Received 30 December 2015; accepted 2 February 2016; published 5 February 2016
2.1. Trial Design and Participants
A study with simple randomization was conducted in a large university hospital in Denmark that serves 1.2 million people. Patients were included in this study if they were offered open surgery for gynecological cancer, could communicate in Danish, and accepted the ward standard of anesthesia without anxiolytic premedication. Moreover, participants should have an adult close relative who agreed to accompany the patient to the operating ward on the day of surgery, if necessary. Exclusion criteria were permanent mental disability and those under the age of 18. No changes were made after commencement.
2.2. Interventions and Standard Care
All the participating nurses were educated in Denmark and had a bachelor in nursing minus three months. No special training was performed in regard to this study. They all worked with gynecological cancer care and had the intention to support patients in the best possible way during the perioperative period. They used their resources and competences in interaction with individual patients; i.e. they were present, attentive and helpful, and chatted at the patient’s convenience. Moreover, they used additional anxiety-reducing strategies appropriate to the immediate situation: informed (provided factual information), predicted (predicted a better outcome than what the patient expected), empowered (gave information about a matter of course before explaining options available for action, and eventually encouraged to take action), and/or distracted (talked about something other than the threatening situation).
Prior to accept of participation all patients were informed about the different support combinations and how these potentially could benefit the patients.
In all groups: a hospital porter transferred the patient in bed from the ward to the operating ward. Five to ten minutes later, a nurse anesthetist came and stayed with the patient throughout the patient’s remaining period on the ward an in theatre. The nurse anesthetist had a list of questions to put to the patient, inserted a drip and assisted the anesthesiologist in inserting, if appropriate, an epidural catheter. The operating nurse and the surgeon would stop by to ask some questions, which is required by law, and would leave. In the operating theatre, the operating nurse and the nurse anesthetist would help the patient to be warm and to lie in a way that would avoid pressure sores. All those involved stayed with the patient until she was heavily sedated.
The differences between groups:
1. Standard care: a surgical nurse met the patient on arrival on the operating ward, and stayed with the patient until she was sedated at the operating theatre. As standard, we have long had an idea of this care as optimal.
2. The Ward Model: a nurse known to the patient accompanied her to the operating ward. A project nurse with several years of experience in gynecological cancer care met the patient at the outpatient clinic prior to scheduled surgery in order to establish a connection and to answer questions. On the day of surgery, approximately 15 minutes before transfer from the gynecology ward, the same project nurse visited the patient again and accompanied her to the operating ward, and into the operating theatre. This care combination might be superior, as a known nurse might be beneficial for patients.
3. The Caregiver Model: a relative accompanied the patient to the operating ward.
On the day of surgery, a close relative of the patient’s choosing was prepared to follow the patient. (S)he was shown a photo of the operating theatre, was asked to put on protective clothing, and was informed to leave the operating theatre quietly once the patient was heavily sedated. The close relative accompanied the patient to the operating ward and into the operating theatre, and acted according to the patient’s wishes and as appropriate to the situation. As a close relative know the patient and visa versa, it might be beneficial to patients to have a close relative by their side at the theatre.
4. The Anaesthesia Model: only the nurse anesthetist attended to the patient 5 - 10 minutes after arrival on the operating ward. The nurse anesthetist and the operating nurse accompanied the patient from the operating ward into the operating theatre. In this care combination the nurse anesthetist and the patient had undisturbed time to establish a relationship, which might be beneficial to patients, as the nurse anesthetist would be the one keeping a special eye on the patient when she would become sedated.
Patient measures and data collection
The primary outcome was a change in nervousness from admission to immediately prior to anesthesia and cancer surgery. Patients expressed their nervousness intensity as a VAS (Visual Analog Scale) score, ranging from zero to ten, where zero meant not at all nervous and ten meant the most nervous the patient thought she could ever be. This measure was collected with one decimal upon admission (primarily noted by the same secretary for all patients) and immediately before being anaesthetized (noted by the nurse anesthetist who was present). The VAS score was tested in the study population before study start (n = 20), and a clinically meaningful difference was stated to be 0.9 - 1.0. The secondary outcome was global satisfaction at discharge, measured by use of a questionnaire filled in by the patients immediately before discharge and handed to a nurse in a sealed envelope. The questionnaire had a four-point Likert scale, including the possibility to choose “don’t know” and included the five items: global questions regarding information received practical help, support, surgery and medicine, as well as the overall impression of the hospital visit. It was qualitatively validated in the study population (n = 12) before study start using a think-aloud test with silent observations, retrospective interviewing, and a small correction to one question after testing on the first four patients  .
Patient characteristics and time spent
Specific patient characteristics were given by the patients before randomization and collected later from the electronic patient journal. The nurses involved measured time spent on support that was in addition to that provided by the nurse anesthetist in each group.
2.4. Sample Size
Based on a prior study  , we assumed a mean change of three on the VAS scale in the standard care group, and a common standard deviation of two. With these assumptions, we calculated that, with 64 patients in each group, we would have 80% power at the two-sided 5% level to detect a difference of at least one in mean change of nervousness.
2.5. Randomization and Masking
Participants were assigned to one of the four groups using simple randomization via a computer-generated random number list that was printed out and placed in numbered sealed envelopes by the first author. Nurses at the outpatient clinic collected written informed consent, and telephoned the study nurse who allocated patients to a group using the next numbered sealed envelopes. Patients became participants as soon as the envelope was opened. The nature of the interventions made it impossible to mask either patients or healthcare professionals. The statistician was masked until the analyses and interpretations were completed.
2.6. Control of Complex Intervention
All the nurses were regularly verbally incited to keep the interventions as described and to report immediately, if any deviations were observed. Moreover, first author made three to five minutes semi-structured interviews with random participants in all groups with the prompt: Please describe for me, the care you received before anesthesia.
2.7. Statistical Analyses
Specific patient characteristics were summarized by descriptive statistics number and percentage for categorical parameters and mean, range of scores, and standard deviation for continuous parameters. VAS score at admission and immediately prior to surgery were summarized, together with the corresponding change. The primary analysis was by linear regression of the difference in VAS score adjusted for group and baseline score, with the standard care group as reference. The difference in change in VAS score between the groups is given by the mean, together with the 95% CI. The normality assumption of the difference was checked by visual inspection of the quantile-quantile plot. The primary analysis was on the full analysis set, i.e. only those with two VAS score measurements. Additionally, we assumed that missing scores were “Missing At Random” i.e. the missing values did not depend on unobserved data. Under this assumption, the missing values were imputed using multiple-imputation with a model containing group and baseline value. Linear regression was performed, including the imputed values of the difference adjusted for group and baseline value. Furthermore, we repeated the primary analysis with the Minimal Model as reference. The questionnaire scores were summed and both a Kruskal- Wallis equality-of-populations rank test and pairwise Wilcoxon rank sum tests were performed. Stata 13.1 StataCorp LP, College Station TX USA was used for all statistical analyses.
2.8. Ethical Issues
Before randomization, patients were informed of the standard care to non-participants, including the possibility to be accompanied by a relative up to (but not into) the operating theatre. The study number at the Biomedical Research Committee System Act at the Scientific Committee for Southern Region in Denmark is S-20082000- 41, and the study adheres to the Declaration of Helsinki  and the Ethical Guidelines for Nursing Research in the Nordic Countries  . It was ethical acceptable to ask patients about their nervousness right before anesthesia as this theme had been addressed to the patients earlier, the question was asked friendly by the nurse anesthetist, and it is not necessarily desirable to remove nervousness entirely before sedation   .
In total, 1116 patients were assessed for eligibility from December 2008. In May 2012, the desired sample size was reached and inclusion ceased. In total, 350 patients provided written informed consent and were randomized (Figure 1). Among the non-participants, many were offered minimal invasive surgery, some stated that they did not have relatives, several declined to participate without giving a reason, and some did not want to rely on help from relatives. Of those randomized, 13 did not want to participate in the end (due to overwhelming developments in their disease or unspecified situations), and 42 experienced changes in treatment plan (to minimal inva-
Figure 1. Flow chart. *ITT: Intention to treat, **FAS: Full Analysis set.
sive surgery or surgery at another hospital). A total of 295 patients provided specific characteristics and these were well balanced across the randomization groups (Table 1). All 295 patients provided a VAS score on admission, and approximately 1.5 days passed between this and the next VAS measures. A total of 37 patients were lost to follow-up, primarily due to business on the ward and premedication given before the VAS evaluation, leaving 258 as complete cases. Patients were in 73 cases follow by a relative of their choice, which were mainly a husband or a daughter. Only few had chosen a friend to follow them. The interventions were delivered consistently apart from four cases where other nurses reported the care to be not as attentive as it could be (one in the ward model and one in standard care) or the care could not be delivered by the same nurse at the outpatient clinic and in the ward (two in the ward model). All interviewed patients (n = 30) described the right care according to their allocation, and a few of those receiving the caregiver model commented, that they were surprised about their relatives reaction which could be both positive and negative.
Descriptive statistics for VAS of nervousness reveals no apparent clinically relevant difference on admission (Table 2). We observed a mean increase in nervousness in all four groups. The largest increase related to the Anesthesia Model, where only a nurse anesthetist was present. This is depicted, for illustration purposes only, by the regression lines in Figure 2. Patients who received supportive care in the form of the Ward Model exhibited the greatest benefit, with a mean difference in change in nervousness of −0.38 [CI: −1.140 to 0.385] p = 0.330, but no experimental model proved to be superior to standard care with statistical significance or clinical relevance. The Caregiver Model achieved a mean difference of 0.23 [CI: −0.498 to 0.967] p = 0.528, and the Anesthesia Model achieved a mean difference of 1.05 [CI: 0.298 to 1.794] p = 0.006. In the additional analysis, the Anesthesia Model was found to be significantly inferior to all the other care combinations. However, in those receiving support from the Caregiver Model, the difference was not clinically relevant: −0.81 [CI: −1.532 to −0.011] p = 0.027. The Ward Model achieved a mean difference of 1.42 [CI: −2.174 to −0.674] p < 0.001. When we performed an analysis for robustness by multiple imputations, no substantial changes were found (data not shown). With regard to patient satisfaction, we found a similar picture. No experimental models proved superior to standard care (Table 3), but in the additional analysis, the Anesthesia Model proved significantly inferior to both standard care (p = 0.013), the Ward Model (p ≤ 0.001), and the Caregiver Model (p = 0.004). The
Figure 2. Regression lines for differences in Visual Analogue Score (VAS) in the four groups at the time before and after intervention.
Table 1. Patient characteristics at admission and at discharge, and the staging.
Table 2. Distribution of the Visual Analogue Score (VASa) of nervousness and the differences between them within the original assigned groups.
aVAS score of nervousness is from 0 to 10, with 0 as no nervousness at all and 10 as the most extreme nervousness patients could ever think of, bFAS: Full Analysis Set, cITT: Intention to Treat.
Table 3. Distribution of patient satisfaction score in the four original assigned groups and differences between groups.
aScore min = 0, max = 15.
Anesthesia Model was the most economically efficient, as it involved no additional time to that spent by the nurse anesthetist, while, for standard care, the Ward Model and the Caregiver Model, respectively, 35, 57, and 6 additional minutes per patient were spent by staff.
We investigated the effect of different healthcare combinations and the involvement of a relative on the nervousness of cancer patients before anesthesia and cancer surgery and we were not able to show that any of the tested interventions were better than standard care. In our additional analyses, standard care, the Ward Model and the Caregiver Model proved significantly better than the Anesthesia Model on both measures: global satisfaction at discharge and VAS of nervousness before sedation and surgery. However, a clinically relevant improvement was reached only by standard care and the Ward Model. Patients’ nervousness was not eradicated, but an increase in nervousness was to some degree counteracted. Preoperative anxiety and recovery have been in focus in research for more than fifty years. Within the last two decades, reviews acknowledge that, a reduction of preoperative anxiety to zero would not necessarily be of benefit to patients   . Preoperative anxiety is, nevertheless, a good predictor of recovery      , and a recent systematic review recommends a reduction in higher levels of anxiety before sedation for surgery for a life-threatening disease  . Because anxiety level is influenced by personality characteristics  , a recommended preoperative anxiety level would be misleading. Our study adds more scientifically tested care combinations to support patients in counteracting an increase in preoperative anxiety. The procedures carried out by the nurse anesthetist could be painful. Moreover, in our study, the anesthetist came to the patient five to ten minutes after the patient arrived on the operating ward. Pain inflicted by healthcare professionals might provoke anxiety, and in our unconscious mind, pain might be linked to threat and danger  . Research on patients’ experiences of dialogues prior to, during and after surgery with the same nurse, however, revealed that patients considered the talks they had and the continuity to be anxiety-reducing   and even that they “prevented and alleviated suffering in surgical procedures”  . This idea is supported in a study about children’s healthcare. It is less the case that the patient suffers pain from the actual procedures conducted, but from the environment in which they take place  . The inherent waiting time at the operating ward before the anesthetist nurse showed up may have contributed to patients’ perception of the situation. This is supported by research performed by Cobley et al., who interviewed 124 patients by means of a 16-item questionnaire within 36 hours of surgery and found that arrival at the operating ward was a distressing time for patients, and for some it was even unbearable  . Having someone to talk to during the perioperative period is preferred by females  , and females might indeed benefit by not being left alone at the final preoperative stage. However, we lack research on adult patients’ experiences in general, the effects of unpleasant or painful preoperative procedures, and being left alone on the operating ward.
We found no statistical or clinically relevant difference between standard care and the Ward Model or the Caregiver Model. These interventions both offered ongoing attention and care from the time before leaving the ward and until the patient was anaesthetized. In this way, support was provided throughout the period that Coleby et al. identified as specifically distressing before surgery: waiting on the ward and being transferred to, and arriving at, the operating ward  . Moreover, the patients who received these support interventions had the opportunity to make a connection with their respective supporters, which might also be beneficial for patients   . However, compared to standard care, the benefit for patients was not statistically significant or clinically relevant in counteracting an increase in nervousness and did not show an increase in global satisfaction with the healthcare. It is notable that compared to the Anesthesia Model in the additional analyses; the Caregiver Model did not result in clinically relevant changes in patients’ experienced nervousness. Relatives might have more personal strength than patients, but new research has found a link between the level of anxiety in cancer patients and their relatives  . Moreover, a new systematic review did not find a convincing effect of family caregivers’ supportive care in cancer  . Therefore, it is not recommended to rely fully on supportive care from relatives in the preoperative period.
Strengths and Limitations
We planned the sample size assuming a standard deviation of two. Evidently, this was optimistic for some groups (Table 2). Thus, the study is slightly under-powered. However, apart from the Minimal Group, none of the experimental groups show clinically relevant differences compared to the standard care group. Hence, higher power would not be likely to alter the clinical conclusions.
The satisfaction questionnaire used was only qualitatively validated. However, the few questions were validated in the study population ahead of study start. Both of the patient measures were based solely on self-report and therefore influenced by personality. Given the size of the study, we can be fairly confident that any unobserved characteristics were well-balanced across the groups. However, we acknowledge that there is some potential for bias because our randomization was not blinded to the patients, the healthcare professionals or the relatives. Patients might have been more nervous by knowing they were allocated to the model with only one supporting nurse. Patients were, however, informed about the possible benefits of all the care combinations, and one good care relation is expected to be enough   . Moreover, following this logic, patients who knew they were to receive care from an additional person and even be followed from the ward might have been even less nervous, but we did not observe this. The healthcare professionals could have promoted their own intervention and disturbed others’, but this was not reported. On the contrary, others performance was reported if it was not as attentive as possible judged by their healthcare professional co-workers. At the study site, each nurse’s primary aim was to strive to help the individual patients as well as possible, and the study’s aim was secondary to this, which is reflected in the missing VAS scores from the operating room. The fact that these are missing, of course, constitutes a limitation in its own right, but the number of complete cases is still convincing, and the robustness analysis did not hint at any problems with reliability. Our interventions ran over more years. The intervention did not change, but the context changed slightly with a few changes in staff, and increased offers of minimal invasive surgery which excluded such patients and leaved those with larger tumors as participants in our study. However, all changes would have affected even in the four arms, due to the study design. The interventions were complex and it was not possible to control whether the best possible anxiety reduction strategy was used in the specific interactions. However, we expected that the healthcare professionals would need to be free to use their knowledge and skills to adapt care to suit individual patients’ needs   . Our approach was to stay as close as possible to usual clinical practice and this enhanced the ecological validity of the study, and, consequently, its external validity  . Of females acceptable for inclusion, 47% declined to participate. As is inherent in almost all such research, we may therefore have included only the strongest.
Previous studies report that females have a greater tendency to have preoperative anxiety, a tendency to higher anxiety, and more females than males prefer to have a partner by their side in the preoperative period  -  . Eventually generalizability to male patient populations should therefore be done with great caution.
Supportive care by a surgery nurse, from the time of arrival at the operating ward, and by a nurse anesthetist, from 5 - 10 minutes after the arrival and until the patient has been anesthetized, will help female cancer patients to counteract an increase in preoperative nervousness and feel satisfied at discharge. The support and company of a ward nurse from ten minutes before leaving the ward and until the patient is anesthetized provide a similar level of benefit. However, support from a surgery nurse and a nurse anesthetist might be attractive to offer as standard, because this is the most time efficient of the four care combinations. Relatives might, to some degree, be helpful in alleviating female patients’ preoperative suffering due to nervousness, but it is not recommended at any time to rely fully on their support. Furthermore, the sole support of a nurse anesthetist from 5 - 10 minutes after arrival at the operating ward is not recommended. How nurse anesthetists’ procedures and patients’ time alone on the operating ward affect adult patients in general should be the focus of further research.
We thank the patients and the relatives who participated in this study. Moreover, we thanks Jette Johansen, Annette Henriksen, Inger Juhl and Kristian Kidholm who contributed to the study’s conception and we acknowledge the important collaboration in randomization and data collection with nurse anesthetists, surgery nurses, ward nurses, and the secretary where the study took place. Lorna Campbell copy-edited this article. The study was financially supported by non-partisan funds: The Tryg Foundation and Sister Marie Dalgaard’s Foundation.