Delays in the diagnosis of oesophagogastric cancer: a consecutive case series

Introduction

The Japanese make strenuous efforts to diagnose cancer of the stomach and oesophagus at an early stage. Patients have wide en bloc resection of the primary tumour and its draining lymph nodes to achieve clear margins of resection, proximally, distally, and circumferentially (so called R0 resection). Operative mortality is about 2% and five year survival after resection for gastric cancer is 86% and for all patients who have resection 64%.1 For patients with oesophageal cancer operative mortality is 4% and five year survival 30%.2

In Britain, however, both gastric and oesophageal cancer have been regarded as fatal diseases until recently. There is no screening programme, and even patients who develop suspicious symptoms such as weight loss, anaemia, dysphagia, and vomiting may remain undiagnosed for many months. When cancer of the stomach or oesophagus is finally diagnosed surgery is usually far less extensive than in Japan. Operative mortality is between 5%3 and 15%4 after resection and five year survival of gastric cancer is 5-10% for all cases and just 20-30% after resection.5 The results for oesophageal cancer are even worse. In Yorkshire operative mortality was reported to be 13-15% and five year survival just 1-2% overall and 6-10% after resection.6

We have previously shown that when Japanese surgical methods were applied in Britain the five year survival of British patients with gastric cancer was similar to that of Japanese patients with the same stage disease.3 5 The task, then, is to attempt to diagnose the disease in Britain at the earliest possible stage and to perform more radical resections. We describe the results of a study of 115 consecutive patients with gastric or oesophageal cancer who were referred to our department over 16 months, starting in January 1994. We examined the length of delay in diagnosis and the reasons.

Subjects and methods

One of us (SY) interviewed each patient at first presentation to our department. All data were verified by a second author (IGM). Dates were recorded according to the patients' recollection and cross referenced with the patients' notes. Details of the patient's first symptoms, the number of visits to the general practitioner before referral to hospital, and of any relevant drug treatment were recorded. We followed the patients' subsequent clinical course.

The time to diagnosis was measured from the date when the patient first experienced the symptoms that led to diagnosis. In the case of patients with long standing dyspepsia, this date was taken to be when the patient first noticed a significant change in these symptoms. The primary end point was the establishment of a histological diagnosis of malignancy and the second endpoint was the date when the patient had definitive surgery (if appropriate). The overall delay in weeks was recorded for each patient and divided into four periods:

1. The time from first symptoms to the patient first seeking medical advice.

2. The time from first seeking medical advice to referral for investigation by endoscopy or barium meal or for a hospital consultant's opinion.

3. The time from referral to first attendance at hospital for investigation or outpatient consultation.

4. The time from first attendance at hospital to establishment of a definitive histological diagnosis.

For patients who had surgery we also recorded the time from histological diagnosis to the operation.

Tumours were staged according to the 1987 TNM classification,7 8 whenever possible from operative specimens. We used statistics appropriate for non-parametric data. Grouped data were compared by the Mann-Whitney U test, corrected for ties and small numbers.

Results

We recruited 115 patients (70 men and 45 women) over 18 months. Eighty eight patients had cancer of the stomach and 27 cancer of the oesophagus. Of the 88 gastric cancers, 27 were located predominantly in the upper third of the stomach, 33 mainly in the middle third, and 28 in the lower third, The median age of the patients when they first developed symptoms was 66 years (range 31 to 89 years).

The first symptoms or signs were dyspepsia or indigestion in 19 (17%), dysphagia in 41 (24%), abdominal or chest pain in 48 (28%), nausea or vomiting in 27 (16%), heartburn in 7 (4%), weight loss in 20 (12%), early satiety in 27 (16%), and anaemia in 19 (17%). Some patients experienced more than one symptom.

All patients had endoscopy and staging investigations (chest x ray film, abdominal ultrasonography, and thoracoabdominal computed tomography). Twenty one (18%) had stage I disease, 13 (11%) stage II disease, 47 (41%) stage III disease, and 34 (30%) stage IV disease (table 1)).

Eighty (91%) of the patients with gastric cancer and 25 (92%) of the patients with oesophageal cancer were treated surgically. Of the surgical procedures, 78 (74%) were regarded as potentially curative resections, 24 (23%) were palliative resections, and three (3%) palliative bypasses or laparotomy.

The median delay from the onset of symptoms to a definitive histological diagnosis was 17.1 weeks for patients with gastric cancer and 17.3 weeks for patients with oesophageal cancer. Table 2) shows the breakdown of this delay. Overall, delay in consulting a doctor accounted for 29% of the total, delay in referral 23%, delay in being seen at hospital 16%, and delay in establishing the diagnosis at the hospital 32%.

Twenty nine patients experienced delays in diagnosis of more than 28 weeks, and among these the median delay was 39 weeks (43 weeks if the time to treatment is included). For these patients, delay in presenting to a doctor accounted for 18% of the overall delay, delay in being referred for investigation and treatment 33%, delay in being seen at the hospital for investigation or consultation 13%, and delay in establishing a histological diagnosis at the hospital 36%.

Among the patients who had surgery there was a further median delay of 3.9 weeks before operation. However, as this extra delay did not alter the conclusions of any of the subsequent analyses and since not every patient had surgery, only the delay between the onset of symptoms and histological diagnosis is considered below.

General practitioners referred 51 patients immediately for investigations or consultation. The remaining 64 patients made further visits before being referred (16 made one further visit, 16 two further visits, 18 three further visits, and 14 patients four or more visits). Fifty five patients were given a prescription at the initial consultation: 23 for simple antacids, 19 for H2 receptor antagonists, seven for proton pump inhibitors, and six for other drugs.

When patients were seen at the hospital the diagnosis was established within four weeks in 70 patients (61%) and within eight weeks for 88 (76%). For 17 patients the delay was between 12 and 87 (median 33) weeks.

Relations with delay in diagnosis

We found no significant relation between the nature of the first symptoms and delay in diagnosis. Even when dysphagia was the main first symptom the diagnosis was not made significantly more rapidly than in patients with less dramatic symptoms. Similarly no relation was found between diagnostic delay and tumour location. The median delay was 17 weeks for oesophageal carcinomas, 14 weeks for lesions of the upper stomach, 16 weeks for middle stomach lesions, and 22 weeks for antral and pyloric lesions.

Use of our open access endoscopy service reduced the delay in diagnosis. Overall the median delay for the 65 patients referred directly to the open access dyspepsia clinic was 14 weeks compared with 25 weeks for the 50 who were more conventionally referred (P<0.001).

For patients with stomach cancer there was no clear relation between tumour stage and delay in diagnosis (fig 1). For oesophageal cancer, however, the median delay was 6.7 weeks in patients with stage I and II disease but 20.9 weeks in those with stage III and IV disease (P<0.02, fig 22).

Fig 1

Relation between stage of gastric cancer and delays in establishing histological diagnosis (thick line=median, box=interquartile range, whiskers=full range)

• Download figure
• Open in new tab
• Download powerpoint

Fig 2

Relation between stage of oesophageal cancer and delays in establishing histological diagnosis (thick line=median, box=interquartile range, whiskers=full range)

• Download figure
• Open in new tab
• Download powerpoint

In patients with gastric cancer we found no relation between delay in diagnosis and the success of potentially curative resection (17.7 weeks for successful operations v 17.8 for unsuccessful). The relation was also non-significant for patients with oesophageal cancer (15 v 24 weeks, P=0.2).

Discussion

Although our department has a well established open access endoscopy service, the median interval between first symptoms and histological diagnosis was 17 weeks and the median interval between first symptoms and definitive treatment 21 weeks. Furthermore, in a quarter of patients the median interval to diagnosis was 39 weeks and to treatment 43 weeks. Even though some patients had to visit their general practitioner several times before being referred for investigation, the longest delays were in hospital. It seems likely that the delays in other hospitals in Britain are likely to be similar if not longer.

We found that the patients were usually quick to seek medical advice for their symptoms. Much of the delay in diagnosis could therefore be avoided if general practitioners referred patients promptly for investigation and a sense of urgency was imparted to the hospital's diagnostic process.