1 INTRODUCTION

About seven in 1000 people are affected by any kind of epilepsy (Fiest et al., 2017) and despite best medication one-third of them do not become seizure-free (Pati & Alexopoulos, 2010). For some of these patients epileptologists recommend hospitalization in an epilepsy monitoring unit (EMU). However, it is not only patients with drug-resistant epilepsy who are treated as inpatients in the EMU. Patients are also admitted because it is unclear whether their seizures are of epileptic or non-epileptic origin, or when the frequency of their seizures should be quantified. The aim of the stay is to observe changes in the patient's behaviour and analyse the brain electric activity during a seizure. Therefore, patients are continuously monitored via electroencephalograms, audio- as well as video recordings and several additional tests, for example single-photon emission computed tomography (Rosenow et al., 2016). Since patients should have seizures during their stay, antiseizure medication is tapered off and additional seizure facilitating procedures, such as sleep deprivation, are undertaken. Because of an enhanced risk of injury, different safety measures, such as restricted mobility, padded guard rails and surveillance by professional staff are standard in this setting (Kobulashvili et al., 2016; Rosenow et al., 2016). Mostly, EMUs are a part of neurological or neurosurgical wards in epilepsy centres. The size of these units ranges from one to seven or more beds in single, double or multi-bed rooms (Kobulashvili et al., 2016). The rooms may be equipped with TV, radio and internet access. The length of the stay depends on the admission indication and the patient's individual situation. Mostly, it is 3–7 days; and with fees ranging up to €2200 per day, hospitalization in the EMU can be costly (Kobulashvili et al., 2016). Physicians, technicians and nurses look after the patients, with the nursing staff either caring only for the patients in the EMU or additionally for the other patients on the ward. EMU patients report violation of privacy, boredom and concerns about their health. But they also hope that based on the diagnostic findings their quality of life will improve. This should be reached by a reduction in seizure frequency or by seizure freedom (Egger-Rainer et al., 2017). There are few EMU beds in epilepsy centres and the waiting time for inpatient admission can be several weeks to months. It is, therefore, important that occupancy is well planned and that patients keep their admission appointment. Because of past experiences in the EMU some patients object to a necessary readmission. Others end their stay prematurely because of the perceived stress. In an American study, 4.1% of the patients left the EMU early (Caller et al., 2014), and in an Australian study the figure was 12.5% (Andrewes et al., 1999).

Experienced comfort can help the patients to overcome stressful situations (Kolcaba, 2003). Comfort is a basic human need and people strive for a high level of comfort (Gropper, 1992), whereby ‘[t]he comfort level appears to be the maximum level that a patient can bear or tolerate without becoming distressed’ (Morse et al., 1994, p. 194). During a hospital stay, it is the nurses who have to ensure this experience (Kolcaba, 2003). To reach this goal, nurses must know the patient's current comfort level and possible intervening variables on nursing comfort measures.

1.1 Background

In the mid-19th century, Florence Nightingale described the importance of comfort for sick people and defined the care of a high level of patient comfort as a central nursing task. It is demanding to ensure patient comfort because the nurse must recognize the patient's individual comfort needs in a momentary situation. But the situation can change quickly and so can the patient's needs (Gropper, 1992). Kolcaba (2003) describes that comfort is a two-dimensional holistic construct. This construct comprises three comfort types: (1) Relief, meaning that the patient experiences having a specific comfort need met; (2) ease, relating to a state in which the patient experiences calm or contentment; and (3) transcendence, meaning that the patient can rise above a specific problem. The three comfort types are experienced in four comfort contexts: physical, psychospiritual, sociocultural and environmental. According to Kolcaba's comfort theory, each patient has specific comfort needs that arise in stressful situations. Nurses address these needs by setting individualized comfort measures within the framework of comfort care. If the comfort measures are appropriate, the patient comfort can increase. The patient feels strengthened and empowered to cope with the stressful situation and to engage in health-seeking behaviours, for example following necessary restrictions. However, nurses must be aware of intervening variables that might render the intervention ineffective (Kolcaba, 2003). Comfort levels are assessed with dedicated comfort questionnaires, such as the Epilepsy Monitoring Unit Comfort Questionnaire (EMUCQ) (Egger-Rainer et al., 2020). To provide comfort-enhancing interventions nurses select from three different measure types: (1) Technical comfort measures, which aim at physiological functions; (2) coaching, which targets anxiety reduction and realistic planning of recovery; and (3) comfort food for the soul, which means unexpected measures that target transcendence. Due to the holistic nature of comfort, measures that increase one comfort type also have positive effects on the other ones. Therefore, total comfort is greater than the sum of its parts (Kolcaba, 2003). As already mentioned, the effectiveness of the intervention depends not only on its appropriateness, but also on the intervening variables. These variables interact with each other and are inherent in a situation. They may be parts of persons, for example sociodemographic characteristics, health history and culture. Or they may comprise different factors such as the patient's previous experiences, the caring model and local conditions (Kolcaba, 2003). Since the intervening variables cannot be addressed by nursing interventions, or only to a limited extent (Kolcaba, 2003), it is all the more important that nurses learn about them. However, little is known which intervening variables may be relevant in the EMU. Therefore, we used data from the EMUCQ validity study (Egger-Rainer et al., 2020) to gain deeper insight into the topic.

2 THE STUDY 2.1 Aims

The aim of the study was to find out which variables might be associated with comfort of EMU patients.

2.2 Design

We used an exploratory, quantitative study design and conducted a secondary analysis of data obtained in the multi-centre EMUCQ validity study (Egger-Rainer et al., 2020). The statistical approaches that are used in this analysis have been specified in advance, and are described in detail in the data analysis section below. However, the explanatory variables of the respective models were not selected based on previous knowledge, but rather in a data-driven way.

2.3 Sample/Participants

Participants were recruited consecutively in 10 EMUs comprising a total of 51 beds, in Austria and South Germany. Patients with at least 18 years of age were eligible if they were hospitalized in the EMU for 5 days or longer, literate in German, not mentally disabled and had signed the informed consent form. The sample size was determined according to standard methodological recommendations for factor analysis. For more details please refer to the previous study (Egger-Rainer et al., 2020).

2.4 Data collection

Data collection lasted from October 2018 to November 2019. The local researchers approached eligible patients and explained to them the aim of the study. Patients who agreed to participate in it filled out a questionnaire and reported the occurrence of seizures on the second and on the last day of their stay in the EMU. Additionally, at the first time point participants completed a form with sociodemographic data regarding gender, age, marital status, educational level and occupation. Patients also reported their reason for referral. Information on mobility restrictions was provided by the local researchers of the respective EMUs. The data collection procedure is reported in detail elsewhere (Egger-Rainer et al., 2020).

2.5 The questionnaire

The EMUCQ was used to collect the data. This questionnaire was based on Kolcaba's General Comfort Questionnaire (Kolcaba, 2003) and was specially designed to assess patient comfort in the EMU. It is a self-completion, paper-pencil instrument for adult patients. The EMUCQ consists of 42 items and shows the three comfort types in three subscales. Patients rate their comfort on a six-point Likert-type scale in which 1 means ‘strongly disagree’ and 6 means ‘strongly agree’. Before statistical analysis is performed, negatively worded items are coded inversely and, therefore, higher scores mean higher comfort levels. The minimal and maximal reachable scores are different for the comfort type ‘relief’ (range 13–78) the comfort type ‘ease’ (range 18–108) and the comfort type ‘transcendence’ (range 11–66). For total comfort, a minimum of 42 points and a maximum of 252 points can be reached (Egger-Rainer et al., 2020). Kolcaba (2003) recommended discussing the results analogously to the comfort contexts, too. Thus, following Kolcaba's theoretical assumptions, the EMUCQ items were assigned to four subscales corresponding to the comfort contexts: physical (range 10–60), psychospiritual (range 11–66), sociocultural (range 6–36) and environmental (range 15–95) comfort. The assignment of the items to the subscales of the two comfort dimensions can be found in the appendix in Table A1.

2.6 Ethical considerations

In the conduct of the study, the provisions of the Declaration of Helsinki were followed. Approval was obtained of the local ethics commissions in Baden-Württemberg (B-F-2018-099), Erlangen (440_18Bc), Innsbruck (1143/2018), Linz (1104/2018), Munich (18-573), Salzburg (415_EP/73/700-2016), Tübingen (589/2018BO1) and Vienna (1672/2018). All patients were informed verbally and in writing by the local researchers and were free to participate in the study.

2.7 Data analysis

Data of the first data collection time point, which was the second day of the patients’ stay in the EMU, were used for analysis. To explore the sociodemographic characteristics of the participants, frequency, mean and standard deviation (SD) were calculated, as appropriate. For the comfort scores we report range, mean and SD. The Statistical Package for the Social Sciences (SPSS) 23 was used in this regard. An exploratory model building approach was adopted to analyse potential associations between the comfort scores and demographic/clinical characteristics. Only complete patient data sets were included in the analysis. At first, after some basic descriptive analyses, a multiple linear regression model was considered. The model included gender, age, marital status, educational level, occupation, mobility, reason for referral, occurrence of seizures and centre as explanatory variables and the overall comfort score as outcome variable. In a second step, the final model was obtained by excluding some explanatory variables that were highly correlated, in order to avoid problems due to multicollinearity. Finally, two models––including gender, occupation and centre––fitted the data almost equally well; although, one of them did not yield estimates for all explanatory variables, due to computational issues. Thus, this model is only reported as supporting information (Table S1) and the other one was chosen as the main model in the present manuscript. For the explanatory variables included in the latter model, additionally, linear regression models for the subscales of the EMUCQ, which are also reported as supporting information (Tables S2 and S3), were employed. In order to further assess the stability of the results, several additional sensitivity analyses using different sets of explanatory variables were conducted. Furthermore, the model appropriateness was also assessed by using residual and QQ plots. Results of the linear regression models are reported as regression coefficients along with standard errors (SE) and corresponding p-values. The significance level was set at α = .05. All analyses were conducted by using R version 4.0.2 (R Core Team, 2020).

2.8 Validity and reliability

Validity and reliability for the three factorial solution of the EMUCQ is reported in detail in Egger-Rainer et al. (2020). For the three subscales and the total comfort scale the convergent validity was supported with all correlations ≥.03 (p < .001). The internal consistency was good (Cronbach alpha values .77–.88). For the four factorial solution of the EMUCQ internal consistency ranged from .53 to .82 in this study.

3 RESULTS 3.1 Sample characteristics

The 267 patients, who participated in the study, were between 18 and 84 years of age with a mean age of 39.39 (SD 16.36) years. A total of 140 (52.4%) patients were female, 79 (29.6%) had finished high school and 89 (33.3%) were full-time employees working 38–40 h per week. While epilepsy as the reason for referral was reported by 109 (40.8%) patients, 38 (14.3%) patients did not report any reason. Already 53 (19.9%) patients had had a seizure in the EMU before filling out the questionnaire. Full information regarding the participants’ sociodemographic information can be found in Table 1. Most patients had to stay in bed except for going to the bathroom/toilet. In certain cases they were accompanied by a nurse. Depending on the severity of the disease, selected patients were allowed to walk around the room and take their meals sitting at a table in four centres (B, E, F, H). In centre F, some patients were also allowed to leave the EMU for about 30 min several times a day to go to the cafeteria or to go smoking. Patients with invasive EEG-electrodes were confined to bed in all centres. One centre did not report the regulations.

TABLE 1. Descriptive statistics for patient characteristics and EMUCQ total comfort scores (n = 267) Frequency EMUCQ Score (total comfort) n (%) Rangea Mean (SD) Gender Male 124 (46.5) 113–232 179.74 (25.59) Female 140 (52.4) 83–235 182.94 (26.31) Missing 3 (1.1) Age 18–20 years 29 (10.9) 125–221 178.83 (25.81) 21–40 years 123 (46.1) 83–232 179.36 (24.93) 41–60 years 82 (30.6) 121–235 181.11 (27.53) 60–84 years 32 (12.0) 143–230 192.47 (24.11) Missing 1 (0.4) Marital status Single 115 (43.1) 83–232 178.74 (26.86) Married/Unmarried couple 132 (49.4) 123–235 183.88 (25.24) Divorced 12 (4.5) 129–211 175.25 (21.14) Widowed 7 (2.6) 139–224 189.57 (29.80) Missing 1 (0.4) Education Compulsory School 71 (26.6) 83–233 183.32 (26.02) High School 47 (17.6) 129–235 183.60 (23.61) High School Diploma 32 (12.0) 121–221 174.19 (27.67) Vocational Training 75 (28.0) 123–230 182.29 (25.24) University Graduate 36 (13.6) 113–231 180.58 (28.57) No school leaving certificate 2 (0.7) 177–192 184.50 (10.61) Missing 4 (1.5) Occupation High School Student ≥18 years 8 (3.0) 125–193 162.63 (25.06) University Student 13 (4.9) 137–221 187.77 (23.88) Apprentice 18 (6.7) 136–216 179.00 (23.04) Housewife/Househusband 16 (6.0) 139–233 187.69 (22.69) Self-employee 8 (3.0) 152–232 185.00 (27.32) Employee full-time 89 (33.3) 113–231 179.71 (25.76) Employee part-time 35 (13.1) 129–229 177.97 (22.57) Unemployed/Sick leave 36 (13.5) 121–219 179.11 (25.82) Retired 42 (15.8) 143–235 191.81 (25.64) Missing 2 (0.7) Reason for referral Evaluation of seizure frequency 62 (23.2) 113–231 185.87 (24.93) Epilepsy 109 (40.8) 123–235 179.34 (25.31) Unclear epilepsy syndrome 7 (2.6) 125–205 167.86 (30.37) Presurgical evaluation 29 (10.9) 145–222 181.21 (24.59) Vertigo 3 (1.1) 163–190 179.00 (14.18) Optimizing of therapy 19 (7.1) 129–219 178.53 (26.12) Missing 38 (14.3) Seizure occurred No 194 (72.7) 113–235 181.55 (25.27) Yes 53 (19.9) 123–231 180.87 (26.21) Missing 20 (7.5) Center A 30 (11.2) 129–235 190.77 (27.21) B 31 (11.6) 140–224 187.00 (23.63) C 26 (9.7) 127–222 176.92 (24.22) D 56 (21.0) 113–230 176.29 (24.41) E 27 (10.1) 83–222 177.41 (31.83) F 16 (6.0) 152–232 199.19 (24.30) G 30 (11.2) 123–217 172.50 (25.55) H 10 (3.7) 139–221 179.70 (25.35) I 33 (12.4) 145–216 181.88 (22.31) J 8 (3.0) 143–223 183.63 (24.44) Abbreviation: SD, standard deviation. a EMUCQ scores range from 42 to 252 points. 3.2 Comfort scores

The EMUCQ scores for total comfort ranged from 83 to 235 points with a mean of 181.32 (SD 25.95). Higher comfort levels could be observed in women compared to men, and in patients with over 60 years of age compared to younger patients. Regarding marital status, widowed patients reported the highest total comfort, while divorced patients reported the lowest. In terms of education, the means of almost all comfort levels ranged between 180.58 (SD 28.57) and 184.5 (SD 10.61), with patients with a high school diploma reporting the lowest comfort level (174.19 [SD 27.67]). The mean total comfort score of high school students ≥18 was 162.63 (SD 25.06) points. It was about 29 points lower than that for retired persons (191.81 [SD 25.64]), who had the highest comfort scores. Patients, who reported an unclear epilepsy syndrome as their admission diagnosis reached the lowest comfort levels. The comfort scores were approximately the same whether the patients had already had a seizure in the EMU or not. On average, the observed comfort scores were highest in centre F while they were lowest in centre G. For detailed information please refer to Table 1. Results of the comfort scores of the subscales can be found as supporting information. Table S4 shows the EMUCQ scores by occupation and Table S5 shows the EMUCQ scores by centre.

3.3 Influential variables of patient comfort

Gender, occupation and centre were identified as relevant influential variables. They were broadly consistent across the different models calculated. Therefore, for the sake of clarity, only the results of the final model are reported in detail in Table 2. The appropriateness of this model is supported by the residual (Figure 1) and the QQ plots (Figure 2). On average, women had a total comfort score 4.69 points higher than men had. The corresponding statistically non-significant p = .139 indicated a trend. Statistically significant differences could be observed depending on occupation. While high school students ≥18 years reported the lowest comfort, housewives or househusbands had higher comfort levels on average by 22.66 points (p = .033), university students by 25.69 points (p = .019) and retired persons by 28.2 points (p = .003). With a comfort score 8.57 points higher than in the reference centre A, patients in centre F reported the highest comfort (p = .262). Statistically significant lower comfort scores were observed in centre D (−18.44; p = .001) and in centre G (−19.20; p = .003).

TABLE 2. Results of the linear regression analysis for the main model using ‘total comfort’ as outcome (n = 263) Variable Coefficient (SE) p-Value† Gender Male Reference – Female 4.69 (3.16) .1392 Occupation High school student ≥18 years Reference — University student 25.69 (10.89) .0192 Apprentice 13.34 (10.69) .2133 Housewife/Househusband 22.66 (10.54) .0325 Self-employee 17.12 (12.35) .1670 Employee full-time 14.54 (9.22) .1160 Employee part-time 10.53 (9.65) .2760 Unemployed/sick leave 16.09 (9.64) .0965 Retired 28.20 (9.50) .0033 Center A Reference — B −4.88 (6.25) .4356 C −12.73 (6.55) .0531 D −18.44 (5.58) .0011 E −9.82 (6.66) .1419 F 8.57 (7.62) .2618 G −19.20 (6.40) .0030 H −10.90 (8.92) .2232 I −8.52 (6.27) .1754 J −7.46 (9.67) .4412 Note Multiple R-squared: 0.1524; adjusted R-squared: 0.08983. Abbreviation: SE, standard error. † p-values <.05 were considered statistically significant; missing data sets were excluded.

Fitted values versus residuals of linear regression analysis for the main model using ‘total comfort’ as outcome

QQ-Plot of linear regression analysis for the main model using ‘total comfort’ as outcome

The other models can be found as supporting information. Table S2 shows the results for the model based on the three subscales ‘comfort type’ and Table S3 shows those for the four subscales ‘comfort context’. In these models it can be seen that high school students ≥18 years generally reported the lowest and retired persons the highest comfort, with the exception of subscale ‘relief’. In this model it was unemployed persons or persons on sick leave whose comfort levels were lowest (−0.28; p = .946), and university students whose comfort levels were highest (10.61; p = .026). Over all, patients of centre F achieved the highest comfort scores. Only for psychospiritual comfort, it was the patients of reference centre A, and not those of centre F (−0.42; p = .873), who scored higher. On average, the comfort scores of the subscales were between 0.08 (relief) and 3.17 (environmental comfort) points higher in women than in men. However, in subscale ‘psychospiritual comfort’ women reported lower comfort than men did (−0.39; p = .725). Statistically significant differences in relation to gender could be observed in the subscales ‘transcendence’ (2.2; p = .043), ‘sociocultural comfort’ (1.39; p = .012) and ‘environmental comfort’ (3.17; p = .032).

4 DISCUSSION

The aim of this study was to find out influential variables on comfort of EMU patients. It is known that there are factors that influence comfort measures and cannot always be addressed by nursing interventions (Kolcaba, 2003). However, explanations can be sought that account for the intervening character of the variables and these aspects can possibly be addressed. Studies dedicated to exploring influential variables pertaining to comfort are scarce. But there are several reports that, relevant variables on related concepts: patient satisfaction, patient experience, patient well-being and patient perspective on quality of care. According to Kolcaba's comfort theory, patients rate the quality of hospital care better if their comfort level is high during the hospital stay (Kolcaba, 2003). Therefore, we used these studies to compare our results.

The model that best fitted our data included gender, occupation and centre as the most important influential variables. However, the identified variables account for only about 9% of the variance. This small number indicates that the individual differences in comfort scores are highly specific and can be explained by key demographic characteristics only to a limited extent.

Several studies found that women were more satisfied and rated the quality of hospital care better than men did (Chumbler et al., 2016; Grøndahl et al., 2011). In our study, female patients reported a higher comfort compared to male patients, too. The only exception was psychospiritual comfort, in which slightly lower comfort levels could be observed in women. This could be due to the presence of items denoting anxiety and depression in this subscale. Women with and without epilepsy tend to be more anxious and depressed than men (Gaus et al., 2015). Also, in a study on patients in the preanesthesia stage, higher scores of anxiety and depression in women came along with lower comfort levels compared to men (Seyedfatemi et al., 2014). Our study supports the influencing character of the occupational status. A similar result could be observed in a Chinese study, where occupation showed a strong association with patient experience of a hospital stay (Min et al., 2019). In our study, high school students ≥18 years reported the lowest and retired persons the highest comfort levels. Probably it is the fear of the future that presses on the comfort of high school students ≥18 years. Unlike retired persons, the participating high school students ≥18 years were in a life span in which they had to decide on their future profession. A newly diagnosed or poorly controlled epilepsy can limit the career options. It is also possible that absence from school while being in hospital causes worries, as the high school students ≥18 years know that they have to catch up on their studies. Notably the differences are between the centres. Differences in the relationship between patients and health care professionals, the caring behaviour of nurses and the person-centeredness of the ward environment possibly might be a reason for this (Chumbler et al., 2016; Edvardsson et al., 2017). Especially, large windows, exposure to daylight, the view of the windows, noise levels, temperature, equipment with room dividers and the decoration with balanced colour schemes are mentioned in the literature (Eijkelenboom & Bluyssen, 2019). A reason for the influencing character of the centres may also be seen in the different restrictions imposed on mobility. Moving freely is part of internal personal control and essential for patient comfort (Egger-Rainer et al., 2017).

Age, marital status and educational level did not show substantial effect in our study. This finding is supported by studies assessing patient experience in hospitals (Min et al., 2019). However, consistent with previous studies, the comfort levels were higher in older than in younger EMU patients (Danielsen et al., 2007; Grøndahl et al., 2011). It is assumed that older patients are better able to deal with unpleasant situations, because they have already experienced several ups and downs (Da Rocha & Ciosak, 2014). Young patients with epilepsy complain that they miss a substantial part of life, and they even think of suicide (Falcone et al., 2020). Since these problems cannot simply be pushed aside by patients, it is possible that they lead to discomfort during hospitalization. Particularly, not knowing whether the results of the examination will yield a better treatment option f