Serological Testing for Coeliac Disease in Patients With Symptoms of Irritable Bowel Syndrome: A Cost-Effectiveness Analysis

S.M. Mein; U. Ladabaum


Aliment Pharmacol Ther. 2004;19(11) 

In This Article

Materials and Methods

We constructed a decision analytic model to estimate the clinical and economic consequences of two serological testing strategies for coeliac disease compared with usual care for IBS in a cohort of 1000 patients with 's'-IBS. We also evaluated up-front endoscopic biopsy, or adding biopsy in patients already undergoing oesophago-gastroduodenoscopy (EGD) for concomitant upper abdominal symptoms. The outcomes of interest were number of coeliac disease cases detected, QALYs gained, and costs. QALYs reflect length of life adjusted for the relative desirability (utility) of a health state, with perfect health assigned a utility of 1 and death a utility of 0. For example, 1 year in a state with a utility of 0.8 represents 0.8 QALYs.

It is possible that a diagnosis of coeliac disease followed by strict adherence to a gluten-free diet may decrease the risk of long-term complications,[9,21,22,23,24,25] and thus perhaps decrease direct medical expenditures. It is unknown whether a gluten-free diet prolongs life expectancy in coeliac disease.[13] We made the conservative assumptions that life expectancy was not improved by uncovering coeliac disease in patients with 's'-IBS, and that the benefit of diagnosing coeliac disease was confined to symptomatic improvement resulting in improved quality of life (higher health state utility compared with 's'-IBS). Because many patients with coeliac disease continue to require some medical care, we made the conservative assumption that the ongoing costs of caring for patients with IBS and coeliac disease were equivalent. These assumptions represent biases in the model against testing for coeliac disease in 's'-IBS.

MEDLINE searches through September 2003 were conducted with the terms 'irritable bowel syndrome', 'coeliac disease', and 'gluten-free diet', each combined with 'quality of life', 'utility', 'quality-adjusted life-year' (QALY)', and 'cost'. Separate searches were conducted for 'prevalence of coeliac disease in IBS' and 'prevalence of IBS in coeliac disease'. Additional searches were made for 'antibodies for coeliac disease', 'antigliadin antibodies', 'antiendomysial antibodies', 'tissue transglutaminase' each combined with 'sensitivity' and 'specificity'. All relevant publications were retrieved and pertinent data extracted. Costs were derived from published literature as well as other sources as noted below.

Serological tests for coeliac disease have high sensitivity and specificity. The sensitivity and specificity of tissue transglutaminase antibodies (TTG) range from 85 to 99%.[26,27,28,29,30,31] It is generally accepted that current TTG tests are over 90% sensitive and specific. Antigliadin IgG is less sensitive than antigliadin IgA, but IgA deficiency is more common in coeliac disease patients than in the general population.[32,33,34]. Antigliadin IgG and IgA combined have high sensitivity (89-100%) but lower specificity (70-95%) than TTG. We modelled serological testing with TTG antibody alone, and with a panel of antibodies (TTG, antigliadin IgG and IgA, and quantitative IgA to exclude IgA deficiency). We assumed a higher sensitivity, but a lower specificity, for the panel compared with TTG alone ( Table 1 ).

A decision analytic model was constructed (using Data-Pro, TreeAge Software, Inc., Williamstown, MA, USA) to focus on three strategies in patients with 's'-IBS: no testing for coeliac disease, testing with TTG antibody alone, and testing with the panel of antibodies described above (Figure 1). EGD with small intestinal biopsy followed any positive TTG or antigliadin IgA antibody test, and was also performed when only the antigliadin IgG antibody was positive in a patient who was IgA deficient. We also modelled up-front EGD with biopsy in all patients, and the addition of biopsy in patients already undergoing diagnostic EGD. In these endoscopic strategies, confirmatory serological testing with the antibody panel was performed if histology was abnormal, because villous blunting and lymphocytic infiltration are highly suggestive but not absolutely specific for coeliac disease. We accounted for potential complications, including death, with EGD and biopsy, and the associated costs. Small bowel biopsy was assumed to be a perfect diagnostic gold standard in the clinical setting of suspected coeliac disease and the presence of coeliac disease antibodies.

Decision analytic model of testing for coeliac disease in patients with symptoms of the irritable bowel syndrome. Positive serological tests (a) were followed by endoscopy and small bowel biopsy (b), and potential complications were accounted for. Up-front endoscopic biopsy was also evaluated.

Patients without coeliac disease and those with undiagnosed coeliac disease were assumed to experience the natural history of IBS symptoms and to incur IBS-related costs. Patients diagnosed with coeliac disease were assumed to be treated and have improvements in quality of life (see below), and to incur the same costs as IBS-related costs. Detection of coeliac disease cases occurred upon entry into the simulation (by testing for coeliac disease). Because the purpose of identifying coeliac disease in patients with 's'-IBS is to improve quality of life, we projected gains in QALYs and costs for the remainder of patients' lives. This was accomplished by calculating remaining life expectancy as a function of age from the USA Life Tables,[46] multiplying remaining life expectancy by the utility of the IBS state or by the utility of the treated coeliac disease state to calculate QALYs, and multiplying remaining life expectancy by the annual costs of IBS care to calculate disease-specific long-term costs. Both QALYs and costs were discounted at an annual rate of 3%.[47] In the base case, we evaluated 35-year-old patients (remaining life expectancy of 42.8 undiscounted years), and we examined the impact of age in sensitivity analyses.

In the base case, we assumed a coeliac disease prevalence in 's'-IBS of 3%.[10,11,12] This key parameter was examined closely in sensitivity analyses. The model inputs are listed in Table 1 .

The utilities of the IBS state and the treated coeliac disease state have not been measured directly. However, there is a body of literature on quality of life in both of these states, including studies that use the Short Form-36 (SF-36). A regression equation has been published that predicts the quality of well-being index, a measure of utility on a scale from 0 to 1, from SF-36 scores.[48,49] This equation and published SF-36 scores have been used previously to perform a cost-utility analysis of alosetron treatment in women with severe diarrhoea-predominant IBS.[41]

The utility of the IBS state for this analysis was calculated from published SF-36 data for IBS patients in the USA,[4,5] as described previously,[41] yielding an IBS state utility of 0.689. It is believed that quality of life with treated coeliac disease is better than with untreated coeliac disease.[18,19,20] In some populations, the quality of life of patients with treated coeliac disease may be comparable with the average in the general population,[15] but in others it may be worse.[35] The increase in utility achieved by diagnosing and treating coeliac disease in a patient with 's'-IBS was derived as follows.

The utility of the treated coeliac disease state was derived from a study of Swedish patients with coeliac disease who had been on a gluten-free diet for 10 years with variable adherence.[35] Using SF-36 data from this report, we calculated a utility of 0.717 for the coeliac disease patients compared with an average of 0.726 for the general population in Sweden, resulting in a utility decrement for treated coeliac disease of 0.009 compared with the general population. Notably, this estimate takes into account imperfect adherence to a gluten-free diet, as only 78% of patients reported strict adherence with a gluten-free diet, while 12% had a partial gluten-free diet and 10% reported eating essentially a normal diet. To compare the utility of treated coeliac disease against the utility of IBS in the United States, we subtracted the utility decrement of 0.009 experienced by the Swedish coeliac disease patients from the average utility of the general USA population (0.722, as described previously[41]) yielding a utility of 0.713 for treated coeliac disease in the USA. Thus, the increase in utility achieved by diagnosing coeliac disease in a patient with 's'-IBS in the USA was estimated as 0.713-0.689 = 0.024.

The length of hospitalization and the utility of the hospitalized state following a major endoscopy complication were estimated as described previously.[41]

Costs, not charges, were the basis for inputs in the model and were derived from published literature, our institution's costs, and 2000 Medicare data.[40,42] All costs were adjusted to reflect 2003 dollars by using the annual medical services component of the consumer price index. The base case cost for standard IBS care was taken from a recent study of IBS costs in a health maintenance organization ($450/patient-year after the index year, updated to year 2003 dollars[40]), as described previously.[41] There are no available data comparing costs of IBS and coeliac disease care. We assumed that medical costs for treated coeliac disease were not different from the costs of IBS care. If symptomatic improvement on a gluten-free diet actually leads to decreased resource utilization, then our assumption about the cost of chronic care would represent a bias against testing for coeliac disease. The potential additional cost of a gluten-free compared with a regular diet was not explicitly included in the analysis.

The cost of serological testing was estimated based on Medicare reimbursement[42] and laboratory costs to our institution. National Medicare reimbursement for TTG was $32-43 and for a serological panel $50-85 in the year 2000.[42] In California in the year 2003, Medicare reimbursement for TTG was $22 and for a panel $67-73. As a client for an outside laboratory, the University of California, San Francisco is charged $68 for TTG and $127 for the panel. The list prices for non-clients are $136 and $253. We used the laboratory client rate for the base case ( Table 1 ).

The risks of major complications or death with EGD and biopsy, and the associated costs, were estimated from published data and year 2003 national Medicare data.[36,37,38,39,42,43,44,45,50,51]

The number of coeliac disease cases uncovered in a population with 's'-IBS was a function of the sensitivity of the tests used. The average QALYs and costs per subject with a given strategy were determined by the test performance characteristics and costs of testing, the proportions of patients in the 's'-IBS state or the coeliac disease state and the associated costs of care, the number and costs of endoscopy-related complications and deaths, the utilities of the health states modelled, and the age and life expectancy of patients.

If one strategy detected more coeliac disease cases than another strategy at a higher cost, we calculated a cost per coeliac disease case detected (incremental cost-effectiveness). If one strategy yielded more QALYs per patient than another strategy but at a higher cost, we calculated the cost per QALY gained (incremental cost-utility).

Because the inputs used in the base case of any decision analysis are estimates that can vary, we examined the effect of changing the value of each variable in sensitivity analyses. In one-way sensitivity analyses, we assessed the impact of varying one input at a time on the incremental cost per coeliac disease case detected and the cost per QALY gained between the principal strategies. The variable ranges tested were based on published data ( Table 1 ). Two-way sensitivity analyses were conducted on critical variables. Finally, a probabilistic analysis (Monte Carlo simulation) was performed in which all model inputs were varied simultaneously for 5000 iterations (ranges in Table 1 ), with results reported as medians and interquartile ranges.[52]


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.