Study design, setting, and period
The retrospective cross-sectional study was conducted from July 15 to August 14, 2021 at Hiwot Fana Comprehensive Specialized Hospital (HFCSH), Eastern Ethiopia. HFCSH is a specialized hospital in Eastern Ethiopia, which is found in Harar town, 526 kilometers from Addis Ababa, the capital city of Ethiopia. It is serving as a referral hospital for the entire eastern part of the country including Eastern Oromia, Dire Dawa city administration, Harari, and Somali regional states. There are different wards and clinics within the hospital that provide services for the community such as medical ward, pediatric ward, surgery ward, gynecologic and obstetrics ward, oncology ward, antenatal clinic, dental clinic, a tuberculosis clinic, ante-retroviral therapy clinic, dermatology clinic, and ophthalmologic clinic. The medical ward has 22 beds.
Population
All adult patients with a diagnosis of AHF treated at HFCSH were source population. All adult patients with a diagnosis of AHF treated at HFCSH in the past five years (from June 01, 2016 to May 31, 2021) who fulfilled the inclusion criteria were the study population.
Eligibility criteria
All adult patients whose was age less than 18 years with a confirmed diagnosis of AHF and treated during the data retrieval period at the medical ward of HFCSH were included in this study. However, all patients with unknown outcome (referred or disappeared) and lost and/or incomplete patients’ medical records were excluded from the study.
Sample size determination
For the first two objectives sample size of the study was calculated using single population proportion formula by considering the following assumptions: prevalence of in-hospital mortality (P) = 17.2% taken from another similar study (20), confidence level (CI) = 95% and corresponding Z score of 1.96, margin of error (W) = 5%. Thus sample size was, n = (Zα/2)2 p (1-p) /w2 = (1.96)2*0.172*0.828/ (0.05)2 = 218.8. For the objective of factors associated with in-hospital mortality sample size was calculated using Epi-Info version 7.2.4.0 (Table 1).
Table 1: Sample size calculation for factors associated with in-hospital mortality of acute heart failure patients using Epi-Info software (version 7.2.4.0).
Determinants of treatment outcome
|
Factors
|
Assumptions
|
% Exposed
|
% unexposed
|
Reference
|
Sample size (N)
|
Smoking
|
CI:95% Power:80% Ratio:1:1
|
44.4
|
15.6
|
(20)
|
80
|
Diabetes mellitus
|
CI:95% Power:80% Ratio:1:1
|
50.0
|
15.6
|
(20)
|
60
|
Pulmonary hypertension
|
CI:95% Power:80% Ratio:1:1
|
15.1
|
41.2
|
(20)
|
94
|
Chronic kidney disease
|
CI:95% Power:80% Ratio:1:1
|
39.59
|
22.44
|
(25)
|
230
|
Heart rate > 100beats/min at presentation
|
CI:95% Power:80% Ratio:1:1
|
34.30
|
47.9
|
(25)
|
412
|
CI = Confidence interval
Therefore, by adding 10% for contingency on the largest sample size, the calculated sample size was found to be 454 adult patients with AHF.
Sampling procedures and sampling techniques
The 907 adult patients were treated at HFCSH for AHF during the last five years. That is 121, 144, 201, 199, 242 patients from June 01, 2016, to May 31, 2021, each year respectively. For this study, record numbers of those patients’ medical records were obtained from the medical registration book of the Medical ward of HFCSH. Finally, 454 patients’ medical records included in the study were selected using proportionally allocated simple random sampling technique (Figure 1).
Data collection method
Data were collected from patients’ medical records using an English version of structured and pretested data abstraction format prepared by reviewing different pieces of literature on similar topics, which consists of socio-demographic information, vital signs at presentation, clinical presentations, comorbidities, laboratory and radiologic findings (Additional file 1).
Record numbers of the individual patients’ medical cards were obtained from medical registration book of Medical ward, and the numbers of those patients’ medical cards to be included in the study were randomly selected using lottery method. Then using randomly selected numbers, patients’ medical cards were drawn from the hospital card room. In-hospital mortality was confirmed by principal investigator from discharge summary. Finally, the data were collected from those medical cards by two clinical pharmacists and one clinical nurse.
Operational definitions
Acute heart failure:- sign and symptoms of new-onset of heart failure or decompensation or worsening of chronic stable heart failure (ADHF) (2, 28)
In-hospital mortality: - AHF patient died in the hospital after he/she was admitted to the medical ward which confirmed by physician’s death summary (29).
Elevated blood pressure: - systemic blood pressure of greater than 120/80 mmHg.
Data quality control
Necessary training was given for data collectors on the contents of the data collecting tool, data collection procedures and consent was obtained from head of hospital before data collection. A pretest was conducted on randomly selected 5% of the sample size (twenty-three AHF patient’s medical records) before the actual data collection. Then, adjustments were made on the tool for final data collection. The supervisors carried close supervision out daily during the data collection time. The principal investigator checked data for completeness, clarity, consistency, and accuracy.
Data analysis procedures
The collected data were entered to Epi-data (version 3.1), transferred to and analyzed using statistical package for social science (SPSS) version 21.0®. Quantitative variables were reported as a median and inter-quartile range. Categorical variables were presented using percentages and frequency. Bivariate and multivariate logistic regression was used to analyze factors associated with treatment outcome. Each variable with no missing value were checked for association against the outcome variable in the binary model. Then, Variables with P< 0.25 in the bivariate model analyses were entered into the multivariate model.
The Binary Logistic Regression model fitness was checked by Hosmer-Lemeshow statistic. Multicollinearity test was carried out to see the correlation between independent variables using variance inflation factor (VIF). Crude odds ratio (COR) and adjusted odds ratio (AOR) were calculated with the 95% confidence interval to measure the strength of the association between the outcome and independent variables. The variable with a p-value less than 0.05 in the multivariable analysis was considered significantly associated with in-hospital mortality.