Page 9 - IJPS-4-1

Alves L C and Pereira C C et al.

have estimated the length of life with depressive symptoms by race over time in the population. The goal of this study is
to investigate racial disparities in the trend of depression-free life expectancy over a 15-year period (1998–2013).
2. Data and Methods
The study used the cross-sectional household data from 1998, 2008 and 2013. Data came from two surveys, Pesquisa
Nacional por Amostra de Domicílios (PNAD – Brazilian National Household Survey) and Pesquisa Nacional de
Saúde (PNS – National Health Survey), and also from Tábuas de Mortalidade (Life Tables) from IBGE (Brazilian
Institute of Demographic Geography and Statistics) (IBGE, 2014). PNAD is a population-based survey with national
representativeness, held annually, to obtain information about the household, the individuals residing in the household,
migration, education, labor and fertility characteristics. In 1998, PNAD included a health supplement in its questionnaire
that is collected every five years, and the 2008 wave was the most recent available information. In 2013, the National
Health Survey (PNS), also a nationally representative household-based survey was carried and collected health
information. It also replaced PNAD’s health supplement (IBGE, 2014).
We used prevalence data from the 1998 and 2008 PNAD and from the 2013 PNS. The prevalence of depression was
estimated based on self-reported depression (as diagnosed by a health professional). The demographic variables were:
age, sex, and race (white, blacks and browns collapsed as a single category, and others). The decision to collapse blacks
and browns comes from the fact the differentiation between the two categories is difficult given the level of miscegenation
encountered in Brazil. Another factor we considered in our analysis is the fact that the word used for Blacks is “pretos”
and it is sometimes used as a racial slur (Chiavegatto Filho, Beltrán-Sánchez, and Kawachi, 2014). This connotation may
lead people to self-declare themselves as “pardos” instead of “pretos”.
We estimated the depression-free life expectancy (DFLE) for the Brazilian population in 1998, 2008, and 2013 by
constructing life tables which combine mortality information and prevalence of chronic diseases, in this case, depression,
as proposed by the Sullivan method (1971). The Sullivan method is the most widely one to estimate healthy life
expectancy (Imai and Soneji, 2007) or disease-free life expectancy (Alves and Arruda, 2017).
The expected years with and without depression were calculated by applying the age- and sex- specific cross-sectional
prevalence rates of the disease, respectively, to the person-years lived in different age categories derived from period life
tables (Jagger, Hauet, and Brouard, 2001), as follows:

and

where DFLE is the average number of years that an individual will live without depression (healthy), starting
x
from exact age x; whereas life expectancy with depression (LED ) is the average number of years that an individual
x
will live with the disease given the current age-specific prevalence and mortality scheme, starting from exact age
x. π is the proportion with depression in age group x to x+n, which is the prevalence obtained based on the
n xi
PNAD and PNS. L is the number of person-years lived in age interval and l is the numbers surviving to age x.
n x
x
Both are obtained from the life table generated based on estimates provided by Mortality Reporting System.
is a proportion without depression in age group x to x+n. is the number of person-years lived in age
interval without depression. is the number of person-years lived with depression in age interval x to x+n.
is the total years lived without depression from age x onward. It was obtained by the sum of the all
from age x until the last age group (e.g., 80 years or more in the present study). is years
lived with depression from age x onward (Andrade, Corona, and Lebrão, 2014; Jagger, Hauet, and Brouard, 2001).
The total life expectancy (TLE) at each age x, e , is calculated by dividing the total number of years lived from
x
that age by the numbers surviving at age x. Life expectancy with and without depression was estimated by race and
sex. All statistical analyses were performed with the aid of the STATA/MP for Windows version 15.0 (STATA Corp.,
Inc., College Station, TX) and Microsoft Excel 2016.
3. Results
Table 1 shows the prevalence of depression by race among the Brazilian population in 1998, 2008, and 2013. These data
reveal that the prevalence of depression increased for total population in the period, regardless of sex and race. On the

International Journal of Population Studies | 2018, Volume 4, Issue 1 3