Systematic review of health-related quality of life models

Of the 100 articles, 57 used an existing HRQOL model as a guide and 25 derived a HRQOL model. Interestingly, 18 articles used an existing HRQOL model as an initial guide and then also derived a revised model based on the findings. Figure 1 shows that of the 100 full-text articles included in the review, 77 either derived a HRQOL model (n = 25) or were guided by a HRQOL model that was used only once or twice (n = 52). There was little consensus among the models used, with each article essentially citing a different model. These 77 articles are summarized in Additional file 1: Tables SA and SB included as an appendix for this paper. Of the 25 articles that derived a HRQOL model, 24 were disease-specific, and 1 was a consensus paper on HRQOL (Additional file 1: Table SA). The disease-specific models were classified as using a uniquely derived HRQOL model based on the findings. For example, Barr and Schumacher [17] identified six categories of HRQOL specific to individuals receiving medical nutrition therapy. Similarly, Klassen, Pusic, Scott, Klok, &; Cano [18] examined the impact of breast conditions and surgery to develop a quality of life framework specific to breast surgery patients. Because there was such a wide variation in disease states, HRQOL domains, and particular characteristics, findings could not be adequately synthesized. Of the 52 articles that were guided by a HRQOL model that was used only once (n = 46) or twice (n = 6), only three HRQOL models were cited twice (Additional file 1: Table SB). Those used twice were Maslow’s hierarchy of needs (n = 2), Mishel’s Uncertainty in Illness Theory (n = 2), and Stewart’s conceptual model of factors affecting dying patients and families (n = 2) (See Additional file 1: Table SB for details).

As depicted at the bottom of Figure 1, there remained a total of 23 articles that cited the same model 4 or more times. As shown in Table 2 (and in Figure 1), the most common existing HRQOL models found in the literature were those by Wilson and Cleary [12] (n = 14), Ferrans and colleagues [13, 19] (n = 3), and the World Health Organization (WHO) [20, 21] (n = 4). Two additional articles used a combination of two of these models. Ferrans et al. [13] used the Wilson and Cleary [12] model as a guide to derive a revised model of HRQOL [13]. Valderas and Alonso [22] used both Wilson and Cleary [12] and the WHO [20] models. Schematic diagrams for each of the three most common HRQOL models have been published in Wilson and Cleary [12], Ferrans and colleagues [13], and the World Health Organization [20, 21], and are described in more detail in the results section. The largest group (n = 10) of the 23 articles in Table 2 reported observational studies (descriptive or correlational) and focused on patients with chronic illness, with sample sizes ranging from 61 [23] to 917 [24]. Literature reviews (n = 6) and instrument development studies (n = 3) were also found. Only one randomized controlled trial was found [25], along with one mixed-methods study [26], one qualitative study [27], and one article that described a model revision [13].

Table 1 details the critique of the three most commonly used HRQOL models found in the literature over the past 10 years using criteria by Bredow [10]. Wilson and Cleary’s [12] model of HRQOL combines two paradigms, biomedical and social science. Their model is a taxonomy that includes five major well-defined domains: biological, symptoms, function, general health perception, and overall HRQOL. However, the definitions for two other domains, individual and environmental characteristics, were not made explicit. Each domain is related to the others, and reciprocal relationships may exist. The authors suggest that environmental and individual factors are associated with outcomes, thus affecting total HRQOL.

Ferrans, Zerwic, Wilbur, and Larson [13] published a revision of Wilson and Cleary’s [12] HRQOL model. The five major domains of the original model were retained. Ferrans and colleagues [13] made explicit the definitions for individual and environmental characteristics, and they simplified the depiction of the model by removing non-medical factors and labels on the arrows portraying the relationships in the figure. In addition, they contributed further theoretical background about the main concepts in the model [19] and provided examples of instruments to enhance measurement. According to Ferrans et al. [13], the model depicts dominant causal associations; however, reciprocal relationships are implied. An explicit assumption is that understanding relationships among these components will lead to the design of optimally effective clinical interventions. The revised conceptual model could be applied to any health care discipline.

The World Health Organization International Classification of Functioning, Disability, and Health (WHO ICF) is a model designed to provide a description of health and health states, while providing a unified and standard language that can be used across disciplines and cultures [3, 20, 21]. The WHO ICF has evolved over time from a focus on “consequences of disease” in 1980 to “components of health” in 2001 [20, 21]. The more recently developed WHO ICF-CY covers infants, children, and adolescents [3]. The WHO has conceptualized HRQOL as an individual’s perception of his or her health and health-related domains of well-being [3, 21, 43]. Health and health-related domains have been further conceptualized in terms of functioning within the WHO ICF model. The WHO ICF model includes components within two main parts. Part 1 focuses on functioning and disability (body functioning and structures, activities, and participation), whereas Part 2 addresses contextual factors (environmental and personal). The main concepts are well-defined overall, with explicit propositions and assumptions. However, unlike the models by Wilson and Cleary [12] and Ferrans and colleagues [13], the WHO ICF is not specific to HRQOL. Cieza and Stuki [43] assert that the WHO ICF categories under functioning can serve as the basis for the operationalization of HRQOL but are not the only potential application of the WHO ICF. For example, Miller and colleagues [44] used the WHO ICF as a framework to organize a comprehensive overview of nursing and interdisciplinary care of the stroke patient. The WHO ICF serves more as a mapping and classification framework than as a guide for hypothesis generation in the area of HRQOL.

Critique of the internal components of the HRQOL models using the Bredow criteria [10] indicated many similarities and some differences. All were fairly complete in the descriptions and definitions of HRQOL, with some gaps in the influence of management of therapeutic regimens and self-management on quality of life. Most existing models focus on the influence of symptoms rather than on management related to the condition. For example, for those with diabetes, both symptoms (such as hypoglycemia) and management (such as frequent checking of glucose levels) are important influential factors for HRQOL. Within the WHO ICF model [3], there was some definitional overlap between activities and participation. The Ferrans et al. [13] model was more complete and clear than that of Wilson and Cleary [12] because of the revisions and because of better definitions for individual and environmental factors. The relationships among the concepts were less clear in the Wilson and Cleary model [12], whereas the Ferrans et al. model [13] added clarity. Both the Wilson and Cleary [12] and Ferrans et al. [13] models implied potential reciprocity. In contrast, the WHO ICF [3] was explicit with the depiction of causal and reciprocal relationships.

There were greater variations among the three models when critiquing model fit with operational application. The models were similarly parsimonious, yet the complexities of multiple relationships were described. They made sense for use in real-world settings and have been used to guide research and practice. A major difference is that the Wilson and Cleary [12] and Ferrans et al. [13] models specifically explain HRQOL, whereas the WHO ICF model [3] describes health related to functioning and disability. In addition, though the Wilson and Cleary [12] and Ferrans et al. [13] models were primarily intended for application to individuals, the WHO ICF model [3] could be used to explain the health of individuals, families, communities, populations, and cultures. With the former, adaptations may be needed for use with families, communities, and individuals unable to report their own HRQOL, such as infants and young children and those with cognitive impairment. Empirical evidence for use of the models for intervention research is limited. However, the Ferrans et al. [13] model and the WHO ICF model [3] have robust potential for guiding the design of interventions that could be tested and applied in practice settings. The WHO ICF may be more applicable to practice situations for needs assessments, matching treatments with conditions, and evaluating outcomes because it is primarily a classification and mapping system. All three models were at a similar level of development emerging from the two paradigms of biomedical and social sciences.

Our search strategies were limited to selected databases (PubMed, MEDLINE, CINAHL, and PsychINFO) and keywords (e.g., quality of life, health-related quality of life, conceptual framework, conceptual model, and theory). Given that standard keywords were used within each search engine, any article indexed by that search engine would have been captured; however, follow up manual searches and review of reference lists might have revealed additional citations. The search strategies were specifically designed to capture articles that were guided by or derived HRQOL models that were further analyzed in detail by the reviewers. All reviewers were doctorally-prepared and a research librarian assisted with the searches. Because the aim of our paper was to identify the most frequently-used HRQOL models found in the literature over the past 10 years, a complete synthesis of disease-specific models was not undertaken. Future work to analyze uniquely derived disease-specific HRQOL models may provide unique HRQOL domains that might further inform the three more commonly used HRQOL models. For example, Klassen et al. [18] qualitatively derived a HRQOL model for women who had undergone breast surgery. Their model consisted of six themes (satisfaction with breasts, satisfaction with process of care, satisfaction with overall outcome, psychosocial well-being, sexual well-being, and physical well-being). The satisfaction with process of care theme further informs both Ferrans and colleagues’ model [13] and the WHO ICF [3, 20, 21] as an important characteristic of the environment. Sexual well-being could further inform the functional status domains in all three models. Identifying domains that are unique to disease-specific models, or particular characteristics such as feedback or recursive patterns to address dynamic changes in HRQOL with time, may further inform or strengthen the rationale for using the three existing HRQOL models.