ADULT MORTALITY DIFFERENTIALS BY GENDER AND REGIONS IN SURINAME IN - - PDF document

1

ADULT MORTALITY DIFFERENTIALS BY GENDER AND REGIONS IN SURINAME IN RECENT YEARS

Andrea Idelga Fernand Jubithana Anton de Kom Universiteit Van Suriname Andrea.jubithana-fernand@uvs.edu Bernardo Lanza Queiroz Departament of Demography Universidade Federal de Minas Gerais lanza@cedeplar.ufmg.br Abstract In this paper, we investigate the quality of vital records registration in Suriname and its main regions and investigate mortality differentials by gender and regions in the last two censuses. Suriname is one of the less populated countries in the world and producing adequate mortality estimates for the country is also a challenge. We use data from the 2004 and 2012 censuses and death counts from the Central Bureau of Citizen Issues (CBB). We evaluate quality of mortality data using the Death Distribution Methods. We find that, the urban coastal and rural coastal area have more disadvantage in mortality than the rural interior area and male mortality is almost twice female mortality for all regions. We also show differences in causes of deaths. Keywords: mortality differentials, Suriname, causes of death, data quality

2 Introduction Mortality differentials are present in all societies. Measuring and understanding differentials might help one to investigate trends and the evolution of life expectancy and the health conditions of different population sub-groups. Suriname is a small country and the second less populated in the America´s. There are not, to our knowledge, many studies on mortality and mortality differentials in the country. By being colonized by the Netherlands, some argue that vital records in the country are of good quality. In this paper, we, first, investigate the quality of vital records registration in the country and main regions and, second, analyse mortality differentials by gender and regions in the period between the last two censuses. Studies show that women are healthier than men, however the report of their health is worse

• n surveys (Case, A; Paxson, C, 2004). But, male mortality is higher than female in several

regions of the world. In the first place there may be sex differences in the distribution of chronic conditions as a result of biological, psychosocial or behavioural factors (Verbrugge, 1989; Lawlor et al, 2001; Molarius and Janson, 2002). In the second place women suffer from health conditions that contribute relatively little to mortality risk in relation to men who have health conditions which have large effects on the probability of death (Case, A; Paxson, C, 2004). On the other hand, studies have shown that women make more use of health care than men and that it may be the reason that they know more about their health and thus are more accurate health reporters (Verbrugge 1989, Idler 2003). There is some evidence about the view that men provided more complete information about their health in case of open-ended questions (MacIntyre, Ford and Hunt, 1999). Studies for developed countries have shown that the gap of life expectancy at birth between man and women have been narrowed for recent years (Glei, A; Horiuchi, S, 2007), due to changes in causes of death. In most of the developing countries like Brazil sex differences in mortality does not show reduction (Simoes, 2002; Queiroz et al, 2017). In the study of the United States (Preston, S; Wang, H, 2005) changes in sex mortality differentials is related to histories of cigarette smoking and has a cohort based structure. In most European countries narrowing sex differentials had been observed (Gjonca, et al., 2005)

3 Mortality differentials exist between and within countries (Kibele, 2012). Even between and within countries large and small differences in life expectancy at birth can be observed (Human Mortality Database 2008b). With respect to the differences in mortality by region the case of Germany (Kibele, 2012) shows variation in regional mortality. Historical studies also present differential mortality of the urban and rural area (Condran, G; Crimmins, E, 1980). For the United States, Fenelon (2012) shows an increasing gap in life expectancy and mortality risks across the northern and southern part of the country. Wilmoth, Boe, and Barbiere (2010) argued that there is a variation in health and mortality by race / ethnicity, socio-economic status, sex and geography in the United States of America. The geographic adult mortality differentials seem to be higher in the USA compared to Western Europe. In the mortality study of the USA (Fenelon, 2013) results show that high mortality is concentrated in space and clustered in the South and that this region is relatively poor with the presence of few health advantage. Suriname has a small scale population (541638 habitants in Census 2012) and differences exist in mortality between the urban coastal area, rural coastal area and the rural interior area. The country consists of ten districts whereby the urban coastal area covers the capital district Paramaribo and district Wanica with about 70% of the population. The rural coastal area (111224 habitants in Census 2012 consists of the districts Para, Commewijne, Saramacca, Nickerie and Coronie. The districts Marowijne, Sipaliwini and Brokopondo are part of the rural interior area (71268 habitants in Census 2012). It is relevant to mention that Suriname is characterized by international and internal

• migration. According to data of the General Bureau of Statistics (2011) the internal migrants

move more from the rural interior area to the urban coastal area, from the rural coastal area to the urban coastal area and from the rural interior to the rural coastal area. The internal migration in 2012 from rural interior area to urban coastal area and from rural coastal area to urban coastal area was 78.04 % and 77.49 % of the internal migrants, respectively. Data and Methods We investigate regional and gender differences in adult mortality in Suriname in the most recent period. The population data is from the Census 2004 and 2012 and the average death counts data between 2004 and 2012 is from the Central Bureau of Citizen Issues (CBB). In

4

• rder to estimate adult mortality we used traditional demographic methods called Death

Distribution Methods (DDM´s): 1) General Growth Balance (GGB) method proposed by Hill (1984); 2) Synthetic Extinct Generation (SEG) method proposed by Benneth and Horiuchi (1981); 3) Combined GGB-SEG method (Hill, You and Choi, 2009). The DDM methods require that assumptions about the population are made regarding the nature of the typical data errors 1) any change in census coverage had been proportionately constant by age, 2) no age misreporting of the population and deaths, and 3) proportionately constant omission of deaths by age (Hill, You, and Choi, 2009). An important assumption in the use of the methods GGB, SEG, and Hybrid GGB-SEG is that the population does not experience net migration. GGB method is a generalization of the Growth Balance method for stable populations proposed by Brass (1975) and it is derived from the population balance equation. The GGB is generalized for non-stable populations when two or more censuses are available (Hill, 1987). The GGB method is mathematically presented by the next equation:

 

) ( ) ( * * ln 1 ) ( ) ( ) (

2 1 2 1 2 1

             

 

x N x D C k k k k t x r x N x N (1) The slope 



C k k

2 1 2 1 *

• f the linear equation estimates the coverage of death recording

relative to an average of the coverage of the two censuses (HILL, 1987; HILL, K., 1987; HILL, K.; You, D.; TIMAEUS, I, 2003; HILL, K.; You, D.; CHOI, Y, 2009). Moreover, the intercept        

2 1

ln 1 k k t

• f the linear equation above represents the

age invariant change in census coverage between two censuses. The SEG method is based on the insight of Vincent (1951) that in a closed population with good reporting of deaths the population of age a at time t could be estimated by accumulating the deaths to that cohort after time t until the cohort was extinct (Hill, You and Choi, 2009). According to Hill, You, and Choi (2009) the SEG method calculates the estimated population that is composed of a sum of those who died multiplied by the sum of the growth rates of the

5 population in each age group divided by the observed population. See the mathematical notation in equation 2.

 

) 2 ( ) ( ) ( ) ( ) ( ) ( a N dx e x D a N a N a c

w a x dy y r

x a





  

  

The application of the combined GGB-SEG method Hill et.al (2009) is also conducted. The application of this method requires first the application of GGB to estimate the change in census coverage        

2 1

k k and then use the estimates to correct data of one or the two census. Secondly the SEG method is applied using the adjusted census population in order to obtain the level of coverage of the mortality data. The GGB and SEG methods give different responses to migration, whereby SEG is strongly affected by emigration, tending to underestimate death and overestimate mortality (Hill and Choi, 2004; Dorrington et al., 2011). According to Hill et al. (2009), the GGB, SEG and combined GGB-SEG methods underestimate coverage (overestimate adjusted mortality) in populations affected by immigration. The GGB method is more sensitive to coverage errors that change with age (Hill et al.,2009). The three methods require a closed population or small migration flows to improve the use of

• estimates. There are methodologies in the literature that allow dealing with this problem. A

simpler alternative, suggested by Hill et al (2009), is to consider only age groups that are not greatly influenced by migration flows. The most appropriate way of deciding which age interval to use in the production of under-registration estimates should involve the assessment

• f diagnostic charts produced by the GGB method. In this paper, we use a series of

combinations of age ranges and we pay special attention to cases that exclude peak migration ages. Since the literature does not indicate the best method to obtain estimates of data quality and mortality function, we use all methods with different variations of the age range to obtain our

• estimates. Hill (2017) argues that the combination of the GGB to the SEG using the age range

5 to 65 produces more robust estimates. We argue that for some countries, such as Suriname,

6 a procedure that applies all methods with different age ranges gives some robustness of our conclusions and allow us to test how variations in data quality could affect the methods and

• ur results.

Evaluation of the data quality In order to evaluate the population data of Suriname Whipple´s Index, Coale and Kisker (1986) measure and the index concentration in single ages (ICSC) are applied for Census 2004 and 2012. The results of the Whipple´s Index for Suriname reveal that the Census data

• f 2004 and 2012 present for male and female population values below 105 which indicates

that the data is very good. Values out of the acceptable interval (greater than 125) indicates that there exist a possibility on erros of age or sex, or selective missed and or age or sex of not registered migration (General Bureau of Statistics (GBS), 2013). Evaluation of the population data applying the Coale and Kisker (1986) measure (Table 1) shows that data of Suriname is reasonable to good for the Census 2004 and 2012. Comparison of the Surinamese data is done with data of other small scale population islands (Aruba, Curaçao, Jamaica and Cuba) which are considered to have good data according Luy (2010). Table 1: Coale and Kisker (1986) measure for population of Census 2004 and 2012 for Suriname

Source: GBS Census data Suriname 2004 and 2012

Results for the female popuation of Suriname are the closest to the data of Aruba, Curacao for group P95+/P70+ for Census 2004 and 2012. However, for the group P80+/P60+ the female data for both censuses is the closests to Aruba. Aanalyzing the data of the female population for the groups P90+/P60+ and P70+/P60+ we observe that the results of the Coale Suriname Census 2004 SEX P95+/P70+ P80+/P60+ P90+/P60+ P70+/P60+ Female 0.0099 0.1193 0.0114 0.4436 Male 0.0045 0.0943 0.0168 0.4249 Suriname Census 2012 SEX P95+/P70 P80+/P60+ P90+/P60+ P70+/P60+ Female 0.0104 0.1448 0.0196 0.4694 Male 0.0058 0.1220 0.0137 0.4462

7 & Kisker measure for Suriname (Census 2004 and 2012) result in values close to Curacao and Aruba, respectively. Male data for group P70+/P60+ is closest to the data of Curacao. Resuming male and female data of Census 2004 and 2012 of Suriname reveal that exageration of age does not occurred. It seems thus that age is well informed. Results of the age heaping index for the population of Suriname by sex for Census 2004 and 2012 presented in Figure 1 are considered to be good for the ages ending in “0” and “5” for young and adult ages. For the age 90 of the male population of census 2004 and census 2012 the highest values are observed, and thus indicating exageration of age. Exageration of age at 90 years for male in Suriname can be attributed to the small scale of the population and the real exageration of age. Figure 1: Index concentration in single ages for population Census 2004 and 2012 by sex for Suriname

Source: data GBS

For the main regions of Suriname the population census data of 2004 and 2012 is also considered to be good (Table2). Table 2: Whipple´s Index for central regions of Suriname, Census 2004 and 2012 REGIONS CENSUS 2004 CENSUS 2012 Male Female Total Male Female Total Urban Coastal 101.35 103.47 102.42 101.72 102.64 102.19 Rural Coastal 95.65 99.47 97.43 100.86 99.28 100.12 Rural Interior 103.68 96.33 99.98 103.98 100.30 102.08

Source: General Bureau of Statistics, Census 2004 and 2012 0.80 20.80 40.80 60.80 80.80 100.80 120.80 140.80 160.80 180.80 200.80 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 ICSC Ages Male 2004 Female 2004 Male 2012 Female 2012

8 The average death count data for the main regions of Suriname, presented in table 3, can be considered of reasonable quality after applying the Coale and Kisker measure (1986). For the male population only for some age groups (D95+/D70+, D80+/D60+, D90+/D60+ and D70+/D60+) the values of the rural interior area are higher than for the other regions. Furthermore, the values for the main regions of Suriname are also between the values of Aruba and Curacao for the male population. But for the female population the values of the main regions of Suriname are close to that of Jamaica. Table 3: Coale and Kisker (1986) measure for the average death count data 2004-2012, in the main regions of Suriname UrbanCoastal D100+/D80+ D100+/D70+ D95+/D70 D80+/D60+ D90+/D60+ D70+/D60+ Female 0.0146 0.0078 0.0407 0.4024 0.1001 0.7520 Male 0.0094 0.0039 0.0196 0.2805 0.0527 0.6797 Rural Coastal D100+/D80+ D100+/D70+ D95+/D70 D80+/D60+ D90+/D60+ D70+/D60+ Female 0.0157 0.0157 0.0220 0.3230 0.0600 0.7310 Male 0.0110 0.0106 0.0140 0.2714 0.0333 0.6938 RuralInterior D100+/D80+ D100+/D70+ D95+/D70+ D80+/D60+ D90+/D60+ D70+/D60+ Female 0.0183 0.0101 0.0510 0.4379 0.1081 0.7938 Male 0.0103 0.0052 0.0209 0.3784 0.0729 0.7432

Source: CBB death count data Suriname 2004-2012

Completeness of death registration relative to the population. For the male population completeness of death registration relative to the population is mainly above the unity varying between 2% and 31%, applying the DDM´s with age segments excluding high proportion of peak migration. The rural coastal area presents under registration between 5% and 18% applying the GGB and GGB-SEG. Over registration in the rural coastal area is observed for male between 9%-18%, applying SEG. For male population under-registration is about 17% in the rural interior area applying SEG. However, applying GGB and GGB-SEG depending on the age segment excluding high proportion of peak

9 migration under (13% -28%) or over (4% -37%) registration is observed in the rural interior area. Application of the different DDM´s using age segments excluding high proportion of peak migration for female population shows that in the urban coastal area completeness of death registration relative to the population is above (1.0374 – 1.2774) or under (0.9617 – 0.9822) the unity. This result can be explained by the in-migration of the population in the urban coastal area. However in the rural coastal area under registration is notable between 14% and 20%, applying the GGB and GGB-SEG method. The application of the SEG method for female population in the rural coastal area results in completeness of death registration to

• ver registration of about 4% -7%. The rural interior area is characterized by under

registration between 23% -43%, depending on the application of the DDM method and age segment excluding high proportion of peak migration. Sex differences in probabilities of dying between 15 and 60 years of age (45q15) on country level In order to analyse sex differentials of mortality, the ratio of probabilities of dying between 15 and 60 years of age, male by female are presented for Suriname and its main regions. The 45q15 are determinate using different Death Distribution Methods and age segments including and excluding high proportion of peak migration. Figure 2 shows that for the country the ratio of 45q15 male by female for age segments excluding peak migration (30+ to 65+, 30+ to 75+, 35+ to 65+ and 35+ to 75+) the methods GGB, SEG, GGB-SEG observed and GGB-SEG adjusted have nearly the same ratios. In other words, differentials in mortality between male and female applying the DDM´s with the above age segments indicate that the applied methods produce nearly the same ratios.

10 Figure 2: Ratio probabilities of dying between 15 and 60 years, male by female population of Suriname, 2004-2012

Source: Data GBS and CBB

However, applying the different DDM´s with age segments including high proportion of peak migration only age segment 15+ to 75+ presents approximately the same ratios 45q15 male by female. A probably explanation for this result may be the length of the age segment. The

• ther age segments with high proportion of peak migration present results with more

dispersion of the ratios between the different DDM´s. It is important to mention that the ratio male by female 45q15 is between 1.54 and 1.85 in Suriname. The different DDM´s have the strong assumption of zero net migration which results in less and more dispersion of the ratios, using age segments without high proportion and with high proportion of peak migration, respectively. In the process of analysing the 45q15 it is important to consider causes of death. In recent years (2008-2013) main causes of death for adult mortality were: 1) cardio vascular disease; 2) External causes of death; 3) Diabetes Mellitus and 4) Neoplasm. Figure 3 presents the sex ratios for main causes of death for adults in Suriname in recent years. The results of the ratio

• f 45q15 male by female above the unity can be better understood, considering the period

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 15+ to 55+ 5+ to 65+ 15+ to 75+ 10+ to 60+ 15+ to 65+ 30+ to 55+ 30+ to 65+ 30+ to 75+ 35+ to 65+ 35+ to 75+ Ratio 45q15 male by female Age segments GGB SEG GGB-SEG observed GGB-SEG adjusted 45q15 male= 45q15 female

11 results of causes of death of recent years. External Causes of death present at least twice sex ratios in comparison to the three other main causes of death. Figure 3: Sex ratios causes of death for adult mortality 2008-2013

Source: Data GBS and CBB

Sex differences in probabilities of dying between 15 and 60 years of age (45q15) for the main regions in Suriname. For the main regions differences in probabilities of dying 45q15 are diverse between the urban coastal area, rural coastal area and rural interior. The rural interior area with the lowest Human Development index (0.599 in 2009) presents results of the ratios 45q15 male by female population different than for the urban coastal and rural coastal area. Results of the ratio 45q15 male by female of the urban coastal and rural coastal area are for the age segments excluding high proportion of peak migration opposite in terms of dispersion. The ratios 45q15 male by female applying DDM´s without high proportion of peak migration are more dispersed in the urban coastal area compared to the rural coastal area (figure 4).

0.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 400.00 2008 2009 2010 2011 2012 2013 Sex ratios Years Cardio vasculair disease External causes of death Diabetus Melitus Cancer

12 Figure 4: Ratio probabilities of dying between 15 and 60 years, male by female Population urban coastal area Suriname, 2004-2012

Source: Own elaboration based on Data GBS and CBB, 2004-2012

Moreover in the urban coastal area for the different DDM´s the ratios male by female 45q15 are overall dispersed for the age segments including and excluding high proportion of peak

• migration. The urban coastal area is the area which receive, almost all the migrants from the

rural coastal and rural interior area. For the GGB-SEG method the ratios 45q15 male by female are lower than in the case of the rural coastal area (figure 5).

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 15+ to 55+ 5+ to 65+ 15+ to 75+ 10+ to 60+ 15+ to 65+ 30+ to 55+ 30+ to 65+ 30+ to 75+ 35+ to 65+ 35+ to 75+ Ratio 45q15 male by female Age segments GGB SEG GGB-SEG observed GGB-SEG adjusted 45q15 male = 45q15 female

13 Figure 5: Ratio probabilities of dying between 15 and 60 years, male by female population rural coastal area of Suriname, 2004-2012

Source: Own elaboration based on Data GBS and CBB, 2004-2012

Considering analyse of ratios 45q15 male by female in the rural interior area it is remarkable that dispersion for all the age segments are notable. Besides that the ratios 45q15 male by female are for some age segment (15+ to 55+, 30+ to 55+and 35+ to 65+) applying the GGB and GGB-SEG adjusted close above or under the unity (45q15 male=45q15 female). This result indicates that 45q15 for male and female are nearly the same in the rural interior area by application of the two mentioned above methods and that female has a disadvantage in the

• mortality. A plausible explanation may be that the rural interior area is characterized by

mainly outmigration of adult male population. The GGB-SEG observed method presents the highest ratios male by female 45q15 indicating greater differences in male and female 45q15 (figure 6). As literature (Hill, You, and Choi (2009) argues SEG is sensitive to migration and GGB is more sensitive to age reporting.

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 15+ to 55+ 5+ to 65+ 15+ to 75+ 10+ to 60+ 15+ to 65+ 30+ to 55+ 30+ to 65+ 30+ to 75+ 35+ to 65+ 35+ to 75+ Ratio 45q15 male by female Age segments GGB SEG GGB-SEG observed GGB-SEG adjusted 45q15 male= 45q15 female

14 Figure 6: Ratio probabilities of dying between 15 and 60 years, male by female population rural interior area Suriname, 2004-2012

Source: Own elaboration based on data GBS and CBB 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 15+ to 55+ 5+ to 65+ 15+ to 75+ 10+ to 60+ 15+ to 65+ 30+ to 55+ 30+ to 65+ 30+ to 75+ 35+ to 65+ 35+ to 75+ Ratio 45q15 male by female Age segments GGB SEG GGB-SEG observed GGB-SEG adjusted 45q15 male= 45q15 female

15 Geographic differences in probabilities of dying between 15 and 60 years of age (45q15) The three main regions in Suriname (urban coastal, rural coastal and rural interior area) present differential in socio- economic development. Access to basic and specialized health services are better in the urban coastal area than in the rural coastal and rural interior area. Suriname had in December 2016 a GDP of about 7662 USD1. The economy of Suriname depends nowadays mostly on the gold and oil industry and in the past on Alumina, bauxite,

gold and oil, which have made up 75% of total exports.2Labour force information is

presented in table 4. As can be seen in Census 2012 the difference between male and female unemployment was about 3.52. Nearly one and a half time more men were employed in Suriname in both censuses in relation to female. Table 4 : Labour force information of population 15-64 in Suriname in Census 2004 and 2012 Sex CENSUS 2004 CENSUS 2012 Employed Unemployed Employed Unemployed Male 65.82 7.81 68.50 6.11 Female 35.52 9.66 39.42 9.63 Total 50.69 8.77 53.84 7.88

Source: Census data 2004 and 2012, GBS

As the literature reveals education is linked to mortality in the sense that the association between education and health survival are well established (Ross, C.; Masters, R.; and Hummer, R , 2012). In Suriname mean years of schooling of the population in 2009 in the ten districts vary between 2.9 years in district Sipaliwini (rural interior area) to 9.7 years in the capital district Paramaribo. Figure 7 presents information on the mean years of schooling and the Human Development index for the ten districts. Two districts of the rural interior area (Brokopondo and Sipaliwine) produce the lowest mean years of schooling and HDI. The districts of the rural coastal area have more dispersion in terms of years of schooling, however for the HDI their values lie nearly on one vertical line. District Wanica and Paramaribo (urban coastal area) are dispersed in terms of mean years of schooling and HDI.

1 https://tradingeconomics.com/suriname/indicators (accessed on September 29th in 2017)

2 http://www.worldbank.org/en/country/suriname/overview (accessed September 29th in 2017)

16 Figure 7: Mean years of schooling versus Human Development index (HDI) Suriname, 2009

Source : Human Development Atlas Suriname 2013

Figure 8: Map of Suriname

Paramaribo (9.7) Wanica (9.1) Commewijne (8.2) Coronie (7.8) Saramacca (8.7) Nickerie (8.6) Marowijne (8.1) Para (6.8) Brokopondo (5.4) Sipaliwini (2.9) 2 4 6 8 10 12 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 Mean years of schoolingi HDI mean years of schooling

17

Source: http://suriname.startsuper.nl/14455-landkaart-suriname.html

Application of the GGB, SEG and GGB-SEG with age segments including and excluding high proportion of peak migration have produced for the different geographic areas for male and for female population different 45q15. Results show that for male and female population differential in 45q15 are overall higher in the urban coastal area compared to the rural interior

• area. Moreover 45q15 are higher when the different DDM´s are applied with age segments

excluding high proportions of peak migration. The lower 45q15 in rural interior area compared to the coastal areas are not an expected result, because of its moderate socio– economic development. Considering the fact that the rural interior area has a small scale population (71268 in Census 2012) and is characterized by outmigration results of lower 45q15 can be better understood. In case of the male population (Table 5) application of GGB-SEG adjusted, age segment 35+ to 75+ produce the highest 45q15 (0.2053) in the rural interior area, but still a lower value than for the urban and rural coastal area.

18 Table 5: Probability of dying by DDM´s for male population of Suriname, 2004-2012, by age segments including and excluding high proportion of peak migration MALE MAIN REGIONS SURINAME METHODS GGB SEG GGB-SEG (observed according to age segments) GGB-SEG (adjusted according to age segments) Urban Coastal Age segments excluding proportions Peak Migration Probabilities of dying

15 45 q

30+ to 55+ 0.1945 0.2188 0.2505 0.1923 30+ to 65+ 0.2146 0.2202 0.2486 0.2243 30+ to 75+ 0.2303 0.2231 0.2469 0.2307 35+ to 65+ 0.2200 0.2194 0.2477 0.2232 35+ to 75+ 0.2335 0.2228 0.2460 0.2335 MEAN 0.2186 0.2210 0.2402 0.2235 SDV 0.0155 0.0020 0.0176 0.0108 Unadjusted

15 45 q

0.2494 0.2494 _ _ Rural Coastal Age segments excluding proportions Peak Migration Probabilities of dying

15 45 q

30+ to 55+ 0.2189 0.1834 0.2113 0.2156 30+ to 65+ 0.2360 0.1897 0.2098 0.2372 30+ to 75+ 0.2455 0.1960 0.2087 0.2341 35+ to 65+ 0.2366 0.1932 0.2097 0.2280 35+ to 75+ 0.2465 0.1997 0.2084 0.2349 MEAN 0.2367 0.1924 0.2096 0.2300 SDV 0.0111 0.0062 0.0011 0.0087 Unadjusted

15 45 q

0.2168 0.2168 _ _ Rural Interior Age segments excluding proportions Peak Migration Probabilities of dying

15 45 q

30+ to 55+ 0.1262 0.1959 0.1697 0.1589 30+ to 65+ 0.1314 0.1904 0.1676 0.1621 30+ to 75+ 0.1826 0.1920 0.1647 0.2019 35+ to 65+ 0.1306 0.1879 0.1694 0.1633 35+ to 75+ 0.1857 0.1902 0.1641 0.2053 MEAN 0.1513 0.1913 0.1671 0.1783 SDV 0.0301 0.0030 0.0026 0.0232 Unadjusted

15 45 q

0.1640 0.1640 _ _

Source: Data CBB and GBS

19 Analysing the mean of 45q15 for the different main regions it is obvious that the urban and rural coastal area have the highest values compared to the rural interior area. The highest difference in mean of 45q15 for male population is 0.854, applying the GGB method and it is between the rural coastal area and rural interior area. The GGB-SEG observed presents the highest difference in mean 45q15 (0.731) between the urban coastal and rural interior area indicating a higher disadvantage in mortality for male in the urban coastal area. Table 6 presents the difference in mean for the different DDM´s and regions for male population. The

• nly advantage of male regarding adult mortality in the urban coastal area is compared to the

rural coastal area and it is when applying the GGB and GGB-SEG Adjusted. Table 6: Differences between main regions in mean of 45q15 for male population Difference between areas Differences in mean GGB SEG GGB-SEG Observed GGB-SEG Adjusted Urban Coastal – Rural Coastal 0.181 (-) 0.286 (+) 0.031 (+) 0.007 (-) Urban Coastal – Rural Interior 0.673 (+) 0.297 (+) 0.731 (+) 0.452 (+) Rural Coastal – Rural Interior 0.854 (+) 0.001 (+) 0.425 (+) 0.517 (+)

Source: Data GBS and CBB Note: The positive sign (+) means greater 45q15 for the first term in the difference; The negative sign (-) means a smaller 45q15 in the first term of the difference

For the female population (Table 7) application of GGB-SEG adjusted, age segment 30+ to 75+ presents the highest 45q15 in the rural interior area (0.1750) and also the highest 45q15 for all the areas. Applying the GGB and SEG method considering age segments excluding high proportion of peak migration 45q15 for female population is lower in the urban coastal area compared to the rural coastal and rural interior area. The GGB-SEG observed results in 45q15 higher in the urban coastal and rural coastal area compared to the rural interior area. However the GGB-SEG adjusted provides the highest 45q15 in the rural interior area. The female population in the rural interior area presents some advantage in mortality compared to the urban coastal area and rural coastal area applying the GGB-SEG observed (Table 7).

20 Table 7: Probability of dying by DDM´s for female population and main regions of Suriname, 2004-2012, by age segments excluding high proportion of peak migration FEMALE MAIN REGIONS SURINAME METHODS GGB SEG GGB-SEG (observed according to age segments) GGB-SEG (adjusted according to age segments) Urban Coastal Age segments excluding proportion Peak Migration Probabilities of dying

15 45 q

30+ to 55+ 0.1129 0.1148 0.1361 0.1329 30+ to 65+ 0.1098 0.1153 0.1363 0.1186 30+ to 75+ 0.1154 0.1166 0.1358 0.1374 35+ to 65+ 0.1116 0.1152 0.1361 0.1334 35+ to 75+ 0.1165 0.1166 0.1356 0.1385 MEAN 0.1132 0.1157 0.1360 0.1322 SDV 0.0027 0.0008 0.0003 0.0080 Unadjusted

15 45 q

0.1369 0.1369 _ _ Rural Coastal Age segments excluding proportion Peak Migration Probabilities of dying

15 45 q

30+ to 55+ 0.1663 0.1235 0.1308 0.1605 30+ to 65+ 0.1560 0.1267 0.1315 0.1565 30+ to 75+ 0.1554 0.1295 0.1315 0.1521 35+ to 65+ 0.1553 0.1290 0.1316 0.1524 35+ to 75+ 0.1552 0.1317 0.1315 0.1518 MEAN 0.1576 0.1281 0.1314 0.1547 SDV 0.0049 0.0031 0.0003 0.0038 Unadjusted

15 45 q

0.1353 0.1353 _ _ Rural Interior Age segments excluding proportion Peak Migration Probabilities of dying

15 45 q

30+ to 55+ 0.1161 0.1260 0.1055 0.1473 30+ to 65+ 0.1301 0.1275 0.1047 0.1372 30+ to 75+ 0.1308 0.1292 0.1047 0.1750 35+ to 65+ 0.1347 0.1274 0.1042 0.1621 35+ to 75+ 0.1323 0.1293 0.1045 0.1622 MEAN 0.1288 0.1279 0.1047 0.1568 SDV 0.0073 0.0014 0.0005 0.0147 Unadjusted

15 45 q

0.1057 0.1057 _ _

Source: Data CBB and GBS

21 Table 8 presents the difference in mean 45q15 for the different methods between the main regions for female population. Results show that for female population the mean 45q15 (table 7) is higher in the rural coastal and rural interior area compared to the urban coastal area applying GGB, SEG and GGB-SEG adjusted. Summarizing for female population the urban coastal area has overall an advantage in mortality compared to the rural areas, which is an expected result. Table 8: Differences between main regions in mean of 45q15 for female population. Difference between areas Differences in mean GGB SEG GGB-SEG Observed GGB-SEG Adjusted Urban Coastal – Rural Coastal 0.044 (-) 0.012 (-) 0.005(+) 0.023(-) Urban Coastal – Rural Interior 0.156 (-) 0.012(-) 0.031(+) 0.025(-) Rural Coastal – Rural Interior 0.288 (+) 0.000 (+) 0.027 (+) 0.002(-)

Source: Data GBS and CBB Note: The positive sign (+) means greater 45q15 for the first term in the difference; The negative sign (-) means a smaller 45q15 in the first term of the difference

Conclusion In this paper, we evaluate the quality of mortality data in Suriname and studies regional and gender differentials in adult mortality. The analysis suggests that the mortality information collected in Suriname is an important source of mortality studies, especially for population

• subgroups. Although there are problems associated with data sources and demographic

methods, data quality in Suriname allows interesting regional and temporal analysis. We also find a large difference in male and female mortality and across regions of the country. Regarding the Death Distribution Methods , it should be noted that there is no gold standard to be adopted and the different methods can present different results due to violation of their assumptions and other problems. In the case of this paper, an important limitation refers to the assumption of population closed to the regions of the country. However, when we apply the different variations of death distribution methods using different age groups as an adjustment, seeking to minimize the effects of international and /or internal migration and errors in the age declaration, we believe that our results are more robust than others available.

22 Since we are working with small number of cases, in the next steps of the papers we might try to smooth death counts to adjust for possible random fluctuations specially for the regional analysis. One option is to use the method proposed and applied to Brazil by Gonzaga and Schmertmann (2016). Differential in adult mortality in Suriname are notable by sex and regional. As expected, female mortality is depending on the applied DDM´s and age segment in some cases half time lower than male mortality. Regional differential in adult mortality show that in case of male population applying the GGB and GGB-SEG, the urban coastal and rural coastal area have more disadvantage in mortality than the rural interior area. In case of the application of SEG the 45q15 for male population in the rural coastal and rural interior area are almost the

• same. The advantage of 45q15 for the female population in the urban coastal and rural coastal

area in relation to the rural interior area applying, GGB, SEG and GGB-SEG Adjusted are an expected result considering socio and economic development of those regions. It is important to consider that advantage of male mortality in some cases in the rural interior area can be explained by more outmigration of the male population and the small scale population in that area. Completeness of death registration relative to the population in the rural interior is also a result of outmigration. The rural coastal area and rural interior area present the highest differences in 45q15 for male and female population using GGB method, which is less sensitive to migration. However, for the GGB-SEG observed the highest differences in 45q15 are between the urban coastal area and the rural interior area, with greater advantage of mortality for male than female in the urban coastal area. Application of the SEG method shows no difference in 45q15 between the rural interior and rural coastal area for female population.

23 References BENNETT, N. G.; HORIUCHI, S. Mortality estimation from registered deaths in less developed countries. Demography, v.21, 1984, p.217-234. CASE, A.; PAXON, C. Sex differences in morbidity and mortality. Centre of health and well- being. Princeton University, 2004. FENELON, Andrew. Geographic divergence in mortality in the United States. Population and development review, v. 39, n. 4, p. 611-634, 2013. GENERAL BUREAU OF STATISTICS (Suriname) - GBS / Censuskantoor. Zevende Algemene Volks- en Woningtelling in Suriname. LANDELIJKE RESULTATEN. Volume I. Demografische en Sociale karakteristieken. Augustus 2005. GENERAL BUREAU OF STATISTICS (Suriname) - GBS / Censuskantoor. Resultaten Achtste Volks- en Woningtelling in Suriname (Volume I). Demografische en Sociale karakteristieken en Migratie. September 2013. GJONCA, A. C. TOMASSINI, B. TOSON, and S. SMALLWOOD. Sex differences in Mortality, A comparison of the United Kingdom and other developed countries. Health statistics quarterly 26:6-16. GLEI, A; HORIUCHI, S. The narrowing sex differential in life expectancy in high-Income population: Effects of differences in the age pattern of mortality. Population studies, Vol.61, No2, 2007, pp.141-159 Gonzaga, M. R., & Schmertmann, C. P. (2016). Estimating age-and sex-specific mortality rates for small areas with TOPALS regression: an application to Brazil in 2010. Revista Brasileira de Estudos de População, 33(3), 629-652. HILL, K. Estimating census and death registration completeness. Asian and Pacific Population Forum, v. 1, n.3, 1987, p. 8 - 13, 23, 24.

24 HILL, K. Methods for measuring adult mortality in developing countries: a comparative

• review. The Global Burden of Disease 2000 in Aging Populations. Research paper, 2001,

n.1.13. HILL, K. Analytical methods to evaluate the completeness and quality of death registration: Current state of knowledge. United Nations Population Division-Technical paper 2017/02. HILL, K.; You, D.; CHOI, Y. Death distribution methods for estimating adult mortality: sensitivity analysis with simulated data errors. Demographic Research, v.21/9, 2009, p.235- 253 HILL, K.; CHOI Y.; TIMAEUS, I. Unconventional approaches to mortality estimation. Demographic Research, v.13, 2005, p.281-300. KIBELE, E.; Regional mortality differences in Germany, Demographic Research Monographs, Springer Science+ Business Media Dordrecht, 2012. LUY, Marc.; A classification of the Nature of Mortality Data Underlying the Estimates for the 2004 and 2006 United Nations `World Population Prospects. Comparative Population

• Studies. Federal Institute for Population Research. Vol.35, 2(2010): 315-334.

PAMPEL, F.; Cigarette use and the narrowing sex differential in mortality. Population and development Review, Vol.28, No1, 2002, pp.77-144. PRESTON, S.; WANG, H. Sex differentials in the United States: the role of cohort smoking

• patterns. Working Paper, 2005.

QUEIROZ, Bernardo Lanza et al. Completeness of death-count coverage and adult mortality (45q15) for Brazilian states from 1980 to 2010. Revista Brasileira de Epidemiologia, v. 20, p. 21-33, 2017. ROSS, Catherine et al. Educationand gendergaps in health and mortality. Demography, 49 (4): 1157-1183, 2012.

25 TIMAEUS, I. M.; Measurement of adult mortality in less developed countries: A comparative review. Population Index, v.57, n.4, p.552-568, 1991. Wilmoth JR, Boe C, Barbieri M. Geographic Differences in Life Expectancy at Age 50 in the United States Compared with Other High-Income Countries. In: National Research Council (US) Panel on Understanding Divergent Trends in Longevity in High-Income Countries; Crimmins EM, Preston SH, Cohen B, editors. International Differences in Mortality at Older Ages: Dimensions and Sources. Washington (DC): National Academies Press (US); 2010. 12.

Available from: https://www.ncbi.nlm.nih.gov/books/NBK62591/