Intensive Care Med (2007) 33:978–985 DOI 10.1007/s00134-007-0600-8
C. Jones C. Bäckman M. Capuzzo H. Flaatten C. Rylander R. D. Griffiths
Received: 14 August 2006 Accepted: 28 February 2007 Published online: 24 March 2007 © Springer-Verlag 2007
C. Jones · R. D. Griffiths University of Liverpool, Intensive Care Research Group, Division of Metabolic & Cellular Medicine, School of Clinical Sciences, Faculty of Medicine, L69 3GA Liverpool, UK C. Jones (u) · R. D. Griffiths Whiston Hospital, Critical Care Unit, L35 5DR Prescot, UK e-mail:
[email protected] Tel.: +44-151-4261600; +44-151-4262382 Fax: +44-151-4301628 C. Bäckman Vrinnevisjukhuset, Intensive Care Unit, Norrköping, Sweden M. Capuzzo University Hospital of Ferrara, Section of Anaesthesia and Intensive Care Medicine, Ferrara, Italy H. Flaatten Haukeland University Hospital, Intensive Care Unit, Bergen, Norway
ORIGINAL
Precipitants of post-traumatic stress disorder following intensive care: a hypothesis generating study of diversity in care
C. Rylander Sahlgrenska University Hospital, ICU, Gothenburg, Sweden
from 3.2% to 14.8% in the different study ICUs. Independent of case mix and illness severity, the factors found to be related to the development of PTSD were recall of delusional memories, prolonged sedation, and Abstract Objective: This prospecphysical restraint with no sedation. tive observational study was designed Conclusion: The development of to explore the relationships bePTSD following critical illness is tween post-traumatic stress disorder associated with a number of different (PTSD), patients’ memories of the precipitating factors that are in part intensive care unit (ICU) and sedarelated to how patients are cared tion practices. Design: Prospective for within intensive care. This study multi-centre follow-up study out raises the hypothesis that the impact to 3 months after ICU discharge. of care within the ICU has an impact Setting: Two district general hospitals on subsequent psychological morbidand three teaching hospitals across ity and therefore must be assessed Europe. Patients and participants: in future studies looking at the way Two hundred and thirty-eight recovpatients are sedated in the ICU and ering, post-ventilated ICU patients. how physical restraint is used. Interventions: None. Measurements and results: Assessment of patients’ Keywords ICU · Critical illness · memories of ICU was undertaken Delusional memories · Physical at 1–2 weeks post ICU discharge. restraint · PTSD Patients’ psychological recovery was assessed by examining the level of PTSD-related symptoms and rate of PTSD by 3 months post ICU. The rate of defined PTSD was 9.2%, ranging
Introduction A significant proportion of patients have severe psychological problems following critical illness, with rates of post-traumatic stress disorder (PTSD) at 4–25% [1–4]. For a diagnosis of PTSD to be made, the individual must have experienced or witnessed a traumatic event and felt terror. The disorder has three symptom groups: reexperiencing,
e.g. nightmares; avoidance, e.g. blocking thinking about the event; and physiological arousal, e.g. sleeplessness. Finally, the symptoms must cause significant distress and impair day-to-day functioning [5]. Some of the ICU PTSD studies referenced above have examined only PTSD-related symptoms, while others have looked at the full diagnosis. While critically ill patients are ventilated they may receive considerable doses of opiates/sedative
979
drugs, and the possible long-term psychological effects are not properly understood. Many patients do not recall their stay in ICU and yet can clearly recall nightmares or hallucinations [6]. One investigation suggested that having sedation breaks may reduce the incidence of PTSD, but this was not a prospective study [7]. Padded straps to restrain patients are used as an alternative to sedation in some European countries. Opiates to relieve pain are, however, given in conjunction with restraint. The psychological impact of physical restraint without sedation is not clear. A previous study showed that the patients displaying severe PTSD-related symptoms were those who had no factual recall of ICU but still suffered vivid delusional memories of events in ICU, such as staff trying to kill them [8]. What is not clear is how such delusional memories are formed and what possible influence is played by sedation, opiates or physical restraint. This prospective study was designed to explore the relationships between psychological morbidity, memories of ICU and sedation practices.
Methods A common survey technique was used in five adult ICUs across Europe with experience in ICU follow-up (RACHEL group). The number of general adult ICU beds in the units ranged from 8 to 12, with nurse:patient ratios from 1:2 – in one unit – to 1:1. All the study units were mixed general ICUs. By intention the units had differing case mixes reflecting the diversity of adult ICU practice across Europe. Each individual hospital’s local research ethics committee approved the study protocol. Inclusion and exclusion criteria Patients 18 years old and over who were ventilated and had an ICU stay of 48 h or more were candidates for inclusion in the study. While they were on ICU their relatives were approached for assent to testing the patients for delirium and drug withdrawal reactions. The patients themselves were informed of this afterwards on the general wards and asked for their written consent to use this information and for 3 months’ follow-up. Patients admitted following a suicide attempt were excluded from the study; this was because individuals with PTSD have a significant suicide risk [9], so PTSD may have pre-dated ICU admission Patients with a pre-existing or concomitant psychotic illness, e.g. schizophrenia, were also excluded due to the significant rate of PTSD in this population [10]. Patients with a history of minor psychiatric problems such as anxiety or depression were
not excluded, but such problems were recorded and included in the building of the PTSD model, as they may increase vulnerability to PTSD [11]. Finally, patients living more than 30 km from the hospital or already in another research study were excluded. All patients consecutively admitted to the participating ICUs during 9 months were screened for enrolment in the study. Data collection Patient status and treatment in the ICU While the patient was in ICU a record was made of the following: 1. Sedative and opiate drug given, type, duration, total daily and peak dosage. To provide equipotent doses for analysis, dosages of all opioids were converted to morphine equivalents and benzodiazepines to lorazepam equivalents (see Table 1) [12]. 2. Level of sedation, using the Motor Activity Assessment Scale (MAAS) [13]. 3. The presence of any withdrawal symptoms, using a combination of signs and symptoms of opioid and/or benzodiazepine withdrawal according to Cammarano (1998) [12]. 4. The presence and duration of delirium once sedation had been stopped, using the Confusion Assessment Method (CAM-ICU) [14]. 5. Use and duration of any physical restraint. 6. Descriptors of the patient that may contribute to development of delirium, e.g. smoking history (smoker, previous smoker, never smoker), premorbid hypertension (drug treatment) [15] and alcohol or drug abuse. 7. Descriptors of the patients and ICU stay: gender, age, length of stay, duration of artificial ventilation, diagnostic category, admission type, APACHE II severity score.
Table 1 Factors for conversion of opioids and benzodiazepines to morphine and lorazepam equivalents [12] Drug Opioids Morphine (1 mg i.v.) Ketobemidon (1 mg i.v.) Heroin (1 mg i.v.) Alfentanil (1 mg i.v.) Sufentanil (1 mg i.v.) Fentanyl (1 mg i.v.) Benzodiazepines Lorazepam (1 mg i.v.) Midazolam (1 mg i.v.) Diazepam (1 mg i.v.)
Relative potency 1 1 2 10 50 100 1 1/3 1/5
980
Memories and symptoms after ICU discharge
PTSD Tools
1. At 1–2 weeks post ICU discharge the patients’ recall of the ICU experience was assessed using the ICU Memory Tool [16]. This consists of 14 items examining recall of hospital and ICU admission and discharge. A checklist of memories is included, falling into three groups: factual memories, such as the presence of an endotracheal tube; memories of feelings such as anxiety or pain; and delusional memories, e.g. nightmares or paranoid delusions. This time point was chosen for assessment of memories as most patients had recovered from delirium by then. It was also consistent with other studies conducted by the authors [8]. In addition, memories of ICU have been noted to change over time [8], and it was felt that questioning at this time point would capture those memories most likely to arouse PTSD-related symptoms during recovery. Patients were also asked about any previous psychological problems such as anxiety or depression. Research into PTSD following other traumatic events suggests that this may increase the risk of developing PTSD [11]. They were also asked whether they had experienced anything more traumatic then their stay in ICU, in the effort to detect any pre-existing PTSD. 2. At 2 months post ICU discharge patients were telephoned or visited at home and the PTSS-14 questionnaire was administered to assess the level of PTSD symptoms [17]. 3. At three months post ICU discharge, patients attended a follow-up appointment and the PTSS-14 was repeated with the Posttraumatic Diagnostic Scale (PDS) [18]. The PTSS-14 was repeated to examine the change in memories for ICU over time.
The PTSS-14 was developed from the PTSS-10 and has been previously validated to measure the level of PTSDrelated symptoms following critical illness [17]. The questionnaire consists of 2 parts: Part A is used to identify the presence of traumatic memories of ICU experience; allowing the recording of the patients’ recall of nightmares, pain, anxiety or breathlessness from ICU. Part B consists of 14 items scoring from 1 to 7 each, identifying the level of PTSD-related symptoms. The PDS is an interview tool and allows a diagnosis of PTSD to be made. It is one of the few measures in the PTSD literature that assesses all criteria, including functional impairment [19]. In addition, the PDS makes an assessment of previous traumatic events, for example assault or combat etc., and allows the recognition of pre-existing PTSD. Those patients found to have pre-existing PTSD were excluded from the analysis. All tools were translated into Swedish, Norwegian and Italian and validated by back-translation. Statistical analysis Analysis was performed using SPSS for Windows 12.0 and EQS 6.1 for Windows. Questionnaire data were treated as ordinal and analysed using non-parametric statistics [20], with the exception of scores derived from multi-item scales. In these cases the variances of the group being compared were checked for equality (Levene’s test) and parametric tests were applied. Structural equation
Table 2 Recruitment and drop-outs across the study centres. PTSD is defined as preadmission when symptoms were reported by the patients beginning more than 6 months before the assessment and ICU-related when beginning after ICU stay
Total admissions staying ≥ 48 h (n) Died in ICU Refused Excluded Age < 16 years Distance too great for follow-up Missed (staff leave) Already in another study Recruited to study Withdrew Died post ICU discharge Lost to follow-up Completed follow-up PTSD Preadmission PTSD (ICU)
Whiston
Norrk¨oping
Centre Bergen
Gothenburg
Ferrara
143 54 (38%) 17 (12%) 20 (14%) 0 0 0 0 52 (36%) 0 0 2 (3.8%) 50 (96%)
81 12 (15%) 0 12 (15%) 5 (6%) 0 14 (17%) 0 38 (47%) 0 7 (18%) 0 31 (82%)
79 10 (13%) 15 (19%) 0 0 0 0 0 54 (68%) 2 (4%) 13 (24%) 5 (9%) 34 (63%)
335 34 (10%) 8 (2.4%) 74 (22%) 16 (4.8%) 100 (30%) 46 (14%) 5 (1.5%) 52 (15%) 0 1 (2%) 9 (17%) 42 (81%)
161 31 (19%) 11 (7%) 11 (7%) 0 0 0 0 108 (67%) 0 17 (16%) 10 (9%) 81 (75%)
2 5 (10%)
0 1 (3.2%)
1 2 (5.9%)
0 2 (4.8%)
(Data are reported as numbers of patients and percentages)
4 12 (14.8%)
981
modelling was used to model possible precipitants of PTSD [21]. Structural equation modelling concentrates on the pattern of covariation between variables and goes beyond the restrictions of logistic regression, and is used extensively in psychological research. Model fit is signified by a p-value greater than 0.5 and a root mean square error of approximation (RMSEA) of less than 0.05. Structural equation modelling looks for a statistically non-significant result as it looks for the closeness of the model fit [22].
Fig. 1 CONSORT diagram study recruitment
Results Recruitment at the five ICUs ran over 9 months with 3 months for follow-up, with staggered start dates (to ease training and supervision), from 2003 to 2005. During the study period 799 patients were admitted and had an ICU stay of at least 48 h and were ventilated. Of the 799, 141 died in ICU, 51 refused to take part and 238 were excluded because of previous psychotic illness, attempted suicide, age < 16 years, distance too great for follow-up
982
or unresolved confusion. A total of 304 patients were recruited into the study, of whom 238 completed their 3-month follow-up questionnaires (see Table 2 for centre breakdown and CONSORT diagram, Fig. 1). The case mix of the admissions to the various centres varied in terms of the proportion of emergency admissions (55–100%; chi-square p = 0.001), age (median 54–73 years; Kruskal–Wallis p = 0.0001) and APACHE II scores (median 12–15; Kruskal–Wallis p = 0.004). The rate of PTSD was slightly lower where the ICU admission was elective—4 (7.6%) patients out of 52 versus 18 (11.6%) out of 154 emergency admissions—but the difference was not statistically significant (p = 0.66). Those patients admitted to ICU following trauma were not more likely to develop PTSD (p = 0.54). In addition, the presence of previous psychological problems differed between study centres (9–52%; p = 0.0001) and previous traumatic events prior to ICU admission on the PDS (p = 0.014). Despite this the rate of pre-existing PTSD was low (Table 3). The sedation practices between study units differed both in the mixture of sedative and opiate drugs used and in the doses given, with median daily lorazepamequivalent doses varying between 0 mg and 54 mg per 24 h (Kruskal–Wallis p > 0.001) and propofol dosages between 0 mg and 4,260 mg per 24 h (p > 0.0001) (see Table 1 for equivalences). The occurrence of withdrawal symptoms from sedative and opiate medication varied (1–48%; p > 0.0001). The incidence of delirium also varied, between 14% and 65% (p > 0.0001), median 41%. The development of delirium was more common in patients receiving high daily doses of benzodiazepines for
sedation (median dose 24 mg vs. 13 mg; Kolmogorov– Smirnov Z p = 0.003). The median lorazepam-equivalent dose across all patients receiving benzodiazepines was only 4.1 mg. No individual benzodiazepine was found to be related to the development of delirium, however, a number of different benzodiazepines were used and some patients received more than one in a day. Patients receiving high daily doses of opiates were also more likely to develop delirium (median dose 88 mg vs. 43 mg; p = 0.039). Those receiving high daily doses of propofol as their main sedative drug were not more likely to become delirious (p = 0.634). Those patients showing withdrawal symptoms from sedation/analgesia were more likely to be found to be delirious (25 out of 32 patients; p < 0.0001), with withdrawal symptoms lasting from 1 to 6 days and 32 (13.2%) of patients showing withdrawal symptoms. None of the variables recorded about the patients’ premorbid lifestyle were found to be associated with the development of delirium, i.e. a previous history of smoking (p = 0.37), premorbid hypertension (p = 0.37), drug abuse (p = 0.72) or alcohol abuse (p = 0.66). The patient’s recall of the ICU experience varied between study centres. Recall of delusional memories was common, varying between 44% and 77% (median 57%). Recall of feelings such as pain was not consistent over the follow-up period, with some patients reporting remembering pain at 2 weeks post ICU but not at 2 and 3 months, and vice versa. Agreement between the different test time points was seen in only 54% of patients. A similar pattern was seen with recall of anxiety in ICU at the three time points, with only 53% agreement. Only one patient who consistently reported
Table 3 Patient characteristics according to the ICU of admission Centre
Whiston
Norrk¨oping
Bergen
Gothenburg
Ferrara
All units
Nurse:patient ratio Age, years (median, range)∗ APACHE II (median, range)# ICU stay, days (median, range)∗ Days of ventilation (median, range)∗ Emergency admissions∗ Male/female Previous trauma# Previous psychological problems* Diagnostic groups Cardiac Gastrointestinal surgery Multiple Organ Failure Neurological Respiratory Trauma
1:1 60 (19–80) 15.5 (7–30) 10 (2–52) 7 (1–90) 52 (100%) 32:20 30 (58%) 20 (40%)
1:1 55 (22–85) 18 (3–36) 13 (3–49) 7.6 (1–47) 29 (93%) 19:12 21 (68%) 16 (52%)
1:1 54 (19–80) 16 (5–28) 9 (3–36) 5 (1–60) 33 (97%) 19:15 12 (35%) 3 (9%)
1:1 63 (24–84) 19 (7–33) 5 (2–49) 2.5 (1–38.7) 26 (60%) 27:16 29 (67%) 6 (14%)
1:2 73 (23–83) 13 (10–31) 5 (2–76) 1.67 (1–69.2) 44 (55%) 52:29 55 (68%) 16 (20%)
– 61 (17–86) 16 (3–36) 7 (2–76) 3 (2–90) 184 (91%) 149:92 147 (59%) 61 (24.4%)
3 18 4 4 19 4 52
3 14 1 3 6 4 31
2 3 6 6 15 2 34
8 16 6 6 12 4 52
15 31 6 0 25 4 81
31 82 23 19 77 18 250
Values for emergency admissions, gender, previous trauma, previous psychological problems and diagnostic categories are reported as absolute numbers with percentages in parentheses ∗ Chi-square/Kruskal–Wallis two-tailed p > 0.0001 # two-tailed p = 0.016
983
Fig. 2 Structural modelling of factors associated with the development of PTSD
recall of pain at all time points and three patients con- Previous history of psychological problems sistently reporting recall of anxiety were diagnosed with PTSD. These patients also reported recalling delusional Those patients with a previous history of psychological problems were more likely to receive prolonged sedation memories. (Mann–Whitney U, p = 0.0001) despite this history not necessarily being known while the patient was in ICU. Structural equation modelling They were also more likely to recall delusional memories (p = 0.0001). A number of models were tested to best predict the development of PTSD post ICU. In order to use structural equation modelling, latent constructs of the observed data were Physical restraint constructed. The hypothesized structural model contained paths: Those patients subjected to restraint who developed PTSD 1. From a previous history of psychological problems had longer periods of agitation, defined as a MAAS score to prolonged sedation and opiates, to recall of delusional of greater than 4: 4 days versus 1 day (Mann–Whitney U memories for ICU to PTSD. p = 0.039). Two (40%) had a history of previous psycho2. From prolonged sedation and opiates to PTSD. logical problems, while four (80%) had experienced previ3. From physical restraint with little or no sedation to ous traumatic events. PTSD. This model showed a good fit to the data (Fig. 2) [chi-square = 7.88, p = 0.72, RMSEA = 0.0001 (90% Discussion confidence interval 0.0001–0.05); comparative fit = 1.00, GFI 0.90, AGFI = 0.75]. R2 values are shown on Fig. 2 The main findings of the study are that the aetiology of the development of PTSD following critical illness is multiabove each variable. factorial. In addition, there may be an individual patient susceptibility through a previous history of psychological problems. This fits with other studies looking at PTSD in Competing models ICU patients [23]. The low incidence of PTSD overall in The hypothesized model was tested against three compet- this study may be due to the use of an interview tool to ing models and best fitted the data. As there was a differ- diagnose the disorder rather than a cut-off point on ence in case mix between the study centres, APACHE II a screening tool as previously used. In addition, the PDS score and age were included in the initial model, but there allows the recognition of pre-existing PTSD which would otherwise have been attributed to the ICU stay. was no fit to the data and so they were removed.
984
One factor that makes this population of patients fairly unusual is that the mechanism of the PTSD was through recall of delusional memories rather than factual traumatic memories. Many delusional memories were of events in ICU that were misinterpreted by the patient at the time, e.g. staff trying to kill them by injection. This loss of safety makes these memories particularly traumatic. Interestingly, patients with a history of previous psychological problems, often unknown to the ICU staff at the time of admission, received more sedation than those with no history, although the reason for this is not apparent. It may be that staff were responding to physiological expressions of anxiety, and this is worthy of further research. The study could be criticised for failing to document the use of psychotropic drugs prior to ICU admission, as the acute cessation of normal antidepressant medication while patients are critically ill may lead to withdrawal symptoms such as anxiety, agitation and nightmares [24]. In hindsight this information would have been helpful, but in reality it is difficult to achieve with emergency admissions. The finding that high doses of benzodiazepine increased the risk of delirium agrees with a recently published study showing that the use of lorazepam is an independent risk factor for the development of delirium in ICU patients [25]. However, no individual benzodiazepine, such as lorazepam, was found to have a particular association with the development of delirium and this may reflect the wide range of drugs used. The lack of a relationship between premorbid smoking and delirium could be explained by the widespread use of nicotine patches in the study units. The finding that physical restraint with no sedation predisposed patients to developing PTSD cannot be explained by patients finding the recall of this experience traumatic, as only one patient could recall being restrained when questioned 2 weeks post ICU discharge. Just over half of these patients could recall delusional memories. Factors such as age, length of ICU stay and number of hours ventilated did not explain the increased rate in restrained patients. The only factor which differed between those restraint patients who developed PTSD and those who did not was the length of time they were agitated (MAAS score > 4). The data on delirium for these patients are not complete, but it would seem that some patients were restrained who were not delirious when they were formally tested. Without these patients Ferrara would have had a PTSD rate of 11.1%, closer to the rate in the rest of the study units. The differing recall of memories of pain during the stay in ICU at 2 weeks, 2 months and 3 months draws into question the results of earlier studies of patients with adult respiratory distress syndrome (ARDS), where the recall of pain was one of the aversive memories found to increase the risk of PTSD [26]. It is possible that the re-
call of pain may be influenced by the patient’s recall of the time in hospital after the ICU stay or after hospital discharge. Data collected in dedicated ICU outpatient clinics have shown that some patients have significant problems with pain, particularly in their joints, during their recovery and this may colour their recall of pain while in ICU [27]. This study was observational and designed to be hypothesis-generating, since earlier work had highlighted a number of associations and possible causations but these had been examined in particular small or local populations. This study recruited a large number of patients across five European ICUs with a diversity of case mix, with the intention of reflecting diversity of care, and identifies some common factors. It is apparent from the different approaches to analgesia-sedation, choice of agents and the depths desired with the use or not of physical or chemical restraint that we achieved a cross-section of European practice. The factors identified are intimately connected with everyday patient care, and that such diversity exists just across five European ICUs suggests it is far from clear what is the ideal approach to analgesia-sedation or use of restraint. This study does not lead to the formulation of any suggestions for change in practice since it has a number of limitations. Foremost, this is not a randomized controlled study but only observational to raise hypotheses, since change and comparison of different practices have not been prospectively studied. In order to obtain reliable data the research group was restricted to units with experience of ICU follow-up and care after ICU. These ICUs may or may not be typical, and their care could potentially already be affecting the incidence of PTSD. One of the study ICUs (W) had already seen a decrease in the rate of PTSD from 16% in an earlier study [14] to 10% in this study, and this may reflect changes in care during critical illness and a greater role for psychological support after ICU. Although this study is not small, a far larger study might identify or clarify the significance of the factors identified. However, we also realize that in such an observational study the factors we identify may not be causal, and therefore amenable to direct intervention, but merely markers of disease or other processes. Inevitably over this time frame a number of patients are lost to follow-up and the final data have a potential for distortion. As the CONSORT diagram shows, we were able to analyse data in 78% of recruited patients with only 9.2% withdrawing or becoming lost to follow-up, the difference due to patient deaths post-ICU. It is not possible to quantify the effect of exclusion criteria or patients missed to investigator absence. A limitation of recruiting patients at difference study sites is the problem of consistency of data collection. To negate this problem the different sites had staggered start dates to allow the lead investigator to visit each centre and
985
train the staff involved. This meant that the study ran for over a year and also reduced the risk of a seasonal effect. In conclusion, this study showed a 1:10 risk for developing PTSD. This risk may be affected by the nature of ICU care. Studies are needed in this area to understand whether the risk of developing PTSD can be reduced through changes within the ICU or whether emphasis
should be put on recognition and help for patients after ICU discharge. Acknowledgements. Funding from the Stanley Thomas Johnson Charitable Foundation. We would like to acknowledge the hard work of our data collectors: Tracy Slater, Neven Ivcic, Anne-Khristine Muri, Ulla Löwenmark, Eva Rosell, Wiveka Etemad, Mia Hending.
References 1. Jones C, Twigg E, Lurie A, McDougall M, Heslett R, HewittSymonds M, Griffiths RD (2003) The challenge of diagnosis of stress reactions following intensive care and early intervention: a review. Clinical Intensive Care 14:83–89 2. Schynder U, Morgeli H, Nigg C, Klaghofer R, Renner N, Trentz O, Buddeberg C (2000) Early psychological reactions to life-threatening situations. Crit Care Med 28:86–92 3. Cuthbertson BH, Hull A, Strachan M, Scott J (2004) Post Traumatic Stress Disorder after critical illness requiring general intensive care. Intensive Care Med 30:450–455 4. Capuzzo M, Valpondi V, Cingolani E, Gianstefani G, De Luca S, Grassi L, Alvisi R (2005) Posttraumatic stress disorder-related symptoms after intensive care. Minerva Anaesthesiol 71:167–179 5. American Psychiatric Association (2000) Diagnostic and statistical manual of mental disorders (4th edn., text rev.). American Psychiatric Association, Washington, DC 6. Jones C, Griffiths RD, Humphris G (2000) Disturbed memory and amnesia related to intensive care. Memory 8:79–94 7. Kress JP, Gehlbach B, Lacy M, Pliskin N, Pohlman AS, Hall JB (2003) The long-term psychological effects of daily sedative interruption on critically ill patients. Am J Respir Crit Care Med 168:1457–1461 8. Jones C, Griffiths RD, Humphris G, Skirrow PM (2001) Memory, delusions, and the development of acute posttraumatic stress disorder-related symptoms after intensive care. Critical Care Med 29:573–580 9. Courtois CA, Bloom SL (2000) Inpatient treatment. In: EB Foa, TM Keane, MJ Friedman (eds) Effective treatments for PTSD. Guilford Press, New York, p 218
10. Williams-Keeler L, Milliken H, Jones B (1994) Psychosis as precipitating trauma for PTSD: a treatment strategy. Am J Orthopsychiatry 64:493–498 11. Yehuda R (1999) Risk factors for post traumatic stress disorder. American Psychiatric Press Inc, Washington, DC 12. Cammarano WB, Pittet J-F, Weitz S, Schlobohm RM, Marks JD (1998) Acute withdrawal syndrome related to the administration of analgesic and sedative medications in adult intensive care unit patients. Critical Care Med 26:676–684 13. Devlin JW, Boleski G, Mlynarek M, Nerenz DR, Peterson E, Jankowski M, Horst HM, Zarowitz BJ (1999) MAAS: a valid and reliable sedation scale for use with mechanically ventilated patients in an adult surgical intensive care unit. Crit Care Med 27:1271–1275 14. Ely W, Margolin R, Francis J, May L, Truamn B, Dittus R, Speroff T, Gautam S, Margolin R, Hart RP, Bernard GR, Inouye SK (2001) Evaluation of Delirium in Critically Ill Patients: Validation of the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU). Crit Care Med 29:1370–1379 15. Dubois MJ, Bergeron N, Dumont M, Dial S, Skrobik Y (2001) Delirium in an intensive care unit: a study of risk factors. Intensive Care Med 27:1297–1304 16. Jones C, Humphris GH, Griffiths RD (2000) Preliminary validation of the ICUM tool: a tool for assessing memory of the intensive care experience. Clinical Intensive Care 11:251–255 17. Twigg E, Jones C, McDougall M, Griffiths RD, Humphris GH (2003) Early screening tool for post traumatic stress disorder following critical illness. Proceedings of the Intensive Care Society 540P 18. Foa EB, Cashman L, Jaycox L, Perry K (1997) The validation of a self-reported measure of posttraumatic stress disorder: The Posttraumatic Diagnostic Scale. Psych Assess 9:445–451
19. Elhai JD, Gray MJ, Kashdan TB, Franklin CL (2005) Which instruments are most commonly used to assess traumatic event exposure and posttraumatic effects? A Survey of traumatic stress professionals. J Traumatic Stress 18:541–545 20. Cohen J (1969) Statistical power analysis for the behavioural sciences. Academic Press, New York 21. Shipley B (2002) Cause and correlation in biology. A user’s guide to path analysis, structural equations and causal inference. Cambridge University Press, Cambridge 22. Miles J, Shevlin M (2003) Structural equation modelling: navigating spaghetti junction. Psychologist 16:639–641 23. Hull AM, Cuthbertson BH (2005) Life interrupted: risk factors for posttraumatic reactions. In: Ridley S (ed) Critical care focus: the psychological challenges of intensive care. Blackwell, Oxford, pp 9–19 24. Anderson IM, Nutt DJ, Deakin JFW (2000) Evidence-based guidelines for treating depressive disorders with antidepressants: a revision of the 1993 British Association for Psychopharmacology guidelines. J Psychopharmacol 14:3–20 25. Pandharipande P, Shintani A, Peterson J, Truman Pun B, Wilkinson GR, Dittus RS, Bernard GR, Ely EW (2006) Lorazepam as an independent risk factor for delirium in intensive care unit patients. Anesthesiology 104:21–26 26. Schelling G, Stoll C, Meier M, Haller M, Briegel J, Manert W, Hummel T, Lenhart A, Heyduck M, Polasek J, Meier M, Preuß U, Bullinger M, Schüffel W, Peter K (1998) Healthrelated quality of life and posttraumatic stress disorder in survivors of adult respiratory distress syndrome. Critical Care Med 26:651–659 27. Jones C, Griffiths RD (2000) Identifying post intensive care patients who may need physical rehabilitation. Clin Intensive Care 11:35–38
