The global activity limitation indicator and self-rated health: two complementary predictors of mortality

Data

Data are drawn from the 2001 Belgian Health Interview Survey (HIS). The 2001 HIS is
a national survey in which 12,111 individuals living in Belgium were interviewed 23]. The survey was carried out by Statistics Belgium and exempt from ethics approval
by law. Participants were selected from the National Register using a multistage sampling
procedure. Household response rate was 61.3%. The GALI and SRH questions were part
of the self-administered questionnaire restricted to individuals aged 15 years and
older. Vital and migration status of the participants were followed up: records of
the survey were linked to the National Register using a unique identifier present
in both data sources. After the approval of the Belgian Commission for the protection
of privacy, we obtained a mortality and emigration follow-up of HIS participants until
31 December 2010 (approximately 10 years). The linkage was successful for 97% of the
records. The final HIS sample included 8,583 individuals aged 15 years and older,
of whom 902 died and 132 emigrated. Emigrants were considered as lost to follow-up
at the date of emigration.

Measures

Age was obtained from the National Register and was categorised into four groups:
persons aged 15–49 (reference category), persons aged 50–64, persons aged 65–74 and
persons aged 75 or more. Given their low probability of death, participants aged 15–49
were grouped in a single category. Sensitivity analysis ensured that the age group
cut-offs did not affect the results. Other variables, apart from mortality, were drawn
from the baseline questionnaire in 2001. Education was measured using the highest
education level achieved within the household. We recoded this measure into four categories:
primary education level or lower, lower secondary, higher secondary, and higher education
24].

Activity limitation was assessed with the GALI question: ‘For the past 6 months or
more, have you been limited in activities people usually do because of a health problem?
Yes, strongly limited/Yes, limited/No, not limited’ 2]. Within the ICF framework 25], the GALI corresponds to the societal perspective of functioning, i.e. the performance
of roles and social involvement in activities. The reference to ‘activities people
usually do’ fosters normative comparisons and allows to account for differences in
performance of roles by age, gender, environmental and living conditions. SRH was
measured on a five-point scale, based on the question: ‘How is your health in general?’
26]. To ease comparison with the GALI, we recoded SRH into three categories: (very) bad,
fair, and (very) good.

Statistical methods

We used Poisson regression models to estimate the effect of the GALI and SRH on the
mortality rate during the follow-up period 27]. The first model (Model 1) estimates Mortality Rates Ratios (MRRs) for sociodemographic
variables alone. Model 2 and Model 3 estimate MRRs for the GALI and SRH separately,
while adjusting for sociodemographic factors. Variants of these models additionally
estimate interaction effects between gender, age and education and each measure of
ill health (results not reproduced). The full model (Model 4) includes all predictors
(without interactions), allowing to compare the relative predictive ability of the
GALI and SRH. For significant moderators, stratum-specific estimates of the GALI and
SRH are presented for each value of the sociodemographic moderator. The impact of
the follow-up period was assessed by comparing MRRs at different times of the follow-up,
adjusting for all covariates. We did not use other measures of health and functioning
to adjust the relationship between the GALI or SRH and mortality because we aimed
to assess the importance of the GALI compared to SRH as predictors of mortality. Adjusting
for other health and functioning measures would affect the GALI and SRH unequally
and prevent rigorous comparison between the two global instruments 19]. All analyses were conducted using Stata 12 (StataCorp, College Station, USA).