The risk of oesophago-gastric cancer in symptomatic patients in primary care: a large case-control study using electronic records

Background:

Over 15 000 new oesophago-gastric cancers are diagnosed annually in the United Kingdom, with most being advanced disease. We identified and quantified features of this cancer in primary care.

Methods:

Case–control study using electronic primary-care records of the UK patients aged ⩾40 years was performed. Cases with primary oesophago-gastric cancer were matched to controls on age, sex and practice. Putative features of cancer were identified in the year before diagnosis. Odds ratios (ORs) were calculated for these features using conditional logistic regression, and positive predictive values (PPVs) were calculated.

Results:

A total of 7471 cases and 32 877 controls were studied. Sixteen features were independently associated with oesophago-gastric cancer (all P<0.001): dysphagia, OR 139 (95% confidence interval 112–173); reflux, 5.7 (4.8–6.8); abdominal pain, 2.6 (2.3–3.0); epigastric pain, 8.8 (7.0–11.0); dyspepsia, 6 (5.1–7.1); nausea and/or vomiting, 4.9 (4.0–6.0); constipation, 1.5 (1.2–1.7); chest pain, 1.6 (1.4–1.9); weight loss, 8.9 (7.1–11.2); thrombocytosis, 2.4 (2.0–2.9); low haemoglobin, 2.4 (2.1–2.7); low MCV, 5.2 (4.2–6.4); high inflammatory markers, 1.7 (1.4–2.0); raised hepatic enzymes, 1.3 (1.2–1.5); high white cell count, 1.4 (1.2–1.7); and high cholesterol, 0.8 (0.7–0.8). The only PPV >5% in patients ⩾55 years was for dysphagia. In patients <55 years, all PPVs were <1%.

Conclusion:

Symptoms of oesophago-gastric cancer reported in secondary care were also important in primary care. The results should inform guidance and commissioning policy for upper GI endoscopy.

The Aarhus statement: improving design and reporting of studies on early cancer diagnosis

Early diagnosis is a key factor in improving the outcomes of cancer patients. A greater understanding of the pre-diagnostic patient pathways is vital yet, at present, research in this field lacks consistent definitions and methods. As a consequence much early diagnosis research is difficult to interpret. A consensus group was formed with the aim of producing guidance and a checklist for early cancer-diagnosis researchers. A consensus conference approach combined with nominal group techniques was used. The work was supported by a systematic review of early diagnosis literature, focussing on existing instruments used to measure time points and intervals in early cancer-diagnosis research. A series of recommendations for definitions and methodological approaches is presented. This is complemented by a checklist that early diagnosis researchers can use when designing and conducting studies in this field. The Aarhus checklist is a resource for early cancer-diagnosis research that should promote greater precision and transparency in both definitions and methods. Further work will examine whether the checklist can be readily adopted by researchers, and feedback on the guidance will be used in future updates.

Features of childhood cancer in primary care: a population-based nested case-control study

BACKGROUND

This study investigated the risk of cancer in children with alert symptoms identified in current UK guidance, or with increased consultation frequency in primary care.

METHODS

A population-based, nested case-control study used data from the General Practice Research Database. In all, 1267 children age 0-14 years diagnosed with childhood cancer were matched to 15,318 controls. Likelihood ratios and positive predictive values (PPVs) were calculated to assess risk.

RESULTS

Alert symptoms recorded in the 12 and 3 months before diagnosis were present in 33.7% and 27.0% of cases vs 5.4% and 1.4% of controls, respectively. The PPV of having cancer for any alert symptom in the 3 months before diagnosis was 0.55 per 1000 children. Cases consulted more frequently particularly in the 3 months before diagnosis (86% cases vs 41% controls). Of these, 36% of cases and 9% of controls had consulted 4 times or more. The PPV for cancer in a child consulting 4 times or more in 3 months was 0.13 per 1000 children.

CONCLUSION

Alert symptoms and frequent consultations are associated with childhood cancer. However, individual symptoms and consultation patterns have very low PPVs for cancer in primary care (e.g., of 10,000 children with a recorded alert symptom, approximately 6 would be diagnosed with cancer within 3 months).

An estimation of the worldwide economic and health burden of visual impairment

This study aims to provide a rigorous estimate of the worldwide costs of visual impairment (VI), and the associated health burden. The study used a prevalence-based model. Prevalence rates for mild VI (visual acuity (VA) worse than 6/12 but not worse than 6/18), moderate VI (VA worse than 6/18 but not worse than 6/60) and blindness (VA worse than 6/60) were applied to population forecasts for each World Health Organisation (WHO) subregion. The limited available country cost data were extrapolated between subregions using economic and population health indicators. Age and gender subgroup population numbers were derived from United Nations' data. Costs and the health burden of VI were estimated for each world subregion using published disease prevalence rates, health care expenditures and other economic data. The study includes direct health care costs, indirect costs and the health burden of VI. The total cost of VI globally was estimated at $3 trillion in 2010, of which $2.3 trillion was direct health costs. This burden is projected to increase by approximately 20% by 2020. VI is associated with a considerable disease burden. Unless steps are taken to reduce prevalence through prevention and treatment, this burden will increase alongside global population growth.

Time to diagnosis and mortality in colorectal cancer: a cohort study in primary care

Background:

The relationship between the diagnostic interval and mortality from colorectal cancer (CRC) is unclear. This association was examined by taking account of important confounding factors at the time of first presentation of symptoms in primary care.

Methods:

A total of 268 patients with CRC were included in a prospective, population-based study in a Danish county. The diagnostic interval was defined as the time from first presentation of symptoms until diagnosis. We analysed patients separately according to the general practitioner's interpretation of symptoms. Logistic regression was used to estimate 3-year mortality odds ratios as a function of the diagnostic interval using restricted cubic splines and adjusting for tumour site, comorbidity, age, and sex.

Results:

In patients presenting with symptoms suggestive of cancer or any other serious illness, the risk of dying within 3 years decreased with diagnostic intervals up to 5 weeks and then increased (P=0.002). In patients presenting with vague symptoms, the association was reverse, although not statistically significant.

Conclusion:

Detecting cancer in primary care is two sided: aimed at expediting ill patients while preventing healthy people from going to hospital. This likely explains the counterintuitive findings; but it does not explain the increasing mortality with longer diagnostic intervals. Thus, this study provides evidence for the hypothesis that the length of the diagnostic interval affects mortality in CRC patients.

Pathways to the diagnosis of ovarian cancer in the UK: a cohort study in primary care

OBJECTIVE

To identify the routes patients with ovarian cancer take between first symptom presentation and diagnosis.

DESIGN

Cohort study.

SETTING

The study took place in 39 general practices in Devon, UK.

POPULATION

All ovarian cancer patients identified in the practices, with a diagnosis between 2000 and 2007 inclusive.

METHODS

All patients had their cancer symptoms, referrals, and diagnoses identified and dated using their doctors' records.

MAIN OUTCOME MEASURES

Numbers of patients taking specific routes to diagnosis, together with the time taken to diagnosis.

RESULTS

Three main routes to diagnosis emerged. The first was the expected route of outpatient referral: 195 (92% of the total) had at least one of the seven ovarian cancer symptoms or an abdominal mass. A total of 123 (58%) were referred to a specialist, although only 65 (31%) were referred to a gynaecologist. Thirty-five (17%) were initially investigated within primary care by ultrasound scanning, and a further 35 (17%) were admitted as emergencies. The interval from first symptom to referral was similar across the different pathways, with a median (interquartile range) time between the first symptom presenting to primary care and first investigation or referral being 2.5 (0, 27.5) days. The median interval from first symptom reported in primary care to diagnosis was 74.5 (32, 159) days.

CONCLUSIONS

Only a minority of ovarian cancer patients follow the expected route to diagnosis, of urgent referral to a gynaecologist. In most women, GPs rapidly identified the need to investigate. Avoidable delays generally occurred after the decision to investigate was made.

What happens to new-onset headache in children that present to primary care? A case-cohort study using electronic primary care records

The aim was to describe the consulting behaviour and clinical outcomes of children presenting with headache in primary care. This was a historical cohort study using data from the UK General Practitioner Research Database. Cases were children aged 5-17 years who presented to primary care with primary headache (migraine, tension-type headache, cluster headache) or undifferentiated headache (no further descriptor). Controls were age, sex and practice matched. Their records were examined for consultations, referrals, relevant treatments and specific diseases in the subsequent year. Children with headache (n = 48 575) were identified and matched to controls. At presentation, 9321 (19.2%) of headaches were labelled primary, 549 (1.1%) secondary and 38 705 (79.7%) received no formal diagnosis. Of the latter group, 2084 (5.4%) received a primary headache diagnosis in the subsequent year. Following a diagnosis of migraine, 258 (3.5%) had received a triptan and 1598 (21%) were using propranolol or pizotifen. Total consultations were higher in cases than in controls in the year before the headache: cases ages 5-8 years, mean (s.d.) 5.0 (4.0) consultations; controls 4.0 (3.5) consultations. In 1 year controls had 43 430 consultations, of which 256 (0.6%) were for headache, of whom 64 (25%) were referred to secondary care. Headache was a risk factor for benign and malignant tumours, cerebrovascular disease, primary disorders of raised intracranial pressure and depression. This risk was reduced if a diagnosis of a primary headache disorder could be made. Although there is an increased likelihood of a serious pathology with headache presentations, the risk is small particularly if a diagnosis of a primary headache is made. General practitioners are likely to be underdiagnosing migraine. This study can inform management guidelines for new presentations of headache in primary care, particularly when a secondary pathology is suspected.

The CAPER studies: five case-control studies aimed at identifying and quantifying the risk of cancer in symptomatic primary care patients

Background:

This paper reviews the background to five primary care case-control studies, collectively known as the CAPER studies (Cancer Prediction in Exeter). These studies, on colorectal, lung, prostate and brain tumours, sought to identify the particular features of cancer as reported to primary care. They also sought to quantify the risk of cancer for symptoms and primary care investigations, both individually and paired together.

Methods:

Two studies were on colorectal cancer: the former with 349 cases used hand searching and coding of entries, while the latter obtained 6442 cases from a national electronic database. The lung and prostate studies had 247 and 217 cases, respectively, and used manual methods. The brain study also used a national electronic database, which provided 3505 cases.

Results:

Generally, the symptoms matched previous series from secondary care, though the risks of cancer, expressed as positive predictive values, were lower. Rectal bleeding in colorectal cancer, and haemoptysis in lung cancer both had positive predictive values of 2.4%. The risk of a brain tumour with headache was one in a thousand.

Interpretation:

The results identify areas where current guidance on urgent referral for investigation of suspected cancer could be improved.

What happens to new-onset headache presented to primary care? A case-cohort study using electronic primary care records

In the UK, 4% of general practitioner consultations are for headache, yet the natural history of these presentations is unknown. The objective of this study was to describe the outcome of new headache presentations to the general practitioner. This was a prospective case-control study in adults over a period of 1 year using data from the General Practitioner Research Database, UK. Records of patients who presented with primary headache (migraine, tension-type headache, cluster headache) or undifferentiated headache (no further descriptor) were examined for the subsequent year for subarachnoid haemorrhage, primary brain tumour, benign space-occupying lesion, temporal arteritis, stroke and transient ischaemic attack. We identified 21,758 primary headaches and 63,921 undifferentiated headaches. The likelihood ratio was 29 (9.9, 92) for a subarachnoid haemorrhage after an undifferentiated headache and increased with age. The 1-year risk of a malignant brain tumour with new undifferentiated headache was 0.15%, rising to 0.28% above the age of 50 years. For primary headache the risk was 0.045%. The risk for a benign space-occupying lesion was 0.05% for an undifferentiated and 0.009% for a primary headache. The risk of temporal arteritis was the highest of the conditions studied, 0.66% in the undifferentiated and 0.18% in the primary headache group. Accepting the limitations of this approach, our data can inform management guidelines for new presentations of headache in primary care and confirm the need for follow-up, even if a primary headache diagnosis is made.

The risk of gallbladder cancer from polyps in a large multiethnic series

BACKGROUND

The aim of this study is assess whether patients with Indian ethnic background are at an increased risk of developing gallbladder cancer (GBC) if they have been diagnosed with ultrasonic abnormalities of the gallbladder.

METHODS

Between January 1998 and July 2006, 137,655 abdominal ultrasound examinations were performed in Leeds Teaching Hospitals NHS Trust. After the exclusion of repeat scans and those performed for renal or pelvic disease, 71,431 reports were included in this analysis. Patients in whom the diagnosis of GBC has been made without histology have been identified from the database of Northern and Yorkshire Cancer Registry and the presence of GBC was correlated with ultrasonic gallbladder abnormalities.

RESULTS

Gallbladder polyps (GBP) were detected in 3.3% of patients and these were larger than 10 mm in 0.1% of the cases. Age above 60 years, Indian ethnic background, single GBP larger than 10mm, the presence of gallstones, severe gallbladder wall thickening and irregular thickening were independently associated with the higher odds of developing GBC. The prevalence of malignancy in those with GBP was significantly higher among patients with Indian ethnic background compared to Caucasian patients, 5.5% versus 0.08%, p<0.001.

CONCLUSIONS

The presence of GBP, irrelevant of size, amongst patients of Indian ethnic decent, is an indication for further investigation and/or cholecystectomy.

The importance of anaemia in diagnosing colorectal cancer: a case-control study using electronic primary care records

Although anaemia is recognised as a feature of colorectal cancer, the precise risk is unknown. We performed a case–control study using electronic primary care records from the Health Improvement Network database, UK. A total of 6442 patients had a diagnosis of colorectal cancer, and were matched to 45 066 controls on age, sex, and practice. We calculated likelihood ratios and positive predictive values for colorectal cancer in both sexes across 1 g dl⁻¹ haemoglobin and 10-year age bands, and examined the features of iron deficiency.In men, 178 (5.2%) of 3421 cases and 47 (0.2%) of 23 928 controls had a haemoglobin <9.0 g dl⁻¹, giving a likelihood ratio (95% confidence interval) of 27 (19, 36). In women, the corresponding figures were 227 (7.5%) of 3021 cases and 58 (0.3%) of 21 138 controls, a likelihood ratio of 41 (30, 61). Positive predictive values increased with age and for each 1 g dl⁻¹ reduction in haemoglobin. The risk of cancer for current referral guidance was quantified. For men over 60 years with a haemoglobin <11 g dl⁻¹ and features of iron deficiency, the positive predictive value was 13.3% (9.7, 18) and for women with a haemoglobin <10 g dl⁻¹ and iron deficiency, the positive predictive value was 7.7% (5.7, 11). Current guidance for urgent investigation of anaemia misses some patients with a moderate risk of cancer, particularly men.

Offspring birthweight is not associated with paternal insulin resistance

AIMS/HYPOTHESIS

Low birthweight is associated with insulin resistance and other insulin resistance-related phenotypes: diabetes, hypertension, and vascular disease in later life. The underlying mechanism is unclear. The foetal insulin hypothesis proposes that a single genetic predisposition to beta cell dysfunction/insulin resistance results in both reduced insulin-dependent foetal growth in utero, hence low birthweight, and predisposition to type 2 diabetes. The aim of this study was to test whether, as predicted by the foetal insulin hypothesis, there is an association between measures of paternal insulin resistance and offspring birthweight.

SUBJECTS AND METHODS

The Exeter Family Study of Childhood Health (EFSOCH) is a community-based study within central Exeter (UK), established to test the foetal insulin hypothesis prospectively. Associations were tested between offspring birthweight and paternal insulin resistance, calculated by homeostasis model assessment analysis in 986 families using data relating to singleton, non-diabetic, UK white pregnancies. Ethics approval was given by the North and East Devon local ethics committee.

RESULTS

Offspring birthweight was not significantly correlated with log paternal insulin resistance (r=0, p=0.91), log HDL cholesterol concentration (r=-0.02, p=0.47) or log triglyceride concentration (r=0, p=0.99) when corrected for paternal BMI and common confounders. Multiple linear regression analysis confirmed that paternal insulin resistance was not an independent predictor of offspring birthweight.

CONCLUSIONS/INTERPRETATION

Results from a young, adult, non-diabetic population do not support the foetal insulin hypothesis as an explanation for the association of low birthweight with insulin resistance.

The mortality of colorectal cancer in relation to the initial symptom at presentation to primary care and to the duration of symptoms: a cohort study using medical records

The association between the staging of colorectal cancer and mortality is well known. Much less researched is the relationship between the duration of symptoms and outcome, and whether particular initial symptoms carry a different prognosis. We performed a cohort study of 349 patients with primary colorectal cancer in whom all their prediagnostic symptoms and investigation results were known. Survival data for 3–8 years after diagnosis were taken from the cancer registry. Six features were studied: rectal bleeding, abdominal pain, diarrhoea, constipation, weight loss, and anaemia. Two of these were significantly associated with different staging and mortality. Rectal bleeding as an initial symptom was associated with less advanced staging (odds ratio from one Duke's stage to the next 0.50, 95% confidence interval 0.31, 0.79; P=0.003) and with reduced mortality (Cox's proportional hazard ratio (HR) 0.56 (0.41, 0.79); P=0.001. Mild anaemia, with a haemoglobin of 10.0–12.9 g dl⁻¹, was associated with more advanced staging (odds ratio 2.2 (1.2, 4.3); P=0.021) and worse mortality (HR 1.5 (0.98, 2.3): P=0.064). When corrected for emergency admission, sex, and the site of the tumour, the HR for mild anaemia was 1.7 (1.1, 2.6); P=0.015. No relationship was found between the duration of symptoms and staging or mortality.

What are the clinical features of lung cancer before the diagnosis is made? A population based case-control study

BACKGROUND

Over 38,000 new cases of lung cancer occur each year in the UK. Most are diagnosed after initial presentation to primary care, but the relative importance of the various clinical features is largely unknown.

METHODS

A population based case-control study was undertaken in all 21 general practices in Exeter, Devon, UK (population 128 700). 247 primary lung cancers were studied in subjects aged over 40 years diagnosed between 1998 and 2002 and 1235 controls matched by age, sex and general practice. The entire primary care record for 2 years before diagnosis was coded using the International Classification of Primary Care-2. Univariable and multivariable conditional logistic regression analyses were used to identify and quantify clinical features independently associated with lung cancer. The main outcome measures were odds ratios and positive predictive values for these variables.

RESULTS

Seven symptoms (haemoptysis, loss of weight, loss of appetite, dyspnoea, thoracic pain, fatigue and cough), one physical sign (finger clubbing), and two abnormal investigation results (thrombocytosis and abnormal spirometry) were associated with lung cancer in multivariable analyses, as was cigarette smoking. After excluding variables reported in the final 180 days before diagnosis, haemoptysis, dyspnoea and abnormal spirometry remained independently associated with cancer.

CONCLUSIONS

This study provides an evidence base for selection of patients for investigation of possible lung cancer, both for clinicians and for developers of guidelines.

Clinical features of colorectal cancer before diagnosis: a population-based case-control study

Most colorectal cancers are diagnosed after the onset of symptoms. However, the risk of colorectal cancer posed by particular symptoms is largely unknown, especially in unselected populations like primary care. This was a population-based case–control study in all 21 general practices in Exeter, Devon, UK, aiming to identify and quantify the prediagnostic features of colorectal cancer. In total, 349 patients with colorectal cancer, aged 40 years or more, and 1744 controls, matched by age, sex and general practice, were studied. The full medical record for 2 years before diagnosis was coded using the International Classification of Primary Care-2. We calculated odds ratios for variables independently associated with cancer, using multivariable conditional logistic regressions, and then calculated the positive predictive values of these variables, both individually and in combination. In total, 10 features were associated with colorectal cancer before diagnosis. The positive predictive values (95% confidence interval) of these were rectal bleeding 2.4% (1.9, 3.2); weight loss 1.2% (0.91, 1.6); abdominal pain 1.1% (0.86, 1.3); diarrhoea 0.94% (0.73, 1.1); constipation 0.42% (0.34, 0.52); abnormal rectal examination 4.0% (2.4, 7.4); abdominal tenderness 1.1% (0.77, 1.5); haemoglobin <10.0 g dl⁻¹ 2.3% (1.6, 3.1); positive faecal occult bloods 7.1% (5.1, 10); blood glucose>10 mmol l⁻¹ 0.78% (0.51, 1.1): all P<0.001. Earlier diagnosis of colorectal cancer may be possible using the predictive values for single or multiple symptoms, physical signs or test results.