Burden and bacteriological profile of sepsis among adult medical emergencies presenting to a national referral hospital in Kampala, Uganda

Study design and setting

A prospective observational cohort study was conducted from December 2018 to July 2019 among patients presenting to the medical emergency ward of Kiruddu national referral hospital during the study period. Kiruddu Hospital, located on Buziga Hill, Makindye Division in Kampala is approximately 10 kilometers southeast of the central business district of Kampala. The medical emergency ward has a 25-bed capacity and receives approximately 976 patients per month. The unit acts as a national referral emergency unit caring for adult non-surgical patients. It is run by different cadres of doctors and nurses and has access to the main hospital laboratory and radiology departments for investigations. Patients received in the unit are assessed, stabilized, and then sent to one of the various sub-specialty medical wards based on the presumptive diagnosis. During the study period, the hospital had no facilities for carrying out culture and antimicrobial susceptibility testing and no formal protocol for sepsis identification or management was in place at the unit.

We assessed patients who presented to the unit during the study period for suspected sepsis. Sepsis was defined as the presence of a suspected infection plus two or more clinical proxies for organ dysfunction including oxygen saturation (SPO2) <90%, estimated urine output in the preceding 24 hours <400mls, jaundice, petechiae &/or ecchymoses, Glasgow Coma Scale (GCS) less than 15, systolic blood pressure (SBP)<90mmhg and absent bowel sounds. All adults aged 18 years and above who met the above study definition for sepsis and consented to participate were enrolled after providing written informed consent. Consent was obtained from the next of kin of very ill patients.

Ethics approval and consent to participate

We obtained approval from the Makerere University School of Medicine Research and Ethics Committee (SOMREC). We also obtained verbal consent before screening all patients and written informed consent before enrollment into the study.

Data collection and study procedures

The sample size was calculated at 250 and the convenience sampling method was used to assess all patients who presented to the emergency ward. After an initial assessment by the primary healthcare team, we assessed patients for sepsis using the study definition based on an adaptation of the sepsis management recommendations for low-income settings published in 2012 (8). A study questionnaire was administered to either the participant or attendant for very ill patients to collect data on the patient’s history including the sociodemographic and medical history and history of antimicrobial use in the past 30 days. A physical examination was performed at enrolment to identify the possible site(s) of infection, vital parameters to determine the qSOFA score, functional status using the Karnofsky performance scale (KPS), and organ dysfunction. Venous blood was drawn for aerobic blood culture, complete blood count, liver and renal function tests. Other tests included a thick blood smear for malaria or malaria rapid diagnostic test and HIV test (9). Treatment including antimicrobials and fluids was initiated by the attending clinicians and documented by the study team.

Blood culture and antimicrobial susceptibility testing

Eight milliliters of blood were drawn aseptically and inoculated into an aerobic blood culture bottle (BD BACTEC). The blood samples were transported at room temperature to MBN laboratory within 8 hours of collection, where they were inserted into a blood culture machine (BD BACTEC FX40) immediately to last a protocol length of 5 days. Any growth was detected automatically by the machine and the following procedures were performed; gram-staining and sub-culture on agar media, identification of colonies that grew on the agar, and antibiotic susceptibility testing using the Kirby Bauer disk diffusion method. The susceptibility testing and interpretation were based on the Clinical Laboratory standard institute (CLSI, 2018). All test results were availed to the primary team managing the patients as soon as they were ready. The MBN laboratory participates in an external quality assurance program under the Oneworld Accuracy system.

Data processing and analysis

All complete data were entered into the computer using a data entry template developed using Epidata version 3.1 software packages with in-built quality control checks. The final data was cleaned, frozen, backed up, and exported to STATA version 14.0 for analysis. Sepsis prevalence was calculated as the proportion of patients with sepsis out of those assessed during the study period. The outcomes were described as proportions for categorical data, means with standard deviations for normally distributed numerical data, and median with interquartile ranges for skewed data. The proportion of aerobic blood culture samples with microbial growth was calculated as a percentage and the isolates were described. The antimicrobial susceptibility patterns (including multidrug resistance) of organisms isolated from blood culture were described.

In-hospital outcomes

Of the 243 patients with sepsis, 143 (58.9%) died after an average length of hospital stay of 4.9 days. The median duration of hospital stay was longer among those who survived than those who died (7(5,13) vs 3(1,6)) days. None of the patients got transferred to an intensive care unit (ICU) during the study period. Patients with a Glasgow coma scale (GCS) less than 15 at admission were more likely to die in the hospital compared to those with a GCS of 15 (p=0.001) (Table 2).

Antimicrobial susceptibility patterns

The aerobic blood culture positivity rate was 18.9% (46/243) of all blood culture samples. The most commonly isolated organism was Staphylococcus aureus (S.aureus) with 31 isolates.

Of the S.aureus isolates, 20, (64.5%) were Methicillin-resistant (MRSA), 31, 100% were resistant to penicillin and 100% were sensitive to vancomycin. Of the 9 Enterobacteriaceae isolates, 8 were resistant to amoxicillin-clavulanic acid, all were resistant to ampicillin and more than half (5) were resistant to ceftriaxone while all were sensitive to amikacin and Imipenem as detailed in table 4 below.

Of the Enterobacteriaceae, two Isolates were producers of Extended Spectrum Beta-lactamases (ESBL).

In-hospital outcomes

We found a mortality rate of 57.8% within an average duration of hospital stay of 4.9 days. The median duration of hospital stay was longer among those who survived than those who died (7(5,13) vs 3(1,6)), meaning patients died early on during hospitalization. This mortality rate is higher than that found by previous studies in Uganda. Jacob and colleagues found an in-hospital mortality rate of 23.7% and an overall mortality rate of 43% up to 30days post-discharge (Jacob et al., 2009). Another study from western Uganda reported in-hospital mortality of 30% (13). The differences in mortality rates could be explained by the severity of illness of patients in this study where over 42% had a Karnofsky score of ≤ 20%. None of the patients in this study was admitted to the ICU within 30 days of in-hospital follow-up. This lack of intensive care management plus the presence of drug-resistant bacteria as evidenced by 8 out of the 9 Enterobacteriaceae spp isolates resistant to ceftriaxone, most likely contributed to the high mortality rates in this study. Ceftriaxone is the most common antibiotic prescribed empirically in the medical emergency ward. In addition, 64.5% of the S.aureus isolates were MRSA yet vancomycin was not readily available. Studies from other settings found similarly high mortality rates; for example, one recent study from India reported in-hospital mortality of 63.6% (14). Overall, this indicates that sepsis contributes significantly to mortality in hospitals worldwide.

Aerobic blood culture and antimicrobial susceptibility patterns

The aerobic blood culture growth rate was 18.9% in this study with Staphylococcus aureus (S. aureus) as the most frequently isolated pathogen (31 of the 46 isolates). More than half of the S. aureus isolated were MRSA. The relatively low culture positivity rate may have been because more than half (53.5%) of the patients had been treated with antimicrobials within the 30 days before admission. Irrational use of antimicrobials is rampant in many low-income settings including in Uganda. A study by Mukonzo et al (2013) found that 41% of all antibiotic sales at community pharmacies were over the counter and not accompanied by a prescription (15). Also, one blood culture set was obtained yet two or more might have increased the yield. However, other studies in Uganda where aerobic blood cultures were performed reported similar rates of culture positivity. For example, a recent study summarizing antimicrobial susceptibility patterns reported culture growth in 14% of the samples. This study also reported S. aureus as the most frequently identified growth and 32% of these were methicillin-resistant (16). An earlier study reported a similar rate of culture-positive sepsis of 20% and more than half of the patients (57.7%) had used antimicrobials before admission., but the commonest isolate was non-Typhoidal Salmonella species(17). Low blood culture yields have also been reported in many studies outside the Ugandan setting. The high rate of drug-resistant pathogens identified provides evidence for the growing problem of antimicrobial resistance in Uganda. This is likely contributed to by the inadequate anti-microbial stewardship in the country.