SPSS26.0. Internal consistency reliability was assessed by Cronbach’s α coefficient and split-half reliability coefficient. When Cronbach’s α for SR-PMSS was greater than 0.8, Cronbach’s α for each factor was greater than 0.7, split-half reliability coefficient for SR-PMSS and all factors were greater than 0.7, it showed acceptable internal consistency reliability.^33,34 The temporal stability of SR-PMSS was assessed in terms of test–retest reliability. We conveniently selected 30 patients from 501 samples to measure repeatedly with 2 weeks interval and calculated the Pearson correlation coefficient of the total scores between measures before and after 2 weeks. Higher Pearson correlation coefficients indicated higher levels of temporal stability. Pearson correlation coefficient greater than 0.7 showed acceptable temporal stability.³⁵

Results

Sample characteristics

A total of 515 eligible patients participated in this study, 304 (60.68%) of them were male, the average age was (56.81 ± 14.47) years old with the range of 20–92, 56 (11.18%) were single, 132 (26.35%) had an education level of primary school or below, 114 (22.75%) had a full-time job, 248 (49.50%) lived in urban at present, 158 (31.54%) were very or moderately satisfied with their family economic condition, 467 (93.21%) had the chronic disease (Table 1).

Table 1.

Patient characteristics (N = 501).

Variables	Group	n (%)
Gender	Male	304 (60.68)
	Female	197 (39.32)
Age (years)	18–45	98 (19.56)
	46–60	193 (38.52)
	⩾61	210 (41.92)
Marital status	Single	56 (11.18)
	No single	445 (88.82)
Education	Primary school or below	132 (26.35)
	Junior middle school	163 (32.53)
	Senior high school or secondary technical school	111 (22.16)
	Junior college	48 (9.58)
	Undergraduate and above	47 (9.38)
Occupational situation	Full-time job	114 (22.75)
	Part-time job	13 (2.59)
	Unemployed	226 (45.11)
	Retired	148 (29.54)
Residence type	Urban	248 (49.50)
	Rural	253 (50.50)
Economic condition	Very satisfied	16 (3.19)
	Moderately satisfied	142 (28.34)
	Neutral	187 (37.33)
	Moderately dissatisfied	124 (24.75)
	Very dissatisfied	32 (6.39)
Chronic condition	Chronic	467 (93.21)
	Non-chronic	34 (6.79)

Validity of SR-PMSS

Content validity

The SR-PMSS has 21 items with good content validity, that is, I-CVI for each item was 0.833–1.00, S-CVI/Ave for the scale was 0.968, and S-CVI/UA for the scale was 0.810 (Supplementary materials).

Exploratory factor analysis

A total of 120 samples were selected randomly from the 501 samples to perform EFA. The results of the KMO test and Bartlett’s test showed that the KMO value was 0.888 (>0.7) and Bartlett’s test was statistically significant (χ² = 1522.034, p < 0.001), indicating data suitability for factor analysis. Then, principal component analysis with varimax orthogonal rotation was used for factor analysis and five factors with eigenvalues > 1.0 were extracted. The cumulative variance contribution rate was 67.766% and factor loadings for each item in its belonging factor were > 0.4 (Table 2). In this stage, the structure of the scale with five factors fitted well with theoretical underpinning, demonstrating the plausibility of the theoretical framework.

Table 2.

Exploratory factor analysis on self-reported patient medication safety scale (N = 120).

Item	Factors^a
Item	1	2	3	4	5
1	0.736	−0.095	0.252	0.112	0.152
2	0.509	0.485	−0.178	0.152	−0.004
3	0.498	0.404	0.186	0.387	0.092
4	0.545	0.226	0.053	0.265	0.333
5	0.142	0.750	0.280	0.189	0.143
6	0.039	0.833	0.244	0.146	0.145
7	0.112	0.749	0.402	0.243	0.138
8	0.067	0.847	0.200	0.241	0.093
9	0.081	0.791	0.265	0.286	0.133
10	0.024	0.516	0.614	0.142	0.139
11	0.116	0.401	0.766	0.175	0.049
12	0.113	0.272	0.730	0.252	0.134
13	0.094	0.438	0.540	0.306	0.134
14	0.259	0.115	0.248	0.689	0.118
15	0.076	0.299	0.040	0.731	0.191
16	0.147	0.247	0.201	0.721	0.068
17	0.136	0.278	0.168	0.712	0.136
18	0.167	0.075	0.480	0.550	0.105
19	−0.209	0.072	0.088	0.458	0.583
20	0.215	0.106	0.239	0.139	0.709
21	0.274	0.218	0.006	0.076	0.784
Eigenvalues	1.000^b	9.069	1.255	1.837	1.070
Explained variations (%)	8.252	22.005	13.125	15.679	8.707

Factor 1: medication environmental safety; Factor 2: behavioral safety to acquire medication information; Factor 3: behavioral safety to take medication; Factor 4: behavioral safety to medication monitoring and feedback; Factor 5: medication outcome safety.

1.00034 was approximately equal to 1.000.Bold values indicated factor loadings greater than 0.5.

Confirmatory factor analysis

We used 381 remaining samples to conduct CFA. The results of fit indices for SR-PMSS indicated an excellent model fit (χ²/df = 2.235 < 3, RMSEA = 0.057 < 0.08, RMR = 0.047 < 0.05, GFI = 0.909 > 0.9, AGFI = 0.883 > 0.85, CFI = 0.950 > 0.9, IFI = 0.951 > 0.9, TLI = 0.942 > 0.9, PCFI = 0.806 > 0.5, PNFI = 0.775 > 0.5). The structure equation model of CFI is shown in Figure 1.

Figure 1.

Structure equation model of confirmatory factor analysis for SR-PMSS.

Convergent validity

The results of convergent validity are shown in Table 3. The AVE was > 0.5 in two factors and in three 0.4–0.5, and CR values for each factor were greater than 0.7.

Table 3.

Convergent validity for SR-PMSS.

Path			Estimate	AVE	CR
Item 1	←	Factor 1	0.500	0.409	0.732
Item 2	←	Factor 1	0.648
Item 3	←	Factor 1	0.704
Item 4	←	Factor 1	0.686
Item 5	←	Factor 2	0.786	0.708	0.924
Item 6	←	Factor 2	0.835
Item 7	←	Factor 2	0.868
Item 8	←	Factor 2	0.880
Item 9	←	Factor 2	0.836
Item 10	←	Factor 3	0.744
Item 11	←	Factor 3	0.891	0.652	0.882
Item 12	←	Factor 3	0.827
Item 13	←	Factor 3	0.760
Item 14	←	Factor 4	0.650	0.492	0.828
Item 15	←	Factor 4	0.769
Item 16	←	Factor 4	0.708
Item 17	←	Factor 4	0.690
Item 18	←	Factor 4	0.683
Item 19	←	Factor 5	0.465	0.465	0.714
Item 20	←	Factor 5	0.763
Item 21	←	Factor 5	0.773

AVE, average variance extracted; CR, construct reliability; SR-PMSS, self-reported patient medication safety scale.

Discriminant validity

The results of discriminant validity showed the square root of AVE values for each factor were greater than the other correlation coefficients (Table 4).

Table 4.

Discriminant validity for SR-PMSS.

	Factor 1	Factor 2	Factor 3	Factor 4	Factor 5
Factor 1	0.409
Factor 2	0.376*	0.708
Factor 3	0.406*	0.763*	0.652
Factor 4	0.514*	0.585*	0.630*	0.492
Factor 5	0.359*	0.434*	0.488*	0.540*	0.465
$\sqrt{AVE}$	0.640	0.841	0.807	0.701	0.682

AVE, average variance extracted; SR-PMSS, self-reported patient medication safety scale.

p < 0.001.

Reliability of SR-PMSS

The Cronbach’s α coefficient for SR-PMSS was 0.929 and for each factor ranged from 0.683 to 0.928, the overall split-half reliability coefficient for SR-PMSS was 0.855 and for each factor ranged from 0.693 to 0.931, indicating acceptable internal consistency reliability. The results of test–retest reliability showed that the Pearson correlation coefficient for SR-PMSS was 0.978 (p < 0.001) and for each factor ranged from 0.743 to 0.945 (p < 0.001). The detailed results are shown in Table 5.

Table 5.

The reliability coefficients for SR-PMSS.

Factors	Items	Cronbach’s α	Split-half reliability	Test-retest reliability
1	4	0.714	0.693	0.840*
2	5	0.928	0.931	0.945*
3	4	0.857	0.879	0.743*
4	5	0.838	0.851	0.898*
5	3	0.683	0.745	0.812*
SR-PMSS	21	0.929	0.855	0.978*

SR-PMSS, self-reported patient medication safety scale.

p < 0.001.

Discussion

The development process of SR-PMSS was scientific

In this study, we developed SR-PMSS following the scale development process in order. The basic process of health measure development included the following steps: conceptual definition, item generation, pre-test, item simplification, and formal survey to test reliability and validity.³⁶ In the process of conceptual definition, we selected the related concept based on a literature review and clarified the scientific connotation of this scale with its factors based on the Donabedian SPO framework, which can guide indicator and item generation. For item generation, we constructed the index system of medication safety with 5 factors and 13 indicators based on group roundtable discussions and expert interviews, then the initial item pool was developed, and two-round expert consultation was used to evaluate the plausibility, scientificalness, and adequacy of the indicators and items. To ensure the linguistic clarity of items, ten patients of different age, gender, and levels of education were invited to take part in an interview. We have simplified the description of items and kept the statement as short as possible to improve the feasibility and acceptability of SR-PMSS. For the pre-test, the rule of thumb was used to identify the sample size, that is, the sample size should be 5–10 times the number of items of the scale, so the minimum sample size was 110.²⁵ For item simplification, we adopted comprehensive methods to analysis the item quality. Item discrimination analysis was used to assess the sensitivity of items, the Correlation coefficient method was used to assess the representation and independence of items, and Cronbach’s α reliability analysis was used to assess the internal consistency of the scale. For the formal survey, considering the impossibility of EFA to identify the association among factors, we performed EFA and CFA to determine the factor structure of SR-PMSS.³⁷ Overall, the development process of SR-PMSS was scientific in this study.

The validity evaluation of SR-PMSS

The validity reflected the effectiveness and accuracy of a tool to measure the concept that is intended to measure, and usually be evaluated by content validity and construct validity.³⁸ In this study, the results showed I-CVI for each item was greater than 0.78, S-CVI/Ave for the scale was greater than 0.9 and S-CVI/UA for the scale was greater than 0.8, which indicated SR-PMSS had good content validity. The results of EFA showed five common factors in line with the theoretical framework extracted with a cumulative variance contribution rate greater than 60% and all factor loadings were greater than 0.4, which indicated SR-PMSS had good construct validity. The CFA showed a good model fit and most indices were inside normative ranges. AGFI in this study was less than 0.9 (>0.85), which was acceptable according to previous studies.^24,39 Some previous studies did not involve convergent validity and discriminant validity to evaluate the validity of a scale,^11,15 but the two indexes were important for evaluating the relationship between items and factors. In this study, the values of AVE for two factors were greater than 0.5, for another three factors were greater than 0.4 but less than 0.5, and all the CR values for each factor were over 0.7, indicating acceptable convergent validity.⁴⁰ In addition, the square root of AVE for all factors was greater than all the correlation coefficients which indicated good discriminant validity of SR-PMSS.

The reliability evaluation of SR-PMSS

The reliability reflected the internal consistency and stability of measurement tools.³⁸ The findings showed that the Cronbach’s α coefficient for SR-PMSS was greater than 0.9 and for each factor ranged from 0.683 to 0.928, the split-half reliability coefficient for SR-PMSS was greater than 0.8 and for each factor ranged from 0.693 to 0.931. A previous study indicated the relationship between the number of items and Cronbach’s α coefficient. It stated that the Cronbach’s α coefficient of less than 0.5 was usually unacceptable when the number of items of its factor was less than 5.⁴¹ The Cronbach’s α coefficient for factor 5 with three items was less than 0.7 but still above 0.5, and the split-half reliability coefficient for factor 1 with four items was less than 0.7 but near 0.7, which indicated acceptable internal consistency reliability. The results of test–retest reliability showed that the Pearson correlation coefficient for SR-PMSS was greater than 0.9 and for each factor ranged from 0.743 to 0.945 with all coefficients above 0.7, indicating good temporal stability.

Application prospect of SR-PMSS

SR-PMSS consisted of five factors based on the Donabedian SPO framework: medication environmental safety, behavioral safety to acquire medication information, behavioral safety to medication monitoring and feedback, behavioral safety to take medication, and medication outcome safety. Medication environment may be one of the predicting factors of patient medication safety.⁴² For example, patients with economic hardship, weak family support, poor access to professional medication services, and medication may have more risks of non-compliance with medication therapy resulting in adverse drug events.^43,44 Behavioral safety to acquire medication information, take medication, and medication monitoring and feedback reflected the actual medication process. At present, most patients are used to seeking and reading medication information before taking prescription medications,⁴⁵ and high-quality medication information sources may directly affect the outcomes of medication safety in a positive way.⁴⁶ Taking medication correctly may be easy for most patients, but when the several numbers of medications taken by older patients with comorbidities, medication-related risks may increase and even adverse drug events occur.⁴⁷ Monitoring and responding to the effects of medication was essential for medication adjustment and disease control.⁴⁸ In addition, medication outcome safety was the direct consequence. Patients with adverse drug outcomes (e.g. Patient forgetfulness, mistakes, and adverse reactions) may have more risks of suffering from medication unsafe problems.⁴⁹

SR-PMSS can be used in a wider population than IIMSS. All individuals currently taking or had taken prescription medication can be the targeted population. Also, SR-PMSS had 21 items with simple wording, only 8–10 minutes were needed for completing it. Meanwhile, SR-PMSS with good reliability and validity comprehensively addressed the scientific connotation of patient medication safety. It can provide a useful measurement tool for clinical healthcare workers and managers to identify patients’ unsafe medication issues. Thus, corresponding health counseling and behavioral interventions can be formulated and developed aiming at different stages of medication process, improving the situation of patient medication safety.

Limitation

This study had several limitations. First of all, we recruited patients from a tertiary hospital and two community healthcare centers in Changsha city of China, so selection bias may occur. Second, the AVE values for three factors were less than 0.50. Therefore, further study is necessary for the improvement of convergent validity. Then, SR-PMSS was a self-assessment and self-reported scale, so the scale entries involve patient recall content measurement bias and recall bias may occur. The scale does not have more refined objective indicators, so patients’ subjective outcomes may differ from healthcare providers’ perceptions. Third, we did not grade the scale scores. In the future, a longitudinal study can be conducted to observe whether patients experience future medication safety events in conjunction with the scale scores and grade them according to cutoff values. Finally, the application of SR-PMSS across culture and medical backgrounds should be further evaluated and explored.

Conclusion

The SR-PMSS is a scientific, reliable, and convenient instrument with good reliability and validity to evaluate the status of patient medication safety. It can provide support for healthcare providers to improve the management of medication safety with respect to patients’ medication process.

Supplemental Material

sj-docx-1-taw-10.1177_20420986231152934 – Supplemental material for Development and psychometric assessment of self-reported patient medication safety scale (SR-PMSS)

Supplemental material, sj-docx-1-taw-10.1177_20420986231152934 for Development and psychometric assessment of self-reported patient medication safety scale (SR-PMSS) by Ning Qin, Yinglong Duan, Shuangjiao Shi, Xiao Li, Haoqi Liu, Feng Zheng, Zhuqing Zhong and Guliang Xiang in Therapeutic Advances in Drug Safety

Supplemental Material

sj-docx-2-taw-10.1177_20420986231152934 – Supplemental material for Development and psychometric assessment of self-reported patient medication safety scale (SR-PMSS)

Supplemental material, sj-docx-2-taw-10.1177_20420986231152934 for Development and psychometric assessment of self-reported patient medication safety scale (SR-PMSS) by Ning Qin, Yinglong Duan, Shuangjiao Shi, Xiao Li, Haoqi Liu, Feng Zheng, Zhuqing Zhong and Guliang Xiang in Therapeutic Advances in Drug Safety

Footnotes

Acknowledgements

The authors thank all the experts for their availability and valuable comments on the revision of SR-PMSS. All participants are gratefully acknowledged. The authors also thank the National Natural Science Foundation of China for supporting this study.

Declarations

ORCID iD

Zhuqing Zhong

Supplemental material

Supplemental material for this article is available online.

References

World Health Organization. Conceptual framework for the international classification for patient safety version 1.1: final technical report January 2009, 2010, https://apps.who.int/iris/handle/10665/70882

Schwendimann

Blatter

Dhaini

, et al. The occurrence, types, consequences and preventability of in-hospital adverse events – a scoping review. BMC Health Serv Res 2018; 18: 521.

Khalil

Shahid

Roughead

Medication safety programs in primary care: a scoping review. JBI Database System Rev Implement Rep 2017; 15: 2512–2526.

Alqenae

Steinke

Keers

RN.

Prevalence and nature of medication errors and medication-related harm following discharge from hospital to community settings: a systematic review. Drug Saf 2020; 43: 517–537.

Gurwitz

Field

Harrold

, et al. Incidence and preventability of adverse drug events among older persons in the ambulatory setting. JAMA 2003; 289: 1107–1116.

Garzón González

Montero Morales

de Miguel García

, et al. Descriptive analysis of medication errors notified by primary health care: learning from errors. Aten Primaria 2020; 52: 233–239.

Elliott

Camacho

Jankovic

, et al. Economic analysis of the prevalence and clinical and economic burden of medication error in England. BMJ Qual Saf 2021; 30: 96–105.

Walsh

Hansen

Sahm

, et al. Economic impact of medication error: a systematic review. Pharmacoepidemiol Drug Saf 2017; 26: 481–497.

Khalil

Kynoch

Hines

Interventions to ensure medication safety in acute care: an umbrella review. Int J Evid Based Healthc 2020; 18: 188–211.

10.

Winterstein

Johns

Campbell

, et al. Development of a medication safety and quality survey for small rural hospitals. J Patient Saf 2017; 13: 249–254.

11.

Lin

Lyu

, et al. Development and the reliability and validity test of nurse medication unsafe behavior scale. Chin J Mod Nurs 2018; 24: 1241–1246.

12.

Park

Seomun

Development and validation of the medication safety competence scale for nurses. West J Nurs Res 2021; 43: 686–697.

13.

Hohl

Hunte

, et al. Clinical decision rules to improve the detection of adverse drug events in emergency department patients. Acad Emerg Med 2012; 19: 640–649.

14.

Hohl

Badke

Zhao

, et al. Prospective validation of clinical criteria to identify emergency department patients at high risk for adverse drug events. Acad Emerg Med 2018; 25: 1015–1026.

15.

Wang

Zhang

, et al. The development and psychometric testing of inpatients’ involvement in medication safety scale (IIMSS). J Nurs Manag 2019; 27: 1648–1654.

16.

World Health Organization. 5 moments for medication safety: Flyer, 2019, https://www.who.int/publications/i/item/WHO-HIS-SDS-2019.3

17.

Donabedian

Evaluating the quality of medical care. Milbank Q 1966; 44: 166–206.

18.

Ayanian

Markel

Donabedian’s lasting framework for health care quality. N Engl J Med 2016; 375: 205–207.

19.

Runciman

Hibbert

Thomson

, et al. Towards an international classification for patient safety: key concepts and terms. Int J Qual Health Care 2009; 21: 18–26.

20.

Gala

Moshokgo

Seth

, et al. Medication errors and blood pressure control among patients managed for hypertension in public ambulatory care clinics in Botswana. J Am Heart Assoc 2020; 9: e013766.

21.

Morisky

Ang

Krousel-Wood

, et al. Predictive validity of a medication adherence measure in an outpatient setting. J Clin Hypertens 2008; 10: 348–354.

22.

Vrijens

De Geest

Hughes

, et al. A new taxonomy for describing and defining adherence to medications. Br J Clin Pharmacol 2012; 73: 691–705.

23.

de Souza

Carissimi

Costa

, et al. The mood rhythm instrument: development and preliminary report. Braz J Psychiatry 2016; 38: 148–153.

24.

Zhong

Shi

Duan

, et al. The development and psychometric assessment of Chinese medication literacy scale for hypertensive patients (C-MLSHP). Front Pharmacol 2020; 11: 490.

25.

Carneiro

AV.

Estimating sample size in clinical studies: basic methodological principles. Rev Port Cardiol 2003; 22: 1513–1521.

26.

Anderson

Gerbing

DW.

The effect of sampling error on convergence, improper solutions, and goodness-of-fit indices for maximum likelihood confirmatory factor analysis. Psychometrika 1984; 49: 155–173.

27.

Sapnas

Zeller

RA.

Minimizing sample size when using exploratory factor analysis for measurement. J Nurs Meas 2002; 10: 135–154.

28.

Shi

Sun

Content validity index in scale development. Zhong Nan Da Xue Xue Bao Yi Xue Ban 2012; 37: 152–155.

29.

Clark

Watson

Constructing validity: new developments in creating objective measuring instruments. Psychol Assess 2019; 31: 1412–1427.

30.

Baumergartner

Homburg

Applications of structural equation modeling in marketing and consumer research: a review. Int J Res Mark 1996; 13: 139–161.

31.

Reichardt

Coleman

SC.

The criteria for convergent and discriminant validity in a multitrait-multimethod matrix. Multivariate Behav Res 1995; 30: 513–538.

32.

Hamid

Sami

Mohmad Sidek

MH.

Discriminant validity assessment: use of Fornell & Larcker criterion versus HTMT criterion. J Phys Conf Ser 2017; 890: 012163.

33.

McKelvie

SJ.

Split-half reliability of two facial imagery questionnaires. Percept Mot Skills 1998; 87(Pt. 1): 953–954.

34.

Bland

Altman

DG.

Cronbach’s alpha. BMJ 1997; 314: 572.

35.

Kimberlin

Winterstein

AG.

Validity and reliability of measurement instruments used in research. Am J Health Syst Pharm 2008; 65: 2276–2284.

36.

McDowell

Jenkinson

Development standards for health measures. J Health Serv Res Policy 1996; 1: 238–246.

37.

Marsh

Morin

Parker

, et al. Exploratory structural equation modeling: an integration of the best features of exploratory and confirmatory factor analysis. Annu Rev Clin Psychol 2014; 10: 85–110.

38.

Ahmed

Ishtiaq

Reliability and validity: importance in medical research. J Pak Med Assoc 2021; 71: 2401–2406.

39.

Bagheri Hosseinabadi

Etemadinezhad

Khanjani

, et al. Evaluating the relationship between job stress and job satisfaction among female hospital nurses in Babol: an application of structural equation modeling. Health Promot Perspect 2018; 8: 102–108.

40.

Bagozzi

Fornell

Larcker

DF.

Canonical correlation analysis as a special case of a structural relations model. Multivariate Behav Res 1981; 16: 437–454.

41.

Hinton

McMurray

Brownlow

SPSS explained. 2nd ed. London: Routledge, 2014.

42.

Gao

, et al. Influence of the medication environment on the unsafe medication behaviour of nurses: a path analysis. J Clin Nurs 2018; 27: 2993–3000.

43.

Olalemi

Muyibi

Ladipo

MM.

Perceived family support and medication adherence amongst hypertensive outpatients in a tertiary hospital, Ibadan, Nigeria. West Afr J Med 2020; 37: 481–489.

44.

Sinopoulou

Rutter

Price

, et al. Implementing an emergency department pharmacy service and its effect on medication safety. Int J Pharm Pract 2021; 29: 394–396.

45.

Monkman

Schmidt

Nøhr

Online medication information for citizens: a comparison of demands on eHealth literacy. Stud Health Technol Inform 2020; 270: 1026–1030.

46.

Carpenter

Elstad

Blalock

, et al. Conflicting medication information: prevalence, sources, and relationship to medication adherence. J Health Commun 2014; 19: 67–81.

47.

Assiri

Shebl

Mahmoud

, et al. What is the epidemiology of medication errors, error-related adverse events and risk factors for errors in adults managed in community care contexts? A systematic review of the international literature. BMJ Open 2018; 8: e019101.

48.

Zaugg

Korb-Savoldelli

Durieux

, et al. Providing physicians with feedback on medication adherence for people with chronic diseases taking long-term medication. Cochrane Database Syst Rev 2018; 1: CD012042.

49.

Uhlenhopp

Aguilar

Dai

, et al. Hospital-wide medication reconciliation program: error identification, cost-effectiveness, and detecting high-risk individuals on admission. Integr Pharm Res Pract 2020; 9: 195–203.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB