In-Hospital Mortality and Long-Term Survival of Patients With Acute Intoxication Admitted to the ICU* Raya Brandenburg, BSc1; Sylvia Brinkman, PhD2,3; Nicolette F. de Keizer, PhD2,3; Jan Meulenbelt, MD, PhD1,4,5; Dylan W. de Lange, MD, PhD1,3,4
Objective: To assess in-hospital and long-term mortality of Dutch ICU patients admitted with an acute intoxication. Design: Cohort of ICU admissions from a national ICU registry linked to records from an insurance claims database. Setting: Eighty-one ICUs (85% of all Dutch ICUs). Patients: Seven thousand three hundred thirty-one admissions between January 1, 2008, and October 1, 2011. Interventions: None. Measurements and Main Results: Kaplan-Meier curves were used to compare the unadjusted mortality of the total intoxicated population and for specific intoxication subgroups based on the Acute Physiology and Chronic Health Evaluation IV reasons for admission: 1) alcohol(s), 2) analgesics, 3) antidepressants, 4) street drugs, 5) sedatives, 6) poisoning (carbon monoxide, arsenic, or cyanide), 7) other toxins, and 8) combinations. The case-mix adjusted mortality was assessed by the odds ratio adjusted for age, gender, severity of illness, intubation status, recurrent intoxica-
*See also p. 1563. 1 Department of Intensive Care and Emergency Medicine, University Medical Center Utrecht, University of Utrecht, Utrecht, The Netherlands. 2 Department of Medical Informatics, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands. 3 National Intensive Care Evaluation Foundation, Amsterdam, The N etherlands. 4 Dutch National Poisons Information Center, University Medical Center, University of Utrecht, Utrecht, The Netherlands. 5 Institute for Risk Assessment Sciences, University of Utrecht, Utrecht, The Netherlands. Dr. Brinkman is employed by National Intensive Care Evaluation (NICE) (foundation pays the department of medical informatics to process and analyze data for the registry). Dr. de Keizer is employed by NICE (foundation pays the department of medical informatics to process and analyze data for the registry). Dr. de Lange served as a board member of the NICE foundation and received support for the development of educational presentations (lecturing fees by congress organizations). The remaining authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000245
Critical Care Medicine
tion, and several comorbidities. The ICU mortality was 1.2%, and the in-hospital mortality was 2.1%. The mortality 1, 3, 6, 12, and 24 months after ICU admission was 2.8%, 4.1%, 5.2%, 6.5%, and 9.3%, respectively. Street drugs had the highest mortality 2 years after ICU admission (12.3%); a combination of different intoxications had the lowest (6.3%). The adjusted observed mortality showed that intoxications with street drugs and “other toxins” have a significant higher mortality 1 month after ICU admission (odds ratioadj = 1.63 and odds ratioadj= 1.73, respectively). Intoxications with alcohol or antidepressants have a significant lower mortality 1 month after ICU admission (odds ratioadj = 0.50 and odds ratio= 0.46, respectively). These differences were not found in the adj adjusted mortality 3 months upward of ICU admission. Conclusions: Overall, the mortality 2 years after ICU admission is relatively low compared with other ICU admissions. The first 3 months after ICU admission there is a difference in mortality between the subgroups, not thereafter. Still, the difference between the in-hospital mortality and the mortality after 2 years is substantial. (Crit Care Med 2014; 42:1471–1479) Key Words: drug overdose; intensive care; intoxication; long-term outcome; mortality; poisoning
A
substantial proportion of patients with an acute intoxication who are seen at an emergency department (ED) are admitted to the ICU (3.7–40%) (1–3). Depending on the setting and the kind of ICU, 3.4–13.8% of the ICU population consists of patients with an acute intoxication (4, 5). However, the in-hospital mortality as a result of intoxications is strikingly low. In the Western hemisphere, both ICU mortality and in-hospital mortality of intoxicated patients are significantly lower than that of the general ICU population (0.2–4.0% vs 16.2–16.4%) (6–12). In the Netherlands, the reported in-hospital mortality of intoxicated patients is even lower: 0–1.4% (13–15). As shown, there are a lot of data about the in-hospital mortality of intoxicated ICU patients. However, we feel that long-term outcome should also be taken into account. Indeed, www.ccmjournal.org
1471
Brandenburg et al
little is achieved when a patient dies soon after hospital discharge, and for that reason, the long-term outcome should be taken into consideration (16). Additionally, long-term outcome reduces the bias of transfers and discharge timing that might influence the in-hospital mortality statistics (17). Also, analyzing long-term mortality could help to identify patients at high risk to die after hospital discharge and to provide these patients with specialized follow-up. Despite its importance, studies about the long-term outcome of intoxicated ICU patients are scarce. A Finnish study (1996) showed a low in-hospital mortality of only 3.6% but an excessive death rate thereafter. When these authors compared the patients with intoxications to the general population (matched for gender and age), the ratio between observed and predicted mortality was 6.3 after 5 years of follow-up (18). Intoxication with dissolvent (ethylene glycol and methanol) had the highest 5-year mortality (35.3%). However, this study distinguished only three main types of intoxications, sedatives and psychotropic drugs, dissolvent, and “other.” Additionally, this study was published in 1996 and the pattern of intoxications has changed significantly over the last years; the numbers of intoxications with γ-hydroxybutyrate (GHB) and antidepressants have increased, whereas the number of intoxications with benzodiazepines has decreased (19, 20). Within the area of critical care medicine, prognostic models such as Acute Physiology and Chronic Health Evaluation (APACHE) IV (21) are frequently used for case-mix adjustment. The APACHE IV model distinguishes seven types of intoxicants as a reason for ICU admission. The aim of this study is to use APACHE IV intoxication subgroups to assess and compare the (case-mix adjusted) in-hospital and long-term mortality of ICU patients admitted with acute intoxication.
MATERIALS AND METHODS Subjects All adult ICU patients admitted to a Dutch ICU between January 1, 2008, and October 1, 2011, with an intoxication reason for admission and satisfying the APACHE IV inclusion criteria were considered eligible. We used the APACHE IV intoxication reasons for ICU admissions: 1) alcohol, 2) analgesics (aspirin, acetaminophen), 3) antidepressants (e.g., cyclic antidepressants and lithium), 4) street drugs (opiates, cocaine, and amphetamine), 5) sedatives (hypnotics, antipsychotic, and benzodiazepines), 6) poisoning due to carbon monoxide, arsenic, or cyanide, 7) other toxins, and 8) combinations of these diagnosis, maximum two. The diagnosis is based on clinical suspicion and does not need to be laboratory confirmed for the APACHE IV inclusion criteria. Data Collection The Dutch National Intensive Care Evaluation (NICE) registry (22) contains demographic, physiological, and clinical data recorded in the first 24 hours of ICU admission of all consecutive ICU patients admitted to participating ICUs. It includes, 1472
www.ccmjournal.org
among others, the APACHE IV severity of illness score and reason for ICU admission (21), as well as the ICU and in-hospital mortality. During the study period, approximately 85% of all Dutch ICUs recorded data on all their admissions in the NICE registry. All participating ICUs are mixed medical-surgical units located in university hospitals (n = 7), teaching hospitals (n = 27), or nonteaching hospitals (n = 47). The mortality after hospital discharge is obtained by linking the NICE registry to a national insurance claims database of Vektis (23). The records from the NICE registry and insurance claims database are linked by a deterministic linkage algorithm (24) that used the hospital of admission, gender, date of birth, ICU admission date, and ICU discharge date. Details on the linkage process are described in more detail elsewhere (25). The insurance claims database contains information on the vital status of patients. The status of the patients (either alive or death) on January 1, 2012, and if relevant, the date of death, was extracted from the insurance claims database. The data used in this study have been encrypted in a way that all patient identifying information, such as name and patient identification number, has been removed. In the Netherlands, there is no need to obtain consent to make use of such registries without patient identifying information. The data are officially registered according to the Dutch Personal Data Protection Act. Statistical Analysis To assess in-hospital mortality and the mortality at 1, 3, 6, 12, and 24 months after ICU admission, we used Kaplan-Meier curves and logistic regression models. The latter is used, as the proportional hazard assumption of the more frequently used Cox model was not fulfilled. The case-mix adjusted mortality was analyzed by calculating the odds ratio (ORadj), while adjusting for age, gender, severity of illness expressed as the APACHE IV score, intubation status and the comorbidities chronic respiratory insufficiency, chronic cardiovascular insufficiency, chronic renal insufficiency, cerebrovascular accident, and the use of vasopressors. Patients who are readmitted to the ICU within the year of follow-up could have a new critical illness with accompanying risk of death or are admitted for another intoxication, implying a chronic use of substances. Therefore, we also adjusted for such recurrent intoxication. The ORadj and corresponding 95% CI were calculated for each ICU subgroup with the whole intoxicated ICU population (excluding the subgroup of interest) as reference. All statistical analyses were performed using PASW statistics 18 (SPSS, Chicago, IL) and the R environment for statistical computing version 2.15.1.
RESULTS From January 1, 2008, to October 1, 2011, a total of 246,053 patients were admitted to the participating Dutch ICUs, including 9,129 patients (3.7%) with at least one diagnosis of intoxication (Fig. 1). Of the 9,129 intoxicated patients, 8,176 records (89.6%) could be linked with the insurance claims database of Vektis. Finally, only the 7,331 patients fulfilling the APACHE IV inclusion criteria (22) were included in our analysis (Fig. 1). June 2014 • Volume 42 • Number 6
Clinical Investigations
Mortality of Intoxicated Patients A total of 85 patients died while being admitted to an ICU (ICU mortality rate, 1.2%), and another 69 patients died after ICU admission while they were still in the hospital (in-hospital mortality rate, 2.1%). The mortality at 1, 3, 6, 12, and 24 months after ICU admission for all patients with intoxications was 2.8%, 4.1%, 5.2%, 6.5%, and 9.3%, respectively. Figure 2 shows the unadjusted mortality of the whole intoxication population and of the intoxication subgroups in Kaplan-Meier curves. Table 3 shows the percentage of the long-term mortality for each APACHE IV subgroup. The ORadj for the i n-hospital and long-term mortality of the APACHE IV subgroups, compared with the whole intoxication population while excluding the intoxication subgroup of interest, is the highest for intoxications with analgesics, street drugs, poison, and other toxins (Table 4). Intoxications with antidepressants had a lower adjusted in-hospital mortality than the other types of intoxication Figure 1. Out of 246,053 ICU admissions, 7,331 patients were included in the final analysis. NICE = National (ORadj, 0.44; 0.21–0.93). The Intensive Care Evaluation, APACHE = Acute Physiology and Chronic Health Evaluation. intoxications with street drugs and other toxins have a higher mortality 1 month after ICU All 7,331 intoxicated patients could be followed for at least 3 , 1.63 and 1.73, respectively). Intoxications admission (OR months after ICU admission. As the survival of the intoxicated adj with alcohol or antidepressants have a lower mortality 1 month patients was assessed on January 1, 2012, patients admitted after , 0.50 and 0.46, respectively). No after ICU admission (OR July 1, 2011, could not be followed for a period for longer than 3 adj differences were found among the intoxication groups in the months after ICU admission. In our dataset, 7,331, 6,775, 5,592, adjusted mortality 3, 6, 12, or 24 months after ICU admission. and 3,240 intoxicated patients could be followed for a period of, respectively, 3, 6, 12, and 24 months after ICU admission. All patients had complete data regarding the covariates used for DISCUSSION case-mix correction in the logistic regression models. Despite a low in-hospital mortality of 2.1%, intoxicated Baseline characteristics of the patients are shown in patients at Dutch ICUs still have a mortality of 9.3% 2 years Table 1. Less than half of these intoxicated patients was male after ICU admission. The type of intoxication has a significant (44.7%) and had a mean age of 41.9 years (sd, 15.6 yr). The influence on the in-hospital mortality and mortality 1 month mean APACHE IV score was 42.1 (sd, 27.7). The most freafter ICU admission, but mortality 3, 6, 12, and 24 months quent type of intoxication was “sedatives” (n = 2,350; 32.1%), after ICU admission did not show these differences. and the least frequent type of intoxication was “poisoning Our ICU and in-hospital mortality rate is reasonably with carbon monoxide, arsenic, or cyanide” (n = 43; 0.6%). comparable to studies in other western countries (6–10). For Recurrent intoxication was found in 6.1% for the whole study example, in a small American retrospective study (n = 199), period (Table 2). patients with an intentional overdose had an ICU mortality Critical Care Medicine
www.ccmjournal.org
1473
Brandenburg et al
Table 1. Characteristics of the Intoxicated ICU Patients and Several Subgroups Based on the Acute Physiology and Chronic Health Evaluation IV Score Variable
All Intoxications
Alcohol
Analgesics
Number
7,331
792
324
Patient with a recurrent intoxication (n, %)
413 (6.1)
24 (3.2)
23 (8.0)
59 (7.0)
Number of deaths (n, %)
154 (2.1)
11 (1.4)
9 (2.8)
9 (1.0)
Age (mean/sd)
41.9 (15.6)
38.5 (16.3)
41.0 (17.0)
42.2 (13.9)
3,280 (44.7)
547 (69.1)
97 (29.9)
286 (31.2)
Acute Physiology and Chronic Health Evaluation IV score (mean/sd)
42.1 (27.7)
42.9 (26.9)
33.8 (25.4)
40.1 (27.4)
Length of ICU stay in days (mean/sd)
1.3 (2.5)
0.9 (1.7)
1.0 (1.1)
1.4 (1.7)
1,388 (18.9)
152 (19.2)
25 (7.7)
201 (21.9)
GCS: 3–8
2,221 (30.3)
313 (39.5)
34 (10.5)
237 (25.9)
GCS: 9–13
1,546 (21.1)
180 (22.7)
43 (13.3)
215 (23.5)
GCS: ≥ 14
3,512 (47.9)
295 (37.2)
246 (75.9)
449 (49)
5,146 (70.2)
572 (72.2)
209 (64.5)
626 (68.3)
Other hospital
182 (2.5)
15 (1.9)
9 (2.8)
34 (3.7)
Nursing ward
1,010 (13.8)
85 (10.7)
70 (21.6)
126 (13.8)
380 (5.2)
36 (4.5)
11 (3.4)
54 (5.9)
Cerebrovascular incident
39 (0.5)
12 (1.5)
4 (1.2)
3 (0.3)
Chronic cardiovascular insufficiency
40 (0.5)
8 (1)
1 (0.3)
1 (0.1)
Chronic renal insufficiency
89 (1.2)
4 (0.5)
9 (2.8)
7 (0.8)
Dysrhythmia
207 (2.8)
14 (1.8)
6 (1.9)
42 (4.6)
Chronic respirator insufficiencies
129 (1.8)
14 (1.8)
1 (0.3)
12 (1.3)
Inotropic medication
450 (6.1)
37 (4.7)
9 (2.8)
70 (7.6)
Sex: male (n, %)
Mechanical ventilation within 24 hr (n, %)
Antidepressants
916
GCS at admission (n, %)
Admission route (n, %) Emergency department
Other Comorbidities (n, %)
From ICU discharged to (n, %) Home
3,089 (42.1)
420 (53)
123 (38)
332 (36.2)
Nursing ward same hospital
2,480 (33.8)
271 (34.2)
102 (31.5)
327 (35.7)
808 (11)
36 (4.5)
46 (14.2)
124 (13.5)
85 (1.2)
7 (0.9)
4 (1.2)
8 (0.9)
545 (7.4)
36 (4.5)
34 (10.5)
83 (9.1)
Other hospital Mortuary Other
GCS = Glasgow Coma Score. a Most frequent combinations are alcohol and sedatives (22%), antidepressants and sedatives (18%), other toxin and sedatives (9%), alcohol and street drugs (8%), and alcohol and other toxin (8%). The comorbidities are based on the Acute Physiology and Chronic Health Evaluation IV definitions: cerebrovascular incident (evident prior to admission or within the first hour of admission), chronic cardiovascular insufficiency (New York Heart Association stage IV heart failure), chronic renal insufficiency (reported prior to admission in combination with a serum creatinine in first 24 hr > 177 umol/L), dysrhythmia (24 hr before admission, in combination with hemodynamic instability), chronic respiratory insufficiency (oxygen at home or chronically ventilated), and inotropic medication (minimal 1 hr in the first 24 hr of admission).
of 2.7% (8). A larger German retrospective study (n = 3,924) showed a mortality of only 0.2% in patients younger than 65 years, whereas patients older than 65 years had a mortality of 2.2% (7). 1474
www.ccmjournal.org
However, our in-hospital mortality is higher than mortality in other previously published Dutch studies (0.0–1.4%) (13–15). This could be explained by the smaller sample size of these previous studies and the fact that June 2014 • Volume 42 • Number 6
Clinical Investigations
Sedatives
Poison: Carbon Monoxide, Arsenic, Cyanide
Other Toxin
834
2,350
43
1,079
993
29 (3.7)
150 (7.0)
2 (4.7)
56 (5.6)
70 (7.8)
29 (3.5)
43 (1.8)
3 (7.0)
45 (4.2)
5 (0.5)
36.5 (15.1)
44.5 (15.7)
44.3 (17.9)
43.4 (16.5)
41.2 (13.5)
597 (71.6)
831 (35.4)
29 (67.4)
485 (44.9)
408 (41.1)
48.4 (30.2)
42.6 (27.0)
40.0 (28.5)
42.5 (29.7)
38.9 (25.0)
1.4 (2.7)
1.3 (2.5)
1.8 (2.4)
1.7 (3.9)
1.0 (1.1)
238 (28.5)
386 (16.4)
12 (27.9)
201 (18.6)
173 (17.4)
343 (41.1)
742 (31.6)
9 (20.9)
241 (22.3)
302 (30.4)
141 (16.9)
595 (25.3)
2 (4.7)
148 (13.7)
222 (22.4)
346 (41.5)
1,000 (42.6)
32 (74.4)
682 (63.2)
462 (46.5)
583 (69.9)
1,588 (67.6)
30 (69.8)
743 (68.9)
795 (80.1)
Street Drugs
Combination of Intoxicationsa
20 (2.4)
58 (2.5)
2 (4.7)
20 (1.9)
24 (2.4)
118 (14.1)
361 (15.4)
4 (9.3)
165 (15.3)
81 (8.2)
131 (5.6)
2 (4.7)
55 (5.1)
41 (4.1)
50 (6) 5 (0.6)
8 (0.3)
0 (0)
6 (0.6)
1 (0.1)
5 (0.6)
8 (0.3)
0 (0)
15 (1.4)
2 (0.2)
15 (1.8)
30 (1.3)
0 (0)
18 (1.7)
6 (0.6)
26 (3.1)
47 (2)
0 (0)
47 (4.4)
25 (2.5)
28 (3.4)
47 (2)
1 (2.3)
19 (1.8)
7 (0.7)
71 (8.5)
111 (4.7)
5 (11.6)
115 (10.7)
32 (3.2)
403 (48.3)
858 (36.5)
10 (23.3)
408 (37.8)
535 (53.9)
282 (33.8)
819 (34.9)
19 (44.2)
422 (39.1)
238 (24)
65 (7.8)
316 (13.4)
7 (16.3)
117 (10.8)
97 (9.8)
18 (2.2)
16 (0.7)
3 (7)
25 (2.3)
4 (0.4)
36 (4.3)
216 (9.2)
2 (4.7)
62 (5.7)
76 (7.7)
they did not only analyze ICU patients but also included patients seen on the ED or admitted to the general ward, which implicates that their “acute presentation” was less severe. Critical Care Medicine
In contrast, the mortality observed in our study 1 year after ICU admission (6.5%) was lower than that found in previous studies. Forsberg et al (26) found a mortality of 10.9%. However, they analyzed a different population, namely only www.ccmjournal.org
1475
Brandenburg et al
Percentage of ≥ 2 ICU Admissions for the Acute Physiology and Chronic Health Evaluation IV Intoxication Subgroups
Table 2.
Intoxicant Group
% Recurrent Admissions
Total
6.1
Alcohol
3.2
Analgesics
8.0
Antidepressants
7.0
Street drugs
3.7
Sedatives
7.0
Poison
4.7
Other toxin
5.6
Combination
7.8
illicit intoxications instead of all intoxicated patients admitted to an ICU. Patients with illicit intoxications may have a higher long-term mortality due to psychiatric illness than patients without a psychiatric illness. Unfortunately, due to the observational character of our study, we were not able to distinguish patients with illicit and patients with accidental intoxications. Similarly our 2-year mortality rate (9.3%) was much lower compared with O’Brien et al (27) who reported 25% mortality 31 months (median follow-up time) after ICU admission. This study analyzed poisonings in comatose patients at an ED instead of all intoxicated patients admitted to an ICU.
Presumably the differences in case-mix and follow-up time explain the marked differences in outcome between the mentioned studies. The overall 1-year crude mortality after ICU admission for intoxicated patients (6.5%) is rather low in comparison with the 26.6% found in another Dutch cohort of mixed ICU patients (28). We feel that the most plausible explanation for a better overall outcome in intoxicated patients might be that the acute physiological derangements are not associated with severe and permanent organ failure. It probably represents a short-term and reversible physiological disturbance, which dissolves rapidly after elimination of the toxin. The lower mortality could also be explained by the fact that intoxicated patients are actually not severely ill at ICU admission (e.g., having less comorbidities) or that the intoxicated patients received better care than other ICU patients. Most likely the intoxicated patients are less severely ill at ICU admission compared with other ICU patients, which implies that the patients may have received intensive care treatment unnecessarily. Additionally, the cohort of intoxicated patients is very young with a mean age of 41.9 years (sd, 15.6) and could therefore have a lower mortality. Indeed, patients are often admitted for “observation” while they are not suffering from any organ failure at that point. Such patients might be observed at lower and less expensive levels of care (e.g., medium care and observatory). Still the difference between the in-hospital and the long-term mortality is substantial (2.1% and 9.3%, respectively). Therefore, we feel that an ICU admission due to intoxication should not be seen as a separate event but as a warning sign for mortality in the next 2 years. Perhaps, a specialized follow-up program such
Figure 2. Kaplan-Meier curve showing the unadjusted long-term survival of the Acute Physiology and Chronic Health Evaluation IV intoxication subgroups.
1476
www.ccmjournal.org
June 2014 • Volume 42 • Number 6
Clinical Investigations
Table 3. Percentage of the Long-Term Mortality of the Acute Physiology and Chronic Health Evaluation IV Intoxication Subgroups Mortality in Months After ICU Admission Intoxicant Groups
3
6
9
12
15
18
21
Total
4.0
5.0
5.9
6.4
7.1
7.7
8.6
9.2
Alcohol
2.7
3.2
4.0
4.4
4.9
5.5
7.1
7.1
Analgesics
4.3
5.0
5.3
5.7
5.7
6.8
7.4
7.4
Antidepressants
2.3
3.5
4.4
5.1
6.1
6.2
6.2
7.2
Street drugs
5.8
7.0
7.8
8.1
9.3
9.6
11.3
12.3
Sedatives
4.4
5.8
6.9
7.6
8.2
9.0
9.7
10.3
Poison
7.0
9.4
9.4
9.4
9.4
9.4
9.4
9.4
Other toxin
5.9
6.6
7.8
8.2
9.0
9.6
10.8
11.4
Combination
2.0
2.9
3.1
3.3
4.4
4.9
5.6
6.2
as mental health treatment could be advantageous for these patients. We could also demonstrate certain subgroups having a higher mortality rate 1 month after hospital discharge, for example, the patients intoxicated with street drugs. These patients could probably benefit from a program starting soon after hospital discharge. In our analysis, we adjusted for recurrent intoxication as this might influence the long-term mortality of intoxicated patients. The number of recurrent intoxications in our dataset could be estimated by using encrypted patients’ numbers and date of birth in the NICE registry to identify admissions of the same patients. This method has limitations as the admission of the recurrent intoxication should take place in an ICU participating in the NICE registry, and even small spelling errors in the patient name and number could lead to false-negative results leading to an underestimation of the recurrent intoxication percentages. However, we were able to show that 6.1% of the intoxicated patients had a recurrent intoxication within the study period (Table 3). These percentages varied between Table 4.
24
the intoxication subgroups from 3.2% in alcohol intoxication patients to 8.0% in analgesics intoxication patients. Novack et al (29) found that a recurrent deliberate intoxication attempt was protective of ICU admission (OR 0.4 [0.16– 0.95]), but that more research is necessary to understand the group of repeated offenders. Although we analyzed all intoxicated patients admitted to an ICU instead of only the deliberate intoxications, implying that the populations are not the same, these numbers suggest that just a fraction of recurrent intoxications will end up at the ICU again. For street drug intoxications, the amount of recurrent admissions was lower than we had expected (4%). We did expect a lot of readmissions in this group because it represents a high-risk lifestyle, illustrated by an Australian survey of 76 currently GHB users. In this study, 33% of the patients that had experienced an overdose requiring medical treatment would not stop using GHB and were sure that they would overdose again (30). We do not have an adequate explanation for this low readmission rate in our street drugs intoxication group.
Adjusted Odds Ratio for Mortality Up to Two Years of Follow Up
Type of Intoxication
In-Hospital Mortality
1 Mo After ICU Admission
3 Mo After ICU Admission
6 Mo After ICU Admission
12 Mo After ICU Admission
24 Mo After ICU Admission
Alcohol
0.56 (0.27–1.18)
0.50 (0.27–0.95)
0.60 (0.36–1.00) 0.63 (0.40–1.01) 0.70 (0.45–1.10) 0.82 (0.50–1.33)
Analgesics
2.33 (1.03–5.26)
1.50 (0.71–3.15)
1.54 (0.84–2.82) 1.53 (0.89–2.65) 1.28 (0.72–2.27) 1.45 (0.77–2.75)
Antidepressants 0.44 (0.21–0.93)
0.46 (0.24–0.87)
0.63 (0.40–1.02) 0.75 (0.50–1.13) 0.97 (0.67–1.43) 0.94 (0.62–1.43)
Street drugs
1.75 (1.06–2.90)
1.63 (1.05–2.52)
1.41 (0.97–2.04) 1.38 (0.97–1.95) 1.23 (0.85–1.78) 1.25 (0.82–1.90)
Sedatives
0.70 (0.46–1.07)
0.81 (0.57–1.15)
1.00 (0.76–1.33) 1.09 (0.84–1.40) 1.10 (0.86–1.42) 0.93 (0.70–1.23)
Poison
5.76 (1.52–21.79) 3.21 (0.90–11.49) 1.82 (0.52–6.38) 2.05 (0.68–6.17) 2.05 (0.68–6.13) 0.94 (0.21–4.21)
Other toxin
1.94 (1.25–3.01)
1.73 (1.18–2.52)
1.31 (0.94–1.83) 1.02 (0.74–1.42)
Combination
0.39 (0.15–1.01)
0.81 (0.45–1.46)
0.75 (0.46–1.21) 0.80 (0.53–1.21) 0.66 (0.43–1.01) 0.78 (0.51–1.20)
a
1.06 (0.77–1.46) 1.23 (0.87–1.74)
Poisoning due to carbon monoxide, arsenic, or cyanide. The adjusted odds ratios for the in-hospital and long-term mortality of the Acute Physiology and Chronic Health Evaluation IV subgroups. The whole intoxication population, excluding the intoxication subgroup of interest, is used as the reference group.
a
Critical Care Medicine
www.ccmjournal.org
1477
Brandenburg et al
The strength of this study is the large number of admissions from a national quality registry (n = 7,331). However, some limitations should be addressed. First, we only knew the main types of intoxication (e.g., analgesics and antidepressants). Within these groups, each drug or toxin might be associated with a different risk for severe illness or death. For example, a study by Isbister et al (31) showed that there is even a difference in mortality between the different types of benzodiazepines in overdosed patients. Second, the diagnosis of intoxication is often based on clinical suspicion not on laboratory confirmation of the intoxication. Whenever laboratory analysis is performed, 11% more substances are detected (32). This could be explained in two ways: 1) it suggests that physicians underdiagnose the amount of substances taken when treating an intoxicated patient. Obviously unrecognized intoxications might have a profound influence on the outcome if they are left untreated, and 2) precisely because the diagnosis does not need to be laboratory confirmed for the APACHE IV inclusion criteria, it could also be that some admissions turn out to have a nontoxicological diagnosis. This would result in an overestimation. However, the large amount of admissions included in our database will probably lessen the impact of these issues. Third, we did not evaluate the treatment given before the ICU at the ED or at the ICU. It might be that the patient has had adequate treatment at the ED. Some patients will have stayed on the ED for a considerable time and arrive at the ICU completely stabilized. In such patients, the APACHE IV score at ICU admission will underestimate the true severity of illness of the intoxication (so-called lead time bias) (33). Fourth, although the data were collected in a prospective fashion by all the ICUs for quality purposes, the analysis is basically a retrospective one. Therefore, only associations can be detected and true cause-and-effect relationships can only be assumed. Future research should focus on finding factors explaining why some types of intoxications have higher short-term mortality than other intoxications. This might help to develop a preventive strategy in these patients. Furthermore, it is imperative to learn more about the costs of intoxicated ICU patients, especially when it could be that some intoxicated patients are admitted to the ICU for observational reasons only. These patients could possibly be observed at a (less expensive) lower level of care (e.g., medium care). Unfortunately, there are very limited data in the Netherlands about ICU admission costs in general, and it is lacking for specific patient groups, such as intoxicated patients. Also it would be interesting to analyze the patients in the APACHE IV subgroups in more detail. Unfortunately, with our current dataset, we only knew the main group of substances taken and not specifically which substance it was. For example, it would be useful to find out what specific type of street drugs is most toxic. Since the in-hospital and long-term mortality of intoxicated patients admitted at an ICU is relatively low (2.1% and 9.3%, respectively), such a prospective study would require many patients. A large international prospective study could possibly be advantageous. 1478
www.ccmjournal.org
CONCLUSION Overall, the mortality 2 years after ICU admission is relatively low compared with other ICU admissions. Despite this, the difference between the in-hospital mortality and the mortality after 2 years is substantial. We feel it is advisable that intoxicated ICU patients receive a specialized and if necessary psychiatric follow-up to prevent future events to prevent long-term sequelae and death. The mortality is statistically different for the APACHE IV intoxication subgroups until 3 months after ICU admission; thereafter, we did not find a difference between the types of intoxications.
REFERENCES
1. Burillo-Putze G, Munne P, Dueñas A, et al; Clinical Toxicology Working Group, Spanish Society of Emergency Medicine (SEMESTOX): National multicentre study of acute intoxication in emergency departments of Spain. Eur J Emerg Med 2003; 10:101–104 2. Kristinsson J, Palsson R, Gudjonsdottir GA, et al: Acute poisonings in Iceland: A prospective nationwide study. Clin Toxicol (Phila) 2008; 46:126–132 3. Heyerdahl F, Bjornas MA, Hovda KE, et al: Acute poisonings treated in hospitals in Oslo: A one-year prospective study (II): Clinical outcome. Clin Toxicol (Phila) 2008; 46:42–49 4. Lam SM, Lau AC, Yan WW: Over 8 years experience on severe acute poisoning requiring intensive care in Hong Kong, China. Hum Exp Toxicol 2010; 29:757–765 5. Henderson A, Wright M, Pond SM: Experience with 732 acute overdose patients admitted to an intensive care unit over six years. Med J Aust 1993; 158:28–30 6. Schwake L, Wollenschläger I, Stremmel W, et al: Adverse drug reactions and deliberate self-poisoning as cause of admission to the intensive care unit: A 1-year prospective observational cohort study. Intensive Care Med 2009; 35:266–274 7. Mühlberg W, Becher K, Heppner HJ, et al: Acute poisoning in old and very old patients: A longitudinal retrospective study of 5883 patients in a toxicological intensive care unit. Z Gerontol Geriatr 2005; 38:182–189 8. Hamad AE, Al-Ghadban A, Carvounis CP, et al: Predicting the need for medical intensive care monitoring in drug-overdosed patients. J Intensive Care Med 2000; 15:321–328 9. Liisanantti JH, Ohtonen P, Kiviniemi O, et al: Risk factors for prolonged intensive care unit stay and hospital mortality in acute drug-poisoned patients: An evaluation of the physiologic and laboratory parameters on admission. J Crit Care 2011; 26:160–165 10. Clark D, Murray DB, Ray D: Epidemiology and outcomes of patients admitted to critical care after self-poisoning. J Intensive Care Soc 2011; 12:268–273 11. Brinkman S, Bakhshi-Raiez F, Abu-Hanna A, et al: External validation of Acute Physiology and Chronic Health Evaluation IV in Dutch intensive care units and comparison with Acute Physiology and Chronic Health Evaluation II and Simplified Acute Physiology Score II. J Crit Care 2011; 26:105.e11–105.e18 12. Juneja D, Singh O, Nasa P, et al: Comparison of newer scoring systems with the conventional scoring systems in general intensive care population. Minerva Anestesiol 2012; 78:194–200 13. Duineveld C, Vroegop M, Schouren L, et al: Acute intoxication: Difference in management between six Dutch hospitals. Clin Toxicol 2012; 50:120–128 14. Meulendijks CF, van den Berg EJ, Fortuyn HD, et al: Predicting the need for hospital admission in patients with intentional drug overdose. Neth J Med 2003; 61:164–167 15. Vermes A, Roelofsen EE, Sabadi G, et al: Intoxication with therapeutic and illicit drug substances and hospital admission to a Dutch university hospital. Neth J Med 2003; 61:168–172 June 2014 • Volume 42 • Number 6
Clinical Investigations 16. Brinkman S, Bakhshi-Raiez F, Abu-Hanna A, et al: Determinants of mortality after hospital discharge in ICU patients: Literature review and Dutch cohort study. Crit Care Med 2013; 41:1237–1251 17. Vasilevskis EE, Kuzniewicz MW, et al: Relationship between discharge practices and intensive care unit in-hospital mortality performance. Evidence of a discharge bias. Med Care 2009; 47: 803–812 18. Niskanen M, Kari A, Halonen P: Five-year survival after intensive care– comparison of 12,180 patients with the general population. Finnish ICU Study Group. Crit Care Med 1996; 24:1962–1967 19. Miró O, Nogué S, Espinosa G, et al: Trends in illicit drug emergencies: The emerging role of gamma-hydroxybutyrate. J Toxicol Clin Toxicol 2002; 40:129–135 20. Staikowsky F, Theil F, Mercadier P, et al: Change in profile of acute self drug-poisonings over a 10-year period. Hum Exp Toxicol 2004; 23:507–511 21. Zimmerman JE, Kramer AA, McNair DS, et al: Acute Physiology and Chronic Health Evaluation (APACHE) IV: Hospital mortality assessment for today’s critically ill patients. Crit Care Med 2006; 34:1297–1310 22. Dutch National Intensive Care Evaluation foundation. Available at: http://www.stichting-nice.nl. Accessed July 24, 2012 23. Vektis (in Dutch). Available at: http://www.vektis.nl. Accessed February 2, 2013 24. Roos LL, Wajda A: Record linkage strategies. Part I: Estimating information and evaluating approaches. Methods Inf Med 1991; 30:117–123
Critical Care Medicine
25. Brinkman S, De Jonge E, Abu-Hanna A, et al: The use of linked registries to assess long-term mortality of ICU patients. Stud Health Technol Inform 2012; 180:230–234 26. Forsberg S, Höjer J, Ludwigs U: Prognosis in patients presenting with non-traumatic coma. J Emerg Med 2012; 42:249–253 27. O’Brien BP, Murphy D, Conrick-Martin I, et al: The functional outcome and recovery of patients admitted to an intensive care unit following drug overdose: A follow-up study. Anaesth Intensive Care 2009; 37:802–806 28. Meynaar IA, van den Boogaard M, Tangkau PLet al: Long-term survival after ICU treatment. Minerva Anestesiol 2012; 78:1324–1332 29. Novack V, Jotkowitz A, Delgado J, et al: General characteristics of hospitalized patients after deliberate self-poisoning and risk factors for intensive care admission. Eur J Intern Med 2006; 17:485–489 30. Degenhardt L, Darke S, Dillon P: The prevalence and correlates of gamma-hydroxybutyrate (GHB) overdose among Australian users. Addiction 2003; 98:199–204 31. Isbister GK, O’Regan L, Sibbritt D, et al: Alprazolam is relatively more toxic than other benzodiazepines in overdose. Br J Clin Pharmacol 2004; 58:88–95 32. Heyerdahl F, Hovda KE, Bjornaas MA, et al: Clinical assessment compared to laboratory screening in acutely poisoned patients. Hum Exp Toxicol 2008; 27:73–79 33. Tunnell RD, Millar BW, Smith GB: The effect of lead time bias on severity of illness scoring, mortality prediction and standardised mortality ratio in intensive care—A pilot study. Anaesthesia 1998; 53:1045–1053
www.ccmjournal.org
1479
Objective: To assess in-hospital and long-term mortality of Dutch ICU patients admitted with an acute intoxication. Design: Cohort of ICU admissions from a national ICU registry linked to records from an insurance claims database. Setting: Eighty-one ICUs (85% of all Dutch ICUs). Patients: Seven thousand three hundred thirty-one admissions between January 1, 2008, and October 1, 2011. Interventions: None. Measurements and Main Results: Kaplan-Meier curves were used to compare the unadjusted mortality of the total intoxicated population and for specific intoxication subgroups based on the Acute Physiology and Chronic Health Evaluation IV reasons for admission: 1) alcohol(s), 2) analgesics, 3) antidepressants, 4) street drugs, 5) sedatives, 6) poisoning (carbon monoxide, arsenic, or cyanide), 7) other toxins, and 8) combinations. The case-mix adjusted mortality was assessed by the odds ratio adjusted for age, gender, severity of illness, intubation status, recurrent intoxica-
*See also p. 1563. 1 Department of Intensive Care and Emergency Medicine, University Medical Center Utrecht, University of Utrecht, Utrecht, The Netherlands. 2 Department of Medical Informatics, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands. 3 National Intensive Care Evaluation Foundation, Amsterdam, The N etherlands. 4 Dutch National Poisons Information Center, University Medical Center, University of Utrecht, Utrecht, The Netherlands. 5 Institute for Risk Assessment Sciences, University of Utrecht, Utrecht, The Netherlands. Dr. Brinkman is employed by National Intensive Care Evaluation (NICE) (foundation pays the department of medical informatics to process and analyze data for the registry). Dr. de Keizer is employed by NICE (foundation pays the department of medical informatics to process and analyze data for the registry). Dr. de Lange served as a board member of the NICE foundation and received support for the development of educational presentations (lecturing fees by congress organizations). The remaining authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000245
Critical Care Medicine
tion, and several comorbidities. The ICU mortality was 1.2%, and the in-hospital mortality was 2.1%. The mortality 1, 3, 6, 12, and 24 months after ICU admission was 2.8%, 4.1%, 5.2%, 6.5%, and 9.3%, respectively. Street drugs had the highest mortality 2 years after ICU admission (12.3%); a combination of different intoxications had the lowest (6.3%). The adjusted observed mortality showed that intoxications with street drugs and “other toxins” have a significant higher mortality 1 month after ICU admission (odds ratioadj = 1.63 and odds ratioadj= 1.73, respectively). Intoxications with alcohol or antidepressants have a significant lower mortality 1 month after ICU admission (odds ratioadj = 0.50 and odds ratio= 0.46, respectively). These differences were not found in the adj adjusted mortality 3 months upward of ICU admission. Conclusions: Overall, the mortality 2 years after ICU admission is relatively low compared with other ICU admissions. The first 3 months after ICU admission there is a difference in mortality between the subgroups, not thereafter. Still, the difference between the in-hospital mortality and the mortality after 2 years is substantial. (Crit Care Med 2014; 42:1471–1479) Key Words: drug overdose; intensive care; intoxication; long-term outcome; mortality; poisoning
A
substantial proportion of patients with an acute intoxication who are seen at an emergency department (ED) are admitted to the ICU (3.7–40%) (1–3). Depending on the setting and the kind of ICU, 3.4–13.8% of the ICU population consists of patients with an acute intoxication (4, 5). However, the in-hospital mortality as a result of intoxications is strikingly low. In the Western hemisphere, both ICU mortality and in-hospital mortality of intoxicated patients are significantly lower than that of the general ICU population (0.2–4.0% vs 16.2–16.4%) (6–12). In the Netherlands, the reported in-hospital mortality of intoxicated patients is even lower: 0–1.4% (13–15). As shown, there are a lot of data about the in-hospital mortality of intoxicated ICU patients. However, we feel that long-term outcome should also be taken into account. Indeed, www.ccmjournal.org
1471
Brandenburg et al
little is achieved when a patient dies soon after hospital discharge, and for that reason, the long-term outcome should be taken into consideration (16). Additionally, long-term outcome reduces the bias of transfers and discharge timing that might influence the in-hospital mortality statistics (17). Also, analyzing long-term mortality could help to identify patients at high risk to die after hospital discharge and to provide these patients with specialized follow-up. Despite its importance, studies about the long-term outcome of intoxicated ICU patients are scarce. A Finnish study (1996) showed a low in-hospital mortality of only 3.6% but an excessive death rate thereafter. When these authors compared the patients with intoxications to the general population (matched for gender and age), the ratio between observed and predicted mortality was 6.3 after 5 years of follow-up (18). Intoxication with dissolvent (ethylene glycol and methanol) had the highest 5-year mortality (35.3%). However, this study distinguished only three main types of intoxications, sedatives and psychotropic drugs, dissolvent, and “other.” Additionally, this study was published in 1996 and the pattern of intoxications has changed significantly over the last years; the numbers of intoxications with γ-hydroxybutyrate (GHB) and antidepressants have increased, whereas the number of intoxications with benzodiazepines has decreased (19, 20). Within the area of critical care medicine, prognostic models such as Acute Physiology and Chronic Health Evaluation (APACHE) IV (21) are frequently used for case-mix adjustment. The APACHE IV model distinguishes seven types of intoxicants as a reason for ICU admission. The aim of this study is to use APACHE IV intoxication subgroups to assess and compare the (case-mix adjusted) in-hospital and long-term mortality of ICU patients admitted with acute intoxication.
MATERIALS AND METHODS Subjects All adult ICU patients admitted to a Dutch ICU between January 1, 2008, and October 1, 2011, with an intoxication reason for admission and satisfying the APACHE IV inclusion criteria were considered eligible. We used the APACHE IV intoxication reasons for ICU admissions: 1) alcohol, 2) analgesics (aspirin, acetaminophen), 3) antidepressants (e.g., cyclic antidepressants and lithium), 4) street drugs (opiates, cocaine, and amphetamine), 5) sedatives (hypnotics, antipsychotic, and benzodiazepines), 6) poisoning due to carbon monoxide, arsenic, or cyanide, 7) other toxins, and 8) combinations of these diagnosis, maximum two. The diagnosis is based on clinical suspicion and does not need to be laboratory confirmed for the APACHE IV inclusion criteria. Data Collection The Dutch National Intensive Care Evaluation (NICE) registry (22) contains demographic, physiological, and clinical data recorded in the first 24 hours of ICU admission of all consecutive ICU patients admitted to participating ICUs. It includes, 1472
www.ccmjournal.org
among others, the APACHE IV severity of illness score and reason for ICU admission (21), as well as the ICU and in-hospital mortality. During the study period, approximately 85% of all Dutch ICUs recorded data on all their admissions in the NICE registry. All participating ICUs are mixed medical-surgical units located in university hospitals (n = 7), teaching hospitals (n = 27), or nonteaching hospitals (n = 47). The mortality after hospital discharge is obtained by linking the NICE registry to a national insurance claims database of Vektis (23). The records from the NICE registry and insurance claims database are linked by a deterministic linkage algorithm (24) that used the hospital of admission, gender, date of birth, ICU admission date, and ICU discharge date. Details on the linkage process are described in more detail elsewhere (25). The insurance claims database contains information on the vital status of patients. The status of the patients (either alive or death) on January 1, 2012, and if relevant, the date of death, was extracted from the insurance claims database. The data used in this study have been encrypted in a way that all patient identifying information, such as name and patient identification number, has been removed. In the Netherlands, there is no need to obtain consent to make use of such registries without patient identifying information. The data are officially registered according to the Dutch Personal Data Protection Act. Statistical Analysis To assess in-hospital mortality and the mortality at 1, 3, 6, 12, and 24 months after ICU admission, we used Kaplan-Meier curves and logistic regression models. The latter is used, as the proportional hazard assumption of the more frequently used Cox model was not fulfilled. The case-mix adjusted mortality was analyzed by calculating the odds ratio (ORadj), while adjusting for age, gender, severity of illness expressed as the APACHE IV score, intubation status and the comorbidities chronic respiratory insufficiency, chronic cardiovascular insufficiency, chronic renal insufficiency, cerebrovascular accident, and the use of vasopressors. Patients who are readmitted to the ICU within the year of follow-up could have a new critical illness with accompanying risk of death or are admitted for another intoxication, implying a chronic use of substances. Therefore, we also adjusted for such recurrent intoxication. The ORadj and corresponding 95% CI were calculated for each ICU subgroup with the whole intoxicated ICU population (excluding the subgroup of interest) as reference. All statistical analyses were performed using PASW statistics 18 (SPSS, Chicago, IL) and the R environment for statistical computing version 2.15.1.
RESULTS From January 1, 2008, to October 1, 2011, a total of 246,053 patients were admitted to the participating Dutch ICUs, including 9,129 patients (3.7%) with at least one diagnosis of intoxication (Fig. 1). Of the 9,129 intoxicated patients, 8,176 records (89.6%) could be linked with the insurance claims database of Vektis. Finally, only the 7,331 patients fulfilling the APACHE IV inclusion criteria (22) were included in our analysis (Fig. 1). June 2014 • Volume 42 • Number 6
Clinical Investigations
Mortality of Intoxicated Patients A total of 85 patients died while being admitted to an ICU (ICU mortality rate, 1.2%), and another 69 patients died after ICU admission while they were still in the hospital (in-hospital mortality rate, 2.1%). The mortality at 1, 3, 6, 12, and 24 months after ICU admission for all patients with intoxications was 2.8%, 4.1%, 5.2%, 6.5%, and 9.3%, respectively. Figure 2 shows the unadjusted mortality of the whole intoxication population and of the intoxication subgroups in Kaplan-Meier curves. Table 3 shows the percentage of the long-term mortality for each APACHE IV subgroup. The ORadj for the i n-hospital and long-term mortality of the APACHE IV subgroups, compared with the whole intoxication population while excluding the intoxication subgroup of interest, is the highest for intoxications with analgesics, street drugs, poison, and other toxins (Table 4). Intoxications with antidepressants had a lower adjusted in-hospital mortality than the other types of intoxication Figure 1. Out of 246,053 ICU admissions, 7,331 patients were included in the final analysis. NICE = National (ORadj, 0.44; 0.21–0.93). The Intensive Care Evaluation, APACHE = Acute Physiology and Chronic Health Evaluation. intoxications with street drugs and other toxins have a higher mortality 1 month after ICU All 7,331 intoxicated patients could be followed for at least 3 , 1.63 and 1.73, respectively). Intoxications admission (OR months after ICU admission. As the survival of the intoxicated adj with alcohol or antidepressants have a lower mortality 1 month patients was assessed on January 1, 2012, patients admitted after , 0.50 and 0.46, respectively). No after ICU admission (OR July 1, 2011, could not be followed for a period for longer than 3 adj differences were found among the intoxication groups in the months after ICU admission. In our dataset, 7,331, 6,775, 5,592, adjusted mortality 3, 6, 12, or 24 months after ICU admission. and 3,240 intoxicated patients could be followed for a period of, respectively, 3, 6, 12, and 24 months after ICU admission. All patients had complete data regarding the covariates used for DISCUSSION case-mix correction in the logistic regression models. Despite a low in-hospital mortality of 2.1%, intoxicated Baseline characteristics of the patients are shown in patients at Dutch ICUs still have a mortality of 9.3% 2 years Table 1. Less than half of these intoxicated patients was male after ICU admission. The type of intoxication has a significant (44.7%) and had a mean age of 41.9 years (sd, 15.6 yr). The influence on the in-hospital mortality and mortality 1 month mean APACHE IV score was 42.1 (sd, 27.7). The most freafter ICU admission, but mortality 3, 6, 12, and 24 months quent type of intoxication was “sedatives” (n = 2,350; 32.1%), after ICU admission did not show these differences. and the least frequent type of intoxication was “poisoning Our ICU and in-hospital mortality rate is reasonably with carbon monoxide, arsenic, or cyanide” (n = 43; 0.6%). comparable to studies in other western countries (6–10). For Recurrent intoxication was found in 6.1% for the whole study example, in a small American retrospective study (n = 199), period (Table 2). patients with an intentional overdose had an ICU mortality Critical Care Medicine
www.ccmjournal.org
1473
Brandenburg et al
Table 1. Characteristics of the Intoxicated ICU Patients and Several Subgroups Based on the Acute Physiology and Chronic Health Evaluation IV Score Variable
All Intoxications
Alcohol
Analgesics
Number
7,331
792
324
Patient with a recurrent intoxication (n, %)
413 (6.1)
24 (3.2)
23 (8.0)
59 (7.0)
Number of deaths (n, %)
154 (2.1)
11 (1.4)
9 (2.8)
9 (1.0)
Age (mean/sd)
41.9 (15.6)
38.5 (16.3)
41.0 (17.0)
42.2 (13.9)
3,280 (44.7)
547 (69.1)
97 (29.9)
286 (31.2)
Acute Physiology and Chronic Health Evaluation IV score (mean/sd)
42.1 (27.7)
42.9 (26.9)
33.8 (25.4)
40.1 (27.4)
Length of ICU stay in days (mean/sd)
1.3 (2.5)
0.9 (1.7)
1.0 (1.1)
1.4 (1.7)
1,388 (18.9)
152 (19.2)
25 (7.7)
201 (21.9)
GCS: 3–8
2,221 (30.3)
313 (39.5)
34 (10.5)
237 (25.9)
GCS: 9–13
1,546 (21.1)
180 (22.7)
43 (13.3)
215 (23.5)
GCS: ≥ 14
3,512 (47.9)
295 (37.2)
246 (75.9)
449 (49)
5,146 (70.2)
572 (72.2)
209 (64.5)
626 (68.3)
Other hospital
182 (2.5)
15 (1.9)
9 (2.8)
34 (3.7)
Nursing ward
1,010 (13.8)
85 (10.7)
70 (21.6)
126 (13.8)
380 (5.2)
36 (4.5)
11 (3.4)
54 (5.9)
Cerebrovascular incident
39 (0.5)
12 (1.5)
4 (1.2)
3 (0.3)
Chronic cardiovascular insufficiency
40 (0.5)
8 (1)
1 (0.3)
1 (0.1)
Chronic renal insufficiency
89 (1.2)
4 (0.5)
9 (2.8)
7 (0.8)
Dysrhythmia
207 (2.8)
14 (1.8)
6 (1.9)
42 (4.6)
Chronic respirator insufficiencies
129 (1.8)
14 (1.8)
1 (0.3)
12 (1.3)
Inotropic medication
450 (6.1)
37 (4.7)
9 (2.8)
70 (7.6)
Sex: male (n, %)
Mechanical ventilation within 24 hr (n, %)
Antidepressants
916
GCS at admission (n, %)
Admission route (n, %) Emergency department
Other Comorbidities (n, %)
From ICU discharged to (n, %) Home
3,089 (42.1)
420 (53)
123 (38)
332 (36.2)
Nursing ward same hospital
2,480 (33.8)
271 (34.2)
102 (31.5)
327 (35.7)
808 (11)
36 (4.5)
46 (14.2)
124 (13.5)
85 (1.2)
7 (0.9)
4 (1.2)
8 (0.9)
545 (7.4)
36 (4.5)
34 (10.5)
83 (9.1)
Other hospital Mortuary Other
GCS = Glasgow Coma Score. a Most frequent combinations are alcohol and sedatives (22%), antidepressants and sedatives (18%), other toxin and sedatives (9%), alcohol and street drugs (8%), and alcohol and other toxin (8%). The comorbidities are based on the Acute Physiology and Chronic Health Evaluation IV definitions: cerebrovascular incident (evident prior to admission or within the first hour of admission), chronic cardiovascular insufficiency (New York Heart Association stage IV heart failure), chronic renal insufficiency (reported prior to admission in combination with a serum creatinine in first 24 hr > 177 umol/L), dysrhythmia (24 hr before admission, in combination with hemodynamic instability), chronic respiratory insufficiency (oxygen at home or chronically ventilated), and inotropic medication (minimal 1 hr in the first 24 hr of admission).
of 2.7% (8). A larger German retrospective study (n = 3,924) showed a mortality of only 0.2% in patients younger than 65 years, whereas patients older than 65 years had a mortality of 2.2% (7). 1474
www.ccmjournal.org
However, our in-hospital mortality is higher than mortality in other previously published Dutch studies (0.0–1.4%) (13–15). This could be explained by the smaller sample size of these previous studies and the fact that June 2014 • Volume 42 • Number 6
Clinical Investigations
Sedatives
Poison: Carbon Monoxide, Arsenic, Cyanide
Other Toxin
834
2,350
43
1,079
993
29 (3.7)
150 (7.0)
2 (4.7)
56 (5.6)
70 (7.8)
29 (3.5)
43 (1.8)
3 (7.0)
45 (4.2)
5 (0.5)
36.5 (15.1)
44.5 (15.7)
44.3 (17.9)
43.4 (16.5)
41.2 (13.5)
597 (71.6)
831 (35.4)
29 (67.4)
485 (44.9)
408 (41.1)
48.4 (30.2)
42.6 (27.0)
40.0 (28.5)
42.5 (29.7)
38.9 (25.0)
1.4 (2.7)
1.3 (2.5)
1.8 (2.4)
1.7 (3.9)
1.0 (1.1)
238 (28.5)
386 (16.4)
12 (27.9)
201 (18.6)
173 (17.4)
343 (41.1)
742 (31.6)
9 (20.9)
241 (22.3)
302 (30.4)
141 (16.9)
595 (25.3)
2 (4.7)
148 (13.7)
222 (22.4)
346 (41.5)
1,000 (42.6)
32 (74.4)
682 (63.2)
462 (46.5)
583 (69.9)
1,588 (67.6)
30 (69.8)
743 (68.9)
795 (80.1)
Street Drugs
Combination of Intoxicationsa
20 (2.4)
58 (2.5)
2 (4.7)
20 (1.9)
24 (2.4)
118 (14.1)
361 (15.4)
4 (9.3)
165 (15.3)
81 (8.2)
131 (5.6)
2 (4.7)
55 (5.1)
41 (4.1)
50 (6) 5 (0.6)
8 (0.3)
0 (0)
6 (0.6)
1 (0.1)
5 (0.6)
8 (0.3)
0 (0)
15 (1.4)
2 (0.2)
15 (1.8)
30 (1.3)
0 (0)
18 (1.7)
6 (0.6)
26 (3.1)
47 (2)
0 (0)
47 (4.4)
25 (2.5)
28 (3.4)
47 (2)
1 (2.3)
19 (1.8)
7 (0.7)
71 (8.5)
111 (4.7)
5 (11.6)
115 (10.7)
32 (3.2)
403 (48.3)
858 (36.5)
10 (23.3)
408 (37.8)
535 (53.9)
282 (33.8)
819 (34.9)
19 (44.2)
422 (39.1)
238 (24)
65 (7.8)
316 (13.4)
7 (16.3)
117 (10.8)
97 (9.8)
18 (2.2)
16 (0.7)
3 (7)
25 (2.3)
4 (0.4)
36 (4.3)
216 (9.2)
2 (4.7)
62 (5.7)
76 (7.7)
they did not only analyze ICU patients but also included patients seen on the ED or admitted to the general ward, which implicates that their “acute presentation” was less severe. Critical Care Medicine
In contrast, the mortality observed in our study 1 year after ICU admission (6.5%) was lower than that found in previous studies. Forsberg et al (26) found a mortality of 10.9%. However, they analyzed a different population, namely only www.ccmjournal.org
1475
Brandenburg et al
Percentage of ≥ 2 ICU Admissions for the Acute Physiology and Chronic Health Evaluation IV Intoxication Subgroups
Table 2.
Intoxicant Group
% Recurrent Admissions
Total
6.1
Alcohol
3.2
Analgesics
8.0
Antidepressants
7.0
Street drugs
3.7
Sedatives
7.0
Poison
4.7
Other toxin
5.6
Combination
7.8
illicit intoxications instead of all intoxicated patients admitted to an ICU. Patients with illicit intoxications may have a higher long-term mortality due to psychiatric illness than patients without a psychiatric illness. Unfortunately, due to the observational character of our study, we were not able to distinguish patients with illicit and patients with accidental intoxications. Similarly our 2-year mortality rate (9.3%) was much lower compared with O’Brien et al (27) who reported 25% mortality 31 months (median follow-up time) after ICU admission. This study analyzed poisonings in comatose patients at an ED instead of all intoxicated patients admitted to an ICU.
Presumably the differences in case-mix and follow-up time explain the marked differences in outcome between the mentioned studies. The overall 1-year crude mortality after ICU admission for intoxicated patients (6.5%) is rather low in comparison with the 26.6% found in another Dutch cohort of mixed ICU patients (28). We feel that the most plausible explanation for a better overall outcome in intoxicated patients might be that the acute physiological derangements are not associated with severe and permanent organ failure. It probably represents a short-term and reversible physiological disturbance, which dissolves rapidly after elimination of the toxin. The lower mortality could also be explained by the fact that intoxicated patients are actually not severely ill at ICU admission (e.g., having less comorbidities) or that the intoxicated patients received better care than other ICU patients. Most likely the intoxicated patients are less severely ill at ICU admission compared with other ICU patients, which implies that the patients may have received intensive care treatment unnecessarily. Additionally, the cohort of intoxicated patients is very young with a mean age of 41.9 years (sd, 15.6) and could therefore have a lower mortality. Indeed, patients are often admitted for “observation” while they are not suffering from any organ failure at that point. Such patients might be observed at lower and less expensive levels of care (e.g., medium care and observatory). Still the difference between the in-hospital and the long-term mortality is substantial (2.1% and 9.3%, respectively). Therefore, we feel that an ICU admission due to intoxication should not be seen as a separate event but as a warning sign for mortality in the next 2 years. Perhaps, a specialized follow-up program such
Figure 2. Kaplan-Meier curve showing the unadjusted long-term survival of the Acute Physiology and Chronic Health Evaluation IV intoxication subgroups.
1476
www.ccmjournal.org
June 2014 • Volume 42 • Number 6
Clinical Investigations
Table 3. Percentage of the Long-Term Mortality of the Acute Physiology and Chronic Health Evaluation IV Intoxication Subgroups Mortality in Months After ICU Admission Intoxicant Groups
3
6
9
12
15
18
21
Total
4.0
5.0
5.9
6.4
7.1
7.7
8.6
9.2
Alcohol
2.7
3.2
4.0
4.4
4.9
5.5
7.1
7.1
Analgesics
4.3
5.0
5.3
5.7
5.7
6.8
7.4
7.4
Antidepressants
2.3
3.5
4.4
5.1
6.1
6.2
6.2
7.2
Street drugs
5.8
7.0
7.8
8.1
9.3
9.6
11.3
12.3
Sedatives
4.4
5.8
6.9
7.6
8.2
9.0
9.7
10.3
Poison
7.0
9.4
9.4
9.4
9.4
9.4
9.4
9.4
Other toxin
5.9
6.6
7.8
8.2
9.0
9.6
10.8
11.4
Combination
2.0
2.9
3.1
3.3
4.4
4.9
5.6
6.2
as mental health treatment could be advantageous for these patients. We could also demonstrate certain subgroups having a higher mortality rate 1 month after hospital discharge, for example, the patients intoxicated with street drugs. These patients could probably benefit from a program starting soon after hospital discharge. In our analysis, we adjusted for recurrent intoxication as this might influence the long-term mortality of intoxicated patients. The number of recurrent intoxications in our dataset could be estimated by using encrypted patients’ numbers and date of birth in the NICE registry to identify admissions of the same patients. This method has limitations as the admission of the recurrent intoxication should take place in an ICU participating in the NICE registry, and even small spelling errors in the patient name and number could lead to false-negative results leading to an underestimation of the recurrent intoxication percentages. However, we were able to show that 6.1% of the intoxicated patients had a recurrent intoxication within the study period (Table 3). These percentages varied between Table 4.
24
the intoxication subgroups from 3.2% in alcohol intoxication patients to 8.0% in analgesics intoxication patients. Novack et al (29) found that a recurrent deliberate intoxication attempt was protective of ICU admission (OR 0.4 [0.16– 0.95]), but that more research is necessary to understand the group of repeated offenders. Although we analyzed all intoxicated patients admitted to an ICU instead of only the deliberate intoxications, implying that the populations are not the same, these numbers suggest that just a fraction of recurrent intoxications will end up at the ICU again. For street drug intoxications, the amount of recurrent admissions was lower than we had expected (4%). We did expect a lot of readmissions in this group because it represents a high-risk lifestyle, illustrated by an Australian survey of 76 currently GHB users. In this study, 33% of the patients that had experienced an overdose requiring medical treatment would not stop using GHB and were sure that they would overdose again (30). We do not have an adequate explanation for this low readmission rate in our street drugs intoxication group.
Adjusted Odds Ratio for Mortality Up to Two Years of Follow Up
Type of Intoxication
In-Hospital Mortality
1 Mo After ICU Admission
3 Mo After ICU Admission
6 Mo After ICU Admission
12 Mo After ICU Admission
24 Mo After ICU Admission
Alcohol
0.56 (0.27–1.18)
0.50 (0.27–0.95)
0.60 (0.36–1.00) 0.63 (0.40–1.01) 0.70 (0.45–1.10) 0.82 (0.50–1.33)
Analgesics
2.33 (1.03–5.26)
1.50 (0.71–3.15)
1.54 (0.84–2.82) 1.53 (0.89–2.65) 1.28 (0.72–2.27) 1.45 (0.77–2.75)
Antidepressants 0.44 (0.21–0.93)
0.46 (0.24–0.87)
0.63 (0.40–1.02) 0.75 (0.50–1.13) 0.97 (0.67–1.43) 0.94 (0.62–1.43)
Street drugs
1.75 (1.06–2.90)
1.63 (1.05–2.52)
1.41 (0.97–2.04) 1.38 (0.97–1.95) 1.23 (0.85–1.78) 1.25 (0.82–1.90)
Sedatives
0.70 (0.46–1.07)
0.81 (0.57–1.15)
1.00 (0.76–1.33) 1.09 (0.84–1.40) 1.10 (0.86–1.42) 0.93 (0.70–1.23)
Poison
5.76 (1.52–21.79) 3.21 (0.90–11.49) 1.82 (0.52–6.38) 2.05 (0.68–6.17) 2.05 (0.68–6.13) 0.94 (0.21–4.21)
Other toxin
1.94 (1.25–3.01)
1.73 (1.18–2.52)
1.31 (0.94–1.83) 1.02 (0.74–1.42)
Combination
0.39 (0.15–1.01)
0.81 (0.45–1.46)
0.75 (0.46–1.21) 0.80 (0.53–1.21) 0.66 (0.43–1.01) 0.78 (0.51–1.20)
a
1.06 (0.77–1.46) 1.23 (0.87–1.74)
Poisoning due to carbon monoxide, arsenic, or cyanide. The adjusted odds ratios for the in-hospital and long-term mortality of the Acute Physiology and Chronic Health Evaluation IV subgroups. The whole intoxication population, excluding the intoxication subgroup of interest, is used as the reference group.
a
Critical Care Medicine
www.ccmjournal.org
1477
Brandenburg et al
The strength of this study is the large number of admissions from a national quality registry (n = 7,331). However, some limitations should be addressed. First, we only knew the main types of intoxication (e.g., analgesics and antidepressants). Within these groups, each drug or toxin might be associated with a different risk for severe illness or death. For example, a study by Isbister et al (31) showed that there is even a difference in mortality between the different types of benzodiazepines in overdosed patients. Second, the diagnosis of intoxication is often based on clinical suspicion not on laboratory confirmation of the intoxication. Whenever laboratory analysis is performed, 11% more substances are detected (32). This could be explained in two ways: 1) it suggests that physicians underdiagnose the amount of substances taken when treating an intoxicated patient. Obviously unrecognized intoxications might have a profound influence on the outcome if they are left untreated, and 2) precisely because the diagnosis does not need to be laboratory confirmed for the APACHE IV inclusion criteria, it could also be that some admissions turn out to have a nontoxicological diagnosis. This would result in an overestimation. However, the large amount of admissions included in our database will probably lessen the impact of these issues. Third, we did not evaluate the treatment given before the ICU at the ED or at the ICU. It might be that the patient has had adequate treatment at the ED. Some patients will have stayed on the ED for a considerable time and arrive at the ICU completely stabilized. In such patients, the APACHE IV score at ICU admission will underestimate the true severity of illness of the intoxication (so-called lead time bias) (33). Fourth, although the data were collected in a prospective fashion by all the ICUs for quality purposes, the analysis is basically a retrospective one. Therefore, only associations can be detected and true cause-and-effect relationships can only be assumed. Future research should focus on finding factors explaining why some types of intoxications have higher short-term mortality than other intoxications. This might help to develop a preventive strategy in these patients. Furthermore, it is imperative to learn more about the costs of intoxicated ICU patients, especially when it could be that some intoxicated patients are admitted to the ICU for observational reasons only. These patients could possibly be observed at a (less expensive) lower level of care (e.g., medium care). Unfortunately, there are very limited data in the Netherlands about ICU admission costs in general, and it is lacking for specific patient groups, such as intoxicated patients. Also it would be interesting to analyze the patients in the APACHE IV subgroups in more detail. Unfortunately, with our current dataset, we only knew the main group of substances taken and not specifically which substance it was. For example, it would be useful to find out what specific type of street drugs is most toxic. Since the in-hospital and long-term mortality of intoxicated patients admitted at an ICU is relatively low (2.1% and 9.3%, respectively), such a prospective study would require many patients. A large international prospective study could possibly be advantageous. 1478
www.ccmjournal.org
CONCLUSION Overall, the mortality 2 years after ICU admission is relatively low compared with other ICU admissions. Despite this, the difference between the in-hospital mortality and the mortality after 2 years is substantial. We feel it is advisable that intoxicated ICU patients receive a specialized and if necessary psychiatric follow-up to prevent future events to prevent long-term sequelae and death. The mortality is statistically different for the APACHE IV intoxication subgroups until 3 months after ICU admission; thereafter, we did not find a difference between the types of intoxications.
REFERENCES
1. Burillo-Putze G, Munne P, Dueñas A, et al; Clinical Toxicology Working Group, Spanish Society of Emergency Medicine (SEMESTOX): National multicentre study of acute intoxication in emergency departments of Spain. Eur J Emerg Med 2003; 10:101–104 2. Kristinsson J, Palsson R, Gudjonsdottir GA, et al: Acute poisonings in Iceland: A prospective nationwide study. Clin Toxicol (Phila) 2008; 46:126–132 3. Heyerdahl F, Bjornas MA, Hovda KE, et al: Acute poisonings treated in hospitals in Oslo: A one-year prospective study (II): Clinical outcome. Clin Toxicol (Phila) 2008; 46:42–49 4. Lam SM, Lau AC, Yan WW: Over 8 years experience on severe acute poisoning requiring intensive care in Hong Kong, China. Hum Exp Toxicol 2010; 29:757–765 5. Henderson A, Wright M, Pond SM: Experience with 732 acute overdose patients admitted to an intensive care unit over six years. Med J Aust 1993; 158:28–30 6. Schwake L, Wollenschläger I, Stremmel W, et al: Adverse drug reactions and deliberate self-poisoning as cause of admission to the intensive care unit: A 1-year prospective observational cohort study. Intensive Care Med 2009; 35:266–274 7. Mühlberg W, Becher K, Heppner HJ, et al: Acute poisoning in old and very old patients: A longitudinal retrospective study of 5883 patients in a toxicological intensive care unit. Z Gerontol Geriatr 2005; 38:182–189 8. Hamad AE, Al-Ghadban A, Carvounis CP, et al: Predicting the need for medical intensive care monitoring in drug-overdosed patients. J Intensive Care Med 2000; 15:321–328 9. Liisanantti JH, Ohtonen P, Kiviniemi O, et al: Risk factors for prolonged intensive care unit stay and hospital mortality in acute drug-poisoned patients: An evaluation of the physiologic and laboratory parameters on admission. J Crit Care 2011; 26:160–165 10. Clark D, Murray DB, Ray D: Epidemiology and outcomes of patients admitted to critical care after self-poisoning. J Intensive Care Soc 2011; 12:268–273 11. Brinkman S, Bakhshi-Raiez F, Abu-Hanna A, et al: External validation of Acute Physiology and Chronic Health Evaluation IV in Dutch intensive care units and comparison with Acute Physiology and Chronic Health Evaluation II and Simplified Acute Physiology Score II. J Crit Care 2011; 26:105.e11–105.e18 12. Juneja D, Singh O, Nasa P, et al: Comparison of newer scoring systems with the conventional scoring systems in general intensive care population. Minerva Anestesiol 2012; 78:194–200 13. Duineveld C, Vroegop M, Schouren L, et al: Acute intoxication: Difference in management between six Dutch hospitals. Clin Toxicol 2012; 50:120–128 14. Meulendijks CF, van den Berg EJ, Fortuyn HD, et al: Predicting the need for hospital admission in patients with intentional drug overdose. Neth J Med 2003; 61:164–167 15. Vermes A, Roelofsen EE, Sabadi G, et al: Intoxication with therapeutic and illicit drug substances and hospital admission to a Dutch university hospital. Neth J Med 2003; 61:168–172 June 2014 • Volume 42 • Number 6
Clinical Investigations 16. Brinkman S, Bakhshi-Raiez F, Abu-Hanna A, et al: Determinants of mortality after hospital discharge in ICU patients: Literature review and Dutch cohort study. Crit Care Med 2013; 41:1237–1251 17. Vasilevskis EE, Kuzniewicz MW, et al: Relationship between discharge practices and intensive care unit in-hospital mortality performance. Evidence of a discharge bias. Med Care 2009; 47: 803–812 18. Niskanen M, Kari A, Halonen P: Five-year survival after intensive care– comparison of 12,180 patients with the general population. Finnish ICU Study Group. Crit Care Med 1996; 24:1962–1967 19. Miró O, Nogué S, Espinosa G, et al: Trends in illicit drug emergencies: The emerging role of gamma-hydroxybutyrate. J Toxicol Clin Toxicol 2002; 40:129–135 20. Staikowsky F, Theil F, Mercadier P, et al: Change in profile of acute self drug-poisonings over a 10-year period. Hum Exp Toxicol 2004; 23:507–511 21. Zimmerman JE, Kramer AA, McNair DS, et al: Acute Physiology and Chronic Health Evaluation (APACHE) IV: Hospital mortality assessment for today’s critically ill patients. Crit Care Med 2006; 34:1297–1310 22. Dutch National Intensive Care Evaluation foundation. Available at: http://www.stichting-nice.nl. Accessed July 24, 2012 23. Vektis (in Dutch). Available at: http://www.vektis.nl. Accessed February 2, 2013 24. Roos LL, Wajda A: Record linkage strategies. Part I: Estimating information and evaluating approaches. Methods Inf Med 1991; 30:117–123
Critical Care Medicine
25. Brinkman S, De Jonge E, Abu-Hanna A, et al: The use of linked registries to assess long-term mortality of ICU patients. Stud Health Technol Inform 2012; 180:230–234 26. Forsberg S, Höjer J, Ludwigs U: Prognosis in patients presenting with non-traumatic coma. J Emerg Med 2012; 42:249–253 27. O’Brien BP, Murphy D, Conrick-Martin I, et al: The functional outcome and recovery of patients admitted to an intensive care unit following drug overdose: A follow-up study. Anaesth Intensive Care 2009; 37:802–806 28. Meynaar IA, van den Boogaard M, Tangkau PLet al: Long-term survival after ICU treatment. Minerva Anestesiol 2012; 78:1324–1332 29. Novack V, Jotkowitz A, Delgado J, et al: General characteristics of hospitalized patients after deliberate self-poisoning and risk factors for intensive care admission. Eur J Intern Med 2006; 17:485–489 30. Degenhardt L, Darke S, Dillon P: The prevalence and correlates of gamma-hydroxybutyrate (GHB) overdose among Australian users. Addiction 2003; 98:199–204 31. Isbister GK, O’Regan L, Sibbritt D, et al: Alprazolam is relatively more toxic than other benzodiazepines in overdose. Br J Clin Pharmacol 2004; 58:88–95 32. Heyerdahl F, Hovda KE, Bjornaas MA, et al: Clinical assessment compared to laboratory screening in acutely poisoned patients. Hum Exp Toxicol 2008; 27:73–79 33. Tunnell RD, Millar BW, Smith GB: The effect of lead time bias on severity of illness scoring, mortality prediction and standardised mortality ratio in intensive care—A pilot study. Anaesthesia 1998; 53:1045–1053
www.ccmjournal.org
1479
