Page 72 - IJPS-11-2

International Journal of
Population Studies Health-care access for the elderly living alone

for example, the probability of hospital admission or the 2. Methods
length of hospital stay – more than those not living alone
(Dreyer et al., 2018; González-González et al., 2011; 2.1. Data and sample
Guzman et al., 2004; Mu et al., 2015; Turner et al., 2016). This study employed a cross-sectional design, employing
These inconsistent findings between Thailand and other data from a nationwide 2019 HWS dataset. The HWS
countries are probably due to the different measures of dataset, a biannual survey dataset released by the
health-care access. That is, the use of health-care services National Statistical Office of Thailand, includes a
analyzed in previous studies in other countries presumed nationally representative sample with comprehensive
equal access to health care (i.e., these studies assumed that sociodemographic and essential health information about
all participants could easily access health care). the entire Thai population. The National Statistical Office
Previous studies have provided a comprehensive carried out the HWS in 2021 during the COVID-19
understanding of access to health care for older adults pandemic but did not release it. Thus, the 2019 HWS
living alone. However, since they made limited efforts to dataset used in this study was the most recent dataset at
control for confounding effects, it remains unclear whether the time this study was conducted.
the high rate of UHN among older people living alone is In this study, individuals aged 65 years or older were
due to their living arrangement or socioeconomic status, defined as older people. Among all older people in the
taking previous findings from Thailand as an example. 2019 HWS dataset, those who reported needing health care
Specifically, the positive association between living alone during the past month were selected as the study sample
and UHN may have been confounded by income. We for the analysis. This study included two separate samples
assumed two conditions: (i) low income is associated with for each research objective: a matched sample obtained
a high prevalence of UHN while living alone is not and using the PSM method for the first objective and an entire
(ii) older people living alone have lower incomes than sample of data for the second objective.
those not living alone. Under these conditions, older
people living alone are likely to have more UHN than those The matched sample was obtained using the PSM
not living alone. However, the positive association between method using 1:1 matching with a caliper width of 0.05.
living alone and UHN could be due to income rather than Propensity scores were estimated from a binary logistic
living arrangements. Thus, a study that addresses the issue regression (BLR) analysis using living arrangement
of confounding effects is needed to evaluate the precise (living alone versus not living alone) as the dependent
relationship between living alone and UHN. variable and all other socio-demographic variables as
the independent variables. A caliper width of 0.05 was
In addition, as mentioned previously, multiple studies
in Thailand have indicated socioeconomic inequalities determined after evaluating the equality between the
in health-care access among older adults. However, most groups and the number of matched pairs (Austin, 2011;
of these studies were conducted using a relatively limited Baek et al., 2015). We repeated the PSM method by
number of samples from particular regions (Kullanit & decreasing the caliper width by 0.025 from 0.2, which is the
Taneepanichskul, 2017; Osornprasop & Sondergaard, optimal caliper width suggested by Austin (2011). When
2016; Srisatidnarakul & Bunthumporn, 2020). Thus, a the caliper width was set higher than 0.05, some variables,
study using nationally representative data is needed to such as income and education, were not statistically equal
verify whether these findings are generalizable to the entire between older people living alone and those not living
nation of Thailand. alone. Meanwhile, when the caliper width was set below
0.05, the number of matched pairs decreased significantly,
Therefore, the present study used Thailand’s 2019 indicating that only particular groups of older people (e.g.,
Health and Welfare Survey (HWS) dataset to investigate high-socioeconomic older people) were matched between
UHN among older people living alone in Thailand. the two groups. A caliper width of 0.05 ensured both the
Specifically, this study has two objectives. The first number of matched pairs and the level of equality between
objective was to analyze the effect of living alone on UHN the two groups.
after adjusting for potential confounding effects. To adjust
for confounding effects, we employed a propensity score The original 2019 HWS dataset included 4,611 older
matching (PSM) method and equalized sociodemographic people who needed health care during the past month, 746
characteristics between older people living alone and of whom were living alone and 3,865 of whom were not.
those not living alone (Austin, 2011; Baek et al., 2015). After the PSM was performed, 733 matched pairs were
The second objective was to analyze the sociodemographic obtained, accounting for approximately 98% of the original
determinants of UHN using the entire sample of data. dataset of older people living alone.

Volume 11 Issue 2 (2025) 66 https://doi.org/10.36922/ijps.1218