Table 1.

Procedures Undertaken in the Development and Validation of the Quality of Life (QoL) Instrument for the Young Elderly in Sri Lanka (QLI-YES).

Stage	Aims	Procedures	Data Collection	Data Analysis
Part 1: Development of the QLI-YES
One	Domain and item selection	QoL instrument review	Review of 24 QoL instruments: Identified from literature from 1990 to 2009. Instruments were obtained from primary sources and secondary sources: patient reported outcomes and quality of life instruments database (PROQOLID)15 and collection of QoL instruments in the commonly used text “Measuring Health”16	Identify the items considered as important for one’s quality of life and the broader domain the item represented
		Focus group discussions (FGDs)	Two FGDs with (60–74 years old) participants (22), one with care providers (9) and one with health professionals (8). All participants were selected purposively The group also reviewed QoL-BREF, SF-36, AQoL, and EQ5D	Include additional items to the list Decide on the mode of administration, length of the questionnaire, wording and the response scale for the instrument
		Key informant interviews	Nine interviews: hospital administrator, a Buddhist monk, 2 medical experts, 2 elderly care workers, social worker, and 2 elderly persons	Include additional items to the list
Two	Content validation	Panel of experts	Comprising a sociologist, a community worker, a psychiatrist, an elderly retired teacher, and a physician assessed the content validity	The panel of experts assessed the compiled list of items and their domains for their content validity
Three	Item reduction	Field study 1 Impact method	Field study 1: 147 elderly participants (60- to 74-year age group) were selected. Multistage stratified sampling was used with Grama-Niladari (GN)—the lowest level of administration as the primary sampling unit (PSU). A random sample of GN areas was selected. Fifteen clusters of 10 individuals each, with each cluster having the ratio 4:4:2 for the number of participants in the age groups, 60–65, 65–70, and >70 years, respectively, were selected using the EPI method where the starting point was randomly selected from a list of junctions. Those bedridden and who were unable to understand the questions based on impairment of cognition were excluded. Two trained medical officers administered the questionnaires. The participants were requested to rate the items based on its importance on a 5-point scale (1 = not important to 5 = extremely important)	The frequency of endorsement, the importance and the impact (the product of “frequency” multiplied by its mean importance) were calculated.
		Factor analysis		Principal components analysis—PCA (with varimax rotation) was conducted for items in each domain. Items with eigen value of greater than 1 were identified.
Four	Construction of QLI-YES	Item presentation	The questions were formulated to enquire about the level of subjective satisfaction or contentment the respondent had for the items or events selected as important	Perceived satisfaction was assessed based on a 1–5 response scale
Part 2: Validation of the QLI-YES
Five	Validation	A field validation study was conducted	Field study: An area having both urban and rural settings and all ethnic and religious groups in one Divisional Secretariet (DS) area (Kotte an area with a population of 107 508) was selected. A multistage stratified systematic sampling technique was used to select sample of 200 elderly participants (60- to 74 year age group) based on 20 clusters of 10 individuals each, with each cluster having the ratio 4: 4: 2 for the number of participants in the age groups, 60–65, 65–70, and >70 years, respectively. The QLI-YES was administered to all people who had been resident for more than 6 months and who were older than 60 years but had not completed 74 years. Households where individuals were bedridden or were physically or mentally incapable of understanding verbal instructions were skipped in selecting the sample
		1. Construct validity—CFA	The construct validity of each domain and the overall QLI-YES was assessed by performing confirmatory factor analysis (CFA)	A 2-step process was followed. First, each domain (single factor) was tested using CFA for a congeneric model to obtain fitting models. In the next step a multifactor model (whole instrument) was used to test the hypothesized 6-factor structure of the QLI-YES using maximum likelihood estimation Absolute fit measures, relative fit measures and comparative fit index (CFI) and fit measures based on the noncentral chi-square distribution was used in assessing model fit.17
		2. Predictive (known group) validity	QoL among who had experienced a significant life event during the past year and those having chronic medical conditions	One-way analysis of variance (ANOVA) test used to assess the statistical significance
		3. Criterion validity	Relevant domain scores of the QLI-YES was compared with 3 independent measures: WHOQoL-BREF;ADL/IADL, and AMTS	Pearson’s correlation
		4. Reliability assessment	Internal consistency was assessed on separate domains as well as the entire instrument Test-retest reliability	Cronbach’s α statistic. In addition, intraclass correlation coefficients (ICC) were assessed.18 ICC scores <0.4 are indicative of poor agreement; scores from 0.41 to 0.60 signify moderate agreement; scores from 0.61 to 0.80 signify strong agreement; and scores in excess of 0.81 indicate excellent agreement19 Kappa coefficient