Page 25 - IJPS-4-2

Chirinda W, et al.

same period as this present study, also concluded that the country was progressing well toward the improvement of the
population’s health status.
Second, although we did not specifically analyze HIV and ART data, our results likely lend support to recent findings
that people living with HIV are living longer and healthier lives than their past cohorts partly due to the successful roll-
out of ART (Bor, Herbst, Newell et al., 2013; Shisana, Rehle, Simbayi et al., 2014). Nevertheless, more research on the
separation of analysis of HIV and non-HIV populations is needed to validate our speculation.
Third, we only stratified the analyses by gender. This can be expanded to include other differentials such as education
and race that have been also frequently studied in the existing literature (Feng, Zhu, Zheng et al., 2016; Gu, Dupre, Warner
et al., 2009). The only drawback would be the unavailability of mortality data disaggregated by these variables, which
would be required to apply the method used in this study. Related to this, due to the relatively small sample size of at
oldest-old ages in the SABSSM surveys, we had an open interval at 80 years and above to have reliable estimates.
Fourth, the exclusion of institutionalized individuals from the SABSSM surveys can bias the prevalence of self-rated
health and HLE estimates upward. It is important to include this segment of the elderly population, since they are likely
to be different in terms of health from those residing in the community. Unfortunately, there are no comparable surveys
that collect information on health among institutionalized elderly populations in South Africa. Fortunately, the effect of
excluding the institutionalized population in the current study is negligible because the number of institutionalized older
adults in South Africa is relatively small (Phillips and Noumbissi, 2004).
Fifth, as the sample size for single-age is not sufficiently large to calculate the prevalence of self-rated health, we used
prevalence in 10-year age group to assume that the prevalence of self-rated health within the age group is the same for
each of single year of age. Such a practice may introduce some bias than that based on the single-age rate if it would be
available. However, the bias should not be substantial since such a practice is similar to the construction of abridged life
tables.
Six, no covariate was included in the analysis. Given that self-rated health is closely linked to some major demographics
and psychosocial factors (Feng, Zhu, Zheng et al., 2016), it is possible that the improvement in HLE in South Africa over
the period of 2005–2012 may be attributable to differences in covariates even if disease condition, physical and cognitive
functions, and reporting pattern had no change over time. Studies incorporating major covariates into modeling would
shed light on investigating the possible root causes or underlying associates linked to the change of HLE.
Seven, we relied on the life tables produced by the United Nations as a basis to estimate the healthy expectancy.
Although the data quality in South Africa is still a concern, as in most developing countries, we are confident that the
United Nations life tables in the World Population Prospects are relatively robust given that these estimates are obtained
from multiple sources under systematic examinations (United Nations, 2017).
5. Conclusions

Based on self-rated overall health in the South African National HIV Incidence, Prevalence, Behaviour and Communication
Survey in 2005, 2008, and 2012, we concluded that the health condition of older adults in South Africa was improved
over the period of 2005–2012. Gender differences were evident with women having a longer life expectancy, HLE, and
unhealthy life expectancy, but lower proportions of life spent in healthy years. Our findings highlight the need for gender-
sensitive health interventions among older adults. We also observed an increase in the proportion of HLE in 2012 for
both men and women compared to those in 2005. This suggests some evidence of compression of morbidity in terms of
self-rated health for South African older adults in the period of 2005–2012.

Authors’ Contributions
W. Chirinda and Y. Saito designed the study, and W. Chirinda performed the analysis and drafted. W. Chirinda, Y. Saito,
D. Gu and N. Zungu revised the manuscript and interpreted the results.

Ethics
Not applicable as the dataset used in this study is a publicly available source.

Conflicts of Interest
No conflict of interest was reported by the authors.

International Journal of Population Studies | 2018, Volume 4, Issue 2 19