SELF-RATED HEALTH

SELF-RATED HEALTH
Draft: January 25, 2001, updated: December 7, 2009
Contact: Dorly Deeg, Henrike Galenkamp

Background
A summary measure for health status is the perception of one's own health: self-perceived or self-rated health. Self-rated health is likely to be more closely linked to quality of life as compared to specific conditions. In many studies, it has proven to be an excellent predictor of mortality (Idler & Benyamini, 2007). In most surveys, a single question is used to determine self-rated health. Also, the first question of the widely used Short Form-36 (SF-36) and of the abbreviated SF-12, the quality of life questionnaire from the Medical Outcomes Study (Ware et al, 1993), is a question about self-rated health.
The single question is used in various wordings and response formats. Most widely used, and adopted by the Netherlands Health Interview Survey of Statistics Netherlands (NCBS, 1989, 1995), is the question on the perception of health in general. An individual's response to this questions, however, depends on this individual's norms and standards about what constitutes (good) health. Because this standard is not explicit in the general health question, it may be useful to include a standard in the wording of the question. There are various options to do this: comparison with age peers or comparison with one's own health as it was 5 or 10 years ago.
Limitations of a single question are the likelihood of chance fluctuations in responses, and the limited response range. In longitudinal research, a single question would not be very sensitive to (relevant) change. To determine perceptions of health that circumvent these limitations, a scale has been developed by the RAND corporation in the United States: the General Health Perception Questionnaire (GHPQ) (Brook et al., 1979; König-Zahn et al., 1993). This scale has been translated into Dutch (Vragenlijst Algemene Gezondheidsbeleving, VAGB) and validated for the Dutch situation by LASA researchers (Kriegsman et al., 1995).

Measurement instruments in LASA

		LSN -A 1992	LASA -B 1992- 1993	LASA -C 1995-1996	LASA -D 1998-1999	LASA -E 2001-2002	LAS 2B* 2002-2003	LASA -F**2005-2006	LASA -G 2008-2009
1	Single question "general"	x	x	x	x	x	x	x	x
	Single question "age peers"	x	x	x	x	x	x	x	x
2	Single question "10 years ago"	x	-	-	-	-	x	-	-
3	Single question ‘3 (or 4) years ago’	-	-	-	-	-	-	x	x
4	GHPQ	-	x	x	x	x	x	x	x
5	SF-12	-	-	-	x	x		x	x

* LAS2B is the first measurement from the second cohort.
** From this measurement on (LASA-F), the first and second cohorts are merged.

1. Single questions ‘general’ and ‘age peers’ are available in the main interview (LSNa030, LASA*036), and in telephone interviews in LASA-C, D, E, F and G (LASA*702 for respondents, LASA*602 for proxy).
2. Single question ‘10 years ago’ is available in the main interview only at baseline measurements, LSNa030 and LASA2b036.
3. Single question ‘3 or 4 years ago’ was added in 2005, available in the main interview (LASAf036, LASAg036).
4. GHPQ is available in self-administered questionnaire (LASA*113, LASA*313).
5. SF-12 is available in self-administered questionnaire (LASA*133, LASA*333, LASA*533).

Self-rated health in the LASA main interview is evaluated using two questions, taken from the Netherlands Health Interview Survey: the perception of one's health in general and the perception of one's health in comparison with age peers (NCBS). There are five response categories: from (1) 'excellent' to (5) 'poor', and from (1) 'much better' to (5) 'much worse', respectively. These responses can be dichotomized, which is often done between (2) and (3), i.e. between 'excellent or good' and 'less than good', and between '(much) better' and 'equally good or worse', respectively (Deeg, 1998). The two items are moderately to strongly correlated (r = 0.43 in LASA-B) and thus might be used as one sum score. However, this is not done in the literature. The test-retest reliability, as calculated from the LSN-cycle and the first LASA-cycle (time interval: 10 months), are 0.75 and 0.67, respectively.

In 2005 at the F measurement, the question ‘how is your health compared to 3 or 4 years ago’ was added. This was done to measure a possible response shift (Sprangers, 1999) in self-rated health since the previous measurement. Incongruence between response to this question and the true response at the previous measurement is indicative of a response shift.

A multi-item scale is included in the LASA-self-administered questionnaire, i.c. an abbreviated version of the GHPQ (Brook et al., 1979; König-Zahn et al., 1993). This version consists of eight questions, four about current health perception, and four about the expectation of future health. The reliability of this version of the scale as a whole, based on the first LASA cycle, and evaluated using Cronbach's alpha, is 0.78. The two subscales have a reliability of 0.73 (current) and 0.63 (future; item 3 omitted to obtain a better scale), respectively.

Longitudinal analyses of self-rated health can be based on the score of 5 response categories or on the dichotomous score. The variable based on the 5 points score is generally considered to be normally distributed, and thus can be used in analysis of variance or regression models. The dichotomous variable can be used in transition analysis methods.

References
Brook RH, Ware JE, Davies-Avery A, et al. Overview of adult health status measures fielded in RAND's Health Insurance Study. Medical Care 1979; 17(suppl): 1-131.

Deeg DJH. Ervaren gezondheid verschilt naar tijd en plaats [Self-rated health differs across time and place]. In: Broese van Groenou MI, Deeg DJH, Knipscheer CPM, Ligthart GJ (eds). VU-Visies op Veroudering [VU-visions on aging]. Amsterdam: Thela Thesis, 1998, pp. 131-136.

Idler EL, Benyamini Y. Self-rated health and mortality: A review of twenty-seven community studies. J Health Soc Behavior 1997; 38: 21-37.

König-Zahn C, Fürer JW, Tax B. Het meten van de gezondheidstoestand: beschrijving en evaluatie van vragenlijsten. 1. Algemene gezondheid [The measurement of health status: description and evaluation of questionnaires. 1. General health]. Assen: Van Gorcum, 1993.

Kriegsman DMW, Eijk JTM van, Deeg DJH. Psychometrische eigenschappen van de Nederlandse versie van de RAND General Health Perception Questionnaire: de Vragenlijst Algemene Gezondheidsbeleving (VAGB). [Psychometric properties of the RAND General Health Perception Questionnaire in the Netherlands.] Tijdschrift Sociale Gezondheidszorg 1995; 73/6: 390-398.

Netherlands Central Bureau of Statistics (NCBS). Gezondheidsenquête [Health Interview Survey]. Heerlen: Central Bureau of Statistics, 1989.

Netherlands Central Bureau of Statistics (NCBS). Vademecum gezondheidsstatistiek Nederland 1995 [Vademecum health statistics The Netherlans 1995]. Voorburg/Heerlen/Rijswijk: NCBS/Ministry of Health, Welfare and Sports, 1995.

Sprangers MAG. Integrating response shift into health-related quality of life research: a theoretical model. Social Science & Medicine 48, 1999: 1507-1515.

Ware JE, Snow KK, Kosinski M, Gandek B. (1993). SF-36 Health Survey: Manual and interpretation guide. Boston: The Health Institute, New England Medical Center.