A tool for recognizing asthma nonadherence during disease management: A prospective cohort study

Abstract

BACKGROUND:

The adherence of asthma patients remains an urgent issue that needs to be addressed, and there is still a lack of tools for the rapid identification of nonadherence.

METHODS:

A number of factors associated with increased risk of nonadherence were compared with adherence assessed using the Medication Adherence Report Scale for Asthma (MARS-A). Among them was the tool for recognizing asthma nonadherence (TRAN), which is the ratio between the patient’s perceived overall health measured using a 0–100 visual analog scale and their asthma control test score. The development set used data from a multicenter, prospective cohort study in asthma. The TRAN was then validated in a further test set of asthma patients.

RESULTS:

Data from 518 participants who completed 3-month follow-up formed the development set. TRAN was the best predictor of nonadherence as defined by MARS-A (odds ratio = 9.14; 95% confidence interval [CI]: 4.16, 20.10; P < 0.01) in univariate logistic regression analysis. Its area under the curve (AUC) for identifying nonadherent patients was 0.810 (95% CI: 0.753, 0.866), at a cutoff score of 1.08, its sensitivity was 82.0%, specificity was 84.4%, positive predictive value (PPV) was 55.8%, and negative predictive value (NPV) was 95.1%. In the validation group (n = 175), the AUC was 0.75 with PPV and NPV at 56.4% and 90.8%, respectively.

CONCLUSION:

The TRAN is a simple method for the initial screening of asthma patients for potential nonadherence.

Background

Asthma is a chronic, noncommunicable disease that affected more than 262 million people worldwide in 2019.[1] Globally, adequate asthma control is not yet achieved,[2,3] and adherence to asthma controller medication has been considered to be one of the key factors,[4] because it leads to poor clinical outcomes, diminished quality of life, and increased economic burden.[5,6,7,8] Previous studies have shown that medication adherence of asthma patients is far from optimal and with estimates of nonadherence typically ranging between 30% and 75% in both adults and children.[9,10,11,12,13]

Current methods for assessing nonadherence such as electronic device monitoring, gas tank weight, biochemical and pharmaceutical data monitoring, and questionnaires, all retrospectively examine the patients’ adherence over the past several weeks or months.[14,15,16,17] A significant limitation of retrospectively assessing nonadherence is the potential loss of patients from follow-up and then return to their physician during the next exacerbation, perpetuating a cycle of poor asthma control and exacerbations.

Nonadherence behavior can be broadly categorized into intentional and unintentional nonadherence.[18] Ma and colleagues showed that intentional nonadherence constitutes the majority of nonadherence in asthma patients.[12] Clifford and colleagues showed that intentional nonadherers were significantly more likely to doubt their personal need for medication and harbor stronger concerns about using their medication compared to adherers.[19] Moreover, patients’ healthcare-seeking behavior is driven by symptom severity, and severity of symptoms influences the overall health preferences.[20,21] When patients perceive their health status as better than the severity of their disease symptoms, they are more likely to doubt their personal need for medication, leading to weakened beliefs. Therefore, we hypothesized that intentional nonadherence is more likely to occur when the severity of the disease conflicts with an individual’s health preferences.

This study aimed to validate a new tool (tool for recognizing asthma nonadherence [TRAN]) to identify patients at risk of nonadherence in primary care, thereby equipping clinicians with a direct and effective method for timely detection and intervention.

Materials and Methods

Study design

We conducted a multicenter, prospective cohort study at 3 hospitals in Chongqing from September 2022 to April 2023. This 3-month study assessed the contribution of the patients’ overall health preferences and symptoms to predicting nonadherence. It consisted of two clinical visits, at baseline and 3 months later. A schematic representation of the study design is shown in Figure 1. The study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice and was approved by the Medical Ethics Committee of the North Korean General Hospital (Songshan Hospital) and registered on chictr.org.cn (No: ChiCTR2200063343). All participants provided written informed consent.

Figure 1.

Flow of participants through the study.

The validation cohort originated from an earlier study in asthma patients aged 18 years or over receiving specific immunotherapy (ChiCTR2200065686).

Patients and measurement

The inclusion criteria were as follows: 18–75 years of age; diagnosed with asthma based on the criteria established by the Global Initiative for Asthma;[22] and prescribed inhaled corticosteroids (ICS), either alone or in combination with other medication, such as long-acting ß-receptor agonist and/or long-acting M-receptor agonists, in the same inhaler. The exclusion criteria included chronic obstructive pulmonary disease or other active medical problems (e.g., bronchiectasis, cystic fibrosis, pulmonary tuberculosis, lung cancer, and severe heart disease); patients who were unable to complete questionnaires (including illiterate and unable to express opinions through alternative means, unable to understand questions of questionnaires, etc.); and pregnancy.

Formation of the tool

TRAN was composed of the ratio of the patient’s perceived overall health to the level of asthma control. Overall health was measured using a visual analog scale (VAS) in which 0–100 score was used to assess the patients’ perception of their overall health in the previous 4 weeks, 100 corresponded to the best health patients could imagine, and zero means the worst health. Asthma control was assessed using the asthma control test (ACT). It comprises five questions, and each question is answered on a 5-point scale, with a total score ranging from 5 to 25.[23] The calculation formula for TRAN is VAS/(ACT * 100/25).

Formation of the groups

The Medication Adherence Report Scale for Asthma (MARS-A) questionnaire was used as a reference for assessing adherence.[24,25,26] It has 10 questions. Patients completed the questionnaire based on their medication adherence over the previous 4 weeks. The nonadherent group was defined as patients with a mean MARS-A score <4.5 points. In the validation cohort, patients were divided into the “risk group” and the “nonrisk group” according to the cutoff value of TRAN, which was generated from data of the development set.

Statistical analysis

Means and standard deviations were computed for variables including age, ACT, VAS, TRAN, forced expiratory volume in 1 s (FEV 1), and FEV 1% predicted. An independent sample t-test was used to compare continuous variables between adherent and nonadherent groups. A Chi-square test was used to test categorical variables. Univariate tests of association between each clinical variable and the nonadherence assessed using the MARS-A were performed using logistic regression. Receiver operating characteristic curve analysis was performed, and area under the curve (AUC) values (95% confidence intervals [CIs]) were computed for TRAN in the development set. The sensitivity, specificity, positive and negative predictive value (NPV), and diagnostic odds ratio (OR) including 95% CIs were calculated for different cutoff values All statistical tests were two-sided; P < 0.05 was considered statistically significant. All analyses were carried out using SPSS 27(IBM Corp., Armonk, NY, USA).

Results

The baseline characteristics of the 518 patients from the development cohort and 175 patients from the validation cohort are summarized in Table 1. The subjects’ age range was 20–74 years, and 200 of these subjects were men. In the development cohort, during the 3-month follow-up period, 62.2% achieved asthma control, defined as ACT ≥20. In the development and validation cohorts, the number of nonadherent patients defined by MASR-A score <4.5 was 100 (19.3%) and 42 (24%), respectively. The proportion of nonadherent patients with only primary school education was significantly higher than that of adherent patients in both the development and validation cohorts (both P < 0.001).

Table 1.

Baseline study characteristics.

	Development cohort			Validation cohort
	Adherence group	Nonadherence group	P	Adherence group	Nonadherence group	P
n	418	100		133	42
Age, mean (SD)	50.0 (12.5)	43.7 (12.8)	0.37	46.1 (13.1)	44.8 (13.3)	0.59
Males, n (%)	166 (39.7)	34 (34.0)	0.29	53 (39.8)	12 (28.5)	0.03
ACT score, mean (SD)	20.7 (5.00)	15.9 (5.0)	<0.001	20.7 (4.3)	16.3 (4.2)	<0.001
VAS score, mean (SD)	78.9 (15.1)	76.2 (17.0)	0.15	81.5 (13.9)	76.3 (17.1)	0.045
TRAN, mean (SD)	1.00 (0.37)	1.36 (1.06)	<0.001	1.01 (0.17)	1.20 (0.30)	<0.001
FEV1 (postbronchodilator), mean (SD)	2.33 (0.90)	2.16 (0.88)	0.08	2.49 (0.89)	2.16 (0.87)	0.03
FEV1% predicted (postbronchodilator)	78.9 (21.1)	73.0 (24.5)	0.02	80.6 (16.2)	71.4 (22.5)	0.02
Primary school, n (%)	22 (5.3)	21 (21.0)	<0.001	6 (4.5)	9 (21.4)	<0.001
Junior high school, n (%)	79 (18.9)	12 (12.0)	0.10	29 (21.8)	6 (14.3)	0.29
Senior high school, n (%)	72 (17.2)	13 (13.0)	0.31	25 (18.8)	6 (14.3)	0.50
Senior high school, n (%)	245 (58.6)	54 (54.0)	0.40	73 (54.9)	21 (50.0)	0.58

SD=Standard deviation, ACT=Asthma control test, VAS=Visual analog scale, TRAN=Tool for recognizing asthma nonadherence, FEV₁=Forced expiratory volume in 1 s

Univariate logistic regression analyses were used to evaluate factors associated with nonadherence. A total of 9 variables were analyzed, of which 6 were associated with nonadherence at P < 0.05 [Table 2]. The strongest predictor was TRAN (OR 9.14; 95% CI: 4.16, 20.10; P < 0.01). ACT, FEV 1% pred, junior high school versus primary school, senior high school versus primary school, and junior college or above versus primary school also showed a significant but lower OR for an association with nonadherence.

Table 2.

Univariate logistic regression models for nonadherence in the development cohort.

	OR (95% CI)	P
Age	0.992 (0.975–1.010)	0.365
Gender: Male versus female	1.279 (0.809–2.021)	0.293
ACT	0.852 (0.817–0.889)	<0.001
VAS	0.989 (0.976–1.003)	0.118
TRAN	9.141 (4.158–20.097)	<0.001
FEV1 (L)	1.014 (0.759–1.355)	0.923
FEV1% predicted (postbronchodilator)	0.988 (0.979–0.998)	0.015
Education: Junior high school versus primary school	0.159 (0.068–0.373)	<0.001
Education: Senior high school versus primary school	0.189 (0.082–0.438)	<0.001
Education: Junior college or above versus primary school	0.231 (0.119–0.450)	<0.001

ACT=Asthma control test, VAS=Visual analog scale, TRAN=Tool for recognizing asthma nonadherence, FEV₁=Forced expiratory volume in 1 s, OR=Odds ratio, CI=Confidence interval

In the development dataset, the AUC of TRAN for predicting nonadherence was 0.810 (95% CI: 0.753, 0.866) [Figure 2], with an optimal cutoff value for TRAN of 1.08. The discriminative performance of the TRAN is shown in Table 3. With a cutoff value of 1.08, sensitivity was 82% and specificity was 84.4%. At a cutoff value of 1.1, the sensitivity was 80% and specificity 85.2%. Of note, when the cutoff value is set at ≥1.3, the sensitivity was markedly decreased to only 39%.

Figure 2.

Receiver operating characteristic curves for prediction of the nonadherence in development cohort. TRAN: Tool for recognizing asthma nonadherence.

Table 3.

The diagnostic efficacy of tool for recognizing asthma nonadherence for asthma nonadherence.

TRAN ODI	Sensitivity (95% CI)	Specificity (95% CI)	PPV (95% CI)	NPV (95% CI)	Diagnostic OR (95% CI)
TRAN ≥1.08	82.0% (72.8%–88.7%)	84.4% (80.5%–87.7%)	55.8% (47.4%–63.9%)	95.1% (92.3%–97.0%)	24.7 (13.9–44.0)
TRAN ≥1.1	80.0% (70.6%–87.1%)	85.2% (81.3%–88.4%)	56.3% (47.8%–64.6%)	94.7% (91.8%–96.6%)	23.0 (13.1–40.2)
TRAN ≥1.2	62.0% (51.7%–71.4%)	89.5% (86.0%–92.2%)	58.5% (48.5%–67.9%)	90.8% (87.5%–93.3%)	13.9 (8.3–23.1)
TRAN ≥1.3	39.0% (30.0%–49.3%)	92.6% (89.5%–94.8%)	55.7% (43.4%–67.4%)	86.4% (82.8%–89.4%)	8.0 (4.6–13.7)

TRAN=Tool for recognizing asthma nonadherence, CI=Confidence interval, OR=Odd ratio, PPV=Positive predictive value, NPV=Negative predictive value, ODI=Optical density index

The included patients had a wide age range, spanning from 20 to 74 years. Therefore, we stratified the patients into three age groups: ≥60 years, 59–40 years, and <40 years. In the development cohort, the ≥60 years group comprised 64 patients, with a TRAN AUC of 0.828 (95% CI: 0.682, 0.974). The 59–40 years group included 236 patients, with a TRAN AUC of 0.824 (95% CI: 0.756, 0.892). The <40 years group consisted of 218 patients, with a TRAN AUC of 0.788 (95% CI: 0.681, 0.895). Across different age strata, TRAN demonstrated robust diagnostic performance, with minimal differences in AUC values among the age groups.

TRAN also showed a good discrimination in the validation cohort, with an AUC value of 0.75 (95% CI: 0.64, 0.85). Its positive predictive value and NPV were 56.4% and 90.8%, respectively. In this cohort, we categorized patients into a risk group and a nonrisk group based on TRAN score ≥1.08. In the “risk group,” 56.4% (31/55) of subjects were nonadherent as defined by MARS-A and 9.2% (11/110) were nonadherent in the “nonrisk group [Figure 3].” The rate of nonadherence in the “risk group” was significantly higher than in the “nonrisk group” (risk ratio 6.15, 95% CI: 3.34, 11.31; P < 0.001).

Figure 3.

Nonadherence in two groups in validation cohort, (a) adherence/nonadherence subjects in risk group and nonrisk group, (b) adherence/nonadherence rates in risk group and nonrisk group.

Discussion

In this study, we developed and validated a straightforward and effective predictive tool to identify the risk of treatment nonadherence. It utilizes the ratio of the patient’s perceived overall health to their level of asthma control. It has two components – a VAS which has been successfully applied in other studies to measure overall health preferences[27,28] and the ACT which is widely used to assess the level of asthma control.[29,30] Its purpose is unlike other questionnaires that quantify the result of nonadherence behaviors because it is designed to predict the likelihood of the nonadherence.

The greatest strength of the TRAN tool lies in its ability to identify individuals potentially at high risk of nonadherence. With a cutoff value of 1.08, the tool demonstrated strong sensitivity (82.0%) and specificity (84.4%). In clinical practice, high sensitivity is essential for a screening tool to swiftly identify nonadherent patients and minimize the risk of such patients going undetected. Simultaneously, robust specificity helps prevent inappropriate medical interventions, thereby enhancing the efficiency of healthcare resource utilization. Furthermore, TRAN does not require additional assessments and can be easily derived through a simple calculation, which does not significantly extend the duration of clinical visits. This makes it suitable for routine clinical use. In the clinical workflow, the TRAN tool could potentially be integrated into the electronic health record system to facilitate the early identification of patients at risk of nonadherence. Although the false-positive rate in the risk group of the validation cohort was slightly higher (43.6%), the patients classified as “at risk” were 6 times more likely to exhibit medication nonadherence compared to those judged by the TRAN to be not at risk. Compared to the general population, the TRAN tool allows us to focus more on asthma patients at high risk of nonadherence. Of course, this remains an ongoing concern, as continuous assessment of patients is essential in the management and treatment of chronic diseases such as asthma. Hence, TRAN has significant value as a straightforward initial screening tool because it narrows the size of the population that requires intervention. Providing physicians with a convenient and practical method to identify patient adherence could instigate a paradigm shift in care.

The TRAN is a tool designed from a behavioral perspective to discover behavioral problems in asthma patients. Dissatisfaction with the current state or hope of achieving a more satisfactory state can be the first action step. Dissatisfaction with their current level of asthma control or a willingness to achieve better control may be key factors that encourage the patient to follow the advice of healthcare providers. Misleadingly positive self-health perceptions blind the patient’s perception of the problem at hand. This leads to a lack of motivation to follow recommendations, thereby disrupting the link between medication adherence and symptom relief or disease progression.[19,31,32] Consequently, a diminished belief in the necessity of medication significantly increases the likelihood of nonadherence.[33] A longitudinal study of individuals living with HIV demonstrated that beliefs about necessity and concern were predictive of adherence behavior.[34] To effectively address intentional nonadherence, motivational and cognitive behavioral interventions may be more successful than educational strategies alone in reshaping patients’ beliefs about ICS.[35] Therefore, openly discussing these beliefs and collaboratively establishing a mutually agreeable disease management plan is a critical first step in optimizing patient outcomes. When clinicians and nursing staff are aware of patients’ “hidden” beliefs and behaviors, they can further guide targeted behavioral change interventions through tailored assessments, engage in shared decision-making with patients, conduct motivational interviews, and provide self-management interventions in various formats to enhance patient adherence. Certainly, more research and evidence are needed in the future to validate these approaches. These approaches may lead to higher-quality decisions that better align patients’ needs with evidence-based recommendations.[36,37]

This study has some potential limitations. First, we selected the MARS-A self-report scale as the foundation for measuring adherence in this study, rather than relying on objective standards such as electronic monitoring devices which were regarded as the “gold standard.”[15] However, this study does not specify the use of fixed inhalation devices, leading to the involvement of multiple electronic monitoring devices. That approach necessitates substantial resources and funding to acquire different types of electronic devices, with the possibility that suitable devices may not be available for each type. Restricting the study to one or two electronic monitoring devices may impact adherence due to patients’ preferences for certain inhalation devices. Moreover, excessive intervention could adversely affect patient adherence. The MARS-A integrates the Morisky Scale and Drug Attitude Inventory Scale to assess medication adherence from three aspects: attitude, belief, and behavior.[24] Furthermore, compared to these objective methods, MARS-A also demonstrates strong criterion validity.[38] Future studies should validate the TRAN by comparing it with objective measures, such as electronic monitoring devices or medication container weight assessments. Finally, the TRAN is designed as an initial screening tool and should be used in conjunction with professional psychobehavioral assessment questionnaires to further explore the motivations behind nonadherence.

Conclusion

This study shows that TRAN has practical value in the preliminary screening of asthma nonadherence. An optimal cutoff was determined at which it demonstrated good sensitivity and specificity. More prospective studies are required to further validate this potentially useful tool.

Authors’ contributions

Qiao Zhang and Yi Lan are co-first authors of this work. Data collection – Qinqin Wang, Linjun Liu, Hong Li; Writing – original draft preparation, Qiao Zhang; writing – review and editing, Yi Lan; visualization, Shuo Wang, Xuxu Gou; supervision, Qianli Ma, Xiuqing Liao. All authors have read and agreed to the published version of the manuscript.

Ethical statement

The study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice and was approved by the Medical Ethics Committee of the North Korean General Hospital (Songshan Hospital) and registered on chictr.org.cn (No: ChiCTR2200063343).

Data availability statement

The datasets analyzed during the current study are not publicly available due to patient privacy concerns but are available from the corresponding author on reasonable request.

Patient consent

The written informed consent was obtained from all patients for this study.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Not applicable.