Study design
This was a hospital-records-based cohort study of critically ill patients who attended the ED of a Tanzanian National Hospital for three years, from January 2019 to December 2021.
Setting
The study was conducted at the Muhimbili National Hospital (MNH) ED in Dar es Salaam, Tanzania. MNH has two campuses; for this study, only data for the Upanga campus has been analyzed. The ED receives acute patients from all hospitals in the country and directly from the community and admits them at MNH or transfers them to the specialized national hospitals of Jakaya Kikwete Cardiac Institute (JKCI) or the Muhimbili Orthopedic Institute (MOI). Obstetric emergencies are handled in the obstetric unit. Trauma cases seen at the ED are transferred to MOI or MNH for further management, depending on the nature of their injury. The three hospitals, MNH, JKCI, and MOI, are located within the same campus and have a total bed capacity of 1,982 beds.
At the ED, patients are received, registered, and triaged based on their vital signs and illness severity and then managed accordingly. The goal is for stable patients to leave the ED within two hours and up-to seven hours for severely sick patients. In addition, resuscitation rooms and a mini-ICU at the ED receive severely sick patients that require emergency or critical care.
Dar es Salaam is the most populous city in Tanzania. During the COVID-19 pandemic, over 40,000 confirmed cases have been reported in Tanzania, with Dar es Salaam having the highest number of cases (16, 17).
Data Source
Data were extracted from the ED electronic information system. Clinicians and triage nurses at the ED enter information about the patients into the electronic system as they are providing care. The extracted data included the time the patient arrived at and was discharged from the ED, demographic information, the presenting complaint, diagnoses at the ED (the first listed diagnosis was used as the main diagnosis), interventions provided, laboratory results, ED outcome status, and the corresponding admission ward at MNH or transfer facility.
Study Population And Eligibility Criteria
All critically ill patients who presented to the ED within three years, from 1st January 2019 to 31st December 2021, were included in the study. Critical illness was primarily defined as present in any patient who either had a severe derangement of one or more vital signs as measured at triage or received a critical care intervention at the ED. The critical interventions included oxygen therapy, intubation, suction, tracheostomy, cardiopulmonary resuscitation, adrenaline/epinephrine, atropine, and chest tube insertion. The vital signs included systolic blood pressure (SBP), pulse rate (HR), respiratory rate (RR), blood oxygen saturation (SpO2%), and the Glasgow coma scale (GCS). The age-dependent cut-offs for severe derangement of vital signs were based on previous literature (Table 1) (1, 18–21).
Table 1
Cut-offs for severely deranged vital signs by age
Vital Sign | Age | Severely Deranged Cut-off |
Respiratory rate per minute | < 1 month | < 20 or > 80 |
1 month-<1 year | < 15 or > 60 |
1 year- <5 years | < 10 or > 50 |
5 years- 12 years | < 8 or > 40 |
> 12 years | < 8 or > 30 |
Saturation (%) | All | < 90% |
Pulse rate per minute | < 1 month | < 80 or > 200 |
1 month- <1 year | < 80 or > 180 |
1 year- <5 years | < 70 or > 170 |
5 years- 12 years | < 60 or > 150 |
> 12 years | < 40 or > 130 |
Glasgow Coma Scale | All | ≤ 8/15 |
Systolic Blood Pressure (mmHg) | < 3 months | < 50 |
3 months- <1 year | < 70 |
1 year- <4 years | < 75 |
4 years- <12 years | < 80 |
≥ 12 years | < 90 |
Critical illness is notoriously difficult to define (2, 4), and we used two additional, alternative definitions. First, we used a broader definition for critical illness, including patients who received care in the resuscitation room or the mini-ICU in the ED in addition to those in the primary definition. The second alternative definition was more restrictive and included only patients with one or more severely deranged vital signs. Records of patients who lacked data for their age or were dead at arrival were excluded from the analysis. Psychiatric patients with non-somatic complaints were excluded, as were those in the electronic information system but had received care in the ED of MNH's additional campus (Fig. 1).
Data Extraction, Management, And Analysis
Anonymized data were extracted from the electronic information system in Excel format and were imported into STATA 17 (StataCorp, Texas, USA). Eligible patients were identified, and duplicates were removed. The data were cleaned and checked for their validity and completeness, and outliers were identified and cleaned or removed if not plausible. Descriptive statistics using frequencies, percentages, median and interquartile range (IQR) were used to summarize the patients’ demographic and clinical characteristics.
Missing vital signs were imputed as not being severely deranged. A sensitivity analysis where missing vital signs were imputed as severely deranged was done. Imputation was not done for the Glasgow Coma Scale for all patients and not for systolic blood pressure for children under 15 years, as there was significant non-random missing data.
The burden of critical illness, defined by the primary and alternative definitions and the sensitivity analysis, was calculated and presented. Using the primary definition, the total number and proportion of critically ill patients per year and their mean daily/monthly numbers were calculated. The proportions were further disaggregated by the demographic of the patients. The proportion of critical illness by year was done, and the chi-squared test was used to compare the observed proportion of critical illness in 2020 and 2021 to that in the first year, 2019. The trend of the monthly proportion of patients who were critically ill was presented with line graphs. Diagnoses of critically ill patients were classified according to the first listed diagnosis and grouped using the organ system affected. Outcomes at the ED were presented using pie charts, and admission wards were categorized and presented using bar graphs.