The mean duration of ICU stay was 5.8 days (median 3, ... 3 days (range 0â33, mean 7.9 days). ..... Grace M. Quality of life measures before and one year after.
Anaesthesia, 1999, 54, pages 727–732 ................................................................................................................................................................................................................................................
ICU admission in patients infected with the human immunodeficiency virus – a multicentre survey J. K. Gill,1 L. Greene,2 R. Miller,3 A. Pozniak,4 J. Cartledge,3 M. Fisher,2 M. R. Nelson1 and N. Soni1 1 Chelsea and Westminster Hospital, 369 Fulham Road, London SW10 9NH, 2 Royal Sussex County Hospital, Brighton, 3 University College London Hospitals, and 4 King’s College Hospital, London, UK Summary
We conducted a retrospective study to assess the reasons for admission to the intensive care unit, and subsequent outcome, in patients infected with the human immunodeficiency virus (HIV). Four hospitals in the south of England participated, all with specialist HIV units. Data were collected on 127 patients admitted to ICU on 133 separate occasions between June 1993 and October 1997. The mean age on admission was 38 years (range 23–60 years). Ninety-four patients (70.7%) were documented HIV-positive before admission and 36 (27%) were diagnosed HIVpositive for the first time during admission; 36.1% were admitted with Pneumocystis carinii pneumonia. Overall ICU mortality was 33%, in-hospital mortality was 56% and the eventual mortality at the end of follow-up (March 1998) was 72%. Survival was highest in those admitted with respiratory HIV-related disease or HIV-unrelated illness. Associations with poor outcome included a prior AIDS-defining illness, a CD4 cell count of less than 100 cells.ml¹1 and admission secondary to sepsis. Keywords Immune system; human immunodeficiency virus. Intensive care; outcome. ...................................................................................... Correspondence to: Dr J. K. Gill Accepted: 1 December 1998
Since the onset of the human immunodeficiency virus (HIV) epidemic, there has been a debate on the role of intensive care units (ICUs) in the management of critically ill HIV-positive individuals. During the 1980s, HIV-infection had a poor prognosis and high morbidity, occasionally leading to ICU admission with poor survival [1]. This led to the perception among clinicians that ICU admission in HIV was associated with a poor outcome. In recent years, the number of HIV-positive patients developing serious illness appears to have decreased considerably [2]. Possible reasons for this include earlier diagnosis of HIV infection because of increased awareness, the use of prophylactic agents against opportunistic infections such as Pneumocystis carinii pneumonia (PCP) and improved management of HIV-related disease within specialist HIV units. The use of highly active antiretroviral therapy (HAART) has had a major impact on disease progression in patients infected with the HIV virus [3]. Much of the available data on ICU survival originates Q 1999 Blackwell Science Ltd
from studies undertaken by centres in the USA and France. Criteria for ICU admission vary both between and within geographical areas, as may the epidemiology of HIVassociated disease. Significant variations have been found in ICU use by hospital characteristics and geographical location that remained significant after controlling for severity of illness and patient demographic characteristics [4]. In addition, studies show that hospital mortality rates were as high as twice ICU mortality rates [5, 6]. It is clear that a contemporary survey of current practice as regards ICU admission and outcome in HIV-positive patients was warranted. To determine the outcome of ICU and hospital admission in HIV-positive patients a study was conducted in a group of hospitals with large HIV practices. Methods
Data were collected retrospectively from the case notes of patients admitted to the ICUs of four hospitals with specialist HIV/AIDS units using a standard form which 727
J. K. Gill et al. • ICU admission in HIV patients Anaesthesia, 1999, 54, pages 727–732 ................................................................................................................................................................................................................................................
had been piloted previously. Patients were included if they were admitted between June 1993 and October 1997 and had documented evidence of a positive HIV antibody test, were diagnosed HIV-positive during their ICU admission, or if HIV disease was clinically evident. Subjects fitting the study criteria were identified using hospital databases. Demographic data were collected, as well as the risk factors for HIV transmission, CD4 cell count, diagnosis of AIDS as defined by the 1993 Expanded European AIDS case definition [7], treatment with antiretroviral drugs, reason for hospital presentation and ICU admission, length of time from hospital admission to ICU transfer, APACHE II score, need for invasive ventilation, duration of ICU stay, ICU outcome and hospital outcome. Cases were labelled ‘HIV-related’ if admission was secondary to a bacterial infection, opportunistic infection/malignancy, or as a complication of HIV treatment, and ‘non-HIVrelated’ in other conditions, such as drug overdose. It was noted whether patients had survived to March 1998, when the observation period ended (now labelled as overall outcome or survival). Data analysis Patients admitted on more than one separate occasion were treated as individual cases for each admission. However, readmission to the ICU within a week of ICU discharge was treated as a single case for the purposes of survival analysis. For those cases where date of death was not documented, we censored survival time as a time midway between the date last seen alive and the end of our observation period. Analysis was performed using SPSS statistical software (SPSS for Windows, SPSS Inc., Chicago, IL, USA). Kaplan–Meier survival curves were constructed to illustrate survival from ICU admission to the end of the study period according to the admission CD4 cell counts. The log-rank test was used to compare overall survival in patients with HIV-related disease and non-HIV-related admissions. Chi-squared tests were used to analyse the effect of AIDS status and ventilation requirement on overall survival. Logistic regression analysis was used to assess the relationship between admission APACHE II score and outcome. A p-value of less than 0.05 was considered to be statistically significant throughout. Results
Descriptive analysis Data were collected on 127 patients who were admitted to ICUs on 133 separate occasions during the period June 1993 to October 1997. Table 1 demonstrates the basic demographic characteristics. The mean age on admission 728
Table 1 Demographic characteristics of patients admitted to ICU. Characteristic Sex Male Female Ethnic origin White Black Asian Other Missing data HIV transmission risk factor Homosexual sexual intercourse Intravenous drug use Heterosexual sexual intercourse Bisexual sexual intercourse Haemophiliac/blood transfusion Missing data
n
%
115 18
86.5 13.5
100 19 5 6 3
75.1 14.3 3.8 4.5 2.3
84 14 20 3 1 11
63.2 10.5 15.0 2.3 0.8 8.3
was 38 years (median 36 years, range 23–60 years). Ninetyfour subjects (70.7%) were documented HIV-positive prior to ICU admission, 36 (27%) were diagnosed HIVpositive for the first time during admission, and in three admissions (two patients) HIV disease was strongly clinically suspected but a confirmatory test was not performed. Of the 94 subjects known to be HIV-infected before ICU admission, 56 had an AIDS diagnosis prior to ICU admission, the remainder had not. Twenty-three of these 94 patients (24.5%) had received antiretrovirals pre-admission with eight patients taking nucleoside reverse transcriptor inhibitor (NRTI) monotherapy, seven taking dual NRTI therapy and eight on combination therapy including a protease inhibitor. Sixty-seven subjects had not received antiretroviral therapy and data were not available in four cases. The CD4 cell count on admission, or most recent prior to admission, in 122 cases was a median of 40 cells.ml¹1 (range 0–945, interquartile range 16–100 cells.ml¹1). The CD4 cell count was not available in 11 cases. One hundred and twelve (84.2%) patients were admitted to ICU with a HIV-related condition, 19 (14.3%) admissions were non-HIV related. In two (1.5%) cases, HIV relation was unclear with one patient admitted with acute renal failure and uraemic encephalopathy of unknown cause and one admitted with peritonitis of unknown cause. All the patients diagnosed as HIV-positive for the first time on ICU had a HIV-related condition. APACHE II scores on admission to ICU were documented in 117 cases, but missing in the remainder. The median score was 15. Primary reasons for hospital admission Table 2 shows the reasons for admission and subsequent outcomes. Q 1999 Blackwell Science Ltd
Anaesthesia, 1999, 54, pages 727–732 J. K. Gill et al. • ICU admission in HIV patients ................................................................................................................................................................................................................................................
Table 2 Primary reasons for hospital Admission n %
admission and mortality rates. Respiratory PCP Non-PCP Gastrointestinal Neurological Sepsis Other Missing data Total
Respiratory/PCP Of the 48 patients admitted with PCP, 30 were confirmed by induced sputum examination, bronchoalveolar lavage, lung biopsy or post-mortem. Respiratory disease/non-PCP Six subjects had bacterial infections, one had tuberculosis, six had other nonbacterial infections (e.g. CMV, fungal), five had pulmonary Kaposi’s sarcoma or lymphoma. The remainder included interstitial pneumonitis (two patients), laryngeal obstruction (one patient) and exacerbation of asthma (one patient). Gastrointestinal Of 23 cases, 11 had postoperative complications following gastrointestinal surgery, nine had gastrointestinal bleeding, two had pancreatitis and one was admitted with hepatic failure. Neurological Diagnoses included cryptococcal meningitis (three patients), cerebral toxoplasmosis (three patients), viral encephalitis (four patients), which was thought to be secondary to CMV in one patient, cerebral lymphoma (two patients), uraemic encephalopathy (one patient), pneumococcal meningitis (one patient), status epilepticus in a known epileptic (one patient), Guillan-Barre´ syndrome (one patient) and tuberculous meningitis (one patient). Sepsis Of these 14 patients, four had staphylococcal septicaemia (one also had falciparum malaria), one had cryptococcosis, one enterobacter sakazakii septicaemia and one disseminated toxoplasmosis. The causative organism was not identified in the remainder of cases. Other Other reasons for hospitalisation included drug overdose (three patients), investigations (two patients), acute renal failure (two patients), cardiopulmonary arrest (one patient), Q 1999 Blackwell Science Ltd
48 21 23 17 14 10 133
36.1 15.8 17.3 12.8 10.5 7.5 100
ICU mortality n %
Hospital n %
Overall n %
15 6 7 5 7 4
31 28 30 29 50 40
23 11 16 10 10 4
48 52 69 59 71 40
44
33
74
56
32 14 19 14 12 5 2 98
67 67 83 82 86 50 72
oral ulceration (one patient) and Stevens Johnson syndrome (one patient). The most common reason for ICU admission was for respiratory support (54.1%) followed by cardiorespiratory compromise including post cardiopulmonary arrest (21.1%). Other reasons included postoperative care (10.5%), monitoring (6%), sedation (5.3%), renal dialysis (1.5%) and for a procedure (1.5%). The mean time from hospitalisation to ICU transfer was 5.5 days (median 3, range 0–30 days). One hundred and two (76.7%) patients required ventilation of the lungs, 12 (9%) had noninvasive assisted ventilation and 16 (12%) did not require any form of assisted ventilation. The degree of ventilation was not documented in three (2.3%) patients. The mean duration of ICU stay was 5.8 days (median 3, range 0–40 days). Survival ICU, hospital and overall mortality rates are shown in Table 2. In those that did not survive to be discharged from ICU, the median time from ICU admission to death was 3 days (range 0–33, mean 7.9 days). For those subjects discharged from ICU, but not surviving to the end of the study period, the median time from ICU discharge to death was 27 days (range 0–1200, interquartile range 5–187, mean 139.6 days). ICU mortality in those requiring invasive ventilation was 40.2% and hospital mortality 62.7%. Of the patients admitted with PCP, 40 (83.3%) were ventilated. In this group, ICU mortality was 37.5% and hospital mortality 50%. Figure 1 illustrates the Kaplan–Meier survival curves for those subjects with an admission CD4 cell count >100 cells.ml¹1 and those with a CD4 cell count of #100 cells.ml¹1 (log-rank, 5.36; p ¼ 0.0206) showing that survival is significantly higher in subjects with a CD4 cell count >100 cells.ml¹1. Survival curves were also used to compare outcomes in those admitted with PCP and those admitted with other HIV-related conditions (log-rank, 6.3; p ¼ 0.012). This 729
J. K. Gill et al. • ICU admission in HIV patients Anaesthesia, 1999, 54, pages 727–732 ................................................................................................................................................................................................................................................
Figure 1 Survival stratified by CD4 cell
count.
indicated that overall survival was significantly higher in those admitted with PCP rather than other HIV-related illnesses. Also compared were those patients admitted with HIV-related conditions other than PCP and those admitted with non-HIV-related illnesses (log-rank, 5.03; p ¼ 0.025). Hence, patients with a HIV-related disease other than PCP were shown to have a significantly worse outcome when compared with patients admitted with non-HIV-related disease. The end-of-study outcome was compared for those patients requiring invasive ventilation in the ICU, against those who did not (Chi-squared, 10.1; p ¼ 0.001), indicating a significantly better outcome in those who were not ventilated. Overall outcome in those with a previous AIDS diagnosis was compared with those without, excluding cases where HIV was first diagnosed during ICU admission (Chi-squared, 7.03; p ¼ 0.008), showing a significantly higher survival in those without a prior AIDS diagnosis. To assess the relationship between the admission APACHE II score and overall outcome, a multivariate logistic regression model was applied including APACHE II score, CD4 cell count, prior AIDS diagnosis, use of antiretroviral therapy and HIV relation of reason for admission, p ¼ 0.025, suggesting the higher the APACHE II score, the greater the probability of death. Discussion
Published studies of ICU admission in HIV-positive individuals have often focused on PCP and other respiratory 730
causes for admission [8, 9]. However, in this study, just under half the admissions were with a variety of nonrespiratory conditions, which is consistent with the findings of De Palo et al. [10] whose prospective study found 46% of cases were admitted to ICU with nonrespiratory disease. An interesting finding is that 27% of admissions were diagnosed HIV-positive for the first time on ICU admission. In each of these cases, admission was with a HIVrelated illness. The recognition and early diagnosis of HIV infection in all patients admitted to the ICU is important as this influences management. This can be a difficult issue in the unconscious or sedated patient with life-threatening illness, as it is not always possible to obtain consent for HIV testing from a patient without a clearly defined legal guardian. In the units participating in this study, the practice is to test if there is a clinical suspicion of HIV infection and if this diagnosis would affect clinical management. This cohort demonstrates a low ICU mortality, but a considerably higher hospital mortality. This is consistent with other studies [5, 6] where hospital mortality rates have been found to be twice the ICU mortality. When making the decision whether to admit a patient to ICU, it is important to consider hospital mortality as well as ICU outcomes. ICU mortality rates were low for patients admitted with respiratory, gastrointestinal and neurological disease, but were higher in those admitted with sepsis or other conditions. However, hospital mortality was highest in patients with gastrointestinal and neurological disease, or sepsis, and this was reflected in the end-of-study mortality. This is Q 1999 Blackwell Science Ltd
Anaesthesia, 1999, 54, pages 727–732 J. K. Gill et al. • ICU admission in HIV patients ................................................................................................................................................................................................................................................
consistent with the findings of Thyrault et al. [11] who reported an excess mortality in AIDS patients with septic shock admitted to the ICU when compared to HIVnegative patients. In this study, the high mortality in the subjects admitted with sepsis may be related to a low CD4 cell count, which in 12 cases, was a median of 22 cells.ml¹1 (mean 56.8, range 0–380 cells.ml¹1); CD4 cell counts were not documented in two cases. Of particular interest is the group of patients admitted with ‘other’ conditions. ICU mortality was highest in this group, but of those surviving ICU all but one were still alive at the end of the study. The majority of deaths in this group were in patients who had HIV-related conditions, whereas all but one of the survivors were admitted with non-HIV-related disease. This is consistent with our findings that patients admitted with HIV-related disease, other than PCP, have a significantly worse outcome than patients with non-HIV-related conditions. In addition, survival is higher in those admitted with PCP compared with those with other HIV-related illnesses. It is acknowledged that it is possible that eligible cases may have been missed either because they were not identified on the database, or because the case records were missing, and that this study is limited by its retrospective nature. In addition, it is important to appreciate that the mortality rates in this study may differ considerably from the outcomes in other ICU units. All the participating hospitals have specialist HIV units with considerable experience and the ICUs involved are accustomed to the diagnosis and management of these patients. There are data indicating that survival is higher in patients admitted to hospitals with more HIV experience [4]. This may be extended to the ICU. In this survey the APACHE II score on admission was collected. This has been evaluated previously. Smith et al. [12] suggested that the APACHE II system significantly underestimated mortality in AIDS patients with PCP requiring mechanical ventilation, but accurately predicted outcome in all other patients with AIDS, and in mechanically ventilated patients without PCP. Brown & Crede [13] conducted a similar study and found that APACHE II significantly underestimated the mortality rate in the HIVpositive population with a CD4 cell count # 200 cells.ml¹1, particularly if patients were admitted with pneumonia or sepsis. In this study, the APACHE II score significantly correlated with survival when using a multivariate model including CD4 cell count, prior AIDS diagnosis, use of antiretroviral therapy and HIV relation of reason for admission. However, as only the APACHE II score was documented rather than its predictive value, we are unable to comment on the applicability of this system to HIVinfected patients. In conclusion, ICU mortality is low in this study, Q 1999 Blackwell Science Ltd
particularly for patients admitted with non-HIV-related illness or respiratory HIV-related disease. In this study only a few subjects were receiving combination antiretroviral therapy. With the advent of HAART it is likely that hospital admission in HIV-positive patients will change in two ways. First, patients may well benefit from HAARTwith a change in the natural history of their HIV infection and, second, the nature of disease presentation to ICU in the HIV-positive population may alter considerably. Although antiretroviral drugs have many beneficial effects, it is possible that adverse reactions to these potent drugs may in turn lead to ICU admission. As yet, there are no data available on how HAART has influenced physicians in their decision to admit patients to ICU, or on the expectations of patients. A large prospective study is needed to help clinicians make appropriate use of the ICU in HIV-positive individuals. Acknowledgments
We thank Professor B. Gazzard, Chelsea and Westminster Hospital, for his support, Dr A. Webb, Clinical Director of the ICU, University College London Hospitals, for supplying information from the ICU database, Dr N. Lucas, Research Fellow, Department of Anaesthetics, Chelsea and Westminster Hospital, for help with data collection and Miss N. Ives, Statistician, Department of Epidemiology, Chelsea and Westminster Hospital, for statistical advice. References 1 Rosen MJ, Cucco RA, Teirstein AS. Outcome of intensive care in patients with the acquired immunodeficiency syndrome. Journal of Intensive Care Medicine 1986; 1: 55–60. 2 Centres for Disease Control and Prevention. Update. Trends in AIDS incidence, deaths and prevalence ¹ United States, 1996. Morbidity and Mortality Weekly Report 1997; 46: 165–73. 3 Brodt HR, Kamps BS, Gute P, Knupp B, Staszewski S, Helm EB. Changing incidence of AIDS ¹ defining illnesses in the era of antiretroviral combination therapy. AIDS 1997; 11: 1731–8. 4 Curtis JR, Bennett CL, Horner RD, Rubenfeld GD, DeHovitz JA, Weinstein RA. Variations in intensive care unit utilization for patients with human immunodeficiency virus-related Pneumocystis carinii pneumonia: importance of hospital characteristics and geographic location. Critical Care Medicine 1998; 26: 668–75. 5 Konopad E, Noseworthy TW, Johnston R, Shustack A, Grace M. Quality of life measures before and one year after admission to an ICU. Critical Care Medicine 1995; 23: 1653–9. 6 Casalino E, Mendoza-Sassi G, Wolff M, et al. Predictors of short- and long-term survival in HIV-infected patients admitted to the ICU. Chest 1998; 113: 421–9. 7 Ancelle-Park RA. Expanded European AIDS case definition. Lancet 1993; 341: 441. 731
J. K. Gill et al. • ICU admission in HIV patients Anaesthesia, 1999, 54, pages 727–732 ................................................................................................................................................................................................................................................
8 Rogers PL, Lane HC, Henderson DK, Parrillo J, Masur H. Admission of AIDS patients to a medical intensive care unit: causes and outcome. Critical Care Medicine 1989; 17: 113–17. 9 Wachter RM, Luce JM, Hopewell PC. Critical care of patients with AIDS. Journal of the American Medical Association 1992; 267: 541–7. 10 De Palo VA, Millstein BH, Mayo PH, Salzman SH, Rosen MJ. Outcome of intensive care in patients with HIV infection. Chest 1995; 107: 506–10.
732
11 Thyrault M, Gachot B, Chastang C, et al. Septic shock in patients with the acquired immunodeficiency syndrome. Intensive Care Medicine 1997; 23: 1018–23. 12 Smith RL, Levine SM, Lewis ML. Prognosis of patients with AIDS requiring intensive care. Chest 1989; 96: 857–61. 13 Brown MC, Crede WB. Predictive ability of Acute Physiology and Chronic Health Evaluation II scoring applied to human immunodeficiency virus-positive patients. Critical Care Medicine 1995; 23: 848–53.
Q 1999 Blackwell Science Ltd