Treatment Outcome of Hyperglycemic Emergency and Predictors in Ethiopia

Hyperglycemic Emergency (HE) denotes critical cases of decompensated diabetes mellitus (DM). Diabetic ketoacidosis (DKA) and hyperosmolar hyperglycemic state (HHS) are the extreme cases of HE. This study aims to assess the treatment outcome of HE and predictors. Method: Four-year medical records of DM patients admitted for HE at Hiwot Fana Specialized University Hospital (HFSUH) were reviewed retrospectively. The data abstraction tool was adapted from the preceding studies. Data were entered into and cleaned by Epi-Info TM 7 software. The statistical analysis was executed using the statistical package for social sciences software (SPSS) version 24. Chi-square test and student’s t-test were done to compare categorical and continuous variables. Logistic regression with the level of α set at 0.05 and AOR of 95% CI was done to determine the predictors. Statistical signicance was established at AOR ≠ 1 within a 95% CI and P-value < 0.05. The model was veried using the Hosmer-Lemeshow goodness of t test (P = 0.392).


Conclusion
Mortality and duration of hospital stay stand high among DM patients admitted for HE at HFSUH. Infection, comorbidity, and lower admission GCS ≤ 8 are the independent predictors of HE mortality.
Operative strategies targeting the predictors ought to be devised to control the morbidity and mortality among DM patients hospitalized for HE.

Background
Hyperglycemic Emergency (HE) denotes critical cases of decompensated diabetes mellitus (DM). Diabetic ketoacidosis (DKA) and hyperosmolar hyperglycemic state (HHS) are the extreme cases of HE (1). DKA usually occurs in type 1 DM; HHS often rises in type 2 DM. However, each case can appear in any of the two DM types (2). Patient's admission features comprise a varying degree of hyperglycemia, dehydration, electrolyte imbalance, acidosis, altered mentation. The treatment entails uid and electrolyte and insulin therapy alongside vigilant assessment and management of the precipitating conditions (3,4).
HE remains one of the most important causes of global morbidity and mortality among individuals living with DM. Studies from various countries had accentuated this connotation (5)(6)(7)(8). The fatality rate for HHS is unsurpassed in contrast to DKA (8)(9)(10)(11)(12). The signi cant predictors of mortality in adult DM patients hospitalized for HE include infection, comorbidities, coma, age extremes, cardiac arrhythmias (13).
Immensely augmented by the detrimental meddling of the extensive socioeconomic glitches in the effective deterrence and management, the HE morbidity and mortality reside even more pronounced in Africa. A study by Ogbera et al in 2009 reported a collective DKA and HHS mortality rate of 20% from an urban hospital in Nigeria (14). Later in 2017, Olugbemide et al reported a pooled DKA and HHS mortality rate of 34% from a specialist teaching hospital in Nigeria. (15). Correspondingly, Mbugua et al described a 30% mortality rate of DKA from Kenyatta National Hospital (KNH), Nairobi, Kenya (16).
HE stands the domineering why and wherefore of hospitalizations among DM patients in Ethiopia (17,18 above. Patients whose medical records lacked adequate information were excluded from this study. The data abstraction tool was adapted from the preceding studies. Sociodemographic, clinical, biochemical, hospital stay duration, and mortality data were extracted exclusively from the patient's respective medical records. Data were entered into and cleaned by Epi-Info TM 7 software. The statistical analysis was executed using the statistical package for social sciences software (SPSS) version 24. Chi-square test and student's t-test were done to compare categorical and continuous variables, respectively. Logistic regression with the level of α set at 0.05 and AOR of 95% CI was tted to determine the predictors. Variables with p-values ≤ 0.2 on a bivariable binary logistic regression analysis were considered in a multivariable binary logistic regression model. Statistical signi cance was established at AOR ≠1 within a 95% CI and P-value < 0.05.
The model was veri ed using the Hosmer-Lemeshow goodness of t test (P = 0.392).

Operational de nitions
DKA was denoted to an admission blood glucose level >250 mg/dL with signi cant ketonuria (urine dipstick ketone level ≥ +2).
HHS was denoted to an admission blood glucose level >600 mg/dL without signi cant ketonuria (urine dipstick test result: no or +1 urine ketone).
The primary study outcome mortality was referred to in-hospital mortality of DM patients diagnosed with HE and started on the HE management protocol.

Results
Patients fulfilled the inclusion criteria and included in this study were 321 of the entire 613 DM patients admitted and treated for HE at HFSUH within the four years.

Sociodemographic characteristics
The study participants were comprised of 192 (59.8%) females and 129 (40.2%) males. The female to male sex ratio was 1.5:1. The mean admission age of the participants was 43.3 ± 18.4 years. Their age ranged from 15 to 85 years. Patients admitted from the urban areas were 163 (50.8%) and the remaining 158 (49.2%) patients were from the rural areas.

Clinical admission features
The clinical admission features of the study participants considered in this study were the vital signs measured at admission for HE. Compared with DKA, both admission systolic and diastolic blood pressure records stood significantly higher while admission Glasgow Coma Scale (GCS) score remained significantly lower among DM patients hospitalized with HHS. The clinical admission features of the study participants as DKA compared to HHS are displayed in Table 1.    Table 3 illustrates the predictors of mortality among DM patients hospitalized for HE.

Discussion
The ndings of this study depicted that HHS was the major diagnosis of DM patients hospitalized for HE. The majority of the participants were known DM patients on treatment. Comorbidities were common. Infection was the foremost precipitant of HE followed by noncompliance followed by newly diagnosed DM. Mortality and duration of hospital stay were high. Infection, comorbidity and lower admission GCS score ≤ 8 were the independent predictors of mortality.
HHS was the major HE case in our study. This nding goes in line with studies from Nigeria and Taiwan (15,(23)(24)(25)(26) but stands against a study reported that DKA was the commonest HE (22). The higher prevalence of HHS in our study might be explained by the fact that majority of our study participants were T2DM patients and HHS often raids patients with uncontrolled T2DM (27). In addition, the higher percentage of known DM patients in our study might also have lifted the HHS prevalence as it was reported that HHS is common in patients with a prior history of DM (25). This implies that HHS is on the rise and demands great attention.
The rate of comorbidity in our study was 32.4%. This is higher than the 22.1% rate reported by a study from Jimma (22). The relatively higher proportion of known DM patients in our study might justify this higher prevalence of comorbidity. This is because comorbidities are more likely to be diagnosed and controlled in known DM patients on follow up than in newly diagnosed DM patients at admission with HE as they may be masked by the resultant HE complications. HTN for instance can be obscured by the hypotension from the HE. The substandard DM care has obfuscated the effort to control the morbidity and mortality among DM patients in Ethiopia (28). Therefore, works need to be done to standardize the DM care.
The 48% rate of infection documented in our study was higher than the rates, ranging from 22.5-32%, reported from South Africa, Saudi Arabia and Colombia (5,29,30). The reason for the upstretched infection rate documented among our study participants might be the incapacitating poverty restraining the healthcare systems in the Sub-Saharan Africa (31). Comprehensive health promotion and infection prevention strategies need to be implemented to control infection and its lethal complications.
The noncompliance rate in our study was 38.6%. This is higher than the 32.3% and 34% rate reported respectively from Jimma and Kenya (16,22). The comparatively higher percentage of known DM patients on treatment in our study might explain this disparity. The nancial hitches and unjust traditional tenets have been linked with missed insulin doses and antidiabetic medication discontinuation among Tropical and Sub-Saharan African DM patients (31,32). Hence, it would be matter-of-factly to devise strategies to ensure accessible standard DM care in the region and abate the imminent dreadful consequences of noncompliance like HE.
The median length of hospital stay in our study was 7 days. Even though this is shorter than the ndings of studies from Nigeria and Colombia (5,23,24), it still is longer than the reports from Jimma and South Africa (22,33). The inconsistency may be due to the difference among the settings in the effective and e cient management of DM patients presenting with HE. As DFU is associated with long hospital stay (23,33), the relatively higher proportion of patients with DFU in our study may further justify the lengthier hospital stay duration.
The overall HE mortality recorded in the present study was 16.5%. Despite being lower than the mortality rates reported from Kenya (16) and Nigeria (14,15), it stands higher than a series of mortality rates, 2.3%-9.8%, reported from Jimma (22), South Africa (33), Thailand (6), Australia (7), and Colombia (5) Availability of data and materials All generated data are comprised.

Competing interests
No competing interests.

Funding
Funding was received for this research work from the University of Gondar.
Authors' contributions PA conceived and designed the study, developed the data collection tool, acquired the fund, supervised the data collection, analyzed the data and wrote the manuscript. GB, AA and BM contributed to the study design, development of the data collection tool, statistical analysis and manuscript drafting. We all the authors red and approved the nal manuscript.