WITH CRONIC MORBIDITY - THE EFFECT OF INCOME: A COMPARATIVE STUDY OF - - PDF document

WITH CRONIC MORBIDITY - THE EFFECT OF INCOME: A COMPARATIVE STUDY OF CERTAIN CAUSES OF MORBIDITY IN BRAZIL AND THE US

Kaizô Iwakami Beltrão† Sonoe Sugahara‡

† EBAPE/FGV, kaizo.beltrao@fgv.br ‡ ENCE/IBGE, sonoe.pinheiro@ibge.gov.br

ii

ABSTRACT

This study analyzes the prevalence of certain causes of chronic diseases morbidity conditions in Brazil and the US. Most conditions are highly correlated with socio-economic strata. Socio-economic strata boundaries, however, are quite difficult to define and different authors would have different definitions of what these strata boundaries are. Among the several options available we opted in this text to use per capita family income as a proxy for social strata. Given this, most information displayed here is also disaggregated by per capita family income (five strata). A comparison of similar statistics on self-reported morbidity in Brazil and the US are computed and analyzed. For reported morbidity, Brazilian data were drawn from PNS 2013. US information

• n the subject was drawn from NHIS 2012/13/14. In both surveys morbidity

conditions are self-reported. To take into account age/sex distribution that can heavily influence such rates, standardization procedures were undertaken using the Brazilian and the American population. The relative position of the regions and of the social strata changed in some instances after the standardization. Age/sex specific rates were also analyzed, broken down as well by the five income brackets. Though the general shape of the curves corresponding to Brazil and the States are similar, they are not quite the same, varying in concavity and inflection points and, sometimes, even in the direction of the trend for some segments.

iii

SUMMARY

ABSTRACT ........................................................................................................................... ii I - Introduction ........................................................................................................................ 1 II - Concepts ........................................................................................................................... 2 III - Self-reported Brazilian prevalence of selected morbidity conditions by age/sex/income

• Comparison to similar statistics in the US ........................................................................... 6

III.1 – Arthritis/Rheumatism ............................................................................................... 6 III.2 – Diabetes .................................................................................................................. 13 III.3 – Bronchitis/Asthma .................................................................................................. 20 III.4 – Hypertension .......................................................................................................... 26 III.5 - Heart Condition ....................................................................................................... 33

• III. 6 - Stroke ..................................................................................................................... 40
• III. 7 - COPD

..................................................................................................................... 47 II – Final Comments ............................................................................................................. 54 BIBLIOGRAPHY ............................................................................................................... 54 ANNEX A – MAIN CHARACTERISTICS OF THE SURVEYS ................................. 57 A.1 Main characteristics of PNS ....................................................................................... 57 A.2 Main characteristics of NHIS ..................................................................................... 57 ANNEX B – POPULATION DISTRIBUTION – BRAZIL ............................................ 59 ANNEX C – POPULATION DISTRIBUTION – USA ................................................... 63 ANNEX D – GRAPHS OF CRUDE AND STANDARDIZED RATES BRAZIL & THE US .......................................................................................... Erro! Indicador não definido. D 1 – Arthritis/Rheumatism ............................................ Erro! Indicador não definido. D 2 – Diabetes ................................................................. Erro! Indicador não definido. D 3 – Bronchitis/Asthma ................................................ Erro! Indicador não definido. D 4 – Hypertension ......................................................... Erro! Indicador não definido. D 5 - Heart Condition ..................................................... Erro! Indicador não definido. D 6 - Stroke ..................................................................... Erro! Indicador não definido. D 7 - COPD ..................................................................... Erro! Indicador não definido.

1

I - Introduction

Although infectious diseases are important death causes, there is a significant growth of non-communicable chronic diseases: cardio-vascular, bronchitis/asthma, diabetes, cancer, mental problems, Chronic Obstructive Pulmonary Disease (COPD) etc. There is a vast literature showing health and death inequalities among social classes. For example, the Registrar General of England and Wales has been collecting data on occupational mortality related to every decennial census since 1851 (Coggon et al., 2009). They show that the social gap has been widening. The first documented study on mortality differentials seems to be due to Stevenson (1928). Other countries followed suit: Australia (AIHW 2005), Germany (GEYER and PETER. 1999), Finland (MARTIKAINEN, 1995), Korea (Lee et al., 2016). Mortality differentials are also due to differences in lifestyle such as smoking and obesity, between the socio-economic groups. Those in lower socio-economic groups are in disadvantage: smoke more and more often, follow a poorer diet with less fruits and greens, consume more alcohol, exercise less often and are more often overweight. This study analyzes the prevalence of certain causes of morbidity conditions in Brazil and the US. As seen in the literature, most conditions are highly correlated with socio-economic

• strata. Socio-economic strata boundaries, however, are quite difficult to define and different

authors would have different definitions of what these strata boundaries are. Among the several options available, we opted in this text to use per capita family income as a proxy for social strata. Wilkinson and Pickett (2007) claim that relative deprivation is a core mechanism for why income inequality affects health in societies. Given this, most information displayed here is also disaggregated into five per capita family income brackets of equal size. Similar income brackets are used for the US. A comparison of statistics on self-reported morbidity in Brazil and the US are computed and analyzed. In the second chapter, we define the rates used in the comparisons and in the graphs analyzed in the study. In the third chapter, we tabulate and comment self-reported Brazilian prevalence of seven selected morbidity conditions by age/sex/income, namely: Arthritis/Rheumatism, Diabetes, Bronchitis/Asthma, Stroke, Hypertension, Heart condition

2 and COPD. A comparison to similar US statistics is also shown. Prevalence is reported on a crude and age-specific basis, as well as standardized both by Brazilian and by the US population distribution. Crude specific and standardized rates are present in tables in the main text and their corresponding graphs in Annex D. Data used for health conditions in the US are from the 2012, 2113 and 2014 NHIS - National Health Interview Survey. NHIS is a sample household survey conducted annually by the US Department of Health and Human Services. Corresponding data for Brazil are from PNS 2013. PNS is also a sample household survey and was conducted by IBGE (see Annex A for a brief description of these surveys). The authors made all tabulations using SPSS (Statistical Package for Social Sciences). Five-year age groups were used with an 80- year-and-over open-ended age interval. Both surveys consider morbidity solely for adults

• interviewees. As a proxy for socio-economic strata we used per capita family income.

Income from all sources was used, there included: salary, pension benefits, rents, revenues from investments, etc. Since Brazilian income distribution is so biased and in a completely different level from that of the US, we decided to use quintiles as the grouping factor. Graphs with age/sex distribution of the Brazilian population disaggregated by per capita family income in the above-mentioned brackets are available in Annex B. We are supposing that poverty is a relative concept. Graphs with the age/sex distribution of the US population disaggregated into quintiles of per capita family income are in Annex C.

II - Concepts

Prevalence rates were computed as the ratio of persons of a given age group, sex and per capita family income bracket having declared to be suffering a specific morbidity condition and all persons in this same group: i) ) , , ( ) , , , ( ) , , , ( r s a pop m r s a pop m r s a P  for age/sex/income specific morbidity rate; ii)    

a a

r s a pop m r s a pop m r s P ) , , ( ) , , , ( ) , , , ( for crude morbidity rate by sex and income bracket;

3 iii)     

r a r a

r s a pop m r s a pop m s P ) , , ( ) , , , ( ) , , , ( for crude morbidity rate by sex; iv)     

s a s a

r s a pop m r s a pop m r P ) , , ( ) , , , ( ) , , , ( for crude morbidity rate by income bracket; and v)      

s r a s r a

r s a pop m r s a pop m P ) , , ( ) , , , ( ) , , , ( for crude morbidity rate; where ) , , , ( m r s a P is the prevalence rate of m morbidity, for age group a, sex s and living in households1 with level r per capita family income; ) , , ( r s a pop is the population of age group a, sex s and living in families in the per capita income bracket r; and ) , , , ( m r s a pop is the subset of this population suffering from m morbidity. In the formulae, the absence of a parameter indicates the summation over all groups. For example, the absence of parameter r (income) indicates all levels of income and the absence of parameter a (age group) indicates all ages. To take into account age/sex distribution, which can heavily influence such rates, a standardization procedure was undertaken using Brazilian and US populations. Standardized prevalence rates were computed as the weighted average of the prevalence rates using as weights the age/sex distribution of a given population: i)    

a p a p p

r s a pop r s a pop m r s a P m r s P ) , , ( ) , , ( * ) , , , ( ) , , , ( for standardized morbidity rate by sex and income bracket; ii)     

r a p r a p p

r s a pop r s a pop m r s a P m s P ) , , ( ) , , ( * ) , , , ( ) , , , ( for standardized morbidity rate by sex;

1 IBGE differentiate households and families in the household, but we are using the terms interchangeably,

but meaning the larger unit.

4 iii)     

s a p s a p p

r s a pop r s a pop m r s a P m r P ) , , ( ) , , ( * ) , , , ( ) , , , ( for standardized morbidity rate by income bracket; and iv)      

s r a p s r a p p

r s a pop r s a pop m r s a P m P ) , , ( ) , , ( * ) , , , ( ) , , , ( for standardized morbidity rate; where ) , , , ( m r s a P is the smoothed prevalence rate of m morbidity, for age group a, sex s and per capita family income bracket r, ) , , , ( m r s Pp  is the standardized prevalence rate of m morbidity, for sex s and per capita family income bracket r, ) , , ( r s a popp is the standard population of age group a, sex s and income bracket r. Here also, in the formulae, the absence of a parameter indicates the summation over all groups. Smoothing of the prevalence rates was conducted with a moving average across ages2. As already mentioned, data for Brazil are drawn from the 2013 PNS. Data for the US were drawn from the NHIS – National Health Interview Survey - for three consecutive years: 2012, 2013 and 2014. The use of a three years average for the US information was necessary to reduce the sample variance, since the sample rate in the American survey was much smaller than similar figure for the Brazilian one. Variables used were extracted from both Family and Sample Adult Records. In this section we have used data from these two sources. For each of the conditions considered we will present a standard set of 31 graphs referring to reported morbidity, namely: i) Crude rates by sex - Brazil & US - per capita family income bracket 1; ii) Crude rates by sex - Brazil & US - per capita family income bracket 1 – standardized by the Brazilian population; iii) Crude rates by sex - Brazil & US - per capita family income bracket 1 – standardized by the US population; iv) Crude rates by sex - Brazil & US - per capita family income bracket 2; v) Crude rates by sex - Brazil & US - per capita family income bracket 2 – standardized by the Brazilian population;

2 Because of the smoothing process used, there are some differences between the crude rates of a given

country and the corresponding standardized rate using the country population. This difference is usually of the

• rder of less than 5%.

5 vi) Crude rates by sex - Brazil & US - per capita family income bracket 2 – standardized by the US population; vii) Crude rates by sex - Brazil & US - per capita family income bracket 3; viii) Crude rates by sex - Brazil & US - per capita family income bracket 3 – standardized by the Brazilian population; ix) Crude rates by sex - Brazil & US - per capita family income bracket 3 – standardized by the US population; x) Crude rates by sex - Brazil & US - per capita family income bracket 4; xi) Crude rates by sex - Brazil & US - per capita family income bracket 4 – standardized by the Brazilian population; xii) Crude rates by sex - Brazil & US - per capita family income bracket 4 – standardized by the US population; xiii) Crude rates by sex - Brazil & US - per capita family income bracket 5; xiv) Crude rates by sex - Brazil & US - per capita family income bracket 5 – standardized by the Brazilian population; xv) Crude rates by sex - Brazil & US - per capita family income bracket 5 – standardized by the US population; xvi) Crude rates by sex and per capita family income bracket – Brazil; xvii) Crude rates by sex and per capita family income bracket – Brazil – standardized by the Brazilian population; xviii) Crude rates by sex and per capita family income bracket – Brazil – standardized by the US population; xix) Crude rates by sex and per capita family income bracket – US; xx) Crude rates by sex and per capita family income bracket – US – standardized by the Brazilian population; xxi) Crude rates by sex and per capita family income bracket – US – standardized by the US population; xxii) Age specific rates by sex - Brazil & US; xxiii) Age specific rates by sex - Brazil & US - per capita family income bracket 1; xxiv) Age specific rates by sex - Brazil & US - per capita family income bracket 2; xxv) Age specific rates by sex - Brazil & US - per capita family income bracket 3; xxvi) Age specific rates by sex - Brazil & US - per capita family income bracket 4; xxvii) Age specific rates by sex - Brazil & US - per capita family income bracket 5; xxviii) Age specific rates by per capita family income bracket - Brazil - males; xxix) Age specific rates by per capita family income bracket - Brazil - females; xxx) Age specific rates by per capita family income bracket - US - males; xxxi) Age specific rates by per capita family income bracket - US - females; As can be seen on the list, the first 15 graphs are comparisons of Brazilian and US aggregated rates for each of the five per capita family income brackets: crude rates and rates standardized by the Brazilian and the US population. The next three graphs (xvi-xviii) compare crude and standardized rates of the Brazilian population disaggregated into the four levels of per capita family income. The following three (xix-xxi) do the same for the US population. The following graph (xxii) presents age specific morbidity rate by sex and

6

• country. The subsequent five graphs (xxiii-xxvii) compare age specific rates by sex and

country for each of the five levels of income. The last four (xxviii-xxxi) compare the different levels of per capita family income for each combination of country and sex.

III - Self-reported Brazilian prevalence of selected morbidity conditions by age/sex/income - Comparison to similar statistics in the US

We present herein the self-reported prevalence rates of certain selected morbidity conditions, namely Arthritis/Rheumatism, Diabetes, Bronchitis/Asthma, Hypertension, Heart condition, Stroke and COPD, by sex and income bracket, among the Brazilian

• population. We also present comparisons between Brazilian and American prevalence rates
• f similar morbidity statistics by age, sex and income bracket. Graphs follow the headings

in the list above. III.1 – Arthritis/Rheumatism When we consider the prevalence rates of Arthritis/Rheumatism for the four combinations

• f sex and country (see Graph 1), all curves present an ascending trend as a function of age,

females present higher rates than males. Brazilians present higher rates than Americans. Graph 1

7 When we consider all combinations of sex and per capita family income brackets (see section D 1 in Annex D), without standardization, we see that for all income brackets, the US population presents a lower crude rate of occurrence of arthritis/rheumatism than the

• Brazilian. When standardized either by the Brazilian or by the US population, this pattern

changes slightly and the gaps between gaps widens for all combinations of income bracket and standard population, although narrower for the first two income brackets. For both sexes combined, the pattern is the same as the one found for each sex separately. When comparing the income brackets in Brazil (see section D 1 in Annex D) rates roughly increase with income but decrease when standardized. The same does not hold true for the US: with exception of the first income bracket, crude rates roughly decrease with income, even when rates are standardized. Table 1 - Crude and Standardized rates of Arthritis/Rheumatism by sex and income bracket – Brazil 1998 and the US 1995/1996

males total

income1 income2 income3 income4 Income5

BR Tot

3,48% 2,44% 3,43% 3,94% 3,64% 3,45%

tot_BR

3,26% 2,52% 3,69% 3,37% 3,44% 2,99%

tot_US

4,05% 2,92% 4,63% 4,10% 4,43% 3,67%

USA Tot

1,92% 1,83% 1,97% 1,88% 1,87% 2,00%

Tot_BR

1,53% 1,59% 1,78% 1,58% 1,37% 1,43%

8

Tot_US

1,81% 1,88% 2,10% 1,87% 1,65% 1,71%

Females

total income1 income2 income3 income4 Income5

BR Tot

9,03% 6,52% 6,76% 10,54% 9,62% 10,02%

Tot_BR

8,61% 9,74% 7,41% 9,16% 8,89% 8,13%

Tot_US

10,17% 11,61% 8,53% 10,73% 10,37% 9,90%

USA Tot

3,29% 3,16% 3,60% 3,46% 3,24% 3,03%

Tot_BR

2,63% 2,83% 3,07% 2,78% 2,34% 2,33%

Tot_US

3,07% 3,22% 3,55% 3,25% 2,76% 2,77%

both sexes

Total income1 income2 income3 income4 Income5

BR Tot

6,41% 4,72% 5,28% 7,56% 6,67% 6,74%

Tot_BR

6,07% 6,08% 5,73% 6,52% 6,23% 5,58%

Tot_US

7,18% 7,11% 6,68% 7,65% 7,38% 6,75%

USA Tot

2,63% 2,59% 2,86% 2,69% 2,57% 2,49%

Tot_BR

2,09% 2,28% 2,49% 2,19% 1,86% 1,84%

Tot_US

2,43% 2,61% 2,87% 2,55% 2,19% 2,16%

Source: IBGE, PNS 2013 & NHIS 2012/13/14

When one considers age-specific rates by country and sex, the pattern is quite consistent across income brackets: males present lower rates than females, and American adults lower rates than Brazilians (see Graph 2 to Graph 6). Upon comparing the five income brackets for Brazil, no clear pattern is discernible neither for males nor for females (see Graph 7 and Graph 8). For the US rates roughly decrease with income (see Graph 9 and Graph 10).

9 Graph 2 Graph 3

10 Graph 4 Graph 5

11 Graph 6 Graph 7

12 Graph 8 Graph 9 Graph 10

13 III.2 – Diabetes When we consider the prevalence rates of diabetes for the four combinations of sex and country (see Graph 11), all curves present an ascending trend as a function of age, females present higher rates than males. Brazilians present higher rates than Americans. Graph 11

14 When we consider all the different per capita family income brackets (see section D 2 in Annex D), we see that either crude or standardized rates do not follow the same pattern for males, females or both sexes combined. US crude rates are all, with no exception higher than those of Brazil are. US standardized rates, though, present lower values than the Brazilian ones only for the last income bracket for females, When comparing the different per capita family income bracket levels in Brazil (see section D 2 in Annex D) crude rates for males, females and both sexes combined, show no discernible trend as a function of income level. The same holds true for standardized

• values. A discernible pattern emerges for the US population: for males, females and both

sexes combined, similarly to what happens with arthritis data, with the exception of the first bracket, crude rates monotonically decrease with income. Table 2 - Crude and Standardized rates of Diabetes by sex and income bracket – Brazil 1998 and the US 1995/1996

males total income1 income2 income3 income4 income5 BR tot 6,44% 5,25% 4,83% 7,17% 6,41% 7,02% tot_BR 5,23% 6,06% 5,03% 5,13% 5,24% 5,20% tot_US 6,62% 7,77% 6,50% 6,46% 6,44% 6,75% USA tot 10,86% 11,61% 13,34% 11,98% 10,14% 8,59% tot_BR 8,27% 9,79% 11,80% 9,53% 7,07% 5,95% tot_US 10,28% 12,04% 14,31% 12,00% 8,85% 7,47%

15

females total income1 income2 income3 income4 Income5 BR tot 7,84% 5,51% 7,46% 8,52% 9,18% 7,47% tot_BR 6,70% 7,32% 8,55% 6,35% 7,66% 5,28% tot_US 8,12% 8,65% 10,50% 7,69% 9,29% 6,49% USA tot 10,31% 12,03% 14,08% 10,57% 9,23% 6,75% tot_BR 8,20% 10,82% 12,00% 8,37% 6,63% 5,30% tot_US 9,64% 12,48% 13,99% 9,91% 7,85% 6,36% both sexes total income1 income2 income3 income4 income5 BR tot 7,27% 5,42% 6,50% 8,00% 7,98% 7,27% tot_BR 6,05% 6,73% 7,17% 5,81% 6,60% 5,24% tot_US 7,34% 8,09% 8,78% 7,04% 7,96% 6,46% USA tot 10,57% 11,85% 13,74% 11,26% 9,67% 7,71% tot_BR 8,25% 10,34% 11,92% 8,95% 6,86% 5,69% tot_US 9,77% 12,09% 13,97% 10,71% 8,17% 6,83%

Source: IBGE, PNS 2013 & NHIS 2012/13/14

When one considers age-specific rates by country and sex, a pattern is discernible across Brazilian income levels: males present the lowest rates and the greatest difference with females at income bracket 2 and the narrowest at income bracket 5. In the US estimated rates are lower for males in the first two income brackets. For the other three brackets, there is a crossover and females present lower rates (see Graph 12 to Graph 16). Upon comparing the five different income brackets, two opposite patterns emerge for Brazil: diabetes reported-morbidity levels for young males decrease with the per capita family income and for females, this is roughly true for all age groups. For the US population though there are multiple crossovers, a clear pattern emerges with rates decreasing with income (see Graph 17 to Graph 20).

16 Graph 12 Graph 13

17 Graph 14 Graph 15

18 Graph 16 Graph 17

19 Graph 18 Graph 19

20 Graph 20 III.3 – Bronchitis/Asthma When we consider the prevalence rates of Bronchitis/asthma for the four combinations of sex and country (see Graph 21), all curves present a rather flat aspect as a function of age. Americans and females present higher rates than respectively, Brazilians and males. Graph 21

21 When we consider all the different per capita family income brackets (see section D 3 in Annex D), we see that either crude or standardized rates follow the same pattern for males, females or both sexes combined: US crude rates are all, with no exception higher than those

• f Brazil are.

When comparing the different per capita family income brackets in Brazil (see section D 3 in Annex D and Table 3) in Brazil we perceive a slight increase with income for males, females and both sexes, standardized or not. For the US population there is a seesaw pattern with the first income brackets as local maximum. Table 3 - Crude and Standardized rates of Bronchitis/Asthma by sex and income bracket – Brazil 1998 and the US 1995/1996

Males total income1 income2 income3 income4 Income5 BR tot 3,57% 2,79% 2,92% 3,76% 3,60% 4,03% tot_BR 3,75% 3,29% 3,33% 3,59% 3,56% 4,45% tot_US 3,69% 3,15% 3,29% 3,54% 3,57% 4,29% USA tot 10,61% 11,77% 11,47% 10,56% 8,98% 10,64% tot_BR 11,27% 12,02% 11,47% 11,22% 9,82% 11,78% tot_US 10,86% 11,69% 11,33% 10,83% 9,39% 11,23% females total income1 income2 income3 income4 Income5 BR tot 5,11% 3,77% 4,31% 5,35% 5,60% 5,71% tot_BR 5,07% 3,66% 4,17% 5,06% 5,58% 6,00%

22

tot_US 5,04% 3,63% 3,97% 5,12% 5,44% 5,90% USA tot 14,08% 15,73% 15,06% 13,90% 13,01% 13,08% tot_BR 14,44% 16,53% 15,15% 13,86% 13,70% 13,32% tot_US 14,13% 16,03% 15,20% 13,74% 13,33% 12,95% both sexes total income1 income2 income3 income4 Income5 BR tot 4,44% 3,39% 3,76% 4,70% 4,68% 4,92% tot_BR 4,50% 3,53% 3,90% 4,45% 4,64% 5,28% tot_US 4,44% 3,42% 3,75% 4,45% 4,57% 5,14% USA tot 12,41% 14,03% 13,43% 12,26% 11,04% 11,81% tot_BR 12,88% 14,53% 13,45% 12,58% 11,77% 12,48% tot_US 12,49% 14,07% 13,40% 12,28% 11,36% 11,98%

Source: IBGE, PNS 2013 & NHIS 2012/13/14

When one considers age-specific rates by sex and country, both countries show rather flat patterns, with Brazilian rates lower than US rates. Males present lower rates than their female counterparts do in US. Brazil presents crossovers, with the exception of income bracket 5, with males presenting lower rates than females (see Graph 22 to Graph 26). Upon comparing the five income brackets for the Brazilian population, one can see that the curves are very similar in shape and in values, with crossovers for males and females. This same pattern occurs for males and females in the US (see Graph 27 to Graph 30). Graph 22

23 Graph 23 Graph 24

24 Graph 25 Graph 26

25 Graph 27 Graph 28

26 Graph 29 Graph 30 III.4 – Hypertension When we consider the prevalence rates of Hypertension the four combinations of sex and country (see Graph 31), all curves present an ascending trend as a function of age. Brazilian

27 females present higher rates than males. US present the opposite pattern, with males with higher rates. Brazilians present higher rates than Americans, with exception of females in the 35-64 age bracket. Graph 31 For females, when we consider all brackets of per capita family income (see section D 4 in Annex D), without standardization, we can see that the US population presents a higher rate

• f occurrence of hypertension than the Brazilian, with exception of females in the fifth

income bracket. However, when standardized either by the US or by the Brazilian population, this pattern changes for females of the fourth and fifth income bracket where the US presents lower rates. For males, Brazilian crude and standardized rates are consistently lower than those of the US for all income bracket. When comparing the different per capita family income brackets in Brazil (see Table 4 and section D 4 in Annex D) crude rates increase with income for males and females up to the third income bracket and decrease for females in the fourth and fifth income bracket. Once rates are standardized, male and female rates are rather flat, but female present the lowest rate decrease in the fifth income bracket. For the US population, crude rates are rather flat for males but decrease for females from the second income bracket on. When standardized,

28 either by the US or the Brazilian population, both male and female rates decrease with income from the second bracket on. Table 4 - Crude and Standardized rates of Hypertension by sex and income bracket – Brazil 1998 and the US 1995/1996

males total income1 income2 income3 income4 income5 BR tot 20,07% 15,28% 15,70% 23,65% 19,53% 21,85% tot_BR 17,18% 16,35% 16,16% 18,09% 16,91% 17,49% tot_US 20,89% 20,47% 19,50% 21,72% 20,51% 21,33% USA tot 30,76% 29,60% 31,38% 30,65% 30,82% 31,04% tot_BR 24,59% 25,95% 27,64% 25,35% 23,35% 22,92% tot_US 29,16% 30,49% 32,88% 30,19% 27,61% 27,24% females total income1 income2 income3 Income4 Income5 BR tot 26,03% 18,92% 22,30% 30,48% 28,42% 26,18% tot_BR 23,08% 23,50% 24,38% 23,86% 24,64% 19,45% tot_US 26,98% 26,76% 28,39% 27,97% 28,71% 23,29% USA tot 29,60% 31,30% 34,48% 31,16% 28,78% 23,81% tot_BR 23,70% 27,53% 28,89% 25,11% 20,86% 18,94% tot_US 27,62% 31,39% 33,50% 29,16% 24,62% 22,69% both sexes total income1 income2 income3 income4 Income5 BR tot 23,49% 17,55% 19,75% 27,74% 24,42% 24,15% tot_BR 20,44% 20,49% 20,99% 21,43% 21,12% 18,53% tot_US 23,93% 23,72% 24,46% 25,06% 24,69% 21,98% USA tot 30,16% 30,57% 33,08% 30,91% 29,78% 27,58% tot_BR 24,18% 26,80% 28,34% 25,26% 22,12% 21,07% tot_US 27,89% 30,49% 32,74% 29,16% 25,63% 24,55%

Source: IBGE, PNS 2013 & NHIS 2012/13/14

When one considers age-specific rates by country and sex a pattern for Brazil is discernible across income levels: males present lower rates than females for most all age group. For the US estimated rates, the gender gap is smaller compared to the Brazilian population for the first four income brackets. However, US male rates are higher than female rates. Brazilian and American age specific rates are all convex curves, with the exception of males in the first income bracket (see Graph 32 to Graph 36). Upon comparing the five different per capita family income brackets across sex for Brazil, no uniform pattern emerges. For Brazilian, males and females, reported morbidity for hypertension does not present any variation as a function of income and values for all income brackets are not statistically different from one other. For American males and females, rates decrease with income (see Graph 37 to Graph 40).

29 Graph 32 Graph 33

30 Graph 34 Graph 35

31 Graph 36 Graph 37

32 Graph 38 Graph 39

33 Graph 40 III.5 - Heart Condition When we consider the prevalence rates of Heart condition the four combinations of sex and country (see Graph 41), all curves present an ascending trend as a function of age. US males present higher rates than females. US pattern is steep than that of Brazil for both

• sexes. Males crossover at around 55 years of age and females close the gap at higher age

groups. Graph 41

34 When we consider all household income brackets, male rates for Brazil are consistently below their American counterparts with exception of the highest income group. For females, Brazilian rates are below those of the US only for the first two income brackets (see Table 5 and section D 5 in Annex D). When standardized, either by the US or the Brazilian population, males keep the same pattern recognized by crude rates. For females,

• nly in the first income bracket are Brazilians with a lower rate than their US counterparts.

When comparing the different income brackets (see Table 5 and section D 5 in Annex D), there is no discernible pattern for Brazil: crude and standardized rates present zig-zag values both for males and females. For the US population, no clear pattern emerges from the graphs: cruse rates for males also present a zig-zag pattern; for females and both sexes combined the maxima is reached in the second income bracket with declining rates from then on; when standardized, males, females and both sexes combined present the same pattern as the one observed for crude rates of females and both sexes combined (see Table 5 and section D 5 in Annex D). Table 5 - Crude and Standardized rates of Heart Condition by sex and income bracket – Brazil 1998 and the US 1995/1996

males total income1 income2 income3 income4 income5 BR tot 4,31% 3,08% 3,34% 4,62% 3,90% 5,37%

35

tot_BR 3,81% 4,02% 3,38% 3,39% 3,43% 4,85% tot_US 4,84% 5,45% 4,04% 4,26% 4,52% 5,93% USA tot 5,78% 5,84% 6,19% 5,94% 6,18% 5,12% tot_BR 4,11% 4,58% 5,07% 4,27% 3,86% 3,51% tot_US 5,54% 5,96% 6,61% 5,80% 5,32% 4,88% females total income1 income2 income3 income4 income5 BR tot 4,71% 2,74% 4,03% 6,59% 4,42% 4,87% tot_BR 4,19% 3,11% 4,30% 4,98% 3,81% 3,63% tot_US 4,99% 3,46% 4,93% 6,00% 4,49% 4,46% USA Tot 3,38% 4,60% 4,87% 3,77% 2,54% 1,62% tot_BR 2,42% 3,49% 3,61% 2,68% 1,62% 1,44% tot_US 3,10% 4,39% 4,53% 3,43% 2,10% 1,92% both sexes total income1 income2 income3 income4 income5 BR Tot 4,54% 2,87% 3,76% 5,78% 4,19% 5,10% tot_BR 4,01% 3,55% 3,89% 4,28% 3,64% 4,18% tot_US 4,78% 4,15% 4,49% 5,15% 4,38% 4,95% USA Tot 4,54% 5,13% 5,47% 4,83% 4,32% 3,44% tot_BR 3,25% 3,95% 4,26% 3,46% 2,72% 2,64% tot_US 4,11% 4,88% 5,28% 4,38% 3,50% 3,42%

Source: IBGE, PNS 2013 & NHIS 2012/13/14

When one considers age-specific rates by country and sex, a pattern is discernible across all income brackets: an increase with age. Brazilian males present higher rates than females in the first and fifth income bracket and Brazilians presents lower rates than the US with exception of the highest income bracket (see Graph 42 to Graph 46). Among Brazilians, males and females rates increase with age but there is no discernible pattern across income brackets. Among Americans rates decrease roughly with income (see Graph 47 to Graph 50).

36 Graph 42 Graph 43

37 Graph 44 Graph 45

38 Graph 46 Graph 47

39 Graph 48 Graph 49

40 Graph 50

• III. 6 - Stroke

When we consider the prevalence rates of strokes for the four combinations of sex and country (see Graph 51), all curves present an ascending trend as a function of age. Both countries present a crossover around the age 50: females present higher rates than males at younger ages and the reverse is true for older ages. Graph 51

41 When we consider all possible combinations of income brackets and standardization (see Table 6 and section D 6 in Annex D) for females, we see that as a general rule, Brazilians present lower rates of occurrences of Stroke than their US counterparts. For both sexes combined the sole exception is the crude rate of the fifth income bracket which presents higher crude for Brazil then for the US. For males, only the first three income brackets present the pattern. For the fourth income group, Brazilan crude rates for males are still smaller than American’s but standardized rates by any of the standard population are

• higher. For the last income bracket, American rates are higher for crude and standardized

rates. When comparing the different per capita family income brackets in Brazil (see Table 6 and section D 6 in Annex D), crude rates present a single maximum at the third income group, for males, females and both sexes combined. When standardized, rates monotonically decrease with per capita family income, with the exception of those for females and both sexes combined at the middle income bracket. A different behavior is discernible among the US population (see Table 6 and section D 6 in Annex D). For all combinations of sex and standardizing population, rates monotonically decrease with income with the exception

• f a peak at the second income bracket.

42 Table 6 - Crude and Standardized rates of Stroke by sex and income bracket – Brazil 1998 and the US 1995/1996

males total income1 income2 income3 income4 income5 BR Tot 1,73% 1,48% 1,61% 2,33% 1,77% 1,45% tot_BR 1,52% 2,56% 1,85% 1,67% 1,56% 1,22% tot_US 2,04% 3,74% 2,39% 2,24% 2,09% 1,64% USA Tot 2,63% 3,87% 4,14% 2,85% 2,16% 1,27% tot_BR 1,93% 3,05% 3,48% 2,17% 1,37% 0,95% tot_US 2,55% 3,98% 4,46% 2,84% 1,89% 1,26% Females total income1 income2 income3 income4 income5 BR Tot 1,48% 1,03% 1,22% 2,08% 1,56% 1,31% tot_BR 1,31% 1,45% 1,43% 1,58% 1,30% 0,93% tot_US 1,61% 1,74% 1,72% 1,94% 1,62% 1,17% USA Tot 2,76% 3,40% 4,40% 2,93% 2,17% 1,33% tot_BR 2,06% 2,81% 3,37% 2,13% 1,51% 1,27% tot_US 2,55% 3,33% 4,17% 2,68% 1,86% 1,60% both sexes total income1 income2 income3 income4 income5 BR Tot 1,59% 1,20% 1,37% 2,18% 1,65% 1,37% tot_BR 1,39% 1,90% 1,58% 1,61% 1,42% 1,05% tot_US 1,72% 2,37% 1,93% 2,00% 1,77% 1,30% USA Tot 2,70% 3,60% 4,28% 2,89% 2,16% 1,30% tot_BR 2,00% 2,91% 3,40% 2,15% 1,44% 1,10% tot_US 2,46% 3,50% 4,17% 2,67% 1,78% 1,34%

Source: IBGE, PNS 2013 & NHIS 2012/13/14

When one considers age-specific rates by country and sex, a pattern is quite clear in the US data: rates for males and females are very close in value but consistently lower for males of younger age groups when compared to females and higher for older age groups, mainly fot those in lower income brackets. Upon comparing the five different income brackets one can see that reported morbidity levels with respect to Stroke decrease with income for both males and females. There are some minor exceptions among the younger age groups,

• though. The situation in Brazil is somehow different: the gender gap is wider but males also

present higher rates than females at older age groups, and lower at younger age groups (see Graph 52 to Graph 60). Among females, rates are roughly ordered by income in a descending order by income.

43 Graph 52 Graph 53

44 Graph 54 Graph 55

45 Graph 56 Graph 57 Graph 58

46 Graph 59

47 Graph 60

• III. 7 - COPD

When we consider the prevalence rates of Chronic Obstructive Pulmonary Disease (COPD) for the four combinations of sex and country (see Graph 61), all curves present an ascending trend as a function of age with a narrow gender gap and Brazilians with a later shape than Americans. Therefore, Brazilians present higher rates for younger age groups and lower rates for older age groups than Americans. Graph 61

48 When we consider all possible combinations of income brackets and standardization (see Table 7 and section D 7 in Annex D) for males, females and both sexes combined, we see that as a general rule, Brazilians present lower rates of COPD than their US counterparts. Exceptions are the higher income bracket for all combinations of rates and sex and for the standardized rates of the fourth group. When comparing the different per capita family income brackets in Brazil (see Table 7 and section D 7 in Annex D), crude rates increase with income for males, females and both sexes combined. The exception is the first income racket for females. When standardized, rates for males present a zig-zag pattern and for females, with the exception of the first bracket, rates increase with income. A different behavior is discernible among the US population (see Table 7 and section D 7 in Annex D). For all combinations of sex and rates, crude or standardized, rates monotonically increase with income, with the exception of the first bracket. Table 7 - Crude and Standardized rates of COPD by sex and income bracket – Brazil 1998 and the US 1995/1996

males total income1 income2 income3 income4 Income5 BR Tot 1,86% 0,88% 1,80% 1,90% 2,06% 2,09% Tot_BR 1,88% 1,09% 2,13% 1,53% 2,34% 1,78%

49

Tot_US 2,07% 1,33% 2,39% 1,72% 2,43% 2,10% USA Tot 2,90% 3,63% 4,72% 3,11% 2,74% 1,43% Tot_BR 2,12% 2,93% 4,01% 2,31% 1,99% 1,03% Tot_US 2,76% 3,73% 5,10% 3,08% 2,50% 1,42% Females total income1 income2 income3 income4 Income5 BR Tot 1,87% 1,54% 1,24% 1,87% 1,99% 2,34% tot_BR 1,76% 1,88% 1,24% 1,59% 1,87% 2,07% tot_US 1,93% 2,05% 1,37% 1,72% 2,06% 2,26% USA Tot 3,03% 3,87% 5,05% 3,27% 2,27% 1,23% tot_BR 2,30% 3,48% 4,20% 2,48% 1,49% 1,00% tot_US 2,82% 4,07% 5,09% 3,06% 1,89% 1,29% both sexes total income1 income2 income3 income4 Income5 BR Tot 1,87% 1,28% 1,46% 1,88% 2,03% 2,22% tot_BR 1,81% 1,52% 1,61% 1,57% 2,10% 1,97% tot_US 1,96% 1,70% 1,76% 1,70% 2,21% 2,16% USA Tot 2,97% 3,77% 4,90% 3,19% 2,50% 1,33% tot_BR 2,21% 3,22% 4,10% 2,40% 1,74% 1,03% tot_US 2,73% 3,84% 5,03% 3,01% 2,12% 1,31%

Source: IBGE, PNS 2013 & NHIS 2012/13/14

When considering age-specific rates, one can say that in most instances American males present higher rates than females for old age groups. Both sexes present a clear hierarchy among per capita family income brackets: higher per capita family income are associated with lower morbidity rates of COPD (see Graph 62 to Graph 70). Age-specific rates for Brazil are rather flat and the curves present several crossovers.

50 Graph 62 Graph 63

51 Graph 64 Graph 65

52 Graph 66 Graph 67

53 Graph 68 Graph 69

54 Graph 70

II – Final Comments

When we consider the prevalence rates of arthritis/rheumatism, diabetes, hypertension, heart condition and stroke for the four combinations of sex and country, all curves present an ascending trend as a function of age. However, COPD for Brazil, prevalence rates present a rather flat aspect as a function of age, for both sexes. Bronchitis/asthma, on the

• ther hand, present flat rates for both countries with a small decrease at the end.

The gender gap favors females for arthritis/rheumatism and bronchitis/asthma in both countries and favors males for hypertension and heart conditions in the US. The gender gap is rather narrow for diabetes, stroke and COPD in both countries, and for hypertension and heart condition in Brazil. Brazil presents higher rates than the US for both sexes for arthritis/rheumatism and stroke, and for heart conditions only for females. The opposite occurs for diabetes and bronchitis/asthma for both sexes and for hypertension for males. All the other combinations

• f sex and morbidity present crossovers between the two countries.

BIBLIOGRAPHY

55

AIHW 2005. Australian health inequalities 2: trends in male mortality by broad occupational group. AIHW bulletin no. 25. Cat. no. AUS 58. Canberra: AIHW. Viewed 4 September 2017 <http://www.aihw.gov.au/publication-detail/?id=6442467713>. BARROS, M. B. de A. et al. Tendências das desigualdades sociais e demográficas na prevalência de doenças crônicas no Brasil, PNAD: 2003-2008. Ciência & Saúde Coletiva, Rio de Janeiro: Associação Brasileira de Saúde Coletiva - Abrasco, v. 16, n. 9, p. 3755-3768, set. 2011. Disponível em: <http://www.scielo.br/pdf/csc/v16n9/a12v16n9.pdf>. CENTERS FOR DISEASE CONTROL AND PREVENTION – CDC. National Health Interview

• Survey. Available at <https://www.cdc.gov/nchs/nhis/index.htm>

COGGON, David; HARRIS, E. Clare; BROWN, Terry; RICE, Simon & PALMER, Keith T. Occupational mortality in England and Wales, 1991–2000. London: National Health Statistics, 2009, 52 p. Available at:

https://www.google.com.br/url?sa=t&rct=j&q=&esrc=s&source=web&cd=4&ved=0ahUK Ewij7dbv4KTWAhVMhpAKHRUhCicQFghJMAM&url=https%3A%2F%2Fwww.ons.go v.uk%2Fons%2Frel%2Foccupational-mortality%2Foccupational-mortality-in-england-and- wales%2Foccupational-mortality%2Foccupational-mortality-in-england-and-wales--1991- 2000.pdf&usg=AFQjCNFLDLvm2_Y_IRsbSdJfHjaHUGndrw FREITAS, M. P. S. de et al. Amostra mestra para o sistema integrado de pesquisas

• domiciliares. Rio de Janeiro: IBGE, 2007. 67 p. (Textos para discussão. Diretoria de

Pesquisas, n. 23). Available at: <http://www.ibge.gov.br/home/estatistica/indicadores/sipd/texto_discussao_23.pdf>. GEYER, Siegfried; PETER, Richard. Occupational status and all-cause mortality - A study with health insurance data from Nordrhein-Westfalen, Germany. European Journal of Public Health, 9, 2, 1999. LEE, Hye-Eun; KIM, Hyoung-Ryoul; CHUNG, Yun Kyung; KANG, Seong-Kyu; KIM, Eun-A. Mortality rates by occupation in Korea: a nationwide, 13-year follow-up study. Journal of Occupational and Environmental Medicine. 2016 May; 73(5): 329–335. MALTA, D. C. et al. Inquéritos nacionais de saúde: experiência acumulada e proposta para

• inquérito de saúde brasileiro. Revista Brasileira de Epidemiologia, Rio de Janeiro:

Associação Brasileira de Saúde Coletiva - Abrasco, v. 11, supl. 1, p. 159-167, maio 2008. Available at: <http://www.scielosp.org/pdf/rbepid/v11s1/16.pdf>. MARTIKAINEN, Pekka. Socioeconomic mortality differentials in men and women according to own and spouse's characteristics in Finland. Sociology of Health & Illness.

• Vol. 17. No. 3, 1995, ISSN 0141-9889. pp. 353-375

STEVENSON, T. H. C. The Vital Statistics of Wealth and Poverty. Journal of the Royal

Statistical Society, Vol. 91, No. 2 (1928), pp. 207-230. Available at http://www.jstor.org/stable/2341530 TRAVASSOS, C. & CASTRO, M. S. M. Determinantes e desigualdades sociais no acesso e na utilização de serviços de saúde. In: GIOVANELLA, L. et al. (Org.). Políticas e sistema

56 de saúde no Brasil. Rio de Janeiro: Ed. Fiocruz: Centro Brasileiro de Estudos de Saúde - Cebes, 2008. p. 215-243. WILKINSON, R. G.; PICKETT, K. E. The problems of relative deprivation: Why some societies do better than others. Social Science and Medicine. 2007; 65, 9, p. 1965–1978

57

ANNEX A – MAIN CHARACTERISTICS OF THE SURVEYS

A.1 Main characteristics of PNS The Brazilian National Health Survey (Pesquisa Nacional de Saúde - PNS) was conducted by IBGE for the first time in 2013. Until then, health information was collected by PNAD in specific supplements included in the questionnaire (namely, 1981, 1986, 1998, 2003 and 2008). From 1998 on, questions in the supplement included information on illness, disability, health expenditures and access to health services, as well as information on related subjects such as lost-time morbidity from either work or school. The PNS is a household survey, independent from PNAD but part of the Integrated System

• f Household surveys (Sistema Integrado de Pesquisas Domiciliares – SIPD) and,

therefore, uses the same Master sampling frame (FREITAS et al., 2007). It was designed to collect health information and provide several indicators with the desired precision to assure continuity of the great majority indicators available from PNAD health supplements. The PNS considered three main axis: the performance of the National Health System, the health conditions of the Brazilian population and the monitoring of Chronic Non- communicable Diseases and related risk factors (MALTA et al., 2008). In addition, inequality indicators received special attention, taking into account evidences that point to the importance of social factors as determinants of morbi-mortality, of health behavior and access to health services (BARROS, M. et al., 2011; TRAVASSOS and CASTRO, 2008). A.2 Main characteristics of NHIS NHIS has been collected continuously since 1957. The survey is conducted by the NCHS – National Center for Health Statistics, Center for Disease Control and Prevention (CDC). The sample design has undergone changes following each decennial census, though. It follows a stratified, multistage probability design that permits a continuous sampling of the civilian non-institutionalized population residing in the United States. Since 1995, the NHIS is designed to produce more reliable estimates for the Black population and the Hispanic population than a national household survey of the same size where all households have the same probability of sample selection.

58 The survey is designed so that the sample scheduled for each week is representative of the target population, and the weekly samples are additive over time. The weekly samples are consolidated to produce quarterly files. The annual sample is designed so that tabulations can be provided for each of the four major geographic regions. Because interviewing is done throughout the year, there is no seasonal bias for annual estimates. The main goal of NHIS is to provide a statistical picture of illness and disability in the civilian population, as well as information on related subjects such as time lost from work

• r school because of illnesses and medical care received by persons who fall ill.

The sample is divided into six different groups and each group answers a different set of chronic-condition-related questions. One specific card (out of a set of six: skin and musculoskeletal; impairments; digestive; miscellaneous; circulatory; and respiratory) is presented to the person being interviewed who informs the presence or absence of each condition listed on the card. Since there are six different sub-samples, one cannot estimate the probability of multiple occurrences if the conditions are not on the same list.

59

ANNEX B – POPULATION DISTRIBUTION – BRAZIL

Population distribution by sex and age group for each of the income brackets can be found from Graph B 1 to Graph B 4. As can be seen the population of the first income bracket is younger but also has a high proportion of cases among the elderly population, presenting an hourglass shape (see Graph B 7) when considering the relative size by age

• group. It is also the largest group of them all.

Graph B 1

60 Graph B 2 Graph B 3

61 Graph B 4 Graph B 5

62 Graph B 6 Graph B 7

63

ANNEX C – POPULATION DISTRIBUTION – USA

US Population distribution by sex and age group for each of the income brackets considered can be found in Graph C 1 to Graph C 4. The breakdown points for the US are just São Paulo quartiles converted into US$. The other two graphs refer to the joint age/sex distribution and the distribution by age-group. As can be seen the population in the first income bracket is composed of the very young, but with an overrepresentation of females. It is the smallest of all the groups, followed in size by income bracket 2. Income bracket 4 is by far the largest group. Graph C 1 Graph C 2

64 Graph C 3

65 Graph C 4 Graph C 5

66 Graph C 6 Graph C 7