INTRODUCTION AND PURPOSE
Multiple authors documented the use of music with infants and children undergoing medical procedures.1–4 Several studies examined the effects of music on pain associated with intramuscular injections,2 lumbar punctures,3 and intravenous cannulation.4 Others reported the effects of music on behavior and physiology of hospitalized neonates5–10 including increased oxygen saturation,5–7 improved heart rate,7 lower state lability,5,8 decreased agitation,5,9 increased sleeping time,7 steady weight gain,9 and increased rate of non-nutritive sucking.10 Keller11 and Barrera et al12 described positive effects of music therapy on increasing comfort and improving well-being of children with cancer. Another study found a significant increase in the number of smiles observed in 60 hospitalized children as the result of music therapy when compared with play therapy.13
Although many studies investigated the behavioral and physiological responses to music exhibited by infants and children,2–13 only a few published reports supported the use of music with children during physical therapy. Gouvier et al14 described the use of music as a type of positive feedback and aversive auditory stimulation as a type of negative feedback during physical and occupational therapy sessions conducted with a 21-year-old patient with head injury. They reported a 19.5% to 57.5% decrease in inappropriate behavior during occupational and physical therapy sessions.14 Walmsley et al15 described teaching head control to 5 children with quadriplegic cerebral palsy and mental retardation while using music as a type of biofeedback and reported improvement in head control in 3 of 5 study participants.15
Grasso et al16 investigated the benefits of recorded music used as an adjunct to daily chest physical therapy in infants and toddlers with cystic fibrosis. The authors assessed the child’s and parents’ enjoyment of chest physical therapy using a questionnaire. They found that children and their parents enjoyed chest physical therapy intervention significantly more when a treatment music tape was used during the session, as compared to no music or a familiar music tape. The researchers suggested that providing families of infants and toddlers with cystic fibrosis with prerecorded music tapes to use as an adjunct would assist them in establishing a positive routine for chest physical therapy.16
The issue of increasing the child’s comfort during physical therapy was explored by Rahlin et al17 in a single-subject research conducted with an infant with Erb’s palsy. This study investigated the effects of music on gross motor progress, the amount of crying during therapy, and parent satisfaction with physical therapy services. The authors used an A-B-A withdrawal single-subject design,18 with music introduced in the intervention period and no music played during baseline periods.17 The patient’s progress was assessed using the Toddler and Infant Motor Evaluation™ (TIME™).19 The amount of crying during therapy sessions was documented in a Crying Log, and parent satisfaction with physical therapy services was assessed with a questionnaire at the end of each of 3 study periods.17 The subject’s progress in the music intervention period increased on 3 of 5 primary subtests of the TIME™. The amount of crying decreased and parent satisfaction increased when music was played during therapy. The study results indicated a need for further research, with added methods of control for investigator bias, to be conducted with a group of infants and toddlers to generalize these findings to a wider patient population.17
The purpose of this research was to continue investigation of the effects of music on the amount of time infants and toddlers cried during physical therapy by using an A-B-A withdrawal multiple single-subject design. It was hypothesized that the amount of time that patients cry would decrease when music is used during therapy sessions, as documented in a Crying Log.
An A-B-A withdrawal multiple single-subject design was used, with music played during therapy in the intervention period. No music was introduced during the baseline periods. There were 6 physical therapy sessions in each period of the study. All therapy sessions were videotaped and then analyzed by 2 research assistants (RAs) unaware of the purpose and design of the study. They documented the amount of crying the children exhibited during each therapy session in a Crying Log (Appendix A).
We initially planned to recruit 10 to 15 participants as a sample of convenience representing male and female infants and toddlers, aged 4 to 33 months, receiving physical therapy at home through a private home care pediatric physical therapy practice and at the Outpatient Pediatric Therapies Department of Advocate Lutheran General Hospital in Park Ridge, IL.
Children who received physical therapy more frequently than 1 time per week were included after participating in physical therapy for at least 4 weeks. Children who were seen by the therapist 1 time per week were included after having received physical therapy for at least 6 weeks. This was to allow children to habituate to their therapist before data collection. The projected time for receiving therapy in the program had to be at least 6 months to be included in this study. Children whose gross motor skill levels were that of a beginning walker or below were included. The study participants were to have or be at risk for a developmental disability based on one of the following criteria: (1) a specific diagnosis that made the child eligible for early intervention services or (2) a gross motor delay of 20% or greater demonstrated by the child at the initial physical therapy evaluation. The percentage of delay was determined using one of the following standardized assessment tools: the Alberta Infant Motor Scale,20 Peabody Developmental Motor Scales II,21 or the TIME™.19
Children were excluded from the study if, at the time of entry, they had a known hearing problem, a chronic ear infection that affected their hearing, a documented sensory integration disorder, or a sensory impairment that necessitated continuous use of music during therapy. Children who never cried during therapy sessions were also excluded.
Planned Control for Extraneous Variables.
The following means of control for extraneous variables were planned for this study. If a participant developed a short-term (7–10 days) ear infection during the study, this would be documented and discussed as a possible confounding variable. If a participant developed a long-term ear infection or a hearing problem, he or she would be withdrawn from the study.
If a participant was taking medication that might influence his or her mood, 3 possible case scenarios were considered. First, if the child’s medication had an effect on his or her mood throughout the study, this was not seen as a problem because the study participants served as their own controls. Second, if a child started taking such a medication during the study, the investigator needed to contact the child’s physician to discuss possible side effects. If the medication was very likely to affect the child’s mood and if the side effects (eg, excessive calming, excessive excitability) were not likely to resolve quickly (within 1 week), the participant would be withdrawn from the study or this confounding variable would be documented as a limitation of the study. Third, if, according to the child’s physician, the side effects of the medication were short term and would disappear within 1 week, this would be documented and discussed as a possible confounding variable.
Eleven children, 6 girls and 5 boys, aged 5 to 33 months at the time the study was initiated, were recruited. Because of unexpected staff changes, subject recruitment from Advocate Lutheran General Hospital was difficult, and 10 of 11 participants were obtained from the private pediatric home care physical therapy practice of the first author. Table 1 summarizes demographic and diagnostic information for the study participants.
Nine children, 4 girls and 5 boys, completed the study. Participant 2 withdrew from the study after her third birthday because she was no longer eligible for early intervention services, and her parents decided not to continue with private physical therapy as they had planned. Participant 7 was withdrawn from the study by the principal investigator. Before her recruitment into this study, she demonstrated significant irritability and crying during physical therapy sessions. However, after the first session was conducted in the baseline period of the study, this child stopped crying. Her calm behavior continued through session 4 when, based on the exclusion criteria, the decision to discontinue her participation in this research was made in collaboration with her parents.
It was initially planned that there would be up to 4 or 5 physical therapists involved in treating children and collecting data during this study, including the authors of this article and 2 or 3 other therapists employed by the Outpatient Pediatric Therapies Department at Advocate Lutheran General Hospital. However, for the reasons discussed earlier, only 2 investigators participated in this research.
The Crying Log was developed to document the number of minutes that the participants cried throughout the study and to record music selection during physical therapy sessions in the intervention period (Appendix A). A modified definition of crying given by Riccilo and Watterson22 was used for the purposes of this research. Crying was defined as “a condition when the child’s eyes were partially or fully closed, and when a characteristic facial grimacing and vocalization were present, with or without tears.”17
Music Session Log.
The Music Session Log (Appendix B) was developed to record the length of the session in minutes, the type of music used, and the number of minutes the music was played.
Video and Audio Equipment.
Two 8-mm camcorders with tripods were used to record physical therapy sessions. To replay the video recordings, a camcorder was connected to a television set, and a remote control was used to start and stop the video tape. Two CD players were used to play music in the intervention period of this study.
The RAs used an electronic stopwatch to measure the time participants cried during therapy sessions as recorded on video.
Approval of this study was initially obtained from the Institutional Review Boards at Midwestern University, Downers Grove, IL, and at Advocate Lutheran General Hospital, Park Ridge, IL. Later, approval was also obtained from the Institutional Review Board at Rosalind Franklin University of Medicine and Science in North Chicago, IL. Each participant’s parent signed and received a copy of the informed consent.
Therapy Sessions, Music Selections, and Video Recordings.
There were 6 therapy sessions in each of the study periods. During the baseline periods, there was no music or singing allowed during the therapy sessions. Two types of music were introduced into therapy sessions during the intervention period. As in the previous research,17 soothing, instrumental classical music and recorded or therapist-sung children’s songs were used. The therapist decided which type of music to play depending on the participant’s responses at each session as described elsewhere.17 Music to which the child responded with the least amount of crying was played. If the children cried, the therapist attempted to comfort them by distracting them with a toy, using a pacifier, talking to them, picking them up, rocking them, or letting the parents comfort them. These calming techniques were used regardless of the study period. If the child stayed indifferent to both types of music but was not crying, the therapist played the children’s songs. If the participant stayed indifferent to both types of music and continued to cry, the therapist played the soothing instrumental music for up to 10 minutes. If the child was still crying, the therapist stopped playing music.
All therapy sessions were videotaped. The treating therapist documented the type(s) of music played at each therapy session by recording a specific code on the videotape label. The code included the participant’s number, the therapist’s initials, the session number, and the letter code for the types of music used. In addition, the therapist filled out a Music Session Log after each therapy session in the music intervention period of the study.
RAs and Videotape Scoring.
The RAs were Doctor of Physical Therapy students. Several pairs of RAs processed the video recordings over the span of 5 years of data collection and analysis (2003–2008). Two RAs processed the video recordings for variability of intervention as described in the next section. Three pairs of RAs scored the tapes for the amount of crying, but only 1 pair was involved in processing data for each study participant at any given time. Before processing the study tapes, the intrarater and interrater reliabilities of the Crying Log were established for each pair of the RAs.
The RAs scored the videotapes continuously throughout the data collection procedure, using the Crying Log and a stopwatch. The RA transferred the videotape label information to the Crying Log, then replayed the videotape, and used the stopwatch to measure the amount of time that the child cried during the therapy session. The stopwatch was started when the child started crying and stopped when he or she stopped crying. The total number of minutes that the child cried within a therapy session was calculated by adding the number of minutes he or she cried during all episodes of crying in that session and recorded in the Crying Log. The total amount of crying was also calculated in seconds for the ease of plotting these data on graphs.
Control for Investigator Bias
The investigators were also the treating therapists who were aware of the stage of the study at each point in time. However, the RAs who were processing the video recordings for the amount of crying time were unaware of the study purpose and design. One of them scored only the videotapes that were recorded during the music period of the study, whereas the other scored the rest of the tapes. When transferring the coding information from the videotape labels to the Crying Log, the RAs were unaware of the meaning of the code designated for the types of music used during the study.
Because the treating therapists were aware of the study design, the variability of intervention in the baseline and music intervention periods of the study could be another source of investigator bias. This was addressed by having 2 RAs perform the analysis of the videotapes for variability of intervention. One RA recorded the intervention techniques used by the therapist in baseline periods of the study, whereas the other RA recorded this information for the same study participant treated during the music intervention period. Then the intervention recording forms were compared for changes in intervention between the study periods to determine whether the therapist used similar intervention techniques in baseline and intervention periods of the study, therefore, assessing whether the investigator bias might have affected the study results.
Before initiating the scoring procedure for this study, pilot projects were conducted on 10 subjects to establish the intrarater and interrater reliabilities for the RAs in scoring data and recording the results in the Crying Log. A Pearson r was used to analyze the association between the 2 sets of Crying Log data, and paired t tests were used to analyze the difference between them.
The celeration line approach and descriptive statistics were used to analyze Crying Log data to compare the amount of crying among the 3 periods of the study. Descriptive statistics included calculating measures of central tendency and variability for Crying Log data collected across the 3 periods of the study. Music Session Log data were analyzed by calculating the percentage of minutes of crying based on the amount of time music was played during therapy sessions in the intervention period.
Music Selection and Length of Playing Time
The types of music used during 6 physical therapy sessions in the music period of the study are shown in Table 2. Based on the Music Session Log data, music was played for the entire session time in 70% of all music sessions. In 30% of all music sessions (16 sessions total), music was stopped because it did not seem to help the child’s crying. In 13 of these 16 sessions, music was stopped in the last 5 to 15 minutes of a 45- to 60-minute session. In 2 instances, music was stopped approximately halfway into the session. In 1 instance, it was stopped at 20 minutes, and the session had to be cut short at 30 minutes because of the infant’s crying.
Crying Log Data
Reliability of the Crying Log.
The intrarater and interrater reliabilities of the Crying Log for each pair of the RAs were established on 10 subjects with Pearson r ranging from of 0.98 to 1.00 and from 0.94 to 0.99, respectively. The t test results showed no statistical difference between the analyzed data sets (2-tailed p > 0.05).
Celeration Line Approach.
The celeration lines were constructed for each participant (Figs. 1–4) as described by Portney and Watkins.18 First, all data points representing the crying time during each therapy session were plotted on a graph separated into 3 study periods by vertical lines. Then a best-fit or split-middle line was drawn for the data points in the first baseline period, with equal number of data points positioned above and below the line. This was the celeration line that demonstrated the data trend for this study period. To determine whether there was a difference in data between the baseline and intervention periods of the study, this line was continued into the intervention period, and the data points above and below the line in the intervention period were counted. A greater number of data points below the line indicated a decrease in crying time in the music intervention period of the study (Fig. 1), whereas equal number of points above and below the line would have indicated that there was no difference in data between the study periods. Statistical significance was determined by using a binomial table to assess the number of data points required to fall above or below the celeration line for the results to be significant. The same process was repeated to compare data in intervention and second baseline periods (Figs. 1–4).18
Visual analysis of Crying Log data using the celeration line approach produced variable results. Four participants (1, 4, 8, and 9) cried less in the music period of the study compared with the first baseline period and subsequently cried more when the music intervention was removed in the second baseline period (Fig. 1). However, only for participant 1, these results were statistically significant for both celeration lines (Fig. 1A). For participants 8 and 9, the results were statistically significant for an increase in crying when music intervention was removed (Fig. 1D, E).
On the graph constructed for participant 4 (Fig. 1B), 2 outlier points were found, one in the first baseline period and the other one in the music period of the study. In the first baseline period, during session 1, the child cried considerably more than in sessions 2 through 6, when the data achieved relative stability. This outlier data point significantly affected the slope of the first celeration line. Session 9 in the music period also showed a peak of crying time (Fig. 1B). When the outlier points were removed from the graph, it appeared that participant 4 cried less in the music period and more in the second baseline period (Fig. 1C), but these results were not significant based on data analysis using a binomial table.
Participants 6 and 10 cried significantly less in the music period compared with the first baseline period of the study as determined by all data points in the music period being below the first celeration line (Fig. 2). However, the amount of crying did not change in the second baseline period as compared with the music period (Fig. 2).
Participants 3 and 5 cried more in the music period as compared with the first baseline period of the study and subsequently cried less in the second baseline period when music intervention was removed (Fig. 3). These results were statistically significant for both celeration lines. Participant 11 cried more in the music period of the study, and crying time continued to increase in the second baseline period (Fig. 4). The results were statistically significant only for the second celeration line.
Means, standard deviations, and ranges for the amount of time each child cried across the 3 periods of the study are summarized in Table 3. Based on the mean crying times and on the minimum amount of time cried, 7 of 9 participants cried less when music was used during therapy compared with the baseline periods of the study.
Comparison of Results Based on Celeration Line Approach and Descriptive Statistics.
Table 4 shows the effects of music and of subsequent music removal based on the celeration line approach compared with descriptive statistics. The same results were achieved using the 2 data analysis approaches for participants 1, 8, 9, and 11. Similar results were obtained for participants 6 and 10, and different results were demonstrated for participants 3, 4, and 5.
Variability of Intervention.
The analysis of videotapes for variability of intervention showed that the treating therapists had used similar intervention techniques in baseline and intervention periods of the study. As expected, when the child progressed in his or her motor skills, some modifications to specific therapeutic activities were made throughout the study, but similar activities were used by the therapists in the last 2 sessions of the first baseline period and in the first several sessions of the music intervention period. The same was true for the last 2 sessions of the music period and the first several sessions of the second baseline period. Calming techniques were used by the therapists and the caregivers as needed across the 3 periods of the study.
This study examined the effects of music on the amount of time that 9 infants and toddlers cried during physical therapy sessions. On the basis of the celeration line analyses, most participants showed a positive response to the use of music during therapy. This finding supported the study hypothesis and was consistent with previous research.17 Based on the current results, the following issues need to be discussed: research design, control for extraneous variables, control for investigator bias, the variability of results obtained using the celeration line approach and descriptive statistics, and generalization of the results.
Romeiser Logan et al 23 identified the A-B-A withdrawal multiple single-subject design as level IV evidence for single-subject research, after randomized, controlled, alternating treatment18 and concurrent and nonconcurrent multiple baseline18 research designs (levels I–III). This design was selected for this study for the following reasons: (1) it allowed the investigators to replicate previous research conducted with 1 subject17; (2) according to Portney and Watkins description of this design,18 withdrawing the intervention in the second baseline period helped to demonstrate that changes in the participants’ behavior were related to the use of music in the intervention period; (3) application of a concurrent multiple baseline design18,23 was not feasible because of limited availability of participants for concurrent enrollment in the study; and (4) use of a nonconcurrent multiple baseline design would have required dropping a participant from the study if baseline data were found to be too variable.18,23 This was not possible because of the reasons discussed below.
According to Portney and Watkins,18 presetting the length of the study periods in a single-subject design helps to control for the effect of maturation. However, it is recommended that the baseline phase be extended until data stability is achieved. Similar to previous research,17 the length of the study periods was preset, and, therefore, it was not possible to know whether data stability was achieved by the end of the first baseline period. On the one hand, this may be considered a limitation of this study. On the other hand, this design replicated previous research17 and maintained control of investigator bias, because the RAs who scored the videotapes after the data collection was completed were unaware of the study purpose and design.
Romeiser Logan et al23 proposed a quantitative method for evaluating quality of single-subject designs. They rated a single-subject research design as methodologically strong if it could be assigned 11 to 14 points by answering 14 specific questions. They reported a 4-rater agreement of 75% in evaluating 3 articles scored on this scale. Our evaluation of the current study using this method yielded a score of 12 points, which suggests that it may be considered strong based on the criteria suggested by Romeiser Logan et al.23
Control for Extraneous Variables and Investigator Bias
Planned control for extraneous variables was built into the study design as described in the Participants section. However, it was not possible to foresee and eliminate all extraneous variables related to history as illustrated by the study findings related to participant 4. As stated in the Results section, on the celeration line graph constructed for participant 4 (Fig. 1B), session 9 in the music period showed a peak of crying time, which was most likely due to an extraneous variable. The analysis of this child’s medical record revealed that this was possibly because the family had moved to a new apartment, and, per her mother’s report, the child had difficulty adjusting to the new environment and cried with all of her therapists during that time. In addition, the previous physical therapy session scheduled for the week before had been cancelled because of the participant having a fever due to teething. At session 9, she was chewing on her fingers, constantly crying and not responding to her mother’s and the therapist’s attempts to comfort her. Our inability to control for these extraneous variables may be considered another limitation to this study.
Maturation was another potential threat to internal validity of this research. It was anticipated that during this study, the participants would grow older, and their responses to the environment, including therapy sessions, might change. The withdrawal of music in the second baseline period of the study helped control for this effect.
Another anticipated effect related to maturation was that in the course of this research, children might habituate to the treating therapist and demonstrate improved behavioral responses as compared with the time of their entry into the study. This possible threat to internal validity was controlled by 2 methods: (1) the withdrawal of music in the second baseline period; and (2) the fact that according to the inclusion criteria, a child could be enrolled in the study only after being in therapy for at least 4 to 6 weeks to ensure the habituation to the therapist before participation in this research.
Because crying in infants is known to decrease linearly with age,24 another concern was that over the course of the study, the participants might mature enough to modulate their mood without the need to use external stimuli for calming. It was initially planned to investigate this possible maturation effect by analyzing study results for 3 age groups (<1 year, 1 year to 2 years 6 months, and >2 years 6 months). Because of the small number of subjects expected to be recruited in each of the age groups, the results of this analysis were not expected to be statistically significant, but certain trends in the participants’ responses to music across the age groups might be discovered. However, the results showed that the participants’ age did not seem to affect the study results.
Methods of control for investigator bias in this research are described in the corresponding section. RAs unaware of the study purpose and design processed the video recordings of the therapy sessions for the amount of crying time. Other RAs analyzed the video tapes for variability of intervention. The variability of intervention results described earlier suggests that the investigator bias did not have an affect on the study results.
Variability of Results
Results of this study were variable as evident from the celeration line graphs constructed for each participant (Figs. 1–4). Music used during therapy sessions appeared to have a calming effect on 6 of 9 participants, seemed to increase crying in participants 3 and 5, and did not have any effect on the amount of crying demonstrated by participant 11. The trends demonstrated by 11 of 18 celeration lines constructed for 9 participants were statistically significant (Figs. 1–4). Data analysis using descriptive statistics also produced variable results (Tables 3 and 4), suggesting that music had a calming effect on 7 of 9 participants, increased crying in 1 child, and made no difference for 1 participant. Measures of central tendency and variability added to the description of the study results23 and mostly supported the findings based on the visual analysis of data. Both methods of data analysis used in this study are considered acceptable for this research design.18,23 However, if these methods produce variable results, the celeration line approach should be considered a more accurate and acceptable method of data analysis for a single-subject research, which is characterized by taking repeated measurements of a target behavior.18 As such, 6 of 9 participants benefited from the use of music in this study; 1 showed no response, and 2 children seemed to respond negatively to music.
It is unclear why music had a negative effect on participants 3 and 5 compared with other children involved in this research. It is possible that music had an irritating effect on them or other factors besides music affected their behavior. Participant 11 had a seizure disorder that did not seem to respond well to medical management. Per his mother’s report, he had 1 to 2 seizures per day on most days during the study. He would always fall asleep after a seizure and then wake up unhappy and irritable. His mother also admitted that his sleep-wake cycle was not well established. When he was hungry, he cried until fed and then would fall asleep. In addition, he sometimes fell asleep during therapy without being fed. It is possible that music did not have any effect on this child’s behavior, because his irritability was not just related to the therapy process but also to his physiological needs, such as feeding and sleeping.
According to Brazelton,25 there may be a variety of reasons for infants and children to cry, including discomfort, pain, fatigue, hunger, and boredom. In addition, crying may result from overstimulation or may be used by the child as a limit-testing or an attention-seeking behavior. Finally, it may be a part of disorganization described by Brazelton25 in conjunction with spurts in development. This may explain the variability of results demonstrated in this research. Because behavior, and specifically crying, is multidimensional, it is possible that it may not be successfully controlled by a single specific intervention, such as music. Although music may help to decrease crying caused by some of the stressors listed earlier, it may have little or no effect on crying in other situations.
When physical therapists take into consideration the need to maximize a patient’s comfort along with maximizing his or her rehabilitation potential, they give careful thought to the needs of the whole patient and thus improve the quality of care they provide. Therefore, it seems that there is a need for deeper understanding of the behavior of infants and toddlers during therapy. This may be accomplished by assessing multiple aspects of therapy-related behaviors and by investigating whether music and other interventions may have a positive effect on these behaviors, ultimately leading to a better outcome of a therapy session.
Generalization of Results
This study replicated a previous single-subject research study17 with multiple participants to strengthen the external validity of the findings. Additional methods of control for investigator bias discussed earlier were included in the current study as compared with previous research.17 Because 6 of 9 children who participated in this project responded positively to the use of music during their physical therapy sessions, these results may be generalized to infants and toddlers with similar characteristics as the study participants.
The results suggest that (1) infants and toddlers with or at risk for developmental disabilities who do not have hearing problems may benefit from the use of music during physical therapy to reduce crying; and (2) infants and toddlers who are teething, who have difficulty tolerating changes in their environment, and who have seizures and state regulation problems may be less responsive or respond differently to the use of music during therapy sessions. Physical therapists and other professionals who work with children aged birth to 3 years, as well as parents who perform home exercise programs with their infants and toddlers, may want to consider using music as a calming technique. However, because behavior is multidimensional, although music may benefit many children, its use may be irritating or irrelevant for others. Therefore, further research is needed to investigate the effects of music on different aspects of infant and toddler behavior during therapy.
The authors would like to thank the study participants and their families. The authors also thank the following research assistants who spent many hours analyzing the video recordings of therapy sessions for this research when they were Doctor of Physical Therapy students at Midwestern University and Rosalind Franklin University of Medicine and Science: Michelle Leach, PT, DPT; Edyta Puto, PT, DPT; Nicole Senkow, PT, DPT; Ayesha Futrell, PT, DPT; Rachel Serek, PT, DPT; Sonia Paul, PT, DPT; and Amber Gore, SPT.
© 2009 Lippincott Williams & Wilkins, Inc.