Approaches to Operationalizing Recovery in Research
Early longitudinal studies of outcome, particularly of schizophrenia, employed definitions of recovery that were broader than the medical definition. As we saw in Chapter 1, most of these studies used both medical outcome measures and assessments of functioning. For example, the Vermont Longitudinal Study (Harding et al., 1987a, 1987b), as well as an interview protocol determining current status on a number of functional and social dimensions, included two measures of functioning, the Global Assessment Scale (Endicott, Spitzer and Fliess, 1976), and the Strauss– Carpenter Level of Functioning Scales, which rate quantity of employment and level of intimate relationships (married/unmarried/dating) on a scale of 0 to 4. Similarly, in a longitudinal follow-up study of schizophrenia, Harrow et al. (2005) added at least half-time work and the absence of poor social functioning to measures of symptoms and hospitalizations. Criteria weremeasured using the Strauss–Carpenter Level of Functioning Scales and the Levenstein–Klein–Pollack Scale (Levenstein, Klein and Pollack, 1966), on which symptoms, hospitalizations, vocational and social functioning, life adjustment and self-support are rated on an eight-point scale. In assessing one-year outcomes of treatment for early psychosis,Whitehorn et al. (2002), in addition to positive and negative symptoms, used two measures of functioning: the Global Assessment of Functioning (GAF; American Psychiatric Association, 1994) and the Social and Occupational Functioning Scale (SOFAS; Goldman, Skodol and Lave, 1992). The SOFAS was derived from the GAF, and uses similar descriptions of social and occupational functioning, but omits references to symptoms. Both of these measures use a scale of 0–100, described in 10 subdivisions, and are based on clinical judgement.
Recognizing the lack of a universal definition of recovery for research, Liberman and colleagues (2002) attempted to introduce consistency by proposing dimensions that reflect the diagnostic criteria for schizophrenia, i.e., symptom remission, vocational functioning, independent living and social relationships. The criteria specified thresholds that were to be attained and maintained for two years. For example, in the case of peer relationships, the criteria specified that the person have a social interaction with a peer outside the family at least once per week. As Liberman and colleagues noted, the thresholds and durations chosen were largely arbitrary, and determined by consensus of professionals. There is a judgemental element to these definitions; assessment of recovery ought not involve moral judgements on a person’s lifestyle. After all, if ‘normality’ is the goal of mental health services, who decides what is ‘normal’ (Ralph and Corrigan, 2005)? We noted earlier that Harding (1994) questioned the definition used in the Vermont study (Harding et al., 1987a1987b), as they found groups of people who, while not meeting all the study criteria for recovery, were nonetheless happy with their way of life. Focus groups made up of researchers, mental health professionals and consumers showed that, although there was broad consensus on the dimensions chosen by Liberman et al.
(2002), the operational definitions gained approval from a greater percentage of researchers than of consumers or practitioners. Consumers preferred to define recovery as an ongoing process, rather than an endpoint or goal.
While acknowledging that there are common elements to recovery, Anthony et al. (2003) pointed out that quantitative approaches do not allow for the diversity of recovery goals among subjects. These authors therefore urge that quantitative approaches be supplemented by the use of subjective, qualitative measures such as narrative analysis and interviews in determining best practices. However, qualitative procedures such as interviews and narratives, and multidimensional batteries such as those described above, are burdensome on consumers, staff and researchers and are often impractical for large-scale studies or routine clinical use – an area in which outcome measurement is becoming more in demand.
Early longitudinal studies of outcome, particularly of schizophrenia, employed definitions of recovery that were broader than the medical definition. As we saw in Chapter 1, most of these studies used both medical outcome measures and assessments of functioning. For example, the Vermont Longitudinal Study (Harding et al., 1987a, 1987b), as well as an interview protocol determining current status on a number of functional and social dimensions, included two measures of functioning, the Global Assessment Scale (Endicott, Spitzer and Fliess, 1976), and the Strauss– Carpenter Level of Functioning Scales, which rate quantity of employment and level of intimate relationships (married/unmarried/dating) on a scale of 0 to 4. Similarly, in a longitudinal follow-up study of schizophrenia, Harrow et al. (2005) added at least half-time work and the absence of poor social functioning to measures of symptoms and hospitalizations. Criteria weremeasured using the Strauss–Carpenter Level of Functioning Scales and the Levenstein–Klein–Pollack Scale (Levenstein, Klein and Pollack, 1966), on which symptoms, hospitalizations, vocational and social functioning, life adjustment and self-support are rated on an eight-point scale. In assessing one-year outcomes of treatment for early psychosis,Whitehorn et al. (2002), in addition to positive and negative symptoms, used two measures of functioning: the Global Assessment of Functioning (GAF; American Psychiatric Association, 1994) and the Social and Occupational Functioning Scale (SOFAS; Goldman, Skodol and Lave, 1992). The SOFAS was derived from the GAF, and uses similar descriptions of social and occupational functioning, but omits references to symptoms. Both of these measures use a scale of 0–100, described in 10 subdivisions, and are based on clinical judgement.
Recognizing the lack of a universal definition of recovery for research, Liberman and colleagues (2002) attempted to introduce consistency by proposing dimensions that reflect the diagnostic criteria for schizophrenia, i.e., symptom remission, vocational functioning, independent living and social relationships. The criteria specified thresholds that were to be attained and maintained for two years. For example, in the case of peer relationships, the criteria specified that the person have a social interaction with a peer outside the family at least once per week. As Liberman and colleagues noted, the thresholds and durations chosen were largely arbitrary, and determined by consensus of professionals. There is a judgemental element to these definitions; assessment of recovery ought not involve moral judgements on a person’s lifestyle. After all, if ‘normality’ is the goal of mental health services, who decides what is ‘normal’ (Ralph and Corrigan, 2005)? We noted earlier that Harding (1994) questioned the definition used in the Vermont study (Harding et al., 1987a1987b), as they found groups of people who, while not meeting all the study criteria for recovery, were nonetheless happy with their way of life. Focus groups made up of researchers, mental health professionals and consumers showed that, although there was broad consensus on the dimensions chosen by Liberman et al.
(2002), the operational definitions gained approval from a greater percentage of researchers than of consumers or practitioners. Consumers preferred to define recovery as an ongoing process, rather than an endpoint or goal.
While acknowledging that there are common elements to recovery, Anthony et al. (2003) pointed out that quantitative approaches do not allow for the diversity of recovery goals among subjects. These authors therefore urge that quantitative approaches be supplemented by the use of subjective, qualitative measures such as narrative analysis and interviews in determining best practices. However, qualitative procedures such as interviews and narratives, and multidimensional batteries such as those described above, are burdensome on consumers, staff and researchers and are often impractical for large-scale studies or routine clinical use – an area in which outcome measurement is becoming more in demand.
