Infarction Stroke Risk Prediction Model for Indonesian Population: A Case-Control Study

Background: Stroke is the main cause of death and disabilities in Indonesia and the world. Various prediction model for stroke have been developed. This study attempts to develop a model used to predict infarction stroke in Indonesia. Objective: This study aims to develop a model to predict infarction stroke risks. Method: This study is an observational research applying case-control research design. The number of samples used in this study were 310 individuals, consisting of 155 members of case group and 155 members of control groups. The writers used discriminant analysis to conduct statistical analysis on the data. Results: Valid and reliable risk factors of stroke used to develop prediction model for infarction stroke in this study are systolic blood pressure, diastolic blood pressure, triglyceride levels, stroke history, hypertension history, dyslipidemia history, vegetable consumption, sleep duration, snoring, exercises, and emotional stresses. Conclusion: This study comes up with a prediction model for infarction stroke risks. The prediction model is expressed by following formula: Infarction stroke risk = 0.929 x Systolic Blood Pressure + 0.886 Diastolic Blood Pressure + 0.160 x Triglyceride Levels + 0.850 x Hypertension History + 0.332 x Stroke History + 0.084 x Dyslipidemia History + 0.124 x Vegetable Consumption + 0.245 x Emotional Stresses + 0.346 x Snoring Habit – 0.193 x Exercise Habit – 0.190 x Sleep Duration


Introduction
Stroke has become the main cause of death and disabilities in Indonesia and the world.The proportion of stroke cases is increasing.Recently, stroke mostly occurs at productive age¹.Productive individuals can be saved from stroke through promotive actions and primary preventions 2 .One of primary preventions is by predicting the risk of stroke based on stroke risk factors.Various prediction models have been developed but these models were applied on populations outside Indonesia and conducted based on limited risk factors as its variables 3,4,5,6,7 .Previous stroke prediction models developed in Indonesia are based on clinical and laboratory indicators 8 .The development of a new stroke prediction model that is able to predict risks of infarction stroke by adding demographic indicators, behavioral indicators, and psycho-spiritual indicators becomes urgent.

Methods Research Design
This study applied case-control research design.The case group consists of stroke patients at Prof. Dr. Aloei Saboe Hospital, Gorontalo (both infarction stroke and hemorrhagic stroke).The control group of this study consists of healthy patients or patients with non-stroke diagnosis.The case and control groups were matched based on their age and sex.This study was conducted for two years, starting from December 2012 to December 2014.

Research Subject
The samples of this study were all patients diagnosed with stroke at Prof. Dr. dr.Aloei Saboe Hospital, Gorontalo.The samples were chosen based on consecutive sampling method, namely by matching all new patients into inclusion criteria and exclusion criteria until the ideal number of samples reached.The inclusion criteria of this sampling method were willingness of the subjects to participate in this research, the subjects were in compos mentis consciousness state, and the subjects came from Malay Sub-races.Compos mentis condition becomes one of inclusion criteria because this condition enables the writers to collect research data easily through interviews and laboratory examinations.Since all the subjects were Indonesian, they were members of Malay sub-race.Meanwhile, the exclusion criteria of this sampling method were aphasia and loss of consciousness.Aphasia and loss of consciousness were the exclusion criteria of this study because these conditions can make communication with the subject more difficult.Based on these criteria, the number of the subjects participated in this study were 310 respondents divided into two groups: case group and control group, consisting of 155 respondents on each group.

Data Collection
Physical examinations on the subjects consisted of different types of examination.Head CT Scan examination was conducted to obtain stroke diagnosis.This examination were conducted using Siemens Sensation 64 model Multi Slice Head CT Scanner with mAs 300 Slice, 3.0 mm, KV: 120, rotation time 1 second.Blood pressure examination was conducted using mercury sphygmomanometer.The examination was conducted while patient in lie down position; cuff was installed around patient's upper arm, 2.5 cm above cubital fossa; the process was repeated twice.Blood sugar examination were conducted based on hexokinase method and cholesterol level examination were conducted based on colorimetric enzymatic method.These examinations were carried out using Hitachi 912 autoanalyzer.Uric acid examination were conducted based on uricase method using Hitachi 912 autoanalyzer.The writers carried out interviews on the patients by using validated questionnaire to obtain data related to the patients and his/her behaviors, such as age, sex, medical history, smoking habit, dietary habit, exercise habit, and religious activities.The writers used Analog Anxiety Scale (AAS) to obtain data related to emotional stresses experienced by the patients and validated patience level proposed by Prasetyono (2014) to measure patience level of the patients.

Statistical Analysis
The writer conducted statistical analyses on the collected data.The statistical analyses consisted on measuring each variable, discriminant analysis, and analyzing each variable factors composing stroke index.

Results
Based on the obtained data, 152 respondents were male (49%) and 158 respondents were female (51%).Based on age categories, 31% respondents aged 50-59 years old; 29% were aged 60-69 years old; 27% were aged 40-49 years old.Based on their residence, 45.8% of the respondents lived in urban areas while 54.2% lived in rural areas.The writers conducted further analysis on the 27 variables of stroke risk factors based on Stepwise Discriminant Analysis.The results of this analysis indicated that only 11 of these risk factors with the highest lambda (λ) value representing five main indicators.As presented on Chemoreceptors on blood vessel responds this condition and triggers the patient's awakening to breathe.Normal breathing may improve oxygen levels in blood and the patient may sleep 12 .However, continuous sleep apnea may drastically reduce oxygen level on brain and triggers infarction stroke 11 .Diagnosis on snoring may be conducted by an instrument named polysomnogram 12 .However, due to lack of facilities, polysomnogram examination cannot be conducted.
Sleep duration is also one of stroke risk factors as indicated by risk factor score -0.190.A previous found that prevalence of stroke is higher on individuals who sleep for less than 6 hours/day or more than 9 hours/day compared to individuals who sleep for 7-8 hours/day 13 .Obstructive Sleep Apnea (OSA) also increases mortality of stroke 14 .OSA also correlates with worse functional impairment and affects the length of rehabilitation period 15 .OSA is found on 44-72% of post-stroke patients.OSA can cause functional damage through intermittent nocturnal hypoxia, reduced cerebral perfusion, and fragmented sleep 16 .Although the data showed there is no significant changes of sleep duration, previous researches indicated that changes on sleeping pattern may affect the occurrence of stroke.However, the writers could not analyze the effects of sleeping pattern change due to lack of polysomnogram facility used to conduct the analysis.
Emotional stresses are also one of stroke risk factors (as indicated by score of this factor as much as +0.245).The group exposed to emotional stresses has the highest because the stresses may affect cerebral hemodynamic functions 17 .A study conducted by Hacinski found that emotional stress might improve the risk of stroke by 1.5 -2 times 18 .Empirical findings has proven the effect of psychological factors (including psychological stresses) on cardiovascular disorders.The results of INTERHEART study, a semi-quantitative research on subjective perception of psychological stresses involving participants from 52 countries, indicated strong correlation among the aspects of stress, including financial stress, and life-related stress 19 .Previous prospective and case-control studies reported that severe self-perceived psychological stress, life-related stresses, and failures in overcoming stresses individually correlated with stroke risk 20 .Through sub-group analysis, several studies showed significant correlation between selfperceived psychological stresses with fatal stroke 21 .
A study conducted by Iso in Japan proves that severe self-perceived psychological stress correlates with stroke mortality 22 .This study reveals that Triglyceride levels is one of risk factors of stroke (as indicated by the score +0.160).A longitudinal study conducted for 7.2 years by Berger found that 68.1% patients with triglyceride level higher than 200 mg/dl are most likely experiencing stroke 23 .Different studies show that the increasing of triglyceride levels by 90 mg/dl improves the risk of stroke by 70% for female patients and by 30% for male patients 24 .Every increasing triglyceride levels by 1 mmol/L independently correlates with increasing stroke cases by 14-37%.Triglyceride metabolism abnormalities trigger atherogenesis by increasing CAMs expression on vascular system 25 .Dyslipidemia history is one of the risk factors of stroke (indicated by score + 0.084).A metaanalysis of 45 cohort observational prospective involving 45000 individuals found that there was no correlation between total cholesterol levels and infarction stroke 26 .Data of previous studies indicating the effect of hypercholesterolemia on stroke were not consistent.Surprisingly, there are consistent findings indicating correlation between low total cholesterol level and high incidents of intra-cerebral hemorrhagic stroke and sub-arachnoid hemorrhagic stroke on oriental populations.Related to these findings, some experts opine that low total cholesterol serum may weaken intra-cerebral arterial endothelium that causes bleeding during hypertension 27

Table 1 :
Overview of Research Variables