Guidelines for the Evaluation of Cardiorespiratory Physiotherapy in Stroke Patients

Evaluation of stroke patients is prioritized over therapeutic interventions to restore cardiorespiratory capacity. This study aimed to develop a clinically applicable guideline to evaluate cardiorespiratory physiotherapy in stroke patients based on a literature review and a modified Delphi survey. The literature search included 13,498 articles in PubMed, EMBASE, CINAHL, and Cochrane Library electronic databases. We surveyed previous articles between January 2010 and June 2019. After the option elimination process, a total of 27 documents were selected and analyzed (draft: 18, modified Delphi survey: 9). The results of this research are roughly divided into two categories. First, 31 draft items were extracted, and a modified Delphi survey questionnaire was created from a literature review. Second, an expert was asked to make two modified Delphi surveys and to modify, delete, and supplement the entries in the statistical analysis at each level to finalize the steps to 20 items. The guidelines developed in this study reflect the selective use of cardiorespiratory physiotherapy evaluation methods in the clinical setting, based on the health status of individual patients. Moreover, the guidelines may help physiotherapists make informed decisions based on expert knowledge, thereby playing a crucial role in the patient-centered treatment planning process.


Introduction
The central nervous system regulates heart rate, cardiac contractility, blood pressure, and vasomotor tone, and diseases related to the regulation of autonomous blood flow occur after the onset of stroke [1]. Globally, stroke is the second leading cause of death and disability and is closely related to human mortality [2]. This necessitates an increased demand for rehabilitation programs to improve patient functional levels and quality of life.
A local brain injury in a stroke patient causes neurological damage, resulting in disorders such as a change in consciousness level, loss of motor and sensory function, and impaired cognitive and perceptual abilities. This results in limitations in functional movement and a decrease in physical activity [3]. Stroke requires a dynamic process of repair and remodeling to maintain neural circuits, which is shaped by the behavioral experience. Neuroplasticity depends on sufficient resolution of neurological dysfunction and restoration of blood flow, and the resolution of neural dysfunction may depend on neural plasticity [4]. The spontaneous and therapeutic-induced mechanisms of plasticity that promote functional activity along the site of the stroke injury provide a positive impact on recovery [5]. The rehabilitation strategy for stroke patients mainly focuses on posture and paralysis-related exercises [6]. However, focusing only on exercises that improve cardiac function is insufficient. As a result, physical activity is limited in most stroke patients, except during function-related therapy, which involves 1 h of therapy targeting the metabolic function of the heart and comprises 30 min of physiotherapy and 30 min of occupational therapy [7]. Thus, stroke patients have low cardiorespiratory endurance due to reduced physical activity [8]. Due to an inefficient energy metabolism, such as hemiplegic gait and spasticity, stroke patients exhibit increased muscle fatigue than the general population; incomplete cardiorespiratory control, abnormal oxygen transfer in the body, and decreased lung volume lead to an abnormal increase in the oxygen demand of the body and change in the cardiorespiratory function [9].
According to previous studies, the diaphragm and intercostal muscles, responsible for inhalation and exhalation in stroke patients, are mainly controlled by the unilateral cerebral cortex and subcortical spinal nerve pathways [10][11][12]. In particular, because the diaphragm is an important muscle responsible for approximately 10-70% of ventilation in the sitting and prone positions, a unilateral diaphragm paralysis can cause a significant decrease in respiratory function [13]. Based on this theory, previous studies have reported that for muscles on the paralyzed side, the gait speed correlates with the amount of oxygen intake, and muscle loss leads to low oxygen intake [14]. In addition, maximum oxygen intake, gait speed, and endurance in subacute stroke patients are less than 50% compared to those in healthy subjects [15].
For cardiac and respiratory rehabilitation, the American Heart Association (AHA) and the American Stroke Association (ASA) offer a multidisciplinary program that efficiently manages the exercise, diet, smoking, stress, and education of patients, and its psychological impact to reduce symptoms and help patients return to daily life [16,17]. Studies evaluating pulmonary function, cardiorespiratory endurance, and the 6-Minute Walk test (6MWT) have been actively conducted to evaluate the cardiorespiratory function of stroke patients [18,19]. However, there is still a lack of awareness of cardiorespiratory physiotherapy in stroke patients, and no studies have proposed a standardized guideline for evaluating cardiorespiratory physiotherapy conducted by researchers.
This study aimed to outline evaluation methods for cardiorespiratory physiotherapy in stroke patients, and a modified Delphi survey was used to achieve consensus between the literature review and an expert panel. The Delphi method, developed by the Research and Development (RAND) Corporation, is widely used to assist healthcare professionals to participate in developmental research by compensating for negative aspects such as time and location constraints [20].
Therefore, this study examined data from previous studies that evaluated cardiorespiratory physiotherapy in stroke patients and conducted a modified Delphi survey of experts' opinions to develop clinically applicable guidelines and evaluation items for cardiorespiratory physiotherapy in stroke patients.

Search Strategy
To develop a guideline to evaluate cardiorespiratory physiotherapy in stroke patients, we systemically searched for articles published in accordance with the direction and purpose of this study between 1 January 2010 and 30 June 2019.
We used the PubMed, EMBASE, CINAHL Plus with Full Text, and Cochrane Library databases. To increase the sensitivity and specificity of the literature search, operators (AND, OR) and truncated search (*) functions were used, and synonyms and alternate words of the keyword and Medical Subject Headings (MeSH), Emtree, and CINAHL subject headings were used (Tables 1-4). In addition, in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA), key questions in the system developed to evaluate cardiorespiratory physiotherapy in stroke patients (participants, intervention, comparisons, outcomes, study design [PICO-SD]) were selected [21].

#9
Stroke AND (MM "Outcome assessment") OR (MM "Outcomes of education") OR (MM "Patient assessment") AND Pulmonary Notes: Truncation search (*) is a comprehensive search method for searching terms.

Inclusion Criteria
In accordance with the PICO-SD, the subjects were adult stroke patients over 18 years of age, and the intervention measures included a single intervention (the single intervention group) or a combination of two or more interventions (the combination group). The results consisted of a single variable or two or more variables, including exercise interventions related to cardiac and respiratory physiotherapy. The study design included descriptive observational studies, case series; observational analytical, cross-sectional, case-control, and cohort studies; systematic reviews; meta-analyses; and randomized controlled trials (RCTs).

Exclusion Criteria
Stroke patients under the age of 18 years were excluded. In order to capture the latest research trends, the literature search to develop the draft guidelines was limited to studies published after 2010, but after the first and second modified Delphi surveys, we did not limit the year of publication in order to reflect the different opinions of experts. In addition, the language was limited to English.

Draft Development Process
A draft guideline was developed by listing the evaluation items to be considered in applying cardiorespiratory physiotherapy in stroke patients. The final selected research papers were independently reviewed by two researchers and one expert as the third figure.

Ethical Considerations of Participants
Although this study has no human contact or human samples, it required the collection of personal information from the experts; therefore, the study was approved by the Institutional Review Board (IRB) of the Nambu University in Gwangju Metropolitan City, Korea (HBU-IRB-1041478-2017-HR-017) and was conducted in accordance with the ethical standards of the Declaration of Helsinki.

Expert Selection
A panel of nine experts consisting of four full-time professors from the Department of Physical Therapy that oversaw academics and research in the field of cardiorespiratory physiotherapy and five clinical experts in charge of the clinical practice of cardiorespiratory physiotherapy was formed. We intended to fully reflect the opinions of experts in the development of clinically applicable guidelines.
The experts agreed to participate in the study and consisted of licensed physiotherapists with at least 10 years of clinical or research experience in cardiorespiratory physiotherapy. Most experts are active instructors in the Korean Academy of Cardiorespiratory Physiotherapy and are the core resources of cardiorespiratory physiotherapy in Korea.

Modified Delphi Data Collection
The aim and method of the study were explained to the experts via phone beforehand. The panel consisted of cardiorespiratory physiotherapy experts that agreed to participate in the study (first trial: 9, second trial: 9). From 15 July, 2019 to 31 August, 2019, drafts using Likert scales and free text responses were sent to experts, and data consultation and content collection were conducted in person or via email.

Confirmation of the Final Evaluation Items
The modified Delphi survey was repeated twice with the panel of experts, and after analyzing the survey results and reaching an agreement based on the statistical process, the final evaluation items were confirmed.

Data Analysis
The statistical analysis was conducted based on the data collected through the modified Delphi survey to the expert panel (first: 9 people, second: 9 people) using the SPSS version 20.0 statistical software (SPSS Inc., Chicago, IL, USA). In addition, descriptive statistical analysis was used to confirm the general characteristics of the expert panel and the verification of item understanding. In the modified Delphi survey, suitability of the items was measured using the Likert scale, with 1 point for "strongly disagree," 2 points for "disagree", 3 points for "neither agree nor disagree", 4 points for "agree", and 5 points for "strongly agree", and the fitness values for each item were displayed.
Based on the displayed fitness values of the Likert scale, content validity and reliability were analyzed, and validity and suitability were calculated to revise and supplement the opinions of the expert panel.

Content Validity and Reliability
To measure the accuracy and consistency of the measurement tool to judge the panel, the Content Validity Ratio (CVR) and Cronbach's α coefficient were analyzed. In this study, based on 18 experts (9 in the first trial and 9 in the second trial), the minimum value of CVR was 0.78 [22].

Validity and Stability
Convergence and consensus were analyzed to assess the validity of the surveys, which served as a criterion for judging the degree of convergence of the responses and the degree of consensus among the panel members. To determine convergence and consensus, the indicator of convergence had to approach 0 and that of consensus had to approach 1, for a given item to have a high level of validity [23,24]. In addition, stability was measured to judge the consensus among the panel of experts [25]; a coefficient of variation of 0.50 or less will lead to no additional survey, and 0.50 to 0.80 indicates that the survey result is relatively stable, whereas 0.80 or more necessitates an additional survey [26].

Data Extraction Criteria
A literature search was carried out using the electronic databases, and the following number of articles were retrieved: PubMed (2789 articles), EMBASE (7399 articles), CINAHL Plus with Full text (1250 articles), and Cochrane library (2060 articles). A total of 13,498 articles were identified, and through a systematic screening of the collected articles following the procedure of the PRISMA statement and using Endnote software (X9.1, Clarivate Analytics. Philadelphia, PA, USA), the final dataset to be analyzed was selected ( Figure 1). We identified 10,876 documents, excluding duplicate documents. Disagreements among reviewers were resolved by consensus; otherwise, a third reviewer was consulted. If not, it was resolved by a third external expert who is a physical therapist and statistical expert. Later, 18 articles were selected for draft development. The selected drafts were revised and analyzed by a panel of experts through the first and second modified Delphi surveys, after which 9 articles were added based on the liberal opinions of each panel member in the first and second surveys, leaving a final set of 27 articles. Tables 5 and 6 present the results of the surveys conducted in this study.  In combination with the CPET, the 6MWT will indicate when deficits preclude walking alone as the primary exercise modality for optimizing cardiovascular fitness 10 Menezes et al. [

General Characteristics of Selected Studies
Characteristics of the studies relating to the evaluation of cardiorespiratory physiotherapy in stroke patients were examined based on three main categories: the publication type, publication year, and study design (Table 7). Concerning the publication type, all 27 articles (100%) were published in academic journals. Concerning the publication year, 5 articles (18.5%) were published before 2010, 11 (40.7%) were published from 2011 to 2015, and 11 (40.7%) were published from 2016 to the present date (June 2019). With regard to the study design, 7 articles (25.9%) were experimental studies, 17 (63.0%) were observational studies, 2 (7.4%) were systematic reviews, and 1 (3.7%) was a systematic review with meta-analysis.

Draft Analysis
Based on the PRISMA statement, the 18 selected articles were analyzed for draft development by two investigators in this study and by a third external expert. Based on the title and abstract, studies related to the evaluation of cardiorespiratory physiotherapy were categorized according to the study design, and after compliance with the structure summary items based on the PICO-SD, duplicated studies were deleted; finally, a total of 31 draft items were considered (Table 8).

First Modified Delphi Study
In the first modified Delphi survey, the mean goodness of fit of the evaluation items was 4.05 ± 0.43 points based on the Likert scale responses. Regarding the content validity ratio, 16 items had a minimum score (less than 0.78) evaluated by the panel of 9 experts and were deleted following the analysis (i.e., mini-mental state examination (MMSE-K), visual analog scale (VAS), and Faces pain rate scale (FPRS), temperature test, light touch and pressure test, proprioception test, Albert test, range of motion (ROM), manual muscle test (MMT), Hand dynamometry, modified Ashworth scale, functional reach test (FRT), Uni-pedal stance test (UPST), Berg balance scale (BBS), dynamic gait index (DGI), modified Barthel index, and functional independence measure (FIM)). Internal consistency reliability for the evaluation items was set at α = 0.895. The mean convergence was 0.57, while the mean consensus and mean stability were 0.69 and 0.21, respectively ( Table 9).
The results of the first modified Delphi survey showed that the accuracy of the statements for each item and the item validity exceeded the minimum scores; nevertheless, appropriate revision and modification were carried out to accommodate additional opinions from the panel and the items of low content validity (Table 10). Six items were newly formulated by the expert panel, including the body mass index (BMI) [45], spatiotemporal gait parameters [51], thoracic contour and chest expansion test [49], the trunk impairment scale (TIS) [52], cardiopulmonary exercise testing (CPET) [50], and the stroke impact scale (SIS) [48].

Open Opinion Content Feedback Content
To address the need for clearly dividing the categories of the main theme according to individual items.
-The evaluation items were divided into five categories: basic physical examination, level of consciousness, balance, cardiorespiratory, and activities of daily living.
To extract the evaluation items through a multidisciplinary approach.
-Cooperation among health professionals of different fields was challenging.
-A considerable length of time and high cost were anticipated. -Thus, a guideline has been developed in this study based on the findings of previous studies, for cardiorespiratory evaluation and intervention items that are required for stroke patients.
To extract the evaluation items according to the advancement of each phase of stroke.
-The evaluation and intervention items showed differences according to the health status of individual patients, which are problematic. -A practical method in the clinical setting seemed to be selective application based on the guideline developed in this study, according to each patient's health status.
To include the description of factors to consider for the discontinuation of exercise by adding an evaluation parameter related to chest pain and leg pain.
-For stroke, the rationale for discontinuation of exercise did not derive from a specific pain as in patients with angina or peripheral vascular disease. -In general, discontinuation of exercise depended on the result of the basic indices such as SPO 2 , HR, and BP, or patient's complaints on the symptoms of headache, vertigo, nausea, and vomit.
To include the description on additional items for the CPET.
-In this study, due to a possibility of falls in stroke patients, GXT description by the cycle had been included based on previous studies; nevertheless, based on a previous study reporting the use of CPET in stroke patients, the item was added.
To add BMI in the basic physical examination category, as it is always included as a test item in cardiorespiratory tests and interventions, and as it is one of the important factors.
-The BMI was added to the category of basic physical examination in evaluation items.
To add spatiotemporal gait parameters used as an indicator of balance in the category of exercise for stroke patients.
-The spatiotemporal gait parameters were added to the category of balance in evaluation items.
To address the need for FVC test, as the stroke-related breathing difficulty is likely to be caused by restrictive lung disease due to reduced lung compliance. To add the Thoracic contour and Chest expansion test, for an asymmetric thoracic cage may lead to local lung collapse. In addition, only after chest expansion test, the lobe and area of the lung with restricted expansion can be detected, for which a suitable chest expansion method or chest mobilization, or segmental breathing for local lung inflation, can be determined.
-FVC test was added as a pulmonary function test, as well as the items; Thoracic contour and Chest expansion test.
-The items; chest mobilization and segmental breathing, were added.
The items; MIP and MEP, should be moved to the cardiorespiratory category as they test the respiratory capacity, and the static lung capacity tests (TV, IRV, ERV, VC, IC) should be added.
-For the pulmonary function test, the items, SVC and maximal-effort expiratory capacity, were revised and subsequently separated. -The items; MIP and MEP, were moved to the cardiorespiratory category.
Cervical and thoracic mobilization.
-Cervical and thoracic mobilization is anticipated to be more helpful during the intervention, rather than limiting to chest mobilization.

Second Modified Delphi Study
The results of the first modified Delphi survey showed stability of 0.21 points and no additional questionnaire was required [26]. On the other hand, a degree of convergence close to 0 and a degree of consensus closer to 1 indicates that the opinions and consensus among the panels are valid [23,24]. According to the first modified Delphi survey, the mean convergence was 0.57 points and the mean consensus was 0.69 points, which were not properly reached and agreed among the panels. In addition, we needed to respond to open questions from expert panels; therefore, we conducted a second modified Delphi survey.
We conducted the second modified Delphi survey with the same panel of 9 experts, and the collection rate was 100%. The mean goodness of fit of the evaluation items was 4.40 ± 0.35 points based on the Likert scale responses. For the content validity ratio, two items with scores less than 0.78 (the minimum score of a panel of 9 experts) were deleted (Glasgow coma scale (GCS) and TIS). Concerning the validity, all other items exceeded the minimum score. The internal consistency reliability was α = 0.815. Likewise, the mean convergence improved from 0.57 to 0.36, the mean consensus improved from 0.69 to 0.84, and the mean stability improved from 0.21 to 0.17 (Table 11).
The accuracy of the statements for each item and the item validity all exceeded the minimum score; nevertheless, appropriate revision and modification were carried out to account for additional opinions from the panel and the items of low content validity (Table 12). The respiratory rate or minute ventilation [44] and the 10-Meter Walking Test (10MWT) [47] were added by expert panelists.

Deriving Final Evaluation Items
To evaluate cardiorespiratory physiotherapy in stroke patients, 31 evaluation items were derived from the literature review and were reorganized through the modified Delphi surveys by expert panelists. Twenty-two items were derived from the first and second survey to achieve consensus, the final 20 items were selected and divided among five main categories (Table 13).

Open Opinion Content Feedback Content
To combine the items; CPET and cycle GXT -The CPET includes the following methods: treadmill, leg cycle, and arm cycle, and among these, the lower leg cycle CPET shows high levels of MHR and RER, which is likely to be useful in evaluating patients with a risk of falls.
-The treadmill-based CPET is a protocol of load-bearing exercise using the degree of slope, so that it is likely to prove more useful in achieving increased heart rate and anaerobic threshold in comparison to 6MWT or ISWT using speed.
To suggest a method of application regarding ISWT and 6MWT -The use of ISWT and 6MWT is determined according to the patient's health status, so that in the clinical setting, it is likely to be more efficient to apply based on the individual patient.
To consider inclusion of SIS items -Among the 9 categories, a correlation of the mobility, ADL/IADL, participation, and physical categories, for items related to movement, and above all, unlike the frequently used FIM and Barthel index, SIS items were likely to prove useful in determining the level of recovery for several symptoms of stroke in clinical setting.
To add the item; Respiratory rate or Minute ventilation, in the category of basic physical examination.
-The items were added because the category of basic physical examination required a respiratory parameter such as respiratory rate or minute ventilation.
To add the item; 10MWT -10MWT and the 6MWT were added, as a previous study reported a significant correlation between the two tests with respect to the physical activity and gait in stroke patients.
To apply Thoracic contour and Chest expansion test -The two items are rarely applied in clinical setting; thus, further investigations are required.
To add an item related to quality of life with respect to the health of the heart -No study was found for the suggested parameter that can be applied to stroke patients.

Discussion
Appropriate clinical decision-making is a prerequisite to achieve adequate levels of evaluation as well as diagnosis, prognosis, and intervention based on the health status of individual patients [53]. However, clinical guidelines to evaluate cardiorespiratory physiotherapy in stroke patients are not yet available. The purpose of this study was to ensure enhanced quality and expertise for cardiorespiratory physiotherapy in stroke patients by providing selected evaluation items with verified evidence and promoting their use in the clinical setting.
As a result, we collected and analyzed 13,498 research papers, 18 articles were selected for the draft development, and after adding 9 articles via the modified Delphi surveys with expert panelists, a total of 27 articles were ultimately selected to determine the evaluation items. In the first modified Delphi survey, the expert panelists were requested to select from a wide range of evaluation items, including those extracted from the literature review on the comprehensive evaluation of stroke patients and those directly related to the cardiorespiratory domain. After statistically assessing the results of the first modified Delphi survey, individual items were assessed and those with a content validity less than 0.78 were deleted, while the remaining items were revised and modified based on stability, convergence, and consensus results and the open opinions provided by the experts to establish the second modified Delphi survey.
After the first modified Delphi survey, 16 items with less than the minimum score of 0.78 for content validity ratio, that is, the MMSE-K, VAS, FPRS, Temperature test, Light touch and pressure test, Proprioception test, Albert test, ROM, MMT, Hand dynamometry, modified Ashworth scale, FRT, UPST, BBS, DGI, modified Barthel index, and FIM, were deleted. The deleted items were found to be those required for primary motor function recovery without direct relation to cardiorespiratory fitness. In addition, the item internal consistency reliability of the first evaluation items was shown to be considerably high at α = 0.895. In determining convergence and consensus, the indicator of convergence should approach 0 and that of consensus should approach 1, for a given item to have a high level of validity [23,24]. The first modified Delphi survey result showed insufficient levels of convergence and consensus among the expert panelists with the mean convergence for the evaluation items at 0.57 and the mean consensus at 0.69. Regarding the stability, which indicates congruency among the experts in their responses, a coefficient of variation of 0.50 or less leads to no additional survey, 0.50 to 0.80 indicates that the survey result is relatively stable, whereas 0.80 or more necessitates an additional survey [26]. The mean stability was 0.21 in this study; therefore, no additional survey was required. However, due to insufficient convergence and consensus among the expert panelists and the need for expert responses for open questions, the second modified Delphi survey was conducted.
In this study, convergence of the open opinions of the expert panels was achieved using a method in which the investigator provided individual feedback to the issues raised as open opinions after the first modified Delphi survey. The open opinions provided by the expert panels included the need to clearly divide the categories of the main theme according to individual items, the need to extract the items through a multidisciplinary approach, and the need to categorize them according to the advancement of each phase of the stroke. The main feedback from the investigator was as follows: first, the evaluation items were divided into five categories: basic physical examination, level of consciousness, balance, cardiorespiratory, and activities of daily living; second, a guideline based on a multidisciplinary approach seemed an ideal method [54]; however, obtaining cooperation among health professionals in different fields was challenging, and the study anticipated that a considerable length of time and high cost were required. Thus, this study developed a guideline based on the findings of previous studies for cardiorespiratory evaluation items that were required for stroke patients; and third, based on the suggestion that the evaluation items should be provided according to the stroke phase (acute, subacute, and chronic), a practical method based on the health status of individual patients in the clinical setting seemed to be the ideal application for the guideline. The items added by the expert panelists included six evaluation items, namely, BMI [45], spatiotemporal gait parameters [51], the thoracic contour and chest expansion test [49], TIS [52], CPET [50], and SIS [48]. For the item of CPET in particular, the description in the draft limited the item to cycle graded exercise test (GXT), but based on a previous study reporting the use of CPET in treadmill and cycle methods, the item was added [16].
After the second modified Delphi survey, items GCS and TIS had less than the minimum score of 0.78 for content validity ratio and were deleted together with the cycle GXT that coincided with the CPET. In the basic physical examination category, respiratory rate or minute ventilation was added, while peak flow and ECG were removed from the basic physical examination category to the cardiorespiratory category. In the first modified Delphi survey, the pulmonary function test was evaluated as two separate items because of a large number of subcategories, but in the second survey, it was treated as a single unified item. Note that the pulmonary function test has been reported to be highly important as it provides basic data for precise exercise diagnosis by measuring accurate pulmonary function and assessing functional capacity, and the diagnosis and prognosis of the condition [55]. Comparing the convergence, consensus, and stability of the second modified Delphi survey results with those of the first modified Delphi survey, the mean convergence was shown to have improved from 0.57 to 0.36, the mean consensus from 0.69 to 0.84, and the mean stability from 0.21 to 0.17, which indicates that the convergence and consensus among the expert panelists increased further in the second modified Delphi survey with a higher level of congruency, suggesting that the experts reached a consensus. The added items were the respiratory rate or minute ventilation [44] and ten-meter walking test (10MWT) [47], implying that a consensus was drawn from the open opinions provided by the expert panelists.
In the second modified Delphi survey, the feedback to the liberal opinions of the expert panelists was as follows: First, the CPET methods include treadmill, leg cycle, and arm cycle exercises, and among these, the lower leg cycle CPET is considered the most efficient based on high levels of maximal heart rate (MHR) and respiratory gas exchange ratio (RER) as a risk of accidental falls. In addition, the treadmill-based CPET is a protocol of load-bearing exercises using the degree of slope, such that it is suggested to be more useful in achieving increased heart rate and anaerobic threshold compared to the 6MWT or incremental shuttle walk test (ISWT) speed [31]. Second, as the application of the ISWT and 6MWT is determined based on the health status of the patient, these items should be applied according to the patient's health status in the clinical setting [46]. Third, for a treadmill walk exercise at 3 km/h, the energy expenditure is 2.43 metabolic equivalents (METs), and when the degree of slope is increased by 10% at the same speed to achieve higher exercise intensity, the energy expenditure is 5 METs. However, on a flat slope, even if the treadmill speed is increased two-fold to 6 km/h, the energy expenditure is only 3.85 METs, suggesting that it is more beneficial to stroke patients if the exercise intensity is increased by increasing the degree of the slope than by increasing the speed [56]. Fourth, it was determined that the Stroke Impact Scale (SIS) was a valid item from the Likert validity score of 4.11 ± 0.60. While the SIS seems to lack significant clinical evidence for its application, as no previous study has reported a definite association between stroke and cardiorespiratory parameters, it is possible that the items related to the movement correlate with mobility, activities of daily living (ADL)/instrumental activities of daily living (IADL), participation, and physical categories among the 9 categories, and above all, unlike the frequently used FIM and Barthel index, the SIS may be useful in determining the level of recovery for several symptoms of stroke [48]. Fifth, in the basic physical examination category, it was suggested that an additional respiratory test was required, such as the respiratory rate or minute ventilation [44]. Sixth, the 10MWT and 6MWT were suggested as additional items, as a previous study reported a significant correlation between the two tests with respect to physical activity and gait in stroke patients [57].
The present study has several limitations. First, when a given item is applied, the effect may vary according to the health status of individual stroke patients. This variation may be resolved through selective application of the proposed evaluation tool in a clinical setting. Second, the study did not take into consideration the impact of differences in clinical experience and academic background among the expert panelists. Third, the Physiotherapy Evidence Database (PEDro), a separate physical therapy database, was not added because the Cochrane Library database contains a wide range of papers and guidelines for randomized and systematic studies and clinical trials. It is highly advisable that this factor be analyzed and implemented in future studies.
In conclusion, the evaluation items suggested in this study were derived from converging evidence from expert opinions and should serve as an efficient tool in the clinical setting for the evaluation of cardiorespiratory physiotherapy necessity in stroke patients.

Conclusions
This study that actively reflects the opinions of expert panels with a modified Delphi survey that utilizes the advantages of non-face-to-face review and prior literature review to present items that can be applied to evaluate cardiorespiratory physiotherapy for stroke patients could be used directly in clinical practice. It is, therefore, important to develop an evidence-based guideline. The final 20 evaluation items can be widely used as a checklist for evaluating and diagnosing patients in clinical practice by selectively providing customized evaluation tools for cardiorespiratory physiotherapy with respect to the level of recovery of motor function in individual stroke patients. In addition, it is expected that it will have a positive impact on programs to plan therapeutic interventions by providing basic data for individual patient evaluations. Future experimental studies investigating the role of cardiorespiratory physiotherapy in stroke patients, in addition to systematic meta-analysis studies to allow a qualitative evaluation to secure its conclusions, are warranted to validate these guidelines.

Conflicts of Interest:
The authors declare no conflict of interest.