Clinicopathological Characteristics of Women with Cervical Cancer at Chandaria Cancer Centre, Moi Teaching and Referral Hospital, Kenya ()
1. Introduction
Cervical cancer is a malignant neoplasm that arises from the cervix. Persistent human papillomavirus infection causes more than 99% of all cervical cancers [1]. It presents with two primary histological variants: adenocarcinoma, which arises from the glandular cells lining the endocervix and squamous cell carcinoma, which arises from the squamous cells lining the ectocervix. Squamous cell carcinoma is more common, accounting for approximately 70% of cases. The interface between these cell types, termed the transformation zone, is a focal point for tumor development, with most malignancies originating there [2].
Cervical cancer remains a leading cause of cancer morbidity and mortality among women worldwide, with the greatest burden borne by low- and middle-income countries where organized screening and access to timely treatment are limited [3]. In 2024, approximately 19.9 million new cancer cases were diagnosed, with 9.7 million associated deaths reported worldwide (excluding nonmelanoma skin cancer), highlighting lung and breast cancers as the most prevalent. Cancer incidence and mortality also varied significantly by Human Development Index (HDI) across countries. Cervical cancer was the ninth most common cancer during the same year, with approximately 604,196 new cases [4].
In 2022, nearly 94% of global cervical cancer deaths, claiming nearly 350,000 lives, occurred in low- and middle-income countries [5]. In Africa, cervical cancer ranked as the leading cause of cancer-related deaths in women and was among the top three cancers affecting women below 45 years in 2018 [6]. During that same year, out of the total cases of cervical cancer globally, 21% are in Africa, with 26% of deaths globally associated with cervical cancer [7].
In Kenya, most women present with advanced disease. Mwaliko et al. [8] reported late-stage presentation in 68.5% of patients at a Kenyan facility, which limits curative treatment options. Patient and tumor characteristics at presentation are central to care planning. Sociodemographic factors such as low income, rural residence and lack of health insurance influence whether women can reach and afford treatment [9] [10]. Reproductive and infection-related factors, including high parity and HIV infection, are linked to cervical cancer risk, with women living with HIV at a six- to eight-fold higher risk [11]. Tumor characteristics, including histological type, stage, grade and nodal status, determine appropriate treatment and the prognosis [2].
Chandaria Cancer Centre at Moi Teaching and Referral Hospital is a major referral center for oncology services in western Kenya. A clear picture of this profile is needed to guide screening outreach, patient support and the documentation practices that underpin future research. This paper therefore describes the sociodemographic, patient-specific and pathological characteristics of women with cervical cancer attending Chandaria Cancer Centre, MTRH.
2. Materials and Methods
2.1. Study Design
This study employed a retrospective descriptive medical-record review.
2.2. Study Population
The study population comprised patients with pathologically confirmed cervical cancer diagnosed between 1st January 2015 and 31st December 2019 at Moi Teaching and Referral Hospital. Systematic sampling was used to select the files for review, and 290 files were included in the analysis.
2.3. Inclusion and Exclusion Criteria
2.3.1. Inclusion Criteria
The study included all files of patients with histologically confirmed cervical cancer who were managed at Moi Teaching and Referral Hospital during the study period. Only patients with a primary cancer of the cervix were included.
2.3.2. Exclusion Criteria
Files of patients with missing or illegible essential records (missing more than 50% of key variables) were excluded.
2.4. Sampling Technique
Systematic sampling was used to select the required sample of 290. Google Sheets was utilized to ensure randomness. The sampling frame comprised all eligible cervical cancer patient records at the Chandaria Cancer Centre, Moi Teaching and Referral Hospital. A complete list of all eligible patient records for the period 1st January 2015 to 31st December 2019 was extracted from the hospital registry and entered into a Google Sheet in the same chronological order as they appear in the filing system. In Google Sheets, the patient file list was entered in column A. To ensure a random starting point for selection, random numbers were generated for each file in column B using the formula = RAND(). This randomization step helped to prevent selection bias.
To identify every third file, the formula = IF(MOD(ROW(), 3) = 0, “Select”) was applied in column C, marking every third file with “Select.” The list was then sorted based on the random numbers in column B to further randomize the order. Subsequently, a filter was applied to column C to display only rows marked with “Select”.
Once the sample was identified, the physical records were retrieved. During retrieval, if a selected file was missing or did not meet the inclusion criteria, the next file in the systematic sequence (i.e., the next 3rd file in the list) was used. Files that were excluded according to the predefined exclusion criteria were documented in a log.
2.5. Data Collection
Data were abstracted from patient files using a structured data collection tool in three parts. Part 1 captured sociodemographic characteristics: age, employment status, marital status, place of residence, health insurance coverage, and history of alcohol and tobacco use. Part 2 captured patient-specific characteristics: contraceptive use, parity, personal history of cancer, comorbidities (diabetes mellitus, hypertension and chronic pulmonary disease), HIV status, Hepatitis B status, history of sexually transmitted infection and prior cervical cancer screening. Part 3 captured pathological characteristics: histological type, FIGO stage at diagnosis, tumor grade, lymph node involvement and lymphovascular space invasion.
Data were abstracted from February 2026 to May 2026 by three people: the principal investigator and two research assistants. The tool was piloted before use, and abstracted entries were checked by the second research assistant. Conflicting or unclear entries were referred to the principal investigator, who re-examined the physical file and resolved the entry. Problems identified with the tool during abstraction were corrected.
Place of residence was classified as urban or rural based on the patient’s recorded primary address. Personal history of cancer was defined as any documented prior malignancy, including previous cervical cancer or distinct second primary tumors. Contraceptive exposure included any documented history of contraceptive use prior to diagnosis. Employment status was categorized into three groups—employed, unemployed, or self-employed—as documented on the initial visit form. Parity was defined as the number of pregnancies that have reached viable gestational age.
Histological type was confirmed from the histology report present in the patient file, and tumor grade was extracted directly from the histology report and categorized as unknown when absent. FIGO stage, lymph node involvement and lymphovascular space invasion were taken from clinical notes, imaging, and surgical or pathology reports. Clinical stage was extracted directly from patient medical records as documented by the treating clinical team at the time of initial diagnosis, based on the prevailing FIGO staging system in use at the time of entry. Disease stage was recorded as main broad categories (Stage I, Stage II, Stage III, and Stage IV), where Stages I and II were categorized as early-stage disease and Stages III and IV were categorized as late-stage (advanced) disease.
2.6. Statistical Analysis
Data were entered and analyzed using the Statistical Package for the Social Sciences (SPSS) version 27. Continuous variables were summarized as means and standard deviations, and categorical variables as frequencies and valid percentages. Valid percentages exclude records with missing data for that variable, and the number of patients with data (n) is stated for each variable. Findings are presented in tables and text.
2.7. Ethical Considerations
Ethical approval for this study was granted by the Institutional Research and Ethics Committee (IREC) of Moi Teaching and Referral Hospital (Approval Reference: FAN 5255). Research licensing was obtained from the National Commission for Science, Technology and Innovation (NACOSTI; License No: NACOSTI/P/26/4183045). Given the retrospective design of the study, IREC granted a waiver of informed consent. To safeguard participant confidentiality, all medical records and study data were anonymized using unique identifiers and stored securely to prevent unauthorized access.
3. Results
A total of 290 patients with cervical cancer treated at Chandaria Cancer Centre, Moi Teaching and Referral Hospital, were included. The characteristics are presented in three parts: sociodemographic characteristics, patient-specific characteristics and pathological characteristics of the tumors.
3.1. The Sociodemographic Characteristics
Table 1. Sociodemographic characteristics of patients (n = 290).
Variable |
Category |
n |
Valid % |
Age, years
(n = 290) |
Mean 52.1 (SD 13.4); range 18 - 90 |
290 |
100 |
Age group
(n = 290) |
<50 years |
125 |
43.1 |
≥50 years |
165 |
56.9 |
Employment
(n = 279) |
Employed |
34 |
12.2 |
Self-employed |
68 |
24.4 |
Unemployed |
177 |
63.4 |
Marital status
(n = 289) |
Single |
22 |
7.6 |
Married |
197 |
68.2 |
Divorced |
19 |
6.6 |
Separated |
1 |
0.3 |
Widowed |
50 |
17.3 |
Residence
(n = 287) |
Urban |
28 |
9.8 |
Rural |
259 |
90.2 |
Health insurance
(n = 281) |
Insured |
140 |
49.8 |
Uninsured |
141 |
50.2 |
The mean age of the patients was 52.3 years (SD = 13.4), with a range of 18 to 90 years. Most patients were unemployed (177, 63.4%), while 68 (24.4%) were self-employed and 34 (12.2%) were employed. Two-thirds of patients were married (197, 68.2%). Fifty patients (17.3%) were widowed, 22 (7.6%) were single, 19 (6.6%) were divorced, and one (0.3%) was separated. The overwhelming majority of patients lived in rural areas (259, 90.2%), compared with 28 (9.8%) in urban areas. Health insurance coverage was evenly split, with 140 patients (49.8%) insured and 141 (50.2%) uninsured. A history of alcohol use was recorded in 13 patients (4.5%) and smoking in 9 (3.1%) (Table 1).
3.2. The Patient-Specific Characteristics
Table 2. Patient-specific characteristics (n = 290).
Variable |
Category |
n |
Valid % |
Contraceptive use (n = 288) |
Yes |
79 |
27.4 |
Parity (n = 277) |
0 |
8 |
2.9 |
1 |
10 |
3.6 |
2 |
19 |
6.9 |
3 |
40 |
14.4 |
4 |
31 |
11.2 |
>4 |
169 |
61 |
Diabetes mellitus (n = 289) |
Yes |
8 |
2.8 |
Hypertension (n = 289) |
Yes |
45 |
15.6 |
Chronic pulmonary disease (n = 288) |
Yes |
4 |
1.4 |
HIV status (n = 289) |
Positive |
66 |
22.8 |
Negative |
182 |
63.0 |
Unknown |
41 |
14.2 |
Hepatitis B status (n = 287) |
Positive |
1 |
0.3 |
Negative |
34 |
11.8 |
Unknown |
252 |
87.8 |
History of STI (n = 289) |
Yes |
1 |
0.3 |
No |
96 |
33.2 |
Unknown |
192 |
66.4 |
Prior cervical cancer screening |
Yes |
3 |
- |
Contraceptive use was reported by 79 patients (27.4%). High parity was common: 169 patients (61.0%) had more than four children, 71 (25.6%) had three to four, 29 (10.5%) had one to two, and only 8 (2.9%) had no children. In total, 86.6% of patients had given birth to three or more children. A personal history of cancer was recorded in 27 patients (9.3%). Comorbid conditions were uncommon, except for hypertension, which was present in 45 patients (15.6%). Diabetes mellitus was recorded in 8 patients (2.8%) and chronic pulmonary disease in 4 (1.4%). Sixty-six patients (22.8%) were HIV positive, 182 (63.0%) were HIV-negative, and the status of 41 (14.2%) was unknown. Hepatitis B status was unknown for 252 patients (87.8%), negative in 34 (11.8%) and positive in only 1 (0.3%). A history of sexually transmitted infection was unknown for 192 patients (66.4%), absent in 96 (33.2%) and present in only 1 (0.3%). Prior cervical cancer screening was documented in only 3 patients (1.0%), so 99.0% of patients had no documented screening before diagnosis (Table 2).
3.3. The Pathological Characteristics
Table 3. Pathological characteristics of the tumours (n = 290).
Variable |
Category |
n |
Valid % |
Histological type (n = 290) |
Squamous cell carcinoma |
269 |
92.8 |
Adenocarcinoma |
20 |
6.9 |
Adenosquamous carcinoma |
1 |
0.3 |
FIGO stage at diagnosis
(n = 286) |
Stage I |
46 |
16.1 |
Stage II |
93 |
32.5 |
Stage III |
122 |
42.7 |
Stage IV |
25 |
8.7 |
Stage I - II |
139 |
48.6 |
Stage III - IV |
147 |
51.4 |
Tumour grade (n = 287) |
Well differentiated |
23 |
8.0 |
Moderately differentiated |
161 |
56.1 |
Poorly differentiated |
73 |
25.4 |
Not known |
30 |
10.5 |
Lymph node involvement |
Yes |
26 |
- |
Lymphovascular space invasion |
Yes |
15 |
- |
Squamous cell carcinoma was the predominant type, found in 268 of 289 patients (92.7%). Adenocarcinoma was recorded in 20 patients (6.9%), and a single case of adenosquamous carcinoma (0.3%) was identified. In total, 21 patients (7.3%) had a non-squamous histology. Stage III was the most common (122 patients, 42.7%), followed by Stage II (93, 32.5%), Stage I (46, 16.1%) and Stage IV (25, 8.7%). Grouped by extent of disease, 139 patients (48.6%) presented with early-stage disease (Stages I - II) and 147 (51.4%) with advanced-stage disease (Stages III - IV) (Table 3).
Most tumors were moderately differentiated (161, 56.1%), followed by poorly differentiated (73, 25.4%) and well differentiated (23, 8.0%). Grade was not known for 30 patients (10.5%). Lymph node involvement was documented in 26 of 287 patients, and lymphovascular space invasion (LVSI) in 15 of 286. Nodal status and LVSI depend on imaging and on surgical or pathological assessment, which were available for only some patients, so these figures may underestimate their true occurrence.
4. Discussion
4.1. Sociodemographic Characteristics
The mean age of patients in this study (52.1 ± 13.4 years) was higher than the 49.5 years reported in Ethiopia by Wassie et al. [12] and close to the older end of the 45 - 64-year range that dominated the Nigerian cohort described by Alegbeleye et al. [13]. This older age profile is consistent with the natural history of cervical cancer, which usually presents after several decades of persistent HPV infection, and it suggests that opportunities for detection through screening at younger ages are being missed locally. The wide age range (18 - 90 years) also shows that the disease affects women across the reproductive and post-reproductive years.
The cohort was predominantly unemployed (63.4%), married (68.2%), and living in rural areas (90.2%). The high unemployment rate is comparable to the 85.6% reported in Ethiopia by Wassie et al. [12]. It reflects the structural gradient linking lower socioeconomic status with heightened vulnerability to cervical cancer [14]. Recent regional evidence from multi-country demographic analyses across sub-Saharan Africa confirms that cervical cancer screening and preventative access are deeply entrenched in pro-rich and pro-education inequalities, leaving vulnerable, low-income populations disproportionately unreached [15]. Furthermore, in post-screening contexts, Isaacson et al. [16] demonstrated that low socioeconomic standing directly compounds loss to follow-up and creates severe structural barriers to initiating definitive treatment among HPV-positive women. The high proportion of rural residents is consistent with the resource-limited catchment area served by MTRH, where geographical distance and travel costs present major structural barriers to timely diagnosis and care [10].
Health insurance coverage was almost evenly split between insured (49.8%) and uninsured (50.2%) patients. That half of the patients had no insurance is important in a setting where the cost of cancer care has been shown to push households into catastrophic expenditure [9]. The low prevalence of alcohol use (4.5%) and smoking (3.1%) may partly reflect under-recording in clinical files, since these behaviors are often poorly documented.
4.2. Patient-Specific Characteristics
High parity defined this cohort. Nulliparous patients constituted 2.9% (n = 8), while those with 1, 2, 3, and 4 children accounted for 3.6% (n = 10), 6.9% (n = 19), 14.4% (n = 40), and 11.2% (n = 31) of the population, respectively. Combining these categories reveals that 25.6% (n = 71) had 3 - 4 children, 61.0% (n = 169) had >4 children, and a cumulative 86.6% (n = 240) had 3 children (or 72.2% with >3 children). Biologically, high parity acts as a documented cofactor in HPV-mediated carcinogenesis through recurrent mechanical micro-trauma to the cervical transformation zone during childbirth alongside prolonged exposure to elevated gestational steroid hormones, which upregulate viral oncogene transcription. Comparison with Alegbeleye et al. [13] [Parity 3 - 4]: Aligning identical parity thresholds reveals that our cohort’s parity 3 - 4 rate of 25.6% (n = 71) is nearly double the baseline demographic rate of 14.1% reported in Nigeria.
When evaluating multiparity against regional literature using harmonized thresholds, our cohort demonstrated a consistently high reproductive burden across comparative benchmarks. Aligning our data to the >3-children threshold used by Wassie et al. [12] in Ethiopia yields a cumulative prevalence of 72.2% (n = 200/277), confirming a sustained, elevated rate of multiparity across both East African oncology populations. Furthermore, standardizing grand multiparity to >4 children reveals that our study’s rate of 61.0% (n = 169/277) exceeds the sub-Saharan African regional baseline of 38.4% reported in multi-country DHS survey data by Hailegebireal et al. [17].
HIV prevalence in this cohort was 22.8%, aligning closely with the 29.0% reported in a similar East African oncology cohort in Rwanda by Park et al. [18] and exceeding the 18.4% reported by Wassie et al. [12]. This figure is substantially higher than the general population prevalence in Kenya, reflecting the well-documented six- to eight-fold increased risk of invasive cervical cancer among women living with HIV [11]. An additional 14.2% of patients had unknown HIV status, suggesting that the true overall burden may be higher. Among other documented comorbidities, hypertension was most common (15.6%), while diabetes mellitus (2.8%) and chronic pulmonary disease (1.4%) were less common. A personal history of cancer was recorded in 9.3% of patients.
Documented prior cervical cancer screening was strikingly low (1.0%), even lower than the 11.5% screening uptake reported among Nigerian patients by Alegbeleye et al. [13], where 88.5% had never been screened. Documentation of infection-related history was also poor: Hepatitis B status was unknown for 87.8% of patients and STI history for 66.4%. These gaps reflect incomplete recording in clinical files rather than a true low prevalence, and they limit what can be said about these exposures in this cohort.
4.3. Pathological Characteristics
Squamous cell carcinoma was the predominant histological subtype (92.7%). This proportion is higher than the 88.1% reported in India by Jain et al. [19] and the 71.8% reported in Nigeria by Alegbeleye et al. [13], but consistent with the global pattern in which squamous cell carcinoma accounts for the large majority of cervical cancers [2]. Adenocarcinoma accounted for 6.9% of tumors, and the small number of non-squamous cases (n = 21) limits any meaningful description of these subtypes.
Just over half of patients (51.4%) presented with advanced-stage disease (FIGO Stages III - IV), broadly comparable to the 51.90% reported in India by Jain et al. [19] but considerably lower than the over 90% late-stage presentation reported nationally for Kenyan facilities by Mwaliko et al. [8]. In either case, more than half of patients presenting with locally advanced or metastatic disease are with the near-absence of documented prior screening in this cohort.
Most tumors were moderately differentiated (56.1%), and one in four was poorly differentiated (25.4%), with grade unknown for 10.5% of patients, which points to incomplete pathology documentation. Lymph node involvement (9.1%) and lymphovascular space invasion (5.2%) were uncommon. These figures are likely to be underestimated because nodal status and LVSI depend on imaging and on surgical or pathological assessment, which were available for only some patients.
4.4. Limitations
It is a single-center review at a referral hospital, so the patients may not represent women with cervical cancer in the wider population, and women who never reached the center are not captured.
Files with missing or illegible records were excluded, and records were sampled in filing order, so selection bias is possible; the diagnosis dates in the sample are concentrated in 2018-2019, and some fall outside the stated study period.
Information was abstracted from clinical files, so documentation bias is likely: unrecorded exposures cannot be distinguished from absent ones for several variables, and unknown categories were large for Hepatitis B, STI history and HIV.
Pathological assessment was unequal across patients, which affects nodal involvement, LVSI and grade, and pathology reports were not independently reviewed. FIGO staging may have been recorded under different editions across the study period, substages were not recorded, and the historical study period may not reflect current practice.
5. Conclusions and Recommendations
5.1. Conclusions
The findings of this study reveal a predominantly older population, married, lives in rural areas and unemployed, with significant socioeconomic vulnerability, as half of the patients were uninsured. The high parity, the notable burden of HIV positivity and the very limited documented history of prior screening reflect the complex clinical and reproductive profile of this population. The predominance of squamous cell carcinoma, the moderately differentiated grade of most tumors, and the presentation of just over half of the patients with recorded stage at Stages III - IV underscore the need for interventions that promote early detection, expand cervical cancer screening, support uninsured and rural patients, and strengthen clinical and pathological documentation within the health system. These findings are partly in keeping with patterns reported in other sub-Saharan African cohorts. As a descriptive record review, however, this study cannot establish the causes of late presentation, and it underscores the need for targeted interventions and prospective studies to improve outcomes for women with cervical cancer in Kenya.
5.2. Recommendations
Link uninsured patients to social support and insurance-enrollment services at diagnosis, as half of the patients had no health insurance.
Offer routine HIV testing and integrate HIV and cancer care, given the high HIV prevalence.
Strengthen medical records and pathology reporting so that Hepatitis B status, STI history, tumor grade, nodal status and lymphovascular space invasion are consistently recorded.
A prospective, multi-center study with standardized data collection would give a more complete and generalizable profile of women with cervical cancer in Kenya.
Future studies should examine how these patient and tumor characteristics relate to treatment outcomes, using time-to-event methods with active follow-up.
Author Contributions
Chelangat Priscah—Conceptualization, Data curation, Formal Analysis, Investigation, Methodology, Writing—original draft. Tracy Irura: Supervision, Writing—review. George Muthui: Supervision, Validation, Investigation, Resources, Writing—review & editing.