Study design and period

Institutional- based retrospective cross-sectional study was conducted from January1, 2019 to December 31, 2021.

Study area

The study area was university of Gondar comprehensive specialized hospital located at Central North Gondar Zone, Amhara Regional state, Ethiopia. The hospital is found 750 km far from Addis Ababa, capital city of Ethiopia and 171 km away from Bahir Dar city of Amhara regional state. It is the largest hospital in Central-North Gondar zone serving for more than 5 million people per year [15]. It is a multidisciplinary comprehensive specialized hospital with 550 beds. University of Gondar comprehensive specialized hospital is the only hospital that gives functional pediatric oncology service in the region which has 01 oncologist pediatrician, 11 BSC comprehensive nurses, and temporary resident and intern doctors in the ward. The unit has 35 beds but no separate radiology, pathology and physiotherapy units.

Study population

All registered children diagnosed with cancer at university of Gondar comprehensive specialized hospital admitted to pediatric oncology ward from January 1, 2019 to December 31, 2021.

Inclusion criteria

All medical records of children diagnosed with cancer aged ≤ 18 years in pediatrics cancer unit at university of Gondar comprehensive specialized hospital from January1, 2019 to December 31, 2021.

Sample size techniques and sampling procedure

Since the total number of the population under investigation was small (200), Census method was applied (all registered children diagnosed with cancer at university of Gondar comprehensive specialized hospital admitted to pediatric oncology ward included as study population). According to pediatric cancer patients’ report from the registration book, on average 5.6 children have been visited pediatric oncology ward per month and sixty seven per year. Therefore, the study population size was all 200 medical charts of children diagnosed with cancer who were registered from January 1, 2019 to December 31, 2021.

Variables of the study

Dependent variable: Delay in Diagnosis.

Independent variables

Socio-demographic characteristics of child (sex, age, residence, health insurance).

Clinical characteristics (sign/symptom, type of malignancy, comorbid illness).

Health care associated characteristics (first visited facility, hospital level, referral, source for referral, medical specialty of care provider).

Operational definition Delay in diagnosis

The time interval between cancer manifestations detected to confirmed diagnosis (if median diagnosis time ≥ 90 days it is delayed) [16].

Patient delayed diagnosis

The time between the onset of symptoms & signs detected by the patients to first health care visit (if median time ≥ 50days) [16].

Physician delayed diagnosis

the time interval from the first health care provider contact to confirmed cancer diagnosis (if median time ≥ 32days) [16].

Data collection tool and procedure

The data were collected by using structured checklist which was adapted from different literatures [11, 16, 17]. The tool was prepared in English version and the data were extracted through reviewing patient’ medical chart by trained data collectors. Two BSc clinical nurses and one BSc nurse (ward head) supervisor were participated for data collection.

Data Quality Control

To assure the data quality, chart review was done on 5%( 10) of the study population to test the checklists’ structure, completeness and essential modifications were made accordingly at University of Gondar comprehensive specialized hospital. Cronbach alpha (0.75) was measured to check reliability of the tool. Face tool validation checked by experts; training for data collectors and supervisor was given prior to data collection for half day.

Data processing and analysis

The data were checked by the principal investigator on a daily basis during data collection for completeness, and consistencies. Collected Data were then coded, entered to EPI data V4.6, cleaned and analyzed using STATA version 14.00 Software. Descriptive analysis was used to describe the frequencies and percentages of the variables in the study. The strength of association was measured using adjusted odds ratio and 95% confidence interval. Binary logistic regression analysis was used to test associations between independent variables and the dependent variable. Variables with P-value < 0.25 in bi-variable analysis were included to the multivariable logistic regression model. Finally, variables with P-value < 0.05 were considered as potential determinants of delayed diagnosis among pediatric cancer patients. The Hosmer-Lemeshow logistic regression model was fitted at (x2 = 10.36; p = 0.241).