Table 7LOWER GASTROINTESTINAL CANCER: signs and symptoms, including risk factors

AuthorSettingDescriptionNo.InclusionExclusionResultsGold StdQuality
Ahsan et al 1998USA reconstructed cohort study to examine the risk of colorectal cancer among first-degree relatives of patients with adenoma compared with that among first-degree relatives of controls without adenoma1554The risk of colorectal cancer was elevated (RR, 1.74 {95% CI, 1.24–2.45) among first degree relatives of patients with newly diagnosed adenomas compared with the risk among first-degree relatives of controls. First degree relatives of patients with adenomas did not have elevated risk for other cancers. The risk for colorectal cancer among family members increased with decreasing age at diagnosis of adenoma in probands. Among first degree relatives of patients who were 50 years of age or younger when the adenoma was diagnosed, the risk was more than four times greater (RR, 4.36 [CI 2.24–8.51]) than that among first degree relatives of patients who were older than 60 years of age when adenoma was diagnosed.
Baquet and Commiskey 1999USdata from several population-based cancer registries were used to identify the incidence of colorectal cancer in different ethnic groups.For men, the age-adjusted incidence rates were highest in Alaskan natives (79.7/100,000), followed by Japanese (64.1/100,000), then African-American (60.7/100,000), white (56.3/100,000), Vietnamese (30.5/100,000), and American Indian (18.6/100,000). For women, rates were highest in Alaskan natives (67.4/100,000), then African-Americans (45.5/100,000), Japanese (39.5/100,000), white (38.3/100,000), Vietnamese (27.1/100,000) and American Indian (15.3/100,000).The particular ethnic groups in the study were those typical of the US, and therefore the findings are not directly applicable to the UK
Bellentani et al 1990Italy, Primary CareBetween Jan 1987–Mar 1988 patients consulting 14 GPs covering 14,000 citizens, or referred to the outpatient gastroenterology unit, either complaining of recurrent abdominal pain or having intestinal problems (as judged by the GP), were studied254All patients referred to the gastroenterology unit, either complaining of recurrent abdominal pain or having intestinal problems.Patients with acute abdomen, acute gastroenteritis or a clear diagnosis of upper gastrointestinal tract disease (gastritis, oesophagitis, peptic ulcer, or dyspepsia).Six parameters were significantly more common in patients with organic disease & weighted as a positive score, namely ESR>17mm; first hour, history of blood in stool, leukocytes>10 000cm3, age>45, slight fever and presence of neoplastic colonic diseases in first-degree relatives.Final diagnosis after investigationThe number of patients with colorectal cancer was only 10. The symptom score was used to detect organic disease rather than colorectal cancer uniquely.
Bonelli et al 1988The relationship between first degree family history of colorectal cancer and the risk of benign or malignant tumours of the large bowel was investigated in a case control study.Two groups of cases :283 patients with adenomatous polyps and 414 patients with adenocarcinoma of the large bowel.

Two groups of controls: 399 polyp free subjects and 456 hospitalised patients.
Data from the two control groups were combined. A 3 fold increase in risk of adenomatous polyps in relatives of patients with colon cancer was observed (OR=3.18, 95% CI 2.06–4.89). The relative risk of colorectal carcinoma among relatives of patients with adenocarcinoma was 2.36 (95% CI 1.54– 3.60). No significant difference in the frequency of first degree relatives with a history of cancer of the large bowel was detected between patients with colorectal cancer and those with adenomatous polyps
Burke et al 1997Studies of cancer risk, surveillance and risk reduction in individuals genetically susceptible to colon cancer were sought through a search of MEDLINE 1990–1995The risk of colorectal cancer in people with confirmed HNPCC was estimated to be 68% to 75% by age 65, although the average age at diagnosis is 45 years. The risk of a new primary after limited resection for a first cancer is also high at 30% after 10 years. Endometrial cancer is the second most common cancer seen in HNPCC.
Cannon-Albright et al 1988Members of the participants family and their spouses were screened by flexible proctosigmoidoscopic examination (60 cm) to determine how frequently colorectal adenomas and cancers result from an inherited susceptibility670Excluding the probands, 18 relatives in the 34 kindreds and 4 of the spouses were found to have a history of colorectal cancer. Adenomas were found in 78 of 407 (19%) of the relatives of probands and in 32 of 263 (12%) of the spouses (P<0.02 for the difference between relatives and spouses). The average age of the 407 relatives was 51 years, and that of the 263 spouses was 52 years. The results suggested that an inherited susceptibility to colonic adenomatous polyps and colorectal cancer was common and that it was probably responsible for the majority of colonic neoplasms observed clinically. An underlying genetic susceptibility was present in the majority of persons with common colonic adenomatous polyps and colorectal cancers.
Chapuis et al 1985AustraliaMen aged above 50 years were interviewed and underwent sigmoidoscopy319Of 351 veterans sampled, 328 (93%) consented to undergo flexible sigmoidoscopy.Those who did not give their consent were excluded from the analysis.Rectal bleeding had a specificity of 86%, a sensitivity of 33% and a positive predictive value of 8% for rectal or sigmoid polyps or cancer. Sigmoidoscopy in apparently healthy subjects will not result in the diagnosis of appreciable numbers of rectal and sigmoid polyps or cancers.After a phosphate enema had been given, flexible sigmoidoscopy was performed with a 60cm fibreoptic sigmoidoscope.Population survey of bleeding. Although flexible sigmoidoscopy was performed in this study by experienced endoscopists, in a minority of patients the instrument was not inserted beyond 40cm. This difficulty could be overcome by better bowel preparation. But the findings were valid for rigid sigmoidoscopy, for which the average length of insertion has been reported as 19.5 cm.
Curless et al 1994UKCase controlled study of 273 patients newly diagnosed with colorectal cancer and 273 age and sex matched community controls.546nonePatients with known colorectal adenoma, carcinoma of IBDLower GI symptoms were reported in a small but clinically significant number of community controls, particularly those aged 70 or older. The odds of 10 of 11 symptoms were greater among patients with cancer: change in bowl habit, abdominal pain, faecal incontinence, tenesmus, mucus, bleeding, change in flatus, anorexia, weight loss, bloating. There was no difference for malaise.A specialist diagnosis of colorectal cancer, after investigations.The study did not include patients consulting in primary care.
Dodds et al, 1999UKRecording of symptoms and signs of patients referred to a specialist service.8438--471(5.6%) patients had colorectal cancers. Rectal bleeding & change in bowel habit were present in 252 (54%), PPV 1:8, LR 2.5; change in bowel habit alone in 110 (23%), PPV 1:17, LR 1.1; bleeding alone 48 (10%), PPV 1:18, LR 3.2; bleeding & perianal symptoms 17 (4%), PPV 1:148, LR 0.36.Diagnosis after full investigation.The study was limited to patients who had been referred to secondary care.
Eaden et al 2000a meta-analysis of the risk of colorectal cancer in patients with ulcerative colitis, and involved a literature search using Medline to identify 194 studies of which 116 met the inclusion criteria116 studies

54, 478 included in the studies.
The overall prevalence of colorectal cancers in any ulcerative colitis patient, was estimated to be 3.7% (95% CI 3.2–4.2%).
41 studies reported Colitis duration. From these, the overall incidence rate was 3/1000 person years duration (95% CI 2/1000 to 4/1000). 19 studies reported incidence stratified into 10-year periods. For the first 10 years, the incidence rate was 2/1000 person years duration, (95% CI 1/1000 – 2/1000), for the second decade 7/1000 person years duration (95% CI 4/1000 – 12/1000), and in the third decade 12/1000 person years duration (95% CI 7/1000 – 19/1000) The overall incidence rate for any child was 6/1000 patient year duration (95% CI 3/1000 to 13/1000). age at onset in adults appeared to have no statistically significant bearing on cancer risk.
some reservations about the primary studies should be noted. Many of the studies in the meta analysis were population based and their inclusion did not rely on contact with gastroenterologists. However, there was a greater likelihood that cancers were detected among those having active follow up as a majority of cases came from surveillance programmes or tertiary referral centres, and very few studies included in the meta analysis used national cancer registry data.
Eisen and Sandler 1994A meta-analysis which combined the results of 7 cohort studies was conducted in order to critically review the risk of colorectal cancer in patients with breast cancer16 studiesThe pooled results of seven cohort studies demonstrated a weak association between breast cancer and the subsequent risk of colorectal cancer [pooled relative risk (RR = 1.5; 95% confidence intervals (CI) = 0.99– 1.31]. Pooled results from another five cohort studies showed that the risk of breast cancer after colorectal cancer was similar (pooled RR=1.10; 95% CI=1.03– 1.17). The combined results from five other cross sectional/case control studies revealed a positive association between breast cancer and colorectal adenomas (pooled RR = 1.74; 95% CI=1.27– 2.21). However, population based cohort studies, a stronger research design, showed essentially no increase in risk of colorectal cancer in women with previous breast cancer.
Fijten et al 1995Netherlands General practiceData collected about patients presenting overt rectal bleeding to the GP (83 GPs in the Netherlands). Outcome measures were sensitivity, specificity, predictive values, odds ratios and a prediction model derived from multiple logistic regression analysis. A further report from the study reported in Fitjen et al, 1993.269Patients presenting with overt rectal bleeding to the GP. There were fewer than expected patients with rectal bleedingPatients younger than 17 yrs or older than 75, pregnant, or urgent admissions (eg massive bleeding or acute abdominal pain)9 patients had colorectal cancer. Age (OR 8), bowel habit change (OR 10) & blood mixed with or on stool (OR 8), were predictive of cancer.Follow up for at least one year, with diagnostic information being extracted from the medical record.Small no of 9 patients with cancer in the study. Does not state how many of those without bleeding had cancer.
Fijten et al 1994The review was undertaken to determine the occurrence and significance of overt blood loss per rectum.Nine studies were found reporting the occurrence of rectal bleeding in the general populationOccurrence rates varied from 2% in the last 2 weeks to 20% in the last year. The positive predictive value of rectal bleeding in the general population was reported in four studies, varying from 3% to 8% for prediction of adenomas and 0% to 1% for carcinomas. The incidence of rectal bleeding without a specified diagnosis was 0.4 per 1000 persons per year. The incidence of bleeding associated with the diagnosis of haemorrhoids was 6.8/1000 consulting persons per year, anal fissure or perianal abscess 3.2, diverticular disease 1.6, colitis 0.8, and cancer 0 per 1000 persons per year. No epidemiologic data on the diagnostic value of rectal bleeding in patients presenting in primary care were found.
The authors of the review estimated from the findings of a single Dutch study that around 0.8 per 1000 persons per year were referred with rectal bleeding by general practitioners to specialists. They went on to estimate the predictive values of rectal bleeding for colorectal cancer from the data they had identified of less than one in 1000 in the general population, two in 100 in general practice, and up to 36 in 100 referred patients.
estimates involved several assumptions and they cannot be taken as precise.
Fitjen et al 1993Netherlands General practiceData collected about consecutively attending patients with rectal bleeding.
There were 83 doctors in study A and 10 involved with study B.
290 study A

48 study B
All patients presenting to the doctor with overt rectal bleeding or a history of recent (ie within the last 3 months) rectal blood loss visible to the patient.Exclusions were younger than 18 years or older than 75 years, pregnancy and urgent admission to a hospital (massive bleeding or acute abdomen).In about 90% of patients, rectal bleeding was due to minor ailments or self- limiting disorders. Incidence of rectal bleeding in study A, was 2.15 per 1000 persons per year. (6.8 per 1000 in study B). Gender differences were not statistically significant. In both studies combined, 3% of the total 313 patients were diagnosed with colorectal cancer.Follow up for a minimum of 12 months.Bias in the selection of patients with clinically relevant bleeding. Initial protocol used for patient selection to the study not made explicit. Problem of non compliance by doctors, resulted in under registration of patients visiting practice with rectal bleeding, particularly younger patients in study A. The studies included only a small number of patients with colorectal cancer. Patient drop out rate affected study A results, but the mean incidence rate was lower in study B.
Fuchs et al 1994A prospective study of men and women who provided data on first degree fmily relatives with colorectal cancer, diet and other risk factors for the diease.

Data were analysed from two ongoing studies: the Nurses Health Study and the Health Professionals Follow-Up Study
32,085 men

87,031 women
The age adjusted relative risk of colorectal cancer for men and women with affected first-degree relatives, as compared with those without a family history of the disease, was 1.72 (95% CI, 1.34–2.19). The relative risk among study participants with two or more affected first-degree relatives was 2.75 (95% CI, 1.34–5.63). For participants under the age of 45 years who had one or more affected first-degree relatives, the RR was 5.37 (95% CI, 1.98–14.6), and the risk decreased with increasing age (P for trend, <0.001). Among women, the relative risk associated with a family history was highest for those younger than 50 years of age and the risk decreased progressively for older women
Goulston 1986Australia Secondary careProspective study of patients above 40 years presenting to 58 GPs with overt rectal bleeding, who were followed through to final diagnosis to determine how successfully GPs & gastroenterologists establish the source of bleeding before full colonic investigation.145All patients aged 40 years or over with rectal bleeding.Patients were excluded if a) their age or general medical condition precluded colonoscopy, b) they were known to have inflammatory bowel disease, colorectal cancer, or polyposis coli c) a coagulation defect or haematological disorder was present d) bleeding was melaena, or e) the patient refused investigation.14 patients had one colorectal cancer and 1 patient had 2. Of the 63 patients in whom GPs predicted an anal source of bleeding only, 11 were ultimately found to be bleeding from a colonic or rectal source (NPV=82.5%); a false negative rate of 34.4%. However, GPs accuracy in attributing bleeding solely to a colonic or rectal source was much lower (PPV 35.6%). The NPV for colorectal cancer as assessed by GPs was 95.4% and the PPV 20.7%.Diagnosis of source of bleeding before and after complete colonoscopic investigation.Seven specialists participating in the study recruited GPs who regularly referred patients to them..

Those participating were possibly well informed about rectal bleeding as a symptom or were more interested and motivated.
Grodstein et al 1999A MEDLINE search dating from January 1966 to September 1998 yielded 18 epidemiologic studies of postmenopausal hormone therapy and colorectal cancerEvidence these studies indicated that postmenopausal hormone therapy was associated with a 20% reduction in the risk of colon cancer in women (RR+0.80, 95% CI, 0.72 to 0.92) and a 19% decrease (RR=0.81, 95% CI, 0.72 to 0.92) in the risk of rectal cancer for postmenopausal women who had ever taken hormone therapy compared with women who never used hormones. Much of the apparent reduction in colorectal cancer was limited to current hormone users (RR+0.66, 95% CI, 0.59 to 0.74)

Of the ten studies with data on current hormone use, nine found a decreased risk of colorectal cancer. The RRs in the 10 studies ranged from 0.4 to 1.0, with a summary RR of 0.66 (95% CI, 0.59 to 0.74) for current use. There was greater protection against colorectal cancer for recent use than for past use, compared with women who never took postmenopausal hormones. The observed decrease was strongest among women currently taking postmenopausal hormones, in whom the risk of colorectal cancer was 34% lower than in never users
A potential bias in the primary studies was the possibility that women who chose to take postmenopausal hormones differed from non users in ways that influenced their risk of colorectal cancer. For example, women taking hormones had to visit their physician and undergo routine stool occult blood tests and endoscopy more frequently, which would reduce the risk of cancer through removal of precancerous lesions. Thus, it may appear that hormone use reduces the risk of cancer when it is simply behavioural characteristics of hormone users that lead to their lower rate of malignancy.
Hamilton 1985A small Case series review supported by a literature review.there was some evidence that patients with Crohn’s disease and colorectal cancer were younger than other patients with colorectal cancer, and the histopathological type of cancer was different, being mucinous in 50% of the case series with Crohn’s in comparison with 9% in other colorectal cancer patients.The quality of the study is not sufficient to provide convincing evidence that Crohn’s disease is associated with a higher risk of development of colorectal cancer,
Helfand et al 1997USPatients with rectal bleeding identified by system review at consultations; these were investigated by sigmoidoscopy and barium enema. The diagnosis was reviewed after 10 years.297 (201 completed all the investigations)--13 of the 201 had colon cancer. The ten year incidence of cancer was not statistically different to that expected in a similar cohort of the general population.Investigation for colorectal disease.Only a small number of patients with cancer were included. The study included patients who reported rectal bleeding on questioning, rather than those consulting because of concern about bleeding.
Jarvinen et al 1995A study was based on an evaluation of the effectiveness of long term screening during a 10 year period.

The colorectal cancer and death rates were compared between two groups of asymptomatic at risk members of 22 families with HNPCC
251 (incl 118 controls)CRC occurred in 6 study subjects (4.5%) and in 14 controls (11.9%; P=0.03), a difference of 7.4% in favour of the study group. The tumour stage was more favourable in the screening group with no deaths caused by CRC compared with 5 of 14 cases in controls. Overall, there were 6 and 12 deaths within the 10 year period in the study and control groups, respectively (P=0.08). The 3 year interval screening programme more than halved the CRC rate in at-risk members of families with HNPCC and seemed to prevent CRC deaths.
Mansson et al 1999Primary Care SwedenBased on analysis of medical records in one district of all subjects with colorectal, pulmonary, breast or prostate cancer reported to the Swedish cancer registry.42nonenoneThe study provided information about the proportion of patients with colorectal cancer who had particular symptoms at initial presentation to the GP.Notification of diagnosis to cancer registryNo patients included who did not have colorectal cancer
Mant et al, 1989Australia Secondary carePatients who complained of rectal bleeding were referred to a specialist for colonic investigation. 248 patients aged 40 years and older with rectal bleeding, consulted 58 GPs over an 11 month.145Patients aged 40 years and older, who consulted for rectal bleeding as a clinical problem.Patients were excluded if 1) age, or general medical condition precluded colonoscopy, 2) known to have inflammatory bowel disease, colorectal cancer, or polyposis coli, 3) a coagulation defect or haematologic disorder was present, 4) bleeding was melenic or 5) the patient refused investigation.16% of patients with haemorrhoids also had a colorectal source of bleeding and in 5% this was a malignancy. There was a slight tendency for patients reporting dark blood to have a colorectal source. Blood mixed with faeces was more likely to come from a colorectal than an anal source.Association between patient history and final diagnosis established through complete investigation.33 patients were not referred to a specialist either because they refused to participate or because the GP decided it was ethically unviable. Symptoms and signs for which there was missing data were family history of colorectal cancer (2 patients), colour of blood, where the blood was seen, whether the blood was separate from or mixed with faeces (5 no data, 21 uncertain); abdominal pain (1 patient), bowel habit change (2 patients)
Metcalf et al 1996UKProspective study in which patients consecutively presenting with rectal bleeding to GPs underwent colooscopy99Age 40 or over-8 (8.1%) had carcinoma. The sensitivity and specificity of presenting symptoms were low.ColonoscopyIt is unlikely that all patients presenting with rectal bleeding were enrolled in the study.
Muris et al 1995NetherlandsOne year prospective observational study in 80 general practices933Aged 18–75; symptoms for minimum of 2 weeks, and with whom GP had a diagnostic problemRefusal of consent4/933 had colorectal cancer; 24 had a neoplasm (colorectal, stomach, pancreas, other). Diagnosis of cancer associated with no specific character to pain; pain relief after defaecation, pain affecting sleep; blood in stool. Regression analysis found male sex, greater age, no specific character to pain, weight loss, ESR above 20mm predicted neoplasm.Diagnosis taken from clinical records, a mean of 18\months after presentation.Few patients with colorectal cancer, and findings not specific to colorectal cancer.
Muris et al, 1993NetherlandsPatients from 11 general practices with a population of 25,000 were followed for 15 months to evaluate abdominal pain.578Cases of non acute abdominal pain were enrolled into the study.Patients younger than 18 years and patients with a condition necessitating immediate referral to or admission into a hospital were not included into the study.Diagnosis of abdominal or stomach pain with unknown cause; irritable bowel syndrome; acute gastroenteritis, presumed viral and disorders of the stomach function accounted for 70% of the final diagnoses. In 17% of cases the patient was referred most often to internal medicine (40%), surgery (24%) and gynaecology (21%). Ten people died during the follow up period of whom 2 cases were caused by malignant colorectal disease. Cases of abdominal complaints in age groups 18–39 295 ; 40–49 80; 50–59 91; 60–75 88.Independent classification of diagnosis 15 months later.A prospective study. The most appropriate work up of patients with abdominal pain is not specified. The authors established that abdominal pain is common in general practice, but did not have a large sample of cancer patients, or elaborate on how their signs and symptoms were different. Only three malignant colorectal diseases were detected.
Murris et al 1993A Medline search was undertaken for publications between 1982 and 1991 that investigated the diagnostic value of rectal examination in patients with abdominal pain and urinary complaintsEight studiesAll the studies were carried out in populations selected by referral, adequate gold standards, based on histological evidence. The sensitivity of rectal examination for detecting rectal carcinoma in the two relevant studies were 50% and 24%; in one of these studies the specificity had been estimated as 95%, and likelihood ratio 4.8.
Norrelund et al 1996General practice in DenmarkDanish GPs recorded information about 3–4 patients each presenting with rectal bleeding in 1989, and for a second sample in 1991.208 patients
2. 209 patients
Age 40 or over, rectal bleedingKnown inflammatory bowel disease, colonic polyps, polyposis coli, colorectal cancer, coagulation defects, melaena32/208 had colorectal cancer; only age and change in bowel habit predictive of cancer. 2. 28/108 who presented with first bleeding had cancer or polyps; 12/48 with change in bleeding pattern had cancer or polyps; 3/45 unchanged bleeding pattern had cancer or polyps. Combined data from both studies indicated only age and change in bowel habit were associated with cancerAt least one year follow upFull ascertainment of presenting patients may not have been achieved.
Pinsky et al 2003Patients who were part of a prostate, lung, colorectal and ovarian cancer screening trial were asked to complete a baseline questionnaire component with a family history section149,332A total of 26 respondents (0.02%) had families meeting the Amsterdam criteria for hereditary nonpolyposis colon cancer. The reported prevalence per 1000 in siblings was 9.4% for colorectal cancer. The ratio of reported to expected rate in men was 0.60 (95% CI, 0.57, 0.62) and in women it was 0.76 (0.73, 079).
Radack and Park 1993A systematic review was undertaken of articles identified by search of Medline for all relevant studies from 1983 until January 1992 to assess the clinical utility of skin tags as a biomarker for colonic polyps10 reportsOnly four of the ten studies reported a statistically significant association between skin tags and colonic polyps; the remaining studies reported outcomes indicating no association.
Only four of the ten studies reported a statistically significant association between skin tags and colonic polyps; the remaining studies reported outcomes indicating no association.
Limitations potentially responsible for the varying outcomes included lack of blinded ascertainment of clinical information, noncomparability of subjects, differing diagnostic investigations of the colon, and uncontrolled confounding. All but one study were performed in a tertiary care setting, seriously limiting the relevance of the results to the “average” subject seen in primary care settings. Variability in study populations, methods of diagnostic evaluation and the control of possible confounders that could affect the potential relationship. For these reasons, the review did not provide a reliable estimate of any association between skin tags and polyps.
SIGN 2003Guidelines for the management of colorectal cancerEvidence based.
Nationally recognised.
Silman et al 1983AustraliaSymptom questionnaires were sent to workers in two large organisations and the results were compared with faecal occult blood testing in the same individuals using haemoccult.916All patients who returned their questionnaires were included.No exclusions were mentioned.No cancers were discovered but adenomas were found in 14 persons out of 916 giving a yield of 1.5%. In 7 of these, an adenoma greater than 10mm in diameter was present giving a yield of 0.8%. All 7 individuals had at least one symptom including 4 with dark bleeding, 2 with bright bleeding and 1 with diarrhoea. Six of the 7 individuals were haemoccult positive.Use of haemoccult test to indicate presence of an adenoma.Population study No cancers were reported The high rate of 12% of large bowel and anal symptoms among the screened population of 916 may have been influenced by some selection since symptomatic persons might have been more ready to participate. The impact of the education programme prior to screening may well have been to encourage those with symptoms selectively to comply.
Stellon et al 1997UKPatients in one general practice over the age of 50 years with proven iron deficiency anaemia were identified, investigated and followed up in general practice over a five year period to evaluate extent to which GI the tract should be investigated. Data collected between Jan 89–Mar 94.26All patients over 50 years of age found to have iron deficiency anaemia were entered in the study from a patient list size that varied between 2,400 and 3,400 patients during the study period.No exclusion criteria mentioned22 of the patients had faecal occult blood tests but only five were found to be positive.
Endoscopy revealed significant disease in eight patients that could have accounted for the iron deficiency. The 22 barium enema examinations gave a diagnosis in two patients.
Five year follow up.

Faecal occult bloods using the haemoccult test, upper GI endoscopy, flexible sigmoidoscopy and the double contrast barium enema
Prospective study. Case series.
Small sample of general practice patients were reported. One patient had caecal carcinoma not reported on the initial barium enema.
Trilling et al 1991USCharts of 173 patients with hemorrhoids from a non selected population were reviewed for treatment management, associated anorectal disease, and sequelae.173All patients who had external or internal hemorrhoids diagnosed and were seen in the model Family Practice centre.No exclusion criteriaDuring the period of chart review, eight cases of colon carcinoma and nine non colon gastrointestinal cancers were diagnosed in the practice in patients without hemorrhoidal disease.
Sigmoidoscopy was performed in 72 patients. Findings were normal in 57 patients and abnormal in 15. Hemorrhoids usually coexist with other anorectal diseases.
Diagnosis after sigmoidoscopyPhysician approach toward hemorrhoid management based upon the practice habits of 10 academic physicians are difficult to generalise to the general population of family physicians.
The study was limited by the small population size and the relatively few patients who had sigmoidoscopy or barium enema to rule out colon cancer.
Winawer et al 1993The incidence of colorectal cancer was retrospectively compared in two cohort studies to test whether the current practice of removing adenomatous polyps of the colon and rectum prevented colorectal cancer.
In group 1 patients who had polyps 1cm or larger and had declined surgical polypectomy, were followed for an average of nine years and 32 colon cancers were detected. In Group 2 a cohort of patients who underwent excision of rectal adenomas were followed for an average of 14 years
Group 1: 226 patients

Group 2: 1618 patients
Group1: Of the cancers detected, 21 (66%) were detected at the same site as the index polyp and 11 (34%) were at sites distant from it.

Group 2: 35 colon cancers were detected. The standardised incidence ratio for colon cancer was 2.1.
Winawer et al 1996A large cohort of patients with newly diagnosed adenomatous polyps were interviewed in order to establish whether their families were at increased risk of colorectal cancer.Colorectal cancer was reported in 68 of 2381 siblings, 133 of 1865 parents, and 29 of 1411 spouse controls. Of the 230 first-degree relatives or spouses reported to have had colorectal cancer 171 (74.3%) had died. The average age was 73.2 years for the parents, 62.3 years for the siblings, and 63.7 years for the spouses. The first degree relatives of patients with adenomas had an increased risk of colorectal cancer as compared with spouse controls (relative risk 1.78; 95 percent confidence interval, 1.18 to 2.67).
The risk of colorectal cancer was higher for the siblings and parents of patients in whom adenomas were diagnosed before the age of 50 years and for the siblings of patients given the diagnosis at 50 to 59 years of age than for the siblings and parents of patients in whom adenomas were diagnosed at 60 years or older. The risk of colorectal cancer increased in the siblings with decreasing age of the index patient at the diagnosis of adenoma (P for trend<0.001).

From: Appendix A, Evidence Tables

Cover of Referral Guidelines for Suspected Cancer in Adults and Children
Referral Guidelines for Suspected Cancer in Adults and Children [Internet].
NICE Clinical Guidelines, No. 27.
Clinical Governance Research and Development Unit (CGRDU), Department of Health Sciences, University of Leicester.
Copyright © 2005, National Collaborating Centre for Primary Care.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.