Psychometric Properties of Quality of Life Assessment Tools in Morbid Obesity: A Review of Literature

Introduction

Obesity is a complex metabolic disease caused by excessive accumulation of fat (1). It is usually defined by body mass index (BMI) (2); BMI ≥ 40 or BMI > 35, which are associated with co-morbidities, are defined as morbid obesity (3).

The universal prevalence of obesity has been alarmingly increasing. World Health Organization (WHO) in 2015 reported that more than 1.9 billion people are overweight and 600 million suffer from obesity (4). In Iran (2010), the prevalence rates of overweight, obesity, and morbid obesity were 28.6%, 10.8%, and 3.4%, respectively (5).

Obesity, particularly morbid obesity, imposes significant health risks to the lives of patients with physical, psychological, socioeconomic, familial, or behavioral problems (6) and undermines their quality of life (QOL). Based on the definition put forth by WHO, QOL is an individual's perception of their position in life in the context of the culture and value systems in which they live. QOL is affected in a complex way by social relationships, personal beliefs (e.g., connectedness to a spiritual being, meaning of life, wholeness, and spiritual strength), and relationship to the environment (7). Spirituality and religiousness have also become a major component of QOL (8).

In a QOL study, an overall outcome index encompassing physical, psychological, social, spiritual, and financial aspects of a certain health state and medical interventions is added to the conventional clinical indices studied. For most patients, including morbid obese patients because of poor QOL (9), QOL is more important than the traditional outcome measures in medical
care (10).

The increasing number of studies on the impact of therapeutic interventions on QOL shows that researchers and clinicians tend to have a holistic attitude toward the patient (11). Numerous studies also showed that the measurement of QOL in obese patients is useful to evaluate the effects of treatment (12, 13) and may influence the development of clinical pathways, service provision, healthcare expenditures, and public health policy. Hence, clinicians, researchers, and policy makers must rely on valid measurement instruments (14).

Although generic instruments for measuring QOL, such as the Short Form-36 (SF-36) (15), provide useful information, they are not designed to measure the specific range of health-related problems experienced by individuals with morbid obesity. A recent study by Kolotkin et al. (16) found differences between weight-related and generic measures of health-related QOL in a one-year weight loss trial, emphasizing on the potential value of using more than one measure in a trial including a disease-specific questionnaire.

For the specific measurement of health-related patient-reported outcomes such as QOL, it is important to evaluate the methodological quality of studies in which the measurement properties of these instruments are assessed. When studies on measurement properties have high methodological quality, their conclusions are more reliable. A checklist containing standards for design requirements and preferred statistical methods is a beneficial tool for this purpose. Former studies developed the Consensus-based Standards for the selection of health status Measurement INstruments (COSMIN) checklist for evaluating the methodological quality of studies on measurement properties (17, 18). The checklist can be used, for instance, in a systematic review of measurement properties, through which the quality of studies on measurement properties of instruments with a similar purpose is assessed, and results of those studies are compared to select the best instrument (19). In the present study, we attempted to review and critique the psychometric properties of some specific tools by COSMIN checklist and their utility for morbidly obese individuals.

Methods

This literature review was performed in 2017 by searching PubMed, Web of Science, PsycINFO, Ovid, Elsevier, ScienceDirect, and google scholar databases. The search terms were “morbid obesity”, “scale”, “instrument”, “questionnaire”, and “index”, “quality of life”, combined with search terms for morbid obesity. We retrieved English-language articles published from 1989 to 2017.

In the sampling stage, the research team selected all the studies reporting validation of morbid obesity QOL-specific questionnaires or their use in clinical trials. The inclusion criteria were inclusion of at least one measure of reliability or validity as defined by the COSMIN criteria (complete reporting of validity, reliability, responsiveness, and interpretability) (20). Articles without available full-texts and with incomplete psychometric properties and non-English publications were excluded (Figure 1).

For each psychometric property defined by the COSMIN criteria, the relevant data were extracted by two independent researchers. The relevant data included the type of psychometric measurement.

Figure 1. Chart of the study process

Results

The initial search yielded 840 articles, after the second evaluation, 50 more relevant articles were obtained. Then, duplicate articles were excluded, and the number of articles decreased to 38. In the review of the articles, only the articles on design, complete psychometric measures of tools, and construction of QOL-related tools were retained in the study (n=21). Finally, the full texts of 19 articles were considered for the evaluation and psychometric properties of eight morbid obesity QOL tools were analyzed using the COSMIN checklist (Table 1).

We identified these assessment tools with varying psychometric properties reported two to eight domains of QOL. While some domains (e.g., sexual life, social health, and eating) have too many assessment tools, other domains have many (e.g., psychological, self-esteem, activity, and work), few

Table 1. Summary of the final articles reviewed

(e.g., physical appearance), or no (spiritual, religious and cultural) tools. The number of items contained in the questionnaires ranges from 6 to 74, and all the questionnaires are self-administered. These tools were criticized in detail based on the COSMIN checklist in Table 2, and their psychometric properties were explained as follow:

Impact of Weight on Quality of Life Questionnaire (IWQOL)

The IWQOL was the first specific questionnaire to assess QOL in obese individuals. The initial version of IWQOL is a self-report tool rated using a 5-point Likert scale (never correct to always correct) with 74 items (21). In this questionnaire, individuals are asked to describe the effects of weight gain on their performance in eight areas of health, social/interpersonal, work, mobility, self-esteem, sexual life, daily activities, and comfort with food (22). This tool was designed by clinicians for obesity treatment based on patients' concerns about their weight. Its completeness and accuracy were approved by the patients. Its construct validity, as well as test-retest and internal consistency reliability were approved (21, 23).

The new version of the questionnaire is available with 31 items and 5 domains including physical performance, self-confidence, sexual life, public distress, and job. Thecorrelation between the original version and its short form is considerably high (from 0.948 [sexual life] to 0.974 [total score]). The short form has also powerful psychometric properties (22). Its internal consistency

Table 2. Tools measuring quality of life in morbid obesity patients

 Table 2 .continued

IWQOL: Impact of Weight on Quality of Life; OP: Obesity-Related Problem; BAROS: Bariatric Analysis and Reporting Outcome System; BOSS: Bariatric and Obesity-Specific Survey; M-A QOL: Moorehead-Ardelt Quality of Life; QOL OD: Quality Of Life, Obesity and Dietetics; BQL: The Bariatric Quality of Life. V1 - Content Validity; V2 - Criterion Validity; V3 - Construct Validity; R1 – Stability; R2 - Internal Consistency.

*Dash (-): It means that the property is not checked in the tool

**Checkmark (√): It means that the property is checked and is properly evaluated in the tool.

reliability coefficients range from 0.90 to 0.94 for the five sub-scales and 0.96 for the total score. Confirmatory factor analysis confirmed the adequacy of the scaling of the tool (24). The scoring of the questionnaire has been calculated from 0 to 100. The initial data indicated that both types of questionnaires have good test-retest and internal consistency reliability (25).

Obesity-Related Problem (OP) Scale

OP scale as a specific tool designed for Swedish obese individuals. This study evaluates the long-term weight loss effects resulting in bariatric surgery and the impact of diet on the psychosocial function. OP scale consists of eight items rated using a 4-point Likert scale. This scale requires the individuals to express their suffering from obesity in special situations such as private parties at their home, friends’ or relatives’ homes, restaurants, social activities, holidays, clothes shopping, swimming in public places, and having intimate relationships (26). The responses are computed as a total score ranging from 0 to 100. The higher scores indicate more optimum psychosocial function (27).

This scale is valid and responsive to weight loss due to surgery and other treatments. There is a dose-response relationship between weight loss and changes in OP scale scores. The function of this scale has been tested in several large samples (obese and non-obese) and in different gender, age, and BMI subgroups. The validity of the assumptions of the scale provided strong support for its construct validity. Homogeneity and dimensionality of this scale were confirmed by exploratory factor analysis (64%). Multi-trait/multi-item analysis test showed the convergent validity of the item, and its item-total correlation was 0.4. The evaluation of OP divergent validity with other QOL tools indicated that the OP scale items have more correlation with their own scale than the other tools (r=0.06-0.54). Its internal consistency reliability was confirmed with α>0.90 (28).

Bariatric Analysis and Reporting Outcome System (BAROS)

The BAROS tool was designed by Oria Moorehead with the goal of universal assessment and standardization of the same reports in bariatric surgery (29), which includes five dimensions of weight, related illnesses, QOL, complications, and new surgeries. The final score classifies the results into five levels from excellent to insufficient (30).

 For designing the BAROS, at first seven dimensions were selected by the American Society of Obesity based on the health risks, list of complications for bariatric surgery, and hospital admission. Then, several psychologists were asked to design a simple, short, and easy-to-respond test to assess QOL after weight loss.

At first, the three dimensions of weight loss, improvement of medical conditions, and QOL were considered. Then, they were upgraded to five dimensions. For global evaluation of the results by this tool and its clinical application, the tool was reformulated and the sixth question about eating behaviors was added to it, and its analysis method was restructured (31).

Laval Questionnaire

The 44-item Laval questionnaire was designed in France (32). Its items were categorized into six dimensions: symptoms, activity/mobility, personal hygiene/clothing, emotions, social interactions, and sexual life. Each dimension was scored on a 7-point Likert scale with higher scores meaning better QOL.

In this questionnaire, the patients were asked to report that during the recent four weeks, how obesity has affected their lives (15). The validation of this questionnaire was performed at the French Laval Hospital, the busiest bariatric surgery center for adult patients.

At first, all patients completed the Laval questionnaire, SF-36, IWQOL-Lite, Beck Depression Inventory (BDI-II), and Rosenberg Self-Esteem Scale (SES). After two weeks, in case of clinical stability and before any interventions, Laval questionnaire was given again to the patients. One year after the surgery, this set of questionnaires was completed again by the intervention group.

The test-retest reliability was significant in all dimensions (symptoms: r=0.93, activity/mobility: r=0.90, personal hygiene/clothing: r=0.85, emotion: r=0.90, social interactions: r=0.87, and sexual life: r=0.84; P>0.01). Cronbach's alpha was above 0.7 in all the dimensions, indicating good internal consistency for all the aspects: (symptoms=0.84, activity/ mobility=0.78, emotions=0.90, social interactions=0.86, and sexual life=0.65).

There was a high correlation between Laval's questionnaire and other questionnaires, except for the SES (construct validity). In assessing the known-groups validity of Laval's questionnaire, at first the questionnaire did not have the ability to distinguish between the two groups (treatment and non-treatment), but it showed a significant difference during follow up. For item reduction and item clustering, the impact method was used instead of factor analysis (15).

Bariatric and Obesity-Specific Survey

Various methods such as literature review, other general and specific tools of diseases, patient and participation’s comments, and discussion with health care professionals were used to generate this questionnaire’s items, which led to the generation of a 200-item pool. These items were selected by a multidisciplinary team consisting of a surgeon, two psychiatrists, an instrument developer, a nurse, and a nutritionist.

To assess the validity of the initial questionnaire, the questionnaire was administrated as a pilot to a small group of patients (12 persons), and each item was evaluated in terms of legibility, relevance to obesity and bariatric surgery, and effect on obesity-related QOL. The items with higher scores were selected in each dimension. Finally, 81 items in 10 dimensions were selected based on the experts’ comments and the pilot group.

For scoring, the Likert scale was used ranging from 0 (zero) to 4 (100%). To perform factor analysis, 200 adults were selected and then divided into three groups: patients with BMI ≥ 35 who visited the bariatric clinic, patients under bariatric surgery, and normal weight persons from the hospital staff. The participants were asked to complete SF-36, Moorehead-Ardelt II Questionnaire (M-AQOL), Hospital Anxiety and Depression Scale (HADS), and BOSS-81. After two weeks, the participants were asked to complete BOSS-81 again. By factor analysis, the items that were not loaded on any of the extraction factors or were not included in any particular factor were removed, and the extracted factors comprising less than 3 items or those that were incomprehensible were deleted.

Finally, the matrix pattern showed 42 items with six factors (incapacity, work/wellbeing, social function, appearance/health, eating patterns, and sexual health). Factor analysis also indicated that this tool had general (factors 1, 2 and 3) and specific (factors 4, 5 and 6) subcategories.

The Cronbach's alpha (α>0.7) indicated acceptable internal reliability. Pearson correlation coefficient reflected that BOSS-42 dimensions were significantly and positively correlated with the dimensions of SF-36 (P<0.001; r=0.36-0.77), HADS (P<0.001; r=0.41-0.78), M-AQLQΠ (r=0.38-0.84; P=0.20-0.92), but it had a non-significant correlation with the participants’ feedback scores (r=0.009–0.127; P=0.20–0.92) (33).

Moorehead-Ardelt II Questionnaire

The M-AQOL, as a part of the BAROS, has been specifically designed for obese patients (30). BAROS includes three domains including weight loss, improvement of medical conditions, and QOL. Points are assigned based on changes in these domains. A maximum of three points is given to each dimension to evaluate the changes after surgical and medical interventions. Each point is defined in five groups from failure to excellent (32).

The original M-AQOL is single-page and easy-to-use with simple drawings and items on all the five dimensions of QOL including self-confidence, physical activity, social life, work condition, and sexual activity. A European study showed that M-AQOL is valid and reliable (32). Since this questionnaire was kept open to receive surgical community suggestions (34), the sixth question related to food perception was added. For scoring, all the questions are scored based on a 10-point Likert scale. This change makes the tool more sensitive (35).

The high Cronbach’s alpha coefficient (0.84) indicates good internal consistency of the instrument and its reliability. The obtained scores had a significant correlation with all aspects of the SF-36, except for the performance domain (P<0.01), which marks its construct validity. The negative correlation between M-AQOL and BDI-II indicates that higher depression diminishes QOL (36).

Quality of Life, Obesity, and Dietetics rating scale

This questionnaire was originally designed through qualitative assessment and face-to-face interviews with obese patients. In these interviews, 31-questions are asked about the relationship between QOL and overweight. This questionnaire is able to identify the points that were poorly recognized by the French translation of the IWQOL, including attitude towards food, diet-related problems, feeling guilt, embarrassment, and functional barriers associated with lower back pain. In this regard, a 17-item scale has been developed to cover these specific points. Therefore, a new combination of the IWQOL and the 17-point scale was generated with 91 items. Due to the large number of items, lack of clinical application, and the shortage of access of physicians and patients, the number of the items was reduced (37).

For the validation of the questionnaire, it was compared with SF-12 (38). Physical dimension of the tool had a significant correlation with SF-12 physical score (r=0.72) and psychological dimension with SF-12 psychic score (r=0.64) (38, 39).

The final version of the questionnaire constitutes 36 items with five domains of physical impact, psychosocial impact, sex life, comfort with food, and diet experience. This questionnaire has an appropriate construct validity and internal consistency reliability (37).

The Bariatric Quality of Life (BQL) Index

Despite evaluation of the psychometric properties of the 31-items IWQOL, this questionnaire is inefficient for evaluating gastrointestinal symptoms after obesity surgery. Therefore, BQL was designed to evaluate weight-related QOL and digestive gastrointestinal symptoms related to the gastrointestinal system.

For the initial design, SF-36 and BAROS were given to 50 patients. They were asked to provide their suggestions about the usefulness of various items. Based on these suggestions and opinions of various surgeons about its face validity, the items were incorporated for designing a new BQL. The new questionnaire was evaluated among 110 patients and modified again. Finally, 19-items were distributed over five factors of psychological well-being, social function, physical function, problems with bariatric surgery, and obesity-related co-morbidities. This questionnaire was scored from 0 to 78.

To evaluate its construct validity, the correlation of this questionnaire with other questionnaires was used. In this regard, six months after surgery, there was a significant correlation between BQL and SF-12 (r=0.79), Gastrointestinal QOL Index, (r=0.52), and BAROS (r=0.64). This tool had good internal consistency reliability (α=0.71-0.86). Factor analysis suggested that BQL includes bariatric QOL items and a secondary section in relation to associated complications and gastrointestinal symptoms (40).

Discussion

In this study, the psychometric properties of QOL assessment tools were investigated in obese, especially morbid obese, individuals. The results showed that although several tools have been designed for measuring QOL, based on the COSMIN criteria, it seems that some of them completely reported dimensions of the concept (OP scale and BQL), the number of items (OP scale), and their psychometric properties such as validity (BAROS).

With respect to usability, some tools required little time and were easy to complete. For example, although M-AQOL is user-friendly, easy, short, simple to evaluate and score, independent from the interviewer, and cost-effective, some graphical symbols of the tool have adjusted its culture sensitivity and the intercultural and linguistic factors affecting on the tool’s reliability. Nevertheless, it seems that this tool does not have good content validity due to the limited number of items (35), while content validity is the main basis and foundation for other psychometric properties of tools, especially for strong construct validity. This drawback is an important area for future tool development.

For responsiveness (sensitivity to changes), six of the eight tools tested responsiveness, which is required to evaluate the impact of clinical trials and interventions. For example, the most important feature of the Laval questionnaire, in addition to having good psychometric properties, is its sensitivity to changes in treatment (15). In addition, the IWQOL, indicated the clinical indicators in their performance in each dimension (responsiveness) (25). These results were consistent with those of the COSMIN checklist, where responsiveness is defined as the ability of an instrument to detect change over time in the construct to be measured and it is related to validity.

For interpretation (qualitative significance of minimal changes with importance in tool score), no tool had testing of interpretability. For instance, although OP scale is responsive to clinical changes over time and it is appropriate for assessing the effects of QOL in obesity interventions, it is unable to recognize clinical indicators and recovery in weight loss function and evaluate all the aspects of QOL (28), which renders an inappropriate interpretation. Despite simplicity, evidence-based nature, and the objectivity of BAROS, it was not able to properly analyze before and after implementation of several clinical, surgical, and endoscopic interventions related to the treatment of obesity. It can also evaluate a few changes in surgical patients pre- and post-operation, and it does have poor responsiveness (29). However, these criteria for the COSMIN's checklist are the main basis and foundation for assessing the applications of tools. These shortcomings in these tools suggest that it is an important area for the improvement of future tool developments. For assessing reliability, four of eight tools had both good stability and internal consistency (i.e., IWQOL, Laval, BOSS, and M-AQOL). However, others had only internal consistency, while reliability is more related to stability.

In evaluation of domains of morbid obesity QOL, few tools address cultural domain. QOL of obese people should be considered in interpersonal, cultural, and religious contexts in which obesity occurs (41, 42). The cultural features of each society are expressed as the values, beliefs, and unique functions of each group that has specific strategies for promoting or maintaining health and preventing diseases. Culture is an important factor in QOL because it defines the purpose of life in health and illness and affects the person’s perception of health and illness, QOL, thoughts, and feelings. Adaptation of health So cultural adaptation is significant in the use and design of tools,  because a tool that is designed for a social or cultural group cannot be applied to another group in another culture.for a social or cultural group cannot be applied to another group in another culture (43, 44).

Currently, some tools have been adjusted to the cultural, religious, and social conditions of certain communities. For example, (i) the Laval questionnaire is only designed for the French community and cannot be easily applied in other cultures (15). Although (ii) QOLOD has good psychometric properties, this questionnaire can be used in clinical trials in French-language countries and cultural adaptation must be made for using it in another country (35). Since the analysis of bariatric surgery has several variables, a multi-dimensional scale is needed to evaluate all its aspects; therefore, (iii) BAROS must be adapted to all the bariatric communities (29). The (iv) psychometric evaluation of BQL is performed for the German version and its English version needs to be validated for clinical application. Its limitations are lack of suitability for other cultures (40).

Since the meaning of a concept may vary from one culture to another (45), experts in questionnaire development agree that the tool should be directly derived from the target population and should be appropriate for the culture and lifestyle of the countries where it is used. This criterion should be considered in item generation, as well. Nevertheless, in designing most of the mentioned tools, obese patients’ views were not considered and the tools were created by deductive method. Moreover, in translated questionnaires attention was not paid to the translation process, cultural adaptation, and linguistic differences, hence the obtained results are not generalizable (46). Therefore, valid and reliable tools, acceptable for the target population and able to measure the concept and identify its differences in the studied population, should be applied.

In addition to neglecting the culture domain in QOL tools development, they have not considered spiritual and religion dimensions. QOL is influenced by complex factors such as relation to the environment, spiritual beliefs, and religion (47). Because religious beliefs can influence QOL, designing tools with acceptable psychometric properties is recommended for assessing the QOL of obese patients based on the religious features of different societies.

Limitations of this literature review were incomplete information regarding the item generation process in some of the tools and limited information on their usability, reliability, validity, interpretability, and responsiveness. Although we did not find much information on responsiveness, a more detailed literature search for each tool would help find evidence for their responsiveness. Finally, some tools included in this review have multiple versions that were not always noted in our sources. Future users of these tools should search for and study different versions that might be more appropriate.

Implications for Practice

Our results revealed that some of the tools focused on the cultural domain, while no tool has considered the spiritual and the religious domains. Despite the lack of a complete and appropriate scale for the measurement of morbidly obese individuals’ quality of life; Laval questionnaire is a proper scale that can be recommended to researchers. However, a vital need is felt for developing an instrument based on cultural and social characteristics of different societies with acceptable psychometrics properties and evaluating responsiveness and interpretability of these tools for measuring quality of life in morbid obese patients. In addition, we suggest designing a valid and reliable tool for research in different age groups and evaluating other tools in patients with morbid obesity in the future studies.

Acknowledgments

The study is a part of a PhD dissertation (NCRCCD-9416) submitted to Ahvaz Jundishapur University of Medical Sciences. The authors would like to thank Ahvaz Jundishapur and Shiraz Universities of Medical Sciences, Center for Development of Clinical Research of Namazee Hospital, and Dr. Nasrin Shokrpour for editorial assistance.

Conflicts of Interest

None declared.