Patients' memories from intensive care unit: A qualitative systematic review

Abstract Aim To identify and synthesize the evidence regarding adult patients' memories from their stay in the intensive care unit. Design A qualitative systematic review and meta‐synthesis. PROSPERO # CRD42020164928. The review employed the guideline of Bettany‐Saltikov and McSherry and the Enhancing transparency in reporting the synthesis of qualitative research guidelines. Methods Systematic search for qualitative studies published between January 2000 and December 2019 in Cumulative Index to Nursing and Allied Health, Medical Literature Analysis and Retrieval System Online, PsycINFO, and Excerpta Medica Database. Pairs of authors independently assessed eligibility, appraised methodological quality using Joanna Briggs's quality appraisal tool and extracted data. The analysis followed the principles of interpretative synthesis. Results Sixteen papers from 15 studies were included in the review. Three themes emerged: (a) memories of surreal dreams and delusions, (b) care memories from sanctuary to alienation and (c) memories of being vulnerable and close to death.


| INTRODUC TI ON
Patients admitted to the intensive care unit (ICU) are critically ill and often require life-sustaining treatment, technological support and continuous monitoring as well as emotional support. A stay in the ICU can be distressing and wearing on the patients (Adhikari et al., 2011;Rose et al., 2011). The lifesaving and highly complex ICU environment may cause patients to encounter multiple stressors, such as being confronted with own mortality, unfamiliar noises, invasive procedures, pain, sedation, delirium and inability to speak (Burki, 2019). Hence, the patients' illness and the medical care provided may alter their ability to comprehend and take in their surroundings.
Memories are defined as the minds ability to take in information, store it and recall it at a later time (Zlotnik & Vansintjan, 2019). Shortterm memory refers to information processed in a limited period of time, while long-term memory allows us to store information for extended periods, including information that can be retrieved consciously or unconsciously (Camina & Güell, 2017). Hence, a memory is the imprint an experience has created in our mind, although it should be understood that the memory has been moulded in our mind and can be influenced and changed (Redelmeier & Kahneman, 1996).
Previous studies have shown that patients have a range of distressing memories from their ICU stay. Three types of ICU memories have been reported: factual, emotional and delusional memories (Samuelson et al. 2008). Factual memories are common, such as memories of ventilator treatment and other procedures, which are reported by 20%-83% of prior ICU patients (Magarey & McCutcheon, 2005;Ringdal et al., 2006;Roberts et al., 2007). Correspondingly, delusional memories such as hallucinations and dreams are reported by 21%-73% of ICU patients (Ringdal et al., 2006;Rundshagen et al., 2002).
Furthermore, patients may recall pain and anxiety, as well as a feeling of being safe (Egerod et al., 2015;Stein-Parbury & McKinley, 2000. Studies show that delusional memories are an important causative factor for developing symptoms of posttraumatic stress disease (PTSD) postdischarge, such as anxiety and depression (Morrissey & Collier, 2016;Jones et al., 2001;Wu et al., 2018).

| Background
A meta-synthesis explored patients' difficulties during critical illness, describing how patients' suffering was altered in relation to sedation protocols (Egerod et al., 2015). Another qualitative review disclosed how patients changed their perception of what was real and unreal during their critical illness, as well as experiencing proximity to death (Cutler et al., 2013). Other reviews have investigated how the ICU affects patients, exploring the prevalence of PTSD in ICU survivors (Davydow et al., 2008), and in relation to delusional memories (Kiekkas et al., 2010). Recall of delusional memories after ICU discharge seems to be associated with PTSD-related symptoms (Kiekkas et al., 2010).
An integrated review including quantitative studies examined the relationship between ICU treatment and memories after discharge in ICU survivors. The results suggest that deep sedation and administration of corticoids contribute to delusional memories (Aitken et al., 2016).
Furthermore, a literature review investigated patients' memories from the ICU exploring patients' general impressions of their stay found that patients recalled positive, neutral and negative experiences (Stein-Parbury & McKinley, 2000).
To the best of our knowledge, no recent qualitative systematic review (SR) has been conducted to synthesize patients' memories of their ICU stay. Such a review could enhance healthcare professionals' understanding of patients' memories of their ICU experiences. By gaining insight into what characterizes patients' memories, healthcare professionals could be further equipped to tailor care on the basis of updated knowledge of patients' recalled experiences from the ICU.

| Aim
The aim of this systematic review was to identify and synthesize the evidence regarding adult patients' memories from their stay in the ICU.

| Design
The qualitative SR was guided by the guideline for SRs described by Bettany-Saltikov and McSherry (2016), which consists of seven steps: (a) formulating an answerable and focused review question; (b) specifying objectives, inclusion and exclusion criteria; (c) conducting a comprehensive and systematic literature search; (d) selecting the studies to include in the review; (e) appraising the methodological quality of the included research papers; (f) extracting the data; and (g) synthesizing, summarizing and presenting the findings. The review was also guided by the "Enhancing transparency in reporting the synthesis of qualitative research (ENTREQ)" guidelines (Tong, Flemming, McInnes, Oliver, & Craig, 2012). The analysis followed the principles of interpretative synthesis. The review was registered in PROSPERO (registration number: CRD42020164928).

| Inclusion and exclusion criteria
Studies published in English, Swedish, Danish or Norwegian language in peer-reviewed journals were included if they met the following criteria: (a) they had a qualitative design, (b) collected data using interview, (c) they included patients submitted to the ICU for 24 hr or more, regardless of diagnosis, and (d) reported patients' memories, recollections or similar concepts from the ICU. Studies were excluded if (a) patients were submitted for less than 24 hr or if not reported, (b) patients were aged 17 years or younger, (c) if proxy reporting by healthcare professionals or next of kin were used and (d) the data were published as letters, comments, conference abstract, doctoral thesis or as any type of review.

| Study selection
On the basis of the inclusion and exclusion criteria, two pairs of authors independently screened papers for inclusion. Rayyan, a web tool that helps expedite the initial screening of abstracts and titles, was used to facilitate the publication selection as well as blinding (Ouzzani et al., 2016). Publications were first considered for inclusion on the basis of title and abstract. The possible eligible studies were then examined by assessing the full text. When there was any doubt whether a publication should be included or not, a third author independently assessed the publication. The database search and the manual search gave a total of 6,926 papers, and the titles and abstracts of 5,815 publications were screened. The full text of 53 publications was read, and the final sample consisted of 15 studies,

| Data extraction
The first author extracted data from the included papers using a standardized data collection form, while the information was checked by the last author. Conflicts were resolved by consensus or in consultation with a third author. The data collection form included the following information: author, country and year of publication, aim, sample size and characteristics, design and method, time

Eligibility
Included of data collection, and main findings related to our aim. In addition, key quotes from the findings section of the included papers were extracted for the qualitative content analysis.

| Quality appraisal
The quality of the included papers was independently assessed by pairs of authors, using the Joanna Briggs Institute appraisal tool for qualitative research (Lockwood et al., 2015). The results of the quality appraisal are shown in Table 2. No papers were excluded on the basis of the results of the quality appraisal.

| Data abstraction and synthesis
For this systematic review, the data abstraction was conducted in line with the principles of an interpretative synthesis (Dixon-Woods et al., 2005). The analyses were guided by inductive qualitative content analysis as described by Graneheim and Lundman (2004), which has been used in previous qualitative SRs (Eriksen et al., 2020;Sibbern et al., 2017;Uhrenfeldt et al., 2013). Conducting a qualitative content analysis gives the opportunity to interpret manifest and as well as latent content (Graneheim et al., 2017).
The findings section of each included paper was read several times to obtain a sense of the material as a whole. The text describing patients' memories from the ICU stay was identified and studied further. Meaning units that were considered to relate to and illuminate the aim were highlighted and extracted from the results section of the included papers. These meaning units were condensed by shortening the text while preserving its core. The condensed meaning units were abstracted by coding and categorizing the material.
Codes were created to describe and interpret the condensed meaning units. The codes generated across the studies were compared according to differences and similarities and gathered into categories to unite the data and conjugate the findings. Categories considered to be related to each other were grouped together. Guided by our aim and through discussions among all the authors, the underlying meanings in these categories were abstracted and interpretated into new themes across the papers. This enabled the analysis to go beyond the content of the included papers. An example of data abstraction and synthesis process is shown in Table 3.

| Characteristics of the included studies
The included studies were conducted in Sweden (N = 2), Australia  Table 4.
The data synthesis revealed three new themes: (a) memories of surreal delusions and dreams, (b) care memories from sanctuary to alienation and (c) memories of being vulnerable and close to death (Table 5).

| Memories of surreal dreams and delusions
In the majority of the studies, patients described having surreal recollections, which many described as the most distinct and scary.
The ICU stay was remembered as a time of confusion and disorientation and many did not know where they were. Patients described being in a state of constant haziness, not knowing if they were awake or asleep and unable to separate day and night. Patients described the border between the real and unreal as being blurred (Adamson et al., 2004;Löf, Berggren, & Ahlström, 2006Meriläinen et al., 2013;Olsen et al., 2017;Pattison et al., 2007;Storli et al., 2008). This led them into a chaotic state of mind: "They found it difficult to differentiate between memories of ICU, the ward and their dreams and nightmares" (Minton & Carryer, 2005).  Patients had several memories encompassing a wide range of dreams, including travelling, the nurses and peculiar people that were out of the ordinary, not necessarily causing fear. These dreams were often described in detail and were perceived as real (Guttormson, 2014;Karlsson & Forsberg, 2008) emphasizing the patients' degree of confusion and disorientation. One patient dreamt that he was physically connected to things (Karlsson & Forsberg, 2008), and others had dreams where family members portrayed hospital staff (Roberts & Chaboyer, 2004). Furthermore, patients recalled comforting or inspiring dreams with relatives or godly figures coming to their aid (Alexandersen et al., 2019;Magarey & McCutcheon, 2005).

TA B L E 3 Example of the analysis process
The majority of patients' recollected dreams were nightmares.
A persistent recollection, represented across the studies, was of patients' upsetting hallucinations and delusions. Several patients saw things such as insects and animals in the room (Löf et al., 2006;Storli et al., 2008), and some saw things emerging through the walls (Adamson et al., 2004;Magarey & McCutcheon, 2005) or blood on the surfaces of the room (Löf et al., 2006). Patients reported having paranoid hallucinations of people trying to hurt and kill them, including nurses (Löf et al., 2006;Minton & Carryer, 2005;Olsen et al., 2017). Some had delusions of being surrounded by dead bodies (Storli et al., 2008). Common to these hallucinations was that they evoked fear, and several described them as horrifying. Furthermore, patients commonly described memories of hallucinations that reoccurred, seemed realistic and were intimidating (Löf et al., 2008;Roberts & Chaboyer, 2004

| Care memories from sanctuary to alienation
In the majority of the papers, patients described memories of the care they were provided in the ICU. Patients recalled nurses surrounding them, performing procedures on them and their fellow patients (Löf et al., 2006;Meriläinen et al., 2013). For many patients, the nurses especially evoked emotions of safety and security. The nurses were remembered by their warm touch, kind voices, reassuring words and as being technically competent and by their ability to collaborate and treat them as a person (Adamson et al., 2004;Laerkner et al., 2017;Löf et al., 2006Löf et al., , 2008Olsen et al., 2017). Patients remembered nurses relieving their discomforts by removing secretions from their airways and alleviating their pain (Adamson et al., 2004;Karlsson & Forsberg, 2008).
Patients' interactions with nurses were not remembered as exclusively reassuring, some had memories of nurses making them feel helpless, like they were not seen or heard (Karlsson & Forsberg, 2008;Laerkner et al., 2017;Meriläinen et al., 2013). Others described feelings of being left alone and that the nurses did not have time for them (Laerkner et al., 2017;Olsen et al., 2017). The patients described times where they felt the nurses were annoyed and perceived that the nurses got tired of them and their needs (Karlsson & Forsberg, 2008;Löf et al., 2008;Magarey & McCutcheon, 2005;Meriläinen et al., 2013). "I don't remember being in pain because the nurses would come… maybe to keep me quiet" (Adamson et al., 2004). Furthermore, patients remembered nurses discussing private issues and other patients while caring for them (Laerkner et al., 2017;Löf et al., 2006;Minton & Carryer, 2005). "I experienced a time when it [for the nurses] was more about talking to each other than taking care of me. I was so annoyed" (Laerkner et al., 2017).

| Memories of being vulnerable and close to death
Patients described memories of being confronted with their vulnerability and dependency, giving them a sense of losing control over their situation. Patients remembered not being able to move, feeling thirsty and feeling dismayed by significant muscle loss. They recalled discomfort and shame related to the fear of not controlling the bowel function and being dependent on others to accommodate their needs (Alexandersen et al., 2019;Meriläinen et al., 2013;Page et al., 2019). Others remembered it as difficult to be awake during mechanical ventilation as it increased their awareness of the severity of their illness (Laerkner et al., 2017). "I don't quite understand why I had to be awake while I was so sick. I just don't. They said it would be better this way, but I found it hard. When my brain was so affected … I couldn't stand it. I felt like a fish out of water" (Laerkner et al., 2017).

TA B L E 4 Characteristics of the included studies
Author, year country Aim Sample size and characteristics

Adamson et al., (2004) Australia
To examine the participant' memories of intensive care and hospitalization at 6 mounts post-discharge, and to explore the impact of the critical illness experience on the recovery  (Ballard et al., 2006). A few remembered how they felt trapped and trying to break free from the lines and equipment in an attempt to regain control (Löf et al., 2006;Minton & Carryer, 2005;Tembo et al., 2012).
The patients' difficulty communicating caused them to be robbed of their independence. They remembered feeling distressed and very frustrated because of the endotracheal tube hindering them in vocalizing their needs (Magarey & McCutcheon, 2005;Olsen et al., 2017;Tembo et al., 2012). In addition, some had problems understanding the nurses. Patients described distress in regard to the ever-changing staff and several speaking foreign languages, thus putting additional pressure on communication (Laerkner et al., 2017;Löf et al., 2006;Magarey & McCutcheon, 2005;Olsen et al., 2017). These challenges made patients feel discouraged. The great strain the deficit in communication represented was made clearer when patients described memories of relief when regaining the ability to express themselves; "The use of a one-way valve was an enormous relief and was described as 'winning the Lottery.' Being able to communicate provided a measure of control in an uncontrollable world" (Olsen et al., 2017). Patients recalled how calm and relieved they felt when nurses understood their gestures and could read their lips (Löf et al., 2006).
Patients remembered how they felt that death was imminent, causing distress and anxiety (Löf et al., 2008;Meriläinen et al., 2013;Tembo et al., 2012). They recounted episodes of horrifying and intense pain, causing great fear (Löf et al., 2006;Magarey & McCutcheon, 2005). The remembered pain was described by some as very distressing (Adamson et al., 2004;Alexandersen et al., 2019;Meriläinen et al., 2013) and all-consuming: "They asked me if I had any pain and where the pain was… I could not say; the pain was everywhere" (Meriläinen et al., 2013). The sense that their life was in danger became manifest, especially in respect to ventilator treatment, and patients recalled panic and fear when they felt like they were not getting sufficient air supply (Karlsson & Forsberg, 2008;Löf et al., 2008;Storli et al., 2008;Tembo et al., 2012). These memories were strong and upsetting; "… I can remember feeling that I wasn't getting enough breath. I thought I was going to suffocate.
My fighting spirit kicked in. I tried to pull it out [tube] … You get a feeling of how you might simply die!" (Tembo et al., 2012). The patients' confrontation with their own mortality was also outlined in how they felt exhausted from the treatment, wanting to give up (Löf et al., 2008;Magarey & McCutcheon, 2005). Despite a high symptom burden and severe medical condition, some patients had no recollections of thinking about death or questioning their survival. They remembered fighting for their recovery (Alexandersen et al., 2019).

| D ISCUSS I ON
This qualitative SR aimed to identify and synthesize the evidence regarding adult patients' memories from their stay in the ICU.
Essentially, the majority of patients' memories were related to negative experiences and feelings from their stay in the ICU, and a range of memories confronted patients with life and death. In contrast, we found some positive memories regarding safety, secure and good care provided by nurses, who also were competent and collaborative.
The patients 'confusion and disorientation of time and place, in addition to being aware of their inability to comprehend what was going on around them, seemed to evoke feelings of fear and reinforce the feelings of helplessness. These findings are in line with a study discovering that patients recalled the chaos and disorientation as the biggest threat in the ICU (Maddox et al., 2001). Therefore, patients could have a great need for knowing what was happening to them, and what was real versus unreal (Hupcey, 2000).
Our findings provide deeper insight into the contents of delusional memories generating intensity and level of fear as well as realistic and threatening nightmares. In addition, we found that most patients recalled that delusions evoked strong emotions, which seems to be consistent with what Storli et al. (2008) contended, the emotional intensity of the delusion is decisive for how firmly the delusion is rooted in the memory. Furthermore, several of the recalled delusions reflected a blurred border between the real and unreal. A review found that experiencing unpleasant events could be misinterpreted by the patients and easily change into delusions and nightmares (Kiekkas et al. 2010). This was displayed in our findings showing that nurses' care and procedures could manifest as threatening nightmares, such as patients being convinced that they were being held captive or that the nurses were trying to kill them. Patients delusional memories from the ICU were associated with the development of diverse aspects of psychological distress, including feelings of anxiety and depression, problems sleeping and PTSD (Kiekkas et al., 2010). However, Aitken et al. (2016) found that it did not matter whether a memory was of a real event or a delusion; the negative outcomes were linked to the fact that the Our findings suggest that patients recalled both positive and negative emotions related to the care provided in the ICU, which reflect the complicated relationship between nurses and patients.
Delmar (2012) claims that all people are fundamentally dependent on one another. Because we bestow our reality on each other, we surrender ourselves to a dependent relationship when interacting.
By influencing each other's frame of mind, our actions and exuded emotions can either expand or constrict the other person's room of action. A person's opportunity to self-express depends on how one is considered by others. According to Delmar (2012), the relationship between a nurse and a patient is not fully voluntary and reciprocal, making it an asymmetrical professional relationship based on dependency and power.
Some patients' room of action seemed to be provided by ICU nurses. How patients relied on nurses to initiate communication when they were intubated, and how patients recalled nurses talking over their heads, made them feel ignored and that the nurses did not have time for them. Nevertheless, ICU nurses may consider the patient and the surrounding equipment as a whole, which may lead to objectifying the patient (Almerud et al., 2007). Consequently, some of the recalled experiences of powerlessness we described in our findings may be due to nurses being ignorant or unaware of their power over the patient. Nurses need to expose the unconscious and invisible actions of power to prevent the constriction of patients (Delmar, 2012). Some of the nurses who attempted to help patients Our findings show how nurses can contribute to patients' needed security as well as helplessness and fear, depending on how they distribute their power in care. Patients can be dependent and still be in control as long as their needs are met (Boggatz et al., 2007). that when we lose our independence, our entire existence is at stake (Lykkegaard & Delmar, 2013). It is worth noting that patients who "nearly died" in the ICU raised questions about their entire existence after surviving (Parker, 1999).

| Strengths and limitations of the review
A strength of this SR was that systematic searches were conducted in four databases, in addition to manual searches, and that study selection, quality appraisal and data extraction were performed by pairs of authors independently. The first author works as an ICU nurse, two other authors have worked as ICU nurses, while the last author has no such experience. The first author's pre-understanding was discussed throughout the process and its influence with the last author considered, in an effort to set aside bias and make it possible to work past the pre-understanding and discover new knowledge.
The analysis was an interactive process where the first author analysed the data, while the last author asked critical questions to facilitate competing interpretations. The process of categorizing and the abstraction and interpretation into themes were discussed among all the authors, and revisions were made.
By limiting the systematic search to publications written in English or Scandinavian languages, studies that could help refine the knowledge of patients' memories from the ICU may have been excluded. The search terms may have limited the results and thereby limited the explored memories. Due to these choices, we may have introduced selection bias. As this is a review of qualitative evidence and the reviewers were not primary investigators of the included studies, there might be nuances and subtleties that were not picked up that could have enriched and deepened the findings.
Nevertheless, the findings from our SR align with and reinforce the findings of the papers included.
Another limitation is related to the quality of the included studies; several studies did not state the researcher's context. Considering the validity of their results is difficult when the researcher's inevitable influence was not described. This implies that the findings presented in this review should be interpreted with some caution and might have limited transferability.

| CON CLUS ION
The unbalanced power relationship between nurses and the patients has potential to cause considerable distress for ICU patients.
Our findings highlight patients' dependency on nurses in the ICU and a high level of vulnerability. The majority of memories brought forth feelings of fear, whether of loss of control, the imminence of death or powerlessness. Memories of confusion and surreal delusions were prominent. The complicated relationship between care giver and care receiver, and its many facets are made prominent in our study. The importance of a reassuring relationship between nurse and patient was highlighted by the recollection of contrasting emotions of dependency and fear in relation to the provided care.
All the studies were from high-income countries, and although patient memories might be similar in other settings, future research is needed in other socioeconomic settings to explore the findings and to broaden our understanding.

ACK N OWLED G EM ENTS
We would like to acknowledge librarian Kari Larsen Mariussen for helping us with building the search strategy and conducting the searches in the databases.

CO N FLI C T O F I NTE R E S T
We have no conflict of interest.

AUTH O R CO NTR I B UTI O N S
CCMM, MTS, MHL and SAS: Substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data, drafting the manuscript or revising it critically for important intellectual content, final approval of the version to be published. Each author should have participated sufficiently in the work to take public responsibility for appropriate portions of the content, agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

E TH I C A L A PPROVA L
Ethical approval was not required since this study is a systematic review.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data availability is not relevant, since all data are available in original articles.