Time to death and its determinants among cervical cancer patients in Tigray, Ethiopia: a retrospective follow up study

Abstract

Background

Cervical cancer remains one of the leading causes of cancer-related deaths among women in Ethiopia. However, limited evidence exists regarding the survival outcomes and associated factors among cervical cancer patients, particularly in conflict-affected regions like Tigray. This study aimed to assess the survival time and its predictors among cervical cancer patients treated at Ayder Comprehensive Specialized Hospital and Axum specialized referral hospitals between 2018 and 2023.

Methods

A retrospective follow-up study was conducted on 91 cervical cancer patients. Data were extracted from medical records using a structured checklist. The primary outcome was time to death from cervical cancer, measured in months. Survival probabilities were estimated using the Kaplan-Meier method, and the log-rank test was used to compare survival across categories. Life tables were used to summarize the probability of survival at specified time intervals. Cox proportional hazards model was employed to identify predictors of mortality, with hazard ratios and 95% confidence intervals reported. The Cox-Snell residual plot was used to assess model adequacy.

Results

The overall median survival time was 19.5 months, with a Four-year cumulative survival rate of just 3.7%. The incidence rate was high at 40.5 per 1000 person-months, based on a total follow-up time of 1431 person-months. Survival outcomes were significantly worse than in comparable Ethiopian studies. Older age and those who underwent chemo and radiotherapy were significantly associated with increased hazard of death. Surprisingly, patients who reported that healthcare providers did not exhibit supportive and respectful behavior had a 62% lower hazard of death (HR = 0.38; 95% CI: 0.18–0.81), a counterintuitive finding that may reflect reverse causality or reporting bias.

Conclusion

Cervical cancer survival in this cohort was markedly low compared to national and international findings. Older age and chemotherapy were associated with higher mortality. The unexpected association between perceived provider behavior and survival warrants further qualitative investigation. To improve outcomes, targeted efforts should focus on strengthening early detection and treatment services, particularly for older women.

Introduction

Cervical cancer remains one of the most common cancers affecting women globally, with a disproportionately high burden in low- and middle-income countries. It is the fourth most prevalent cancer among women worldwide, responsible for an estimated 600,000 new cases and over 350,000 deaths annually. Alarmingly, more than 94% of these deaths occur in low-resource settings, where access to preventive measures, early detection, and effective treatment is often limited. Sub-Saharan Africa, in particular, bears a significant portion of this burden [1]. In Ethiopia, cervical cancer ranks as the second leading cause of cancer mortality among women of reproductive age. According to the World Health Organization, approximately 6300 Ethiopian women are diagnosed with cervical cancer each year, and more than 4800 die from the disease [2].

Globally, the 5-year survival rate for cervical cancer varies significantly depending on the stage at diagnosis and access to treatment. In high-income countries, where organized screening and timely care are available, survival rates can exceed 65%. In contrast, in low- and middle-income countries (LMICs), including much of sub-Saharan Africa, survival rates often fall below 30% due to late-stage diagnosis and limited access to treatment [3, 4]. In Ethiopia, cervical cancer remains the second leading cause of cancer-related death among women, with fewer than 20% surviving beyond five years, primarily due to late diagnosis and inadequate access to specialized care [5, 6].

Cervical cancer is primarily caused by persistent infection with high-risk types of human papillomavirus (HPV), a common sexually transmitted virus. Of the more than 200 known HPV types, HPV-16 and HPV-18 account for approximately 70% of cervical cancer cases worldwide. Risk factors that increase the likelihood of persistent HPV infection and progression to cancer include early sexual activity, multiple sexual partners, smoking, immunosuppression (such as HIV infection), and long-term use of oral contraceptives. Preventive measures such as HPV vaccination and regular cervical screening are essential to reducing the burden of cervical cancer [7, 8].

The impact of cervical cancer extends far beyond premature death, causing chronic pain, disability, and profound social and economic disruption for affected women and their families. Most women diagnosed with cervical cancer are in their most productive years, and the disease often results in significant loss of income and productivity. The financial burden associated with diagnosis and treatment much of which is frequently paid out-of-pocket can be devastating, especially in impoverished regions like Tigray. As women often serve as primary caregivers and key income earners in their households, their illness or premature death disrupts family stability, jeopardizes children’s education and well-being, and increases emotional and psychological strain on family members. Moreover, cervical cancer is often surrounded by stigma, which can lead to social isolation and delays in seeking medical care, further worsening health outcomes [9, 10].

Several factors contribute to the persistently high morbidity and mortality rates from cervical cancer in Ethiopia. Most women are diagnosed at advanced stages when treatment options are limited and survival chances are poor. This is fueled by a lack of widespread awareness of cervical cancer symptoms and inadequate screening and diagnostic services both of which are essential for early detection [11]. Geographic remoteness and financial constraints further restrict access to healthcare, while systemic challenges including shortages of trained oncologists, limited radiotherapy equipment, frequent chemotherapy stock outs, and weak referral mechanisms undermine treatment capacity [12].

In Tigray, the armed conflict that began in late 2020 severely disrupted health services: up to 86% of health facilities were damaged or destroyed, and access to essential cancer diagnostics and treatment was substantially compromised [13]. Additionally, prevalent comorbidities such as HIV, malnutrition, and anemia conditions that impair immune function and reduce treatment efficacy further exacerbate disease progression and mortality in the region [14].

Despite the magnitude of the problem, there is a lack of empirical data on survival outcomes and the determinants of time to death among cervical cancer patients in Tigray. While national-level studies have begun to explore survival rates and prognostic factors, significant regional variations in healthcare access, patient demographics, and disease presentation underscore the need for localized research. This need has become even more urgent in the context of two overlapping crises that have severely strained the region’s health system. The COVID-19 pandemic disrupted cancer screening, diagnostic services, and treatment continuity across Ethiopia, diverting critical resources and reducing patients’ ability or willingness to seek timely care. Compounding this, the armed conflict in Tigray has led to the collapse of health infrastructure, with widespread destruction of facilities, displacement of healthcare professionals, supply chain disruptions, and prolonged service interruptions. These conditions have likely delayed diagnosis, interrupted treatment regimens, and worsened health outcomes for many cervical cancer patients, but the extent of this impact remains undocumented. Understanding how long women survive after diagnosis and identifying the factors influencing their outcomes under such extraordinary circumstances is essential for designing context-specific and crisis-informed interventions. This study is therefore undertaken to fill this critical knowledge gap by analyzing time to death and its predictors among cervical cancer patients in Tigray.

Methods and materials

Study area

The study was conducted in the Tigray region in the aftermath of a devastating war. The region’s healthcare system comprises 2 specialized comprehensive hospitals, 14 general hospitals, 24 primary hospitals, 231 health centers, and 743 health posts. Due to security concerns, some areas specifically the western zone, parts of the southern zone, and peripheral regions were excluded from the study.

Cervical cancer screening services were initiated in 2011 at Mekelle Hospital, Lemlem Karl Hospital, and St. Mary Hospital in Axum. Clients with suspected cervical cancer were referred to Addis Ababa for confirmation until pathology services became available at Ayder Comprehensive Specialized Hospital (ACSH). In 2023, Aksum Referral Hospital also began offering pathology services for confirming precancerous cervical lesions. Prior to this, Aksum Referral Hospital primarily provided follow-up care for patients diagnosed at ACSH.

Among healthcare facilities in the region, Ayder and Aksum Comprehensive Specialized Hospitals offer the most advanced cervical cancer services. As referral centers, they provide screening using VIA (Visual Inspection with Acetic Acid)/Lugol’s iodine and pathology-based diagnostic confirmation. Immediately before the outbreak of war, HPV DNA testing was launched at the Tigray Health Research Institute in Mekelle, serving as a referral hub for six surrounding health facilities.

Treatment services in the region include cryotherapy, thermal ablation, and LEEP for precancerous lesions, as well as surgery and chemotherapy for cervical cancer. Currently, approximately 30 health facilities across the Tigray region provide cervical cancer screening services.

Study design and period

Data were retrospectively collected from cervical cancer registration books, patient charts, and Smart Care records at Ayder Comprehensive Specialized Hospital (ACSH) and Axum specialized referral hospitals, supplemented by information obtained from patients or patients’ partners. The study included cases diagnosed between August 2018 and August 2023, encompassing periods before, during, and after the war (which lasted from November 2020 to November 2023). A retrospective follow-up design was used to assess the survival outcomes of women diagnosed with cervical cancer.

Study population

The study population consisted of all women diagnosed with cervical cancer and registered for follow-up care at Ayder Comprehensive Specialized Hospital (ACSH) and Axum specialized referral hospital between August 2018 and August 2023.

Inclusion criteria

Women with a confirmed diagnosis of cervical cancer during the study period were included if their clinical and demographic data were complete and documented in the cervical cancer registration books, patient charts, or Smart Care records. Among these confirmed cases, only patients who were traceable and had available information on survival status and behavioral characteristics were considered eligible. Patients with other concurrent malignancies were excluded to minimize potential confounding.

Recruitment, data extraction procedure and tool

Data for this study were retrospectively extracted using a standardized data abstraction tool specifically developed for this purpose. The tool was designed to ensure accurate and comprehensive capture of key variables from both paper-based cervical cancer registration logbooks and the hospital’s electronic medical records. Prior to use, the tool was pre-tested on a small sample of patient records to improve clarity and consistency. Trained clinical data collectors manually reviewed and extracted the necessary information, with regular supervision and cross-checking to maintain data quality.

A structured series of steps was followed to accurately identify and follow up all cervical cancer patients enrolled at ACSH between August 2018 and August 2023. Initially, 125 patients were identified as having cervical cancer from the logbooks. Personal details including name, marital status, address, contact information, and key clinical dates were extracted. Using the Medical Record Number (MRN) from the logbooks, each patient’s electronic medical file in the SMART Care system was reviewed to determine the type and sequence of management received. During this step, four patients were excluded because they did not meet the clinical criteria for cervical cancer.

Next, a detailed review of each patient’s physical medical chart was performed to obtain additional clinical information, including medical history, date of diagnosis, treatment initiation, and treatment modalities. This review was essential for identifying and correcting misclassifications between cervical cancer and precancerous cervical lesions. As a result, four additional patients who had been incorrectly labeled as having cervical cancer were excluded.

Following the logbook, SMART Care database, and chart review, a desk assessment was conducted for the remaining 117 patients to identify potential sources for patient tracing. This involved verifying mobile phone numbers for patients or close relatives and confirming the accuracy of addresses and contact information. Only patients who were traceable were eligible for inclusion in the study. The tracing of survival status employed a multi-tiered approach. Patients with active mobile numbers were contacted directly. For those unreachable by phone, close relatives such as spouses, children, or parents were contacted. In cases where contact information was unavailable or incomplete, health extension workers (HEWs) traced patients in their respective kebeles (the lowest administrative unit in Ethiopia) and consulted other community members to verify survival status.

Using this approach, 91 of 117 patients were successfully traced, and their survival status was determined. The remaining 26 patients could not be traced, primarily due to inaccurate contact information or inability to locate their physical addresses. A small number of cases also involved limited cooperation from family members (Fig. 1).

Fig. 1.

Schematic representation of the recruitment and data extraction process for cervical cancer patients in Tigray, Ethiopia (2018–2023) (n = 91)

To ensure the quality and accuracy of data collection, supervisors with master’s degrees in nursing and midwifery were recruited to oversee the process. Additionally, 19 nurses and midwives health professionals with prior training in cervical cancer screening and management were trained and deployed as data collectors. These professionals were stationed in hospitals and health centers across the catchment area. The training covered the study objectives, use of data collection tools, and appropriate methods for approaching and determining the survival status of patients.

Variables

Dependent variable

Survival time (time to death).

Independent variables

The main exposure variables in this study included socio-demographic factors (age at diagnosis, marital status, place of residence, occupation, and educational status), cancer-related symptoms (abnormal vaginal bleeding, pelvic pain, pain during intercourse, and vaginal discharge), as well as information on treatment and diagnostic modalities. Additionally, variables related to the social, economic, and health impacts of cervical cancer, reaction after diagnosis, lifestyle changes, coping mechanisms, and changes in personal relationships were also considered as independent variables.

Operational definition

• Comorbidities: Defined as the presence of one or more additional medical conditions diagnosed before or during cervical cancer treatment.

• Exhibiting supportive and respectful behavior: refers to consistently demonstrating attitudes and actions that acknowledge others’ dignity, value, and perspectives, while offering appropriate assistance, encouragement, and consideration in interpersonal interactions. This is expressed through Communicating respectfully and empathetically, demonstrating consideration, respecting dignity, and autonomy.

• Survival Time: The duration in months between the date of confirmed cervical cancer diagnosis and the date of death or the date of last follow-up (for censored cases). Survival time was calculated using medical records and was measured in months. Patients who were alive at the end of the follow-up period or lost to follow-up were considered censored.

Event (death)

The occurrence of death during the follow-up period was recorded as the primary event. An event was coded as “1” if the patient died during the study period and “0” if the patient was censored (alive at last follow-up or lost to follow-up). Deaths were confirmed through reports provided by the patient’s partner or closest relative.

Data quality assurance

To ensure data quality, standardized data collection tools were used, and data collectors received thorough training on the study protocol and proper record-keeping. Completed questionnaires and medical records were regularly reviewed for completeness and accuracy by supervisors. During data entry, double data entry and cross-checking were performed to minimize entry errors. Logical consistency checks and validation rules were applied within STATA to identify and correct inconsistencies, missing values, and outliers. Regular backups of the dataset were maintained to prevent data loss. Additionally, the assumptions of statistical models were carefully checked to ensure the reliability and validity of the analysis results.

Data analysis

The collected data were entered and managed using STATA version 14, with meticulous attention to data quality. Data cleaning involved checking for completeness, correcting inconsistencies, validating date sequences, and identifying outliers.

Following data preparation, univariate analysis was conducted to describe baseline characteristics of the study population. Frequencies and percentages were calculated for categorical variables, while continuous variables were summarized using means with standard deviations or medians with interquartile ranges, depending on their distribution. Survival analysis began with the construction of life tables to estimate interval-specific survival probabilities and cumulative survival over defined time periods.

Next, Kaplan-Meier survival curves were generated to estimate overall survival probabilities, and differences between groups were assessed using the Log-Rank test. Cox proportional hazards regression was then used for both bivariable and multivariable analyses to identify independent predictors of mortality among women diagnosed with cervical cancer. Variables considered for inclusion in the multivariable Cox proportional hazards regression model were selected using a combination of statistical and clinical considerations. Specifically, variables with a p-value ≤ 0.25 in bivariable analyses were initially considered, and those deemed clinically or biologically relevant based on prior literature were also included, regardless of their bivariable p-value. This approach aimed to balance statistical evidence with subject-matter knowledge, ensuring appropriate adjustment for potential confounders.

Prior to fitting the final model, multicollinearity among independent variables was evaluated using the Variance Inflation Factor (VIF = 1.74). The proportional hazards assumption was further assessed using Schoenfeld residuals and graphical diagnostics, confirming no significant violations. Censoring was assumed to be non-informative, meaning that the likelihood of being censored was independent of the risk of the event of interest. This assumption is commonly applied in survival analyses and is considered reasonable in our study, as censoring largely occurred due to administrative reasons (e.g., end of follow-up) rather than factors related to patient prognosis.

Adjusted hazard ratios with 95% confidence intervals were calculated to measure the strength of associations, and variables with p-values less than 0.05 were considered statistically significant. Final results were reported using life tables, Kaplan-Meier survival curves, hazard ratio tables, and descriptive statistics to support interpretation and clinical relevance.

Ethical consideration

Prior to data collection, ethical approval for this study was obtained from the Health Research Ethics Review Committee (HRERC) of Mekelle University, College of Health Sciences, under ethical registration code of MU-IRB1836/2021. Permission to access patient records was granted by the hospital administration in accordance with institutional and national regulations. Patient confidentiality was rigorously protected by anonymizing all data and assigning unique identification codes in place of personal identifiers. Data were securely stored, with access strictly limited to the research team. As this was a retrospective study utilizing existing medical records, the requirement for informed consent was waived by the ethics committee; however, every effort was made to ensure the protection of patient privacy. For information obtained directly from patients or their partners, verbal informed consent was obtained. The study was conducted in accordance with the principles of the Declaration of Helsinki and all relevant ethical guidelines.

Result

Socio-demographic characteristics of participants

This retrospective study was conducted using the medical records of 91 cervical cancer patients with complete documentation. Of these, 41 (45.1%) were recorded as illiterate, and 46 (50.5%) were identified as housewives by occupation. The majority of patients, 63 (69.2%), were married, and more than 38% were documented as being beyond reproductive age. The mean age of the patients was 53.9 ± 9.9 years (Table 1).

Table 1.

Socio-demographic characteristics of cervical cancer patients in Tigray, Ethiopia (2018–2023 (n = 91)

Variable characteristics		Survival Status		Total N (%)
		Censored	Event
		n(%)	n(%)
Age category	30–40	6(18.2)	3(5.2)	9(9.9)
	41–50	12(36.4)	17(29.3)	29(31.9)
	51–60	10(30.3)	25(43.1)	35(38.5)
	> 60	5(15.2)	13(22.4)	18(19.8)
Educational status	Illiterate	9(27.3)	32(55.2)	41(45.1)
	Primary school	16(48.5)	11(19.0)	27(29.7)
	Secondary school	5(15.2)	8(13.8)	13(14.3)
	College and above	3(9.1)	7(12.1)	10(11.0)
Occupational status	Farmer	6(18.2)	8(13.8)	14(15.4)
	Employed (government/private)	5(15.2)	8(13.8)	13(14.3)
	House wife	14(42.4)	32(55.2)	46(50.5)
	Self Employed	8(24.2)	10(17.2)	18(19.8)
Marital status	Married	26(78.8)	37(63.8)	63(69.2)
	Unmarried	7(21.2)	21(36.2)	28(30.8)

Sign and symptoms of cervical cancer patients

Abnormal vaginal bleeding was documented in the majority of patients, 81 (89.0%). However, over 65% of the participants reported no pain during sexual intercourse (Table 2).

Table 2.

Signs and symptoms of cervical cancer patients in Tigray, Ethiopia, 2018–2023 (n = 91)

Variable characteristics		Survival Status		Total N(%)
		Censored (%)	Event (%)
Abnormal Vaginal Bleeding	Yes	31(93.9)	50(86.2)	81(89.0)
	No	2(6.1)	8(13.8)	10(11.0)
Pelvic or back pain	Yes	24(72.7)	34(58.6)	58(63.7)
	No	9(27.3)	24(41.4)	33(36.3)
Pain during sexual intercourse	Yes	13(39.4)	18(31.0)	31(34.1)
	No	20(60.6)	40(69.0)	60(65.9)
Vaginal discharge	Yes	13(39.4)	23(39.7)	36(39.6)
	No	20(60.6)	35(60.3)	55(60.4)

Diagnosis and treatment related characteristics of cervical cancer patients

The majority of participants, 52 (57.1%), were diagnosed between 2013 and 2015 Ethiopian Calendar (E.C.). Biopsy was the primary method of diagnosis in 71 patients (78%). For 51 participants (56.0%), the diagnosis was initiated only after the development of clinical signs and symptoms. Treatment interruptions occurred in 39 patients (42.9%), with the most common reasons being financial constraints and lack of transportation, affecting 21 participants (23.1%). Additionally, for the majority of patients, 50 (54.9%), the duration of treatment exceeded two weeks (Table 3).

Table 3.

Diagnosis and treatment-related characteristics of cervical cancer patients in Tigray, Ethiopia (2018–2023 (n = 91)

Variable		Survival Status		Total N(%)
		Censored (%)	Event (%)
Year of cervical cancer diagnosis	2018–2020	4(12.1)	35(60.3)	39(42.9)
	2021–2023	29(87.9)	23(39.7)	52(57.1)
Diagnostic method	Tissue punch	4(12.1)	10(17.2)	14(15.4)
	Vaginal swab	12(36.4)	9(15.5)	21(23.1)
	Unknown	17(51.5)	39(51.5)	56(61.5)
Diagnosis with biopsy	Yes	31(93.9)	40(69.0)	71(78.0)
	No	2(6.1)	18(31.0)	20(22.0)
Diagnosis with Pap smear	Yes	6(18.2)	24(41.4)	30(33.0)
	No	27(81.8)	34(58.6)	61(67.0)
Diagnosis with HPV-DNA	Yes	3(9.1)	7(12.1)	10(11.0)
	No	30(90.9)	51(87.9)	81(89.0)
Diagnosis with CT -Scan/MRI	Yes	3(9.1)	5(8.6)	8(8.8)
	No	30(90.9)	53(91.4)	83(91.2)
Time of diagnosis	After sign and symptom developed	22(66.7)	29(50.0)	51(56.0)
	Another medical test	9(27.3)	22(37.9)	31(34.1)
	Routine cervical cancer checkup	2(6.1)	7(12.1)	9(9.9)
Treatment duration	< 1wk	7(21.2)	9(15.5)	16(17.6)
	> 2wks	20(60.6)	30(51.7)	50(54.9)
	1-2wks	6(18.2)	19(32.8)	25(27.5)
Treatment modality	Surgery	8(24)	3(5)	11(12.8)
	Surgery and Chemo	7(21)	2(3.5)	9(9.9)
	Surgery, chemo & Radio	5(15.2)	0(0.0)	5(5.5)
	Chemo & Radiotherapy	7(21.)	41(70.7)	48(52.80)
	Palliative care only	6(18.2)	12(20.7)	18(19.8)
Treatment interrupted	Yes	13(39.4)	26(44.8)	39(42.9)
	No	20(60.6)	32(55.2)	52(57.1)
Reason for treatment interruption	Lack of money & transport	10(71.4)	11(44)	21(53.85)
	Movement restriction & side effect	4(28.6)	14(56)	18(46.15)
Treatment process affected by war	Yes	24(72.7)	39(67.2)	63(69.2)
	No	9(27.3)	19(32.8)	28(30.8)
Effect of the war on treatment	No Medication or hospital services	13(39.4)	20(34.5)	33(36.3)
	No money	7(21.2)	10(17.2)	17(18.7)
	Not mentioned	4(12.1)	6(10.3)	10(11.0)
	Not affected	9(27.3)	22(37.9)	31(34.1)

Socioeconomic effect of the disease and health seeking behavior of the participants

For the majority of participants, 51 (56%), treatment expenses were covered through personal savings (out-of-pocket), with 59 (64.8%) of them receiving financial support from family members or friends. Additionally, 31 (34.1%) of the participants reported selling household assets to cover treatment costs, with 16 (17.6%) specifically selling their house or household equipment. After being diagnosed, 25 (27.5%) of the women’s partners and 21 (23.1%) of other family members had to leave their jobs to assist with treatment. A significant number, 73 (80.2%) of the victims, reported experiencing financial hardship due to the illness.

Following the onset of the disease, 47 (51.6%) of their partners reported emotional distress, 28 (30.8%) experienced a decline in their ability to provide physical support, and 72 (79.1%) reported negative changes in their sexual relationships. As coping mechanisms, 68 (74.7%) of the patients sought help from health professionals, while 46 (50.5%) turned to religious leaders for support (Table 4).

Table 4.

Socioeconomic effects of the disease and health-seeking behaviors among participants in Tigray, Ethiopia (2018–2023) (n = 91)

Variable characteristics		Survival Status		Total N (%)
		Censored	Event
		N (%)	N (%)
Treatment Costs Paid from Personal Savings	Yes	22(66.7)	29(50.0)	51(56.0)
	No	11(33.3)	29(50.0)	40(44.0)
Treatment Costs Covered by Health Insurance	Yes	1(3.0)	4(6.9)	5(5.5)
	No	32(97.0)	54(93.1)	86(94.5)
Treatment Costs Financed Through Loans or Credit	Yes	9(27.3)	12	21(23.1)
	No	24(72.7)	46(79.3)	70(76.9)
Treatment costs covered by family or friends	Yes	22/(66.7)	37(63.8)	59(64.8)
	No	11(33.3)	21(36.2)	32(35.2)
Sale of Assets to Finance Treatment	Yes	13(39.4)	18(31.0)	31(34.1)
	No	20(60.6)	40(69.0)	60(65.9)
Types of Assets Sold to Cover Treatment Costs	Animals	3(9.1)	4(6.9)	7(7.7)
	House or house equipment	8(24.2)	8(13.8)	16(17.6)
	Not mentioned	4(12.1)	7(12.1)	11(12.1)
	No assets sold	18(54.5)	39(67.2)	57(62.6)
Left Job to Seek Treatment	Yes	13(39.4)	22(37.9)	35(38.5)
	No	20(60.6)	36(62.1)	56(61.5)
Partner left job for treatment purpose	Yes	10(30.3)	15(25.9)	25(27.5)
	No	23(69.7)	43(74.1)	66(72.5)
Family left Job for treatment purpose	Yes	7(21.2)	14(24.1)	21(23.1)
	No	26(78.8)	44(75.9)	70(76.9)
Changed residence for treatment purpose	Yes	9(27.3)	7(12.1)	16(17.6)
	No	24(72.7)	51(87.9)	75(82.4)
Emotional impact due to the disease	Yes	9(27.3)	17(29.3)	26(28.6)
	No	24(72.7)	41(70.7)	65(71.4)
Financial impact due to the disease	Yes	31(93.9)	42(72.4)	73(80.2)
	No	2(6.1)	16(27.6)	18(19.8)
Social impact due to the disease	Yes	18(54.5)	23(39.7)	41(45.1)
	No	15(45.5)	35(60.3)	50(54.9)
Emotional Impact on Relationship with Spouse	Yes	20(60.6)	27(46.6)	47(51.6)
	No	13(39.4)	31(53.4)	44(48.4)
Impact on Care and Support from Spouse	Yes	9(27.3)	26(44.8)	35(38.5)
	No	24(72.7)	32(55.2)	56(61.5)
Impact on Physical Support from Spouse	Yes	15(45.5)	13(22.4)	28(30.8)
	No	18(54.5)	45(77.6)	63(69.2)
Impact on Sexual Relationship with Spouse	Yes	24(72.7)	48(82.8)	72(79.1)
	No	9(27.3)	10(17.2)	19(20.9)
Consulted a health professional for coping	Yes	25(75.8)	43(74.1)	68(74.7)
	No	8(24.2)	15(25.9)	23(25.3)
Sought guidance from a religious leader for coping	Yes	21(63.6)	25(43.1)	46(50.5)
	No	12(36.4)	33(56.9)	45(49.5)
Used self-care strategies for coping	Yes	15(45.5)	16(27.6)	31(34.1)
	No	18(54.5)	42(72.4)	60(65.9)

Incidence of mortality and patient survival

A total of 91 cervical cancer patients were observed over a combined follow-up period of 1431 person-months. During this time, the overall incidence rate of mortality among patients who received medical care was 40.5 per 1000 person-months of observation (95% CI: 31.33 to 52.43).

When stratified by age, patients aged over 60 years exhibited the highest mortality rate among all follow-up groups, with 61 deaths per 1000 person-months (95% CI: 35.44–105.11). A similarly elevated mortality rate was observed among patients who underwent both chemotherapy and radiotherapy, at 54 deaths per 1000 person-months (95% CI: 39.9–73.7).

Mortality rates before and after the onset of the war were nearly identical, reported at 40.3 (95% CI: 28.92–56.01) and 40.9 (95% CI: 27.20–61.59) per 1000 person-months, respectively. Furthermore, patients who were unable to fund their treatment through personal savings experienced a higher mortality rate 46.77 per 1000 person-months (95% CI: 32.50–67.31)—compared to 35.76 per 1000 person-months (95% CI: 28.6–54.5) among those who could afford to pay.

In terms of clinical symptoms, patients who presented with vaginal bleeding had a lower mortality rate (39 per 1000 person-months; 95% CI: 29.6 to 51.5) compared to those who did not experience bleeding (53.3 per 1000 person-months; 95% CI: 26.67 to 106.65). Additionally, the mode of diagnosis played a role in patient outcomes. Those diagnosed through routine cervical cancer screening had the highest mortality rate at 45.5 per 1000 person-months (95% CI: 21.67 to 95.35) compared to those diagnosed after showing symptoms or during unrelated medical visits.

The mortality rate also varied based on treatment and financial support factors. Patients who received chemotherapy showed the highest mortality rate among treatment groups, with 52.6 per 1000 person-months (95% CI: 37.0 to 74.8). On the other hand, those who paid for their treatment through personal savings (out-of-pocket) experienced a comparatively lower mortality rate of 35.8 per 1000 person-months (95% CI: 24.8 to 51.5). Furthermore, patients whose care and support from their spouses were negatively affected by the disease reported higher mortality, with an incidence of 43.3 per 1000 person-months (95% CI: 30.45 to 61.56). These findings suggest that social, clinical, and economic factors significantly influence mortality outcomes among cervical cancer patients (Fig. 2).

Fig. 2.

Proportion of mortality among cervical cancer patients in Tigray, Ethiopia (2018–2023) (n = 91)

A total of 91 cervical cancer patients were followed for a minimum of 1 month and a maximum of 60 months. The median survival time was estimated between 12 and 24 months, while the interquartile range (IQR) spanned from 6 to 30 months, indicating variability in survival outcomes. During the follow-up period, 58 patients (63.7%) died, and 33 (36.3%) were censored. The majority of deaths occurred between 6 and 12 months of follow up (Table 5). Some censored observations in the actuarial life table result from the mid-interval adjustment used to calculate survival probabilities and do not represent true censoring.

Table 5.

Survival life table of cervical cancer patients in Tigray, Ethiopia (2018–2023; n = 91)

Time interval in months	No at start	Censored	No at risk	Death	Interval survival proportion	Cum Survival proportion	[95% Conf. Int.]
0–6	91	8	83	3	0.96	0.96	0.8969	0.9887
6–12	80	12	68	17	0.75	0.7437	0.6311	0.8266
12–18	51	7	54	11	0.75	0.5715	0.4471	0.6778
18–24	33	1	32	10	0.71	0.3956	0.2751	0.5136
24–30	22	1	21	8	0.68	0.2484	0.1466	0.3641
30–36	13	1	12	4	0.67	0.1689	0.0843	0.2785
36–42	8	2	6	3	0.50	0.0965	0.0337	0.1995
48–54*	3	1	2	2	0.000	0.0193	0.0005	0.1315

*Intervals are 6 months, and extend to 54 months for data structure purpose

The probability of survival was 94.1% (95% CI: 86%-97.1%) at 6 months post-diagnosis, decreasing to 61% (95% CI: 50%-71.1%) at 12 months, 30% (95% CI: 44.7%–67.8%) at 24 months (95%CI:19%-41.2%), 11.1% (95% CI: 4.3–21.4%) at 36 months, and by 48 months, survival further dropped to 3.7% (95% CI: 0.35 − 0.15%) (Fig. 3).

Fig. 3.

Over all Kaplan–Meier survival estimates among cervical cancer patients who live in Tigray, Ethiopia (2018–2023) (n = 91)

When stratified by financial status, the 6-month survival probability was 89.1% among patients who financed their treatment through other means, compared to 97.9% among those who used personal savings. However, by the end of the follow-up period, cumulative survival had declined sharply to 0.0% and 3.75%, respectively (Fig. 4).

Fig. 4.

Kaplan–Meier survival estimates by financial saving for treatment among cervical cancer patients in Tigray, Ethiopia (2018-2023) (n = 91)

A similar trend was observed when comparing survival by age: at 24 months of follow-up, patients over 60 years had a survival rate of 0.0%, while those aged 30–40 had a survival rate of 50% (Fig. 5). Furthermore, analysis by treatment modality revealed that patients who received only chemotherapy and radiotherapy had no survival at the end of the follow-up period (Fig. 6). Survival probabilities differed significantly across age groups and treatment modalities, with log-rank p-values of 0.03 and 0.005, respectively.

Fig. 5.

Kaplan–Meier survival estimates by age among cervical cancer patients in Tigray, Ethiopia (2018–2023) (n = 91)

Fig. 6.

Kaplan–Meier survival estimates by treatment modality among cervical cancer patients in Tigray, Ethiopia (2018–2023) (n = 91)

Predictors of mortality

On bivariable Cox proportional hazard model, Age, educational status, presence of back pain, treatment modality, Husband reaction after diagnosis, Supportive and respectful behavior of the treatment team, drop out of job of family members for treatment arrangement, exploring new coping strategy, considering treatment type on deciding to join care and support, and year of diagnosis have showed statistical significance.

In the multivariable Cox proportional hazards model, only age, treatment modality, and the supportive and respectful behavior of the treatment team remained independently associated with time to death. Each additional year of age was associated with a 5% (AHR = 1.05; 95% CI: 1.01–1.08) increase in the hazard of death (hazard ratio [HR] = 1.05). Similarly, patients who underwent only chemotherapy and radiotherapy had a 3.56-fold higher hazard of death (AHR = 3.56; 95% CI: 1.34–9.49) compared to those who received surgery alone, neoadjuvant chemotherapy and surgery, or a combination of surgery, adjuvant chemotherapy, and radiotherapy. Interestingly, patients who reported that the treatment team did not exhibit supportive and respectful behavior had a 60% lower hazard of death compared to those who reported receiving such behavior (AHR = 0.40; 95% CI: 0.20–0.83) This counterintuitive finding suggests that supportive and respectful behavior of the treatment team, as perceived by patients, was associated with an increased risk of mortality in this cohort (Table 6).

Table 6.

Predictors of survival time A among women treated at ayder comprehensive specialized hospital, Tigray, Ethiopia (2018–2023) (n = 91)

Variables	CHR (95%)	AHR(95%CI)	Std. Err	z	P-Value
Age	1.05(1.02–1.07)	1.05(1.01–0.08)	0.02	2.64	0.008
Educational status
Illiterate	1.74(0.76–3.98)	1.33(0.44-4.00)	0.75	0.51	0.612
Primary school complete	0.82(0.31–2.12)	0.60(0.20–1.78)	0.33	-0.91	0.361
Secondary School complete	1.47(0.53–4.09)	1.28(0.38–4.32)	0.79	0.40	0.693
College and above	1	1
Back pain
Yes	1.03(0.59–1.80)	1.14(0.59–2.20)	0.39	0.39	0.699
No	1	1
Treatment type
Surgery or CS* or CSR#	1	1
Chemo & Radio therapy	3.55(1.40–8.99)	3.56(1.33–9.49)	1.78	2.54	0.011
Palliative care only	2.13(0.75–6.09)	3.04(0.90-10.27)	1.89	31.79	0.073
Husband felt sad at Diagnosis
No	1	1
Yes	0.61(0.36–1.02)	0.77(0.42–1.42)	0.24	-0.83	0.407
Treatment team had Supportive and respectful behavior
Yes	0.57(0.34–0.98)	0.41(0.20–0.83)	0.15	-2.47	0.013
No	1	1
Family members left job for Rx arrangement
Yes	1	1
No	1.03(0.56–0.88)	0.80(0.42–1.50)	0.26	-0.70	0.482
Explored new coping strategy
Yes	3.36(1.00-11.21)	1.09(0.28–4.24)	0.76	0.13	0.896
No	1	1
Treatment type was considered for deciding care
Yes	2.57(1.31–5.03)	1.64(0.67–4.05)	0.76	1.07	0.282
No	1	1
Year of diagnosis
2018–2020	1	1
2021–2023	1.55(0.87–2.76)	1.77(0.90–3.47)	0.61	1.67	0.096

*Chemotherapy and Surgery, #Chemotherapy, Surgery and radiotherapy

The Cox proportional hazards assumption was assessed using the global Schoenfeld residuals test for the full model, and the assumption was satisfied (global test p = 0.94). All covariates met the proportional hazards assumption. Model fit was further evaluated using Cox-Snell residuals, and the final model demonstrated a good fit, where the cumulative hazard function closely follows the 45° line (Fig. 7).

Fig. 7.

Nelson-Aalen cumulative hazard plot of cox-snell residuals among cervical cancer patients in Tigray, Ethiopia (2018–2023) (n = 91)

Discussion

The aim of this study was to explore the survival time of cervical cancer patients in Tigray, Northern Ethiopia. Our findings showed that the five-year cumulative survival rate was markedly low, declining from 96.5% in the first year to just 3.7% by the end of the fourth year. The survival rates at two, three, and four years were also low, at 14.6%, 8.4%, and 3.4%, respectively. These figures are significantly lower than those reported in other regions of Ethiopia, such as Black lion Specialized Hospital (38.62%), Addis Ababa Oncology Center (18.27%), and the Amhara region (53.15%) [15–17]. This discrepancy may be attributed to the relatively better healthcare infrastructure and cancer management services available in Addis Ababa, the capital city. Differences in community awareness and access to early detection services may also contribute. Furthermore, the study was conducted in the aftermath of the war in Northern Ethiopia, during which many health facilities were damaged or destroyed, likely affecting timely diagnosis and treatment.

Our study also found that the cumulative survival of cervical cancer patients was lower compared to findings from Kenya (7.3%), Bhutan (22.7%), Ghana (32.4%), and Western Kenya (45%), with follow-up periods ranging from 2 to 5 years [18–22]. These discrepancies may be due to differences in follow-up duration and the level of advancement in healthcare systems. In particular, the war crisis that began in the Tigray region in 2020 severely disrupted healthcare infrastructure, likely worsening disease progression and limiting access to timely treatment. For patients with advanced cervical cancer, survival can only be meaningfully extended when comprehensive and continuous care is available.

The overall median survival time was short, at 19.5 months, with a low total person-time of 1,431 person-months and a high incidence rate of 40.5 per 1000 participants at risk. This median survival time is notably lower than findings from studies conducted at Black lion Hospital (37 months), the Addis Ababa Oncology Center (54 months), and Hawassa, Ethiopia (37 months) [15–17]. The disparity may be attributed to differences in healthcare infrastructure and the impact of the war crisis in Tigray, which severely disrupted health services across the region.

This explanation is supported by our finding that mean survival time was longer among patients diagnosed before the war (23 months) compared to those diagnosed after the onset of the conflict (18 months). Early detection of cervical cancer is critical, as it allows for timely and effective management. In both pre- and post-war groups, the majority of participants presented with abnormal vaginal bleeding the most common symptom of cervical cancer. This symptom arises when necrotic or damaged squamous cells lose structural integrity and bleed upon contact.

The incidence rate in this study was significantly higher compared to previous studies conducted at the Addis Ababa Oncology Center (7.34 per 1000 person-months) and Black lion Hospital (31 per 100 person-years), despite having a much shorter total person-time of observation (1431 months in this study versus 8167 months and 738 years in the referenced studies) [16, 23]. This difference may be attributed to disparities in healthcare systems. Limited access to early detection and diagnostic services in under-resourced settings can delay diagnosis and contribute to rapid disease progression to advanced stages. Additionally, lower community awareness about cervical cancer symptoms and the importance of screening may further hinder early detection and timely treatment. While advanced cervical cancer is costly and difficult to manage, it is often treatable when diagnosed early.

According to the proportional hazards model, each additional year of age was associated with an increased risk of death from cervical cancer. Clinically, this indicates that older patients are at slightly higher risk and may benefit from closer monitoring, earlier interventions, or more aggressive management when feasible. This finding aligns with previous studies conducted at Black lion Hospital and in Addis Ababa [15, 23–25]. The increased hazard among older women may be attributed to a higher likelihood of comorbidities and a diminished biological response to disease processes. Furthermore, older patients are often diagnosed at more advanced stages of cervical cancer, which can complicate treatment and significantly reduce survival rates [18, 26, 27].

Patients who received only chemotherapy and radiotherapy had a substantially higher hazard of death compared to those who underwent surgery alone, neoadjuvant chemotherapy followed by surgery, or a combination of surgery with adjuvant chemotherapy and radiotherapy. Clinically, this likely reflects differences in disease severity and resectability rather than a direct causal effect of treatment. Surgery is typically offered to patients with early-stage, resectable tumors, who generally have a more favorable prognosis, whereas patients ineligible for surgery often have more advanced or inoperable disease, limiting the effectiveness of chemotherapy and radiotherapy alone. These findings should therefore be interpreted as indicating that treatment modality is a marker of underlying disease severity rather than a causal determinant of survival. This pattern has been observed in previous studies conducted at Tikur Anbessa Specialized Hospital, in other parts of Ethiopia, and in Prague, where multimodal treatment approaches that included surgery were associated with significantly improved survival [18, 24, 25, 28, 29].

Interestingly, patients who reported that the treatment team did not exhibit supportive and respectful behavior had a 62% lower hazard of death compared to those who did report receiving such behavior (HR = 0.38; 95% CI: 0.18–0.81). This counterintuitive finding may be influenced by several factors and should be interpreted with caution. First, patients receiving more respectful care may have presented with more advanced disease or complications, prompting healthcare providers to offer increased emotional support and attention a form of reverse causality. Second, subjective perceptions of care quality may vary based on patient expectations, health literacy, or psychological state, all of which can influence both reporting and outcomes. Third, residual confounding or measurement bias may exist, particularly if care experience was assessed retrospectively or through self-report without standardized measures. Further qualitative investigation may be necessary to understand the dynamics between patient-provider interaction and survival outcomes in this context.

Limitations

This study has several limitations that should be considered when interpreting the findings. First, the retrospective design relies on existing medical records, which limited data completeness and accuracy. In particular, key prognostic factors such as disease stage, tumor histology, and performance status were unavailable due to incomplete documentation. Consequently, we were unable to fully adjust for these variables in the survival analyses, and residual confounding may have influenced the observed associations. Nevertheless, treatment patterns may serve as a partial proxy for disease stage, which could help reduce variability related to unmeasured prognostic factors.

Second, information on several clinical and socio-demographic variables, including socioeconomic status, psychological factors, and detailed measures of treatment adherence, was not consistently available. For some variables, particularly those related to patient experience or behavioral responses, data were obtained from partners rather than directly from patients, which may have introduced reporting bias or inaccuracies.

Third, we acknowledge that the assumption of non-informative censoring may not have been fully satisfied. If censoring was related to unmeasured factors influencing survival, this could have introduced bias into the estimates. However, given the study design and follow-up procedures, we believe this limitation is unlikely to have substantially affected our conclusions.

Finally, baseline demographic and behavioral characteristics of untraced patients were largely unavailable, limiting our ability to formally compare traced and untraced groups. However, because untraceability was primarily due to missing or incorrect contact information rather than clinical or demographic factors, systematic differences between traced and untraced patients are unlikely.

Conclusion and recommendation

This study revealed critically low survival outcomes among cervical cancer patients in Tigray, Northern Ethiopia, with a four year cumulative survival rate of only 3.7% and a median survival time of 19.5 months. These outcomes are substantially lower than those reported in other regions of Ethiopia and comparable international settings, highlighting serious gaps in early detection, timely diagnosis, and effective treatment.

Despite the widespread disruption of healthcare services during the war in Tigray, survival outcomes did not significantly differ between patients diagnosed before and after the conflict. This suggests that structural weaknesses in cervical cancer care such as delayed presentation, limited access to diagnostic tools, and lack of comprehensive treatment options were already present and may have been long-standing challenges in the region.

Survival was significantly influenced by age and treatment modality. Patients who received surgery either alone or in combination with chemotherapy and/or radiotherapy had better survival outcomes compared to those treated with chemotherapy and radiotherapy alone. This supports the importance of early-stage diagnosis, when surgery is still a viable and potentially curative option.

To improve survival rates, strengthening cervical cancer care must be prioritized. This includes expanding screening programs, increasing public awareness, improving diagnostic infrastructure, and ensuring access to comprehensive treatment pathways. Long-term investment in health system capacity is essential to improve cancer outcomes and reduce preventable mortality in Tigray and similar settings.

Abbreviations

LMICs

Low- and middle-income countries

HPV

Human papillomavirus

ACSH

Ayder comprehensive specialized hospital

HEW

Health extension workers

HR

Hazard ratio

Authors’ contributions

**G.B** . Contributed to the development of the Concept, collected and analyzed data, and wrote the draft and final article. **R.A** . & **M.A** . contributed to the development of the Concept, collected data analyzed data, and wrote the draft and final article. **N.J** . contributed to the development of the Concept, collected data, and reviewed the draft and final article. **W.A** . contributed to the development of the Concept, collected data, and reviewed the draft and final article. **D.A** . contributed to the development of the concept, data analysis and reviewed the draft and final article. **H.H** . contributed to the development of the concept, data collection, and data analysis and reviewed the draft and final article. **E.N** . contributed to the development of the concept, data collection, and data analysis and reviewed the draft and final article **. A.Alemu** . contributed to the development of the concept, data collection, and data analysis and reviewed the draft and final article **. A.Abdissa** . contributed to the development of the concept, data collection, and data analysis and reviewed the draft and final article. **G.K.G** . contributed to the development of the concept, and reviewed the draft and final article. **N.M** . contributed to the development of the concept, and reviewed the draft and final article. **G.L.M.** contributed to the data analysis and reviewed the draft and final article and was responsible for the laboratory testing. **P.B.** contributed to the data analysis and reviewed the draft and final article and was responsible for the laboratory testing. **E.J.K.** contributed to the development of the concept, and data analysis and reviewed the draft and final article. **K.M** . contributed to the development of the concept, and data analysis and reviewed the draft and final article **. S.M.G.** contributed to the development of the concept, and data analysis and reviewed the draft and final article and was responsible for the laboratory testing. All authors read and approved the final manuscript.

Funding

This study was conducted with the financial support of London school of hygiene and tropical medicine, medical research council. However, Author(s) received no financial support for authorship, and/or publication of this article.

Data availability

The raw data file could be provided for research purpose up on request via email of the corresponding author ( [gerezgiher.buruh@mu.edu.et](mailto: gerezgiher.buruh@mu.edu.et) ).

Declarations

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.