Page 31 - IJPS-4-2

Sreerupa, et al.

disproportionately higher prevalence of disease and disability and lower access to and availability of health care, relative
to their urban counterparts (Alam, 2000). It is important to keep in mind that nearly three-fourths of older people in India
live in rural areas. Therefore, it is of interest to assess how changes over time in life expectancy with and without mobility
limitation differ among older Indian men and women and among older persons residing in rural and urban areas of India.
Thus, utilizing data from two cross-sectional national surveys conducted in 1995–1996 and 2004, we test for the
presence of expansion or compression of morbidity among older men and women and among older rural and urban
persons in India. We do so by estimating changes over a decade in absolute and relative life expectancy with and without
mobility limitation by gender and by place of residence.
2. Data and Methods

2.1. Data sources
Information on age-specific death rates (ASDR) from the Sample Registration System (Office of the Registrar General)
was used to construct abridged life tables. The average values of ASDR from 1995 to 1998 and 2002 to 2005 were used as
the ASDR of 1995–1996 and 2004, respectively, to avoid inconsistencies. Information on the age-specific prevalence of
mobility limitation, by gender, and by place of residence, was obtained from publically available de-identified data of two
survey rounds: The NSS on Morbidity and Treatment of Ailments, 52 Round (National Sample Survey Organization,
nd
1998), conducted from July 1995 to June 1996, and the NSS on Morbidity, Health Care, and the Conditions of the
Aged, 60 Round (National Sample Survey Organisation, 2006), conducted from January to June 2004. The prevalence
th
estimates were weighted using appropriate round-specific survey weights. The NSS are nationally representative cross-
sectional household surveys conducted by the National Sample Survey Organization (NSSO), a branch of the Department
of Statistics of the Government of India.
The sample for either round was selected using a two-stage stratified design, with census villages and urban blocks as
the first-stage units for the rural and urban areas, respectively, and households as the second-stage units. The surveys had
a sample of 34,084 older persons from 120,942 households in the 52 NSS Round and 34,831 older persons from 73,868
nd
households in the 60 NSS Round. Data were collected using structured questionnaires in face-to-face interviews at the
th
homes of the respondents.
2.2. Mobility limitation

The physical mobility statuses of the respondents of both NSS rounds were recorded, with the options being (1) physically
immobile and confined to bed (persons unable to move around the house and, in particular, to use the washroom on their
own), (2) physically immobile and confined to their homes (persons able to move within the house but unable to move
outside the house), and (3) physically mobile (National Sample Survey Organization, 1998, National Sample Survey
Organization, 2006). The proportion of individuals categorized in option (1) was <2% in either round. Thus, for our
analysis, those categorized in options (1) and (2) were combined so as to dichotomize physical mobility status into “with
mobility limitation” and “without mobility limitation.”
Our decision to focus on mobility limitation as the measure of health and not on other variables such as disability or
NCDs, was in part due to comparability issues of such variables between the two NSS rounds. While questions and/or
response options relating to disability or NCDs varied, the question on physical mobility status and its response categories
remained unchanged between the rounds.

2.3. Statistical analysis
Using the method devised by Sullivan (1971), data on the prevalence of mobility limitation were combined with abridged
life table data to compute life expectancy with and without mobility limitation for older persons in India, by gender and
by place of residence. With this method, it is possible to see the current mobility health structure of a population adjusted
for mortality by partitioning life expectancy at a given age into two mobility states: Expected years lived without mobility
limitation (~healthy) and with mobility limitation (~unhealthy). Weighted age-specific prevalence rates were used to
divide the life table stable population for each age group into specific mobility states. The total stable population above
age x by mobility states was divided by survivors at age x to compute life expectancy with and without mobility limitation.
Differences, by gender and by place of residence, in absolute and relative estimates of life expectancy with and without
mobility limitation were tested using the approach described by Jagger et al. (2014).

International Journal of Population Studies | 2018, Volume 4, Issue 2 25