International Journal of Public Health and Epidemiology Research

Vol. 5(2), pp. 108-118, July, 2019. © www.premierpublishers.org. ISSN: 1406-089X

Research Article

Survival Analysis of Determinants of Breast Cancer

Patients at Hossana Queen Elleni Mohammad Memorial
Referral Hospital, South Ethiopia: Bayesian Application
of Hypertabastic Proportional Hazards Model
*Getachew Tekle1, Zeleke Dutamo2
1Wachemo University, Faculty of Natural & Computational Sciences, Department of Statistics, Hossana, Ethiopia.
2Wachemo University, Faculty of Health & Medicine Sciences, Department of Public Health, Hossana, Ethiopia.

Breast cancer is one of the most severe diseases in the world and become the public’s ever day’s
agenda in both developed and developing countries. The primary goal of this study was to
identify the determinants of survival time of breast cancer patients at Hossana hospital, south
Ethiopia. Kaplan-Meier estimation method and a new two-parameter probability distribution called
hypertabastic are introduced to model the survival time of the data. A simulation study was
carried out to evaluate the performance of the hypertabastic distribution in comparison with
popular distribution with the help of R and SAS statistical software Packages. One-fourth (25%)
of the total patients survived for only 2 days. 31(35.2%) were censored, and 55(62.5%) were died.
Hypertabastic survival model was found to be best fitting to the breast cancer data and age, level
of education, family history, breast problem before, High fat diet, child late age, early menarche,
late menopause were significant risk factors for the death of breast cancer patients. Awareness
has to be given for the society on causes of breast cancer and screening test and early detection
policies for most risky groups has to be established.

Key words: Survival Data Analysis; Hypertabastic Survival; Proportional hazard; Breast cancer; survival time

INTRODUCTION

Cancer is an emerging public health problem in Ethiopia, characteristics, clinical profile, and problems related to
with breast and cervical cancers accounting for over half early diagnosis and treatment of breast cancer in 72 (62
of all newly diagnosed cancers in womenGebremariam A, females and 10 males) Ethiopian patients, the female to
et al (2019). male ratio being 6.2: 1, and the females in this series
developed breast cancer at a younger age (72% were
Ersumo (2006) reported that breast cancer as one of the premenopausal) and 76% had advanced disease (stages
most severe diseases in the world and become the public’s III and IV) at presentation [Gebremedhin and Shamebo,
every day’s agenda in both developed and developing 1998].
countries.

South Africa’s national Cancer Registry (NCR) revealed

that breast cancer is common in women being the primary *Corresponding Author: Getachew Tekle, Wachemo
cancer among them. It shows that there is a change in an University, Faculty of Natural & Computational Sciences,
incidence rate in age and variation in race group [Wadler Department of Statistics, Hossana, Ethiopia. E-mail:
et al., 2011]. [email protected]; Tel: +251920405199 Fax:
+251465551930
A prospective study conducted at Tikur Anbessa was Co-Author Email: [email protected]; Tel:
designed to obtain information on demographic +251910489741; Fax: +251465551930;

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model
Tekle and Dutamo 109

As per the many literatures, risk factors for breast cancer The primary goal of this study was to identify the
can be separated into avoidable and non- avoidable. determinants of survival time of breast cancer patients at
Breast cancer, like other forms of cancer, can result from HQEMMH, south Ethiopia.
multiple environmental and hereditary risk factors. The
term "environmental", as used by cancer researchers,
means any risk factor that is not genetically inherited METHODOLOGY
[https://en.wikipedia.org/wiki/Risk_factors_for_breast_can
ce]. Description of Study Area and Period

Howlader (2012) put breast cancer asa multifaceted Hadiya zone is one of 13 zones in SNNPR. There are 10
disease that affects women and men of all ages and ethnic woredas and one town administration in Hadiya zone.
groups. Despite decades of productive research on breast Hosanna town is administrative center for Hadiya zone
cancer diagnosis and treatment, preventing this cancer is and which far from capital city of the country by 235 kms.
the only way to reduce the human toll of this disease that The study was conducted in HQEMMH, South Ethiopia
affects 1 in 8 women in their lifetime. from September, 2011 to February, 2017 (range of data
time).
In the Cox model, the baseline hazard function is regarded
as a nuisance parameter, while in parametric models, the Sample size, Sampling Design and Technique
hazard function reflects the time course of the process
under study. In this study, we introduce a new two- A total of 86 random breast cancer patients were
parameter continuous probability distribution called considered for the study. For this study retrospective study
hypertabastic probability distribution. The hypertabastic design was employed. In order to select a fairly
hazard function can assume a different variety of shapes. representative sample of the population for this study,
Some studies tried to identify the epidemiological risk simple random sampling technique was employed.
factors, but the cause of any individual breast cancer is not Table 1: Covariates Considered in the Study
known. As epidemiological researches inform, though not Covariates Codes
informed of the incidence in individual populations the 1. Gender of patients 1=female, 0=Male
patterns of the breast cancer incidence a cross certain 2. Age of patients 0<=18yrs,
population clearly revealed. In relation to this, hereditary 1= 19-25yrs,
syndrome contributed 5% of new breast cancer while 30% 2= >26yrs
of the risk factors are accountable to well-established 3. Breast ca.stage 0=stgI,
factors [Madigan et al., 1995]. 1= stgII,
2=stgIII,
Majority of breast cancers have not been with the exact 3=stgIV
cause(s) so that major challenge for prevention is to 4. Family history 0=no, 1= yes
identify women at risk as precisely as possible and then to 5. Has breast problem 0=no, 1= yes
apply measures such as chemoprevention and lifestyle 6. High fat diet 0=no, 1= yes
changes. Current models can predict probable numbers of 7. Residence of patients 0=rural, 1= Urban
breast cancer cases in specific risk factor strata, but have 8. Marital status 0= separated/divorced/widowed,
modest discriminatory accuracy at the individual level 1= never married,
[Amir et al., 2010]. 2=married
9. Level of education 0=secondary and above
This study attempts to identify factors that have strong 1= no education
associations with the survival experience of breast cancer 2= primary
patients under treatment in one of the government 10. Religion 0= Muslim
hospitals in the regional state of SNNPR at Hossana 1=orthodox
Queen Elleni Mohammed Referral Hospital. 2=protestant,
Hence, the current study is supposed to answer the 3= others
following basic questions: 11. Smoking 0= no, 1= yes
1. What the potential risk factors for breast cancer are 12. First child at delayed 0= no, 1= yes
indentified? age
2. What is the distribution of breast cancer among some 13. Premature start of 0= no, 1= yes
covariates? menarche
3. What is the instant ahead of which 50% of the 14. Late menopause 0= no, 1= yes
individual’s breast cancer patients are likely to 15. Stress 0= no, 1= yes
survive? 16. Type of treatment 0= surgery, 1= chemotherapy,
4. Which model better fits the data? 2=Radiation therapy,
3=Hormone therapy

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model
 Int. J. Public Health Epidemiol. Res. 110

Survival Analysis (SA) Proportional Hazards Model (PHM)

SA consists of a set of specialized statistical techniques The basic model for survival data to be considered in this
used to study time-to- event data. In analyzing such data, study is the PHM. It is anticipated by David Cox (1972) and
the main objective is to determine the length of time has also come to be known as the Cox regression model.
interval for the occurrence of an event. Survival analysis is His paper took a special approach to standard parametric
mainly used for two distinguishing features of time-to- SA and extended the methods of the parameter free KM
event data. Duration times are non-negative values usually parameter coefficient values to regression type point of
exhibiting highly skewed distribution and therefore the view for life-table analyses. Cox advanced to prediction of
assumption of normality is violated. Secondly, censoring survival time in individual subjects by only utilizing
may occur or the true duration is not always observed or variables covering with survival and ignoring the baseline
known, that is, some subjects are potentially being hazard of individuals. Cox assumed only that the hazard
unobserved for the full time to failure. functions of different individuals remained proportional and
constant over time and he made no assumptions about the
The core characteristic of time-to-event data is the baseline hazard of individuals.
existence of repress which occurs when the periods of time
to event incidence of some individuals cannot be The proportional hazards (PH) assumption refers to the
completely observed. The process of censoring and fact that the hazard functions are multiplicatively related.
truncation make these data unsuitable to analyze with That is, for any two individuals with covariates Xi and Xj
traditional regression method and hence, the appropriate the ratio h(t|Xi)/h(t|Xj) is assumed to be constant over
technique is SA. Details on various estimation methods survival time.
developed in SDA that taken censoring and truncation in
to account can be obtained in [Hosmer and Lemeshow The Hazard Function
1998].
It gives a turn of phrase for the chance at time t for an entity
In this study the CPHM was used to examine survival time with a specified measurement of a set of covariates
of breast cancer patient. Kaplan-Meier (KM) estimators denoted by X and it is usually put as follows:
were applied to estimate survival curves of breast cancer h(t , X i , β) = h o (t)exp(β' X i )
patient and the log rank test was used for the comparison (2)
between the covariate categories. As usual we start our in which way ho(t) indicates the baseline hazard function
scheme by generous the connotation of censoring, KM which is found while all X's are set to zero, X i is the vector
and CPHM; we then proceed to model building and of values of the covariates for the ith individual at time t and
assessments. β is the vector of unknown regression parameters that are
Kaplan-Meier Estimation assumed to be the same for all individuals in the study,
which measures the influence of the covariate on the
Kaplan-Meier Estimation is a product limit estimation of the survival experience.
survivorship function which is developed by Kaplan-Meier
(1958). Kaplan-Meier (KM) estimator is used by most So, it can likewise be regarded as linear model, as a linear
software packages because of the simplistic step combination of the covariates for the logarithm
approach. The KM estimator incorporates information from transformation of the HR put as:
all of the observations available, both censored and  h(t , X, β) 
=β X
'
log 
uncensored, by considering any point in time as a series  h0 (t )  (3)
of steps defined by the observed survival and censored The CHF is put as:
times. When there is no censoring, the estimator is simply
the sample proportion of observations with event times H (t ) = H O (t ) exp(β ' X)
greater than t. The technique becomes more complicated  h(t , x,  )
but still manageable when censored times are included. From model log   =  ' x , we acquired the
The KM estimator consists of the result of a number of  h0 (t ) 
conditional probabilities resulting in an estimated survival survivor function shown as:
function in the form of a walk function. It is a parameter
S (t , X, β) = S O (t )
exp(β'X)
free estimator of the S(t) , which a survivor function. ,
Where So (t) is a baseline survival function.
dj
Ŝ (t) =  (1 − )
t j t nj Hypertabastic Proportional Hazards Model
(1)
Where djis the number of individuals who experienced the Tabatabai et al (2007) proposed a new probability
event at time tj, and nj is the number of individuals who distribution, hypertabastic distribution, and hypertabastic
have not yet experienced the event at that time. survival model. Let T be a continuous random variable

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model
Tekle and Dutamo 111

𝑛
representing the waiting time until the occurrence of an
∏(𝛽/𝜎 2 , 𝑡, 𝑥) ∝ ∏[𝑓(𝑡𝑖 /𝑥𝑖 , 𝜃)𝐼(𝛿𝑖=0) ∗ 𝑆(𝑡𝑖 /𝑥𝑖 , 𝜃)𝐼(𝛿𝑖=1) ] ∗ ∏(𝛽/𝜎 2 )
event. The hypertabastic baseline survival function is
𝑖=1
defined by (cited by Hong Li, 2017) ∏(𝜎 2 /𝛽𝑡𝑥) ∝ ∏𝑛𝑖=1[𝑓(𝑡𝑖 /𝑥𝑖 , 𝜃)𝐼(𝛿𝑖=0) ∗ 𝑆(𝑡𝑖 /𝑥𝑖 , 𝜃)𝐼(𝛿𝑖=1) ] ∗
𝑆0 (𝑡) = 𝑃(𝑇 > 𝑡) = 𝑆𝑒𝑐ℎ[𝛼𝑊(𝑡)] (4) ∏(𝛽/𝜎 2 ) ∗ ∏(𝜎 2 ) (10)

Where W (t) =𝛼(1 − 𝑡 𝛽 𝐶𝑜𝑡ℎ(𝑡 𝛽 ))/𝛽 , αand β are the model The posterior distribution for the model specification above
parameters and both positive. Correspondingly, the does not have closed form solution for the parameters. For
hypertabastic baseline hazard function is given by these models, MCMC-Gibbs sampler is implemented
ℎ0 (𝑡) = 𝛼 (𝑡 2𝛽−1 𝐶𝑠𝑐ℎ2 (𝑡 𝛽 ) − 𝑡 𝛽−1 𝐶𝑜𝑡ℎ(𝑡 𝛽 )) 𝑇𝑎𝑛ℎ[𝑊(𝑡)] (5) using the SAS software’s (Asrat Demeke, 2015).

Under the proportional hazards assumption, the above Model Building Strategies for proportional Hazard
authors introduced the hypertabastic proportional hazards model
model. The hazard function for this model is given by (cited
by Hong Li, 2017) Having a basic plan to follow in selecting the covariates for
the model and assessing the adequacy of the model both
ℎ(𝑡/𝑋, 𝜃) = ℎ0 (𝑡)𝑔(𝑋/𝜃) (6) in terms of the individual covariates and from the point of
where X is a p-dimensional vectors of covariates, θ is a view of the overall fit of the model is required for achieving
vector of unknown parameters, g ( X θ ) is non negative this “best” model. It is also highlighted in Hosmer and
function of X satisfying the condition that g (0/θ ) = 1, and Lemeshow (1998) that successful modeling of a complex
𝑝 data set is part science, part statistical methods, and part
g(X/ 𝜃 )=exp[- ∑𝑘=1 𝜃𝑘 𝑋𝑘 ]. Similarly, the hypertabastic
survival function for this model is defined as experience and common sense.

𝑆(𝑡/𝑋, 𝜃) = [𝑆0 (𝑡)]𝑔(𝑋/𝜃) (7) Here the model to buildstarts from single covariate
analysis as recommended by Collet (1994). Collet
All the unknown parameters, including X and θ, can be
suggested the move toward of first doing a single variable
estimated using the maximum likelihood method. If the
analysis to “screen" out potentially significant covariates
sample consists of only right censored data, the
for consideration in the multi variate model in order to
hypertabastic proportional hazards log-likelihood function
recognize the significance of each covariate. All covariates
with log time can be written as
𝛽 𝛽 those are noteworthy at 25% level, the humble level of
𝐿𝐿(𝜃, 𝛼, 𝛽: 𝑥) = ∑𝑛𝑖=1 ln⁡[𝛼(1 − 𝑡𝑖 𝐶𝑜𝑡ℎ(𝑡𝑖 ))/𝛽]𝑔(𝑋𝑖 /𝜃) + significance from one predictor single covariate regression
−1+2𝛽 𝛽 −1+𝛽 𝛽
𝛿𝑖 ln⁡[𝑡𝑖 (𝛼𝑡𝑖 𝐶𝑠𝑐ℎ2 (𝑡𝑖 ) − 𝛼𝑡𝑖 𝐶𝑜𝑡ℎ(𝑡𝑖 ) × model are taken into multi variate model.
𝛽 𝛽
𝑇𝑎𝑛ℎ(𝛼(1 − 𝑡𝑖 𝐶𝑜𝑡ℎ(𝑡𝑖 ))/𝛽]𝑔(𝑋𝑖 /𝜃)))] (8)
0⁡𝑖𝑓⁡𝑡𝑖 ⁡𝑖𝑠⁡𝑎⁡𝑟𝑖𝑔ℎ𝑡⁡𝑐𝑒𝑛𝑠𝑜𝑟𝑒𝑑⁡𝑜𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛 Ethical Consideration
Where𝛿𝑖 = {
1⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒⁡
The principled permission was found from WCU Research
Bayesian Approach and community engagement Vice president Office. And
also authorized permission was obtained from WCU
The Bayesian method is based on specifying a probability Referral Hospital medical director. Careful recruitment and
model for the observed data X, given a vector of unknown training for data collectors were undertaken.
parameters𝜃, leading to the likelihood function L(𝜃/𝑋).

Posterior Distribution RESULTS AND DISCUSSION

The posterior distribution is obtained by multiplying the There is 11(12.5%) chance of having breast cancer for a
prior distribution over all parameters, 𝜃by the full likelihood woman who bear child at a late age.
functions, L (𝜃/𝑋). All Bayesian inferential conclusions are
based on theposterior distribution of the model generated One-fourth (25%) of the total patients survived for only 2
(Asrat Demeke, 2015). days. The median survival time of death in breast cancer
patients are 2.5 days. Thus, the survival time of most
Inference concerning 𝜃 is then based on the posterior patients is minimal as the disease is chronic and fast fatal.
distribution, which is obtained by Bayes’ theorem. Then The proportion of being exposed to the disease for women
posterior distribution of 𝜃 is given by who get stressed/ depressed is 18(20.5%). There is
22(25.0%) chance of being exposed to breast cancer for
𝐿(𝜃/𝑋)𝜋(𝜃)
𝜋(𝜃/𝑋) = (9) late menopause women.
∫ 𝐿(𝜃/𝑋)𝜋(𝜃)𝑑𝜃

Combining the likelihood function with the prior distribution Forty (40) random females age greater than or equal to 20
on (β, σ2) and the full conditional distributions for unknown years were involved in testing the measures of level of
parameters, the posterior distribution can be written as: knowledge and practice of danger causes, near the

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model
 Int. J. Public Health Epidemiol. Res. 112

Table 2: Descriptive Statistics (Total and Percentage for each Category of the Variables)
Variable Category Status of Patient
Number of Number of Total (%)
Event (%) Censored (%)
Sex Male 1(1.1) 0(0) 1(1.1)
Female 54 (63.53) 31(36.5) 85(96.6)
Smoking habit non-smoker 49 (64.47) 27(35.5) 76(83.37)
Smoker 6(60) 4(40.0) 10(11.63)
Muslim 13(61.90) 8(38.1) 21(38.1)
Protestant 17(73.91) 6(26.1) 23(32.75)
Region Orthodox 13(56.52) 10(43.5) 193(26.75)
Others 19(63.16) 7(36.8) 19(22.09)
Urban 10(66.67) 5(33.3) 15(17.44)
Place/residence Rural 45(63.38) 26(36.6) 71(82.56)
Breast problem before No/Normal 49(66.21) 25(33.8) 74(86.04)
Yes/Analgic, bleeding or other 6(50.00) 6(50.00) 12(13.95)
Stage I 26(59.09) 18(40.9) 44(51.16)
II 26(72.23) 10(27.8) 36(41.86)
III 3(50.00) 3(50.00) 6(6.97)
single/unmarried Married 9(56.25) 7(43.8) 16(18.60)
Marital status Divorced/separated/widowed 46(66.67) 23(33.3) 69(80.23)
0(0.00) 1(100.00) 1(1.16)
Level of education Primary 25(60.98) 16(39.0) 41(47.67)
Secondary 19(82.60) 4(17.4) 23(26.74)
Tertiary 0(0) 2(100.0) 2(2.33)
0ther 11(55.00) 9(45.0) 20(23.26)
Family History No 53(63.85) 30(36.1) 83(96.51)
Yes 2(66.67) 1(33.3) 3(3.49)
Radiotherapy surgery 10(66.67) 5(33.3) 15(17.44)
Treatment taken Chemotherapy 17(80.95) 4(19.0) 21(24.41)
s/w or OR 2(33.34) 4(66.7) 6(6.97)
26(59.09) 18(40.9) 44(51.16)
High fat diet No 42(61.76) 26(38.2) 68(79.07)
Yes 13(72.23) 5(27.8) 18(20.93)
Early onset of No 46(63.01) 27(37.0) 73(84.88)
menarche Yes 9(69.23) 4(30.8) 13(15.11)
1st child at late age No 48(64.00) 27(36.0) 75(87.21)
Yes 7(63.64) 4(36.4) 11(12.79)
Stress/depression No 41(60.29) 27(39.7) 68(79.07)
Yes 14(77.78) 4(22.2) 18(20.93)
Knowledge &practice Very good 0(0.00)
Good 5(12.50)
Poor 35(87.5)
beginning exposure actions and premature caution signs
of breast cancer. Among which 35(87.5%) of them have
poor knowledge and practice of risk factors, early detection
measures and early warning signs of breast cancer.

The Kaplan- Meier Estimate of Time-to-death of

Breast Cancer Patients

The mean survival time of cigarette smoker women were

5.220 days with [95% CI: 2.857, 7.583] which was more
than that of non-smoker women by 4.676 days with its
[95% CI: 3.876, 5.475]. The mean survival time for patients
with no family record of breast cancer was greater than
had family history by 4.814and 2.667 [95% CI: 4.012, Figure 1: Fitting survival time (in days) curve of the breast
5.615] and [95% CI: 1.912, 3.421], respectively. cancer patients at HQEMMRH, south Ethiopia from
September, 2011 to February, 2017.

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model
Tekle and Dutamo 113

The above survival curve in the figure-1shows that the As shown in the table 4, figure-2shows S (t) ^hat for the
probability to die for those who stayed longer is getting CPHM of time to death of breast cancer patients on WHO
lower and vice versa, as shown in the numerical table-8 Stage (I, II, III) of the cancer which is similar for other
output, in the appendix. Thus, the probability of surviving categorical variables; the ruined lines show a 95%
for two days is 0.6887 which means 15 patients survived confidence cover about the survival function. Figure-1 has
for the second day of admit ion with (95%CI: 0.5919, shown the same pattern as shown here.
0.801) while only four patients survived for the fourth date
of admit ion with the probability of 0.4720 and with Multivariate Analysis (Cox Proportional Hazards
(95%CI: 0.3648, 0.611). Model)

Result offered in Table 3 indicates the model factor

approximates of coefficients for the predictors in the final
model based on Cox Proportional hazards beside with the
related standard error, Wald statistic/Z-statistic,
significance level/ Pr(>|z|) , hazard ratio/ exp(-estimates)
and 95-percentCI for the HR. The predictors such as age,
smoking habit, level of education, family history, breast
problem/Analgic, bleeding or other, High fat diet, child late
age, early menarche, late menopause passed the first
filtration of variables for multiple covariates analysis and
then forward variable selection method was used to
choose the important covariates worthy of including in
CPHM. So as to fix on if a covariate is important, the p-
value related with each model parameter has been
Figure 2: Kaplan Meier curve by Green Wood method for estimated and variables that have p-value less than or
WHO stage of the cancer with CIs to Compare Survival equals 0.05 cut point or 5% significance level are
Experience of breast cancer Patients at HQEMMRH, south considered as important variables and hence, are included
Ethiopia from September, 2011 to February, 2017. in the final model.
Table 3: Parameter (model) estimates of coefficients for the variables in the last model beside with the related standard
error, Wald statistic/Z-statistic, significance level/ Pr (>|z|), hazard ratio/ exp (-coef) and95-percent CI for the HR.
Estimates exp(Estim) SE z Pr(>|z|) exp(-Estima) lower upper
[.95 .95]
age -5.862e-02 9.431e-01 2.604e-02 -2.251 0.0244 * 1.060e+00 0.896 0.992
LeveleducSecon 8.426e-01 2.322e+00 3.878e-01 2.173 0.0298 * 4.306e-01 1.085 4.966
Breast problem Yes -6.654e-01 5.141e-01 5.956e-01 -1.117 0.0029* 1.945e+00 0.159 1.651
HighfatdietYes 2.883e-01 1.334e+00 6.223e-01 0.463 0.0170* 7.496e-01 0.394 4.517
ChildlateageYes -6.811e-01 5.061e-01 7.731e-01 -0.881 0.0321* 1.976e+00 0.111 2.303
EarlymenarchYes 1.732e+00 5.650e+00 7.495e-01 2.310 0.0209 * 1.770e-01 1.300 24.54
LatemenopauseYes -1.729e-01 8.413e-01 6.584e-01 -0.263 0.0192 * 1.189e+00 0.231 3.057

Table 4: Model Summary for Hypertabastic Proportional Hazards Model for breast cancer Patients at HQEMMRH
Parameter Estimate Approx t Value Gradient
Std Err Objective Function
Α 4.235545 2.756014 1.53683* -6.509327E-8
Β 1.300136 0.257030 5.05829* -0.000000160
Age -0.048327 0.022378 -2.15956* -0.000014768
leveleducSecon 0.657296 0.710347 0.92531* 0.000000155
histroyYes -0.200784 0.344273 -0.58321* -0.000000607
bresatprblemYes -1.107280 0.432010 -2.5630* 7.1233358E-8
HighfatdietYes -0.766543 0.681431 -1.1249* -0.000001279
childlateageYes 0.315151 0.470965 0.6691* 0.000003493
latemonepouseYes -1.018009 0.562178 -1.810* 0.000003604
In final model (model with significant covariates) of Cox late age, and late menopause. The values of the Wald

Proportional hazards the survival moment of patients had value for every  coefficients hold that the  values 

breast cancer was significantly affected as a result of i i

age, level of education, family history, breast were significantly dissimilarwith zero at α= 5% level of
problem/Analgic, bleeding or other, High fat diet, child significance.

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model
 Int. J. Public Health Epidemiol. Res. 114

Table 5: Model comparison inputs for the Hypertabastic Proportional Hazards and Cox proportional Models for breast
cancer Patients at HQEMMRH
Cox proportional Hazards Model Hypertabastic Proportional Hazards Model
Criterion WithoutCovariates WithCovariates Criterion WithoutCovariates WithCovariates
-2 LOG L 151.715 145.208 -2 LOG L 151.715 134.007
AIC 151.715 162.007 AIC 151.715 157.208
SBC 151.715 180.149 SBC 151.715 164.983

Hypertabastic Proportional Hazards Model Analysis The number of iteration used in the analysis for all of the
and Comparison with Cox Proportional Hazards Model parameters was set up to be 10,000 with ‘thinning’ number
of 1; the first 2000 was discarded (burn-in) and the rest of
The above Table 5 gives the clear visualization for the the chains was used to summarize the posterior
comparison of the two models applied for this data. distribution and this way means that 10,000 MC samples
Comparison of the two models is based on comparison were used.
criterion -2 LOG L and AIC values where the model with
smaller value fit the data better. Hence, Hypertabastic The Bayesian model discussed below the posterior
Proportional Hazards Model has smaller criterion values means, standard deviations and Monte Carlo (MC) errors
as compared to Cox Proportional Hazards Model and it for beta and sigma parameters were calculated to assess
better fits the data on breast cancer patients in this study. the accuracy of the simulation.
Based on the results (Table-4) from Hypertabastic
Proportional Hazards Model, age of patient, level of The first step in evaluating the results is to review the
education (Secondary), family history of cancer, earlier convergence diagnostics. Figures 3 to8 (Appendix) display
breast problem (Analgic, bleeding or other), High fat diet the Bayesian diagnostic graphs for age, level of education
consumption, late age child and late menopause are (secondary), family history, earlier breast problem, high fat
significantly contributing the risk of breast cancer for diet, late age child, and late menopause coefficient
females in this study which is also similar with model result parameters. The time series/ chain plots given in the
of Cox Proportional Hazards Model. figures confirm that convergence has been achieved for
the estimated parameters, since the chains appear to be
Simulation Study overlapping. That is, we are reasonably confident that
convergence has been achieved as all the chains appear
To evaluate the performance of the hypertabastic model to be overlapping one another. Also, the kernel density
we conduct a simulation study in which we compare the plots in the figures show the smooth, unimodal bell-shaped
overall fit of it with the Cox Proportional Hazards model. parameters. The autocorrelation plots displayed in the
In Bayesian inference, uncertainty with respect to figures decreases to near zero indicate efficient sampling.
parameters is at any point in time quantified by probability A more convincing evidence for convergence is observed
distributions. This means that a distribution needs to be in the Brook, Gelman and Rubin (BGR) diagnostics, which
specified for all parameters in advance. These prior compares the within-chain and the between-chain
distributions (Uniform prior for this study specifically) variability. The figures indicate that between-chain to the
reflect the priori expectations with respect to the parameter within-chain variability is one for each parameter or
values where in this study the Uniform distribution was converge to approximately one indicating the good
used as prior distribution. convergence of chains.

Table 6: Posterior Summary Statistics for parameters from Bayesian analysis of the Hypertabastic model
Quantiles Posterior Intervals
Description N Mean Standardev 95% Equal-Tail 95% HPD Interval
Deviation 25% 50% 75% Interval
Age 10000 0.052 0.0267 -0.070 -0.0518 -0.0342 0.098 0.0016 0.1506 0.1141
Leveleduc 1vs 2 10000 0.238 0.5783 -0.145 0.2373 0.6262 0.890 1.2649 1.454 3.584
History 0 vs 1 10000 2.939 4.9578 0.837 1.5690 3.1225 0.279 14.47 0.078 9.611
Breast problem 0 vs 1 10000 2.919 2.2615 1.515 2.3121 3.5670 0.695 8.777 0.375 7.077
Highfatdiet 0 vs 1 10000 1.181 0.8987 0.603 0.9351 1.4660 0.267 3.5687 0.134 2.851
Childlateage 0 vs 1 10000 2.100 2.1670 0.847 1.4719 2.5626 0.302 7.770 0.093 5.780
Latemenopause 0 vs 1 10000 1.956 1.7675 0.931 1.4603 2.3710 0.415 6.3862 0.269 5.046

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model
Tekle and Dutamo 115

Table-6: Posterior Summary Statistics for parameters from Bayesian analysis of the Hypertabastic model (Continued…)
Maximum Likelihood Estimates
Parameter DF Estimate MC St.Error z Pr > |z|
age 1 -0.0483 0.0255 -0.9230 0.0350
leveleduc2 1 0.1871 0.5499 0.7175 0.0020
history1 1 0.2036 0.9059 -0.5295 0.04560
bresatprblem1 1 0.6144 0.5856 1.2643 0.0029
Highfatdiet1 1 0.1473 0.6120 -1.0869 0.0170
childlateage1 1 -0.4471 0.7576 1.0198 0.3078
latemonepouse1 1 -0.2648 0.6482 -0.0826 0.0189

Table 7: comparison of goodness-of-fit (based on DIC, pD) for the two models
Bayesian Cox Proportional Hazards Model Bayesian Hypertabastic Proportional Hazards Model
AIC (smaller is better) 161.187 AIC (smaller is better) 160.241
BIC (smaller is better) 188.400 BIC (smaller is better) 187.603
DIC (smaller is better) 1132.411 DIC (smaller is better) 1120.200
pD (Effective Number of Parameters) 275.343 pD (Effective Number of Parameters) 271.139

Based on the above Table-7 result, Bayesian A q-q plot is made to check whether the Hypertabastic
Hypertabastic Proportional Hazards Model would be the Proportional model presented an enough fit to the data by
better fit the data as its criterion (DIC and pD) are smaller having 2 dissimilar groups of population. We shall
than that of the Bayesian Cox Proportional Hazards Model. graphically check the sufficiency of the model by
Hence, the last inference in this study would be based on evaluating the significantly dissimilar groups of patients by
Bayesian Hypertabastic Proportional Hazards Model. stage, first child at late age, and educational level of
patients. The facts come into view to be roughly linear for
Parameter Interpretation all variables stage, first child at late age, and educational
level of patients. Therefore, the Hypertabastic Proportional
Thus, an extra year of age reduces the daily hazard of
model comes into view to be the best in explaining survival
breast cancer with a value of exp (-0.0483) = 0.9528479,
time of the patients in this study.
on average that is, by 100 % - (100%*0.9528479) =
4.71521percent, by keeping other variables in the model.
Similarly, the hazard of breast cancer patient for a woman
DISCUSSION
learnt secondary education increased with 1.2058, or
20.58%. Thus, the hazard of breast cancer for woman
Hypertabastic Proportional Hazards model in classical
learnt secondary education (95%CI: 0.1506, 0.1141;
approach and in Bayesian approach were fitted. Both
P=0.0020) is 1.2058 times higher than woman learnt other
methods give almost consistent results but most of the
educational level or never learnt. The hazard of breast
parameters in Bayesian approach had smaller standard
cancer for woman with breast problem before like Analgic,
error than the corresponding classical model. Therefore,
bleeding or other (95%CI: 0.375, 7.077; P=0.0029) is
Bayesian Hypertabastic Proportional Hazards gives better
1.848547 times higher than that of breast normal woman.
fit than classical counterpart.
The hazard of breast cancer for woman consuming high
fat diet (95%CI: 0.134, 2.851; P=0.0170) 1.1587 times
In average, the study subjects of patients are young in age
higher than that of consuming free of fat diet. The risk of
and age was found to be among the leading threat factor
breast cancer for woman late menopause (95%CI: 0.269,
for the disease as the study result reveals. This result is
5.046; P=0.0192) is 0.7674 times reduced as compared to
similar to a study done on the disease occurrence with age
the counterpart.
amongst women in the UK 2006-2008 which revealed that
the hazard of getting the disease boosts through age.
Further Checking for the Model Adequacy
Thus, a woman is more than 100 times more likely to
develop breast cancer in her 60s than in her 20s [Richard
Statistical tests for Proportional hazards model
et al., 2000; Ahammad Basha Shaik et al., 2015].
assumptions and model diagnostics were done using
numerical and graphical techniques.
Out-of-the-way from rising age, among the strongest
The test of correlation (rho) is insignificant that indicates recognized threat cause for a woman being detected with
proportional hazards assumption is fulfilled and the global breast cancer is with a close female relative (i.e., mother,
test is greater than 0.05 the assumptions did satisfy by the sister, or daughter) had a history of breast cancer. This
covariates in the model. Hence, the non-parametric relationship is interrelated in most researches with a dual
modeling approach was applied by overwhelming the raise in risk [Burstein H., 2008].
parametric models for this reason.

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model
 Int. J. Public Health Epidemiol. Res. 116

Male gender has a great deal of lesser threat of emergent with the disease. It also showed that sex, place of
of the disease than female gender. This is supported by residence of the patients, stages of the disease, marital
other studies stated as around 99-percent of the disease status were not significant for the survival probability of
cases are detected in women, among most urbanized patients with breast cancer.
countries; which stand for the uppermost prevalence of
men breast cancer which way male comprises 5–15 Hypertabastic PH model was fitted because the
percent of the disease cases, as manifested in a few supposition of it was satisfied or not violated. Using the QQ
African countries [Richardet al., 2000]. plot technique Hypertabastic PH model was found to be
best fitting to the breast cancer data. The results showed
As the current study shows, lesser age of initial childbirth, that the predictors: age, level of education, family history,
contrasted to the average age of 24, breast problem/Analgic, bleeding or other, High fat diet,
[http://www.medscape.com/viewarticle] having additional child late age, late menopause were significant threat
offspring (about 7% lowered risk per child), and cause for the death of patients from the disease. And
breastfeeding (4.3% per breastfeeding year, with an though not included in the model, Knowledge &practice of
average relative risk around 0.7 [McTiernan and Thomas, pretest or early test of breast by self was very poor among
1986; Byers et al., 1985]) have all been associated with the patients. Thus, Knowledge &practice towards threat
lowered breast cancer risk in huge studies [Breast cancer causes, premature discovery measures and early warning
and hormone replacement therapy, 2008]. Breast cancer signs of breast cancer were not recognized by the patients.
risk increases with premature menarche and delayed The median survival time of death in breast cancer patients
menopause, and it is reduced by premature first full term is 2.5 days. Thus, the survival time of most patients is
pregnancy [Jatoi, 1999]. minimal as the disease is chronic and fast fatal.

In 2009, the Canadian Expert Panel on Tobacco Smoke The ministry of health and policy makers should work on
and Breast Cancer Risk concluded that both active and awareness by letting to know the threat causes for the
passive smoke experiences raise breast cancer hazard. disease and to complete the prescribed treatment without
However, this current study contradicts this result. The considering breast cancer as incurable disease and to
deviation can also be revealed as absence of correlations follow up their cancer status to minimize the risk of death
between breast cancer and smoking in various researches and recognizes breast cancer as an important health
could be because of the potential anti-estrogenic problem and establishing screening test and early
consequence of smoking, which might oppose the detection policies for most risky groups.
unpleasant effects of chemical carcinogens in the breast
[MacMahon B et al., 1982]. In addition, it will be important to open cancer diagnosing
and treatment center in each woreda of the zone.
As Ahammad Basha Shaik et al., (2015) conducted,
survival prototypes of the patients were researched and Awareness has to be given for the society on causes of
survival estimates were computed using the KM method breast cancer. The mass media can play an effective role
and the Hypertabastic Proportional model was considered in this regard and special attention should be given to old
to investigate the consequence of covariates to the age women, because they are the riskiest groups for
survival time of the patients. breast cancer.

Limitation of the Study

CONCLUSIONS AND RECOMMENDATIONS
As the data is gathered from the card of patients in the
Bayesian inference is more powerful than the classical study there were a lot of patients with insufficient
counterpart in model analysis. Thus, in this study the information. Even if the number of patients is high in
authors fitted both Bayesian and classical models for the number, on follow-up it is often a physical and financial
data and the Bayesian approach better fitted the data. burden for them to return hospital for follow-up in the day
Hypertabastic PH model in Bayesian version (in final of appointment; most patients didn’t come again to the
mode) and in Classical version (in other mode) is Hospital to be treated. The other limitation was small
supported to be better fit the data as compared to the Cox number of cases because the Hospital has no cancer
PH version. Hence, the model inference was given based registry center to serve the cancer related patients in the
on the Bayesian Hypertabastic PH model in the most final catchment area. And therefore, the study included only 86
model. patients which were the total number of patients with full
information.
In assessing the significant risk factors the Log Rank test
for both models revealed that, age, level of education, ABBREVIATION
family history, breast problem/Analgic, bleeding or other,
High fat diet, child late age, late menopause had significant Cox Proportional Hazard Model = CPHM
effect on the continued existence probability of patients Cumulative Hazard Function = CHF

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model
Tekle and Dutamo 117

Hazard Ratio = HR B.M.Wadler,C.M. Judge,M.Prout, J.D.Allen, andA.

Hossana Queen Elleni Mohammad C.Geller (2011). “Improving breast cancer control via
Memorial Referral Hospital = HQEMMRH the use of communityhealth workers in South Africa: a
Kaplan-Meier = KM critical review,” Journal of Oncology, Article ID 150423,
Proportional Hazards Model = PHM 8 pages, 2011.
Survival Analysis = SA Collett, D. (1994). Modeling survival data in medical
Wachemo University = WCU research. London: Chapman & Hall.
http://www.medscape.com/viewarticle/517532
http://www.csrwire.com/.
Declaration of Agreement with Ethical Values https://en.wikipedia.org/wiki/Risk_factors_for_breast_can
ce.
Ethical endorsement: All procedures done in the research Gebremariam A, Addissie A, Worku A, et al (2019). Breast
concerning human participants were in harmony with the and cervical cancer patients’ experience in Addis
ethical standards of the institutional and/or general Ababa city, Ethiopia: a follow-up study protocol. BMJ
research commission. Open 2019; 9:e027034. Doi: 10.1136/ bmjopen-2018-
027034.
Informed consent: Not applicable Howlader N, Noone AM, Krapcho M, Neyman N, Aminou
R, Altekruse SF, et al (2009). SEER Cancer Statistics
Fund Status: This study was not funded by anyone. Null Review, 1975-2009 (Vintage 2009 Populations)
fund at all, done by self strive. [internet]. Bethesda (MD): National Cancer Institute;
c2009-3 [updated 2012 Aug 20; cited 2013 Jan 7].
Authors’ clash of concern: There is no clash of concern Hosmer, D. and S. Lemeshow (1998).Applied survival
among the authors. analysis: Regression Modeling of Time to Event
Data.Wiley New York.
Jatoi, I. (1999). Breast cancer screening. American
REFERENCES Journal of Surgery, 177, 518-524.
Kaplan, E.L.; Meier, P. (1958)."Non parametric estimation
Ahammad Basha Shaik , Venkataramanaiah. M , S.C. from incomplete observations" .J.Amer. Statist.Asn,
Thasleema (2015). Statistical Applications of Survival 53(282):457-481. JS TOR 2281868.
Data Analysis for Breast Cancer Data. i-manager’s Li, H. (2017) Survival Analysis for a Breast Cancer Data
Journal on Mathematics, Vol. 4 l No. 2 l April - June Set. Advances in Breast Cancer Research, 6, 1-15.
2015. http://dx.doi.org/10.4236/abcr.2017.61001.
A.GebremedhinandM. Shamebo (1998). “Clinical profile MacMahon B, Trichopoulos D, Cole P, Brown J (1982).
of Ethiopianpatients with breast cancer,” East African Cigarette smoking and urinary estrogens. N Engl J
Medical Journal, vol.175, no. 11, pp. 640–643. Med.;307(17):1062-5.
Asrat Demeke (2015). Bayesian Analysis of Survival of Madigan MP, Ziegler RG, Benichou J, Byrne C, Hoover RN
Breast Cancer Patients: A Case Study at Tikur Anbessa (1995). "Proportion of breast cancer cases in the United
Specialzed Hospital, Addis Ababa, Ethiopia. Master’s States explained by well-established risk factors". J.
thesis. Natl. Cancer Inst.87 (22): 1681–5.
Bast RC, Kufe DW, Pollock RE, et al (2011). Cancer doi:10.1093/jnci/87.22.1681. PMID 7473816.
Medicine (e.5 ed.). Hamilton, Ontario: B.C. Decker. Margolese, Richard G, Bernard Fisher, Gabriel N
ISBN 1-55009-113-1. Retrieved 27 January 2011. Hortobagyi, and William D Bloomer (2000). "118". In
Burstein H( 2008). Malignant tumors of the breast. In: McTiernan A, Thomas DB (September 1986).
DeVita, VT Jr., Lawrence TS, Rosenberg SA, DePinho "Evidence for a protective effect of lactation on risk of
RA, Weinberg RA, editors. DeVita, Hellman, and breast cancer in young women. Results from a case-
Rosenberg’s Cancer: Principles and Practice of control study". Am. J. Epidemiol.124 (3): 353–8.
Oncology. 8th ed. Philadelphia: Lippincott Williams & PMID 3740036.
Wilkins; p. 1606-54. Mohammad A Tabatabai (2007).Hypertabastic survival
Byers T, Graham S, Rzepka T, Marshall J ( 1985). model. Theoretical Biology and Medical
"Lactation and breast cancer. Evidence for a negative Modelling2007,4:40doi:10.1186/1742-4682-4-40.
association in premenopausal women". American (http://creativecommons.org/licenses/by/2.0).
Journal of Epidemiology 121 (5): 664–74. T. Ersumo (2006). “Breast cancer in an Ethiopian
doi:10.1093/aje/121.5.664. PMID 4014158. population, Addis Ababa,” East Africa Journal of
Surgery”, vol. 11, no. 1, pp. 81–86.

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model
 Int. J. Public Health Epidemiol. Res. 118

Accepted 22 June 2019

Citation: Tekle G, Dutamo Z (2019). Survival Analysis of

Determinants of Breast Cancer Patients at Hossana
Queen Elleni Mohammad Memorial Referral Hospital,
South Ethiopia: Bayesian Application of Hypertabastic
Proportional Hazards Model. International Journal of
Public Health and Epidemiology Research, 5(2): 108-118.

Copyright: © 2019 Tekle and Dutamo. This is an open-

access article distributed under the terms of the Creative
Commons Attribution License, which permits unrestricted
use, distribution, and reproduction in any medium,
provided the original author and source are cited.

Survival Analysis of Determinants of Breast Cancer Patients at Hossana Queen Elleni Mohammad Memorial Referral Hospital, South Ethiopia: Bayesian
Application of Hypertabastic Proportional Hazards Model