Study design
Telephone cognitive assessment was first administered to AF patients using T-MoCA. Patients were divided into high educational level (≥ 12 years of education) and low educational level (< 12 years of education). Patients who have completed the telephone interview were invited for an in-person cognitive function assessment using Clinical Dementia Rating (CDR) and Mini-Mental Status Evaluation (MMSE). Using CDR = 0.5 as a reference standard for MCI, the sensitivity and specificity of T-MoCA were analyzed in overall population and educational subgroups. The performance of T-MoCA was also compared with the MMSE scale in MCI screening in AF patients.
Study population
Three hundred and four consecutive patients in the waiting list for AF ablation in Beijing Anzhen Hospital, a tertiary medical center in Beijing, China, were invited from January 2019 to July 2019 to participate in a telephone interview by trained physicians using T-MoCA. Among them, 159 refused to participate and 9 did not finish the interview (2 patients had hearing problems, 2 patients did not speak Mandarin, and 5 patients had other reasons); In total, 136 patients completed the T-MoCA interview (Fig. 1). Within this cohort, 28 patients were not in-person interviewed before they received catheter ablation. In order to avoid possible impact of catheter ablation on cognitive impairment[7], no further in-person interview was performed in those patients. Another 7 patients withdrew their consent to participate. In the end, 101 patients participated in an in-person cognitive interview within one month after the telephone interview.
Data collection
Data on history of hypertension, diabetes, coronary heart disease, heart failure, smoking, and alcohol use were collected in all patients during the in-person interview. Results of laboratory tests were extracted from a medical chart by a cardiologist.
Telephone interviewed Montreal Cognitive Assessment (T-MoCA)
T-MoCA is a simplified version of MoCA from which trail making, visual structure and naming are removed. It consists of digit span, attention, calculation, repetition, verbal fluency, abstraction, recall, and orientation, with a maximum score of 22. Since the Chinese version of MoCA has been widely validated and applied in previous reports[8, 9], the items of T-MoCA Chinese version are picked out from the MoCA while the sequence and combination of the items in T-MoCA were kept same with those in MoCA. Instructions for each item were strictly adhered to standardize the process of interview.
Standard cognitive function assessment
Clinical Dementia Rating (CDR) was used as a standard cognitive function assessment. CDR is a widely used semi-structured clinical measure for global cognitive status. It comprises six domains (memory; judgment and problem-solving; orientation; community affairs; home and hobbies; and personal care). The information in CDR are provided by assessing patients directly and by asking those who are familiar with the patient. All CDR raters were required to be certified (https://knightadrc.wustl.edu/CDR/CDR.htm). An online algorithm was used to calculate CDR scores (https://biostat.wustl.edu/~adrc/cdrpgm/index.html). CDR = 0.5 is used in clinical practice for MCI diagnosis in Chinese speaking populations[10, 11]. Therefore, we used CDR = 0.5 as the reference standard for T-MoCA validation.
The Mini-Mental Status Evaluation (MMSE) is a widely used scale for screening cognitive impairment[12]. Although limited by low sensitivity, it is commonly used by clinicians as it is easy to apply [4]. The threshold of MMSE for MCI varies in different studies[13]. A recently published data for the Chinese community elderly population proposed a stratified criterion (27/28 for those with a ≥ 7 years of education, 24/25 for those with an 1–6 years of education, and 19/20 for illiterates)[14]. In the present study, we used this criterion as a comparison to the T-MoCA in MCI screening.
Statistical analysis
Data analysis was conducted using R3.5.1. Normative data were presented as mean [SD] and compared using the t-test, while non-normative data were presented as median [interquartile range] and compared using the Mann-Whitney U test. Categorical data were presented as count (frequency) and compared using the Chi-square test or Fisher's exact test.
The pROC package was used for the receiver operating curve (ROC) analysis. Sensitivities and specificities for different thresholds were calculated in the overall patients and in patients with low educational level and high educational level respectively. The optimal threshold was defined as the score with highest Youden index among all scores with sensitivity > 75% and specificity > 60%. When reasonable sensitivity and specificity cannot satisfy simultaneously, the optimal threshold was simply determined by Youden index. The area under the curve (AUC) of T-MoCA and MMSE was compared using Delong's test. The net reclassification index (NRI) of optimal thresholds of T-MoCA against the optimal threshold of MMSE in the present study and optimal thresholds suggested by the MMSE normative data of the Chinese community elderly population were calculated to assess whether and to what extent T-MoCA outperforms MMSE in screening MCI.
Sample size is estimated by PASS 15.0 prior to the study. Under the null hypothesis of a sensitivity of 0.5, a minimum of 13 cases has 0.8 power to detect a sensitivity of 0.82 in the previous study at a significance level of 0.05. And a minimum of 36 cases account for 0.9 power to detect a sensitivity of 0.75, under the null hypothesis of a sensitivity of 0.5 and a significance level of 0.05. Therefore, assuming the prevalence of MCI in AF population is 30%-40%, a sample size of 100 patients was considered.
Sensitivity analysis was conducted in patients without a history of stroke.