Objective Patients with dyspepsia with alarm features are suspected of having upper gastrointestinal (GI) malignancy; however, the true value of alarm features in predicting an underlying malignancy for patients with dyspepsia with high background prevalence of Helicobacter pylori infection and upper GI malignancy is uncertain. The aim of the present study was to determine the diagnostic accuracy of alarm features in predicting upper GI malignancy by reviewing an endoscopic database consisting of >100 000 Chinese patients.
Methods A retrospective analysis of prospectively collected data was conducted in a single tertiary medical centre. Consecutive patients who underwent oesophagogastroduodenoscopy (OGD) for dyspepsia in 1996–2006 were enrolled. The data including gender, age, symptoms, and endoscopic and pathological findings were analysed. The main outcome measure was the diagnostic accuracy of individual alarm feature.
Results 102 665 patients were included in the final analysis. Among all the 4362 patients with malignancy, 52% (2258/4362) had alarm features. Among 15 235 patients who had alarm features, 2258 (14.8%) were found to have upper GI malignancy. The pooled sensitivity and specificity of the alarm features were 13.4% and 96.6%, respectively. Only the feature of dysphagia in patients between 36 and 74 years old had a positive likelihood ratio (PLR) >10 for malignancy prediction, while all other alarm features in other age groups had a PLR <10.
Conclusions For uninvestigated Chinese patients with dyspepsia with high background prevalence of H pylori infection and upper GI malignancy, alarm features and age, except for dysphagia in patients between 36 and 74 years old, had limited predictive value for a potential malignancy; therefore, prompt endoscopy may be recommended for these patients. However, less invasive, inexpensive screening methods with high diagnostic yield are still needed to reduce unnecessary endoscopy workload.
- Alarm features
- gastrointestinal tract
- upper gastrointestinal endoscopy
- upper gastrointestinal malignancy
Statistics from Altmetric.com
- Alarm features
- gastrointestinal tract
- upper gastrointestinal endoscopy
- upper gastrointestinal malignancy
Dyspepsia is a very common problem; ∼25–40% of adults in the general population have dyspeptic symptoms,1 and dyspepsia accounts for 2–5% of all consultations in primary care,2 thus many primary care physicians and gastroenterologists manage patients with dyspepsia on a daily basis. Although dyspepsia is so common that many guidelines have been published in recent years,1 3–8 much controversy still exists as to the most appropriate empirical strategy, for example basically all guidelines suggested that alarm features at any age warranted prompt endoscopy, and most recommended an age cut-off of between 50 and 55 years for endoscopy as an initial management strategy, while for young patients without alarm features, either ‘Helicobacter pylori test and treat’ or empirical proton pump inhibitors were the initial management strategies of choice.1 3–8
However, studies from areas with a high background prevalence of gastric cancer demonstrated that 0.4–5% of gastric cancers would have been missed if endoscopy had not been offered9 10; another study from Shanghai also suggested both H pylori ‘test and treat’ and ‘test and endoscopy’ strategies are not suitable for the management of patients with uninvestigated dyspepsia.11 Nonetheless, because dyspepsia is so prevalent in the general population, it is impossible to recommend prompt endoscopy for all patients with dyspepsia; therefore, efforts have been made to minimise unnecessary endoscopy by restricting its use to those above a certain age, generally 45–55 years old, or to those who present with alarm features. Despite these efforts, several studies including a comprehensive systematic review involving a total of 57 363 patients concluded that alarm features have limited predictive value for an underlying malignancy.12–17 However, most of these studies originated from the USA or European countries where the incidence of upper gastrointestinal (GI) malignancy is low, and there was considerable heterogeneity between these studies which had greatly varied sensitivity and specificity12; in addition, the relatively small number of detected malignancies hampered further statistical analysis and limited the generalisability of the conclusions to other populations. To evaluate the diagnostic yield of alarm features in Chinese patients with dyspepsia with a high background prevalence of H pylori infection and upper GI malignancy18 19 and to provide evidence for the management of dyspepsia in countries or areas with similar H pylori infection and upper GI malignancy prevalence, we conducted an endoscopic database review of >100 000 patients over a 10 year period.
The study was conducted in the Digestive Endoscopy Center of Changhai Hospital, which is a university tertiary medical centre. The Digestive Endoscopy Center is an open-access endoscopic unit and all the patients were referred by a doctor in the clinics of Changhai Hospital. The patient population consisted entirely of outpatients of Changhai Hospital. When referring patients for endoscopy, the doctor completed a standard and structured tick-box questionnaire on GI symptoms, and the predominant indication was determined. All the referring doctors have quite similar interpretation of upper GI symptoms. All the patients were told to avoid acid suppression treatment for as long as possible before upper GI endoscopy. The indications for and the purposes of endoscopy were then recorded in the structured questionnaire which was available during endoscopic examination. The waiting time for referral was within 1 week. The oesophagogastroduodenoscopies (OGDs) were performed by experienced endoscopists (n=30) at Changhai Hospital and every participating endoscopist had at least 2 years experience in upper GI endoscopy using forward or side-viewing endoscopes (GIF 200/230/240/260, Olympus Optical, Shanghai, China). All the patients received only topical pharyngeal anaesthesia during the study period. All of the parts of the upper GI tract (oesophagus, stomach and duodenum) were carefully examined for any GI lesion which may be the cause of dyspepsia, including oesophagitis, peptic ulcer (gastric and/or duodenal ulcer) and malignancy. Multiple biopsy specimens were taken in any area where the endoscopists suspected the possibility of malignancy. Because it was not routine practice for our endoscopic centre to perform biopsy for H pylori infection detection, only in a minority of patients who required the endoscopists to perform biopsy for H pylori infection detection was this procedure done. For patients who were suspected to have upper GI malignancy on endoscopy, the final diagnosis was confirmed by histological examination from biopsy specimens. All specimens were evaluated by experienced pathologists of the Department of Pathology of Changhai Hospital.
The indication for OGDs, the patient's age, gender, images of endoscopic examinations, and endoscopic and pathological findings were all recorded in an endoscopic database (EIS (Endoscopy Information System), Angelwin, Beijing, China). Electronic records were available for all endoscopic reports from June 1996 to June 2006. The population for the study consisted of all consecutive patients undergoing first-time diagnostic OGD during the study period; however, patients who had been suspected to have upper GI malignancy by barium study, abdominal CT, signs of metastasis, by an indefinite biopsy result in other centres, with previous endoscopy or previous gastric surgery or with a history of upper GI malignancy, or, in the surveillance programme for peptic ulcer, Barrett's oesophagus were excluded. Written informed consent for OGD was obtained from all patients before the procedure.
The database was searched systematically to identify all patients who underwent first diagnostic OGD for the evaluation of dyspeptic symptoms. Oesophagitis was defined as mucosal breaks or erosions within the oesophagus. The Los Angeles classification has been used for oesophagitis grading since the latter half of 1999. An ulcer was endoscopically defined as a mucosal break ≥5 mm in diameter; pathologically, a peptic ulcer was defined as a defect in the gastric or duodenal wall that extends through the muscularis mucosae (the lowermost limit of the mucosa) into the deeper layers of the wall (submucosa or the muscularis propria).20 The definition of dyspepsia refers to any condition or disease in which there are upper abdominal symptoms including upper abdominal pain or discomfort, loss of appetite, heartburn, regurgitation, bloating, early satiety or belching. The investigation of alarm features in this study only considered the following four alarm features: dysphagia, weight loss, GI bleeding and persistent vomiting. Upper abdominal pain was defined as pain perceived to be in the area of the upper abdomen. Upper abdominal discomfort was defined as a subjective negative feeling in the area of the upper abdomen which did not reach a level of pain. Loss of appetite was defined as the decreased desire for food and the accompanying pleasures of eating. Abdominal bloating was defined as a condition in which the abdomen feels uncomfortably full and tight. Belching was defined as voiding of gas from the stomach. Early satiety was defined as the premature sensation of postprandial abdominal fullness. Regurgitation was defined as the sensation of effortless return of gastric and/or oesophageal fluid into the mouth or throat. Heartburn was defined as a burning and ascending retrosternal sensation without any difficulty in swallowing.21 Dysphagia was defined as the perception of an impediment to the normal passage of swallowed material.22 Weight loss was defined as ≥5% of body weight in the past month. GI bleeding was defined as any evidence of haematemesis and/or melaena; haematemesis was defined as vomiting of blood or coffee ground-like material, and melaena was defined as passage of black, tarry and foul-smelling stools due to the presence of altered blood. Upper GI malignancy was defined as the presence of a malignancy in the oesophageal, gastric or duodenal region as observed by the endoscopist and finally confirmed by pathological examination.
The numbers of patients and occurrence of major endoscopic findings, including upper GI malignancy, peptic ulcer and oesophagitis, according to different age groups were calculated. The age group breakdown was chosen as <35, 36–54, 55–74 and ≥75 years old, because according to a review of guidelines for dyspepsia management, most of the current American and European guidelines set the cut-off age threshold for endoscopy at 55 years old5 and, in addition, Asian guidelines recommended endoscopy for those aged 35–55 depending on the risk of gastric cancer in the region.7 Furthermore, another recent study classified ages ≤35 years as very young patients because 35 years was the threshold age that separated those patients with unusually advanced gastric cancer from those with less advanced disease.23 For these reasons, the age group breakdown was chosen as 20 year intervals. Statistical analysis of the data was performed with SPSS 10.0 for Windows (Statistical Product and Service Solutions). Categorical data were compared by χ2 test with continuity correction if appropriate. Continuous variables are expressed as mean±SD and ranges, and were compared with the Student t test. The diagnostic values of individual alarm feature, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (PLR), negative likelihood ratio (NLR) and their 95% CI were calculated. The overall sensitivity and specificity of alarm features for malignancy prediction were pooled by Meta-DiSc.24 Two-tailed p values <0.05 were considered statistically significant.
There were a total of 110 048 consecutive patients referred for endoscopy from June 1996 to June 2006, and the process of identifying relevant patients is shown in figure 1. Among the 110 048 potentially relevant patients, 7265 were excluded for various reasons. Another 118 patients were excluded because of incomplete information. Finally, 102 665 patients were included. The demographic characteristics of these patients are described in table 1. Generally, the mean age of these patients was 47.6 years (range 10–95), and male patients accounted for 53% of all the patient population. There were 15 235 (14.8%) patients who had alarm features. The indications for endoscopy for the patients with alarm features were as follows: abdominal pain or discomfort (n=10 999, 72.2%); abdominal bloating (n=2377, 15.6%); regurgitation or heartburn (n=1204, 7.9%); belching (n=274, 1.8%); loss of appetite (n=198, 1.3%); and early satiety (n=183, 1.2%). Altogether, 4362 patients were diagnosed as having upper GI malignancy pathologically, including 2871 gastric, 1438 oesophageal, 33 duodenal and 20 synchronous malignancies. Excluding the 15 235 patients who presented with alarm features, the predominant indications for endoscopy for the remaining 87 430 patients were as follows: abdominal pain or discomfort (n=43 668, 50.0%); abdominal bloating (n=25 045, 28.6%); regurgitation or heartburn (n=14 227, 16.3%); belching (n=2189, 2.5%); loss of appetite (n=1 217, 1.4%); and early satiety (n=1084, 1.2%).
Overall, 4362 (4.2%) patients had upper GI malignancies; the prevalence of malignancies increased from 0.6% in patients <35 years old to 18.0% in patients >75 years old, and 29.8% (1302/4362) of patients with malignancy were younger than 55 years old. Similarly, the prevalence of oesophagitis also increased from 2.5% in patients <35 years old to 10.0% in patients >75 years old. The prevalence of peptic ulcer was relatively stable in all age groups, ranging from 18.3% to 21.4% (table 2).
Diagnostic accuracy of individual alarm feature
Among tholse 15 235 patients who had alarm features, 2258 (14.8%) were found to have upper GI malignancy (table 3). Among the four alarm features, dysphagia had the highest sensitivity (29.4%) and PPV (37.8%), while persistent vomiting had the lowest sensitivity (5.7%) and PPV (5.8%). All the four alarm features had very high specificity (ranging from 95.3% to 97.8%) and NPV (ranging from 95.8% to 96.9%). The pooled sensitivity of the alarm features for malignancy prediction was 13.4%, and the pooled specificity was 96.6%. By calculating the PLR and NLR of these alarm features separately for those in the different age groups, it is noted that only the feature of dysphagia in two age groups (36–54 and 55–74 years old) had a PLR >10, while all other alarm features in other age groups had a PLR <10 (table 4).
The present study suggests that except for dysphagia in patients aged between 36 and 74 years old, alarm features and age are not strongly suggestive of potential upper GI malignancy in Chinese patients with uninvestigated dyspepsia. Among all the 4362 patients who had upper GI malignancy, 48% had no any alarm features at presentation. In addition, more than half of the patients with malignancy who were >55 years old had no alarm features. These findings imply that the age for endoscopy as an initial management strategy may be much lower in Chinese patients with high background prevalence of H pylori infection and upper GI malignancy.
The advantages of our study are first, despite the high prevalence of dyspepsia in the general population, relatively few reports on the predictive value of alarm features have been published and little information on the predictive value of individual features was known. Moreover, most of the studies originated from Western populations; except for a study from Singapore and a study from Chinese Taiwan,9 25 almost no large-scale study had been conducted in countries where the prevalence of upper GI malignancies is high. In view of this, to the best of our knowledge, the present study is the largest study of its kind ever conducted on this topic. Secondly, because this study was conducted in a tertiary endoscopic unit, the possibility of clinical heterogeneity is minimised. Thirdly, due to the large sample size of the present study, the number of malignancies detected was quite large (4362 patients); therefore, we had enough statistical power to detect some important difference among various age groups which otherwise may be missed because of a small number of patients. Finally, though similar studies has been reported from Chinese Hong Kong and Chinese Taiwan9 10 where the prevalence of upper GI cancer is also high, none of these studies documented the details of diagnostic accuracy of the individual alarm features, and Liou et al considered alarm features as a whole but did not focus on individual alarm features9; therefore, the present study may be the largest study evaluating the diagnostic accuracy of individual alarm feature in patients of various age groups who had a high background prevalence of H pylori infection and upper GI cancer.
However, the current study had several limitations. (1) The major limitation is the observational design; nevertheless, the large sample size of the present study may compensate for this limitation. (2) The study was conducted in a university tertiary centre; therefore, selection bias may have been introduced. However, we excluded all patients with previous gastric surgery, or with history of upper GI malignancy and patients who had been highly suspected to have upper GI malignancy and thus the selection bias may have been reduced to some degree. (3) The H pylori infection status in most of the present patient population was not available for this study. The reason why routine H pylori testing was not done was because patients had to pay for the costs of such testing and hence most of them refused. During the 10 year period investigated, only 2462 patients (2.4% of the whole population) had their H pylori status recorded, and we were not able to investigate the effectiveness of the H pylori ‘test and treat’ strategy. However, because the prevalence of H pylori infection in Shanghai was as high as 66.4% in the general population,18 we suppose that if the C-13-urea breath test had been used, the majority (∼60%) of the included patients would have been diagnosed as H pylori positive, and it is suggested that in areas with a high prevalence of H pylori, the ‘test and treat’ strategy was unlikely to be beneficial.7 11 (4) There were 118 patients who were excluded from the final analysis due to incomplete information, comprising 75 patients with unclear indications for OGDs and the other 43 patients without documented information of age. However, none of these patients had upper GI malignancy (five patients with oesophagitis, 18 peptic ulcers, 95 normal), and the exclusion of a very small percentage of patients will not change the main conclusions of the present study. (5) The duration of dyspepsia symptoms was not well documented in the study. (6) The present study only included patients who were referred for endoscopy. Therefore, young patients who lack the alarm features may be less likely to undergo endoscopy. This may decrease both the specificity and NPV a little; however, because most of the patients were free of malignancy and the number of true negative patients was quite large, the impact on both specificity and NPV was not substantial. (7) The final limitation is that not all possible alarm features were evaluated but only those with relatively high frequency, while alarm features such as a palpable abdominal mass and anaemia were not studied because the number of patients with these alarm features was too small in our database to allow for statistical analysis.
In our study, the mean age of the patient population (47.6 years), the sex distribution (male/female ratio: 1.13) and the prevalence of major endoscopic findings (28.7%) were comparable with those reported in recent studies.7 14 16 25–30 We observed that there was a significant increase in the prevalence of oesophagitis with age, which was in line with the results of our previous population-based study where the prevalence of gastro-oesophageal reflux disease (GORD) increased with age.31 However, the prevalence of peptic ulcers remained stable in all age groups (∼20%) in our case series, and this finding was in contrast to several studies32 33 where the prevalence of peptic ulcers was age dependent; it is not clear what accounts for this difference. The prevalence of upper GI malignancy in our study (4.2%) was relatively higher than that of the study in Hong Kong10 (0.9%, 23/2627) and that in Taiwan9 (1.25%, 225/17894). We consider that the reason for this difference is that Shanghai has a prevalence of gastric cancer, with an age-standardised incidence rate (per 100 000 per year) of 34.3 in males and 18.9 in females,19 respectively, which is 1.8-fold higher than that of Taiwan. Furthermore, the study carried out in Taiwan did not include patients with oesophageal or duodenal malignancy. Interestingly, although the age-standardised incidence rate of gastric cancer in Shanghai was similar to that of Poland,34 which was 39.1 in males and 14.1 in females, the prevalence of upper GI malignancy in our study was much lower than that of the Polish study35 (4.2% vs 9.7%). It is thought that the dramatic difference may lie in the overselected patient population in the Polish study.36 The prevalence of malignancy increased from 0.6% in patients <35 years old to 18.0% in patients >75 years old in our study. Similarly in the study conducted in Singapore, the authors reported a positive association between cancer frequency and increasing age; the cumulative frequencies of gastric cancer by age group were 0.68 of 1000 OGDs in patients <35 years old, 1.15 in patients <45 years old and 9.60 in patients >45 years old.25
The overall proportion of patients who had alarm features in the whole patient population (14.8%) was comparable with that in the study of Vakil et al (19%, 8669/46,161). The diagnostic accuracy of individual alarm features for upper GI cancers varied substantially in our study. Among the four alarm features we studied, dysphagia had the highest sensitivity though it was only 29.4%, vomiting had the lowest sensitivity (5.7%), while the other two symptoms had intermediate sensitivity. All the four alarm features had very high specificity and NPV, and this may reflect the low prevalence (4.2%) of cancer in our patient population but not a specific attribute of any alarm feature in excluding a potential malignancy; under such conditions, high specificity and NPV would be possible with any unspecific symptom, no matter whether it was an alarm feature or not. However, by calculating the PLR and NLR of these alarm features, it is noted that only the feature of dysphagia in two age groups (36–54 and 55–74 years old) had a PLR >10, which was strongly associated with possible malignancy, while all other alarm features in other age groups had a PLR <10. The results of our study were partially consistent with those of Vakil et al12 in their systematic review; it was found that the sensitivity of alarm symptoms varied from 0% to 83% with considerable heterogeneity between studies, and the specificity also varied significantly from 40% to 98%. In our study, when individual alarm features were considered, we found that the feature of dysphagia was better than the other three alarm features (weight loss, GI bleeding and vomiting) in predicting malignancy, especially in patients between 36 and 74 years old.
The results of the present study not only provide a guide to the management of Chinese patients with dyspepsia for doctors, but also imply that in other countries, for patients who migrate from a high gastric cancer prevalence country, such as China, Japan or Korea, and present with dyspepsia, especially those who have symptoms of dysphasia or a family history of gastric cancer, a prompt endoscopy should be offered instead of just treating H pylori infection.37 For example, in the USA, Asian Americans have the highest gastric cancer incidence and death rates compared with other races.38 For this reason we believe it is difficult to issue a ‘one size fits all’ guideline for all patients with dyspepsia in the world. On the contrary, patients from different parts of the world should be managed differently according to their background prevalence of H pylori infection and family history of upper GI malignancy, as well as age and predominant symptoms.
In Western countries, the prevalence of H pylori and upper GI malignancy is low and the cost of upper endoscopy is high (from > US$1000 in the USA to US$200–500 in most European countries) 39; therefore, a cost-effectiveness analysis under such circumstance revealed that prompt endoscopy conferred a small benefit in terms of cure of dyspepsia but cost more than ‘test and treat’ and was not a cost-effective strategy for the initial management of dyspepsia.40 However, as suggested by Delaney et al, the results of this cost-effectiveness analysis were quite sensitive to the cost of upper endoscopy.41 The cost-effectiveness ratio could be reduced to one-tenth of this value if the cost of endoscopy decreased from US$373 to US$152. In China, for example, the cost of a diagnostic upper endoscopy could be as low as US$30 dollars (ranging from 25–45; unpublished data from a multicentre survey). Thus, a cost-effectiveness analysis in the setting of the Chinese condition may draw a totally different conclusion. In addition, recent Asian studies cast doubt on the safety of the ‘test and treat’ strategy for the management of young patients with dyspepsia without alarm symptoms. Sung et al detected 17.4% upper GI malignancies in these low-risk patients,10 while Li et al11 found, among 202 Chinese patients with GI malignancies, that only 108 (53.5%) presented with alarm features. However, it should be noted that although we detected many young patients with upper GI malignancy by endoscopy, the proportion of patients with cancer among the whole ≤35 years patient group was 0.6% (128/22 011), which suggested that endoscopy, though very cheap in China, may not be the most cost-effective method and other non-invasive, inexpensive screening tools are still needed.
In conclusion, for uninvestigated Chinese patients with dyspepsia with high background prevalence of H pylori infection and upper GI malignancy, age and alarm features, except for dysphagia in patients between 36 and 74 years old, were not accurate for prediction of malignancy. Therefore, doctors may take a more generous approach to avoid missing cancers in the upper GI tract. Taking the wide availability and relatively low cost for upper GI endoscopy in China into consideration, prompt endoscopy may be recommended. However, less invasive, inexpensive screening methods are still needed to reduce the endoscopy workload.
Significance of this study
What is already known about this subject?
Dyspepsia is very common in that >25% of adults in the general population have dyspeptic symptoms.
Nearly all Western guidelines suggested that alarm features at any age warranted prompt endoscopy, and most recommend an age cut-off of between 50 and 55 years for endoscopy as an initial management strategy for dyspepsia.
The diagnostic value of alarm features in Chinese patients with dyspepsia with a high background prevalence of H pylori infection and upper GI malignancy is uncertain.
What are the new findings?
Among 15 235 Chinese patients with dyspepsia who had alarm features, 2258 (14.8%) were found to have upper GI malignancy. Among all the 4362 patients with malignancy, 52% (2258/4362) had alarm features.
The pooled sensitivity and specificity of the alarm features for malignancy prediction were 13.4% and 96.6%, respectively. Only the alarm feature of dysphagia in patients between 36 and 74 years old had a PLR >10 for malignancy prediction, while all other alarm features in other age groups had a PLR <10.
How might it impact on clinical practice in the foreseeable future?
For uninvestigated Chinese patients with dyspepsia, except for dysphagia in patients between 36 and 74 years old, alarm features and age had limited predictive value for a potential malignancy; therefore, prompt endoscopy may be recommended for these patients.
Overall, only a small proportion of patients with dyspepsia were diagnosed to have malignancy after upper GI endoscopy; therefore, less invasive, inexpensive screening methods with high diagnostic yield are still needed to reduce unnecessary endoscopy.
We thank all the doctors and nurses who helped to manage the patients who underwent OGD in our centre during the years of the period studied.
Funding YB is partly supported by the ‘Chen Guang’ project supported by the Shanghai Municipal Education Commission and Shanghai Education Development Foundation (Grant no. 2008CG44), and partly supported by the National Natural Science Foundation of China (Grant No. 30801087).
Competing interests None.
Ethics approval This study was conducted with the approval of the Shanghai Changhai Hospital Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.