Construction and validation of a nomogram prediction model for treatment adherence in patients with ischemic stroke based on PREMISE score and knowledge-attitude-practice model

Objective To construct and validate a nomogram prediction model for treatment adherence in patients with ischemic stroke based on the PREMISE score and the knowledge-attitude-practice (KAP) model.

Methods A simple random sampling method was used to prospectively select 260 patients with ischemic stroke from two hospitals as the study subjects. They were randomly assigned to model construction group (182 cases) and model validation group (78 cases) at a ratio of 7∶3. The PREMISE scores and medication compliance behaviors under the KAP model of the patients were analyzed. The Morisky scale was employed to assess the patients' treatment adherence. Univariate and multivariate logistic regression analyses were conducted to screen the influencing factors of treatment adherence. A nomogram prediction model was constructed, and the Hosmer-Lemeshow goodness-of-fit test and the receiver operating characteristic (ROC) curve were used to evaluate the predictive performance of the model.

Results According to the Morisky scale scoring criteria, patients in the model construction group were divided into poor treatment adherence group (64 cases, accounting for 35.2%) and good treatment adherence group (118 cases, accounting for 64.8%). The results of multivariate logistic regression analysis showed that higher educational level (OR=2.781, 95%CI, 1.515 to5.105), having a stable job (OR=2.106, 95%CI, 1.255 to 3.534), higher annual household income (OR=1.992, 95%CI, 1.161 to 3.417), a modified Rankin scale (mRS) score of 0 (OR=2.618, 95%CI, 1.294 to 5.296), a National Institutes of Health Stroke Scale (NIHSS) score of < 5 (OR=2.248, 95%CI, 1.247 to 4.053), without history of diabetes (OR=3.101, 95%CI, 1.572 to 6.114), patients being aware of the importance of taking medications regularly (OR=4.106, 95%CI, 1.885 to 8.949), being satisfied with the current treatment plan (OR=3.078, 95%CI, 1.596 to 5.934), being concerned about drug side effects (OR=0.468, 95%CI, 0.268 to 0.818), being able to take medications on time (OR=2.728, 95%CI, 1.278 to 5.823), regularly attending follow-up visits (OR=2.489, 95%CI, 1.260 to 4.921), and adjusting lifestyle according to doctors' advice (OR=3.337, 95%CI, 1.830 to 6.084) were all independent influencing factors for patients' treatment adherence (P < 0.05). Based on the above independent influencing factors, a nomogram model was constructed. The ROC curve showed that the area under the curve (AUC) of the model in the model construction group and the model validation group was 0.825 and 0.782, respectively. The results of the Hosmer-Lemeshow goodness-of-fit test showed a result of χ²=6.98, P=0.420 in the model construction group, and χ²=7.35, P=0.395 in the model validation group, indicating good calibration of the model.

Conclusion The nomogram prediction model constructed based on the PREMISE score combined with the KAP model has good predictive performance and calibration. It can be used for the early identification and personalized management of patients with a high risk of poor treatment adherence in ischemic stroke, improving treatment adherence and the long-term prognosis of patients.