Mortality rate and cause of death pattern in Thai Nguyen and Quang Ninh provinces

Nội dung text: Mortality rate and cause of death pattern in Thai Nguyen and Quang Ninh provinces

1. JOURNAL OF MEDICAL RESEARCH MORTALITY RATE AND CAUSE OF DEATH PATTERN IN THAI NGUYEN AND QUANG NINH PROVINCES Pham Ngan Giang, Nguyen Phuong Hoa, Thanh Ngoc Tien, Nguyen Thi Tuyet Nhung Department of Family Medicine, Hanoi Medical University Cause of deaths reflects the burden of diseases in the community and is important information for evidence-based health policy. Objectives of the study were to determine mortality rates and cause of death pattern in Thai Nguyen and Quang Ninh. A cross - sectional study was conducted. One thousand four hundred and seventy seven deaths were recorded at 26 communes in 2014. The survey was used WHO standard verbal autopsy questionnaire. The results showed that overall mortality rate was 4.94‰, mortality rate among males was higher than among females (6.09‰ versus 4.91‰, p < 0.05), urban population had lower rate of death than the rural population (4.73‰ versus 5.56‰, p < 0,05). The results showed that most of the deaths occurred at home (88%), only 4.8% of deaths in health facilities. There was a transition in the cause of death pattern while the leading causes were cardiovascular diseases, cancer and injury. In particular, death from stroke was 20.6%, lung cancer 8.3% and traffic accident 3.7%. In conclusion, it is necessary to collect information about the deaths, which are outside health fa- cilities (at home) and an intervention programs need to prioritize for some of the leading causes of death. Keywords: mortality rate, cause of death, burden of diseases I. INTRODUCTION Mortality statistics and causes of death statistics [4]. (COD) information are important to measure Mortality data on causes of death for population health status, identify key public Vietnam have not been reported to the health issues, set priorities, and improve World Health Organization (WHO) to date health outcomes through effective resource [5; 6]. With a population of over 91 million allocation [1 - 3]. However, an estimated 2/3 [6], there is a critical need for such data for of all deaths were not reported globally. Mil- the above stated purposes. At the national lions of people in Africa and Asia die without level, due to limitations in the availability of leaving any trace in legal records or official data, cause of death patterns in Vietnam has been estimated based on mortality data Corresponding author: Nguyen Phuong Hoa, Depart- from Chinese, Thai and Indian populations ment of Family Medicine,, Hanoi Medical University [7]. Email: nguyenphuonghoa@hmu.edu.vn The absence of complete and valid na- Received: 05 June 2017 tional mortality data limits the evidence Accepted: 16 November 2017 base to estimate the burden of disease in JMR 111 E2 (2) - 2018 85
2. JOURNAL OF MEDICAL RESEARCH Vietnam. At the national level, three organi- home. zations within the Government of Vietnam Therefore, by using verbal autopsy (VA), collect national mortality data: the Ministry this study was conducted to determine mor- of Health (MOH), the Ministry of Justice tality rates and cause of death pattern in (MOJ), and the General Statistics Office Thai Nguyen and Quang Ninh in 2014. (GSO). GSO data provide important indi- II. SUBJECTS AND METHODS cators such as life expectancy and crude death rates [8]. However, this source does 1. Study sites and sample not collect detailed information about COD, The study was implemented in two prov- which is urgently required by the health sec- inces, Quang Ninh and Thai Nguyen, which tor for developing health interventions, prior- are located in the Northern region of Viet- ity setting, and policy formulation. The MOJ nam. In each province, one urban district has legal responsibility over the national and one rural district were chosen to as- civil registration and vital statistics system. sess likely differences between those two For deaths, this system only collects infor- areas. Within each selected district, 6-7 mation about the numbers of deaths by sex communes were chosen as study sites by and age. However, the registration of deaths simple random method. recorded in this system is low for different The study sample comprised all deaths areas. Also, the MOJ system does not have that occurred between 01/01/2014 and any procedures for formal reporting of the 31/12/2014 among residents of the 26 se- causes of death (COD). lected communes. There were 1477 deaths In order to meet the information needs in total, which were listed by combining the of the health sector, the MOH operates a A6 registers, the Justice Clerks’ books and routine death register system at commune some other resources. All deaths in each health stations (CHS). Local commune selected commune were re-investigated to health staffs identify deaths in the commu- ascertain the causes of death, using Ver- nity and record basic demographic data and bal Autopsy (VA) surveys. There were 1365 information on the cause of death for each Verbal Autopsy (VA) interviews conducted. death in an official MOH log-book named The other 112 cases could not undertake VA the “A6 register”. Frankly, data from the A6 mainly due to the movement of population. registers are not used effectively at different 2. Methods levels in the health sector because there is This assessment was based on a no consistent process for compiling data cross-sectional study design. from A6 registers at district, province, and Data collection national levels; therefore the MOH mortality - Making the combined death list database now in the Statistical Handbook All deaths recorded in A6 registers, the of Vietnam MoH was based on mortality Justice Clerk books and some other sourc- data from hospitals only. However, currently es during the defined one-year period be- in Vietnam, the majority of deaths occur at 86 JMR 111 E2 (2) - 2018
3. JOURNAL OF MEDICAL RESEARCH tween 01 January 2014 and 31 December The interviewees were persons who 2014 were extracted onto a separate form. were mainly responsible for taking care of Information collected included reported the decedent before he/she died, and who name, age and sex, date of death, address were able to provide information about the of the deceased. Then, a process of match- symptoms and diseases experienced by the ing death cases from these sources was deceased prior to death. carried out by commune health staffs, who The supervisors were the principal in- were responsible for mortality recording. vestigators and staff in the Provincial/ Dis- Variables used for the matching process trict Health Centre. Supervisors provided were name, sex, age, date of death, and assistance and monitored the interviewers’ address of the deceased. activities to ensure the quality of the VA in- - Implementing Verbal Autopsy surveys terviews. On completion of all VA interviews All deaths identified in the above com- were diagnosed and coded of the Underly- bined list were followed up to conduct the ing cause of death (UCOD), by trained doc- household verbal autopsy (VA) interview tors The UCOD then was coded using Inter- using a standardized questionnaire that national Classification of Diseases version elicits information on signs, symptoms, 10 (ICD-10) by application of the mortality medical history, and circumstances preced- coding rules and guidelines[9]. ing death. The VA questionnaire used for Data analysis and management this assessment is the updated version of Epidata software and SPSS18 were em- the Vietnamese verbal autopsy question- ployed to analyse data. naire, accompanied by a manual and guide The proportions were calculated by for fieldworkers. The original English ver- communes, district, provincial levels, urban/ sion of the VA questionnaires, which was rural areas, sex, broad age groups (0 - 4 recommended by WHO, were translated years, 5 - 14 years, 15 - 59 years, and 60+), into Vietnamese and revised. place of death, type of health facility, and Interviewers were local health workers the last treatment method. Each proportion from commune health stations who have was computed for 95% confidence intervals medical related backgrounds (e.g., medical [10]. assistants, nurses) working at the commune 3. Ethics or village level where the deaths occurred, Respondents of this study were clearly and who have the responsibility for collect- explained all information regarding the ob- ing data and recording it in the A6 registers jectives of this assessment, the detail of at commune health stations. The training of collecting information. Respondents have interviewers emphasized techniques and had complete autonomy in regard to partic- communication skills to motivate the princi- ipation, as well as freedom to withdraw at pal caretaker of the deceased to participate any stage during the interview. Access to in the survey and encourage them to give completed questionnaires and data were accurate and honest answers. JMR 111 E2 (2) - 2018 87
4. JOURNAL OF MEDICAL RESEARCH restricted to authorized personnel to ensure ber of deaths. VA interview could not be car- the confidentiality of each respondent. The ried out in 112 cases (7.4% of total deaths). collected data was only used for the pur- Table 1 describes the death amount and pose of research. the crude death rate in general according to gender and location identified during III. RESULTS the study. The mortality rate of general A total of 1477 deaths were recorded in population was calculated 4.94 per 1000. the reference year, which comprised 746 In comparison to female group, the death cases in Quang Ninh province and 731 cas- proportion in male was higher with statisti- es in Thai Nguyen province. Out of these cally significance (p < 0.001). In regards to 1477 deaths, the COD were re-investigated location, the urban population had the lower in 1365 cases using VA household inter- mortality rate than the rural population (p < views, equivalent to 92.6% of the total num- 0.001) Table 1. Crude death rate by sex and area in 2014 Rate Characteristic Total Number of deaths p value (‰) General population 299,237 1477 4.94 Sex Male 152,357 928 6.09 < 0.001 Female 146,880 549 4.91 Area Urban 226,325 1071 4.73 < 0.001 Rural 72,912 406 5.56 Table 2 describes the distribution of last treatment prior to death. Most of them deaths by age group and some factors re- had visited central/ provincial hospitals lating to death, all the statistics were as- (80%) and in about 29% of cases, a visit to certained by VA. In regard to age, over two a district hospital was reported. Only 4.7% thirds of the deaths were among the elderly. went to a commune health station, 2.1% The proportions of deaths recorded in two saw healers and very few people visited pri- groups under the age of 5 years and 5 - 14 vate doctors. As shown in table 2, although years old are very low (1.8% and 1.2% re- only 6% of VA respondents kept the last spectively). As can be seen, more than 70% treatment documents provided by hospi- of people attended a health facility for the tals, which would be useful for reporting for 88 JMR 111 E2 (2) - 2018
5. JOURNAL OF MEDICAL RESEARCH the mortality register at CHS. This aspect will be given attention in the recommendations to strengthen the COD reporting system. Table 2. Distribution of deaths by age and information before death Characteristic Number of deaths Percent (%) Age group (n = 1365)* 0 - 4 24 1.8 5 - 14 17 1.2 15 - 59 455 33.3 60+ 869 63.7 Treatment at health facility in the last sickness leading to the death? (n = 1365) Yes 959 70.3 No 354 26.0 Unsure/Don’t know 35 2.6 Type of health facility in the last treatment (n = 959) Central/Provincial Hospitals 770 80.3 District hospitals 276 28.8 Commune Health Station 45 4.7 Healers (traditional medicine) 20 2.1 Private Doctor 6 0.6 Others 4 0.4 Recall information about the diagnosis after 885 92.3 discharge from hospital (n = 959) Kept the documents from hospital about the 55 5.7 last treatment (n = 959) *1365 cases were interviewed by VA questionnaire Figure 1 illustrates the places of deaths. Approximately 88.1% of the people died at home and only 5.0% died at a health facility (includes hospitals, commune health station, clinic, etc.). However, as mentioned above, a large number of the decedents who died at home had visited health facilities during their final illness. JMR 111 E2 (2) - 2018 89
6. JOURNAL OF MEDICAL RESEARCH 5% 1.9% 5% At home At Health facility On the way to health facility Others 88.1% Figure 1. Place of death of the study sample Table 3 describes the differentials in mortality pattern by sex. Although stroke was the first leading COD in both males and females, the percentage of females who died due to stroke was higher than those in males. This is similar to some other causes such as pneumonia, stomach cancer and other CVD. The proportion of senility as a cause of death in females was 5 times higher than males, therefore the rate of this cause in male was not listed in the 10 leading COD table. However, some causes such as liver cancer, lung cancer, cirrhosis of liver and HIV, had proportions of male deaths that outweighed those in females. Beside two common groups as CVD and cancers, injury that was marked mainly by road traffic injury also situated in the list of top as a leading cause of death among both males and females (4.2% and 2.8% respectively). Table 3. Distribution of 10 leading CODs by sex Male (n = 835) Female (n = 530) Rank Disease Rate (%) Disease Rate (%) 1 Stroke 18.7 Stroke 23.6 2 Lung cancer 10.4 Senility 10.9 3 Liver cancer 6.7 Lung cancer 4.9 4 HIV/AIDS 4.3 Pneumonia 4.7 Breast, Cervix and Ovary 5 Cirrhosis of liver 4.3 4.2 cancers 6 Road traffic Injury 4.2 Stomach cancer 3.0 90 JMR 111 E2 (2) - 2018
7. JOURNAL OF MEDICAL RESEARCH Other cardiovascular dis- 7 Pneumonia 3.5 3.0 eases 8 Ischaemic heart disease 2.5 Road traffic Injury 2.8 9 Other malignant neoplasms 2.4 Ischaemic heart disease 2.6 10 Other unintentional injuries 2.3 Diabetes mellitus 2.5 All other diseases/ causes 29.5 All other diseases/ causes 30.2 Ill-defined and unknown Ill-defined and unknown COD 9.2 7.5 COD Table 4 describes five specific causes with high rate of mortality in both male and female. It can be recognized that almost all of these causes represent some large groups of disease such as CVD, cancer, Injury, communicable disease. Noticeably, the overall proportion of the non-communicable disease group including stroke, lung cancer and ischemic heart diseases was remarkably higher than the others. Pneumonia was the only representative of the com- municable disease category in this top five leading causes list with the quite low percentage (4.0%). There were statistically significant differences between male and female in stroke and lung cancer with opposing tendencies. While the number of death due to stroke in female was considerably higher than in male, the percentage of lung cancer death among male was more than 2 times higher than those among female. Table 4. Comparison of some leading CODs by sex Both sex Male Female Disease (n = 1365) (n = 835) (n = 530) p value Rate (%) Rate (%) Rate (%) Stroke 20.6 18.7 23.6 0.03 Lung cancer 8.3 10.4 4.9 < 0.001 Road traffic Injury 3.7 4.2 2.8 > 0.05 Pneumonia 4.0 3.5 4.7 > 0.05 Ischaemic heart disease 2.5 2.5 2.6 > 0.05 IV. DISCUSSION This study provides useful statistics contributing to the formation of an up to date mortality data at national level and reveals an empirical evidence of the current situation of deaths re- cording in the routine health management information system in Vietnam. By combining two steps in data collection: making death lists from traditional resources (A6 registers, Justice Clerk books and others) and then ascertaining this list by implementing verbal autopsy inter- JMR 111 E2 (2) - 2018 91
8. JOURNAL OF MEDICAL RESEARCH views, the completeness and the reliability was markedly higher [14]. Possibly, these of our results are at high level. The study differences came from the way of choosing also provides useful observations on the sample while estimates of GSO was based utility of VA methods to identify causes for on demographic models and projections, in deaths occurring outside public health facili- comparison with the locally measure CDR ties. Previously, this model has been used in from this study. Regarding to location, our several researches and was demonstrated findings figured out the significantly higher to be a very effective instrument for evalu- proportion of death in rural community than ating mortality patterns in Vietnam [11 - 13]. in the urban. The fact is that our sample The response rate of this study was population with higher fraction of urban cit- 92.6%, which seemed to be lower than izens could have led to this disparity since those in some foreign reports that found urban communities in Viet Nam probably rates reaching nearly 100% [2], but it was enjoy better health care and health status similar to previous experiences in use of than rural ones [11; 13]. In terms of gender, verbal autopsy surveys in Vietnam [10; 12]. as can be seen, the gap of death rates be- There are many reasons for this phenome- tween male and female showed a statisti- non. One of them had been mentioned by cally significant differential, similar to sever- Hoa et al [10] with missing cases mostly al previous reports [11; 10; 15]. atributable to migration of the household to Besides, our result also pointed out two another district/province after the deaths of third of the deaths occurred in the group of their family member. This explains why their above 60 years old (63,7%), similar to some names were recorded, but the data collec- others studies [14]. As mentioned in some tors could not find them. Another possible other researches, the disease pattern alter- reason for the difference between Vietnam ation has been happening in recent years and other countries comes from an insuf- in Vietnam, from communicable diseases to ficient mortality data recording system, as non- communicable diseases which occur Vietnamese network of primary healthcare mainly in the old people [10; 11; 16]. In ad- has yet to be developed, resulting in a sys- dition, Vietnamese life expectancy tends to temic lack of complete medical records [13]. rise due to better living conditions and the Mortality rate and some relating char- percentage of elder population in Vietnam acteristics society is increasing [6; 15]. Apparently, the The overall death rate of both two prov- combination of two above reasons results inces was 4.94‰, lower than findings from in the high rate of death among the elderly. another source of CDR of Vietnam pop- The low number of reported neonatal and ulation as National crude death rate in child deaths results in very low estimated 2014 from GSO which was estimated to be under 5 mortality rates for our study pop- 6.9‰ [8]. However, compared with the rate ulation. Similar apparent under-reporting of of 3.9‰ reported by Hoa et al in a region- under 5 mortality was also observed in the al research of death rate in 2012, our rate sample mortality surveillance system [10]. 92 JMR 111 E2 (2) - 2018
9. JOURNAL OF MEDICAL RESEARCH Apropos of the demand of treatment Causes of death before death, from our analysis, a majority An important part of this study is to in- of the deceased visited at least one health vestigate the cause of death. It can be facility at the end stage of life, similar to easily recognized the transition of disease several previous reports [13; 17]. Howerv- pattern in Vietnam in recent years from er, the highlight here is the level of medical communicable diseases to non-communi- services, more than 80% of the death went cable diseases (NCD) as being reported in to Central/ Provincial Hospitals. This illus- many researches [10; 14; 16]. Our result trates out several issues in Vietnam such as also revealed the same trend when three insufficient capability of diseases treatment out of the five leading causes of death were of primary healthcare system, the percep- stroke, lung cancer and ischemic heart dis- tion of people to put belief in the medical eases, which are categorized into the group service of high-level hospital. This pattern of NCD. A considerable point here is the may be different in other provinces in Viet- differentials in COD pattern between male nam where access to higher level hospi- and female. While in stroke, senility, or oth- tals is more difficult and where the patients er CVD, the proportions in females were may rely on traditional healers. However, remarkably higher than those in males, a although a few cases still kept medical re- reverse pat-tern was found for in liver can- cord after discharge of health facilities, most cer, lung cancer, cirrhosis of liver, where the of the deceased family members could re- percentages in males outweighed those in call the diagnosis in hospital. This finding females. It is possible that this difference suggests that a good system of discharge stems from habits among Vietnamese high- documents may help to improve the avail- er smoking and drinking males. Also, the ability of reliable information to ascertain the higher traffic injury among males compared cause of death from verbal autopsy. Instead to females can be explained by the preva- of only asking the relatives of the deceased lence of drunk driving among men. The ev- about the COD, the CHS health staffs can idence for this is the increasing proportion ask them to show the discharge documents of traffic accident during big national festi- (if they are available) to get more detailed vals conducive to alcohol overuse, mostly information to support the recording of the among males. There were also some caus- COD. Another aspect is the place of death, es of death related specifically to gender the mortality rate at home in our study was such as breast, cervix and ovary cancers in very high (88.1%), this feature is similar to females, which constitute the fifth cause in previous surveys in Vietnam [13; 14]. It is the list top ten leading causes of death. common practice in Vietnam for terminally V. CONCLUSION ill patients to go back home for the final pe- riod of their lives, even in some cases, the This study identified the mortality pattern patients are just hospitalized for a very short of two provinces in the Northern region of time. Vietnam contributing to the national mortali- JMR 111 E2 (2) - 2018 93
10. JOURNAL OF MEDICAL RESEARCH ty database. With the transition of the cause 7. Rao, C. (2013). Mortality estimates of death pattern, it is necessary to pay more for South East Asia, and INDEPTH mortality attention to collecting information of the surveillance: necessary but not sufficient? Int J Epidemiol, 42(4), 1196 - 1199. deaths outside health facilities as well as 8. GSO (2014). Tỷ suất tử vong. http:// improving the mortality recording system, www.gso.gov.vn. especially at primary healthcare network. 9. WHO (2008). International Statisti- Comprehensive mortality database, and in- cal Classification of Diseases and Related tervention programs need to prioritize these Health Problems: 10th Revision. 2: Instruc- leading causes of death. tion manual. 10. Hoa, N.P., Rao, C., Hoy, D. G. et Acknowledgements al. (2012). Mortality measures from sam- ple-based surveillance: evidence of the We would like to thank the Provincial epidemiological transition in Viet Nam. Bull Health Departments of Quang Ninh and World Health Organ, 90(10), 764 - 772. Thai Nguyen provinces for their help in 11. Ngo, A.D., Rao, C., Hoa, N. P. et al. data collection activities and all local health (2010). Mortality patterns in Vietnam, 2006: workers from commune health stations who Findings from a national verbal autopsy sur- directly conducted interview. Finally, we vey. BMC Res Notes, 3, 78. profoundly thank all respondents who spent 12. Huong, D.L., H.V. Minh, and P. Byass (2003). Applying verbal autopsy to time in sharing information for our research. determine cause of death in rural Vietnam. REFERENCES Scand J Public Health Suppl, 62, 19 - 25. 13. Hoa, N.P., Bích, P.T.N. (2009). Tình 1. WHO (2012). World Health Statistics. hình tử vong tại 7 tỉnh/ thành phố và một số 2. Fottrell, E., Byass, P. (2010). Verbal yếu tố liên quan. Tạp chí Y học Việt Nam, autopsy: methods in transition. Epidemiol 418(2), 64 - 68. Rev, 32, 38 - 55. 14. Hoa, N.P., Nhung, N.T.T. (2012). Tỷ 3. Mathers, C.D., Fat, D. M., Inoue, M. lệ và nguyên nhân tử vong tại một số tỉnh ở et al. (2005). Counting the dead and what Việt Nam năm 2008. Tạp chí nghiên cứu Y they died from: an assessment of the glob- học, 80(3C), 345 - 352. al status of cause of death data. Bull World 15. Chuc, N.T.K., Hoa, N.P. (2009). Mô Health Organ, 83(3), 171 - 177. hình tử vong tại huyện Ba Vì - Hà Nội qua 4. Setel, P.W., Macfarlane, S.B., Sz- hơn 9 năm theo dõi, 1999-2008. Tạp chí ng- reter,S. et al. (2007). A scandal of invisi- hiên cứu Y học, 61(2), 100 - 105. bility: making everyone count by counting 16. Huong, D.L., Minh, H. V., Vos, T. et everyone. The Lancet, 370(9598), 1569 - al. (2006). Burden of premature mortality 1577. in rural Vietnam from 1999-2003: analyses 5. Mahapatra, P., Shibuya, K., Lopez, from a Demographic Surveillance Site. Pop- A. D. et al. (2007). Civil registration sys- ul Health Metr, 4, 9. tems and vital statistics: successes and 17. Tran, B.H., Nguyen, H. T., Ho, H. T. missed opportunities. Lancet, 370(9599), et al. (2013). The Chi Linh Health and De- 1653 - 1663. mographic Surveillance System (CHILILAB 6. World Population Review (2014). HDSS). Int J Epidemiol, 42(3), 750 - 757. Vietnam Population 2014. 94 JMR 111 E2 (2) - 2018