Introduction

It has long been understood that somatic illnesses and their treatment may have a considerable influence on patients’ health-related quality of life (QoL). Since the 1980s a range of generic and disease-specific QoL measures have been developed in efforts to gain an understanding of this influence [1]. Consequently, patient-reported QoL measures have increasingly been included in randomized clinical trials to demonstrate the effect of treatment beyond clinical efficacy and safety [2].

The majority of these QoL questionnaires are based on domains formulated by researchers and health policy makers [3]. However, a repeated finding is that externally defined domains may not reflect the domains that patients consider relevant for their QoL [e.g., 4–6]. For example, Morris et al. [4] compared the health-related QoL domains identified by patients undergoing major surgery with seven commonly used HRQoL instruments. While the domains’ ‘concern about quality of care’, ‘cognitive preparation’ and ‘spiritual wellbeing’ were frequently mentioned as constituting patients’ QoL, these were not assessed by most of the instruments.

While the usefulness of standardized QoL questionnaires has been repeatedly demonstrated and is beyond doubt, we lack a comprehensive overview of QoL domains that patients themselves nominate as constituting their QoL. Such insight is needed to ensure that the relevant domains are addressed and to guide questionnaire selection. We therefore undertook a literature review of qualitative studies that asked patients to identify domains constituting their QoL. To our knowledge, this is the first attempt to provide a comprehensive overview of patients’ self-nominated QoL domains.

Two types of studies are relevant for this review. First, studies using the Schedule for Evaluation of Individual Quality of Life (SEIQoL) [7, 8] are relevant, as they make the perspective of the individual central to defining relevant QoL domains. This widely used individualized measure [9] requires that patients nominate five domains they consider most relevant to their QoL. When patients have difficulty nominating five domains, a prompt list can be used consisting of the cues: family, relationships, health, finances, living conditions, work, social life, leisure activities and religion/spiritual life [10]. The SEIQoL generates an overall index score that is the result of the individual’s rating of his/her functioning in and importance of each self-nominated QoL domain. The SEIQoL thus provides a wealth of qualitative data about the content of the nominated domains, although most studies only report the quantitative results related to the overall index scores. We specifically excluded individualized measures that did not directly ask for life domains relevant for patients’ QoL. For example, the Patient-Generated Index (PGI) [11] was excluded because it asks patients to nominate the five most important areas of life or activities that are affected by their condition as was Cantrill’s ladder [12] that asks patients to describe their worst imaginable and best imaginable life satisfaction. Individualized measures such as the Audit of Diabetes-Dependent Quality of Life (ADDQoL) [13] and the World Health Organization Quality of Life (WHOQoL) [14] were excluded, since they only allow for individual weighting of predefined QoL domains. All of these measures thus have a slightly different scope than that in the current review.

A second cluster of studies is also relevant; these explore somatically ill patients’ self-generated QoL domains to evaluate the content validity of existing, standardized QoL questionnaires or to improve the quality of care. The interview question(s) used to elicit patients’ self-defined QoL domains vary per study, e.g., respondents are explicitly asked what their personal perception of quality of life is, how they would describe quality of life, or what the term quality of life means to them. To differentiate these studies from those using the SEIQoL, we refer to this group of studies as those using study-specific questions.

This review thus includes studies reporting qualitative data originating from the use of the SEIQoL and from studies employing study-specific questions. The domains that patients report and/or researchers aggregate and present may be influenced by several factors. We will address one of these in examining whether the method of enquiry is related to generation of different domains. The use of the SEIQoL prompt list is likely to result in the presentation of QoL domains similar to the prompt list, whereas the use of study-specific questions may result in different QoL domains. We therefore compare the QoL domains presented in studies using the SEIQoL with those in studies using study-specific questions (Appendix 2).

Methods

Literature searches

We conducted two systematic literature searches in the databases PubMed/Medline, CINAHL and PsychInfo for papers published from 1980 on using (1) SEIQoL and (2) study-specific quality-of-life questions. We conducted consecutive literature searches employing the following search terms: SEIQoL, SEIQoL-DW and patient(s) as search terms (literature search 1) and quality of life, QoL, content, definition, item generation, content generation and patient(s) (literature search 2). The literature searches were initiated in March 2007, and updated until March 2008.

Study selection

Two researchers independently assessed the eligibility of all abstracts retrieved by our literature searches in PubMed/Medline and PsychInfo (ETB, MK) and CINAHL (ETB, MV). The researchers involved discussed their findings, and decided on each abstract’s eligibility based on mutual consensus. All studies included in this review met the following criteria: (1) The study presents QoL domains qualitatively generated by respondents residing in Anglo-Saxon (i.e., English speaking) or non-English speaking European countries, which are somatically ill (in contrast to having a psychiatric illness) or have symptoms as the result of their illness at the time of study. (2) The study was published in English between 1980 and September 2008 in an internationally peer-reviewed journal. In addition, the studies met the following methodological quality criteria: (3) The formulation of the interview question(s) is provided. (4) The original data are sufficiently presented to demonstrate the relation between the data and the researchers’ interpretation, i.e., via patients’ quotations or detailed categorization schemes. (5) In studies using multiple assessment points, QoL domains nominated at one separate assessment point are discernible. (6) In studies using study-specific questions, data-analysis is carried out inductively, i.e., without a pre-determined framework for the categorization of nominated QoL domains. In case of multiple publications based on the same patient sample, we only included the paper with the most comprehensive presentation of the qualitative data. Due to the different nature of psychiatric illnesses as opposed to somatic illnesses, and its potential implications for patients’ self-defined QoL domains, we only included studies conducted among somatically ill patients. Reviews and case studies were also excluded.

Categorization of QoL domains

Three researchers (ETB, MS, MV) classified all QoL domains presented in the selected papers in two steps based on mutual consensus. First, most studies reporting data originating from the SEIQoL categorized the self-nominated domains according to the nine domains included in the prompt list. We therefore initially used these same nine domains (e.g., family) or closely related QoL domains (e.g., family-related) for categorization (see Table 1).

Table 1 Categorization of QoL domains included in and highly related to the SEIQoL prompt list
Full size table

Second, two researchers (ETB, MV) independently classified the QoL domains that could not be grouped according to the SEIQoL prompt list domains, into new domains. They discussed the formulation of the domains and the classification with MS until consensus was reached. This iterative process resulted in eight additional domains; psychological functioning, coping/positive attitude, independence, role functioning, feeling of self, cognitive functioning, quality of care, sexuality, and a miscellaneous category (see Table 2).

Table 2 Categorization of QoL domains according to additional, inductively generated domains
Full size table

In order to classify all QoL domains according to the afore-mentioned categorization scheme, we had to tease apart the QoL domains originally presented in 22 papers [8, 17, 18, 24, 26, 28, 29, 32, 34, 36, 38, 39, 41, 43–45, 48, 50–53, 55]. For example, we have separated the single QoL domain family/friends presented in a study by Archenholtz et al. [53] into two QoL domains: family (according to the SEIQoL prompt list) and friends (related to the SEIQoL prompt list cue relationships).

Additionally, we only classified the QoL domains that were presented at the lowest level of abstraction in the articles, since these are closest to the patients’ own definition of QoL. This meant that in 12 papers [8, 16, 18, 22, 38, 46–52, 56] we ignored the overarching themes that authors used to group the self-nominated QoL domains. For example, Cohen and Leis [51] classified the QoL domains ‘physical condition’, ‘physical functioning’, ‘psychological state’ and ‘cognitive functioning’ into the overarching theme ‘own state’. We used the four QoL domains for classification rather than the more abstract construction ‘own state’.

Results

Study selection and characteristics

The literature search for papers using SEIQoL resulted in 61 abstracts (see Fig. 1). Twenty-nine abstracts were excluded based on the inclusion and exclusion criteria presented earlier. The remaining 32 papers [8, 15–45] were examined with regard to our methodological quality criteria, resulting in the further exclusion of six papers [20, 25, 27, 30, 37, 40]. Examination of the references included in the 26 selected papers resulted in one additional paper eligible for this review [7]. Literature search 1 thereby resulted in 27 eligible papers.

Fig. 1
figure 1

Flow chart of the selection of eligible papers resulting from literature search 1 (studies using the SEIQoL)

Full size image

The literature search for papers using study-specific questions yielded a total of 1,765 abstracts (Fig. 2). From these studies, 1,752 were excluded based on the inclusion and exclusion criteria. The remaining 13 papers [46–58] were examined with regard to our methodological quality criteria, which led to the further exclusion of four papers [49, 54, 57, 58]. Additionally, all references quoted in the selected nine papers were examined for eligibility, which did not lead to the inclusion of new papers. Overall, the literature searches yielded a total of 36 eligible papers [27 papers (literature search 1) + 9 papers (literature search 2)] (See Tables 5 and 6 in the Appendices for a summary of the design and results of the included papers).

Fig. 2
figure 2

Flow chart of the selection of eligible papers resulting from literature search 2 (studies using study-specific questions)

Full size image

Half of the included studies were conducted among patients with cancer [15, 16, 18, 19, 21, 22, 24, 28, 34, 35, 41, 44, 47, 50–52, 55, 56], whereas the other studies included patients with a range of other somatic illnesses (see Table 3). In three studies, the patient sample consisted of a combination of both patients with cancer and patients with another somatic illness [38, 39, 45].

Table 3 Patient classification according to somatic illness and method of enquiry for literature searches 1 and 2
Full size table

In most studies, a face-to-face interview was conducted to elicit patients’ QoL domains [7, 8, 15–19, 21–24, 26, 28, 29, 31–37, 39, 41, 42, 44, 45, 48, 50, 51, 55, 56]. In the remaining studies, QoL domains were identified by means of a telephone interview [53], focus groups [47], or a questionnaire employing open-ended questions [43, 46, 52].

Studies using SEIQoL presented a median of 16 QoL domains (range 7–62), and studies using study-specific questions presented a median of 13 QoL domains (range 9–29) (Appendix 1).

Elicited QoL domains

QoL domains categorized according to the SEIQoL prompt list

Table 1 provides the QoL domains categorized according to the 9 domains included in or highly related to the SEIQoL prompt list, as derived from the studies using the SEIQoL and studies using study-specific questions, separately. As the first two columns of Table 1 illustrate, SEIQoL studies are unique in presenting the prompt list domains relationships, finances, and living conditions, whereas family, health, work, social life, leisure activities and religion/spiritual life are also reported by one to two studies using study-specific questions. More interestingly, both types of studies report domains related to the SEIQoL prompt list (see last two columns of Table 1). These domains entail more specific information as opposed to the SEIQoL prompt list domains. For example, we classified the presented domains friends, neighbors, associate with family, lover, and marriage, into the domain relationships-related.

All studies using SEIQoL and study-specific questions report a domain referring to health, either by presenting the SEIQoL prompt list domain health, or in presenting a health-related domain. The majority of the studies employing the SEIQoL report other QoL domains included in or highly related to the SEIQoL prompt list (63–100%), whereas fewer studies using study-specific questions do so (22–89%). SEIQoL studies are unique in presenting the domains marriage and/or partnership and spousal welfare (relationship-related), activity and mobility (health-related) and in presenting specific hobbies (leisure activity-related). Irrespective of the method of enquiry, the domain presented least often is living conditions.

QoL domains categorized inductively

Table 2 displays the classification of the QoL domains that could not be grouped according to the domains included in or highly related to the SEIQoL prompt list. These QoL domains are classified into 8 inductively generated, additional domains. Interestingly, ‘independence’ is mentioned in 74% of the studies using the SEIQoL and is thus more frequently reported than the SEIQoL prompt list domains religion/spiritual life (70%), social life (63%) and living conditions (63%). The other inductively generated domains are less frequently reported in studies using the SEIQoL (4–48%) than in studies using study-specific questions (33–78%). The latter group of studies have more elaborate presentations of domains related to psychological functioning (e.g., the domains relaxation and being without anxiety) and coping/positive attitude (e.g., the domains coping strategies and being able to enjoy things). Conversely, only studies using the SEIQoL (N = 6) present the QoL domain sexuality. Irrespective of the method of enquiry, the domain quality of care is presented least often.

Discussion

Perhaps, one of the most important aspects of patients’ QoL is their evaluation of important life domains. Domains that patients consider important are preferably elicited by qualitative interviews. This information is indirectly captured in standardized questionnaires that use patient-generated item content.

This structured literature review is a first attempt to provide a comprehensive overview of the QoL domains a variety of somatically ill persons themselves consider relevant. The presented domains are found to be robust given that the influence of the method of enquiry on patient’s self-nominated QoL domains appears limited. As expected, SEIQoL studies more frequently report the domains used in the SEIQoL prompt list, whereas studies using study-specific questions report more often the inductively generated domains. However, this finding should not obscure the fact that the domains reported are highly comparable: most domains are presented by both types of studies, albeit with different frequencies. Consequently, the domains listed in Tables 1 and 2 are meaningful and may help future researchers to identify relevant and important domains that may need to be addressed in their studies. Second, our findings confirm that the SEIQoL prompt list covers, to a large extent, relevant domains of patients’ QoL. Researchers wishing to use a more exhaustive prompt list can make use of the current findings. For example, these results indicate that the domains independence, psychological functioning, and coping might be additional candidate domains.

Reflections on reviewing qualitative studies

Our findings need to be considered in the light of this review’s limitations. Firstly, there are inherent limitations in reviewing this qualitative material that have hindered a comprehensive and unequivocal overview. The first consideration lies in the way and level of abstraction and aggregation that is needed to communicate patients’ nominated QoL domains. These abstractions first take place during data collection when the individual patient talks with the researcher, and subsequently at the data recording, analysis and reporting stages. Different studies use different levels of aggregation, which hampers comparisons across studies to a great extent. For example, we cannot be sure whether the presented domain (e.g., family) is mentioned literally by patients or rather is an aggregation of, for instance, the domain (grand) children by the researchers. Similarly, we cannot be sure that the domain sexuality was not mentioned in studies using study-specific questions, since the authors might have aggregated it to the level of relationships.

A second consideration is that in qualitative research the choice of words is of key importance. Some specific words may in fact be synonyms (e.g., financial security versus sufficient income; pain-free versus no pain), whereas slightly different words may be intended to mean entirely different things (e.g., physical capacity versus physical functioning). This interpretative difficulty also holds for QoL domains that are phrased either positively or negatively. For example, is inner peace similar to or different from having no stress? Are positively and negatively formulated words polar ends of the same construct or do they represent different constructs? Consequently, caution is needed when comparing different qualitative domains across studies and across different methods of inquiry.

Reflections on the included studies

Since a review can never be better than its constituent studies, the limitations of the included studies impede a comprehensive review. First, the reviewed studies provided notably little information on the socio-demographic and clinical background of their patient groups. Since the type of somatic illness may affect the specific domains patients consider relevant for their QoL, we have attempted to compare the self-nominated QoL domains among different patient groups, e.g., cancer versus non-cancer. However, information regarding patients’ stage of disease was generally insufficiently presented or lacking. Furthermore, the heterogeneity of diseases did not allow a useful comparison between patient groups.

Second, the majority of studies using the SEIQoL did not provide information on the use of the prompt list. This limits our insight into the process of generating QoL domains, i.e., did patients come up with the domains constituting their QoL themselves, or were they guided in the selection of domains by the prompt list? Additionally, these studies did not describe whether patients experienced difficulty in nominating five QoL domains. However, the requirement to arrive at five QoL domains might result in the nomination of domains that are of lesser importance to the patients. Study reports, in which the use of the prompt list is mentioned, did not differentiate between self-nominated and prompt list-elicited QoL domains. Likewise, studies using study-specific questions did not include information on a possibly minimally required number of QoL domains, or the use of an aid, which may have guided patients to think of specific QoL domains.

Third, in the majority of the reviewed studies using the SEIQoL (N = 18) [15–19, 22–24, 26, 28, 29, 31, 32, 36, 38, 39, 41, 42] and in one study using study-specific questions [55] patients also completed other (QoL) questionnaires, which might have affected the choice of self-nominated domains. Unfortunately, most studies did not provide information about the order in which the various questionnaires were administered. Fourth, most studies took place in the hospital, whereas other studies were conducted at patients’ homes. The site of data collection might have affected patients’ responses [59, 60].

Limitations and strengths

Limitations inherent to our way of conducting this review also merit attention. First, this study’s objective was to review somatically ill persons’ nominated QoL domains. A specific aim was to examine whether the method of enquiry is related to the generation of different QoL domains. In studies using the SEIQoL, patients not only nominate their QoL domains but additionally weigh the relevance of each of these domains. Since only two studies using study-specific questions [53, 55] included such weighting of nominated QoL domains, we were not able to take the weighting of QoL domains into account. In combining all patient-generated QoL domains, we implicitly weighted all domains as equally important. However, patients might find the first two to three mentioned domains more important than the fourth and fifth domain. Therefore, treating all domains as equally important may not be in accordance with the importance patients attach to their domains.

Second, since the research in this area is multidisciplinary, it is difficult to know if we have retrieved an exhaustive list of references. Furthermore, our review encompasses published papers only. However, we are confident that the studies included provide a comprehensive picture of the current research in this area for several reasons. We conducted our literature searches in three different databases using broad search terms to avoid missing relevant papers, which resulted in a large number of abstracts. Additionally, we examined all references quoted in the selected papers for eligibility, and identified abstracts were reviewed for eligibility by at least two persons [61].

Third, even with three persons involved in the inductive categorization of QoL domains not included in the SEIQoL prompt list, other researchers might have proposed other inductive categories. However, it is doubtful whether this would lead to substantially different findings and conclusions. Fourth, this review focused on one individualized measure, the SEIQoL. Whereas this might imply a limitation in our scope, we focused on the most widely used individualized measure that makes the perspective of the individual central to defining relevant QoL domains. As noted earlier, we excluded other individualized measures which were not specifically QoL oriented or that weighted, rather than generated, QoL domains.

Additional guidelines for conducting and reporting qualitative research

Whereas the described limitations are in part inherent to reviewing qualitative data, they also point to shortcomings of qualitative reports that are amenable to improvement. To address these shortcomings, we provide a number of guidelines in addition to more general checklists for conducting and reporting qualitative research [62–64] (see Table 4). Our guidelines supplement these existing checklists in their focus on criteria relevant for this type of qualitative research, e.g., the use of an aid/prompt list to guide respondents in nominating (QoL) domains, and the subsequent distinction between self-nominated and prompt list-based (QoL) domains in reporting the results.

Table 4 Guidelines for conducting and reporting qualitative research aimed at exploring respondents’ self-nominated QoL domains
Full size table

These guidelines might be of particular interest for studies aimed at identifying patient-nominated QoL domains. However, they will also be relevant for other qualitative researchers in enhancing the transparency of the research process and subsequent report of their studies. We hope that this literature review on somatically ill persons’ self-nominated QoL domains, and the provision of guidelines for conducting and reporting qualitative research will stimulate further discussion and improvement of qualitative (QoL) research. Additionally, the guidelines might be helpful to journal editors and reviewers to ensure stringent research.