Stroke is a leading cause of death and disability around the world (1). Ischemic stroke (IS) has high rates of morbidity, disability, recurrence and mortality (2). Insomnia is one of the common complaints of IS survivors. In contrast to other long-term sequelae of IS such as mobility and cognitive impairment, insomnia has received less attention and related research, despite being a risk factor for stroke (3). In accordance with the latest manual by the American Association of Sleep Medicine (AASM), insomnia is defined as persistent sleep problems and daytime socio-occupational dysfunction, which may be actual or perceived despite adequate sleep opportunities (4). According to the International classification of sleep disorders-3 edition (DSM-3) and the Chinese expert consensus on the assessment and management of stroke related sleep disorders (CEC-SSD) (5), stroke-related insomnia consists of two conditions: (i) post-stroke insomnia: insomnia first appears after stroke; (ii) stroke with insomnia: insomnia existing before stroke persists or worsens after stroke and meets the diagnostic criteria for insomnia. According to reports published 30 ~ 68% of poststroke patients were burdened with insomnia (6–8). In addition, insomnia is also thought to increase the risk of psychological problems (depression and anxiety), physical function (disability), and cognitive function (dementia) in IS patients (9, 10).

Strokes are more likely to occur in people who suffer from insomnia. The results of a meta-analysis of 160,867 patients in 15 studies showed that falling/maintaining asleep difficulty, and non-restorative sleep were positively strongly associated with the risk of stroke (11). According to a prospective clinical cohort study of stroke patients, insomnia patients were more likely to be depressed, anxious, disabled, and have difficulty returning to work than stroke patients without insomnia after 1 year after stroke (12). Surveys such as that conducted by Huang et al. (13) have shown that a negatively association between insomnia and the improvement in activities of daily living was found in subacute stroke inpatients. Tang et al. (14) showed that insomnia may make stroke survivors more susceptible to suicide. Kim et al. (10) demonstrated that insomnia had a negative effect on quality of life in stroke patients at the initial phases of rehabilitation. Also, insomnia IS patients have a higher recurrence stroke rate in the first year and a higher mortality rate within 6 years compared with no-insomnia IS patients (15).

Although the prevalence of insomnia continues to rise in many countries around the world (16, 17), patients often treat insomnia as a lifestyle issue rather than a major health problem (18). Also, the importance of insomnia diagnosis has been neglected in the routine medical examination of some IS patients. Clinician assessment and judgment based on the DSM-5 and CEC-SSD remain the criteria for diagnosing insomnia in stroke patients. Polysomnography (PSG) is considered as a common objective measure for the measurement of sleep disorders. However, PSG is not easy to obtain for most clinicians’ daily insomnia diagnosis routine (19), and is mostly used for the diagnosis of sleep disordered breathing (SDB). At the same time, PSG is very expensive for epidemiology and research, takes a long time and is not in line with clinical practice (19). However, multiple insomnia questionnaires have been developed including the PSQI, ISI, and AIS. All three questionnaires were used multiple times and translated into multiple languages and have been widely used in China. These questionnaires are considered to be efficient screening tools for insomnia. They are simple to perform and cost effective and do not demand additional special apparatus or facilities. The use of brief questionnaires for subjective assessment saves time and effort and ensures a high level of patient subjective willingness and compliance. Also, due to the self-explanatory nature of these tools, the need for and reliance on specialists and clinicians can be significantly reduced (20, 21). The three insomnia assessment tools used in this study, PSQI, ISI, and AIS, were not developed specifically for IS patients, and each scale validation was based on primary sleep disorders patients (22). Because of the significant role that the rating scales plays in the diagnosis of insomnia in IS patients, psychometric properties, cut-off values and diagnosis effectiveness of these tools need to be evaluated in IS patients. The purpose of this study was to the establish reliability and determine the optimal threshold values of PSQI, ISI, and AIS in the diagnosis of insomnia of IS patients.

Insomnia is a common psychiatric complaint in IS patients (9). The reported prevalence of insomnia varies widely among studies due to the different definitions and assessment tools used. Among the 569 IS patients contained in this study, 89 were diagnosed as insomnia by clinicians. The prevalence of pre-stroke insomnia in IS patients was found to be 15.64% in this study, lower than that of previously reported levels. Surveys such as that conducted by Leppävuori et al. (49) have shown that 38.6% had insomnia before the stroke and 18.1% had insomnia after the stroke. Another study found a 12% rate of new-onset insomnia after stroke when excluding patients who had insomnia before stroke (50). A study from China found that the prevalence of insomnia in stroke patients was 57.9%, of which 32.2% had insomnia before the stroke and 25.7% had new insomnia after the stroke (51). A 4-year follow-up study of 21,438 insomniacs and 64,314 non-insomniacs found that the incidence of stroke was significantly higher in insomniacs than in non-insomniacs (incidence rate ratio = 1.85; 95% CI: 0.7, 2.05) and those insomniacs had 54% higher risk of developing stroke compared with non-insomniacs (adjusted hazard ratio = 1.54; 95% CI: 1.38, 1.72) (3). There is mounting clinical evidence to support a bidirectional relationship between insomnia and stroke. Insomnia is likely to be an independent risk factor for stroke, and stroke may also be a causative factor for the development of insomnia. Insomnia should be given more attention in IS patients.

This study demonstrated high sensitivity and specificity of the PSQI, ISI, and AIS in screening for insomnia in IS patients. It was discovered that the PSQI scale was the most sensitive (identifying insomnia in 100% of the IS patients when the cutoff value is 9), while DeLong’s test indicated that ISI and ASI were less but similarly sensitive, with ISI identifying insomnia in 91.01% of the IS patients when the cutoff value is 15, and AIS identifying insomnia in 75.28% of the IS patients when the cutoff value is 8. Some studies suggest that the diagnostic sensitivity to distinguish good sleep from poor sleep was 89.6% and the specificity was 86.5% when the total PSQI score > 5 (52). The Youden index of the PSQI was 0.852, with a strong ability to screen insomnia patients from non-insomnia patients. When the effect of morbidity was excluded, the positive likelihood ratio (PLR) for the PSQI was 6.761, which is a higher than ISI and AIS. Also, according to Charles, the ISI cutoff of 10 was the optimal score for identifying insomnia cases in the community people, with an 86.1% sensitivity and 87.7% specificity (36). When diagnosing insomnia on a score of ≥6, the AIS scale had a sensitivity of 93%, and a specificity of 85%, and the overall correct case identification rate is 90% (44). This confirmed the findings of these scales have excellent performance of these scales in screening for insomnia in IS patients in the literature (34, 36, 42). Our analysis showed that, identical to the widely used cutoff values for insomnia diagnosis in the general population in China, the best cutoff values in screening for insomnia in IS patients of the ISI questionnaires were “15.” However, the best cutoff values in screening for insomnia in IS patients of the PSQI and AIS questionnaires were “9” and “8” respectively, higher than the widely used diagnostic cutoff values for insomnia population in China.

Cohen’s kappa consistency statistic check was used to analyze the consistency of PSQI, ISI, and AIS with clinician-based insomnia diagnostic in IS patients. The kappa coefficient of PSQI, ISI, and AIS were 0.643, 0.459, and 0.539 and the asymptotic 95% confidence interval were (0.570, 0.716), (0.385, 0.533), and (0.451, 0.627). Landis and Koch (53) proposed that a kappa in the range of 0.41 ~ 0.60 be considered “moderate” agreement, kappa = 0.61 ~ 0.80 be considered “substantial” agreement, and kappa >0.81 be considered “almost perfect” agreement. This indicates that the consistency between PSQI with clinician-based insomnia diagnostic were substantial consistent. The consistency between ISI and AIS with clinician--based insomnia diagnostic was moderately consistent.

The PSQI focuses on the sleep status of patients in the past 4 weeks. Also, there are four types of responses (“not during the last month,” “less than once a week,” “once or twice a week,” “three or more times per week”) instead of using “very poor,” “poor,” “good,” “very good” and similar language to assess sleep. Such descriptions are important to accurately measure the severity of the IS patient’s sleep-related components. We believe this could help our patients better understand the PSQI question and choose the option that best matches their situation, compared to the AIS and ISI. In addition, these descriptions and response category of sleep quality severity can be translated into numerical rating ranges, so this may explain the greater sensitivity of PSQI in assessing insomnia in IS patients. PSQI is a widely used sleep quality evaluation tool to measure sleep quality and disturbances over the prior month and to discriminate between “good” and “poor” sleepers (22). Meanwhile, PSQI is often used together with the Epworth Sleepiness Scale (ESS) to assess daytime sleepiness, such as in medical students (54), COPD and asthma (55), or women with premenstrual syndrome (56).

Although the ROC curve was used to evaluate the diagnostic performance of the three questionnaires in insomnia, it was not the only thing to consider in choosing an assessment tool. The three questionnaires have similar content and language descriptions of sleep problems. Each of the three questionnaires has advantages and disadvantages when assessing insomnia. The PSQI cannot collect information about the patient’s need of treatment options and other aspects. The PSQI scale has more questions and takes longer to assess compared to the ISI and AIS, so it performs poorly in rapidly assessing the severity of insomnia in clinical patients. It should be emphasized that PSQI is the only one of the three questionnaires that needs to calculate and transform scores. Because of the need to integrate various responses and calculate such variables as sleep efficiency, hand-calculation of scores may be somewhat burdensome (22). This undoubtedly caused additional inconvenience and time consumption for clinical evaluation.

One advantage of ISI is that there are data based on which insomnia can be classified hierarchically. In a study which set out to investigate insomnia in long-term hospitalized older adults, Aluzaite et al. (57) classified ISI scores as follows: 0 ~ 7 points represent no insomnia; 8 ~ 14 points represent sub-threshold insomnia; 15 ~ 21 points represent moderate insomnia; 22 ~ 28 points represent severe insomnia. Under this score classification, the degree of insomnia in patients can be measured. Meanwhile, for PSQI and AIS such thresholds to measure the severity of insomnia in patients have not yet been investigated. In many clinical assessments, PSQI is often used as an index to measure the overall sleep quality, and ISI is used to measure the severity of insomnia (58). ISI focuses on the subjective symptoms of insomnia, the consequences and the degree of distress, and is sensitive to detect changes in sleep conditions brought about by treatment. However, researchers need to be alert to the possibility of false positive results when patients subjectively amplify their insomnia feelings.

The AIS is designed to assessment subjective feeling of sleep difficulty that only the subjective feelings of the interviewee are considered for the rating (41). In addition, the severity of the patient’s sleep difficulty as estimated by the interviewer, such as the hours of sleep or the approximate duration of sleep onset latency (41), was not considered in the AIS score since the interviewer’s judgment may not be consistent with the patient’s subjective assessment. Patients may exaggerate the negative feelings caused by short sleep duration, leading to bias in AIS assessment, which may also be the reason for the lower sensitivity of AIS.

There are several limitations to this study: All subjects were not tested with Polysomnography in this study. We compared the application of the three questionnaires in the diagnosis of insomnia, but did not give grade judgments for the severity of insomnia. Finally, because the information was gathered based on self-reporting, the data may be prone to information bias.

This study demonstrated high sensitivity and specificity and excellent diagnostic ability of the PSQI, ISI, and AIS questionnaires in screening for insomnia in IS patients. We found the best diagnostic cutoffs for PSQI, ISI, and AIS are ≥9, ≥15, and ≥8, respectively. Based on these cutoff values, the PSQI scale was revealed to be the most sensitive, while the ISI was the moderate sensitive and AIS was the least sensitive. Each of the three questionnaires has advantages and disadvantages when assessing insomnia. In the evaluation of insomnia in IS patients, the best questionnaire selection should be made according to the purpose of clinical evaluation (such as screening potential insomnia patients or the diagnosis of clinical insomnia) and considering the sensitivity and specificity of the questionnaire.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Institutional Review Boards of the Tenth People’s Hospital, Tongji University (Approval No. SHSY-IEC-KY-4.0/17–47/01). The patients/participants provided their written informed consent to participate in this study.

SN and YS: study conception and design, manuscript drafts and revisions. SN, SD, LCW, and LW: data collection. SN and QW: data analysis. SN, QW, SD, LCW, and YS: agree with manuscript results and conclusions. All authors approved the final manuscript and act as guarantors for the study.

This research was supported by the grants from the general project of National Natural Science Foundation of China (71774117).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.