|Home | About | Journals | Submit | Contact Us | Français|
Nosocomial infections are linked to rising morbidity and mortality worldwide. We sought to investigate the pattern of nosocomial sepsis, device usage, risk factors for mortality and the antimicrobial resistance pattern of the causative organisms in medical intensive care units (ICUs) in an Indian tertiary care hospital.
We conducted a single-centre based prospective cohort study in four medical ICUs and patients who developed features of sepsis 48hours after admission to the ICUs were included. Patients' demographics, indwelling device usage, microbiological culture reports, drug resistance patterns and the outcomes were recorded. The Acute Physiology and Chronic Health Evaluation (APACHE) III score and the relative risk of variables contributing towards non recovery were calculated.
Pneumonia (49%) was the commonest nosocomial infection resulting in sepsis, followed by urosepsis (21.8%), bloodstream infection (BSI) (10.3%) and catheter-related bloodstream infection (CRBSI) (5%). Sixty three percent of the Acinetobacter baumannii and 64.4% of the Pseudomonas aeruginosa were multidrug-resistant (MDR). Seventy percent of the Klebsiella pneumoniae were extended spectrum beta-lactamase producers and 7.4% were resistant to carbapenems. Forty three percent of the Staphylococcus aureus were methicillin-resistant S. aureus. Resistance to carbapenems was 35.2% in this study. High APACHE III scores (P=0.006 by unpaired t-test) and chronic kidney disease (P=0.023) were significantly associated with non-recovery.
A high degree of multidrug resistance was observed among both Gram-positive and -negative organisms in nosocomial sepsis patients. Carbapenem resistance was a common occurrence. Chronic kidney disease and high APACHE III scores were significantly associated with non-recovery. Male gender and sepsis leading to cardiovascular failure were the independent predictors of mortality.
Nosocomial infections represent a significant problem to patients admitted to health care institutions. Hospital acquired infections are associated with high level of morbidity and mortality and increase the economic burden to the patient and governing establishments.1 Patients admitted to intensive care units (ICUs) are at greater risk of developing nosocomial infections than patients admitted to other hospital wards. Hospital acquired infections affect both developed and resource poor nations. A prevalence study conducted by the World Health Organization (WHO) in 55 hospitals of 14 nations revealed that an average of 9% of hospital patients acquired nosocomial infections.2 The highest rate of nosocomial infections was reported from hospitals in the Eastern Mediterranean (12%) and South East Asia (10%). Rate of healthcare-associated infections among ICU admitted patients across seven different cities in India have been reported to be up to 4.4% including 61.3% central line associated bloodstream infection, 29.6% ventilator associated pneumonia and 9.0% catheter associated urinary tract infections.3
Occurrence of sepsis is one of the most frequently observed complications among ICUs' patients. Clinical sepsis itself increases the risk of mortality in such patients.4 Like other infectious diseases, the clinico-microbial epidemiology of nosocomial infections and related sepsis is dynamic and subject to constant improvements in healthcare system. Therefore, a coherent intensive care of sepsis necessitates having ample acquaintance with the prevailing aetiologic organisms and their antimicrobial susceptibility profile within healthcare sectors of respective geography.
This study was conducted to investigate the pattern of nosocomial sepsis in ICUs, the antimicrobial resistance pattern of the pathogens and the prognostic factors influencing the outcome of the patients.
This study was conducted in four medical ICUs in a tertiary care hospital from January 2012 to December 2013. Patients who developed features of sepsis 48hours after admission to ICU were included. Patients who already had a focus of infection before admission to the ICU were excluded. Written informed consents were obtained from all patients before inclusion in the study. An approval was obtained from the institutional ethics committee prior to commencement of the study.
Standard definitions of infections were followed as formulated by the International Sepsis Forum Consensus Conference (2005).5 Sepsis was defined as presence of any two or more of the following conditions along with proven or suspected microbial aetiology: (1) hyperthermia (oral temperature >38°C) or hypothermia (<36°C); (2) tachypnoea (>24breaths/min); (3) tachycardia (heart rate >90beats/min); (4) leucocytosis (>12000/μl), leucopenia (<4,000/μl), or >10% bands.6 All patients were followed up either until discharge from the hospital or death. Data abstraction tool was used to capture the following details of the patients: age, sex, reason for ICU admission, comorbid conditions like diabetes mellitus, chronic liver disease (CLD), chronic kidney disease (CKD), chronic obstructive pulmonary disease (COPD), human immunodeficiency virus (HIV) infection, details of chemotherapy, steroids or immunosuppressants in the past 6months and coma (Glasgow Coma Scale <6). The use of endotracheal tubes, urinary catheters, central venous catheters, haemodialysis catheters and the number of days of usage prior to onset of sepsis were also noted.
The source of infection for the development of sepsis was identified whenever possible. Ventilator associated pneumonia (VAP) was diagnosed if pneumonia developed 48hours after mechanical ventilation.
Cultures of the relevant clinical specimens were done and interpreted as recommended.7 The culture technique used for catheter tip was Roll-flush-plate technique.8 Sensitivity was done by disc diffusion method9 according to Clinical and Laboratory Standards Institute guidelines. Multidrug-resistant (MDR) and pandrug-resistant (PDR) Pseudomonas aeruginosa and Acinetobacter baumannii were defined based on a review by Falagas et al.10
The patients were started on empirical antibiotic therapy while awaiting culture reports. Based on the organisms isolated and antimicrobial susceptibility profile specific treatments were initiated. Acute Physiology and Chronic Health Evaluation (APACHE) III11 scores were calculated for the worst recorded variables within 24hours of the onset of sepsis. Multi-organ dysfunction after onset of sepsis and the outcomes of the patients were documented.
The variables were presented as mean±standard deviation, median and interquartile range, frequency and percent as appropriate. The association of the different categorical variables, such as, comorbidities, source of infection and microorganism, with non-recovery was assessed and the relative risk (RR) was calculated. The correlation between APACHE III scores and outcome was assessed by unpaired t-test. Univariate logistic regression analysis was done to determine the risk factors of mortality. Further, multivariate logistic regression analysis was carried out using ‘Forward Wald’ method to ascertain the independent predictors of mortality. All tests of significance were two sided with P-value <0.05 having statistical significance. Data was analysed using the Statistical Package for Social Sciences (SPSS) version 16.0 (Chicago, IL, USA).
Out of the 174 patients of nosocomial sepsis 62.6% (109 patients) were males and 37.4% (65 patients) were females. The mean age of males was 53years (±16.25) and of females was 59years (±16.10). The patient demographics are listed in Table 1. A majority of the patients (77%) had at least two indwelling devices.
The pattern of nosocomial sepsis in this study is presented in Fig. 1. The usage of indwelling devices is represented in Fig. 2. Ventilator-associated pneumonia accounted for 56% (48/85) of pneumonia. Catheter-associated urinary tract infections accounted for 72% (33/46) of urosepsis.
Relevant biological samples for microbiological evaluation were sent e.g. in patients suspected to have pneumonia (as per the CPIS scoring), either sputum or ETA or broncho-alveolar lavage (BAL) specimens. In patients with fever and pyuria, urine was sent for culture. Positive cultures were seen in 63.8% (111/174) of the patients. Gram-negative organisms accounted for 75.3% (110 out of 146 micro-organisms) of the infections. Among the patients, 13.8% (24 patients) had at least two organisms grown in culture. The pathogens and their antimicrobial resistance pattern are displayed in Table 2.
The most common organisms causing pneumonia were A. baumannii 24.7% (21/85), P. aeruginosa 24.7% (21/85) and K. pneumoniae 23.5% (20/85). Most of the MDR A. baumannii 89.5% (17/19) were sensitive to either cefaperazone sulbactam and/or netilmicin.
Urosepsis was predominantly caused by Enterobacteriaceae 36% (12/34). The commonest organism was Escherichia coli 26% (9/34). Two thirds of E. coli (6/9) was Extended spectrum beta-lactamase (ESBL) producers. All patients with K. pneumoniae and Enterobacter spp. causing urosepsis were ESBL producers. All Enterococcus spp. were sensitive to vancomycin, teicoplanin, ampicillin and linezolid. The details of the fungal isolates in the study are as follows. Twelve patients diagnosed as UTI (based on Centers for Disease Control and Prevention – National Healthcare Safety Network [CDC-NHSN] guidelines) had grown Candida spp. in urine culture. These patients had shown improvement with empirical antibiotics given for the UTI. Since they recovered without antifungals, the candidal isolate was considered as a coloniser. Four patients in the study had candidemia. Two patients with candidemia died despite treatment.
A. baumannii, P. aeruginosa, Staphylococcus aureus and Candida spp. accounted for the eight patients with catheter-related bloodstream infections (CRBSI). The two patients with coagulase negative Staphylococcal bacteraemia were methicillin sensitive. There were two deaths in CRBSI patients. Both had A. baumannii infection.
Organ system dysfunctions occurred as follows: renal dysfunction in 27% (47 patients), respiratory failure in 23% (41 patients) and metabolic acidosis in 6.9% (12 patients). Fifty seven patients (32.8%) developed septic shock.
Multidrug-resistant organisms were common in this study. Among S. aureus isolates, 43% (6 out of 14 isolates) were MRSA. Multidrug resistance was seen in 63.3% (19/30) of the A. baumannii infections. Eighteen were MDR and one was PDR. Multidrug resistance was seen in 64.4% (21/33) of the P. aeruginosa infections. Sixteen were MDR and five were PDR.
Among the E. coli, 66.7% were ESBL (6/9) producers. Among K. pneumonia, 70% were ESBL producers (19/27). Among Gram-negative isolates resistance to carbapenems was 35.2% (37/105) in this study. Carbapenem resistance was seen in 66.7% (20/30) A. baumannii isolates, 42.4% (14/33) P. aeruginosa isolates, 7.4% (2/27) K. pneumonia isolates and 20% (1/5) Enterobacter spp. isolate. Tigecycline resistance was seen only in 3.3% (1/30) A. baumannii isolates (PDR A. baumannii by definition).
In this study, mortality was 22% (39/174). Sepsis with A. baumannii had a mortality of 40% (12/30). However, as we had a high rate (15%) of discharge against medical advice (DAMA), we calculated non-recovery as a composite of death and DAMA. Sixty three percent (109/174) had recovered and the non-recovery rate was 37%. There was no difference in non-recovery rate between culture negative sepsis (38.1%) and culture positive sepsis (36.9%). The mean APACHE III scores of recovery and non-recovery groups were 38.6 (±18.7) and 81.9 (±26.3), respectively. High APACHE III scores were associated with non-recovery and this was statistically significant (P=0.006, mean difference=−35.3, 95% CI: −41.7–28.1). Patients with CKD had significantly higher non recovery rate than non-CKD patients (P=0.023, RR=1.731, 95% CI 1.0–2.9). Diabetes mellitus and CLD were also associated with high non recovery rate but this was not statistically significant.
Univariate logistic regression analysis showed male gender, rising APACHE score, underlying chronic kidney disease as comorbidity, sepsis leading to cardiovascular failure or renal failure or respiratory failure or metabolic acidosis or circulatory collapse to have statistically significant higher odds of mortality. Multivariate logistic regression analysis further endorses a significant independent association of male gender and sepsis causing cardiovascular failure with mortality. (Table 3)
Nosocomial infections are several times more common in ICU patients. Frequent use of invasive devices, underlying comorbidities, poor immunity and altered sensorium predispose these patients to nosocomial infections. Most patients admitted to ICUs have more than one underlying predisposing condition, which alters immunity and makes them vulnerable to acquire infections from the hospital setting. In this study, 26.4% of the patients had diabetes mellitus type II. Diabetes has been proven to produce dysfunction of both cellular and humoural innate immunity.12
Microbiological culture forms an integral part of treating any infection. The identification of the micro-organism proves invaluable in effective treatment. In the case of nosocomial infections, it may throw light on the environmental source of the infection, which could prevent further spread. This study had 63% culture positivity, which was comparable to large epidemiological studies (Sepsis Occurrence in Acutely Ill Patients [SOAP] study,13 60% and EPIC II study,14 69%). Gram-negative organisms are on the rise in corroboration with the previous study.14 In this study, Gram-negative organisms predominated (75.3%). Multi-drug resistance is of particular concern among Gram-negative organisms. The most frequently isolated Gram-negative organism was A. baumannii whereas in the EPIC II study it was P. aeruginosa.14
Interestingly, A. baumannii was involved in 18.8% of the infections, which was much more than the observations by the EPIC II14 (both 9%) and the relatively older SOAP study (3.6%).13 There is a wide variation in prevalence of A. baumannii infections. This has several implications. One is the prevalent local infection control measures and its efficacy. The knowledge of the pattern of infection in each region can guide empirical antibiotic therapy.
The commonest organisms causing pneumonia, in this study were P. aeruginosa (29%) and A. baumannii (29%), whereas in other studies they were P. aeruginosa and S. aureus.5,15 P. aeruginosa and A. baumannii also accounted for most of the cases of VAP in this study.
The commonest organisms causing urosepsis in the current study was Enterobacteriaecea in 36% cases, similar to the results of Rosenthal et al.2 S. aureus was the commonest organism causing nosocomial blood stream infections in this study. In a study on blood stream infections by Hugonnet et al.,16 coagulase negative Staphylococcus was the most common isolate.
Nowadays, the epidemiology of infections in ICUs has centred on the emergence of MDR organisms. In this study, 63.3% of the A. baumannii and 64.4% of the P. aeruginosa isolated were MDR, 70% of the K. pneumoniae were ESBL and 43% of the S. aureus were MRSA.
Carbapenem resistance is on the rise world-wide.17 In ‘Meropenem Yearly Susceptibility Test Information Collection Program’, meropenem resistance among isolates of K. pneumoniae increased from 0.6% in 2004 to 5.6% in 2008.18 In this study, the prevalence of Carbapenem-resistant K. pneumoniae is 7.4% (2 out of 27 isolates). Carbapenem-resistant Enterobacteriaceae (CRE) infections have limited treatment options. Recently, other carbapenemase subtypes including New Delhi metallo beta-lactamase have posed a great challenge. This emphasises the urgent need to step up measures preventing CRE transmission.
The mortality of patients with nosocomial sepsis in the current study was 22%, as against 14.6% in a study by Januel et al.,19 25% in the EPIC II study14 and 33% in a study by Naidu et al.20 In this study, 15% of patients (who were severely ill) opted to get discharged against medical advice, because of financial constraints and the guarded prognosis offered. The non-recovery rate of 37% was comparable to previous studies.
APACHE III score is widely used in predicting ICU mortality. Considering its advantages, we decided to use it. In this study, the mean APACHE III score of patients who recovered was 38.6 and mean score of those who expired was 74.2. High scores were significantly associated with non-recovery (P(0.006). Male gender and sepsis leading to cardiovascular failure were the independent predictors of mortality.
In the present study, Acinetobacter infections had an attributable mortality of 15.3%. Case-control studies, which studied in particular, the effect of A. baumannii bactremia on outcome of ICU patients conducted by Grupper et al. (N=104)21 in Israel and Playford et al. (N=120)22 in Australia, revealed an attributable mortality of 36.5 and 20%, respectively. In a commentary by Falagus and Rafailidis,23 it was suggested that attributable mortality due to A. baumannii should no longer be a controversial issue and measures should be directed to the development of new antibiotics against MDR and PDR A. baumannii and implementation of infection control.
Till date, APACHE III scoring system has not been used in the setting of nosocomial sepsis. A high APACHE II score was found to be associated with mortality by Agarwal et al.24 (N=201) (OR=1.1; 95% CI: 1.0–1.1).
In treating nosocomial infections, the clinician faces two conflicting issues: (1) an appropriate adequate antimicrobial therapy and (2) limiting excessive antimicrobial use to prevent the emergence of antimicrobial resistance. Some strategies adopted by institutions to decrease MDR include decreased duration of therapy, hospital formulary-based or targetted antimicrobial restrictions, use of antimicrobial protocols and guidelines, scheduled antimicrobial rotation or ‘cycling’. The most effective strategy would be to inculcate among all doctors, an urgent need to refrain from antibiotic use in unwarranted situations. It would be ideal, if the antibiotic use in institutions is monitored, to curb the excessive unwanted use of higher antibiotics. Another preventive measure is to de-escalate the broad spectrum empiric antibiotic to a limited spectrum antibiotic, once the culture and drug sensitivity pattern is available. Diligent use of these various approaches, in combination with frequent reviews of the nosocomial infection spectra, along with strict policies to prevent transmission of health-care associated infections will go a long way in decreasing the burden of nosocomial infections.
This study has highlighted several significant observations. It has prospectively discerned the pattern of nosocomial infections in medical ICUs contributing to sepsis, the associated device utilisation and their microbiological profile with 63% culture positive rates. Such single-centre studies, which evaluate the spectra of pathogens and their antimicrobial resistance patterns will provide a quality check on infection control. We have also investigated APACHE III score as a prognosticating tool, unlike previously studied APACHE II scores. Future scope of the study is to deal with the correlation of prior antibiotic usage and emergence of MDR organisms. Such an effort would help to curtail the MDR burden. Tactlessly, data on number of individuals excluded, ICU admissions and beds available in ICUs during the study period were not captured. This study included only medical ICUs and the pattern of infection cannot be generalised to surgical and cardiac ICUs.
Nosocomial infections remain a challenging problem in management of severely ill patients in ICUs. Multidrug resistance was detected in more than half the isolated strains, especially amongst the predominant microorganisms (A. baumannii, P. aeruginosa, S. aureus and K. pneumoniae). Carbapenem resistance is alarming. It could become a global threat unless measures are taken to curb it. Indwelling devices are channels for spread of resistant organisms in ICUs. Limiting their usage is primordial. Chronic kidney disease and high APACHE III scores were significantly associated with high non recovery. Male gender and sepsis leading to cardiovascular failure were the independent predictors of mortality. Simple preventive measures and periodic review of drug resistance patterns of nosocomial infections will be beneficial for hospitals with significant load of medically ill patients in ICUs.
Conception and design: KS. Clinical evaluation & management: KS, MP. Microbiological assessments: CM, VKE. MP and KS wrote the first draft of the manuscript. All authors contributed in subsequent revisions of the manuscript, read the final version and approved it for submission.
Conflicts of interest
An approval was obtained from the Kasturba Medical College and Kasturba Hospital, institutional ethics committee prior to commencement of the study.