Browse the corpus

Development of the Pediatric Scale for Quality of Recovery (PedSQoR). BACKGROUND: Measuring the quality of a patient's recovery is vital, and reliable patient-centered outcome metrics are needed for clinical investigations and quality improvement. Currently, assessment tools to measure quality of recovery in pediatric patients are lacking. This study aimed to develop a scale to assess the quality of recovery construct in pediatric patients. METHODS: Using a mixed-methods investigative model, item generation was achieved using two complementary approaches. First, a comprehensive review of the literature identified tools and questions that assessed the endpoints relevant to recovery in children. Questions were categorized and then assessed by an expert Delphi panel who determined the most significant domains and items to be included. Concurrently, semistructured interviews were conducted with patients and their families to identify themes related to recovery that were important to patients and families. The resulting pilot questionnaire was administered to patients and their families presenting for elective surgery in the United States and Australia. RESULTS: The literature search identified 41 instruments, comprising 216 questions relevant to recovery. After the initial Delphi round, the item list was reduced to 91 questions, and then to 50 questions after the second round. The themes identified in the semistructured interviews aligned with domains considered important by a panel of experts. A 50-item questionnaire was administered to 1,162 children at multiple timepoints after surgery. Item reduction and factor analysis resulted in the 20-item Pediatric Scale for Quality of Recovery that assesses the domains relevant to physical and psychologic recovery. CONCLUSIONS: The Pediatric Scale for Quality of Recovery scale is a 20-item questionnaire designed to provide a holistic representation of a child's physical, emotional, and psychologic recovery after surgery and anesthesia. It was developed and validated with consumer involvement and a strong patient-centered focus. Once further validation has been established, it is expected to become a standardized endpoint in pediatric perioperative trials and quality improvement projects.

Measuring the quality of a patient’s recovery is vital, and reliable patient-centered outcome metrics are needed for clinical investigations and quality improvement. There are no widely accepted assessment tools to measure quality of recovery in children. This article describes the Pediatric Scale for Quality of Recovery (PedSQoR), which was developed and validated with consumer involvement and a strong patient-centered focus. It is a 20-item questionnaire designed to provide a holistic representation of a child’s physical, emotional, and psychologic recovery after surgery and anesthesia. In perioperative medicine, the capture of patient-reported outcomes is vital for assessing the quality of a patient’s perioperative experience and for measuring the impact of clinical interventions and quality improvement efforts. Although measures of broad outcomes such as morbidity, hospital readmission, and mortality are undoubtedly important, they are insensitive to many dimensions relevant to patients and their caregivers. Patients’ perceived quality of and satisfaction with perioperative care is multifactorial and includes patient experience, expectations, and outcomes relevant to their recovery. Quality of recovery (QoR) is a multidimensional, health-related quality of life construct encompassing dimensions relevant to recovery after surgery and anesthesia including pain experience, physical function and comfort, psychologic well-being, and cognitive and psychosocial impact.1,2

omes relevant to their recovery. Quality of recovery (QoR) is a multidimensional, health-related quality of life construct encompassing dimensions relevant to recovery after surgery and anesthesia including pain experience, physical function and comfort, psychologic well-being, and cognitive and psychosocial impact.1,2 Several QoR scales have been developed to assess outcomes in adult perioperative medicine. The QoR-40 and its shorter version, QoR-15, have been extensively validated and are now included in the list of recommended standardized outcomes for perioperative clinical trials.3 while there have been efforts to measure specific postoperative events, there is currently no equivalent overall QoR measurement tool for pediatric patients.4,5 The design of a QoR tool for a pediatric population presents a number of challenges not present in adult perioperative populations, including different developmental stages and explicit recalls,6 the need for proxy reporting of symptoms, and interpretation through the lens of the whole family unit. Further complexity is introduced when considering the applicability of a single recovery tool for a variety of uses including research, quality improvement, or quality assurance. In addition, the tool must have adequate temporal responsiveness, be sensitive to cultural diversity, and interface with the wide spectrum of periprocedural perturbations experienced by pediatric patients.

ity of a single recovery tool for a variety of uses including research, quality improvement, or quality assurance. In addition, the tool must have adequate temporal responsiveness, be sensitive to cultural diversity, and interface with the wide spectrum of periprocedural perturbations experienced by pediatric patients. The construction and validation of a measurement scale requires a generative process described by psychometric theory.7–10 This process can be summarized into three phases: item development, scale development, and scale evaluation.10 In this article, we describe the development of a scale for assessing the QoR in pediatric patients (Pediatric Scale for QoR, PedSQoR) using psychometric theory. We provide a thorough description of our methodology and results as well as a discussion of the projected impact of this scale for future clinical investigations and quality improvement work.

Measuring the quality of a patient’s recovery is vital, and reliable patient-centered outcome metrics are needed for clinical investigations and quality improvement. There are no widely accepted assessment tools to measure quality of recovery in children.

This article describes the Pediatric Scale for Quality of Recovery (PedSQoR), which was developed and validated with consumer involvement and a strong patient-centered focus. It is a 20-item questionnaire designed to provide a holistic representation of a child’s physical, emotional, and psychologic recovery after surgery and anesthesia.

A mixed-methods approach was employed, involving a working group of pediatric anesthesiologists and psychologists from Australia and the United States. The PedSQoR project was approved by the institutional research ethics committees at low or negligible risk (Children’s Hospital of Philadelphia IRB 20-018401, Boston Children’s Hospital IRB-P00039211, and Queensland Children’s Hospital HREC/2021/QCHQ/71747). Conceptualization was undertaken by the working group. Previous experience in adult literature was considered along with the specific requirements for a pediatric measurement tool. Our goal was to isolate the construct of recovery from procedure-related perturbations with a tool that could be used at the individual and population levels throughout the pediatric age range for all procedures, and would be valid and responsive throughout the time course of recovery. It was recognized that, despite being an important age group, children less than 2 yr of age would require a substantially different question set to assess the recovery process, and as such, this age group was excluded from consideration a priori.

nd would be valid and responsive throughout the time course of recovery. It was recognized that, despite being an important age group, children less than 2 yr of age would require a substantially different question set to assess the recovery process, and as such, this age group was excluded from consideration a priori. QoR was recognized as a complex latent variable that was both multidimensional and hierarchical, with the potential to correlate and be impacted by other constructs that are not immediately evident. Health-related quality of life measures have an accepted minimum core set of domains that comprise the overall construct—physical health, mental health, social functioning, role functioning, and general health perceptions11—with adult QoR items often measuring each of these domains.12,13 It was hypothesized that evaluation of perturbations would require items across each of these domains, and a preliminary conceptual framework was developed. An inductive and deductive approach was used to assemble items across these domains. We undertook a sequential exploratory mixed-methods study that conformed to the COnsensus-based Standard for the selection of health status Measurement INstrument (COSMIN) study design evaluation methodology for patient-reported outcome measurement instruments.14

pproach was used to assemble items across these domains. We undertook a sequential exploratory mixed-methods study that conformed to the COnsensus-based Standard for the selection of health status Measurement INstrument (COSMIN) study design evaluation methodology for patient-reported outcome measurement instruments.14 A subgroup of the authors (W.T.M. and P.A.S.) conducted a qualitative study involving 95 semistructured interviews with patients, parents, and healthcare providers.15 This study used a “ground-up” or emic approach to identify themes and domains for the PedSQoR tool. This approach enables the exploration of important recovery factors from the perspectives of stakeholders (children, parents, and healthcare providers). Free-listing exercises were conducted with 60 patient/parent dyads and 24 healthcare providers (recovery room nurses, anesthesiologists, and surgeons). These exercises were designed to elicit words or phrases associated with recovery from surgery and anesthesia. The interviews were analyzed, and common themes were identified. These themes were then used to develop interview templates for subsequent semistructured interviews. Thirty patient/parent dyads underwent semistructured interviews, and their responses were analyzed. Full details of these results have been reported previously.15

rviews were analyzed, and common themes were identified. These themes were then used to develop interview templates for subsequent semistructured interviews. Thirty patient/parent dyads underwent semistructured interviews, and their responses were analyzed. Full details of these results have been reported previously.15 A review of relevant tools was conducted using PubMed (National Library of Medicine, National Institutes of Health, Bethesda, Maryland), gray literature search, and tangential electronic exploration of related articles (i.e., “snowballing”) to identify measurement instruments that were broadly related to perioperative care or health-related quality of life in children. This search was intentionally extended outside the perioperative sphere and included pediatric mental health tools. Two researchers (C.G. and P.L.-A.) independently rated each item based on criterion validity, clarity, specificity to one dimension, lack of ambiguity, and responsiveness. Duplicates and poorly performing items were excluded if there was concordance between the researchers or if the item was refined. To maintain broad content validity, the resulting items were categorized into domains, subdomains, and dimensions related to recovery based on the International Classification of Functioning, Disability, and Health,16 previously validated tools,17–21 and inductive research.

n the researchers or if the item was refined. To maintain broad content validity, the resulting items were categorized into domains, subdomains, and dimensions related to recovery based on the International Classification of Functioning, Disability, and Health,16 previously validated tools,17–21 and inductive research. A broad selection of people with expertise in pediatric perioperative care was assembled to determine the important aspects of recovery. The expert group consisted of eight pediatric ward nurses; five recovery nurses; three pain nurses; two day ward nurses; one intensive care nurse; one complex care nurse; one rehabilitation nurse; seven anesthesiologists; four anesthetic assistants; one general pediatric surgeon; one ear, nose, and throat surgeon; two surgical fellows; one general practitioner; one social worker; one physiotherapist; one occupational therapist; two child psychologists; and a parent and child with lived experiences of recovery after surgery and anesthesia. Members of this group were asked to suggest the five most important endpoints relevant to patient recovery with an allowance for free-text responses. These responses were then thematically analyzed and collated.

child psychologists; and a parent and child with lived experiences of recovery after surgery and anesthesia. Members of this group were asked to suggest the five most important endpoints relevant to patient recovery with an allowance for free-text responses. These responses were then thematically analyzed and collated. We performed a basic Delphi substudy in which respondents were asked to rate each of the dimensions related to recovery on a nine-point scale based on their perceived importance. Rankings of 7 and above were considered as “important,” and the item-level content validity index was calculated for each dimension. Dimensions with a low content validity index were considered for removal until the scale-level (average method) content validity index exceeded the threshold. Dimensions that were marginal or for which there was no clear consensus among respondents were retained for the subsequent round. The respondents were then asked to rank the items in each area. They were asked to rank based on validity (does it measure the area of interest), reliability (is the question reproducible in other settings and populations), feasibility (is the question easy to understand and not ambiguous) and patient centeredness (does the question assess an endpoint that has a meaningful impact on all pediatric patients’ recovery). The lowest-ranked questions were eliminated while ensuring that all dimensions retained items, and redundant items were dropped or incorporated into one question, in which two or more items were similar.

enteredness (does the question assess an endpoint that has a meaningful impact on all pediatric patients’ recovery). The lowest-ranked questions were eliminated while ensuring that all dimensions retained items, and redundant items were dropped or incorporated into one question, in which two or more items were similar. The results of inductive and deductive approaches were combined to produce a potential item set. Cognitive interviews were conducted with subject matter experts and medically literate consumers, and iterative improvements were made to the wording and directionality of questions. Phrasing and literacy level were adapted to a Year 4 Flesch reading level and were made consistent with the stem “In the last 24 h” and answering to a 5-point proportion of time Likert scale (“none of the time,” “rarely,” “some of the time,” “often,” “always,” and an option for “not applicable or uncertain”). The potential item set plus three global questions and a 100-mm visual analog scale of overall recovery were sent via a digital link to patients and caregivers on the day of surgery for day-stay patients or postoperative day (POD) 1. A subset of patients who consented to multiple survey collections were asked to respond on PODs 1 to 2, then weeks 2 to 5 (11 potential responses).

mm visual analog scale of overall recovery were sent via a digital link to patients and caregivers on the day of surgery for day-stay patients or postoperative day (POD) 1. A subset of patients who consented to multiple survey collections were asked to respond on PODs 1 to 2, then weeks 2 to 5 (11 potential responses). Study participants were recruited from two sites in the United States (Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania; and Boston Children’s Hospital, Boston, Massachusetts) and one Australian site (Queensland Children’s Hospital, Brisbane, Australia). To ensure a representative sample of the target population, day surgery and inpatient populations were evenly recruited across age-bracketed groups chosen based on their similar developmental ages. The four age brackets were 2 to 4 yr, 5 to 7 yr, 8 to 12 yr, and 13 to 17 yr. Younger participants in the first two age brackets had responses provided by a caregiver, whereas older participants could complete the questionnaires themselves or concurrently with a caregiver. Participants electing to provide a single timepoint response were recruited in a 3:1 ratio, with those willing to provide multiple responses over the course of their recovery. Pediatric patients aged 2 to 17 yr old presenting from home for day surgery or surgery requiring postoperative admission American Society of Anesthesiologists (Schaumburg, Illinois) Physical Status III or lower

Study participants were recruited from two sites in the United States (Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania; and Boston Children’s Hospital, Boston, Massachusetts) and one Australian site (Queensland Children’s Hospital, Brisbane, Australia). To ensure a representative sample of the target population, day surgery and inpatient populations were evenly recruited across age-bracketed groups chosen based on their similar developmental ages. The four age brackets were 2 to 4 yr, 5 to 7 yr, 8 to 12 yr, and 13 to 17 yr. Younger participants in the first two age brackets had responses provided by a caregiver, whereas older participants could complete the questionnaires themselves or concurrently with a caregiver. Participants electing to provide a single timepoint response were recruited in a 3:1 ratio, with those willing to provide multiple responses over the course of their recovery. Pediatric patients aged 2 to 17 yr old presenting from home for day surgery or surgery requiring postoperative admission American Society of Anesthesiologists (Schaumburg, Illinois) Physical Status III or lower Parent or guardian is the primary caregiver to the patient; in the event of two parents or caregivers, one parent will complete the survey(s) for the child (2 to 7 yr) or concurrently with the child (8 to 17 yr) Parental or guardian permission (informed consent) and child assent Non-English speakers of patient or parents or guardians Urgent or emergent surgery Developmental delay or severe systemic disease that impacts activities of daily living

Parent or guardian is the primary caregiver to the patient; in the event of two parents or caregivers, one parent will complete the survey(s) for the child (2 to 7 yr) or concurrently with the child (8 to 17 yr) Parental or guardian permission (informed consent) and child assent Non-English speakers of patient or parents or guardians Urgent or emergent surgery Developmental delay or severe systemic disease that impacts activities of daily living Parents or guardians or patients who, in the opinion of the investigator, may be not available to complete the study Data analysis was performed using IBM (USA) SPSS Statistics for Windows version 28. Descriptive statistics were used to identify low response rates and the floor and ceiling effects. Missing data were addressed using the expectation-maximization imputation as recommended by Schlomer et al.22 and Graham et al.23 This process was applied to the individual questionnaire items separately to ensure a complete dataset for subsequent analyses. Analysis was performed on the self-reported (“child,” 8 to 17 yr) and proxy-reported (“carer,” 2 to 7 yr) responses separately, and the coreported data were retained for validity testing.

Data analysis was performed using IBM (USA) SPSS Statistics for Windows version 28. Descriptive statistics were used to identify low response rates and the floor and ceiling effects. Missing data were addressed using the expectation-maximization imputation as recommended by Schlomer et al.22 and Graham et al.23 This process was applied to the individual questionnaire items separately to ensure a complete dataset for subsequent analyses. Analysis was performed on the self-reported (“child,” 8 to 17 yr) and proxy-reported (“carer,” 2 to 7 yr) responses separately, and the coreported data were retained for validity testing. Factor analysis was used to identify the underlying structure of questionnaire items. The analysis began by evaluating the suitability of the data for factor analysis by checking the correlation matrix for significant correlations among variables and assessing the Kaiser–Meyer–Olkin measure of sampling adequacy and Bartlett test of sphericity. A Kaiser–Meyer–Olkin value above 0.70 and a significant Bartlett test indicated that the data were appropriate for factor analysis.

analysis by checking the correlation matrix for significant correlations among variables and assessing the Kaiser–Meyer–Olkin measure of sampling adequacy and Bartlett test of sphericity. A Kaiser–Meyer–Olkin value above 0.70 and a significant Bartlett test indicated that the data were appropriate for factor analysis. The aim of factor analysis is to reveal any latent variables that cause manifest variables to covary. An exploratory factor analysis was performed using principal axis factoring as the extraction method.24 These exploratory procedures statistically analyze the interrelationships between the instrument items and domains to uncover the unknown underlying factorial structure of the construct of interest.25 Factors were extracted based on eigenvalues greater than 1 and by examining the scree plot to identify a clear elbow point.26 To improve interpretability, varimax rotation, an orthogonal rotation method, was applied to maximize the variance of the factor loadings. The exploratory factor analysis results were assessed by reviewing the factor loadings, which reflected the correlation between each item and the extracted factors. Items with loadings less than 0.40 were suppressed and were considered for removal.27 The internal consistency of the identified factors was evaluated using Cronbach alpha, with values above 0.70 deemed acceptable for reliability. Item reduction was iterated through exploration of forced factor loading to three, four, or five factors, and self- or proxy-reported performance. Correlation analysis and performance of the items over time were evaluated to further guide item selection.

alpha, with values above 0.70 deemed acceptable for reliability. Item reduction was iterated through exploration of forced factor loading to three, four, or five factors, and self- or proxy-reported performance. Correlation analysis and performance of the items over time were evaluated to further guide item selection. Confirmatory factor analysis was performed by choosing an optimal number of items to balance accuracy and acceptability. A reliability analysis of the final questionnaire based on the confirmatory factor analysis was conducted. In addition to the overall Cronbach alpha, the “scale if item deleted” statistics were generated and reviewed to determine whether any items negatively affected the reliability of the scale. Items with low item-total correlations were considered for removal if they contributed to a significant increase in Cronbach alpha when omitted. The results provided a measure of reliability for the entire scale and helped ensure that the items consistently reflected the intended construct. The construct validity of the tool was assessed by testing convergence (correlations) between the 20-item score and the overall global QoR items and the 100-mm visual analog scale of overall recovery. Finally, how well the model was able to reproduce the data was measured using common indices of model fit. These include the root mean square error of approximation (RMSEA), the Common Fit Index, the Tucker–Lewis Index (TLI), and the standardized root mean squared residual.

A mixed-methods approach was employed, involving a working group of pediatric anesthesiologists and psychologists from Australia and the United States. The PedSQoR project was approved by the institutional research ethics committees at low or negligible risk (Children’s Hospital of Philadelphia IRB 20-018401, Boston Children’s Hospital IRB-P00039211, and Queensland Children’s Hospital HREC/2021/QCHQ/71747).

Conceptualization was undertaken by the working group. Previous experience in adult literature was considered along with the specific requirements for a pediatric measurement tool. Our goal was to isolate the construct of recovery from procedure-related perturbations with a tool that could be used at the individual and population levels throughout the pediatric age range for all procedures, and would be valid and responsive throughout the time course of recovery. It was recognized that, despite being an important age group, children less than 2 yr of age would require a substantially different question set to assess the recovery process, and as such, this age group was excluded from consideration a priori. QoR was recognized as a complex latent variable that was both multidimensional and hierarchical, with the potential to correlate and be impacted by other constructs that are not immediately evident. Health-related quality of life measures have an accepted minimum core set of domains that comprise the overall construct—physical health, mental health, social functioning, role functioning, and general health perceptions11—with adult QoR items often measuring each of these domains.12,13 It was hypothesized that evaluation of perturbations would require items across each of these domains, and a preliminary conceptual framework was developed.

ical health, mental health, social functioning, role functioning, and general health perceptions11—with adult QoR items often measuring each of these domains.12,13 It was hypothesized that evaluation of perturbations would require items across each of these domains, and a preliminary conceptual framework was developed. An inductive and deductive approach was used to assemble items across these domains. We undertook a sequential exploratory mixed-methods study that conformed to the COnsensus-based Standard for the selection of health status Measurement INstrument (COSMIN) study design evaluation methodology for patient-reported outcome measurement instruments.14

A subgroup of the authors (W.T.M. and P.A.S.) conducted a qualitative study involving 95 semistructured interviews with patients, parents, and healthcare providers.15 This study used a “ground-up” or emic approach to identify themes and domains for the PedSQoR tool. This approach enables the exploration of important recovery factors from the perspectives of stakeholders (children, parents, and healthcare providers). Free-listing exercises were conducted with 60 patient/parent dyads and 24 healthcare providers (recovery room nurses, anesthesiologists, and surgeons). These exercises were designed to elicit words or phrases associated with recovery from surgery and anesthesia. The interviews were analyzed, and common themes were identified. These themes were then used to develop interview templates for subsequent semistructured interviews. Thirty patient/parent dyads underwent semistructured interviews, and their responses were analyzed. Full details of these results have been reported previously.15

A review of relevant tools was conducted using PubMed (National Library of Medicine, National Institutes of Health, Bethesda, Maryland), gray literature search, and tangential electronic exploration of related articles (i.e., “snowballing”) to identify measurement instruments that were broadly related to perioperative care or health-related quality of life in children. This search was intentionally extended outside the perioperative sphere and included pediatric mental health tools. Two researchers (C.G. and P.L.-A.) independently rated each item based on criterion validity, clarity, specificity to one dimension, lack of ambiguity, and responsiveness. Duplicates and poorly performing items were excluded if there was concordance between the researchers or if the item was refined. To maintain broad content validity, the resulting items were categorized into domains, subdomains, and dimensions related to recovery based on the International Classification of Functioning, Disability, and Health,16 previously validated tools,17–21 and inductive research.

A broad selection of people with expertise in pediatric perioperative care was assembled to determine the important aspects of recovery. The expert group consisted of eight pediatric ward nurses; five recovery nurses; three pain nurses; two day ward nurses; one intensive care nurse; one complex care nurse; one rehabilitation nurse; seven anesthesiologists; four anesthetic assistants; one general pediatric surgeon; one ear, nose, and throat surgeon; two surgical fellows; one general practitioner; one social worker; one physiotherapist; one occupational therapist; two child psychologists; and a parent and child with lived experiences of recovery after surgery and anesthesia. Members of this group were asked to suggest the five most important endpoints relevant to patient recovery with an allowance for free-text responses. These responses were then thematically analyzed and collated.

We performed a basic Delphi substudy in which respondents were asked to rate each of the dimensions related to recovery on a nine-point scale based on their perceived importance. Rankings of 7 and above were considered as “important,” and the item-level content validity index was calculated for each dimension. Dimensions with a low content validity index were considered for removal until the scale-level (average method) content validity index exceeded the threshold. Dimensions that were marginal or for which there was no clear consensus among respondents were retained for the subsequent round. The respondents were then asked to rank the items in each area. They were asked to rank based on validity (does it measure the area of interest), reliability (is the question reproducible in other settings and populations), feasibility (is the question easy to understand and not ambiguous) and patient centeredness (does the question assess an endpoint that has a meaningful impact on all pediatric patients’ recovery). The lowest-ranked questions were eliminated while ensuring that all dimensions retained items, and redundant items were dropped or incorporated into one question, in which two or more items were similar.

The results of inductive and deductive approaches were combined to produce a potential item set. Cognitive interviews were conducted with subject matter experts and medically literate consumers, and iterative improvements were made to the wording and directionality of questions. Phrasing and literacy level were adapted to a Year 4 Flesch reading level and were made consistent with the stem “In the last 24 h” and answering to a 5-point proportion of time Likert scale (“none of the time,” “rarely,” “some of the time,” “often,” “always,” and an option for “not applicable or uncertain”). The potential item set plus three global questions and a 100-mm visual analog scale of overall recovery were sent via a digital link to patients and caregivers on the day of surgery for day-stay patients or postoperative day (POD) 1. A subset of patients who consented to multiple survey collections were asked to respond on PODs 1 to 2, then weeks 2 to 5 (11 potential responses).

Pediatric patients aged 2 to 17 yr old presenting from home for day surgery or surgery requiring postoperative admission American Society of Anesthesiologists (Schaumburg, Illinois) Physical Status III or lower Parent or guardian is the primary caregiver to the patient; in the event of two parents or caregivers, one parent will complete the survey(s) for the child (2 to 7 yr) or concurrently with the child (8 to 17 yr) Parental or guardian permission (informed consent) and child assent

Non-English speakers of patient or parents or guardians Urgent or emergent surgery Developmental delay or severe systemic disease that impacts activities of daily living Parents or guardians or patients who, in the opinion of the investigator, may be not available to complete the study

Confirmatory factor analysis was performed by choosing an optimal number of items to balance accuracy and acceptability. A reliability analysis of the final questionnaire based on the confirmatory factor analysis was conducted. In addition to the overall Cronbach alpha, the “scale if item deleted” statistics were generated and reviewed to determine whether any items negatively affected the reliability of the scale. Items with low item-total correlations were considered for removal if they contributed to a significant increase in Cronbach alpha when omitted. The results provided a measure of reliability for the entire scale and helped ensure that the items consistently reflected the intended construct. The construct validity of the tool was assessed by testing convergence (correlations) between the 20-item score and the overall global QoR items and the 100-mm visual analog scale of overall recovery. Finally, how well the model was able to reproduce the data was measured using common indices of model fit. These include the root mean square error of approximation (RMSEA), the Common Fit Index, the Tucker–Lewis Index (TLI), and the standardized root mean squared residual.

High-priority areas from the inductive process for patients included mobility, self-care, and access to a strong social support system after surgery. In contrast, caregivers focused on clear and open communication between themselves and the team of healthcare providers in addition to being equipped with appropriate recovery education. The results of the expert group survey were thematically analyzed and grouped according to the prioritization of the respondents. “Safety and adverse events” and “pain and comfort,” were given the most emphasis, while “functional recovery” and “return to normal physiology” were also highly prioritized. “Care perception, feeling safe and family engagement” and “distress and anxiety” were rated moderately. “Integrated holistic, patient-centered care,” “resolution of problem,” “psychosocial recovery,” and “caregiver distress” were also mentioned. The deductive process identified 41 relevant instruments from the adult and pediatric literature, and 216 relevant items were distilled for review from all sources. Additional free-thought questions from subject matter experts were included to obtain 264 items. Item reduction before categorization resulted in 106 items. The response rate for the first Delphi round was 57% (n = 27). Results from the first Delphi round are included in the Supplemental Digital Content (https://links.lww.com/ALN/D969).

The deductive process identified 41 relevant instruments from the adult and pediatric literature, and 216 relevant items were distilled for review from all sources. Additional free-thought questions from subject matter experts were included to obtain 264 items. Item reduction before categorization resulted in 106 items. The response rate for the first Delphi round was 57% (n = 27). Results from the first Delphi round are included in the Supplemental Digital Content (https://links.lww.com/ALN/D969). Item-level content validity index calculations for each dimension are presented in the Supplemental Digital Content (https://links.lww.com/ALN/D969). Based on these results, the lowest components of “thirst” and “sore throat” were marked for elimination, resulting in a scale-level contact validity index of 0.72, which is above the threshold of 0.70 for a panel of 27 respondents.28,29 “Sleep disturbance” was merged into the physical dimension of “sleep,” and an additional descriptive dimension of “mobility” was added to the physical health domain. For the second round of the Delphi process, respondents rated each of the 106 items within the 24 dimensions identified from the first Delphi round. These ratings were rank-ordered through summation of the respondents’ rankings. For full details of the rankings, please refer to the Supplemental Digital Content (https://links.lww.com/ALN/D969).

he Delphi process, respondents rated each of the 106 items within the 24 dimensions identified from the first Delphi round. These ratings were rank-ordered through summation of the respondents’ rankings. For full details of the rankings, please refer to the Supplemental Digital Content (https://links.lww.com/ALN/D969). Based on the Delphi results, lower-ranked items, and a review of the themes identified in the semistructured interviews, the working group reduced the number of items from 106 to 47, with the inclusion of 3 additional global questions. This list of items was agreed upon by the working group for psychometric evaluation in a representative sample population. The complete dataset was analyzed by searching for questions with more than 10% missing data, floor and ceiling responses greater than 15%, and poor variability assessed by skew. None of the questions were eliminated during this stage. The recruitment phase was between August 2021 and November 2022 but was extended for a further 6 months until June 2023 because of a higher than expected incomplete response rate for the multiple timepoint cohort and insufficient numbers across some age brackets. The initial analysis was performed on 936 usable responses from 1,162 participants (table 1) that were received on the day of surgery and POD 1. Responses from the later timepoints were not included in the initial validation analysis; however, some items were retained in the final item set as they loaded strongly on later timepoints. Demographics of Questionnaire Respondents

The initial analysis was performed on 936 usable responses from 1,162 participants (table 1) that were received on the day of surgery and POD 1. Responses from the later timepoints were not included in the initial validation analysis; however, some items were retained in the final item set as they loaded strongly on later timepoints. Demographics of Questionnaire Respondents Minor surgery was classified as anything performed as a day-case procedure, and major surgery was defined as any procedure requiring an inpatient stay. Kaiser–Meyer–Olkin was calculated to be 0.866 for the self-report version and 0.898 for the proxy version, indicating excellent sampling adequacy. The Bartlett test of sphericity was less than 0.001 for both datasets, indicating that they were appropriate for factor analysis. The scree plots (figs. 1 and 2) indicated an inflection at two factors, supporting the division into physical and mental well-being constructs. Applying the Kaiser eigenvalue-greater-than-1 rule, five factors aligned with the theoretical framework in both the self- and proxy-reported versions, explaining 62% and 65% of the variance, respectively (table 2). This meets an acceptable cumulative percentage of explained variance, which is 55 to 65%.27 Cumulative Percentage of Explained Variance in the Self-report and Proxy Questionnaires Scree plot of eigenvalues for the self-report version of the questionnaire. Scree plot of eigenvalues for the proxy report version of the questionnaire.

The scree plots (figs. 1 and 2) indicated an inflection at two factors, supporting the division into physical and mental well-being constructs. Applying the Kaiser eigenvalue-greater-than-1 rule, five factors aligned with the theoretical framework in both the self- and proxy-reported versions, explaining 62% and 65% of the variance, respectively (table 2). This meets an acceptable cumulative percentage of explained variance, which is 55 to 65%.27 Cumulative Percentage of Explained Variance in the Self-report and Proxy Questionnaires Scree plot of eigenvalues for the self-report version of the questionnaire. Scree plot of eigenvalues for the proxy report version of the questionnaire. Communalities (proportion of common variance) were examined using these factors to consider the items for review. Poorly loaded items, defined as less than 0.4, and strong cross-loadings, defined as less than 0.6 and greater than 0.4 on two separate factors, were reviewed. The factor loadings based on a five-factor solution for the early postoperative period are listed in table 3. Some items were retained even if they loaded poorly, either due to the question loading strongly on the other version (proxy or self-report) or if the loading became stronger over time and the working group felt it measured an important component of recovery. Factor Loadings for the Proxy and Self-report Versions of the PedsQoR Item Set Based on a Five-factor Solution “I had trouble sleeping” had loadings < 0.4 in the early postoperative period. PedSQoR, Pediatric Scale for Quality of Recovery.

Communalities (proportion of common variance) were examined using these factors to consider the items for review. Poorly loaded items, defined as less than 0.4, and strong cross-loadings, defined as less than 0.6 and greater than 0.4 on two separate factors, were reviewed. The factor loadings based on a five-factor solution for the early postoperative period are listed in table 3. Some items were retained even if they loaded poorly, either due to the question loading strongly on the other version (proxy or self-report) or if the loading became stronger over time and the working group felt it measured an important component of recovery. Factor Loadings for the Proxy and Self-report Versions of the PedsQoR Item Set Based on a Five-factor Solution “I had trouble sleeping” had loadings < 0.4 in the early postoperative period. PedSQoR, Pediatric Scale for Quality of Recovery. “I had vomiting” loaded strongly on the proxy version and was therefore included. “I felt sick in my stomach, like I needed to throw up” loaded on the psychologic and cognitive factor in the self-report version; however, it loaded more strongly on the physical symptoms factor in the proxy version (0.754). Therefore, it was retained in that factor. “I have had more nightmares than usual” loaded strongly on the self-report version by days 3 and 4 (0.806 and 0.733) and was retained. “I have been thinking about being in hospital” had a moderate loading and was retained as the working group considered it could be reworded to better reflect intrusive thoughts of a traumatic experience. The revised version is, “I had upsetting thoughts or feelings about being in the hospital.” “I had trouble sleeping” did not load strongly on POD 0 or 1, but by day 2, the proxy version loading was 0.653, and by day 4, the self-report loading was 0.879.

ed to better reflect intrusive thoughts of a traumatic experience. The revised version is, “I had upsetting thoughts or feelings about being in the hospital.” “I had trouble sleeping” did not load strongly on POD 0 or 1, but by day 2, the proxy version loading was 0.653, and by day 4, the self-report loading was 0.879. Further factor analysis on the later timepoints (days 2 to 4) evaluated performance over time of items. In general, loading on items related to pain and vomiting tended to decrease over time, whereas sleep- and anxiety-related loading increased. Item reduction was based on the iterative analysis of forced factor loading to three, four, or five factors and self- or proxy-reported performance with respect to time. The working group considered an optimal number of items in a scale, with a balance between accuracy and acceptability of approximately 20. To ensure accuracy, a minimum of three items pertaining to each factor was required, and the concepts of physical and mental well-being were considered separately with equal weighting (number of items) for each concept. The five-factor confirmatory factor analysis satisfied the requirements and supported the theoretical structure of the item categorization and construct validity of the selected items.

and the concepts of physical and mental well-being were considered separately with equal weighting (number of items) for each concept. The five-factor confirmatory factor analysis satisfied the requirements and supported the theoretical structure of the item categorization and construct validity of the selected items. The reliability of the proxy and self-report versions was excellent, with Cronbach alpha values of 0.906 and 0.922, respectively, based on 20 items in a five-factor solution. Cronbach alpha for the items relating to the concepts of physical or mental well-being across all timepoints was 0.784 and 0.877, respectively. Correlations with broad assessments of recovery were moderate (table 4). Correlations between PedSQoR and Broad Measures of Recovery, the 100-mm VAS of Recovery, Q1—“I Feel that I Am Back to Normal,” and Q49R (Reversed Polarity)—“I Still Have Problems from My Anesthesia or Surgery” Pearson correlation coefficient. Correlation is significant at the 0.01 level (two-tailed). PedSQoR, Pediatric Scale for Quality of Recovery; VAS, visual analog scale.

Correlations between PedSQoR and Broad Measures of Recovery, the 100-mm VAS of Recovery, Q1—“I Feel that I Am Back to Normal,” and Q49R (Reversed Polarity)—“I Still Have Problems from My Anesthesia or Surgery” Pearson correlation coefficient. Correlation is significant at the 0.01 level (two-tailed). PedSQoR, Pediatric Scale for Quality of Recovery; VAS, visual analog scale. The RMSEA was 0.096 (95% CI, 0.093 to 0.099), the Common Fit Index was 0.854, the TLI was 0.831, and the standardized root mean squared residual was 0.066 (recommended less than 0.7). The RMSEA (expected 0.08) was slightly higher, and the TLI (expected 0.9) was slightly lower, which is not unexpected given that multiple concepts (both parent and child questionnaires) were fitted in the single model to make the tool versatile across timepoints and modes of response. Additionally, several items were forced in based on construct validity, as determined by the working group. The final question sets for the proxy and self-report versions of PedSQoR are shown in table 5. The PedSQoR Final Question Set

The RMSEA was 0.096 (95% CI, 0.093 to 0.099), the Common Fit Index was 0.854, the TLI was 0.831, and the standardized root mean squared residual was 0.066 (recommended less than 0.7). The RMSEA (expected 0.08) was slightly higher, and the TLI (expected 0.9) was slightly lower, which is not unexpected given that multiple concepts (both parent and child questionnaires) were fitted in the single model to make the tool versatile across timepoints and modes of response. Additionally, several items were forced in based on construct validity, as determined by the working group. The final question sets for the proxy and self-report versions of PedSQoR are shown in table 5. The PedSQoR Final Question Set Items to be answered based on the stem, “In the last 24 h.” Responses are as follows: 1, always; 2, often; 3, sometimes; 4, rarely; 5, never, or not applicable. The overall score will range from 20 to 100 with higher scores indicating a better quality of recovery or more complete recovery. Separate scores out of 50 should be obtained for the physical and mental well-being domains. Responses of “not applicable” or missing values will be treated in the same way. Total the scores in the physical or mental well-being domains separately and divide by the number of valid responses to find the average for that domain, and then multiply by 10, and round to the nearest whole number to determine the overall score for that domain. Responses are as follows: 1, always; 2, often; 3, sometimes; 4, rarely; and 5, never, or not applicable.

Items to be answered based on the stem, “In the last 24 h.” Responses are as follows: 1, always; 2, often; 3, sometimes; 4, rarely; 5, never, or not applicable. The overall score will range from 20 to 100 with higher scores indicating a better quality of recovery or more complete recovery. Separate scores out of 50 should be obtained for the physical and mental well-being domains. Responses of “not applicable” or missing values will be treated in the same way. Total the scores in the physical or mental well-being domains separately and divide by the number of valid responses to find the average for that domain, and then multiply by 10, and round to the nearest whole number to determine the overall score for that domain. Responses are as follows: 1, always; 2, often; 3, sometimes; 4, rarely; and 5, never, or not applicable. The overall score will range from 20 to 100 with higher scores indicating a better QoR or more complete recovery. Separate scores out of 50 should be obtained for the physical and mental well-being domains. Responses of “not applicable” or missing values will be treated in the same way. Total the scores in the physical or mental well-being domains separately and divide by the number of valid responses to find the average for that domain, and then multiply by 10, and round to the nearest whole number to determine the overall score for that domain.

Based on the Delphi results, lower-ranked items, and a review of the themes identified in the semistructured interviews, the working group reduced the number of items from 106 to 47, with the inclusion of 3 additional global questions. This list of items was agreed upon by the working group for psychometric evaluation in a representative sample population. The complete dataset was analyzed by searching for questions with more than 10% missing data, floor and ceiling responses greater than 15%, and poor variability assessed by skew. None of the questions were eliminated during this stage. The recruitment phase was between August 2021 and November 2022 but was extended for a further 6 months until June 2023 because of a higher than expected incomplete response rate for the multiple timepoint cohort and insufficient numbers across some age brackets. The initial analysis was performed on 936 usable responses from 1,162 participants (table 1) that were received on the day of surgery and POD 1. Responses from the later timepoints were not included in the initial validation analysis; however, some items were retained in the final item set as they loaded strongly on later timepoints. Demographics of Questionnaire Respondents Minor surgery was classified as anything performed as a day-case procedure, and major surgery was defined as any procedure requiring an inpatient stay.

Kaiser–Meyer–Olkin was calculated to be 0.866 for the self-report version and 0.898 for the proxy version, indicating excellent sampling adequacy. The Bartlett test of sphericity was less than 0.001 for both datasets, indicating that they were appropriate for factor analysis. The scree plots (figs. 1 and 2) indicated an inflection at two factors, supporting the division into physical and mental well-being constructs. Applying the Kaiser eigenvalue-greater-than-1 rule, five factors aligned with the theoretical framework in both the self- and proxy-reported versions, explaining 62% and 65% of the variance, respectively (table 2). This meets an acceptable cumulative percentage of explained variance, which is 55 to 65%.27 Cumulative Percentage of Explained Variance in the Self-report and Proxy Questionnaires Scree plot of eigenvalues for the self-report version of the questionnaire. Scree plot of eigenvalues for the proxy report version of the questionnaire. Communalities (proportion of common variance) were examined using these factors to consider the items for review. Poorly loaded items, defined as less than 0.4, and strong cross-loadings, defined as less than 0.6 and greater than 0.4 on two separate factors, were reviewed. The factor loadings based on a five-factor solution for the early postoperative period are listed in table 3. Some items were retained even if they loaded poorly, either due to the question loading strongly on the other version (proxy or self-report) or if the loading became stronger over time and the working group felt it measured an important component of recovery.

on for the early postoperative period are listed in table 3. Some items were retained even if they loaded poorly, either due to the question loading strongly on the other version (proxy or self-report) or if the loading became stronger over time and the working group felt it measured an important component of recovery. Factor Loadings for the Proxy and Self-report Versions of the PedsQoR Item Set Based on a Five-factor Solution “I had trouble sleeping” had loadings < 0.4 in the early postoperative period. PedSQoR, Pediatric Scale for Quality of Recovery. “I had vomiting” loaded strongly on the proxy version and was therefore included. “I felt sick in my stomach, like I needed to throw up” loaded on the psychologic and cognitive factor in the self-report version; however, it loaded more strongly on the physical symptoms factor in the proxy version (0.754). Therefore, it was retained in that factor. “I have had more nightmares than usual” loaded strongly on the self-report version by days 3 and 4 (0.806 and 0.733) and was retained. “I have been thinking about being in hospital” had a moderate loading and was retained as the working group considered it could be reworded to better reflect intrusive thoughts of a traumatic experience. The revised version is, “I had upsetting thoughts or feelings about being in the hospital.” “I had trouble sleeping” did not load strongly on POD 0 or 1, but by day 2, the proxy version loading was 0.653, and by day 4, the self-report loading was 0.879.

Patient-reported outcomes development is a multiphase project that requires a mixed-methods approach. An ideal measure of recovery needs to be valid, responsive, and administrable across the full range of recovery phases and trajectories; multidimensional to have content validity for the construct; able to accurately discriminate important outcomes of interest; be as succinct as possible; and still maintain reliability and accuracy. A previous systematic review of pediatric measures evaluating the recovery process4 established the absence of a psychometrically developed and validated instrument that can be used in pediatric patients. This is a challenging demand in the pediatric population, perhaps reflecting the current absence of such an instrument. Our literature review uncovered three instruments that have been used to evaluate recovery in the pediatric population.19,30,31 These instruments are conceptually different from our proposed instrument and have not been validated across different developmental stages. A patient-centered and -reported instrument that evaluates the construct of recovery from procedure-related perturbations in the pediatric population is required as a core outcome measure. This measure can be used in combination with others to provide an essential component of the core outcome set for pediatric patients. It is vital that we have robust, standardized endpoints in pediatric perioperative medicine so that results from clinical trials can be synthesized and compared, and well-supported recommendations can be made.

d in combination with others to provide an essential component of the core outcome set for pediatric patients. It is vital that we have robust, standardized endpoints in pediatric perioperative medicine so that results from clinical trials can be synthesized and compared, and well-supported recommendations can be made. In the first phase, the working group conceptualized the construct and determined the context and mode of administration. Previous groups17,31–33 have conceptualized recovery along the domains of established quality-of-life measures: physical well-being, mental well-being, social functioning, role functioning, and general health perceptions. Similarly, other organizations, such as the National Institutes of Health for the Patient-Reported Outcomes Measurement Information System34 and the World Health Organization (Geneva, Switzerland) for the International Classification of Functioning, Disability, and Health,35 employ analogous domain frameworks. To ensure content validity, we initially used the comprehensive categorization structure from the International Classification of Functioning, Disability, and Health, followed by the established QoR domains for preliminary conceptualization. Item development adhered to established methodologies for scale development, integrating both deductive and inductive insights. Items were extracted from an extensive range of sources and subsequently refined or pruned using established psychoclinimetric techniques appropriate for scale development. Items were retained only if they maintained applicability across stages of childhood development from postinfancy to adulthood. Therefore, some compromises were required with respect to the inclusion of potentially applicable domains.

ned using established psychoclinimetric techniques appropriate for scale development. Items were retained only if they maintained applicability across stages of childhood development from postinfancy to adulthood. Therefore, some compromises were required with respect to the inclusion of potentially applicable domains. Ideally, the weight of each dimension should be proportional to the importance of that area in the target construct. Through the development process, a patient-centered focus allowed item reduction and selection that adhered to this guiding principle. To improve fidelity and reliability, and to weigh the relative importance of dimensions, increased items or proportional weighting can be used; however, this must be balanced against usability. The current study endeavored to maintain broad applicability and usability across the full spectrum of childhood and recovery phases. To satisfy these requirements, some specific components of recovery are not retained, thus reducing the content validity. For example, the early physiologic recovery phase,36 cognitive recovery, and social and role functions cannot be fully evaluated because items related to these components will show significant floor and ceiling effects or will not be relevant to large proportions of the pediatric population. We attempted to include items that covered the components of cognition, affectivity, interpersonal functioning, and impulse control, with the intention of future evaluation of the PedSQoR for its utility as a screening instrument for significant psychologic disorders. We demonstrated changes in the performance of the items with respect to self- or proxy reporting and postoperative recovery time. These important differences must be considered when selecting items for scales administered at different timepoints during the recovery trajectory.

ent for significant psychologic disorders. We demonstrated changes in the performance of the items with respect to self- or proxy reporting and postoperative recovery time. These important differences must be considered when selecting items for scales administered at different timepoints during the recovery trajectory. Several items were not retained in the final item set due to cross-loading on factors. Examples of cross-loading were the items “When I think about pain it makes me sad,” which loaded onto the psychologic factor (0.571) and the pain factor (0.492) for proxy reporting, and “Pain has stopped me doing the things I normally do,” which had cross-loading on emotional (0.624) and pain (0.533) in the self-reporting group. Best practice suggests that an item should be predominantly loaded onto a single factor (with a ratio of communalities of more than 75% or using absolute values of more than 0.6 and less than 0.4). Alternative items that more specifically represented these dimensions of recovery were included in preference to the cross-loading items.

uggests that an item should be predominantly loaded onto a single factor (with a ratio of communalities of more than 75% or using absolute values of more than 0.6 and less than 0.4). Alternative items that more specifically represented these dimensions of recovery were included in preference to the cross-loading items. There are several limitations to our study. The mode of reporting utilizing a link sent via text message or email was selected to increase the usability and initial development of the instrument; however, further validation is required for paper and face-to-face administration. Additionally, the sample population used for validation consisted of children undergoing elective procedures, so further validation in critically ill children and those undergoing emergency procedures should be performed. The response rate for the Delphi rounds was relatively low at 57% and therefore could have been strengthened by a more complete response from the expert group and by including a broader range of patients and families. The response rate for the multiple-timepoint questionnaires decreased over time and by week 5 was less than 50%. The responses received from the later timepoints may represent more motivated patients and families and may not be truly representative of the target population. Our sample consisted of children from tertiary hospitals in the United States and Australia and may not be representative of all pediatric populations, so the instrument will require testing in a range of different ethnic, socioeconomic, and geographical populations. A 20-item scale such as this is suitable for research and quality assurance activities; however, it may not always be practical for routine clinical use. Further modifications may be required to make it suitable for routine use and for integration into electronic health records.

oeconomic, and geographical populations. A 20-item scale such as this is suitable for research and quality assurance activities; however, it may not always be practical for routine clinical use. Further modifications may be required to make it suitable for routine use and for integration into electronic health records. We recommend that the PedSQoR instrument is used in patients aged 2 to 18 yr at any timepoint after discharge from the postanesthesia care unit. The initial validation of the instrument was carried out using responses from postoperative days 0 and 1; however, data were also analyzed on responses at later timepoints out to 5 weeks after the procedure. Some items loaded more strongly at later timepoints—for example, sleep- and anxiety-related issues—whereas pain and gastrointestinal issues decreased over time as would be expected. The PedSQoR will give an overall score ranging from 20 to 100, with higher scores indicating better recovery, and the physical and mental well-being domains can also be scored separately out of 50. We would encourage users of the instrument to trial it in perioperative research and quality assurance activities to provide further validation and to determine the minimal clinically important difference in scores.

better recovery, and the physical and mental well-being domains can also be scored separately out of 50. We would encourage users of the instrument to trial it in perioperative research and quality assurance activities to provide further validation and to determine the minimal clinically important difference in scores. This third phase of the scale’s evaluation should be performed on new sample datasets.37 The hypothesized dimensions that make up the construct can be tested using structural equation modeling and confirmatory factor analysis (polychoric correlation and RMSEA-based maximum likelihood method), and the model fit should be tested using multiple indices.37,38 Further validation of the PedSQoR scale should be conducted to assess responsiveness and temporal changes. This instrument can be revised for wider validation studies, including translated versions and patient groups that were not included in this study. This scale was developed for pediatric populations from the United States and Australia, and it is envisaged that in the future, the scale will be validated in different languages and distinct patient populations. For example, in Australia, the First Nations peoples will have a unique set of priorities related to recovery after surgery and anesthesia. It is hoped that researchers, in conjunction with the First Nations peoples, can use the PedSQoR scale as a template to codesign a questionnaire specific to Indigenous Australians. This will allow more specific measurement of outcomes for First Nations children and allow more tailored interventions to improve overall health outcomes. Similar processes can occur in specific patient populations worldwide.

he PedSQoR scale as a template to codesign a questionnaire specific to Indigenous Australians. This will allow more specific measurement of outcomes for First Nations children and allow more tailored interventions to improve overall health outcomes. Similar processes can occur in specific patient populations worldwide. We developed a scale for QoR in pediatric patients aged between 2 and 17 yr using a mixed-method approach that conforms to the COSMIN guidelines for scale development. The scale was psychometrically valid and demonstrated good reliability. The PedSQoR instrument requires further validation in specific pediatric populations, and we encourage researchers to include it as an outcome measure in pediatric perioperative trials. We suggest that the PedSQoR scale become a standardized outcome measure after surgery and anesthesia in children. The authors would like to thank the following people for their significant contribution to this work: Paula Hu (B.S.N., M.P.H., The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania), Elsa Taylor (M.B.Ch.B., Department of Pediatric Anesthesia, Starship Children's Hospital, Auckland, New Zealand), Ivy Pham (B.Sc., M.P.H., Boston Children's Hospital, Boston, Massachusetts), Benjamin Telicki (B.A., Department of Anesthesiology, Critical Care and Pain Medicine, Boston Children's Hospital, Boston, Massachusetts). This work was carried out with the support of a grant from the Society for Pediatric Anesthesia in Australia and New Zealand (Bonnells Bay, Australia). The authors declare no competing interests.

The authors would like to thank the following people for their significant contribution to this work: Paula Hu (B.S.N., M.P.H., The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania), Elsa Taylor (M.B.Ch.B., Department of Pediatric Anesthesia, Starship Children's Hospital, Auckland, New Zealand), Ivy Pham (B.Sc., M.P.H., Boston Children's Hospital, Boston, Massachusetts), Benjamin Telicki (B.A., Department of Anesthesiology, Critical Care and Pain Medicine, Boston Children's Hospital, Boston, Massachusetts).

Supplemental material, https://links.lww.com/ALN/D969 Table S1. Item-level content validity indices for the Delphi panel responses for the dimensions of the construct of QoR Table S2. Scale-level content validity index for 25 items and with “sore throat” and “thirst” removed Table S3. Results of the responses from the second Delphi round