Assessment of the impact of eating disorders on quality of life using the disease-specific, Health-Related Quality of Life for Eating Disorders (HeRQoLED) questionnaire
- First Online:
- Cite this article as:
- Muñoz, P., Quintana, J.M., Las Hayas, C. et al. Qual Life Res (2009) 18: 1137. doi:10.1007/s11136-009-9542-7
- 176 Views
To evaluate and compare the quality of life (QOL) in patients with eating disorders (ED) and general population, using the disease-specific Health-Related Quality of Life for Eating Disorders (HeRQoLED) questionnaire.
A total of 358 patients with ED completed the HeRQoLED questionnaire as well as the SF-12 and the Eating Attitudes Test (EAT-26) at baseline; 273 patients completed the same instruments after 1 year of multidisciplinary treatment. A total of 305 individuals recruited from the general population completed the HeRQoLED once. Comparison of means was used to assess change. Multivariate models were created to determine variables predictive of change in HeRQoLED scores.
Patients with anorexia nervosa had higher baseline scores (indicating worse perception of QOL) on the HeRQoLED questionnaire and experienced smaller improvements than patients with other diagnoses after 1 year of treatment. After adjustment by relevant variables, body-mass index (BMI) and EAT-26 scores were associated with changes in QOL. SF-12 scores showed significant improvement in the physical health component but not in mental health. General population had lower baseline scores on the HeRQoLED.
As measured by the disease-specific HeRQoLED and generic instruments, QOL in patients with ED improved after 1 year of treatment, though it did not reach the values of the general population.
KeywordsQuality of lifeEating disordersQuestionnairesProspective studies
Health-Related Quality of Life for Eating Disorders
Short-Form Health Survey
SF-12 physical component summary
SF-12 mental component summary
Eating Attitudes Test
Eating disorders not otherwise specified
Clinical Global Impression
Quality of life
Since the mid-1990s, researchers have evaluated the impact of eating disorders (ED) on quality of life (QOL) [1–11]. Not surprisingly, patients with ED report inferior QOL compared to the general population. A key limitation of many of these investigations is the use of generic instruments rather than disease-specific instruments for measuring QOL. Given the relevance of ED in modern-day society, their high prevalence , and their associated morbidity and mortality , it is important to determine more precisely their impact on QOL.
Disease-specific questionnaires offer several advantages over generic instruments [14, 15]. They are better suited to the specific features of the disorder under investigation. Patients perceive that the questions are directly related to their own pathology, thus improving the likelihood they will satisfactorily complete the instrument. Although disease-specific QOL questionnaires do not allow for comparison between diagnostic groups, dealing with the specific nature of the disorder under investigation manages to avoid some of the errors of interpretation associated with generic questionnaires.
A recent study by Doll et al.  suggested that generic questionnaires such as the Short Form Health Survey 36 (SF-36) may yield erroneous QOL data from patients with anorexia nervosa. The investigators suggested that the SF-36 may not be sensitive to detecting emotional stress in these patients. It is also possible that increased physical activity, which the SF-36 associates with improved QOL, may be a sign of increasing severity of anorexia nervosa rather than improvement.
We developed a QOL instrument specifically for patients with ED. The Health-Related Quality of Life for Eating Disorders (HeRQoLED) questionnaire [16, 17] attempts to remedy the deficiencies of generic QOL questionnaires and confer the advantages of disease-specific questionnaires. In this study, we aimed to evaluate QOL in patients with ED using the HeRQoLED questionnaire, as well as the generic Short Form Health Survey 12 (SF-12), at baseline and again after 1 year of multi-disciplinary treatment. A sample drawn from the general population also completed a modified version of the HeRQoLED to compare ED scores with those from the general population. A secondary aim of the study was to evaluate how QOL changes with treatment by initial diagnostic group (anorexia nervosa, bulimia nervosa and eating disorders not otherwise specified) and by severity of ED.
Materials and methods
Participants and procedures
Patients were recruited from three health centers in Bizkaia, Spain, by four psychiatrists experienced with ED. To be included in the study, participants had to (1) have been diagnosed with an ED according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV)  criteria; (2) consent to receive treatment on a regular basis during the study period in one of the three centers; and (3) voluntarily agree to participate after having been briefed personally by their psychiatrist and after providing informed consent. Patients were excluded if they had a clinically serious multi-organ disorder, cerebral organic deterioration, acute psychosis, or other conditions that would prevent them from completing the questionnaires. The ethics review board of each centre approved the study.
Upon inclusion into the study, three measurement instruments were mailed to the participants: the HeRQoLED, the Eating Attitudes Test (EAT-26), and the SF-12. Patients who did not promptly return the materials were sent two reminders at 15-day intervals. The first reminder included a letter asking the patient to complete the questionnaires that had previously been sent. The second reminder included the same letter plus duplicate copies of each measurement instrument.
Each patient took part in a psychopharmacologic and psychotherapeutic treatment program consisting of cognitive-behavioral therapy, nutritional orientation and counselling, psychoeducation, motivational therapy, social skills training, and therapy to modify a distorted perception of body image.
After 1 year of treatment and follow-up, the three questionnaires were again sent to the patients who had completed the baseline questionnaires. The same protocol was used to improve response rates as for the baseline questionnaires.
Each patient’s psychiatrist completed a clinical questionnaire at baseline and again after the patient had completed 1 year of treatment. It contained sociodemographic and clinical questions, as well as the Clinical Global Impression—Severity of Illness (CGI)  scale.
To create a comparison group, 305 women were recruited from the general population of the surrounding area (i.e., public universities, and centers of advanced education). They were matched with study patients by age categories (in decades). All participants were informed in advance about the characteristics of the research and its voluntary nature.
Patients completed three questionnaires at baseline and after 1 year of treatment: the HeRQoLED, the SF-12, and the EAT-26.
The Health-Related Quality of Life in Eating Disorders (HeRQoLED) [16, 17] questionnaire is the first version of an instrument for evaluating the impact of an ED on QOL. It is composed of 50 items distributed across 8 domains: symptoms, restrictive behaviors, body image, mental health, emotional role, physical role, personality traits, and social relations. Seven focus groups were formed (four with patients with ED, two with family members of patients and one with a multidisciplinary group of experts) to determinate which aspects of the patients’ lives were most affected by the disorder.
Based on an analysis of the contents of the focal groups, the domains most commonly affected by ED were identified. Also, an extensive bibliographic review was carried out to find articles and questionnaires dealing with the identified domains. Finally, members of the research group selected the items that could be used to construct the questions in the HeRQoLED from the reviewed questionnaires.
Responses to each item use an ordinal rating scale with a maximum of 6 options. Items’ response scores for each domain are directly summed up to obtain the score per domain. This score is standardized to a range of 0–100. The lower the score, the better the QOL.
To be applicable to the general population, the social relations domain of the HeRQoLED was slightly modified. Originally, the questions in the social relations domain refer to the impact of the ED in specific areas of social life. To make it applicable to a sample of individuals without ED, the term ‘eating disorders’ in the heading of the question was substituted by “your physical and mental health”. This change allowed the general population sample to respond to every question in the social relations domain. The remainder domains did not need any modification.
Prior studies of the HeRQoLED have shown satisfactory validity, with correlations > 0.40 with the criteria measures . The questionnaire was found to have excellent reliability, stability, and responsiveness—the mean changes detected by the HeRQoLED correlated above 0.30 with the criterion measures .
The Eating Attitudes Test (EAT-26) [20, 21], an abbreviated version of the EAT-40, was used as an indication of symptom severity. The EAT-26 encompasses three domains: (1) concern with dieting; (2) bulimia and food preoccupation; and (3) oral control. The Cronbach alpha for the three domains was 0.90, 0.84 and 0.83, respectively. The higher the score (> 20), the greater the symptomatology of the ED. The EAT-26 is highly correlated with the EAT-40 (r = .98) and has excellent test–retest reliability . The EAT-26 has been validated in the Spanish population  as a general measurement of ED pathology.
The Short Form Health Survey 12 (SF-12), an abbreviated version of the SF-36, is a generic survey of health status that can be summarized in two subscales: a physical component summary (PCS) and a mental component summary (MCS) . It includes eight constructs commonly represented in health surveys: physical functioning, role physical, bodily pain, general health, vitality, social functioning, role emotional and mental health.
Higher values indicate better health perception. The SF-12 is designed to be self-administered, reducing the burden of data collection for health care providers. Most patients can complete the SF-12 in less than 3 min without assistance.
Very high correlations between the SF-12 and SF-36 have been observed, ranging from 0.94 to 0.96 for the PCS and 0.94–0.97 for the MCS .
Patients’ psychiatrists completed the Clinical Global Impression—Severity of illness (CGI)  at baseline and after 1 year of treatment. The CGI-Severity of illness consists of one question (“How mentally ill is the patient at this time?”) with seven possible answers for quantitatively evaluating clinical status and changes in the severity of a patient’s mental health.
This scale requires the clinician to rate the severity of the patient’s illness at the time of assessment, relative to the clinician’s past experience with patients who have the same diagnosis.
Descriptive statistics of sociodemographic and clinical variables were evaluated using means and standard deviations (SD) for quantitative variables, and frequencies and percentages for nominal data. The paired Student’s t-test was used to compare mean scores between the HeRQoLED domains at baseline and after 1 year of treatment.
Univariate analysis was performed to determine which variables were predictive of change in each of the HeRQoLED domains after 1 year of treatment. The change for each domain was defined as the difference of the scores from baseline to 1 year later. General linear models were performed to establish variables for the multivariate analysis. Age, body mass index (BMI), dichotomized EAT-26, CGI score, SF-12 domains (PCS and MCS), and the baseline score in each domain of the HeRQoLED were considered as independent variables. R2 were obtained for all these models.
All statistical analyses were computed with the statistical program SAS System Inc. v9.1. Statistical significance was considered as P < 0.05.
A total of 435 patients fulfilled the inclusion criteria. Of these, 358 (82%) completed the baseline questionnaires: 61 with anorexia nervosa, 47 with bulimia nervosa, 245 with EDNOS, and 5 with no diagnosis recorded. Almost all (96.6%) of the sample were women, with a mean age of 26.8 years. All patients were literates. The mean BMI was 21.4. Of the 77 patients who did not complete the baseline surveys, 7 explicitly refused to participate in the investigation; the rest either did not suitably complete the questionnaires, the correspondence was returned by the postal service, or did not participate for miscellaneous reasons. Of the patients who completed the baseline questionnaires, 273 (76%) also completed the follow-up questionnaires. At that stage, returns from the postal service or inadequate completion of the questionnaires accounted for all of the nonresponders; no patients explicitly refused to participate at this stage.
Changes in the mean scores for the 8 domains on the Health-Related Quality of Life for Eating Disorders (HeRQoLED) questionnaire, the generic SF-12 quality of life questionnaire, and the EAT-26 questionnaire at baseline and after 1 year of treatment
At baseline n = 358
After 1 year of treatment n = 273
Comparison of the Health-Related Quality of Life for Eating Disorders (HeRQoLED) questionnaire domains at baseline and after 1 year of treatment by initial diagnostic category
After 1 year of treatment
Anorexia nervosa (n = 61)
Bulimia nervosa (n = 47)
EDNOSa (n = 245)
Among patients with bulimia nervosa, substantial improvements in HeRQoLED scores were apparent after 1 year of follow-up, although the changes were not statistically significant in the areas of restrictive behavior, body image, and physical role (P < 0.05). Patients with bulimia nervosa experienced a greater improvement in the symptoms (mean = −12.01; SD = 12.91; P < 0.001) and emotional role (mean = −10.63; SD = 26.66; P = 0.03) domains than patients with anorexia nervosa or EDNOS.
Overall, the clearest improvements in QOL measured by the HeRQoLED were observed among the EDNOS patients, with significant gains in all domains (P < 0.05). The magnitude of the gains, however, were lower than observed in bulimic patients in the symptoms, mental health, emotional role, and personality traits domains.
Comparison of the scores from the Health-Related Quality of Life for Eating Disorders (HeRQoLED) questionnaire according to eating disorder severity determined by the EAT-26 and the Clinical Global Impression (CGI), at baseline and after 1 year of treatment
At baseline (n = 358)
After 1 year of treatment (n = 273)
At baseline (n = 358)
After 1 year of treatment (n = 273)
≤20 Mean (SD)
>20 Mean (SD)
≤20 Mean (SD)
>20 Mean (SD)
Patients classified as “Not at all severe/mildly/moderately severe” improved in all domains of the HeRQoLED after 1 year. Nevertheless, patients who were markedly or severely ill had poorer QOL in some of the domains after 1 year (P < 0.001). As with the EAT-26, the differences were similar at baseline and after 1 year of treatment.
Adjusted predictors of Health—Related Quality of Life for Eating Disorders (HeRQoLED) domains after 1 year of treatment
Domain at baseline
Body Mass Index
Model R-square (R2) (%)
With respect to diagnostic groups, only patients diagnosed with anorexia nervosa showed differences in changes in the physical role and social relations domains (P < 0.05) when compared to the EDNOS group.
Differences between the ED and comparison group, both at baseline and after 1 year of treatment, were largest in the areas of restrictive behavior, body image, and personality traits. The lowest score for ED patients was observed in the restrictive behavior domain; the highest score in the mental health domain.
Using the disease-specific HeRQoLED questionnaire, we observed statistically significant improvements in QOL among patients with ED after 1 year of multidisciplinary treatment. In general, patients with bulimia nervosa experienced the largest gains in all domains except in restrictive behaviors, though these improvements were not always statistically significant. EDNOS patients experienced significant, though smaller, improvements in all domains. Patients with anorexia showed the least gains in QOL. As measured by the HeRQoLED, QOL was substantially worse for ED patients, even after 1 year of treatment, than for the general population. These findings support those of Keilen et al. , Spitzer et al. , Padierna et al. [3, 6, 24], Hay , González Pinto et al. , Doll et al. , and de la Rie et al.  who used generic QOL instruments to show that patients with ED suffer from a diminished QOL.
Physical role was the only HeRQoLED domain in which significant improvements were not routinely observed. Patients with anorexia nervosa showed non significant declines in QOL in this domain after 1 year of treatment. Patients with bulimia nervosa showed improvements in the physical role domain after 1 year, but the change was not statistically significant. In contrast, EDNOS patients demonstrated statistically significant improvements in the HeRQoLED physical role domain. Although patients with anorexia nervosa exhibited a deterioration in the HeRQoLED physical role domain, they demonstrated gains in the physical health component of the generic SF-12 (but not in the mental health component). This discrepancy between the HeRQoLED and the SF-12 may be due to the fact that the SF-12 does not contain a specific area for the physical role, but rather a physical component with a broader scope.
In addition, the physical component of the SF-12 associates better QOL with increased physical activity. Among patients with ED, this may be misleading, since the hyperactivity they generally present may be a sign of increasing severity of disease rather than an improvement. In addition, patients with anorexia nervosa may perceive a loss of weight as an improvement in QOL, rather than as the clinical problem it actually is. An alternative interpretation for these opposite trends is that the disease-specific HeRQoLED questionnaire was not sensitive to treatment-related improvements in physical health.
We observed the greatest improvement in HeRQoLED scores in the social relations and symptoms domains. These findings are comparable to those from other investigations. Using a generic QOL instrument, Spitzer et al.  observed the largest gains in physical pain (symptoms) and social role. Padierna et al.  using the generic SF-36, found that patients with ED showed the greatest improvement in social function and experienced a smaller decline in the physical component than we observed using the disease-specific HeRQoLED questionnaire.
QOL scores for patients with bulimia nervosa were in between those of patients with anorexia and EDNOS. In some domains, such as symptoms, mental health, emotional role, or personality traits, patients with bulimia showed significantly greater improvements than those observed for the other two diagnostic categories. In other domains such as physical role, social relations, body image, or restrictive behavior, the improvements were not statistically significant. However, the small sample size for this subgroup limits the interpretation of these findings.
Patients with EDNOS, who comprise the majority of our sample, did not exhibit the largest improvements in QOL but their improvements were statistically significant in all QOL domains, including physical role, after 1 year of treatment and follow-up.
In the present study, severity of the ED was rated using two different and independent parameters: the clinical severity according to the CGI, completed by the psychiatrists, and the EAT-26 questionnaire completed by the patients. The results of the study do indicate a different evolution in the QOL according to the type of ED, or according to the degree of clinical severity, but not with the relationships that both factors can have between themselves.
Nevertheless, the different ED has different severity and evolution. Generally, the EDNOS are, a priori, less serious than anorexia or bulimia nervosa, as an important percentage of them, are mitigated or attenuated forms of these last two conditions. EDNOS cover non-specific syndromes, they are conceptually problematic to define and they form a heterogeneous group of diagnoses (e.g., including binging disorder), making it difficult to give a prognostic judgement about such conditions and, therefore, being common a variable evolution .
To put our findings into context, we compared HeRQoLED scores from our ED patients with a sample drawn from the general population. QOL was substantially lower for the ED population, even after 1 year of treatment. The largest differences in QOL scores between ED patients and the general population were those closely related to the specific symptoms of these disorders, such as restrictive behavior, body image, and personality traits. These domains are most intimately related to the pathology of ED, and thus reflect fundamental differences between the two groups.
Analysis of possible predictors of change in QOL indicates that patients with worse QOL perception at baseline, as measured by the HeRQoLED questionnaire, have higher room for improvement following treatment than those with higher baseline QOL. In multivariate analysis, only the severity of ED, as measured by baseline EAT-26 score, remained as a predictor of change for HeRQoLED scores, having higher room for improvement those with less ED symptomatology; diagnosis groups, CGI, BMI, or age did not appear to influence these scores except for individual QOL domains. Prior studies on the differences in QOL and prognosis among patients with ED have yielded varying results [7, 10, 26–30].
We found several other predictors for QOL after 1 year of treatment. Higher BMI at baseline was significantly associated with poorer QOL in the HeRQoLED questionnaire’s physical role domain. Low scores on the EAT-26, which indicate low ED symptomatology, were associated with statistically significant improvements in QOL in several domains, especially personality traits and social relations. Patients who presented with less severe ED as measured by the CGI experienced greater improvements in QOL after 1 year of treatment than those with more severe disorders.
Patients with the most severe ED showed irregular progress following treatment, at least with regard to QOL, with slight deteriorations in QOL in some domains. The association between severity of illness and improvement in QOL following treatment makes sense when taking into account the fact that less severe pathologies tend to respond better to treatment and require less response time than more serious or complex disorders.
The possibility that generic instruments are open to misinterpretation is supported by the findings of Mond et al. , who observed that patients with restrictive anorexia nervosa presented better QOL scores than patients with all other types of ED; even in sub-domains such as social role their scores were similar to those of the general population. In contrast, the anorexia nervosa patients in our study did not demonstrate significant improvements in their QOL after 1 year of treatment, except in the symptoms and restrictive behavior domains. Although their sample was larger (n = 84) than ours, their study consisted of a cross-category analysis of QOL without any follow-up.
Doll et al.  also found that QOL related to emotional well-being was less affected in patients with anorexia nervosa than among those with other ED. Given that both perceived social role and support can influence adolescent women with regard to later development of ED , our work and others highlight the need for further investigation in this area.
The strengths of our study are the relatively large sample of patients with different ED diagnoses, the follow-up design, and the use of both disease-specific and generic instruments to assess patients’ QOL. When we developed the HeRQoLED questionnaire [16, 17] to measure QOL in patients with ED, no other ED-specific QOL instrument existed. Although three others have been developed since then [31–33], to the best of our knowledge this is the first study in which changes in QOL have been measured with a wide range of instruments among patients with ED.
Several limitations should be noted. Treatment and follow-up was limited to one year. It is possible that longer treatment and follow-up could yield greater improvements among the most severely affected patients.
The loss of patients during follow-up reduced the final sample size, and the number of patients diagnosed with anorexia or bulimia nervosa was small compared with the number of patients with EDNOS. These factors could have influenced the magnitude and the significance of the results. In the comparison of ED patients with the general population, we adapted questions from the social relations domain of the HeRQoLED for the general population. This could have biased the results in this domain.
The comparison group completed the HeRQoLED only once. Asking the participants in this sample to complete the questionnaire again after 1 year would have allowed for more direct comparison of changes in this group compared to the ED group.
Finally, our results cannot be generalized to the population of male individuals with ED since they are underrepresented in our sample. ED is more frequently observed in women. The calculations for the prevalence of women/men oscillated between 1/6 and 1/10 . In the psychiatrics units where the sample of the patients was taken from, the ratio was 1/15. To rely understand the impact of gender, we would have needed to create a sufficiently large sample of males, with the difficulties that that entails given the low prevalence described.
In conclusion, this is the first study to include a disease-specific measure to evaluate QOL in a large sample of ED patients before and after 1 year of treatment and follow-up. It showed the ability of HeRQoLED to asses QOL and track changes in it among ED patients, and provided more information and specific data on diverse aspects of these patients’ QOL, thus enabling a much more detailed analysis of the information. Future studies are needed to make direct comparisons between disease-specific and generic instruments to better delineate changes in QOL among patients with ED.
This work was funded by the Osakidetza Extra-Hospital Mental Health Program Contract 2007 and by the Instituto de Salud Carlos III (file no. 00/0115) grant entitled: “Creation and validation of a specific quality-of-life questionnaire for persons affected by eating disorders”.