Survival and predictors of mortality among acute leukemia patients on follow‐up in Tikur Anbessa Specialized Hospital, Addis Ababa, Ethiopia: A 5‐year retrospective cohort study

Abstract Background Although Ethiopia has more than 78% of leukemia cases and a significant burden of the disease, the survival of leukemia patients in the country is poorly recognized. The purpose of this study was to assess the survival and predictors of acute leukemia patients. Methods A 5‐year retrospective cohort study was conducted including all acute Leukemia patients who visited Tikur Anbessa Specialized Hospital between January 2015 and December 2019. Data were retrieved from patient's medical records between March and April 2020. Using SPSS version 25, the Kaplan–Meier curve and Cox regression models were employed to analyze the data. Results A total of 119 patients with acute leukemia were retrospectively evaluated for 60 months, having 196 person‐years of risk. About 46 deaths (38.7%) were recorded over the follow‐up period, giving a mortality incidence rate of 23.5 (95% CL:18–52) per 100 person‐years. The median survival time was 35 months (95% CI, 28.3–41.7). At 60 months of follow‐up, the predicted overall survival rate after diagnosis for acute leukemia was 21%. The adjusted hazard ratio for acute leukemia subtypes (aHR:4.9, 95% CI:2.3–10.4), history of relapse (aHR:3.9, 95% CI:1.0–7.9), participant age (aHR:1.25, 95% CI:1–1.75), hepatomegaly (aHR:2.7, 95% CI:1.36–5.36), and splenomegaly (aHR:2.29, 95% CI:1.2–4.4). Conclusion The 5‐year overall survival rate was found to be 21%. The finding was remarkably lower than other published reports. Survival among acute leukemia patients was significantly associated with older age, history of relapse, hepatomegaly, splenomegaly, as well as certain subtypes. Therefore, improving early detection and initiation of treatment for all acute leukemia patients is necessary in order to improve patient's survival status.


| BACKGROUND
The global burden of cancer is estimated to have risen to 19.3 million new cases, and 10 million deaths with 50 million of 5-year prevalence in 2020. 1 The American Cancer Society (ASCO) estimated that 13 800 cases of acute myelogenous leukemia (AML) and 6000 cases of acute lymphoblastic leukemia (ALL) were diagnosed in the United States. 2 People with lower socioeconomic status are more likely to be exposed to cancer risk factors like smoking, obesity, and obstacles to effective cancer prevention, early diagnosis, and treatment.Lifestyle of the patients, genetic factors, therapeutic effects, and environmental factors have also been related to an elevated risk of cancer. 3cluding non-melanoma skin cancer, Leukemia was among the top 5 most common cancers, with 4361 new cases reported in Ethiopia in 2020.Leukemia incidence and death rates are predicted to be 46 and 39, respectively in Ethiopia in 2020. 4According to the cell types and the clinical course history Leukemia is classified into myeloid and lymphoid leukemia. 5Accordingly, ALL, chronic lymphocytic leukemia (CLL), AML and chronic myeloid leukemia (CML) are the four major types of leukemia (CML).Acute leukemia is very aggressive disorder that causes poor production of bone marrow cells that in turn leads to immunosuppression.Since acute leukemia has a very high complete remission rate (60%-67%) it requires rapid medical attention. 6cording to studies, improvements in acute leukemia diagnosis, treatment choices, and patient risk classification instruments have enhanced survival rates. 7Data from surveillance, epidemiology and end results (SEER) showed that the 5-year survival rate for patients with ALL increased from 31.6% in 1997-2002 to 39.0% in 2003-2008.However, compared to non-Hispanic whites age-adjusted 5-year relative survival rates for African-Americans and Hispanics remained lower. 6Cancer stage, sex, age, ethnicity, and socioeconomic position were also found to be determining factors of cancer survival.
Black male adolescents and young adults with severe illness stages were shown to have worse survival rates. 8Although leukemia is becoming more prevalent and was a very serious problem in Ethiopia, it is not well studied about adult acute leukemia (AL) patients' survival rates.Thus, this study was designed to assess mortality risk factors among patients above 18 years with acute leukemia who were admitted to Tikur Anbessa Specialized Hospital (TASH) from January 2015 to December 2019.

| Study area, design, and period
A health institutional based retrospective cohort study was conducted from March to April, 2020.
Between January 2015 and December 2019, 203 medical records of adult patients with acute leukemia who were hospitalized to the hematology unit at Tikur Anbessa Specialized Hospital (TASH) were examined.Tikur Anbessa Specialized Hospital is the only public cancer care referral center in the country operating with a total of 700 beds.
The hospital registry indicates that 370 000 to 400 000 cancer cases were monitored annually.

| Definitions
The event in this study was death.
Acute leukemia: a patient with greater than or equal to 25% blasts in the BMA.
The time variable was time from diagnosis to the occurrence of death.
Participants with failure time or unknown outcome due to drops outs were considered as censored.

| Populations
Between January 1, 2015 and December 31, 2019, any adult acute leukemia patients who were older than 18 and received care and therapy at Tikur Anbessa Specialized Hospital were enrolled.

| Eligibility
All patient records indicating an acute leukemia diagnosis with complete data records of (Patient characteristics: age, sex, marital status, residence, ethnicity, occupation, and religion were included. Acute leukemia, sub-types such as AML, ALL, ALL-B and immunophenotype-T, hemoglobin level, white blood cells count phase of chemotherapy medical payment, remission, relapse place of relapse number of relapses, co morbidity co morbidity types splenomegaly, and hepatomegaly) were eligible, and those patient records lacking full information of the above variables were excluded.

| Sampling procedure
All patient records with acute leukemia diagnosis during the course of the investigation were included sequentially using a convenient sampling method.

| Data collection tool
A structured data collection checklist was prepared to extract data from medical records and cancer registry books.In order to ensure the checklist's consistency to the study, it was pretested among 5% of the sample's total participants in a private clinic.
It includes an information about patient's sociodemographic variables including age, sex, marital status, ethnicity, country of residence, religion, and profession.It also included an information on the patient's acute leukemia including the date of diagnosis, length of hospitalization, type and phase of treatment, type of leukemia, a history of relapse, the location and number of relapses, co morbidity, hemoglobin, blood cell counts, and survey endpoints.In order to collect data, two experienced nurses with bachelor's degree were recruited and a daylong training was given.Variables that show statistical significance in bivariate logistic regression analysis at (p < .25)were included into the multivariate Cox regression models to identify independent predictors for the patients' survival status and to estimate adjusted hazard ratios.Both crude and adjusted hazards ratio with 95% confidence interval were reported for variables that were statistically significant.Before the Cox proportional hazard regression model was applied, its multi-collinearity was tested using the variance inflation factor (VIF) and pair-wise correlation.about half of the participants were rural inhabitants.Nearly half of them were married and 44 (37.0%) were farmers by occupation (Table 1).following therapy was 21%.At 12, 24, 36,48, and 60 months, the estimated cumulative survival probability was 89.1%, 67.7%, 44.4%, 28.5%, and 21%, respectively, (Figure 1).

| Hematologic and pathologic characteristics
As the follow-up period lengthens, the likelihood of survival falls, which is a typical feature of survival data analysis.The KM survival curve makes it very evident that the period of the greatest mortality occurred between 12 and 36 months.As the follow-up period lengthens, the likelihood of survival drops which is a typical feature of survival data analysis.The KM survival curve shows it very evident that the period of the greatest mortality occurred between 12 and 36 months.

| Survival experience among different groups of leukemia patients in log-rank test
The log-rank test was performed to determine if there were any statistically significant variations in survival times among the various levels of the categorical factors taken into consideration in the study.

| DISCUSSION
Recent progress in understanding survival and predictors of mortality has arouse growing interest in the epidemiology of hematologic malignancies.In this study, we evaluated survival patterns and predictors of outcomes in acute leukemia patients in Ethiopia.In the present retrospective cohort study, the 1-, 2-, 3-, 4-, and 5-year survival rate of acute leukemia was respectively 89.1%, 67.7%, 44.4%, 28.5%, and 21%, respectively.The results indicated that Advanced age, history of relapse, hepatomegaly, splenomegaly, and leukemia subtype had a significant effect on survival rate.Overall acute leukemia survival rate was high in developed world, while being very low in developing countries indicating that leukemia treatment centers were relatively young in these nations. 9 the other hand, there are not many local studies available for the comparison of acute leukemia survival rates.In this study, the survival rate for a patient with acute leukemia was 21%.This was lower than the rates reported for other developing countries, which was 86% for India 10 and 52% for Nigeria. 11The variation might be due to the fact that there is only one leukemia treatment facility in the country, in which may lack the patients in the present study location from receiving early treatment.In this study, AML is the deadliest form of acute leukemia, and its survival is substantially lower than that of ALL subtypes.The overall survival rate for AML was (19%), which is nearly similar to the study in the USA that showed 18.2%, however, the present study for ALL found 27%, which was lower than USA 44.0%. 12udies conducted in German showed 5-year survival probabilities for AML was 44.3% 13 and for ALL was 43.4% 14 subtype which is higher than current study for both types.
Overall 5-year survival was estimated at 43.4% for Germany which is higher than the current finding, this might be due to scarcity  of Leukemia management centers, lack of skilled human power in the cancer center or it may be due to SES, genetic and geographic variations in study cohort. 15In contrast to the Swedish study, which found no difference in survival across subtypes, the current study found a substantial difference in survival between the AML and ALL subtypes, which are 19.4% and 27%, respectively, 16 this could be there is poor treatment center for leukemia in Ethiopia specially AML subtypes, that may result very poor survival of AML in Ethiopia or it may be due to socio-economic, genetic or geographical variation.Our 5-year acute leukemia survival rate was lower than the survival rate reported on patients of USA in both ALL and AML subtypes. 17The current study found a higher value in induction death (13%) than the US found 5.9% ± 1.9% 18 induction deaths, which is suggestive of a different regimen or low cytotoxicity management.Furthermore, the current study's complete remission rate was very low (26%), compared to the 57% of the American study. 19 comparison to the Canadian study, which covers patients with ALL and AML, both types of acute leukemia had lower overall survival rates (64% and 62%), respectively. 7This is likely because there is inadequate or no medical care for an advanced tumor in the existing setup.
In the current study, leukemia subtype, history of relapse, hepatomegaly, splenomegaly, and age were the main predictors of acute leukemia mortality in Tikur Anbessa Specialized Hospital.
According to the present study findings, the survival rate of AML (19%) subtypes were lower survival rates than ALL (27%) subtypes, which is consistent with the report from the American cancer association that showed AML (66%) subtypes have lower survival rates than ALL (90%) subtypes. 9Another study done in Nigeria also supported the current finding that ALL (77.7%) had better survival than AML (44%) subtypes. 20Although the WBC count and survival of leukemia patients were not significantly correlated in this investigation, a study carried out in the USA suggested that a higher WBC count T A B L E 4 Median survival time, 5-year cumulative survival probability and log-rank test according to hematologic and pathologic characteristics of leukemia patients in Tikur Anbessa Specialized Hospital, Ethiopia (n = 119).(≥50 000/μL) was associated with worse outcomes. 21Our study found that the history of relapses was the primary predictor acute leukemia survival, and those who have relapse history were four times more likely than those of the controls which is similar to the previous study in British 21 and Germany. 22rrent finding shows that, the risk of mortality increased by 1.25 times as people aged (95%CI: 1.3, 1.75) which is consistent with research carried out in Sweden and the United Kingdom 23,24 This result also inline Netherlands showed older age has a lower survival for acute leukemia patients. 25The findings of the current study demonstrated that an individual with acute leukemia who are older than 60 had worse treatment outcomes and survival rates.This is consistent with research of a similar nature conducted in elsewhere. 15Age at diagnosis was one of the risk factors that affected leukemia patients' chances of surviving.Leukemia patients over 65 were shown to have a higher mortality risk than those under 34, which was supported by UK research that found those under 34 had a better outlook. 26According to data from other research, the 5-year survival of acute leukemia patients approaches 40% in younger patients and 15% in older individuals. 14is study also showed that worsened survival among age group >64 years which is similar with previous finding increased age have (>64) have a worse survival. 27The finding was also supported by the Canadian study which showed that age at diagnosis is linked with the survival of acute leukemia cases in which older ages are at a higher risk for death in comparison to the young ones. 23 According to the US study, leukemia subtype and sex are the key determinant factor of survival in leukemia patients.Hence, the study shown that the female patients with a leukemia diagnosis had a better survival rate than male patients. 12There was no correlation between sex and survival status among acute leukemia cases in the current investigation, despite the fact that the study's previous findings indicated that there were survival differences between two genders this may be due to methodological or genetic variation in white and race cohorts.
This study had some limitations.Firstly, we could not determine the impact of chemotherapy and immune-therapy on survival because we lack full course of treatment in most patient's chart.
Secondly, since the sample size were small it lacks generality.The study was retrospective with a possibility of bias from inaccurate survival and predictors information due to the missing data or inadequate reporting by hospital.We do not reported hospital stay due to we have no clear data on patient stay at hospital.Causespecific survival was not determined due to lack of data on specific cause of death, this may overestimate acute leukemia-related mortality rate.

| CONCLUSION AND RECOMMENDATIONS
The survival rate for individuals with acute leukemia cases was very low.Advanced age, history of relapse, hepatomegaly, splenomegaly, and leukemia subtype were explained to be reliable predictors of death in patients with acute leukemia.Therefore, improving early detection and initiation of treatment for all acute leukemia patients is necessary in order to improve the patient's survival status.Further large-scale prospective follow-up studies could be recommended on specific types of acute leukemia.
T A B L E 5 Cox regression analysis of acute leukemia patients on follow-up at Tikur Anbessa Specialized Hospital, Addis Ababa, Ethiopia, 2019 (n = 119).
Data were coded recorded into an excel spread sheet, and then checked for consistency and completeness before being exported into the computer software Statistical Package for Social Sciences (SPSS) version 25.0 for analysis.Descriptive statistics such as frequency, median, mean and standard deviation were used to summarize participants characteristics.The survival rates were calculated using the Kaplan-Meier curve technique.The log-rank test was used to compare the groups' chances of surviving.Bivariate and multivariate Cox proportional hazards regression models were done to identify the factors associated with survival status.

3. 3 |
Incidence and overall survival of acute leukemia patients during the follow-up timeAbout 119 acute leukemia patients were followed for 60 months, yielding 196 person-years-time at risk.In the whole follow up period 46(38.7%)new deaths were observed resulting an overall death incidence rate of 23.5 (95% CL:18-52) per 100 person-years.The median follow-up period for this group is 17 (95% CI, 1-59) months, and the median survival was 35 months (95% CI, 28.3-41.7).In 60 months of follow-up, the estimated survival rate for those with acute leukemia

F I G U R E 1
Overall survival probability of patients starting cancer chemotherapy in Tikur Anbessa Hospital from January 2015 to December 2019.In accordance with the results of the Kaplan-Meier survival test (

T A B L E 3
Median survival time, 5-year cumulative survival probability and log-rank test according to socio demographic characteristics of leukemia patients in Tikur Anbessa Specialized Hospital, Ethiopia (n = 119).
3 | RESULTS 3.1 | Socio-demographic characteristics of the study participants A total of 203 acute leukemia patients were observed from January 2015 to December 2019 in Tikur Anbessa Specialized Hospital (TASH), of which, 119 patient charts were found to have complete data and fulfilled eligibility criteria for this study.The mean age of the study participants was 40.5 ± 20.5 with minimum and maximum age range of 18-88 years old.Regarding to their religion about 53 (44.5%) were Orthodox Christian, 40 (33.6%) were Muslim, and 23 (19.3%) were protestants.More than half of study participants were from Oromia and Amhara regions (28.6% and 24.4%), respectively, and T A B L E 2 Hematologic and pathologic characteristics of acute leukemia patients at Tikur Anbessa Specialized Hospital, Addis Ababa, Ethiopia from January 2015 to December 2019 (n = 119).

Table 4
) patients with ALL subtypes (43 months) (95% CI:37.91-48.08)had a better survival time than those with AML subtypes AML 25(18.28-31.71).Those without any co-morbidities had a longer median survival period (47 months, 95% CI: 28.30-41.69)than those with co-morbidities of any kind.The median survival times for patients with history of hepatomegaly and splenomegaly were 28 (95% CI:20-35.9)and 26 (95% CI:15.90-36.10)days, respectively, compared to those without a history of hepatomegaly and splenomegaly.Patients who did not experience relapse had a median survival time of 23 months.(95% CI:14.82-31.17),those with a history of leukemia