Late Diagnosis among patients with Prostate Cancer at the Uganda Cancer Institute: A retrospective cohort study

Background Late diagnosis is associated with high mortality, morbidity and low quality of life. We estimated time taken from perception of symptoms attributable to prostate cancer to biopsy among patients with prostate cancer at the Uganda Cancer Institute (UCI) and the associated factors. Methods We conducted a retrospective cohort analysis of records of 280 patients with histologically confirmed diagnosis of prostate cancer at UCI from January 2016 to December 2017. Time to diagnosis was obtained from the difference between the approximate date of onset of initial symptoms and date when a biopsy was taken. Late diagnosis was that when an individual was diagnosed with prostate cancer stage III or IV whereas stages I and II were classified as early. We used modified poisson regression to assess factors associated with timing of diagnosis among patients.


Abstract
Background Late diagnosis of prostate cancer is common in Uganda and elsewhere. Diagnosis in advanced stages is associated with high mortality, morbidity and low quality of life. We estimated time taken from perception of symptoms attributable to prostate cancer to biopsy among patients with prostate cancer at the Uganda Cancer Institute (UCI) and the associated factors.

Methods
We conducted a retrospective cohort analysis of records of 280 patients with histologically confirmed diagnosis of prostate cancer at UCI from January 2016 to December 2017. Time to diagnosis was obtained from the difference between the approximate date of onset of initial symptoms and date when a biopsy was taken. Late diagnosis was that when an individual was diagnosed with prostate cancer stage III or IV whereas stages I and II were classified as early. We used modified poisson regression to assess factors associated with timing of diagnosis among patients.

Results
The median time from first perceived symptoms to biopsy for prostate cancer patients was 12 (IQR5-24) months and 76% were diagnosed after 4 months of symptoms. Median age at time of diagnosis of patients was 70 (IQR66-74.5) years and at least 50% were aged between 65-74 years. About 81.8% of the patients were diagnosed late; of which 35.7% were in stage III and 46.1% were in stage IV. Nearly all patients presented with raised prostate specific antigen with median prostate specific antigen of 100.2 (IQR36.02-350) ng/ml of blood at the time of admission. In adjusted analysis, a patients whose biopsies were taken before 5 months of recognising symptoms were two times as likely to have cancer stage I and II compared to those patients in whom the biopsies were taken after 4 months.

Conclusion
More than three in four patients were diagnosed late. Taking a biopsy after 4 months of initiation of symptoms was partially responsible for the delay. To improve time to diagnosis, communities should be educated about symptoms of prostate cancer and advised to seek health care early. Health care workers should be sensitised to suspect prostate cancer among patients to allow timely referral for appropriate specialised assessment and management.

Background
Prostate cancer is one of the big public health problems globally and is the third commonest cancer diagnosed worldwide responsible for 7.1% of all cancer deaths (1). Prostate cancer is the second most diagnosed malignancy and the second-leading cause of cancer-related deaths among men globally (2). The burden of prostate cancer is relatively higher in sub-Saharan Africa where it accounts for 20.3% of all cases of cancer in men with the highest incidence rate reported in Zimbabwe and Uganda (3,4).
In Uganda prostate cancer has been increasing at a rate of 5.2% annually, making it the most rapidly increasing cancer in the country and in sub-Saharan Africa (5). Despite the increase in incidence, approximately 90% of prostate cancer patients are not aware of the disease and as a result they do not take early urinary symptoms seriously hence they are diagnosed late with very advanced disease in stage IV presenting with incurable tumors (5-7). Late diagnosis limits treatment options, increases mortality and leads to low quality of life for patients and their families (8,9). The factors associated with late diagnosis and consequently late treatment have not been well established.
Previous studies in Uganda found the median age of prostate cancer diagnosis to be 70 years among Ugandan men with majority dying within the first year of diagnosis and only 46.9% living 5 years after diagnosis (10). Early diagnosis of prostate cancer may lead to timely management of the disease which can lead to better treatment outcomes. We conducted this study to estimate the timing of diagnosis among prostate cancer patients in Uganda and the associated factors.

Aim, design and setting
We conducted this study to generate information on the timing of diagnosis and associated factors among prostate cancer patients in Uganda. We hope that this study will provide evidence for health care providers and policy makers on the need to design programs that will improve the timing of diagnosis of prostate cancer. This was a retrospective cohort study in which we quantitatively analyzed patient records of men with a histologically confirmed diagnosis of prostate cancer at UCI between January 2016 and December 2017.
We conducted the study at the Uganda Cancer Institute (UCI). UCI is a public specialized tertiary care Specialized treatment for prostate cancer in Uganda is only offered at UCI. All suspected and confirmed prostate cancer patients in the country are supposed to be referred to UCI for management. The guidelines for diagnosis and management of patients with suspected cancer of the prostate cover both symptomatic and asymptomatic cases. The guidelines for patients with symptoms state that prostate cancer should be considered in any male above 50 years of age and or above 40 years of age if he has a first degree relative who has or had prostate cancer and or is of African ethnicity. For asymptomatic men seeking screening, the guidelines recommend counselling to undertake a prostate specific antigen (PSA) test and digital rectal exam (DRE) if PSA is raised. Referral to UCI should be considered after two abnormal PSA at least 6 weeks apart and or abnormal hard prostate on DRE (11).

Description of Materials
We abstracted data from records of all newly diagnosed prostate cancer patients registered at UCI from January 2016 to December 2017. These records included; patient files, registers, summary sheets, doctors' medical notes and referral notes. The data was abstracted using a data abstraction tool. All patients reported symptoms at the time of diagnosis which included frequent micturition, weak flow of urine, urge to urinate frequently at night, blood in the urine, erectile dysfunction and burning sensation during urination. We excluded 15 patient records which did not indicate the date of onset of the initial perceived symptoms and date when diagnosis was made.

Study variables
The primary outcome variable was stage of prostate cancer at the time of diagnosis. Stages 1 and 2 were classified as early, whereas stages 3 and 4 were classified as late. This classification was based on prostate cancer staging by American cancer society (12). Independent variables included time taken to diagnosis, age, family history of prostate cancer, education level, occupation, ethnicity, marital status, religion, co-morbidities and presenting symptoms. The time to diagnosis was obtained from the difference between the recorded estimated time of onset of initial symptoms and the time when a biopsy was done.

Data management
The data were collected, cleaned and entered into Microsoft Excel 2016. Continuous numerical responses were entered as absolute values while categorical responses were coded. The data was stored in a confidential manner in a password-protected computer.

Data analysis
Data were analyzed using STATA version 14. All continuous variables were summarized using medians with interquartile ranges while categorical data were recorded as proportions with percentages. We used Pearson's Chi-square test to examine the associations between independent variables and timing of diagnosis among prostate cancer patients. Modified Poisson regression with robust variances was used at bivariable and multivariable analysis to identify factors associated with timing of diagnosis among prostate cancer patients at UCI. Prevalence ratios (PRs) were used to estimate the strength of association between the outcome and indicator variables and associations were tested at a 95% confidence interval (CI).

Results
The median time to diagnosis for the prostate cancer patients at UCI between January 2016 and December 2017 was 12 (IQR5-24) months. Only 24 % of the patients were diagnosed with in the first four months of the symptoms perception. Their median time to admission at UCI from the initiation of symptoms was 14 (IQR 6-24) months. More than three quarters (81.8%) of the patients were diagnosed late, of which 35.7% were in stage III and 46.1% were in stage IV (table 1).  At bivariable analysis, only the time taken to request biopsy from the time first symptoms appeared was significantly associated with timing of diagnosis with P-value 0.012.

Discussion
We found that the median time to diagnosis for prostate cancer patients was 12 months from the time the patients recognized symptoms that may have been attributable to cancer of the prostate. Most of the patients were diagnosed late with metastatic cancer and very high levels of prostate specific antigen (PSA). Patients who had biopsies within 4 months of initiation of symptoms were two times likely to be diagnosed early compared to patients who had biopsies after 4 months. This is comparable to other studies which found that diagnosis after 3 months from the time of recognition of symptoms was associated with late stage at diagnosis and poor prognosis (13,14).
Late diagnosis could have been due to patients' delay in seeking health care for the symptoms, and or due to health systems delays in requesting and performing of necessary investigations including prostate biopsies. The possible causes of patient delay include lack of knowledge regarding importance of symptoms, trivialization of symptoms, financial and economic barriers, poor health seeking behavior for men, lack of trust in health systems and seeking care from non-professionals such as traditional healers, spiritual healers and Chinese herbalists (15,16). On the other hand, health system delays may be attributable to inadequate number of health care workers able to attend to patients, lack of knowledge and skills to assess patients, limited diagnostic equipment as well as lack of supplies, delayed referrals, and health workers not considering cancer of prostate in their differential diagnosis. Related studies in low and middle income countries also found an association between delays and late diagnosis among cancer patients (16,17). Such late diagnosis results into poor prognosis and death (18).
Related studies in Uganda and other developing countries have found other possible causes of patient and health system delays to include patients non-adherence to the recommended investigations, referral complexity as cancer investigations are done from different locations and lack of large scale population-based prostate cancer screening (5,10,19). We further found that most patients were lost to follow up while others had succumbed to prostate cancer within two years. The high loss to followup may be related to limited capacity of UCI to provide adequate care and follow-up due to limited staffing levels (20) . This implies that the health system in Uganda and the UCI need to do more follow up of patients after confirmatory diagnosis of prostate cancer to improve prognosis of prostate cancer patients.

Strength and limitations
The major strength of this study was that it was conducted in a national referral facility which is the only center that provides cancer treatment in Uganda, this helped us to capture data from all the patients across Uganda.
This study neither differentiated between patient and health system delays nor analysed the reasons for the delays. Future studies are needed to differentiate between patient and health system delays and to elucidate the reasons for these delays Patient files were identified using the prostate cancer patient registration numbers and no personal identifiers were used. We handled all the data that we collected and the patient files with confidentiality.