Mental Health Services Research, Vol. 6, No. 2, June 2004 ( C  2004) Service Quality as Measured by Service Fit and Mortality Among Public Mental Health System Service Recipients H. Stephen Leff, 1,2 James C. McPartland, 1 Stephen Banks, 1 Bruce Dembling, 1 William Fisher, 1 and I. Elaine Allen 1 Service fit, defined as consistency between mental health services judged needed and services received was measured for a random sample of service recipients in a public mental health system ( N = 6588). A variant of small area analysis was used to measure the relationship between catchment area mortality rates from natural causes, suicide, and medicolegal causes and area fit scores for a variety of services. We tested the theory-based hypothesis that service fit would predict interarea variations in mortality better than simple measures of amount of service prescribed and received. We also tested the hypothesis that, controlling for relevant demographic and clinical factors, fit would be protective for mortality from all causes. Findings supported the first hypothesis. With respect to the second, service fit for only certain services was protective. Housing and clubhouses services were particularly protective, suggesting the importance of services providing social support. KEY WORDS: service fit; mental health services; mortality. INTRODUCTION Service Fit This paper describes an approach to measur- ing service quality which we have labeled service fit. It also presents a small area analysis (Wennberg & Gittelson, 1973) of the relationship between service fit and service recipient outcomes, in this case mortality, for service recipients in a public mental health system. Service fit is a measure of the congruence between services prescribed or needed and services received (Kahana, 1982; Leff, Mulkern, Lieberman, & Raab, 1994). It is related to recent efforts to measure adher- ence to practice guidelines (Lehman & Steinwachs, 2001). This type of correspondence has been identi- 1 Human Services Research Institute, 2336 Massachusetts Avenue, Cambridge, Massachusetts. 2 Correspondence should be directed to H. Stephen Leff, PhD, Human Services Research Institute, 2336 Massachusetts Avenue, Cambridge, Massachusetts 02140; e-mail: sleff@hsri.org or chow@hsri.org. fied by Donabedian (1980) and others (Eddy, 1992; Levin, Wilder, & Gilbert, 1978) as a measure of qual- ity. As Donebedian notes: “... once it has been es- tablished that certain procedures used in specified situations are clearly associated with good results, the mere presence or absence of these procedures in these situations can be accepted as evidence of good or bad quality” (p. 83). According to the In- stitute of Medicine: “Efforts to ensure high quality care must prevent or, alternatively, detect and overcome three main problems: (1) overuse of unnecessary care and of inappropriate care, (2) underuse of necessary care, and (3) poor performance (in both the technical and interpersonal senses)” (italics added, Institute of Medicine, 1992, p. 100). As a measure of quality, ser- vice fit falls into the category of practice measures, or measures of the types and amounts of service persons receive (De Geyndt, 1970). Quality also can be mea- sured in terms of the congruence between standards and the conditions of facilities, records, and other administrative apparatus (structure), the most com- mon form of quality assurance practiced by accred- itation organizations, and the congruence between 93 1522-3434/04/0600-0093/0 C  2004 Plenum Publishing Corporation