Outcome predictors of systemic lupus erythematosus requiring admission to the intensive care unit

Abstract

BACKGROUND

Systemic lupus erythematosus (SLE) patients are admitted to the intensive care unit (ICU) for disease flares and infections, both of which carry a high mortality risk. Studies characterizing the outcome predictors of SLE are few in the Asian continent. This study characterized the clinical profile, treatment, and outcome predictors of ICU admissions with SLE.

AIM

To ascertain the reasons for ICU admission among SLE patients and to explore outcome predictors in these patients. The primary outcome was ICU mortality. Secondary outcomes included need for ventilation, inotropes, renal replacement therapy, and length of ICU and hospital stay.

METHODS

A retrospective study of 77 SLE patients was conducted in the medical ICU of a tertiary care teaching hospital in India. Clinical features, treatment, and outcomes of patients admitted between January 2018 and December 2022 were recorded. Factors associated with mortality were explored using bivariate and multivariate logistic regression analysis and reported as adjusted odds ratios with 95% confidence intervals.

RESULTS

The mean (SD) age was 31.1 (10.3) years; 83.1% were female. The median (interquartile) duration of SLE before admission was 12 (1-60) months; SLE was newly diagnosed in the current admission in 23.4%. The median Acute Physiology and Chronic Health Evaluation II score was 16.3 (14.5-18.2) and similar among survivors and non-survivors; 32 had evidence of disease flare, 44 had an infection, and one patient had an intracranial bleed. ICU admission was for respiratory failure (46.7%), hemodynamic instability (32.5%), and status epilepticus (14.3%). Twenty-nine patients (37.7%) had autoimmune hemolytic anemia, and 11 (14.3%) had diffuse alveolar hemorrhage. Immunomodulation included corticosteroids (96.1%), cyclophosphamide (33.8%), mycophenolate (23.4%), plasma exchange (13%), and immunoglobulins (11.7%). All patients received broad-spectrum antibiotics. Respiratory support, inotropes, and renal replacement therapy were required in 93.5%, 51.7%, and 32.5%, respectively. ICU mortality was 50.7% (95% confidence interval: 39%-62%). The mean ± SD hospital length of stay was 18.9 ± 14.3 days. On multivariate analysis, only shock (P = 0.004) was independently associated with mortality.

CONCLUSION

Intercurrent infection and disease flare are common reasons for ICU admission in SLE patients. Despite multimodal therapy, mortality is high. Shock was independently associated with mortality.

Key Words: Systemic lupus erythematosus; Lupus flare; Intensive care unit; Critically ill; Sepsis; Respiratory failure; Mortality; Shock

Core Tip: In this retrospective single-center study conducted in a medical intensive care unit (ICU) at a tertiary care teaching hospital in India, patients with systemic lupus erythematosus (SLE) and admitted to the ICU were noted to have a high mortality rate despite initiating multi-modal therapy. Respiratory failure and shock were the major reasons for ICU admissions in SLE. The differentiation between SLE flare and infection can be challenging and is the key to effective treatment.

INTRODUCTION

Systemic lupus erythematosus (SLE) is an autoimmune disease with protean manifestations. In general, SLE carries a 2-3 times increased risk of mortality adjusted for gender and age; 13.8% require intensive care unit (ICU) admission at some point in their illness[1]. Owing to the progress in diagnosis and treatment, the 5-year survival rates in SLE have now improved from 64%-87% in the 1980s to over 90%[2,3].

Sepsis or septic shock[4] and disease flare, as evidenced by high SLE Disease Activity Index (SLEDAI) 2000 score[1,5,6], are the most common reasons for ICU admission[2,7]. The distinction between SLE flare and superimposed infection is challenging since fever is present in both settings and occurs in 36%-86% of patients[8]. Although newer markers like serum triggering receptors expressed on myeloid cells and neutrophil CD64[9], neutrophil-lymphocyte ratio, NEUT-x, NEUT-y, NEUT-z indices, procalcitonin, C-reactive protein (CRP), erythrocyte sedimentation rate, and composite scoring systems[10] are used to differentiate between infection and disease flare, none of them are conclusive.

Several studies have explored the reasons for mortality among SLE patients admitted to the ICU. High Acute Physiology and Chronic Health Evaluation (APACHE) II[1,11] or Sequential Organ Failure Assessment score[8,10], ventilation[1], respiratory failure[12], inotrope requirement[1,11], cardiac failure[13], use of high dose immunosuppressives[14], intracranial hemorrhage, gastrointestinal bleeding[15], and antiphospholipid syndrome[10] have been associated with mortality. In a United States database cohort, infection was the leading cause of mortality (38.2%), followed by cardiac manifestations (12%)[16]. In another study, a shorter duration between SLE diagnosis and ICU admission was a poor outcome predictor[17]. Newer SLE disease scoring systems like British Isles Lupus Assessment Group[18], SLEDAI 2000[19], and Safety of Estrogens in Lupus National Assessment-SLEDAI[20], along with the traditional ICU-based scoring systems such as Sequential Organ Failure Assessment and APACHE II, may help prognosticate outcomes better among critically ill patients with SLE.

SLE cohorts necessitating ICU admission have been the subject of numerous studies from developed countries and registries, with mortality rates ranging from 18% to 78%[1,5,21]. In studies from India, ICU mortality among SLE patients admitted to the ICU ranges from 21%[5] to 81.3%[21], with pulmonary infection being a predictor of mortality[22]. Since data from India is limited, this study was undertaken to ascertain the reasons for ICU admission among SLE patients and explore the factors that are associated with mortality.

MATERIALS AND METHODS

Settings and study design

This retrospective, single-centre study included adult (> 18 years) patients with a diagnosis of SLE who were admitted to the medical ICU of a tertiary care teaching hospital in India between January 2018 and December 2022. Electronic medical records were searched using the International Classification of Diseases-10 code M-32 to identify patients admitted with a diagnosis of SLE and fulfilling the Systemic Lupus International Collaborating Clinics criteria[23]. Those with an overlap syndrome were also included. Patients with concomitant oncological diseases and those who had undergone transplants (e.g., renal) were excluded.

Detailed methodology and data collection

Patient demographics and clinical characteristics, including duration of the disease, comorbidities, and treatment details (type, duration, adherence to therapy) were noted. The primary reason for ICU admission was ascertained from the clinical records of the patient and recorded as documented by the primary treating team as to whether it was primarily due to disease flare, due to infection, or a combination of both. In situations where the cause of deterioration was unclear or it was related to other factors (e.g., postoperative monitoring not directly related to SLE), it was recorded as “others”. The indication(s) for ICU admission were documented. Illness severity, when documented in the records, was noted. This included the SLEDAI score[21] and APACHE-II score[10]. The lag time between hospital admission and ICU admission was noted. Details of treatment prescribed by the primary medical team and the critical care team were retrieved.

Vital signs on ICU admission and laboratory investigations were extracted from the records. The nature and extent of organ dysfunction were documented. Treatment included empiric and specific antibiotics for an intercurrent infection and immunomodulation for disease flare, which included pulse steroids, intravenous immunoglobulins, and plasma exchange. The decision on escalation or de-escalation of immunomodulation was made by the intensivist in consultation with the treating team (rheumatology, general medicine), based on the reason for deterioration. Organ support, by way of invasive or non-invasive mechanical ventilation, vasoactive agents, and renal replacement therapy (RRT) was provided as per the existing guidelines for the management of organ dysfunction or failure. The primary outcome of interest was ICU mortality. Secondary outcomes included the need for ventilation, inotropes, RRT, and length of ICU and hospital stay.

Statistical analysis

Demographic data, comorbidities, treatment, and outcome details were recorded into data abstraction forms. Summary statistics were used to report demographic and clinical characteristics. Patients were compared across two groups, i.e., survivors and non-survivors. Continuous variables were expressed as mean ± SD or median with interquartile range (IQR), based on the distribution of data. Student t-test was used according to normality assumptions for the comparison of continuous variables, whereas the χ² test was used for categorical variables. A P value of < 0.05 was taken as significant. Bivariate logistic regression analysis was done to look for possible associations between the various parameters and mortality. Clinical and statistically significant (P < 0.2) predictors were included in the multivariable analysis. STATA^TM statistical software version 16 (Stata Corp, College Station, TX, United States) was used for data analysis.

The study was approved by the Institutional Review Board, approval No. 2407170.

RESULTS

Study population and baseline characteristics

From January 2018 to December 2022, 77 patients with SLE, with a mean ± SD age of 31.1 ± 10.3 years, were admitted to the ICU. There was a female preponderance (4.9:1). The baseline characteristics are summarized in Table 1. The median duration of the disease prior to ICU admission was 12 (1-60) months. Eighteen patients (23.4%) were newly diagnosed with SLE during the current admission, and 24 patients (32.5%) had defaulted on their treatment before ICU admission. Comorbidities included hypertension (23.4%) and diabetes (7.8%). The median (IQR) SLEDAI score was available in 31 patients and was 25.6 (21.0-30.2). The APACHE-II score (n = 43) was 16.3 (14.5-18.2). Laboratory characteristics are summarized in Table 2. Of note, summary statistics of the study population showed anemia, without leucocytosis or leucopenia, or thrombocytopenia. Elevated CRP and procalcitonin, anti-nuclear antibody positivity, low complement levels, and high titres of double-stranded DNA were noted (Table 2). Sixteen patients had a positive anti-phospholipid antibody.

Table 1 Baseline characteristics of systemic lupus erythematosus patients admitted to the intensive care unit, n (%).

Parameter	All patients (n = 77)	Survivors (n = 38)	Non-survivors (n = 39)	P value
Age (year), mean ± SD	31.1 (10.3)	31.5 (10.2)	30.8 (10.5)	0.76
Female gender	64 (83.1)	32 (84.2)	32 (82.1)	0.8
Diabetes mellitus	6 (7.8)	3 (7.9)	3 (7.7)	0.98
Hypertension	18 (23.4)	13 (34.2)	5 (12.8)	0.03
Newly diagnosed SLE	18 (23.4)	10 (25.6)	8 (20.5)	0.63
Prior treatment	56 (72.7)	28 (73.7)	28 (71.8)	1.0
Drug defaulter	25 (32.5)	14 (36.9)	11 (28.2)	0.42
Duration of disease, median (IQR) months	12 (1-60)	24 (1-48)	12 (0-72)	0.58
SLEDAI (n = 31), median IQR	25.6 (21.0-30.2)	27.5 (21.2-33.8)	21.6 (15.5-27.7)	0.22
APACHE-II (n = 43), median IQR score	16.3 (14.5-18.2)	15.5 (12.7-18.2)	17.1 (14.5-19.7)	0.37
Lag time, median (IQR) days	3 (0-7)	3 (0-7)	2 (0-7)	0.36
Hemodynamic instability	25 (32.5)	9 (23.7)	16 (41)	0.29
Respiratory failure	36 (46.7)	19 (50)	17 (43.9)	0.65
Status epilepticus	11 (14.3)	8 (21.1)	3 (7.7)	0.11
Low GCS	2 (2.6)	1 (2.6)	1 (2.6)	1.0
Hypertensive urgency	1 (1.3)	0	1 (2.6)	1.0
AKI needing dialysis	1 (1.3)	0	1 (2.6)	1.0
Hematemesis	1 (1.3)	1 (2.6)	0	0.49

All values are percentages unless specified. SLE: Systemic lupus erythematosus; SLEDAI: Systemic Lupus Erythematosus Disease Activity Index; APACHE-II: Acute Physiology and Chronic Health Evaluation score; GCS: Glasgow Coma scale; IQR: Interquartile range; AKI: Acute kidney injury.

Table 2 Laboratory investigations at intensive care unit admission.

Parameter		All patients (n = 77)	Survivors (n = 38)	Non-survivors (n = 39)	P value
Hemoglobin, g/dL		8.3 (2.2)	8.4 (2.5)	8.4 (2.5)	0.6
Leucocyte count/mm3		9952 (7984)	10647 (9309)	9274 (6493)	0.77
Platelet count, lakhs/mm3		1.52 (1.03)	1.72 (1.17)	1.33 (0.85)	0.10
Serum creatinine, mg/dL		2.12 (2.0)	2.3 (2.3)	1.9 (1.6)	0.38
Serum albumin, g/dL		2.42 (0.67)	2.28 (0.65)	2.56 (0.67)	0.08
Admission pH		7.39 (0.12)	7.38 (0.13)	7.41 (0.12)	0.27
Bicarbonate, mmol/L		18.2 (6)	18.2 (6.7)	18.2 (5.1)	0.98
Admission lactate, median (IQR), mmol/L		1.5 (0.9-2.8)	1.7 (0.8-2.7)	1.4 (1.1-2.9)	0.12
CRP, (n = 49), median (IQR), mg/L		25 (5.6-100)	25 (7.2-124)	24 (4-100)	0.56
Procalcitonin, (n = 45), median (IQR), μg/dL		3.1 (0.38-25)	4.93 (0.72-30.8)	2.43 (0.37-23)	0.59
Low serum C3 (n = 73)		51.4 (30.2)	51.5 (30.6)	51.4 (30.3)	0.99
Low serum C4 (n = 71)		13.3 (8.7)	13.3 (7.7)	3.2 (9.8)	0.94
ANA (n = 60)	0	3	1	2	0.36
	1+	4	4	0
	2+	28	15	13
	3+	18	8	10
	4+	7	3	4
dsDNA (n = 73)		452 (374)	490 (374)	414 (374)	0.39

All values are mean ± SD unless specified. CRP: C-reactive protein; IQR: Interquartile range; ANA: Antinuclear antibody; dsDNA: Double-stranded DNA; n: Indicates the number of patients in whom the data were available.

Indications for ICU admission

The lag time from symptom onset to ICU admission was 3 days (0-7). The most common reasons for ICU admission were respiratory failure (46.7%), hemodynamic instability (32.5%), status epilepticus (14.3%), and low Glasgow coma scale (2.6%). One patient each was admitted for acute kidney injury (AKI) requiring RRT, hypertensive urgency, and haematemesis (Table 1). Thirty-two patients had SLE flare as the reason for ICU admission, 44 had some form of infection, and one patient presented with intracranial bleeding.

Organ dysfunction during ICU stay

At the time of admission to the ICU, 11 patients (14.3%) were in shock; 47 patients (61.0%) developed shock at some point during their stay in the ICU. Features of myocarditis were noted in 30 patients (39.5%). Respiratory involvement was frequent and observed in 72 patients (93.5%), while the central nervous system (CNS) was involved in 33 patients (42.9%). Nephritis was present at the time of ICU admission in 61 patients (79.2%), autoimmune hemolytic anemia in 39 (37.7%), and diffuse alveolar hemorrhage (DAH) in 11 (14.3%) patients (Table 3).

Table 3 Organ involvement during intensive care unit stay, n (%).

Parameter	All patients (n = 77)	Survivors (n = 38)	Non-survivors (n = 39)	P value
Central nervous system involvement	33 (42.9)	22 (57.9)	11 (28.2)	0.01
Respiratory involvement	72 (93.5)	33 (86.8)	39 (100)	0.03
Shock during the hospital stay	47 (61.0)	10 (26.3)	37 (94.9)	< 0.001
Myocarditis	30 (39.0)	14 (36.8)	16 (41.0)	0.78
Nephritis	61 (79.2)	31 (81.6)	30 (76.9)	0.62
Serositis	19 (24.7)	7 (18.4)	12 (30.8)	0.21
Autoimmune hemolytic anemia	29 (37.7)	14 (36.8)	15 (38.5)	0.88
Diffuse alveolar hemorrhage	11 (14.3)	2 (5.3)	9 (23.1)	0.03

All values in parentheses are percentages unless specified.

Treatments received and outcomes

Immunomodulation at the time of admission and during ICU stay included corticosteroids (96.1%), cyclophosphamide (33.8%), mycophenolate (23.4%), plasma exchange (13%), and immunoglobulins (11.7%). During the current admission, pulse steroids were administered in 34 patients (44.2%). All patients were administered empiric broad-spectrum antibiotics at admission, given the challenges in differentiating disease flare from infection. Respiratory support, inotropes, and RRT were required in 93.5%, 51.7%, and 32.5% respectively. ICU and hospital length of stay were 8.1 (7.2) days and 18.9 (14.3) days, respectively. ICU mortality was 50.7% [95% confidence interval (CI): 39%-62%] (Table 4).

Table 4 Treatment and outcomes of patients with systemic lupus erythematosus admitted to the intensive care unit, n (%).

Parameter	All patients (n = 77)	Survivors (n = 38)	Non-survivors (n = 39)	P value
Prior use of steroids	52 (67.5)	27 (71.1)	25 (64.1)	0.52
Steroids during stay	74 (96.1)	37 (97.4)	37 (94.9)	0.57
Pulse steroids	34 (44.2)	19 (50)	15 (38.5)	0.31
Intravenous immunoglobulin	9 (11.7)	3 (7.9)	6 (15.4)	0.31
Plasma exchange	10 (13)	7 (18.4)	3 (7.7)	0.16
Cyclophosphamide	26 (33.8)	19 (50)	7 (18)	0.003
Mycophenolate mofetil	18 (23.4)	12 (31.6)	6 (15.4)	0.09
Biologics	1 (4.5)	1 (2.6)	0	0.32
Respiratory support	72 (93.5)	33 (86.8)	39 (100)	0.03
Invasive ventilation	57 (74)	18 (47.4)	39 (100)	< 0.001
NIV alone	15 (19.4)	15 (100)	0
NIV and Invasive ventilation	38 (49.4)	20 (52.6)	18 (46.2)	0.57
Need for inotropes	41 (51.7)	2 (5.2)	39 (100)	0.001
Renal replacement therapy	25 (32.5)	14 (36.9)	11 (28.2)	0.42
Packed RBC transfusion	57 (74)	26 (68.4)	31 (79.5)	0.27
ICU length of stay	8.1 (7.2)	7.4 (6.2)	8.9 (8.1)	0.36
Hospital length of stay	18.9 (14.3)	22.3 (16.2)	16.5 (11.5)	0.04

All values are percentages unless specified. NIV: Non-invasive ventilation; RBC: Red blood cell; ICU: Intensive care unit.

Outcome predictors

Demographic characteristics, duration of disease, treatment default, lag time to presentation, illness severity score, organ dysfunction at presentation, and laboratory tests were similar in survivors and non-survivors (Tables 1 and 2). Need for respiratory support, shock during hospital stay, and DAH were more frequently observed among non-survivors, while survivors had a higher frequency of CNS involvement (Table 3). Among the immunomodulators, patients who were on cyclophosphamide had a lower mortality (Table 4). The combination of ventilatory support and inotropes was associated with 100% mortality; the need for RRT or RBC transfusions was not associated with death (Table 4).

On bivariate logistic regression analysis, sepsis, shock during the hospital stay, need for inotropes, and length of hospital stay were associated with mortality, while hypertension appeared to be protective. On multivariate logistic regression analysis, although thrombocytopenia [odds ratio (OR) = 1.0, 95%CI: 1.0-1.0, P = 0.06], and admission lactate (OR = 1.82, 95%CI: 0.95-3.5, P = 0.07) showed a trend towards association with mortality, only shock during hospital stay (OR = 726, 95%CI: 8-65088, P = 0.004) was independently associated with mortality. The high OR and CI for shock are probably reflective of the small sample size.

DISCUSSION

In this retrospective study of 77 SLE patients admitted to the ICU, SLE was diagnosed 12 (1-60) months prior to ICU admission; 32.5% had defaulted treatment. In 23.4%, the diagnosis was made during the current admission. The lag time from symptom onset to ICU admission was 3 (0-7) days. The median APACHE-II score was 16.3 (14.5-18.2). Respiratory failure and hemodynamic instability were the most frequent reasons for ICU admission, with 41.6% suspected to have a disease flare. Forty-two patients (54.6%) received immunomodulation in the form of pulse steroids, intravenous immunoglobulins, and plasma exchange. Respiratory support, inotropes, and RRT were required in 93.5%, 51.7%, and 32.5%, respectively. ICU mortality was 50.7%. Hospital length of stay was 18.9 ± 14.3 days. On multivariate logistic regression analysis, only shock (P = 0.004) was independently associated with mortality.

The mean ± SD age in our study of 31.1 ± 10.3 years and the female preponderance (83%) are comparable to data from other studies[2,10,21,24]. The incidence of diabetes mellitus was low (7.8%) and similar to the 8% reported by Aragón et al[10]. The lag time from SLE diagnosis to ICU admission in our study was 12 (1-60) months. A systematic review reported a wide range in the time from diagnosis to ICU admission, from 3.6 months (IQR: 0.2-9.4) to 102 months (IQR: 1-180)[1]. Newly diagnosed SLE constituted 23.4% of the patients in our study, contrasting 38.2% in the study by Bhargav et al[21] and 14% who were diagnosed within the last year in a study by Aragón et al[10]. The mean SLEDAI score of 25.3 was higher than the scores reported by Aragón et al[10] (SLEDAI 9), Bhargav et al[21] (SLEDAI 16), and Fatemi et al[25] (SLEDAI 11.3)[10,21,25]. The mean APACHE score in our study (16.3) was similar to that (19.8) reported by Fatemi et al[25].

Studies categorize the reason for ICU admission in SLE patients as infection or flare[1,26]; both have been associated with higher mortality[7,16,22,27]. In our study, disease flare accounted for 41.5% of the ICU admissions, while 57.1% had evidence of infection, of whom 24 (31.1%) also had disease flare. In other studies, the frequency of flare ranged from 11.83%[10] to 55.5%[28], while infection accounted for 27.6%[10] and 42.2%[28] of the admissions. The challenge of categorizing the deterioration as a disease flare or infection resulted in all our patients receiving broad-spectrum antibiotics.

Although 93.5% required some form of respiratory support in our study, the primary reason for ICU admission was respiratory failure in 46.7% of our patients, which was similar (40%) to the study by Alvarez Barreneche et al[28]. Other studies reported respiratory failure in 9.32%[10] to 76.5%[21] of patients. Pneumonia occurred in 40 patients (51.9%) in our study; in other studies, the incidence ranged from 13.8% to 60.2%[21,25,29]. DAH occurred in 14.3% of patients in our study, and was similar to the reports by Fatemi et al[25] (13.8%) and Bhargav et al[21] (16.6%). In our study, DAH was associated with mortality (P = 0.03). Hemodynamic instability as a reason for ICU admission was more frequent (32.5%) in our study when compared with the 11.8% by Aragón et al[10]. Other reasons for ICU admissions in our study were seizures, low Glasgow coma scale, and hematemesis.

In the current study, laboratory parameters at admission, including disease-specific markers like C3, C4, and double-stranded DNA, and infection parameters such as procalcitonin and CRP (Table 2), did not discriminate survivors from non-survivors. A meta-analysis by Liu et al[30] suggested that procalcitonin might not help differentiate infection from flare in SLE cases. A study by Gandino et al[31] observed that SLE-specific parameters were similar among survivors and non-survivors.

In our study, organ involvement (Table 3) included nephritis (79.7%), myocarditis (39%), and CNS involvement (42.9%). The overall incidence of lupus myocarditis worldwide is approximately 9%[32], while neuropsychiatric lupus and cardiac involvement in another study were reported as 54% and 14.7%, respectively[21].

In our study, immunomodulation with pulse steroids (44.2%), immunoglobulins (11.7%), plasma exchange (13%), or biologics (4.5%) did not influence survival (Tables 4 and 5). Aragón et al[10], also did not observe reduced mortality with steroids (37.9%), plasma exchange (11.8%), immunoglobulins (7.5%), and biologics (3.9%) The survival benefit among patients on cyclophosphamide prior to ICU admission in our cohort (Table 4) is interesting and warrants further study, since literature reports increased incidence of infections among patients treated with steroids and cyclophosphamide[33,34]. There are case reports from India regarding the effective use of plasma exchange as a therapeutic modality for SLE with flare[35] or with thrombotic thrombocytopenic purpura[36].

Table 5 Factors associated with intensive care unit mortality.

Parameter	Bivariate analysis OR (95%CI)	P value	Multivariate analysis OR (95%CI)	P value
Age	0.99 (0.95-1.03)	0.76	-	-
Female gender	1.16 (0.35-3.85)	0.8	-	-
Diabetes mellitus	0.97 (0.18- 5.14)	0.98	-	-
Hypertension	0.28 (0.08-0.89)	0.03	0.32 (0.02-4.38)	0.39
Newly diagnosed SLE	1.29 (0.45-3.73)	0.63	-	-
SLEDAI 2K	0.96 (0.89-1.02)	0.23	-	-
Duration of disease	0.99 (0.98-1.00)	0.33	-	-
Drug defaulter	0.67 (0.25-1.75)	0.42	-	-
APACHE-II score	1.04 (0.94-1.16)	0.37	-	-
Lag time	1.04 (0.96-1.13)	0.36	-	-
Shock at admission	3.01 (0.73-12.5)	0.12	-	-
SLE flare	0.55 (0.22-1.54)	0.26	-	-
Sepsis	2.64 (1.03-6.76)	0.04	-	-
Hemoglobin	1.04 (0.84-1.28)	0.60	-	-
Platelet count	0.99 (0.99-1)	0.10	1.0 (1.0-1.0)	0.06
Serum creatinine	0.89 (0.71-1.13)	0.38	-	-
Serum albumin	1.91 (0.92-3.96)	0.08	0.83 (0.26-2.6)	0.75
Admission pH	8.95 (0.18-442)	0.27	-	-
Bicarbonate	0.99 (0.92-1.08)	0.98	-	-
Admission lactate	1.14 (0.95-1.37)	0.12	1.82 (0.95-3.5)	0.07
Anti-nuclear antibody	1.29 (0.74-2.2)	0.36	-	-
Double-stranded DNA	0.99 (0.98-1.01)	0.39	-	-
Non-invasive ventilation	0.77 (0.31-1.88)	0.57	-	-
Shock during the hospital stay	51.8 (10.5-255)	< 0.001	726 (8.0-65088)	0.004
Need for inotropes	5.57 (2.49-12.5)	0.001	-	-
Renal replacement therapy	0.67 (0.25-1.8)	0.42	-	-
Transfusion	1.78 (0.63-5.03)	0.27	-	-
Immunomodulation1	0.77 (0.31-1.88)	0.56	-	-
ICU length of stay	1.03 (0.96-1.10)	0.36	-	-
Hospital length of stay	0.96 (0.92-0.99)	0.04	-	-

¹Immunomodulation included the use of pulse steroids, intravenous immunoglobulins, and/or plasma exchange (n = 42).

OR: Odds ratio; CI: Confidence interval; SLE: Systemic lupus erythematosus; SLEDAI 2K: Systemic Lupus Erythematosus Disease Activity Index 2000; APACHE: Acute Physiology and Chronic Health Evaluation; ICU: Intensive care unit.

The proportion of patients in our study who required invasive mechanical ventilation (74%) was similar to the 81.3% in the study by Bhargav et al[21], but higher when compared to 31.1% in a Columbian study[28] and 27.2% in the study by Aragón et al[10]. Inotropes were required in 51.7% of our subjects. In the various studies, inotropes were required in 8.6%, 50%, and 20% of the patients[10,21,28]. RRT was required for 32.5% of our patients, 36.2% in the study by Aragón et al[10], and 27.4% in the study by Bhargav et al[21].

The mortality of patients admitted to the ICU with SLE varies from 12.1%[28] to 81.3% (Table 6)[21]. The wide mortality range may reflect ethnic variations, differences in illness severity, lag time to presentation to the hospital, the reason for admission to the ICU, the extent and nature of immunosuppression, and the prevalence of multidrug-resistant organisms in the hospital setting. Cohorts from developed countries have reported lower mortality, ranging from 21.2% in a retrospective cohort from the United States[29] to 28% in a retrospective study from the United Kingdom[26]. The mortality from cohorts in developing countries has ranged from 12.1% in Colombia[28] to 81.3% in India[21]. The mortality of 50.7% in our cohort is similar to the mortality figures of 50% from Iran[25], 47% from Taiwan[15], and 57% from Thailand[34] but higher than the mortality of 32.7% reported from a study in Mexico[11].

Table 6 Summary of studies on systemic lupus erythematosus patients requiring intensive care unit admission.

Ref.	Country, study type	No. of patients	Reasons for ICU admission	Mortality (%)	Factors associated with mortality
Current study, 2024	India, retrospective	77	SLE flare (41.6%), infection (57.1%)	50.7	Shock, thrombocytopenia, admission lactate
Bhargav et al[21], 2023	India, retrospective, prospective	102	Pneumonia	81.3	APACHE II > 16, infection, lymphopenia, creatinine > 1.3, hypotension
Gopalan et al[5], 2023	India, retrospective	87	Infection (58%), disease flare (51%)	21	High APACHE II, fever, AKI, low albumin, low hematocrit, low GCS, sepsis, ventilation, inotropes
Aragón et al[10], 2020	Colombia, retrospective	188	Infection	20.2	Higher SOFA, ventilation, inotrope need, APLA
Alvarez Barreneche et al[28], 2019	Colombia, retrospective	33	Lupus flare	12.1	Infection
Fatemi et al[25], 2017	Iran, retrospective	94	Pneumonia, DAH, seizure	50	High APACHE II, septic shock, duration of mechanical ventilation
Siripaitoon et al[34], 2015	Thailand, retrospective	61	Respiratory failure, shock, AKI	57	High APACHE, ventilator-associated pneumonia, inotrope use
Han et al[29], 2013	United States, retrospective	70	Infectious (35.7%) and non-infectious (64.3%)	21.2	Higher APACHE II
Namendys-Silva et al[11], 2009	Mexico, retrospective	104	Infection	32.7	High APACHE II, inotrope use
Hsu et al[15], 2005	Taiwan, retrospective	51	Pneumonia with ARDS	47	Intracranial and GI hemorrhage, septic shock
Williams et al[26], 2002	United Kingdom, retrospective	61	Infection	28	Leucopenia, high severity of illness score, cyclophosphamide

ICU: Intensive care unit; APACHE: Acute Physiology and Chronic Health Evaluation; AKI: Acute kidney injury; GCS: Glasgow coma scale; SOFA: Sequential Organ Failure Assessment; APLA: Anti-phospholipid antibodies; DAH: Diffuse alveolar hemorrhage; ARDS: Acute respiratory distress syndrome; GI: Gastrointestinal.

Although several factors tended (P > 0.05) to be independently associated with mortality in our study and included thrombocytopenia (P = 0.06) and high lactate level (P = 0.07) on admission, only shock during hospital stay (P = 0.004) was significantly associated with mortality. An ICU-based study by Hsu et al[15] showed that gastrointestinal bleeding, intracranial hemorrhage, and septic shock during ICU stay were associated with a greater risk of death. Higher APACHE II scores were indicative of mortality in a few studies[10,29]. Bhargav et al[21] showed that APACHE II scores > 16, infections, lymphopenia, vasopressor support, and elevated creatinine led to poor outcomes[21]. The factors associated with mortality in the various studies are summarized in Table 6.

Limitations

The study must be interpreted in the light of the following limitations. This was a single-center retrospective study with the challenges of missing data for some patients for variables such as the staging of lupus nephritis and SLEDAI score. Given that patients were admitted under multiple units (rheumatology, medical units), there was variation in the type and dose of immunosuppressants used, as well as the choice of antibiotics, and this could have influenced outcomes. The mortality was probably overestimated since only the sickest patients were admitted to the ICU, and this is reflected in the APACHE-II score. Despite these limitations, this study adds to the current literature on the outcomes of SLE patients admitted to the ICU and the factors associated with poor outcomes.

CONCLUSION

SLE patients are primarily admitted to the ICU for the management of disease flares and infections and present with respiratory failure and circulatory shock. Despite multi-modal therapy with immunomodulation, antibiotics, and organ support, mortality is high. Only shock during stay was associated with poor outcomes.