Delay to celiac disease diagnosis and its implications for health-related quality of life
© Norström et al; licensee BioMed Central Ltd. 2011
Received: 9 August 2011
Accepted: 7 November 2011
Published: 7 November 2011
Skip to main content
© Norström et al; licensee BioMed Central Ltd. 2011
Received: 9 August 2011
Accepted: 7 November 2011
Published: 7 November 2011
To determine how the delay in diagnosing celiac disease (CD) has developed during recent decades and how this affects the burden of disease in terms of health-related quality of life (HRQoL), and also to consider differences with respect to sex and age.
In collaboration with the Swedish Society for Coeliacs, a questionnaire was sent to 1,560 randomly selected members, divided in equal-sized age- and sex strata, and 1,031 (66%) responded. HRQoL was measured with the EQ-5D descriptive system and was then translated to quality-adjusted life year (QALY) scores. A general population survey was used as comparison.
The mean delay to diagnosis from the first symptoms was 9.7 years, and from the first doctor visit it was 5.8 years. The delay has been reduced over time for some age groups, but is still quite long. The mean QALY score during the year prior to initiated treatment was 0.66; it improved after diagnosis and treatment to 0.86, and was then better than that of a general population (0.79).
The delay from first symptoms to CD diagnosis is unacceptably long for many persons. Untreated CD results in poor HRQoL, which improves to the level of the general population if diagnosed and treated. By shortening the diagnostic delay it is possible to reduce this unnecessary burden of disease. Increased awareness of CD as a common health problem is needed, and active case finding should be intensified. Mass screening for CD might be an option in the future.
Celiac disease (CD) is a permanent intolerance to gluten found in wheat, rye and barley. Gluten induces an autoimmune reaction in the small intestinal mucosa resulting in inflammation, villous atrophy and malabsorption. The only effective treatment is a gluten-free diet, which usually leads to healing of the intestinal mucosa and recovery from signs and symptoms . Living with untreated CD is associated with a risk for extensive negative health consequences [2, 3]. In some individuals the gastrointestinal symptoms are obvious, but frequently the symptoms are vague, which complicates the identification of CD cases. This has been shown both by an average delay of 11-13 years from symptoms to diagnosis [4–6], and through screening studies. Screening studies have shown five to 10 undiagnosed cases for every diagnosed case in some Western European countries , four of five Swedish adults with CD are undiagnosed , and two of three Swedish children with CD are undiagnosed . In the latter study the prevalence was as high as 3%, indicating that CD is more common in some populations than the often mentioned 1%.
Health-related quality of life (HRQoL) as an aspect of living with CD has been studied frequently [4, 5, 10–17], although several questions still remain. The two generic utility-based instruments, Short Form 36 (SF-36) and EuroQol 5 Dimensions instrument (EQ-5D), are often used to measure HRQoL. SF-36 includes eight multi-item measures that represent different aspects of health . EQ-5D includes both a descriptive system, comprised of five health-related dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) divided into three levels of severity, and a visual analogue scale (VAS) for recording overall health . Both instruments can be translated to a quality-adjusted life year (QALY) score, which enables comparisons between different diagnoses and with the general population. EQ-5D translation algorithms were first developed for the United States  and the United Kingdom (UK) populations , with the latter commonly used for the Swedish population . For CD, SF-36 is still the most frequently used instrument [12, 14–17], but EQ-5D has become increasingly popular [5, 10].
Studies on treated CD adults have shown divergent results, with HRQoL either similar to [4, 5, 10] or worse than that of the general population [14, 17]. Females with treated CD have been shown to experience worse HRQoL than males with treated CD [5, 11–13]. One study using EQ-5D collected retrospective HRQoL data from members of a CD patient organization in the UK . They concluded that the HRQoL before CD diagnosis is quantitatively similar to that of stroke patients, and that after initiated treatment it improved and was as good as that of the general population.
Despite many studies, unanswered questions remain, due in part to the small size of those studies. Previous studies have also lacked detailed analyses of various aspects due to the complex pattern regarding diagnostic delay and factors such as sex, age and time period for diagnosis. The aim of this study was to determine how the delay in diagnosing CD has developed during recent decades and how this affects the burden of disease in terms of HRQoL, and also to consider differences with respect to sex and age.
During 2009 we performed a cross-sectional questionnaire survey, of Swedish adults with CD that was approved by the Regional Ethical Review Board at Umeå University. The questionnaire was in Swedish and is accompanied to this article with an English translation [additional file 1]. For comparison, a questionnaire survey of the general adult population, "Health On Equal Terms" [23, 24], was used. The latter was performed during 2006 and was approved by the ethical review board at the Swedish National Board of Health and Welfare.
We studied members of the Swedish Society for Coeliacs in 2009. The society is open to people with CD and other food intolerances and is the only one of its kind in the country. When joining the society, members self-report their food intolerance. At the time of the study, the society had 15,659 members reporting CD, 11,094 of whom were 20 years of age or older. Ludvigsson and colleagues defined 29,096 individuals diagnosed with CD in Sweden, 26,047 who were still alive, between 1969 and 2008, based on biopsy information from a computerized search of all regional pathology departments . Thus, the society most likely represents about 60% of all persons with CD in Sweden. The general adult population study consisted of the four most northern Swedish counties, with a population of 677,777 persons 20 years of age or older.
Characteristics of celiac disease subjects
Age when responding (years)
Duration to diagnosis (years)
First doctor visit
Age at diagnosis for the whole group
≥ 60 years
Age at diagnosis in relation to
period for diagnosis
Compliance with a gluten-free diet
Household member with celiac disease
Highest educational level
In the general population survey 37,912 randomly selected persons were invited in strata of age, sex, county and municipality. In total, 27,809 (73%) responded, 25,797 of whom were 20 years or older . Fifty-three percent (n = 13,781) were females and the mean age was 52 years.
Questionnaires were administered by the Swedish Society for Coeliacs (CD population) and by Statistics Sweden (general population), both distributed by the postal service to the persons' homes. To facilitate responding in both studies, a prepaid envelope was included and three reminders were sent out when needed. Questionnaire responses were scanned and thereafter checked for consistency.
The CD questionnaire included a section on delay to CD diagnosis with questions about the first appearance of symptoms possibly related to CD, the time of the first visit to a physician for those symptoms, and time of diagnosis. Information that could verify the CD diagnosis and self-reported degree of compliance with a gluten-free diet was also collected. We used the EQ-5D instrument to measure HRQoL. The respondent was asked about the year prior to initiated treatment for CD, referred to as pre-treatment, and about today, the time of responding to the questionnaire. The EQ-5D descriptive system was answered completely by 779 (76%) respondents and EQ VAS by 914 (89%) respondents.
In the general population, the EQ-5D descriptive system was completely responded to by 24,460 persons (95%) aged 20 years or more. EQ-VAS was not included in the questionnaire.
Descriptive statistics were presented using frequency tables, cross-tabulations, and mean and median values. Means were compared with Student's t-test. Delay to diagnosis was estimated as the difference between the year of the first symptoms indicative of CD and the reported year of diagnosis. For a response of no symptoms before diagnosis, the delay was defined as 0 years. The delay from the first doctor visit was estimated in a similar way. Dependency between the time period for CD diagnosis and the delay to diagnosis was analyzed with the Cox proportional hazards model . An exponential of the hazard ratio above 1 implies a shorter delay from first symptoms to diagnosis compared to the baseline, which was a diagnosis before 1980. The descriptive system was translated to a QALY score using UK weights . Linear regression was used to study determinants for the QALY score pre-treatment and today. These analyses included sex, current age, delay from first symptoms indicative of CD to diagnosis (cutoff set to 2 years), and time period for diagnosis. Statistical significance was defined at the 5% level. Microsoft Access was used for data handling, while Stata 11.2 (StataCorp LP, College Station, TX) was used for statistical analysis.
Of the 1,031 respondents, 52% were females. Mean and median ages at diagnosis were 39 and 42 years, respectively. Strict compliance with a gluten-free diet was reported by 979 (96%), and 117 (11%) reported a household member with CD (Table 1).
The mean delay from the first symptoms indicative of CD to diagnosis was 9.7 years and the median delay was 4 years (quartiles 1-14 years). From the first visit to a doctor due to CD-related symptoms to diagnosis, the mean delay was 5.8 years and the median delay was 1 year (quartiles 0-8 years) (Table 1). Both males and females had a mean of at least 9 years from the first symptoms to diagnosis for each 5-year age group from 35-39 up to 65-69 years (data not shown). Excluding those below 20 years, no age group had a shorter mean delay than 6 years.
Relationship between time period for diagnosis and delay from first celiac disease symptoms to diagnosis
Age 20-39 years
Age 40-59 years
(n = 142)
(n = 148)
(n = 105)
Females (n = 150)
Diagnosis year a
EQ-5D descriptive system
(n = 779a)
(n = 24,460a)
Quality-adjusted life year (QALY) scores and EQ-VAS
Factors related to health-related quality of life (HRQoL) pre-treatment for celiac disease (CD) and today
Current age (years)
Long delay c
Diagnosed 1990-1999 d
CD patients experienced poor HRQoL pre-treatment, which improved significantly after diagnosis and treatment. Many experienced a long delay to diagnosis, both from the first symptoms and from the first visit to a doctor. It is promising that patients diagnosed after 2004 reported better HRQoL than those diagnosed earlier and that the delay in diagnosis has decreased during recent decades; however, many still experience too long a delay. There has been a pronounced increase in median age at diagnosis, from 2 to 46 years in recent decades, illustrating that CD has changed from presenting as a childhood disease to being a disease affecting all ages.
This is a large study with a high response rate. A strength is that we used a validated HRQoL instrument and compared HRQoL for the CD population with that of a general population. The approach of asking CD respondents about HRQoL both before and after initiated CD treatment has only been tried once before . Our study is unique in that results for males and females are presented separately, and there is a more thorough analysis of factors that affect HRQoL. Further, we add results concerning how diagnostic delay has changed over time, also taking age and sex into consideration.
We believe that our findings are of great value for an increased understanding of how CD and its diagnostics affect people, despite some potential biases. We cannot be sure that our results are representative for the whole Swedish CD population. However, even if the results are only valid for responders (66% of invited) to the questionnaire within our study population (members of the Swedish Society for Coeliacs which correspond to about 60% of Swedish adults with CD), they nevertheless show an experienced burden for a sample representing about 40% of Swedish adults with CD, irrespective of age. The general population in our study is limited to northern Sweden while the CD population comes from the entire country. However, regarding HRQoL, there was no statistical difference between persons with CD from the northern part of the country and the rest of Sweden. A recall bias might appear as some responders had their CD diagnosis long ago. This might cause both a low estimate of HRQoL pre-treatment, as well as an overestimate of the delay from first symptoms indicative of CD to CD diagnosis. However, the usual recall pattern is to report fewer episodes of ill health, and therefore previous health problems might instead be underestimated.
A long diagnostic delay has been shown earlier in several studies [4–6]. However, awareness of CD has increased in recent decades and serological testing has been introduced and improved. Similar to findings in our study, a shortened diagnostic delay over time was found in the UK , but was not observed in Canada . Neither study considered increasing age at diagnosis, a factor that might affect their results as well.
The positive effects of diagnosis and treatment we report here are also supported by others [5, 10]. Our results confirm that CD females experience a lower HRQoL than males . The difference is even larger before initiated treatment, which indicates that living with untreated CD might be a greater burden for females. It is also interesting that treated CD patients reported a better HRQoL than the general population. This might be a result of an altered frame of reference, after experiencing poorer health, or due to adapting to a healthier life style after being diagnosed with a chronic disorder. Our study subjects reported an unexpectedly high rate of compliance with a gluten-free diet (96%), and it has been shown that better compliance is related to a better HRQoL [11, 17, 27].
Despite increased awareness in society and in health care, many CD cases will be missed due to vague symptoms. Mass screening for CD has been raised as a possible option . CD mass screening fulfils most of the listed criteria for a medical mass screening adapted by WHO from Wilson and Jungner . However, knowledge concerning the natural history of CD and the cost-effectiveness of a CD mass screening is lacking [28, 30, 31]. There are few health economic evaluation studies of a CD mass screening. Existing studies indicate that a screening might be cost-effective [32, 33], and that parents' willingness to pay for a CD mass screening on average is greater than the screening cost . It was recently estimated in the United States that the medical cost for clinically detected CD patients is reduced by $1764 the year following diagnosis as compared to the average cost during the preceding years . The fact that both cost savings and long-term health complications might differ between clinically diagnosed patients and screening detected CD must be taken into account, and there is a need for studies on this particular group [36, 37].
Our study shows that for many individuals there is an unacceptably long delay from the first symptoms to CD diagnosis. Untreated CD results in poor HRQoL that is improved to the level of the general population if diagnosed and treated. By shortening the diagnostic delay it is possible to reduce this unnecessary burden of disease. Increased awareness of CD as a common health problem is needed, and active case finding should be intensified. Mass screening for CD might be an option in the future.
We thank all those who responded to the questionnaire, and the Swedish Society for Coeliacs for their assistance in improving the questionnaire and their crucial administrative help. In relation to the construction of the questionnaire we also thank Professor Alastair Gray and his research team at the University of Oxford. The study was undertaken at the Umeå Centre for Global Health Research at Umeå University.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.