Primary care physicians are often the first point of contact for individuals struggling with various health concerns. However, eating disorders frequently go undiagnosed in this setting. This oversight is concerning because delayed diagnosis, particularly in conditions like bulimia nervosa, can lead to poorer treatment outcomes. Patients might present with seemingly unrelated physical or psychological complaints, masking the underlying eating disorder. To address this issue, effective and easy-to-use eating disorder screening tools are essential for primary care settings.
Existing screening instruments, while developed to improve detection, often fall short in primary care. Many are lengthy and complex, making them impractical for busy clinical environments. Some potentially useful screening questions have emerged from studies in specific populations like university students, but their broader applicability and validation in primary care remained uncertain.
This study aimed to evaluate two screening tools in primary care and university student populations:
- The SCOFF questionnaire, a previously established clinical prediction guide for eating disorders.
- A new set of five questions, termed the Eating disorder Screen for Primary care (ESP), derived from prior research.
The goal was to determine their effectiveness in identifying eating disorders and to compare their performance characteristics in these settings.
Methods
The study received ethical approval, and all participants provided informed consent. Participants were recruited from two groups: university students and patients attending a primary care clinic. Students were recruited through campus-wide announcements and posters, similar to health campaign promotions. Primary care patients were approached consecutively in the clinic waiting area. Exclusion criteria included age outside the 18-65 range, inability to read English, or pre-existing chronic conditions that could significantly affect body weight.
All participants first completed the Questionnaire for Eating Disorder Diagnoses (Q-EDD). This served as the reference standard for diagnosing eating disorders, as it is a validated self-report tool aligned with the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) criteria. The Q-EDD has demonstrated high accuracy compared to DSM-IV interviews.
Following the Q-EDD, a psychiatrist, unaware of the Q-EDD results, administered the ESP and then the SCOFF questionnaire to each participant. The questions for each tool were as follows:
Eating disorder Screen for Primary care (ESP)
- Are you satisfied with your eating patterns? (A “no” answer was considered abnormal).
- Do you ever eat in secret? (A “yes” answer was considered abnormal).
- Does your weight affect the way you feel about yourself? (A “yes” answer was considered abnormal).
- Have any members of your family suffered with an eating disorder? (A “yes” answer was considered abnormal).
- Do you currently suffer with or have you ever suffered in the past with an eating disorder? (A “yes” answer was considered abnormal).
SCOFF Questionnaire
- Do you make yourself Sick because you feel uncomfortably full?
- Do you worry you have lost Control over how much you eat?
- Have you recently lost more than One stone (14 lbs or 7.7 kg) in a three-month period?
- Do you believe yourself to be Fat when others say you are thin?
- Would you say that Food dominates your life?
For SCOFF, a “yes” to any question was considered an abnormal response. The results from ESP, SCOFF, and Q-EDD were then analyzed independently by two clinicians blinded to the other assessments. Statistical analysis was performed to compare the performance of each tool against the Q-EDD standard, using sensitivity, specificity, and likelihood ratios.
Results
The study included 129 university students and 104 primary care patients. After exclusions, the final analysis comprised 225 participants. Demographic data and the distribution of eating disorders, as identified by the Q-EDD, are summarized in Table 1. The overall prevalence of eating disorders in the combined population was 12%.
Table 1. Demographics and the Distribution of Eating Disorders in the 2 Populations Studied
| Combined (N = 233) | Student (N = 129) | Primary Care (N = 104) |
|---|---|---|
| Number excluded | 8 | 0 |
| Number analyzed | 225 | 129 |
| Mean age, y (range) | 29 (18 to 64) | 22 (18 to 44) |
| Female, % | 77 | 77 |
| Mean BMI (range) | 22 (16 to 45) | 22 (16 to 34) |
| With an eating disorder, % (95% CI) | 12 (7.8 to 16) | 12 (6.7 to 18) |
| Anorexia nervosa, n (%) | 6 (22) | 4 (25) |
| Bulimia, n (%) | 11 (41) | 7 (46) |
| Binge eating disorder, n (%) | 9 (33) | 5 (31) |
| Non-binging bulimia, n (%) | 1 (4) | 0 (0) |
Results from both groups were combined due to similar findings. Individual ESP and SCOFF questions were analyzed for their ability to diagnose or exclude eating disorders. The most effective individual questions for identifying an eating disorder were: “Do you worry that you have lost control over how much you eat?”, “Do you make yourself sick when you feel uncomfortably full?”, “Do you currently suffer with or have you ever suffered in the past with an eating disorder?”, and “Do you ever eat in secret?”. Conversely, “Does your weight affect the way you feel about yourself?” and “Are you satisfied with your eating patterns?” were most effective at ruling out an eating disorder. The family history question did not significantly improve ESP’s screening accuracy and was excluded from further analysis.
A multilevel analysis, summarized in Table 2, evaluated the probability of an eating disorder based on the number of abnormal responses to ESP and SCOFF.
Table 2. Multilevel Analysis: The Probability of an Eating Disorder Based on the Number of Abnormal Responses for ESP and SCOFF
| Number of Abnormal Responses | Eating Disorder | No Eating Disorder | LR (95% CI) | Post-test Probability, % |
|---|---|---|---|---|
| ESP: 3 or 4 | 22 | 15 | 11 (6.4 to 18) | 59 |
| 2 | 5 | 43 | 0.85 (0.38 to 2.0) | 10 |
| 0 or 1 | 0 | 140 | 0.0 (0.0 to 0.15) | 0 |
| Total | 27 | 198 | ||
| SCOFF: 4 or 5 | 3 | 2 | 11 (1.9 to 62) | 60 |
| 2 or 3 | 18 | 21 | 6.2 (3.8 to 10) | 46 |
| 0 or 1 | 6 | 173 | 0.25 (0.12 to 0.51) | 3 |
| Total | 27 | 196 |
ESP and SCOFF Performance Comparison
Single Cutoff Analysis
Using a cutoff of 2 or more abnormal responses for SCOFF, the sensitivity was 78% and specificity was 88%. The positive likelihood ratio was 6.6, and the negative likelihood ratio was 0.25.
For ESP, with a cutoff of 2 or more abnormal responses, the sensitivity reached 100%, and specificity was 71%. The positive likelihood ratio was 3.4, and the negative likelihood ratio was 0.0.
Multilevel Analysis
Multilevel analysis revealed that ESP and SCOFF performed similarly in ruling in an eating disorder when using higher cutoffs (3+ for ESP, 4+ for SCOFF). However, ESP was significantly better at ruling out an eating disorder. Zero or one abnormal response on the ESP had a likelihood ratio of 0.0, compared to 0.25 for SCOFF, indicating ESP’s superior ability to identify individuals unlikely to have an eating disorder.
Discussion
This study provides valuable insights into the effectiveness of short eating disorder screening tools in primary care settings. The findings suggest that SCOFF, in this primary care and university student population, was less sensitive than previous studies indicated. The original SCOFF validation study reported 100% sensitivity, whereas this study found 78% sensitivity at a cutoff of 2 or more positive responses. This difference may be due to the derivation study’s comparison of very distinct groups and exclusion of complex cases, potentially leading to an overestimation of SCOFF’s performance.
In contrast, the ESP demonstrated 100% sensitivity at a cutoff of 2 or more abnormal responses, although with a lower specificity (71%) compared to SCOFF (88%). While SCOFF showed higher specificity, ESP’s perfect sensitivity at this cutoff is particularly noteworthy in a screening context where minimizing false negatives is crucial. Missing a case of an eating disorder can have serious health consequences, making a highly sensitive tool valuable for initial screening in primary care.
Limitations of the study include the non-consecutive recruitment of university students, although efforts were made to simulate a health promotion campaign to reach at-risk students. The prevalence of eating disorders in both study populations was also higher than some population-based estimates, possibly due to the inclusion of ‘eating disorders not otherwise specified’ and the sensitivity of the Q-EDD reference standard. However, this higher prevalence does not invalidate the comparative findings of ESP and SCOFF. The study’s sample size also limited the ability to analyze individual question levels in detail.
Despite these limitations, the study highlights the potential of ESP as a practical screening tool for eating disorders in primary care and university student settings. While not significantly different from SCOFF in ruling in eating disorders, ESP excels at ruling them out. A score of one or fewer abnormal responses on the ESP strongly suggests the absence of an eating disorder, making it a useful tool for initial screening and guiding decisions about further, more comprehensive assessments.
In conclusion, the Eating disorder Screen for Primary care (ESP) offers a brief, easily administered, and highly sensitive method for eating disorder screening in primary care. Its ability to effectively rule out eating disorders with a low number of abnormal responses makes it a valuable asset for busy primary care settings.
Acknowledgments
We thank the staff at Highgate Group Practice and NoCTeN for their support.
Funding for this work was provided by University College, London.
REFERENCES
[References from the original article would be listed here]
