Survival of HIV-infected patients admitted to the intensive care unit in the era of highly active antiretroviral therapy

Abstract

We retrospectively studied outcomes for HIV-infected patients admitted to the intensive care unit (ICU) between January 1999 and June 2009. Patient demographics, receipt of highly active antiretroviral therapy (HAART), reason for ICU admission and survival to ICU and hospital discharge were recorded. Comparison was made against outcomes for general medical patients contemporaneously admitted to the same ICU. One hundred and ninety-two HIV-infected patients had 222 ICU admissions; 116 patients required mechanical ventilation (MV) and 43 required renal replacement therapy. ICU admission was due to an HIV-associated diagnosis in 113 patients; 37 had Pneumocystis pneumonia. Survival to ICU discharge and hospital discharge for HIV-infected patients was 78% and 70%, respectively, and was 75% and 68% among 2065 general medical patients with 2274 ICU admissions; P = 0.452 and P = 0.458, respectively. HIV infection was newly diagnosed in 42 patients; their ICU and hospital survival was 69% and 57%, respectively. From multivariable analysis, factors associated with ICU survival were patient's age (odds ratio [OR] = 0.74 [95% confidence interval (CI) = 0.53–1.02] per 10-year increase), albumin (OR = 1.05 [1.00–1.09] per 1 g/dL increase), Acute Physiology and Chronic Health Evaluation (APACHE) II score (OR = 0.55 [0.35–0.87] per 10 unit increase), receipt of HAART (OR = 2.44 [1.01–4.94]) and need for MV (OR = 0.14 [0.06–0.36]). In the era of HAART, HIV-infected patients should be offered ICU admission if it is likely to be of benefit.

Keywords

HIV AIDS intensive care outcome

INTRODUCTION

HIV infection remains an important communicable disease in the UK with serious morbidity, high costs of treatment and significant mortality. The number of people living with HIV in the UK continues to rise annually, with an estimated 86,500 infected at the end of 2009.¹ Of these, a quarter was unaware of their infection. In 2009, there were 6630 new diagnoses of HIV in the UK. Notably, 52% of patients were ‘late presenters’ and at diagnosis of HIV infection had CD4 cell counts <350 cells/μL, i.e. the threshold at which antiretroviral treatment is recommended to begin.

Several studies early in the AIDS epidemic reported high intensive care unit (ICU) mortality rates and poor long-term survival among HIV-infected patients requiring critical care support, which lent weight to their exclusion from ICU environments.^2–4 Following the introduction of highly active antiretroviral therapy (HAART) in the mid-1990s, there was a dramatic improvement in morbidity and mortality rates of HIV-infected individuals.^5–7 This led to a change in paradigm such that HIV infection is now a chronic illness that can be effectively controlled with HAART, with most patients able to achieve a near-normal life-expectancy.⁸ Concurrent with this, new approaches to the management of critically-ill patients with acute respiratory failure have demonstrated improved survival among patients from the general population, related to reduced ventilator-related iatrogenic lung injury and a more optimal utilization of drug therapy.^9–12 Several studies from Europe and the USA have demonstrated improved outcomes for HIV-infected patients requiring ICU admission in the era of HAART.^13–20

A previous study from our centre described ICU outcomes for HIV-infected patients early in the HAART era (1999–2005).¹⁹ It was our impression that outcomes remained good among this patient group and were comparable to general medical patients admitted to the same ICU. In the present study, we have also analysed our recent (2006–2009) ICU experience and describe ICU outcomes among HIV-infected patients over a 10-year period during the era of HAART and identify factors associated with prognosis.

METHODS

Patients

Consecutive admissions of HIV-infected individuals to the ICU at University College London Hospitals (UCLH) between 1 January 2006 and 30 June 2009 were identified. HIV infection was either previously known or newly diagnosed during that admission. Keyword searches of the ICU electronic database were performed to identify all HIV-infected individuals. A further search of the hospital electronic discharge system was performed to identify patients with a new diagnosis of HIV infection who had been admitted to the ICU. These data were added to the existing database (of consecutive HIV-infected individuals admitted to the ICU between 1 January 1999 and 31 December 2005)¹⁸ to generate a combined data set used in this study. All adults admitted to the ICU between 1 January 1999 and 30 June 2009 were identified from the ICU electronic database. Specific diagnoses in those general medical patients (without known HIV infection) were recorded. University College London Hospital is a 665-bed university-affiliated teaching hospital with 35 critical care beds. It serves an urban population with a high prevalence of HIV.

Data collection

Demographic information recorded included age, gender, ethnicity and risk factor for HIV acquisition. Patients' awareness of their HIV status, receipt of HAART and previous AIDS-defining illnesses were also noted. Other variables noted were haemoglobin, serum bilirubin, serum creatinine and the Acute Physiology and Chronic Health Evaluation (APACHE) II score on the day of ICU admission. A peripheral blood CD4 count was noted if performed on, or within four weeks prior to, admission to the ICU. Some patients had an HIV viral load recorded, although this is not routinely measured during acute illness at this HIV centre. The final diagnosis (either at ICU discharge, or death) was noted. The length of ICU admission, duration of mechanical ventilation and/or renal replacement therapy (if required), and the occurrence of pneumothorax were also recorded. HAART was not commenced in any patient during their ICU admission. Outcome was recorded either as death, or survival to ICU discharge and hospital discharge.

Statistical analysis

The probability of survival to ICU discharge was modelled using logistic regression. Robust standard errors were used to account for the small number of multiple admissions. Because of skewness in the distribution of CD4 counts, log₁₀ values were used. Due to the limited number of fatal events in these data, a multivariate model was fitted including only five factors, which were those considered to be of greatest clinical significance. Statistical analysis was performed using STATA Version 10 (STATACORP, College Station, TX, USA). A P value of <0.05 was considered significant.

RESULTS

One hundred and ninety-two HIV-infected patients were admitted to the ICU on 222 occasions over the 10-year period. During this time there were a total of 14,936 admissions to the ICU. Sixteen HIV-infected patients had two admissions, five had three admissions and one required five admissions. Of the 30 repeat ICU admissions, 14 were within two weeks of discharge from the ICU (7 for progression or deterioration of the condition that had originally required ICU admission, 7 for a new diagnosis). In the other 16, readmission occurred between one and 42 months after first admission and was due to a new clinical problem. There was a trend with time for increasing numbers of HIV-infected patients to be admitted to the ICU; HIV-infected patients accounted for 49/5222 (0.94%) of all ICU admissions in 1999–2002, 64/3652 (1.75%) in 2003–2005 and 109/6056 (1.80%) in 2006–2009; P < 0.001. Over the same time period, a total of 2118 HIV-infected patients were admitted to UCLH on 4313 occasions.

The characteristics of patients with HIV infection admitted to the ICU are shown in Table 1. As a group, the median (interquartile [IQR]) APACHE II score in the first 24 hours of the ICU admission was 16 (13–22), and their median (IQR) duration of ICU stay was four days (2–10). Survival to ICU discharge was 78% (149/192) and survival to hospital discharge was 70% for HIV-infected patients with their first admission to the ICU. For all admissions of HIV-infected patients to the ICU, survival to ICU discharge was 79% (175/222) and to hospital discharge 72%.

Table 1

Characteristics of 192 HIV-infected patients admitted to the ICU on 222 occasions

	No. (%) of patients
Characteristics	Cohort 1* 1999–2005	Cohort 2 2006–2009	Combined cohort 1999–2009
No. of patients/no. of admissions	102/113	90/99	192/222
Median (IQR) age in years	39 (32–44)	41(36–50)	40 (33–48)
Male gender	83 (73)	92 (84)	79 (79)
Race, ethnicity
White	54 (48.7)	65 (60)	119 (54)
Black	48 (42.5)	25 (23)	73 (33)
Asian	3 (3.5)	13 (12)	16 (7)
Other	5 (4.6)	0 (0)	5 (2)
Unknown	3 (0.8)	6 (5.5)	9 (4)
HIV risk factor
Heterosexual	54 (50)	31 (39)	85 (45)
MSM	42 (39)	31 (39)	73 (39)
Intravenous drug use	12 (11)	18 (22)	30 (16)
Initial HIV diagnosis made on this hospital admission	31 (27)	11 (10)	42 (19)
Receipt of HAART	42 (37)	65 (60)	107 (48)
Admission diagnosis
Non-HIV associated	37 (33)	72 (66)	109 (49)
HIV Associated	76 (67)	37 (34)	113 (51)
PCP	26 (23)	9 (8)	35 (16)
Clinical characteristics, median (IQR)
CD4 (cells/μL)	75 (10–260)	300 (80–460)	110 (20–340)
APACHE II	18 (12–23)	16 (13–22)	16.5 (13–23)
Albumin (g/L)	24 (19–31)	26.5 (19–35)	25 (19–33)
Complications
Mechanical ventilation	70 (62)	46 (43)	116 (53)
Pneumothorax	12 (11)	3 (3)	15 (7)
Renal replacement therapy	22 (19)	21 (20)	43 (20)

MSM = men who have sex with other men; HAART = highly active antiretroviral therapy; PCP = Pneumocystis jirovecii pneumonia; APACHE = Acute Physiology and Chronic Health Evaluation; IQR = interquartile range

Data shown are n (%) of patients or median values IQR

*Data previously reported in Thorax 2007;62:964–8

ICU diagnoses

Lower respiratory tract diagnoses accounted for the largest proportion of admissions by organ system at 37% (Table 2). Of those with respiratory diagnoses, 65 (79%) had acute lung injury (defined by partial pressure of oxygen in arterial blood/fraction of inspired oxygen (PaO₂/FiO₂) ratio <40 kPa [<300 mmHg] and chest radiographic abnormalities). Most patients with acute lung injury had either Pneumocystis jirovecii pneumonia (PCP) (n = 35; 54%) or bacterial pneumonia (n = 21; 32%). ICU and hospital survival for admissions due to respiratory diagnoses were 74% and 58%, respectively. Of those with respiratory diagnoses who were mechanically ventilated, ICU and hospital survival were 66% and 62%, respectively. Among patients who were not mechanically ventilated, ICU survival was 93% and hospital survival 82%. Nine of 15 patients with pneumothorax had PCP, three had bacterial pneumonia, one developed adult respiratory distress syndrome (ARDS) and two had neurological infection. All patients (except one with PCP) were mechanically ventilated when pneumothorax occurred.

Table 2

Diagnoses in 192 HIV-infected patients admitted to the ICU on 222 occasions

Diagnosis	1999–2005 (n = 113)*	2006–2009 (n = 109)	1999–2009 (n = 222)
LRTI	54	28	82
PCP	26^†‡	9^§	35
Bacterial pneumonia	17^**††	11^‡‡	28
Tuberculosis	7^§§	5	12
Asthma/COPD	3	1	4
CMV pneumonitis	1	0	1
Smoke inhalation	0	1	1
Aspiration pneumonia	0	1	1
Neurological	16	15	31
Intra-cerebral bleed	3***	4	7
Cerebral toxoplasmosis	3	2	5
Cryptococcal meningitis	3	2	5
Uncontrolled epilepsy	2	1	3
Bacterial meningitis	1	1	2
Viral encephalitis	0	2	2
Tuberculous meningitis	1	2	3
HIV encephalopathy	2	0	2
Intracerebral thrombosis	0	1	1
CMV encephalitis	1^†††	0	1
Infection	10	17	27
Sepsis	10^‡‡‡	14^§§§	24
Malaria	0	3	3
Postoperative	7	18	25
Trauma	1	4	5
Cardiovascular	9	5	14
Post cardiac arrest	7	1	8
Pulmonary hypertension	0	1	1
Pulmonary embolus	0	1	1
Cardiac dysrhythmia	1	2	3
Malignant hypertension	1	0	1
Gastrointestinal	7	4	11
Variceal bleed	5	3	8
Acute pancreatitis	1	1	2
Acute liver failure	1	0	1
Renal	0	4	4
Acute renal failure	0	2****	2
Tumour lysis syndrome	0	2	2
Castleman disease	0	6^††††	6
Overdose	6	4	10
DKA	0	4	4
HAART-related	3^‡‡‡‡	0	3

LRTI = lower respiratory tract infection; COPD = chronic obstructive pulmonary disease; CMV = cytomegalovirus; DKA = diabetic ketoacidosis; HAART = highly active antiretroviral therapy; ICU = intensive care unit; PCP = Pneumocystis jirovecii pneumonia; IRIS = immune reconstitution inflammatory syndrome

*Data modified from Thorax 2007;62:964–8

^†Six patients were also treated for other opportunistic infections during their admission to the ICU (3 had CMV pneumonitis, 2 had pulmonary tuberculosis and 1 had CMV encephalitis)

^‡One patient was re-admitted to the ICU with PCP as a manifestation of IRIS

**One patient also had cerebral toxoplasmosis

^††One patient was undergoing chemotherapy for non-Hodgkin's lymphoma

^§§Two patients also had PCP and two others had CMV encephalitis

^†††One patient also had tuberculous meningitis

***Two patients had intracerebral haemorrhages as a complication of end-stage viral liver disease

^‡‡‡Three patients presented with sepsis (without localizing features) as a complication of chemotherapy for HIV-related malignancy

^‡‡‡‡Two patients had lactic acidosis and one had IRIS

^††††Three patients had acute renal failure

^§Three patients were also treated for other infections during their admission to the ICU (one had Salmonella sepsis, the other had varicella-zoster pneumonia)

^‡‡One patient also had pulmonary hypertension

^§§§One patient presented with sepsis (without localizing features) as a complication of chemotherapy for HIV-related malignancy

****One patient also had thrombotic thrombocytopaenic purpura

All patients with PCP had acute lung injury; ICU and hospital survival was 66% and 54%, respectively. Twenty-one (60%) of these had newly diagnosed HIV infection during this hospital admission. Twenty-one of the 35 patients were mechanically ventilated; among this patient group, both ICU and hospital survival was 45%. In patients who were not mechanically ventilated, ICU survival was 93% (13/14) while hospital survival was 64% (9/14).

Twenty-three patients with chronic liver disease were admitted to the ICU on 29 occasions (13.1% of total ICU admissions). In this group ICU and hospital survival was 79% and 62%, respectively. The aetiology of the liver disease was hepatitis C in 22, alcoholic liver disease in four and hepatitis B in one.

Haematological diagnoses were responsible for 14 admissions (6%), mainly due to sepsis and/or acute renal failure (7 with lymphoma, 6 with Castleman disease and 1 with thrombotic thrombocytopaenic purpura). ICU survival was 57% in this group. Six (43%) of these patients required renal replacement therapy. Neurological diagnoses accounted for 31 (14%) of admissions, and of these 58% were due to HIV-related infections or encephalopathy. Among this subgroup outcome was good, with 89% and 81% surviving to ICU and hospital discharge, respectively.

HIV-associated admissions and receipt of HAART

Diagnoses were HIV-related in 113 episodes (51%); ICU and hospital survival in this group was 73% and 70%, respectively, compared with 84% (P = 0.05) and 76% (P = 0.057), respectively, in 109 patients with HIV-unrelated admissions. Patients with HIV- and non-HIV-associated diagnoses had similar admission APACHE II scores (median [IQR; interquartile range] of 18 [14–22], compared with 16 [11–23], P = 0.17); however, the former group had lower admission serum albumin levels (median [IQR] of 28 [18–28] compared with 29 [23–37], P = 0.001).

Of those admitted to the ICU with an HIV-associated problem and who required mechanical ventilation (MV), the duration of MV (median [IQR]) was six (3–17) days compared with two (1–5) for those with a non-HIV-associated admission (P < 0.001). The duration of ICU stay was five (3–11) days for those with an HIV-associated admission compared with two (3–5) for those with a non-HIV-associated admission (P = 0.001).

Among those receiving HAART (n = 107), survival to ICU and hospital discharge was 87% and 80%, respectively, compared with 75% (P = 0.10) and 67% (P = 0.049) in those not in receipt of HAART (n = 115). APACHE II scores were similar in those receiving HAART or not (17 [14–23.5] versus 16 [11–22], respectively; P = 0.171). Patients receiving HAART had higher serum albumin levels (27.5 g/L [20–36] versus 23.5 g/L [18–29]) and CD4 counts (150 [40–380] cells/μL compared with 50 [10–190] cells/μL, P = 0.03 and P = 0.003, respectively). Of patients receiving HAART, HIV viral loads were available in 55. Among these, 44 had undetectable viral loads, four had viral loads ≤1000 copies/mL, while seven had viral loads >1000 copies/mL.

New diagnoses of HIV infection

The majority of patients requiring ICU admission who were newly diagnosed with HIV infection had respiratory disease (71%) or neurological disease (12%) (Table 3). Forty of these 42 patients (95%) were admitted with HIV-related diagnoses. Pneumocystis pneumonia accounted for 50% of the ICU admissions in newly diagnosed HIV-infected patients, compared with only 14/180 (8%) of those with previously diagnosed HIV infection.

Table 3

Characteristics of 42 patients with a new diagnosis of HIV infection made just before or subsequent to admission to the ICU

Characteristics	First cohort 1999–2005, No. (%) of patients	Second cohort 2006–2009, No. (%) of patients	Combined 1999–2009, No. (%) of patients
Age (years); median (IQR)	34 (29–41)	51 (39–55)	35 (30–46)
Male gender	21 (67.7)	9 (81)	30 (71)
HIV risk factor
Heterosexual	20 (65)	4 (36)	24 (57)
MSM	10 (32)	4 (36)	14 (33)
Intravenous drug use	1 (3)	3 (27)	4 (9.5)
Admission diagnosis
PCP	18 (58.1)	3 (27.3)	21 (50)
TB	4 (12.9)	1 (9.1)	5 (11.9)
Bacterial pneumonia	3 (9.7)	2 (18.2)	5 (11.9)
Sepsis	0	3 (27.3)	3 (7.1)
HIV encephalopathy	1 (3.2)	0	1 (2.4)
CMV encephalitis	1 (3.2)	0	1 (2.4)
Cerebral toxoplasmosis	1 (3.2)	0	1 (2.4)
Cryptococcal meningitis	1 (3.2)	0	1 (2.4)
Thromboembolic disease	1 (3.2)	0	1 (2.4)
Postoperative	1 (3.2)	0	1 (2.4)
Castleman/acute RF	0	1 (9.1)	1 (2.4)
Acute on chronic RF	0	1 (9.1)	1 (2.4)
CD4 count, cells/μL; median (IQR)	20 (10–60)	70 (20–310)	20 (10–60)
APACHE II score; median (IQR)	16 (9–22)	16 (15–23)	16 (11–22)
Albumin, g/L; median (IQR)	22 (17–28)	22 (19–28)	22 (17–28)
Mechanical ventilation	22 (71%)	7 (63%)	29 (69%)
Pneumothorax	8 (26%)	1 (9%)	9 (21%)
Renal replacement therapy	3 (9.7%)	4 (36%)	7 (17%)
Length of ICU admission if survived (days); median (IQR)	6 (3–23)	6 (4–12)	6 (4–15)
Survival
To ICU discharge	23 (74)	6 (55)	29 (69)
To hospital discharge	18 (58)	6 (55)	24 (57)

MSM = men who have sex with other men; PCP = Pneumocystis jirovecii pneumonia; TB = tuberculosis; CMV = cytomegalovirus; RF = renal failure; APACHE = Acute Physiology and Chronic Health Evaluation; ICU = intensive care unit; IQR = interquartile range

Data shown are n (%) of patients or median values IQR

Survival to ICU and hospital discharge in patients with newly diagnosed HIV infection was 69% and 57%, compared with 80% and 74%, respectively, in patients with a known diagnosis prior to ICU admission; P = 0.132 and P = 0.012, respectively. APACHE II scores were similar in both groups, namely 16 (11–22) for the newly diagnosed group and 17 (13–25) for the previously diagnosed group. Renal replacement therapy was needed in 7/42 (17%) compared with 36/180 (20%) of previously diagnosed patients, respectively. A higher number of newly diagnosed patients required MV (29/42 [69%] versus 87/180 [48%] in those previously diagnosed, P = 0.017). Pneumothoraces occurred in nine/42 (21%) and six/180 (3%), respectively; P = 0.003. The newly diagnosed patients were more immunosuppressed than those with a previously known HIV diagnosis, with a median (IQR) CD4 count of 20 (10–100) cells/μL versus 140 (20–360) cells/μL, respectively; P = 0.001. They also had longer ICU stays with a median (IQR) of six (4–15) days compared with three (2–10) days; P < 0.001.

General medical patients

During the 10-year study period, an additional 2065 patients without known HIV infection and who had general medical conditions (excluding haematology and oncology patients) were admitted to the ICU on 2274 occasions. These patients were older than HIV-infected patients, median (IQR) age was 61 years (36–72), compared with 40 years (33–48); P < 0.001, but had similar APACHE II scores, 18 (13–25), compared with 16.5 (13–23); P = 0.32. Reasons for the ICU admission of the general medical patients are shown in Appendix 1 (available online only at: http://www.ijsa.rsmjournals.com/cgi/content/full/22/9/498/DC1). Among these general medical patients, their first episode ICU survival was 75% while survival to hospital discharge was 68%. By comparison, among HIV-infected patients first episode ICU and hospital survival was 78% and 70%; P = 0.452 and P = 0.458, respectively. Survival to ICU discharge among the general medical patients did not vary with time, being 74% between 1999 and 2005, and 76% between 2006 and 2009.

Predictors of survival in HIV-infected patients admitted to the ICU

Factors associated with survival to ICU discharge in univariable analysis are shown in Table 4. There was little evidence for a trend with time for better ICU survival (odds ratio [OR] = 1.40, 95% confidence interval [CI] = 0.72–2.72, P = 0.32) for patients admitted in 2006–2009 compared with patients who were admitted to the ICU in 1999–2005.

Table 4

Univariable analysis of factors associated with ICU survival of HIV-infected patients

Characteristics	Survivors no. (%)	Died no. (%)	OR (95% CI)	P value
Age in years, median (IQR)	39 (33–48)	42 (35–52)	0.78* (0.59, 1.03)	0.082
Gender
Female	35 (20)	12 (26)	1–
Male	140 (80)	35 (74)	1.37 (0.64, 2.95)	0.420
Ethnicity
White	93 (53)	26 (55)	1–
Black	58 (33)	15 (32)	1.08 (0.52, 2.25)	0.966
Other	24 (14)	6 (13)	1.12 (0.40, 3.16)
HIV risk factor
Heterosexual	65 (44)	20 (49)	1–
MSM	59 (40)	14 (34)	1.30 (0.59, 2.86)	0.793
Intravenous drug use	23 (16)	7 (17)	1.01 (0.37, 2.79)
Initial HIV diagnosis made on this hospital admission
Yes	29 (17)	13 (28)	1–
No	146 (83)	34 (72)	1.92 (0.90, 4.12)	0.091
Receipt of HAART
No	85 (49)	30 (64)	1–
Yes	90 (51)	17 (36)	1.87 (0.95, 3.66)	0.068
Cohort
1 (1999–2005)	86 (49)	27 (57)	1–
2 (2006–2009)	89 (51)	20 (43)	1.40 (0.72, 2.72)	0.324
Clinical Characteristics, median (IQR)
CD4 (cells/uL) ^†	155 (40–360)	60 (10–230)	1.76^‡ (1.17, 2.64)	0.007
APACHE II	16 (11–22)	22 (16–27)	0.47^§ (0.31, 0.70)	<0.001
Albumin (g/dL)	26 (20–35)	22.5 (17–28)	1.06** (1.02, 1.10)	0.004
Haemoglobin (g/dL)	8.7 (7.4–10.3)	7.4 (6.7–8.5)	1.36^†† (1.15, 1.61)	<0.001
PaO₂/FiO₂ ratio^‡‡	30 (0–50)	16 (0–29)	1.02 (1.00, 1.03)	0.029
Complications
Mechanical ventilation
No	95 (55)	8 (17)	1–
Yes	77 (45)	39 (83)	0.17 (0.07, 0.38)	<0.001
Pneumothorax
No	163 (95)	41 (87)	1–
Yes	9 (5)	6 (13)	0.38 (0.13, 1.13)	0.080
Renal replacement therapy
No	146 (85)	30 (64)	1–
Yes	26 (15)	17 (36)	0.31 (0.15, 0.64)	0.001

MSM = men who have sex with other men; HAART = highly active antiretroviral therapy; APACHE = Acute Physiology and Chronic Health Evaluation; PaO₂/FiO₂ = partial pressure of oxygen in arterial blood/fraction of inspired oxygen; IQR = interquartile range; ICU = intensive care unit; OR = odds ratio; CI = confidence interval

*OR = per 10-year increase

^†CD4 counts available in 175 episodes

^‡OR = per 1-log₁₀ increase

^§OR = per 10-unit increase

**OR = per 1 g/dL increase

^††OR = per 1 g/dL increase

^‡‡PaO₂/FiO₂ ratio recorded in 196 episodes

A P value of 0.1 was used as the cut-off value for the initial selection of factors to be considered for the multivariable analysis. Subsequently, given the small number of events (deaths), a model with five predictors was identified. Factors were grouped by clinical meaning and in each group the single factor of greatest clinical significance was selected for inclusion in the model. The need for MV was selected in preference to development of pneumothorax and admission PaO₂/FiO₂ ratio, as all but one of the patients who developed a pneumothorax was receiving MV at the time and PaO₂/FiO₂ values were not recorded in 26 episodes. Receipt of HAART was selected while awareness of HIV serostatus before ICU admission was discarded along with CD4 count, the latter also due in part to missing values (47 episodes). Albumin was selected while the other nutritional surrogate, haemoglobin, was discarded. The model was completed by including age and admission APACHE II score; time was not included due to its very weak association with survival in univariable analysis. In the multivariable model all factors displayed an association with survival, although the association with age was not statistically significant (Table 5).

Table 5

Multivariable analysis of factors associated with survival from the ICU

Variable	OR (95% CI)	P value
Age (per 10-year increase)	0.74 (0.53, 1.02)	0.068
APACHE II (per 10-unit increase)	0.55 (0.35, 0.87)	0.011
Albumin (per increase of 1 g/dL)	1.05 (1.00, 1.09)	0.041
HAART*	2.24 (1.01, 4.94)	0.047
Need for mechanical ventilation	0.14 (0.06, 0.36)	<0.001

HAART = highly active antiretroviral therapy; APACHE = Acute Physiology and Chronic Health Evaluation; ICU = intensive care unit; OR = odds ratio; CI = confidence interval

*Proportion receiving HAART was significantly different between the two cohorts;

P = 0.003

DISCUSSION

This study is the first description from the UK of a large cohort of HIV-infected patients to show improved mortality of patients in both early and late HAART eras. Our study shows that respiratory disease remains the most common reason for ICU admission, but that the proportion of patients with respiratory diagnoses (specifically, those with PCP) has decreased. Comparing patients in the early and late HAART eras, admission with an HIV-associated diagnosis almost halved (67% versus 34%), while the number of patients receiving HAART prior to ICU admission increased (from 37% to 60%). Receipt of HAART was associated with better survival, fewer HIV-associated admissions, and higher CD4 counts and serum albumin levels. Over the 10-year period of the study, increasing numbers of HIV-infected patients were admitted to the ICU. This parallels the increasing numbers of new diagnoses of HIV-infected patients at this treatment centre and does not reflect changes in case-mix. The APACHE II score on ICU admission did not change with time, inferring comparable severity of illness over the course of the study period.

The improved survival from ICU in the era of HAART demonstrated in this study is in keeping with observed trends in other studies from the USA (New York and San Francisco)^14,17,20 and Europe (Paris and Malaga).^15,16,21 By contrast with these studies, our study showed that ICU survival did not improve with time (1999–2005 = 77% and 2006–2009 = 78%); however, survival to hospital discharge did improve (68% versus 73%). Several factors may account for these findings. First, before the present study began we demonstrated improved outcome for patients with severe HIV-associated PCP, which was unrelated to the use of HAART and which we hypothesized was due to adoption of a low tidal volume ventilator strategy for acute lung injury.²² Subsequently, other groups have shown that the use of a low tidal volume strategy is associated with improved outcome from all-cause HIV-associated acute lung injury.²³ Second, other improvements in general ICU care occurred (but were inconsistently applied during the study period), including tight glycaemic control,^24,25 use of activated protein C (APC)²⁶ and early goal-directed therapy (EGDT) for patients with early sepsis.²⁷ However, a relatively small number of patients in our study were admitted with sepsis and our ICU does not routinely use APC; additionally EGDT is notoriously poorly delivered.²⁸ It is unlikely that these interventions impacted on outcome but as we did not systematically record these interventions in our patients, their potential contribution to survival remains uncertain.

We acknowledge several limitations to our study. First, it includes patients from a single university-affiliated ICU serving an urban UK population. Given the variation in clinical practice and patient demographics in other reports,^{14–18,20,21} this limits the generalizability of our findings and the ability to make direct comparisons with other patient populations. However, similar methods were employed in previous studies performed at this centre, allowing for valid internal comparison between different cohorts.^19,29 Second, the retrospective nature of the analysis gives rise to two shortcomings. First, data relating to length of survival were limited to ICU and hospital discharge; we did not follow up patients after hospital discharge, as we were concerned about incomplete data acquisition. Longer-term follow-up, for up to 12 months after ICU admission, which has been described in other studies,^21,30 was not available, thus limiting direct comparison. Second, data collection was incomplete for some variables including HIV viral load and CD4 count. At our centre the former is not routinely measured on admission to hospital of an acutely unwell patient with known HIV infection; additionally, presentation with severe PCP or other infections may render evaluation unreliable.³¹ We recorded each patient's CD4 count in a non-systematic way, i.e. only if it had been measured either on admission to the ICU or within the four weeks beforehand. This strategy differs from other studies where this variable had been recorded up to six months before ICU admission.^20,30

We found that respiratory diagnoses are becoming progressively less common as a reason for ICU admission among HIV-infected patients. While the incidence of PCP decreased, mortality remains high especially among those who required MV. By contrast, outcome from ICU was good for patients with liver and neurological diagnoses. Several variables were identified as being associated with improved survival from the ICU, namely: low admission APACHE II score; higher serum albumin level; being on HAART and not needing MV.

Survival to ICU and hospital discharge of HIV-infected individuals in our study (78% and 70%, respectively) were comparable with that of general medical patients admitted contemporaneously (75% and 68%, respectively), and were similar to outcomes reported during the study period by the Intensive Care National Audit and Research Centre (ICNARC) case-mix programme for adult patients admitted to general ICUs in the UK (79.4% and 71.4%, respectively).³² While HIV-infected patients and general medical patients in our study had similar APACHE II scores (16.5 and 18, respectively; ICNARC 16.5) inferring a similar illness severity, the general medical patients were older (median age 61 years versus 40 years; ICNARC cohort 63 years). Changes in physiology associated with increasing age and a patient's subsequent ability to withstand critical illness suggest an uncertain comparability between these two groups. Future studies of ICU admissions among HIV-infected patients will reflect the evolving HIV epidemic in the UK as it becomes a chronic health problem affecting an ageing population.

In the present study, almost a fifth of HIV-infected patients had their HIV diagnosis made just prior to or on admission to the ICU. Strikingly, the outcome was significantly worse for these patients with newly diagnosed HIV infection with only 69% surviving ICU and 57% surviving hospital. The British HIV Association (BHIVA) audit of causes of death among HIV-infected patients in the UK reported that 23% of deaths in 2005 were in individuals whose HIV infection had been diagnosed too late for effective treatment.³³ These observations underscore the need, recently emphasized by the UK joint British Association for Sexual Health and HIV/British HIV Association/British Infection Society National HIV testing guidelines,³⁴ to encourage HIV testing in health-care settings outside of genitourinary medicine clinics, with the aim of reducing the number of patients presenting to hospital with late-stage disease. No studies have been undertaken to support an economic benefit from early diagnosis of HIV infection, although it is intuitive that the resultant costs of hospital or ICU admission for patients with late, undiagnosed HIV infection are likely to outweigh the costs of aggressive screening programmes and outpatient management of previously diagnosed HIV disease.

No patient began HAART on the ICU during the study period. This policy is challenged by recent evidence from a multicentre study that early institution of HAART (median 14 days after diagnosis of an opportunistic infection) was associated with improved survival.³⁵ However, while some patients in this study were moderately unwell (and some with PCP received adjunctive corticosteroids) the study specifically excluded patients who were mechanically ventilated, were too ill to take oral medication or who had severely abnormal laboratory results such as anaemia, neutropenia, abnormal liver function tests or elevated serum creatinine levels.³⁵ By contrast, a single centre study from São Paulo demonstrated a survival benefit from starting HAART in critically ill patients, particularly if commenced within four days of ICU admission.³⁰ The survival advantage was still evident after correcting for disease severity (APACHE II score) and need for MV. Currently, the evidence base to support early institution of HAART in critically ill patients in the ICU is unclear.^22,23,30,35 Concerns about gastrointestinal absorption of HAART and drug–drug interactions add to this uncertainty.

In conclusion, since the availability of HAART for the long-term management of HIV-infected patients, the outcome from ICU admission for this patient group has been comparable to that among general medical patients at this UK treatment centre. A low admission APACHE II score, higher serum albumin level, being on HAART and not needing MV were all associated with improved survival. Data showing a better outcome among those with a known diagnosis of HIV infection prior to their ICU admission compared with those with newly diagnosed HIV infection on hospitalization or on ICU admission underscores the need to identify patients with undiagnosed HIV infection at an earlier stage in order to prevent their costly admission to hospital and to the ICU. The overall good outcomes from ICU admission suggest that HIV-infected patients should be offered admission to the ICU if it is likely to be of benefit.

References

Health Protection Agency. HIV in the United Kingdom: 2010 Report. Publication date 26 November 2010. See http://www.hpa.org.uk/web/HPAwebFile/HPAweb_C/1287145367237 (last checked 29 January 2011)

Wachter

, Luce

, Turner

, Intensive care of patients with the acquired immunodeficiency syndrome. Outcome and changing patterns of utilization. Am Rev Respir Dis 1986;134:891–6

Schein

, Fischl

, Pitchenik

, ICU survival of patients with the acquired immunodeficiency syndrome. Crit Care Med 1986;14:1026–7

Smith

, Levine

, Lewis

. Prognosis of patients with AIDS requiring intensive care. Chest 1989;96:857–61

Palella

, Delaney

, Moorman

, Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection: HIV Outpatient Study Investigators. N Engl J Med 1998;338:853–60

Valdez

, Chowdhry

, Asaad

, Changing spectrum of mortality due to human immunodeficiency virus: analysis of 260 deaths during 1995–1999. Clin Infect Dis 2001;32:1487–93

Mocroft

, Ledergerber

, Katlama

, Decline in the AIDS and death rates in the EuroSIDA study: an observational study. Lancet 2003;362:22–9

Lewden

, Chene

, Morlat

, HIV-infected adults with a CD4 cell count greater than 500 cells/mm³ on long-term combination antiretroviral therapy reach same mortality rates as the general population. J Acquir Immune Defic Syndr 2007;46:72–7

The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the Acute Respiratory Distress Syndrome. N Engl J Med 2000;342:1301–8

10.

Steinberg

, Hudson

, Goodman

, Efficacy and safety of corticosteroids for persistent acute respiratory distress syndrome. N Eng J Med 2006;354:1671–84

11.

Eichacker

, Gerstenberger

, Banks

, Cui

, Natanson

. Meta-analysis of acute lung injury and acute respiratory distress syndrome trials testing low tidal volumes. Am J Resp Crit Care Med 2002;166:1510–04

12.

Ranieri

, Suter

, Tortorella

, Effect of mechanical ventilation on inflammatory mediators in patients with acute respiratory distress syndrome: a randomised controlled trial. JAMA 1999;282:54–61

13.

Morris

, Wachter

, Luce

, Improved survival with highly active antiretroviral therapy in HIV-infected patients with severe Pneumocystis carinii pneumonia. AIDS 2003;17:73–80

14.

Narasimhan

, Posner

, DePalo

, Intensive care in patients with HIV infection in the era of highly active antiretroviral therapy. Chest 2004;125:1800–4

15.

Vincent

, Timsit

, Auburtin

, Characteristics and outcomes of HIV-infected patients in the ICU. Impact of the highly active antiretroviral treatment era. Intensive Care Med 2004;30:859–66

16.

Casalino

, Wolff

, Ravaud

, Impact of HAART advent on admission patterns and survival in HIV-infected patients admitted to an intensive care unit. AIDS 2004;18:1429–33

17.

Khouli

, Afrasiabi

, Shibli

, Outcome of critically ill human immunodeficiency virus-infected patients in the era of highly active antiretroviral therapy. J Intensive Care Med 2005;20:327–33

18.

Palacios

, Hildalgo

, Reina

, Effect of antiretroviral therapy on admissions of HIV-infected patients to an intensive care unit. HIV Med 2006;7:193–6

19.

Dickson

, Batson

, Copas

, Survival of HIV-infected patients in the intensive care unit in the era of highly active antiretroviral therapy. Thorax 2007;62:964–8

20.

Powell

, Davis

, Morris

, Survival for patients with HIV admitted to the ICU continues to improve in the current era of combination antiretroviral therapy. Chest 2009;135:11–7

21.

Coquet

, Pavie

, Palmer

, Survival trends in critically ill HIV-infected patients in the highly active antiretroviral therapy era. Crit Care 2010;14:R107

22.

Miller

, Allen

, Copas

, Improved survival for HIV infected patients with severe Pneumocystis jirovecii pneumonia is independent of highly active antiretroviral therapy. Thorax 2006;61:716–21

23.

Davis

, Morris

, Kallet

, Low tidal volume ventilation is associated with reduced mortality in HIV-infected patients with acute lung injury. Thorax 2008;63:988–93

24.

Van den Berghe

, Wouters

, Weekers

, Intensive insulin therapy in critically ill patients. N Engl J Med 2001;345:1359–67

25.

Finfer

, Chittock

, Yu-Shou Su

, , on behalf of the NICE-SUGAR Study Investigators. Intensive versus conventional glucose control in critically ill patients. N Engl J Med 2006;360:1283–97

26.

Bernard

, Vincent

J-L

, Laterre

P-F

, Efficacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med 2001;344:699–709

27.

Rivers

, Nguyen

, Havstad

, Early goal-directed therapy in the treatment of severe sepsis and septic shock. N ENgl J Med 2001;345:1368–77

28.

Reade

, Huang

, Bell

, , for the British Association for Emergency Medicine, the UK Intensive Care Society, the UK Society for Acute Medicine, the Australasian Resuscitation in Sepsis Evaluation (ARISE) Investigators and the Protocolized Care for Early Septic Shock (ProCESS) Investigators. Variability in management of early severe sepsis. Emerg Med J 2010;27:110–5

29.

Gill

, Greene

, Miller

, ICU admission in patients infected with the human immunodeficiency virus – a multicentre survey. Anaesthesia 1999;54:727–32

30.

Croda

, Croda

, Neves

, De Sousa dos Santos

. Benefit of antiretroviral therapy on survival of human immunodeficiency virus-infected patients admitted to an intensive care unit. Crit Care Med 2009;37:1605–11

31.

Shaunak

, Veryard

, Javan

. Severe Pneumocystis carinii pneumonia increases the infectious titre of HIV-1 in blood and can promote the expansion of viral chemokine co-receptor tropism. J Infect 2001;42:3–6

32.

Harrison

, Brady

, Rowan

. Case mix, outcome and length of stay for admissions to adult, general critical care units in England, Wales and Northern Ireland: the Intensive Care National Audit & Research Centre case mix programme database. Crit Care 2004;8:R99–R111

33.

British HIV Association (BHIVA). See http://www.bhiva.org/documents/ClinicalAudit/FindingsandReports/Mortality/Audit.ppt (last checked 20 October 2010)

34.

British HIV Association. UK National Guidelines for HIV Testing 2008. See http://www.bhiva.org/HIVTesting2008.aspx (last checked 30 January 2011)

35.

Zolopa

, Andersen

, Powderly

, Early antiretroviral therapy reduces AIDS progression/death in individuals with acute opportunistic infections: a multicenter randomized strategy trial. PLoS One 2009;4:e5575