Introduction

Evidence shows that treatments accompanied by mechanical ventilation in an intensive care unit (ICU) cause anxiety for patients because the ICU induces both physical and psychological stressors [1–3]. For those undergoing mechanical ventilation, the induced stressors may be due to their inability to breath independently, communicate effectively, or rest normally [4]. Although many drugs such as sedatives, opioids, and neuromuscular blocking agents are effective in reducing patient anxiety [5], clinicians should consider the possibly serious adverse effects [6, 7]. As a result, searching for alternative methods without serious adverse effects could be useful for clinicians, especially for those providing care in an ICU.

Music intervention and aromatherapy—which are complementary and alternative medicines—have been used by clinicians as part of a holistic approach to decrease preoperative anxiety [8]. Music intervention “functions as a distractor, diverting attention away from a negative catalyst and focusing awareness on soothing stimuli” [9]. Therefore, listening to preferred music is often used in a certain period of time via headphones. Aromatherapy uses plant oils that have comforting smells. Many commonly used plant oils include lavender (Lavandula angustifolia), rose (Rosa damascena), bergamot (Citrus aurantium), sandalwood (Santalum album), or rosemary (Rosmarinus officinalis) [10]; commonly used methods for aromatherapy include inhalation, injection, applying oil topically to the skin, and massage.

The major benefits of using music intervention or aromatherapy are low cost and trivial adverse effects [6]. Moreover, the effects of music intervention on reducing anxiety of ICU patients have been supported in numerous clinical trials [11–13]; however, most music intervention studies [6, 11–14] examined only the immediate post-treatment effects. In other words, we do not know the follow-up effects of music intervention. Probing this topic may provide clinicians helpful information to make clinical decisions.

The effects of aromatherapy in reducing anxiety have been reported for general surgical patients before or after operations [15–18], for school teachers with generalized anxiety [19], for mothers of children with attention deficit hyperactivity disorder (ADHD) [20], and for ICU patients [21]. However, we did not know whether the effects of aromatherapy to reduce anxiety are equivalent to those of music therapy. Knowing the treatment effects between music and aromatherapy may provide clinicians useful information to design appropriate interventions for ICU patients. Moreover, we postulated that music intervention has direct intervention effects on reducing anxiety as demonstrated by previous findings [6, 11–14], while aromatherapy may have longer effects than music intervention because the smells may last longer than the sound.

The primary purpose of this study was investigating the treatment effects of music and aromatherapy interventions on reducing anxiety for ICU patients undergoing mechanical ventilation. To reduce anxiety for patients, we used a 30-min intervention—the same duration as other studies on music intervention [11–13] and aromatherapy [18]—and examined the post-treatment effects and follow-up effects for up to 30 min. We hypothesized that patient anxiety would be decreased after receiving a single 30-min session of music intervention or aromatherapy.

Methods

Participants and ethical considerations

We recruited participants from an academic medical center with 1105 beds in Taichung City, Taiwan. All participants received the treatments (either music intervention or aromatherapy) or usual care in private isolation rooms in the medical and surgical ICU. The ICU had 47 beds with 9 private isolation rooms. Although all participants and the research nurses were blinded to the randomized group assignment, the received treatments were apparent when the treatments were given.

The eligible participants fulfilled the inclusion criteria: (1) 18–85 years of age; (2) conscious and mentally clear to voluntarily participate and understand the study purpose; (3) able to understand Mandarin Chinese, Taiwanese (Southern Min), or both; (4) able to communicate using body gestures, writing, or both if necessary; and (5) admitted to the ICU for more than one day. Participants with the following conditions were excluded: (1) hemodynamic instability; (2) treated with continuous intravenous analgesics or sedatives; (3) treated with cortisol drugs; (4) disliked the smell of aroma oil; and (5) allergic to aroma oil [4, 22, 23].

The Institutional Review Board and the Human Research Ethics Committee of the Chung Shan Medical University Hospital (IRB: CSH-2013-A-018) approved the study protocol, which included a provision for withdrawing participants whose health deteriorated during the study. All participants provided a written informed consent.

Instruments

Primary endpoint: patient-reported anxiety

Subjective measures (VAS-A and C-STAI) were used for assessing anxiety levels. The VAS-A asked each participant to evaluate his/her anxiety level using a 100-mm scale with the following instruction: “Please use the 100-mm scale to rate your current anxiety level. The score of 100 on the right side indicates the highest level of anxiety; the score of 0 on the left side indicates the lowest level of anxiety.” Moreover, the test–retest reliability (r = 0.44) and convergent validity (r = 0.60 with STAI) of the VAS-A have been supported [24]. However, it should be noted that both correlations (test–retest and the convergent validity) were not very promising. The relatively low test–retest reliability can be explained by the nature of the anxiety and the single-item scale. Anxiety is easily influenced by transient and situational factors [25, 26], and it could be easily changed across time. Also, the low test–retest reliability has been demonstrated in other single-item scales [27]. In terms of the moderate correlation between the VAS-A and the STAI, we considered that it is acceptable because we assumed that they measured different aspects of the anxiety. In addition, the correlation between the VAS-A and the STAI was higher than the correlation between the VAS-A and a depression measure, Patient Health Questionnaire (r = 0.30; Fisher’s r-to-Z = 3.32; p < 0.01) [24]. Also, the responsiveness of the VAS-A has been supported: The VAS-A can detect changes in anxiety via a significant pre- to during-stressor in VAS-A scores (F [1, 48] = 25.13, p < 0.001) [24]. Therefore, we concluded that using VAS-A as one of our primary measures was acceptable.

The C-STAI includes state and trait anxiety; only state anxiety was used in this study. The state anxiety consists of 20 items; each item was rated on a 4-point Likert scale with a lower score representing a lower level of anxiety. The state anxiety of C-STAI had adequate reliability (α = 0.90), test–retest reliability (r = 0.74), and concurrent validity [28, 29]. Moreover, α values of the C-STAI were satisfactory in our study (α = 0.83 for the baseline measure, 0.78 for the post-test measure).

Secondary endpoint: blood pressure, heart rate, and breathing rate

We used an HP/Philips/Agilent M1205A system, which was calibrated twice per year, to automatically monitor and record blood pressure, heart rate, and breathing rate. In addition to the commonly used systolic and diastolic blood pressures (SBP and DBP), we used mean arterial pressure (MAP) to demonstrate overall blood pressure.

Justification for sample size

G*Power 3.1.5 [30] was used to estimate the sample size with the effect size of Cohen’s d at 0.8 on a two-tailed independent t test. In addition, the following default setups were used for sample estimation: a type I error of 0.05, an allocation ratio of 1 for the two groups, and a power of 0.95. We used an effect size at 0.8 because Han et al. [12] reported a large effect size (their Cohen’s d was ca. 1.67) and Lee et al. [14] reported a small effect size (their Cohen’s d was ca. 0.04) between their music intervention and control groups. Using the effect size at 0.8, which is between the two, seemed appropriate. Also, Wu et al. [20] reported an effect size similar to 0.8 (their Cohen’s d was ca. 0.73) between aromatherapy and control groups. The estimated sample size for each group was 42 with a total of 126. We assumed that 20% of the patients would decline to participate or drop-out; therefore, we invited 160 patients to participate, and finally had 132 retained in our study. The number of retained participants suggested our sample size was acceptable.

Interventions

A caution note—“Do not disturb during the period of music intervention”—was posted outside the room of all participants whether he/she was in either group. Neither routine care nor visitors were permitted during the intervention period (or the resting period in the Control group). A research nurse—who was unfamiliar to the patients in either group—also silently sat at the bedside to care for the patients’ physical needs when necessary. In addition, light and temperature were set up in the same conditions for all groups.

Music intervention

A 30-min session for each participant to listen to music was provided in the Music group between 4:00 and 4:30 PM. Patients were not given other treatments for their illnesses during the music intervention session. The following music programs were provided for participants to select one for intervention according to their preference [11, 14]: Western classical music (e.g., Erik Satie’s Trios Gymnopedies, Mozart’s Piano Concerto no. 26), Chinese classical music (e.g., bamboo flute, rain, and tears), music of natural sounds (e.g., Sylvan Spa [31]; Relax Your Mood [32]), and religious music, including Buddhist [33] and Christian [34].

The characteristics of the music included a slow beat (60–80 beats per minute) corresponding to a normal heart rate, which helped the patients relax [35]. We used an MP3 player placed at the patient’s bedside and delivered the music through headphones. During the intervention, all participants lay on a bed with the lights low and room temperature set at 26 °C, and were free to ask the research nurse to help them adjust the volume.

Aromatherapy

The day before receiving aromatherapy, a research nurse placed one drop of lavender essential oil on the medial forearm of participants assigned to the Aromatherapy group, and observed the reaction of the participants’ skin for the following 20 min. The nurse declined participation of those who had uncomfortable feelings (e.g., red or itchy). Finally, all participants in the Aromatherapy group did not have uncomfortable feelings. Although five participants declined to participate, they declined because of family issues instead of uncomfortable feelings. On the day of receiving aromatherapy, a research nurse who had completed the training of aromatherapy massage let the participant recline in a lateral position. Then she placed the carrier oil on the participant’s four limbs and entire back of the participant, followed by applying 2% lavender essential oil (20 cc; dōTERRA), and massaged the participant’s back of the participant using the technique of stroking for 5 min. Afterward, the research nurse positioned the participant in a supine position for 5 min. The participant was then laid on a bed in any preferred position for the rest of 20 min with the lights low and the room temperature set at 26 °C.

Control group

Participants in the control group had a 30-min rest between 4:00 and 4:30 PM, the same time that the intervention groups were listening to their music intervention or receiving their aromatherapy.

Procedures

ICU head nurses first identified the eligible patients (n = 160), and then, one research nurse randomized the eligible patients into three groups with the allocation ratio as 1. Excel was used to generate the random number for assigning eligible patients into the experimental or control group. Specifically, we used the RANDBETWEEN function with numbers between one and three. A random number of one was assigned to the Control group, of two to the Music group, and of three to the Aromatherapy group. The allocation concealment was accounted for when we used the RANDBETWEEN function. We did not generate all the random numbers in advance, but only when an eligible participant was enrolled. Hence, we did not know in advance which treatment the next person would get. Afterward, the head nurses explained the study purpose to eligible ICU patients and asked for a signed informed consent from those willing to participate. Originally, 56 patients were invited into the Music group and 52 into the Aromatherapy and Control groups each. However, 15 patients in the Music group declined to participate or dropped out; 5 in the Aromatherapy group declined to participate; 8 in the Control group dropped out. Therefore, 132 patients provided signed informed consent (Fig. 1).

Fig. 1
figure 1

Flow diagram of the study. VAS-A visual analogue scale-anxiety, C-STAI State-Trait Anxiety Inventory-Chinese version, HR heart rate, RR respiratory rate, SBP systolic blood pressure, DBP diastolic blood pressure, MAP mean arterial pressure

Full size image

The measures for primary endpoint were collected at 4:00 PM (before intervention) for baseline measures and at 4:30 PM (after intervention) for post-test measures. Blood pressure, heart rate, and breathing rate were monitored from 4:00 to 5:00 PM, and we measured the values every 10 min (e.g., 4:00, 4:10, 4:20). The baseline data and post-test data were collected just before and after the intervention, respectively. In addition, two trained research nurses—blinded to the research purposes based on standardized procedures—took blood samples, measured the C-STAI, VAS-A, and recorded the blood pressure, heart rate, and breathing rate of all participants. The study lasted from August 2013 through December 2014.

Data analysis

All the analyses were completed using SPSS 17.0 for Windows. We used frequencies to demonstrate the demographic and clinical characteristics, and used χ 2 tests to examine the differences in the demographic and clinical characteristics between the three groups. One-way analysis of variance (ANOVA) was used to detect baseline differences with Bonferroni adjustment. Specifically, we first used seven ANOVAs to test each baseline measure (VAS-S, C-STAI, heart rate, breathing rate, SBP, DBP, and MAP). Then, we did three post hoc analyses for every two groups (Music vs. Control; Aroma vs. Control; Music vs. Aroma). Analyses of covariance (ANCOVA) were used to examine the post-test effects, which adjust for the effects of age, sex, and baseline measures. The analysis process of the ANCOVA was the same as the process of the ANOVA (i.e., separate tests for each dependent variable and post hoc tests). Two sets of ANCOVA were used: The first set tested group effects on post-test measures at the end of the interventions (i.e., after receiving 30 min of interventions); the second set tested at the end of the follow-up (i.e., 60 min after baseline measures). Bonferroni adjustments were also used in the ANCOVA.

We used several generalized estimating equations (GEEs) to analyze the effects of both interventions compared with the control group because of the nature of our data in the measures of physiological indicators. That is, heart rate, breathing rate, and blood pressure were measured repeatedly. Five GEE models were separately analyzed. The five dependent variables were heart rate, breathing rate, SBP, DBP, and MAP. Age and gender were included as the confounders in the models, group (music intervention, aromatherapy, and control group) and time of measurement (a total of seven times) as the main factors, and an interaction term (group × time). Control group and the baseline measurements were the references in the GEEs. We were especially interested in the interaction term because we believed that the intervention effects were very likely to differ by time. A positive coefficient in the interaction term indicated that the direction of the group comparison and that of time comparison were in the same direction: For example, music therapy group compared with control group, 10 min after baseline measure compared with baseline measure. We hypothesized that intervention groups would have lowered values in physiological indicators, and that the follow-up measures would also have lowered values; therefore, we anticipated observing positive coefficients for the interaction terms.

Results

Participant characteristics

The participants’ characteristics, including demographic and clinical characteristics, were similar across the three groups (Table 1). Although the Music group seemed to consist more of male participants (53.7%) than the other two groups (31.9 and 34.1%), the differences were not statistically significant (p = 0.08).

Table 1 Participant characteristics
Full size table

Baseline and post-test measures after receiving 30 min of interventions

The post hoc analyses of the ANOVA indicated no significant differences among the three groups in VAS-A, C-STAI, heart rate, breathing rate, SBP, DBP, and MAP at baseline (Table 2). After adjusting for age, sex, and baseline anxiety, the post hoc analyses of the ANOCVA showed that the Music group reported a significantly lower level of self-reported anxiety than the Control group (VAS-A score: 49.56 ± 8.09 vs. 55.94 ± 9.27, p < 0.001; C-STAI score: 2.62 ± 0.23 vs. 2.71 ± 0.18, p = 0.001). The Music group had significantly lower heart rate than the Control group (75.53 ± 9.98 vs. 79.71 ± 8.15, p < 0.001), lower SBP (122.06 ± 14.49 vs. 126.16 ± 14.19, p < 0.001), and lower MAP (83.20 ± 8.57 vs. 84.35 ± 9.13, p = 0.03).

Table 2 Comparison of pre- and post-measures between groups (n = 132)
Full size table

The post hoc analysis of the ANCOVA showed that the Aromatherapy group was significantly more improved than the Control group: VAS-A (52.81 ± 8.34 vs. 55.94 ± 9.72, p = 0.015), heart rate (77.01 ± 6.24 vs. 79.71 ± 8.15, p = 0.001). All other post-test measures (C-STAI, heart rate, breathing rate, SBP, DBP, and MAP) were not significantly different between the two groups (Table 2).

Post-test measures occurred at 60 min after baseline measures, while interventions ended at 30 min after baseline measures

Figure 2 indicates the trends of heart rate, breathing rate, and blood pressure for the three groups from baseline to the end of the follow-up (i.e., 60 min after baseline measures); interventions ended at 30 min after baseline measures. After controlling for age, sex, and baseline measures, and compared with the Control group, post hoc analyses of the ANOCVA showed that the Music group had lower heart rate (mean difference = −2.92, p = 0.001) and lower SBP (mean difference = −4.41, p < 0.001); the Aromatherapy group had lower heart rate (mean difference = −5.27, p < 0.001), slower breathing rate (mean difference = −0.56, p = 0.03), lower SBP (mean difference = −4.98, p < 0.001), and lower MAP (mean difference = −2.13, p = 0.001).

Fig. 2
figure 2

Trends of heart rate, breathing rate, and blood pressure every 10 min. Intervention (i.e., music intervention and aromatherapy) was began after pre-test and ended at 30-min

Full size image

GEE results

GEEs corroborated data presented in Fig. 2, suggesting that the Aromatherapy group had a longer duration of effect than the Music group. For the Music group, all significant and positive interactions were found before the end of the intervention: 20 and 30 min for heart rate [B (SE) = 2.95 (0.63) and 2.85 (0.79), p < 0.001), SBP [B (SE) = 2.98 (0.91) and 2.45 (0.99), p = 0.001 and 0.013]; 20 min for MAP [B (SE) = 1.26 (0.50), p = 0.012]. In contrast, the Aromatherapy group had significant and positive interactions at 10–20 min after the end of the intervention. Specifically, significant interactions were found at 40 and 50 min after baseline measure for heart rate [B (SE) = 4.13 (0.54) and 1.79 (0.53), p = 0.001], SBP [B (SE) = 4.34 (0.76) and 2.94 (0.78), p < 0.001], and MAP [B (SE) = 2.37 (0.55) and 2.22 (0.52), p < 0.001] (Table 3).

Table 3 Generalized estimating equations (GEEs) testing the interaction effects of music intervention and aromatherapy
Full size table

Discussion

Because little is known about the follow-up effects of music and aromatherapy interventions on reducing anxiety for ICU patients, we wanted to investigate whether the reducing effects could last for a certain period of time. We also intended to explore the effects of aromatherapy on ICU patients. Our study seems to be the first to compare the effects between music intervention and aromatherapy in an ICU population. Generally speaking, if we focused on the immediate post-treatment effects, our results showed that both music intervention and aromatherapy had effects on reducing anxiety but on different levels—the effect of music intervention seemed to be stronger than that of aromatherapy.

The effects of our music intervention agree with previous findings [12, 36] that self-reported C-STAI scores decreased, and with those studies in which blood pressure and heart rate were decreased [4, 14]. In contrast, Cooke et al. [1] did not find the effectiveness of music intervention using the Faces Anxiety Scale; however, it has been proposed that the shorter treatment session (15 min vs. 30 min) may explain this result [6]. In addition to the immediate effects after music intervention, our results found that the effects can last for at least 30 min, though we had only physiological parameters to support the effects.

Our results also portrayed the effects of aromatherapy on reducing anxiety for ICU patients though the effects seemed less promising than that of the music intervention. Nevertheless, our findings agree with the effects of aromatherapy found in ICU patients [21, 37]. Karaman et al. [21] showed the effects using Beck Anxiety Inventory, and Cho et al. [37] and we demonstrated the immediate effects of anxiety reduction in VAS-A score and showed the effects in heart rate. Our results of the effect shown by VAS-A is consistent with the results of Braden et al. [15], who also used the VAS-A to measure anxiety and lavender essential oil for aromatherapy. The lack of significant difference in C-STAI anxiety scores between aromatherapy and control patients contradicts the findings of Ni et al. [18]. A possible explanation may lie in the difference between essential oils used with our aromatherapy participants receiving lavender essential oil, while Ni et al. [18] used bergamot essential oil. That is, different types of essential oils may have different treatment effects; however, future studies are needed to further probe this issue. Another possible explanation for the non-significant effects shown in our C-STAI score is the intervention duration: Karadag et al. [21] applied 15 days of lavender essential oil and found that ICU patients reduced their anxiety using the Beck Anxiety Inventory.

An interesting finding is that music intervention had stronger effect than the aromatherapy at the post-test measure; however, as we followed the effects of reducing anxiety, the heart rate and blood pressure of the Aromatherapy group kept decreasing, while those of the Music group were maintained. Finally, the Aromatherapy group had lower heart rate and blood pressure compared with the Music group at the end of the data collection (i.e., 60 min after the beginning of the interventions). We speculate that perhaps the effects of aromatherapy were delayed because the Aromatherapy group needed more time than the Music group to absorb the effects. That is, the music can be directly input into an individual’s mind; the essential oil needs time to be absorbed into the skin. However, we did not have additional evidence to support our speculation, and future studies are encouraged to investigate the potential mechanisms.

Limitations

There are some limitations in the study. First, our randomized controlled trial design was not robust because we did not use a placebo as a control for aromatherapy. However, we justified that using placebo is somewhat difficult. The frequently used placebo for aromatherapy is water [18, 22]; however, patients can easily detect whether they received aromatherapy or placebo due to the smell. Therefore, it seems hard for us to eliminate the placebo effects of aromatherapy. Second, the choices for music intervention and aromatherapy were restricted (only four types of music for the Music group, and only lavender essential oil for the Aromatherapy group). Therefore, patients may not have chosen their most preferable music or oil, and the effect of music intervention and that of aromatherapy might have been diminished. Third, all participants were recruited from the same hospital, and the generalizability of our results might be restricted because the participants might share similar demographic characteristics. Fourth, participants in the Control group may have had increased anxiety because they wore a noise-canceling headphone; however, we tried to diminish the effects using statistical methods (i.e., ANCOVA that accounts for possible confounders). Fifth, although the follow-up effects of music intervention and aromatherapy were supported by physiological parameters, we did not know whether the subjective measures have similar effects because both VAS-A and C-STAI were conducted before and immediately after treatment. Sixth, we randomly assigned the eligible patients into different groups before seeking their consent to participate. This approach could lead to an unbalanced sample size between groups, and result in insufficient statistical power. Fortunately, our sample sizes were nearly equal among the three groups; thus, the limitation did not seriously influence our results. Finally, although we provided the information of psychometric properties for both the VAS-A and the C-STAI based on previous studies, it should be cautioned that we did not test the reliability and validity of both measures in advance. Because reliably and validity vary depending on the goal of the testing procedure and the sample characteristics, we were unable to fully ensure that the VAS-A and C-STAI had excellent psychometric properties.

Relevance to clinical practice

ICU patients undergoing mechanical ventilation may reduce their anxiety through a 30-min session of music listening or aromatherapy. Evidence from objective measures (heart rate and blood pressure) and subjective questionnaires (C-STAI and VAS-S) both support the effects of anxiety reduction. Moreover, the effects were immediate and can last for 30 min or longer. Clinical nurses are encouraged to use music as an intervention to facilitate a healing environment and potentially reduce anxiety among ICU patients using interventions that are low cost, non-invasive, and simple to administer. While aromatherapy has promising results, it requires additional training. We therefore suggest that clinical nurses should learn the technique and apply it when necessary.

Conclusion

Among mechanically ventilated ICU patients, both a music intervention and aromatherapy can reduce anxiety and the effects can last for up to 30 min after cessation of treatment. Clinicians in critical care may apply both treatments as an alternative to drugs.