Noninvasive mechanical ventilation in patients having ... - Springer Link

3 downloads 121 Views 577KB Size Report
Nov 27, 2012 - Jérôme Lambert. Ferhat Meziani. Matthieu Schmidt. David Schnell ... Surgical ICU, Hotel Dieu University. Hospital, Clermont-Ferrand, France.
Intensive Care Med (2013) 39:292–301 DOI 10.1007/s00134-012-2746-2

ORIGINAL

E´lie Azoulay Achille Kouatchet Samir Jaber Je´roˆme Lambert Ferhat Meziani Matthieu Schmidt David Schnell Satar Mortaza Matthieu Conseil Xavier Tchenio Patrick Herbecq Pierre Andrivet Emmanuel Guerot Ariane Lafabrie Se´bastien Perbet Laurent Camous Ralf Janssen-Langenstein Franc¸ois Collet Jonathan Messika Ste´phane Legriel Xavier Fabre Olivier Guisset Samia Touati Sarah Kilani Michael Alves Alain Mercat Thomas Similowski Laurent Papazian Anne-Pascale Meert Sylvie Chevret Benoıˆt Schlemmer Laurent Brochard Alexandre Demoule

Noninvasive mechanical ventilation in patients having declined tracheal intubation

Received: 19 July 2012 Accepted: 22 October 2012 Published online: 27 November 2012 Ó Springer-Verlag Berlin Heidelberg and ESICM 2012

E´. Azoulay ())  D. Schnell  A. Lafabrie  B. Schlemmer Medical ICU, Saint-Louis Hospital AP-HP, Medical ICU, Faculte´ de me´decine, Universite´ Paris-Diderot, Sorbonne Paris-Cite´, 1 avenue Claude Vellefaux, 75010 Paris, France e-mail: [email protected] Tel.: ?33-142-499421 Fax: ?33-142-499426

For the REVA and the FAMIREA study groups. The collaborators are listed in the Appendix. Electronic supplementary material The online version of this article (doi:10.1007/s00134-012-2746-2) contains supplementary material, which is available to authorized users.

A. Kouatchet  S. Mortaza  A. Mercat Medical ICU, Angers Hospital, Angers, France

S. Jaber  M. Conseil Surgical ICU, Saint-Eloi University Hospital, Montpellier, France J. Lambert  S. Chevret Biostatistics Department, Saint-Louis Hospital and Paris 7 University, Paris, France F. Meziani  R. Janssen-Langenstein Medical ICU, Central University Hospital, Strasbourg, France M. Schmidt  T. Similowski  A. Demoule Pulmonary and Critical Care Department, La Pitie´-Salpeˆtrie`re University Hospital, Paris, France

293

X. Tchenio Bourg en Bresse ICU, Bourg en Bresse Hospital, Bourg en Bresse, France P. Herbecq Roubaix ICU, Roubaix Hospital, Roubaix, France P. Andrivet Bligny ICU, Bligny Medical Center, Briis-sous-Forges, France E. Guerot Medical ICU, Pompidou Hospital, Paris, France S. Perbet Surgical ICU, Hotel Dieu University Hospital, Clermont-Ferrand, France L. Camous Medical ICU, Biceˆtre University Hospital, Le Kremlin-Biceˆtre, France F. Collet St-Malo ICU, Saint-Malo Hospital, SaintMalo, France J. Messika Tenon ICU, Tenon University Hospital, Paris, France S. Legriel Versailles ICU, Versailles Hospital, Versailles, France X. Fabre Roanne ICU, Roanne Hospital, Roanne, France O. Guisset Bordeaux ICU, Saint-Andre´ Hospital, Bordeaux, France

S. Touati Saint-Joseph Hospital, Paris, France

posttraumatic stress disorder-related symptoms, and symptoms of anxiety and depression. Post-ICU burden was S. Kilani compared between DNI patients and Quimper ICU, Bretagne Sud Hospital, patients receiving NIV with no treatLorient, France ment-limitation decisions (TLD). Of 780 NIV patients, 574 received NIV M. Alves Saint-Antoine ICU, Saint-Antoine with no TLD, and 134 had DNI University Hospital, Paris, France orders. Hospital mortality was 44 % in DNI patients and 12 % in the noL. Papazian TLD group. Mortality in the DNI Marseille Nord ICU, North University group was lowest in COPD patients Hospital, Marseille, France compared to other patients in the DNI group (34 vs. 51 %, P = 0.01). In the A.-P. Meert Brussels ICU, Jules Bordet Institute, DNI group, HRQOL showed no sigBrussels, Belgium nificant decline on day 90 compared to baseline; day-90 data of patients L. Brochard and relatives did not differ from those Geneva ICU, Hoˆpitaux universitaires in the no-TLD group. Conclude Gene`ve, Geneva, Switzerland sions: Do-not-intubate status was present among one-fifth of ICU Abstract Purpose: Noninvasive patients who received NIV. DNI ventilation (NIV) is a treatment patients who were alive on day 90 option in patients with acute respira- experienced no decrease in HRQOL tory failure who are good candidates compared to baseline. The prevafor intensive care but have declined lences of anxiety, depression, and tracheal intubation. The aim of our PTSD-related symptoms in these study was to report outcomes after patients and their relatives were simNIV in patients with a do-not-intubate ilar to those seen after NIV was used (DNI) order. Methods: Prospective as part of full-code management observational cohort study in all (clinicaltrial.govNCT01449331). patients who received NIV for acute respiratory failure in 54 ICUs in Keywords Mechanical ventilation  France and Belgium, in 2010/2011. Intubation  Palliative care  Results: Goals of care, comfort, and Quality of dying and death  vital status were assessed daily. On Family members  End-of-life  day 90, a telephone interview with Dyspnea  Breathlessness patients and relatives recorded healthrelated quality of life (HRQOL),

Introduction In patients nearing the end of life, acute respiratory failure and dyspnea are extremely common [1]. Causes include acute worsening of chronic conditions (e.g., COPD or lung fibrosis exacerbations) and progressive irreversible processes leading to asphyxia (e.g., progression of a malignancy) [2, 3]. In both situations, endotracheal mechanical ventilation is considered a non-beneficial intervention that deprives patients of their dignity and ability to recognize that family members are present [4, 5]. Among patients with do-not-intubate orders (DNI), those who have reversible causes of dyspnea may benefit from etiological interventions, and noninvasive

mechanical ventilation (NIV) may therefore be warranted as a life-supporting intervention [6, 7]. However, existing studies of NIV in patients with treatment-limitation decisions (TLDs) failed to separate patients with DNI orders who received NIV with survival hopes and from patients under comfort care only who received NIV to alleviate their dyspnea, although these are two very different situations [8, 9]. In addition, most of these studies were conducted retrospectively [6, 10, 11], in a single center, in selected patients [10, 12–14], and focused on short-term outcomes [10, 13, 15]. None described the decision-making process for using NIV. Importantly, health-related quality of life (HRQOL) and symptoms of anxiety, depression, or post-traumatic stress disorder

294

(PTSD) were not evaluated; this drawback is important, as a major concern in DNI patients is that increased survival obtained with NIV might merely indicate prolongation of the dying process, that is, survival with a dismal quality of life and no hope for improvement. Finally, the post-ICU burden in family members of patients given NIV in DNI patients has not been assessed. The primary objective of this prospective cohort study in 54 ICUs in France and Belgium was to determine whether NIV in patients with acute respiratory failure and DNI decisions prolonged life only at the cost of a significant decline in HRQOL scores. To provide perspective, we recorded data from patients without TLDs who were managed with NIV during the same period in the same ICUs. We chose day 90 as the time point for evaluating post-NIV outcomes including HRQOL scores in the patients; and for evaluating anxiety, depression, and PTSDrelated symptoms in both the patients and the relatives.

Patients and methods The study was approved by the institutional review board of the French-speaking Society for Respiratory Medicine (Socie´te´ de Pneumologie de Langue Franc¸aise). Written informed consent was obtained from all patients and relatives. Patients were recruited in 54 ICUs belonging to the REVA and FAMIREA networks in France and Belgium (eTable 1) [16, 17]. Participating ICUs included consecutive adults receiving NIV over a 2-month period between November 2010 and April 2011. All patients with acute respiratory failure who received NIV during the ICU stay were included and evaluated daily in the ICU, at hospital discharge, and 90 days after ICU discharge. Every day, patients were classified into three groups according to NIV goals [8]: improved survival and no TLDs (no TLD group), improved survival in patients who had made DNI decisions but accepted other life-supporting measures (DNI group), and comfort care only. Patients were classified according to the highest group level comfort care group [ DNI group [ noTLD group). DNI patients were defined as patients who themselves declined tracheal intubation and those in whom the healthcare staff considered that tracheal intubation was not appropriate. All data shown in Tables 1 and 2 were collected prospectively. Mortality after NIV was determined overall and within four predefined subgroups: chronic respiratory insufficiency, heart failure, pneumonia, and cancer [13, 15]. The Simplified Acute Physiology Score (SAPS 2) [18] and SOFA score were computed at ICU admission [19]. Poor chronic health status was defined as being mostly confined to the bed or chair or having severe disabilities [20]. The dates of hospital discharge and death were recorded.

Within 5 days after ICU admission, patients completed the SF-36 questionnaire by referring to the past 3 months [21]. Ninety days after ICU discharge, trained social workers coached by psychologists and sociologists of the FAMIREA study group interviewed survivors by telephone. For each patient, a single relative (the surrogate) was interviewed. Patients were asked to complete the SF36 questionnaire to assess HRQOL, Impact of Event Scale (IES) [22, 23] to assess PTSD-related symptoms, and Hospital Anxiety and Depression Scale (HADS) [24] to quantify symptoms of anxiety and depression, in that order. Lower HADS and IES scores indicate less post-ICU burden, but lower SF-36 scores indicate worse HRQOL. In addition, the relatives completed the HADS and IES during a separate telephone call. We used HADS anxiety and depression subscale cutoffs of 8 points; [17, 25] higher IES scores indicate greater symptom severity. A patient or relative was deemed unreachable after ten calls went unanswered. The database was audited via an independent check of all ICU variables in a random sample of 10 percent of patients. Statistical analysis Continuous variables were described as medians (interquartile range) and compared between DNI and no-TLD groups using the Wilcoxon rank-sum test. Categorical variables were described as proportions and 95 % confidence intervals (95 % CIs) and compared between the two NIV groups using the Pearson Chi-square test. Fisher’s exact test was used when appropriate. Our primary hypothesis was that NIV provided a substantial chance of survival in DNI patients, but at the price of a significant quality-of-life impairment, as assessed by the SF36 scores at ICU admission and on day 90. We assumed that a 20 % decrease in the SF36 score was clinically significant. To allow the detection of an at least 20 % decrease in the SF36 score with alpha set at 5 and 90 % power, based on a study where the two SF36 subscales were evaluated separately [26], eight patients were needed for the physical SF36 subscore (from a mean value of 28 and assuming a standard deviation of 7) and 19 for the mental SF36 subscore (from a mean value of 40 and assuming a standard deviation of 5). Therefore, 16–38 DNI survivors were needed.

Results Patient population Figure 1 is the patient flow-chart. Of the 708 patients who received NIV as a life-supporting intervention, 134 (18.9 %) were DNI patients. Half the DNI patients

295

Table 1 Patient characteristics at diagnosis of acute respiratory failure Numbers (%) or Median (interquartile range)

NoTLD patients (N = 574)

DNI patients (N = 134)

P value

Age (years) Female gender Co-morbidities Current smoker Heavy drinker Homeless Heart failure Chronic respiratory failurea Long-term oxygen therapy Chronic home NIV Cancer HIV infection Poor chronic health statusb DNI status existed at ICU admission Treatments implemented before ICU admission Steroids Aerosols Diuretics Direct admission to the ICU Time from hospital to ICU admission Clinical presentation at diagnosis Respiratory rate Lowest SpO2 Oxygen flow (l/min)

66 (57–76) 209 (36.4)

76 (65–83) 62 (46.3)

\0.0001 0.06

113 (19.7) 89 (15.5) 8 (1.4) 101 (17.6) 386 (67.2) 89 (15.5) 58 (10) 70 (12.2) 15 (2.6) 254 (44.2) 0

14 (10.5) 9 (6.8) 1 (0.7) 26 (19.4) 107 (79.8) 32 (24) 17 (13) 33 (24.6) 0 91 (68) 74 (55.2)

0.004 0.003 0.79 0.91 0.10 0.09 0.72 0.03 0.23 0.0001 \0.0001

97 (17) 217 (41) 137 (24) 270 (55.4) 0 (0–3)

22 (16) 68 (51) 46 (34) 66 (52.4) 0 (0–3)

0.62 0.06 0.08 0.93 0.94

32 (27–37) 88 (80–92) 8 (3–15)

30 (25–35) 88 (80–92) 7 (3–15)

0.22 0.70 0.54

The NIV group was assessed by the ICU clinicians according to the pulmonary disease, HIV human immunodeficiency virus, ICU classification developed by Curtis et al. [8] at the time of NIV intensive care unit, SpO2 oxygen saturation by pulse oximetry a initiation and daily thereafter COPD or restrictive or mixed respiratory insufficiency TLD treatment-limitation decision, DNI do-not-intubate, NIV b Bedridden or severely disabled patients noninvasive mechanical ventilation, COPD chronic obstructive

acquired this status during the ICU stay, a median of 2 (1–7) days after ICU admission. In nine patients, DNI status was changed to NIV with no TLD. Table 1 reports the main patient characteristics. DNI patients were older and were more frequently smokers or heavy drinkers; they more often had cancer or a poor chronic health status. The number of failing organs and the proportion of patients receiving life-supporting interventions was not different between the no-TLD and the DNI groups. In the DNI group, acute respiratory failure was more often due to COPD exacerbation than in the no-TLD group. Characteristics of the decision-making processes for noninvasive ventilation in conjunction with treatment-limitation decisions are reported in etable 2. Median overall NIV duration was 2 days. NIV durations per day were 5 (3–9) hours in the no-TLD group and 6 (3–10) hours in the DNI group (P = 0.43). DNI patients refused NIV in 5.2 % of the cases and patients without TLDs refused NIV in 4.3 % of the cases. Table 2 reports on patient’s comfort, as assessed by ICU nurses throughout the ICU stay. Sleep quality and NIV tolerance were similar in the two groups. Also, a similar proportion of patients from the DNI and the no-TLD groups were receiving analgesia (17.9 and 17 %) and were fasting (34 and 35 %).

Mortality Hospital survival was 46 % in the DNI group (Table 2; Fig. 1). Figure 2 reports hospital mortality according to the precipitating factor of acute respiratory failure. Of note, in the DNI group, mortality was lower in COPD patients than in other patients (34 vs. 51 %, P = 0.01). Mortality was also lower in COPD patients in the no-TLD group (7 %). Day-90 quality of life and post-ICU burden in do-not-intubate patients Overall, 154 patients and 175 relatives were interviewed on day 90. Among DNI patients, 25.4 % of patients and 39.6 % of relatives were interviewed (Fig. 1; Table 3). Among no-TLD patients, 41.8 % of patients and 42.5 % of relatives were interviewed. The interviews were conducted 109 (90–141) days after ICU discharge. Reasons for not performing the day-90 interview were as follows: in the DNI group, 60 % of patients had died, 21 % declined the interview, and 3 % were hospitalized; in the no-TLD group, 17.5 % of patients died before day 90, 26 % declined the interview (11 % initially and 15 % on

296

Table 2 ICU management and outcomes Numbers (%) or Median (interquartile range) Severity at ICU admission SAPS 2 score SOFA score Other treatments provided Vasopressors/dialysis Analgesia/sedation Daily evaluation of patient comfort by nurses (from day 1 to day 3) Sleep quality never assessed as ‘‘poor’’ Sleep quality assessed as ‘‘good’’ at least once Sleep time in hours (daily average) Good-to-excellent NIV tolerancea No oral nutrition Cause of acute respiratory failure COPD exacerbation Cardiogenic pulmonary edema Community-acquired pneumonia Other/unknown Lengths of ICU stay Hospital stay ICU mortality Hospital mortality Day-90 mortalityb

NoTLD patients (N = 574)

DNI patients (N = 134)

P value

36 (27–47) 4 (2–7)

41 (35–51) 4 (3–6)

\0.0001 0.22

22 (4)/19 (3) 98 (17)/19 (3.3)

5 (4)/2 (1.5) 24 (17.9)/5 (3.7)

0.70/0.22 0.55/0.87

65/336 (19.3) 138/336 (41.1) 4 (2–6) 166 (28.9) 142/409 (35)

14/88 (15.9) 33/88 (37.5) 4 (2–6) 43 (32) 36/105 (34)

0.81 0.22 0.57 0.17 0.49

276 (48) 67 (11.7) 104 (18.1) 44 (7.7)/83 (14.5)

80 18 19 10

0.04 0.28 0.11 0.77/0.11

6 (3–12) 16 (8–30) 45 (7.8) 69 (12) 100 (17.4)

6 (3–15) 15 (7–29) 37 (27.6) 59 (44) 81 (60.4)

(59.7) (13.4) (14.2) (7.5)/7 (5.2)

0.47 0.56 \0.0001 \0.0001 \0.0001

TLD treatment-limitation decisions, DNI do-not-intubate, COPD skin ulcers, or gastric distension (4.3 %); and anxiety (10.7 %). chronic obstructive pulmonary disease, NIV noninvasive mechani- NIV tolerance did not differ significantly across the groups b cal ventilation, ICU intensive care unit 16 % and 3 % of patients in the no-TLD and DNI groups were a Intolerance to NIV was assessed daily by nurses and included lost to follow-up leaks (3.7 %); nasal congestion or dryness (4 %); eye irritation,

Fig. 1 Patient flow chart

297

died had significantly more depressive and PTSD-related symptoms than relatives of patients who survived, in both NIV groups.

P