Healthcare Provider and Parent Behavior and Children’s Coping and Distress at Anesthesia Induction

Participants in the current study were children with American Society of Anesthesiology physical status I or II who were a part of the National Institutes of Health–funded Behavioral Interactions-Perioperative Study (BIPS). The BIPS is a large-scale multiyear project assessing main effects and moderators of adult behaviors on children’s perioperative distress. Preliminary results of children’s behavior in the BIPS are reported elsewhere.16Children recruited for the BIPS were aged 2–10 yr and undergoing outpatient surgery with general anesthesia. Exclusion criteria included children with chronic illness, children with developmental delay, and children with parents who did not speak English.

The Revised Perioperative Child–Adult Medical Procedure Interaction Scale (R-PCAMPIS) is an observational behavioral coding system designed to capture children’s and adults’ behaviors in the perioperative setting. Based on the originally validated PCAMPIS,17the R-PCAMPIS includes 44 operationally defined verbal and nonverbal behavioral codes. Modifications to the original PCAMPIS were made to facilitate the interface between the coding system and a behavioral collection computer system, Observer® XT (Noldus Inc., Wageningen, The Netherlands). Specifically, the original PCAMIS was modified to differentiate between state codes (i.e. , codes representing behaviors with meaningful durations, such as touch) and event codes (i.e. , codes representing behaviors with meaningful frequencies, such as reassuring comments).

Because of the extremely large amount of data, six theoretically and practically relevant hypothesis driven adult verbal and nonverbal codes are examined in this report: (1) medical reinterpretation (reinterpreting medical equipment and procedures as nonthreatening, or medical play; e.g. , “Are you ready to play the astronaut game?” referring to the mask), (2) nonprocedural talk (distracting talk about topics outside the surgery center; e.g. , “What grade are you in?”), (3) humor (jokes, laughing with intent of engaging the child and improving his or her mood), (4) reassurance (comments with the intent of comforting the child about his or her condition or the course of the procedure; e.g. , “It’s ok, you’re fine”), (5) empathy (statements that express understanding of or identification with children’s emotions; e.g. , “I know you’re scared”), and (6) empathic touch (touch with the purpose of comforting the child; e.g. , holding the child’s hand, rubbing the child’s back or head).

The R-PCAMPIS also contains codes for children’s behavior that are combined to yield profiles of children’s behavior during induction (i.e. , from the time the child left the holding area until the time anesthesia was induced). Operationally defined child codes of nonverbal resistance and cry are included in the Acute Distress subscale. Codes of verbal resistance (e.g. , “Get me out of here”), requesting support (e.g. , “Mommy!”), and negative verbal emotion (e.g. , “I’m scared”) are included in the Anticipatory Distress subscale. An additional profile of Early Regulating Behaviors, including nonprocedural talk (e.g. , talking about school), medical play (e.g. , pointing at “mountains” on the monitor), and information seeking (e.g. , asking questions about the procedure), is also compiled using codes from the R-PCAMPIS and is representative of children’s coping. Data on validity and reliability of these behavior profiles are reported elsewhere.16Specifically, controlling for age, Acute Distress and Anticipatory Distress profiles were significantly positively related to a validated measure of children’s anxiety, the Yale Perioperative Anxiety Scale. The Early Regulating Behavior profile was significantly negatively related to Yale Perioperative Anxiety Scale scores.

All behaviors that are short verbalizations (e.g. , reassurance, negative verbal emotion) are represented by event codes. Nonverbal or prolonged vocal behaviors (empathic touch, cry, nonverbal resistance) are represented by state codes. Scoring of event codes is accomplished by dividing frequency of each event code by the total number of minutes in the observation period, yielding a metric of rate of code per minute. Scoring of state codes is accomplished by dividing the total duration of each state code by the total duration of the observation period, yielding a metric of proportion of observation. These procedures correct for varying durations of anesthesia induction.

One researcher with a master’s degree and two researchers with bachelor’s degrees completed the behavioral coding. All coders underwent a 3-month training protocol under the direction of the first author (J.M.C.). This training process included two phases. First, coders were familiarized with the technological coding interface, Observer® XT, via  administration of a simplified set of behavioral codes. Second, coding of study-independent training videos was accomplished. Multiple raters coded each training video and met, at length, with the first author to discuss reliability statistics and disagreements. Raters were considered “trained” when they met a κ criterion of 0.80 agreement with the first author on training tapes.

Administration of the R-PCAMPIS was facilitated by using Observer® XT, a behavior-analysis software package with the capabilities to code behaviors of one individual, or the interactions of many. This system allows for the linking of particular behaviors (e.g. , reassurance) to the subject who initiated the behavior (e.g. , parent, nurse, anesthesiologist). Data coding was accomplished in passes, with each behavior coded in a separate pass. Real-time second-to-second data coding was used with timed onsets of each event code recorded and the onset and offset of each state code recorded. Although this methodology is time-consuming, it ensures maximum reliability and validity of coding. Coding required approximately 4 h per participant, representing a total of more than 1,500 h of coding.

Interrater reliability of individual behavioral codes was assessed by having two research assistants overlap on 10% of participants. Timed-event κ coefficients were in the excellent range18for all codes (Child Acute Distress composite = 0.79, Child Anticipatory Distress = 0.88, Child Early Regulating Behaviors composite = 0.92, Reinterpretation = 0.91, Humor = 0.94, Empathy = 0.96, Reassurance = 0.92, Nonprocedural Talk = 0.93, Empathic Touch = 0.83). Reliability assessment and discussion was a process repeated weekly throughout coding. One reliability subject was coded per week; after κ values were calculated, coders met with the first author to discuss disagreements. Decisions on valid coding of behaviors were incorporated into the final version of the observational records.

All procedures were approved by the Yale Human Investigation Committee (New Haven, Connecticut). All attending pediatric anesthesiologists and operating room nurses practicing in the surgery centered provided informed consent for participation in this study. Healthcare providers were informed the purpose of this study in vague terms to decrease the impact of demand characteristics on their behavior (i.e. , “to look at interactions among healthcare providers, children, and families in the preoperative period”). Child and parent participants were recruited by phone between 1 week and 1 day before surgery or on the morning of surgery. Parents provided written informed consent, and children provided written assent as age appropriate (children 7 yr old and older). After informed consent, parents completed a demographic questionnaire and measures relevant to the larger BIPS. All children were accompanied to the operating room by one parent, and no child received any sedative premedication.

A trained research assistant using a handheld digital video camera (Sony Handycam DCR-HC21; Sony Electronics Inc., San Diego, CA) videotaped all children from the time the child left the holding area until anesthesia was induced. Digital video files were converted to.mpg files and imported into Observer® XT software for coding, compiling, and analysis.

In previous studies using the CAMPIS in pediatric settings, small to large effect sizes have been found for the relation between specific adult behaviors and child distress (e.g. , r  values ranging from approximately 0.2 to 0.6 in observational studies and Cohen d values of 0.35 to 0.45 in intervention studies). Effect sizes in the middle of this range would be statistically significant at the α= 0.05 level with power of 0.80 in regression analysis with a sample size of 85–100 participants. These effect sizes are based on direct, univariate associations between observed parent behaviors and child distress using the CAMPIS. In the case of the current study, controlling for age using hierarchical regression requires an increase in sample size. Assuming age accounts for 5% of variability in children’s distress, a sample size of n = 264 is needed to detect an additional ΔR  ²of at least 0.025 attributed to adult behaviors using an F test with an α level of 0.05. Based on these computations, a sample size of approximately 300 will provide sufficient power to address the primary aims of this study.

Data were analyzed in a series of steps. First, rates of specific behaviors were compared across adults. Given that rates were not normally distributed and transformations were not successful in reaching normal distribution, nonparametric statistics were used. Friedman analyses of variance were used to compare rates across adults, and follow-up pairwise comparisons were conducted using Wilcoxon signed ranks tests. Bonferroni corrected P  values were used to control for family-wise error. Next, Spearman rank order correlations were used to examine relations among adults’ rates of behaviors and among rates of behaviors combined across adults. Results of these analyses led to empirically driven combinations of codes, which were then examined for their relations with children’s coded distress and regulating behaviors using Spearman correlation and logistic regression.

To account for attrition 338 children and their parents were enrolled in the BIPS project. Of these participants, 45 child–parent dyads were excluded for the following reasons: 25 were excluded because only very partial video data were available, 11 were excluded because the child was given sedative premedication as a rescue intervention, three participants dropped after providing consent without explanation, one child was admitted to the hospital, two children had anesthesia induced using intravenous access rather than mask, two parents spoke languages other than English to their child during induction, and one child had surgery canceled. Therefore, the final sample of participants included in this report is 293 children and their accompanying parent. Forty-eight percent of these children were female, and most were non-Hispanic white (85.7%). Thirty-five percent of children had previous experience with surgery. The most common surgical procedures the participants underwent included tonsillectomy and/or adenoidectomy (n = 96), followed by pressure-equalizing tube placement (n = 47), endoscopy (n = 40), urologic procedures (n = 28), hernia repairs (n = 21), and dermatologic procedures (n = 14). The majority of children were accompanied to the operating room by their mother (n = 241), and the remainder were accompanied by their father (n = 41).

All attending pediatric anesthesiologists (n = 14) and operating room nurses who practiced in the study surgery center were included in this study. Results of a hierarchical model with codes for anesthesiologist identity entered as a covariate indicated that, when adjusted for phase and anesthesiologist behavior, there were no statistically significant differences among anesthesiologists (P > 0.8 for all). In the hierarchical models, this indicates very little variability among anesthesiologists, thus rendering hierarchical models inappropriate. Therefore, nonnested models will be presented here. Unfortunately, the identity of operating room nurses was not available for analyses; therefore, nurse behavior could not be tested in a hierarchical nested model.

Rates of the six behaviors of interest for anesthesiologists, parents, and nurses are shown in table 1. It is notable that the distributions of most behaviors were positively skewed; therefore, median and interquartile range are reported. Friedman analyses of variance indicated significant differences across adults on all behaviors (table 1). Nurses engaged in the lowest rates of all behaviors. Anesthesiologists and parents engaged in similar rates of reassurance, empathy, and humor. Anesthesiologists used the highest rate of medical reinterpretation and parents used the highest proportion of empathic touch.

Spearman rank order correlations were used to examine relations among rates behaviors among adults during anesthesia induction (table 2). There were significant positive correlations between parent–anesthesiologist and parent–nurse rates of reassurance, humor, and nonprocedural talk. Nurse and anesthesiologist rates of humor were significantly positively related; otherwise, no correlations were significant.

Additional analyses were conducted to examine relations among adult behaviors (across nurse, anesthesiologist, and parent). Significant spearman correlations were evidenced among reassurance, empathy, and empathic touch (P < 0.01) and between nonprocedural talk and humor (P < 0.01). The rate of reinterpretation was not significantly correlated with the rate of any other behavior. Based on these results, behaviors were empirically grouped to reduce factors in further analyses. Nonprocedural talk and humor were significantly correlated and therefore were grouped into a factor termed “distracting behavior.” Empathy, empathic touch, and reassurance were also significantly related to one another and therefore were grouped into a factor termed “emotion-focused behavior.” Medical reinterpretation was not significantly related to any other behavior and rates of reinterpretation were not related across adults; therefore, reinterpretation was examined individually for each adult. Relations among adult codes and children’s distress and regulating (i.e. , coping) behaviors are shown in table 3. Emotion-focused behavior was significantly positively related to children’s distress and negatively related to children’s regulatory behaviors. Distracting behavior showed the opposite pattern. Medical reinterpretation by anesthesiologists was significantly positively related to children’s regulatory behaviors, but the same behavior by parents was significantly positively related to children’s distress.

Logistic regression was used to examine the predictive ability of adult behaviors found to be significant in correlational analyses while accounting for child age (table 4). The first logistic regression used binary coded child distress (present/absent) as an outcome. Child age was entered in the first block, and adult emotion-focused behaviors, distracting behaviors, and parent reinterpretation were entered in the second block. The overall model was significant [χ²(4) = 134.4, P < 0.001] and accounted for 50.1% of the variance in child distress status (Nagelkerke R  ²). Adult behaviors accounted for significant variance above and beyond child age [χ²(3) Block = 104.6, P < 0.001], although emotion-focused behaviors were the only adult behavior with a significant β weight. A second logistic regression was conducted with child regulating behaviors (present/absent) as an outcome and child age (step 1) and adult behaviors (emotion-focused, distracting, anesthesiologist reinterpretation; step 2) as predictors. The overall model was significant [χ²(4) = 42.6, P < 0.001] and accounted for 30.4% of the variance in child distress status (Nagelkerke R  ²). Similar to the model for child distress, adult behaviors accounted for significant variance above and beyond child age [χ²Block (3) = 22.1 P < 0.001]. In addition to child age, adult emotion-focused behaviors and anesthesiologist reinterpretation had significant β weights in this model.

The current study examined adult behaviors during induction of anesthesia in children when parents are present. To our knowledge, this was the first study to examine multiple individuals who interact with children during induction of anesthesia and evaluate the relation between discrete, trainable behaviors and children’s distress and coping.

Six hypothesis-driven adult behaviors were examined based on previous literature in pediatric procedural pain. Overall, nurses engaged in the lowest rates of all behaviors when compared with anesthesiologists and parents. Parents and anesthesiologists displayed relatively comparable rates of most behaviors. In terms of differences, anesthesiologists used more medical reinterpretation than parents, and parents used more empathic touch than anesthesiologists. Correlations indicated that anesthesiologists’ and parents’ rates of most behaviors were correlated as were rates of nurses’ and parents’ behaviors. Taken together, these findings highlight the interactive nature of anesthesia induction with parents present. It is likely that parents take their cues from anesthesiologists and nurses on how to behave during the induction; when anesthesiologists use more nonprocedural talk and humor, so do parents. It is also possible that children drive these relations; adults may behave similarly because they are responding to the same behaviors from the child. These findings could have important implications for intervention. Training each parent who will be present at anesthesia induction is a time-consuming process. Results presented here suggest that healthcare personnel may be able to directly affect parents’ behavior by engaging in higher rates of desirable behaviors themselves.

The current study also examined relations among adults’ behaviors and children’s distress and coping during anesthesia induction. Two commonly cooccurring distracting behaviors, nonprocedural talk and humor, were significantly positively related to children’s coping and negatively related to children’s distress. Both nonprocedural talk and humor may reduce children’s distress through extinction (i.e. , by ignoring the behavior, thus removing any positive consequences such as attention) and have the added benefit distracting children by directing attention away from the potentially distressing medical procedure.

One of the most striking findings of the current report is the strong positive relation between adults’ emotion-focused behaviors (reassurance, empathy, empathic touch) and children’s distress. This finding is likely to incite discussion because it is both counterintuitive and contradictory to typical training of physicians in communication skills.18This finding is consistent, however, with at least four randomized controlled trials of emotion-focused type behavior on children’s acute distress.7,8,9,19Each of these studies was conducted by a different research group and used a different methodology, and three of the four included a different pain stimulus, thereby supporting the generalizability of these results. In terms of an explanation of this effect, authors have hypothesized that reassurance may cue children to be distressed by communicating to the child that the situation should be of concern or may serve to direct attention toward the unpleasantness of the situation, thereby increasing distress.20 

This study also examined the behavior of medical reinterpretation, which is defined as attempts to provide information on the induction procedure while reframing the procedure as less threatening (perhaps even fun). Not surprisingly, this behavior was more commonly engaged in by anesthesiologists than by parents or nurses. Interestingly, this behavior showed little concordance across adults: Parents were not necessarily more likely to use reinterpretation when anesthesiologists were (or vice versa ). In terms of outcomes, anesthesiologists’ use of reinterpretation was related to children’s regulating behaviors; children displayed a higher rate of regulating behaviors when anesthesiologists used more reinterpretation. The somewhat contradictory results of reinterpretation being used by parents rather than anesthesiologists were surprising. There was a positive association between parent rate of reinterpretation and children’s distress. The explanation for this finding is unclear. It is possible that parents are unfamiliar with equipment in the operating room and therefore are less successful in interpreting it as nonthreatening. Alternatively, it is possible that parents become more involved in reinterpretation when the child is more distressed. To our knowledge, this is the first time that medical reinterpretation has been described in the literature, and therefore, more research on this behavior and its impact on children’s distress is needed.

Several methodologic issues with the current study should be noted. First, although measures were taken to reduce reactivity (e.g. , participants were informed only of the general nature of the study, duration of study), it may be that participants’ behavior changed as a result of being observed. Second, this study was conducted in one pediatric surgery center, and therefore, results may not be widely generalizable outside such a center or to centers with different standards of practice. For example, although nurses showed significantly lower rates of the coded behaviors in this study, we are aware of other surgery centers in which nurses play the most prominent role in the induction. Third, this study did not take into account adult behaviors in the waiting area or at other times before surgery. There is little question that the behaviors of adults in these time periods influence children’s distress, and we do not intend to minimize these effects here. However, given the relatively large effect sizes found in this study, we assert that interactions during the induction can affect children’s experiences with this procedure. Fourth, this study was conducted with children who did not receive sedative premedication. Although it is our impression that the efficacy of specific anesthesiologist and nurse behaviors would generalize to children who were premedicated, further studies should be conducted to examine interactions with children who have received this intervention. Finally, it is also important to note that given the correlational nature of the current data, it is impossible to conclude that it is adults who are affecting children, as opposed to the children affecting adults. Future work in this area should consider sequential analysis to gather support for causation.21Experimental intervention studies will be needed to confirm these hypotheses. Despite these limitations, this study has methodologic strengths that support the validity of the findings. Specifically, this is the largest scale study collecting observational data of healthcare providers, parents, and children during acute medical procedures. Further, the examination of discrete behaviors strengthens the clinical utility of the findings.

In sum, the current study examined adults’ behaviors during anesthesia induction. Behaviors were identified that were related to increased and decreased distress and coping in children. These behaviors were relatively straightforward and should be easy to teach and incorporate in practice. These results should be considered preliminary, however and should be confirmed via  sequential analyses and randomized controlled trials. Further, future work should consider potential moderators of effects (e.g. , child temperament, previous surgical experience). Once confirmed, these behaviors could be easily incorporated into standard practice and effectively influence children’s distress before and during anesthesia induction.