Six years' audit of laboratory workload and rates of referral for colposcopy in a cervical screening programme in three districts.

OBJECTIVE--To determine laboratory workload and rates of referral for colposcopy in a three district cervical screening programme during 1983-9 to assess the feasibility of accommodating call up of all women at risk, recall at three year intervals (now five year intervals), and investigation of women with all degrees of abnormality. DESIGN--Analysis of computerised screening histories dating back to 1977 of women screened in the Avon cervical screening programme. SETTING--Three district health authorities covering the population of Bristol and Weston-super-Mare, comprising 800,000 people, of whom 250,000 were female residents aged 20 to 64. SUBJECTS--196,977 Women aged 20 to 64 screened in cervical screening programme since 1983. RESULTS--Laboratory workload devoted to follow up of women with abnormalities increased sharply between 1987-8 and 1988-9, with increases of 54% (from 2075 to 3196) in the number of smears for follow up of severe dyskaryosis and invasive cancer, 40% (from 1925 to 2695) for mild and moderate dyskaryosis, and 49% (from 1793 to 2677) for borderline change. The increases were partly explained by the introduction in April 1988 of protocols for follow up and investigation based on guidance in an intercollegiate working party report. The proportion of women with mild and moderate dyskaryosis who were recommended for referral for colposcopy increased steadily from 9.9% in 1983-4 to 79.9% in 1988-9, and for borderline change the proportions were 3.5% and 13.6% respectively. Of all women tested in 1988-9, referral for colposcopy was recommended in 3%. CONCLUSIONS--The increase in laboratory follow up work identified, if it continued, could result in half of existing laboratory capacity in Avon being devoted to follow up work by 1993, with little prospect of maintaining call, recall, and quality control. Investigation of all women with minor cytological abnormalities is neither justifiable nor sustainable and will undermine the benefits of screening by increasing the rate of false positive results and the financial costs.     

Screening of vision in school: could we do better by doing less?

Assessment of vision in schoolchildren is routinely performed, but the effectiveness of the screening programmes has not been reviewed. A survey of health district screening programmes for vision in schools was performed at the end of 1984. The response rate from districts in England and Wales was 81%. All 165 of the districts that responded screened for loss of distant visual acuity; 96% screened for loss of colour vision, 73% for squint, and 67% for loss of near visual acuity. The frequency with which districts screened varied considerably. Some districts screened yearly, and various different types of tests were used. In many districts children were screened in unsuitable places, such as corridors, assembly halls, and toilets. Criteria for referral varied from one district to another, and few districts collected data appropriate for monitoring their screening programmes. Many districts screened more intensively than could be justified on the basis of the conditions tested for and the likely benefit of remedial treatment.     

Vision screening at 8 and 18 months. Steering Committee of Oxford Region Child Development Project.

OBJECTIVE--To determine the effectiveness of an existing screening programme based in the community for ocular and vision defects in infants considered at increased risk of such defects. DESIGN--Children with ocular or vision defect by the age of 2 were ascertained by searching records. Those from populations at high risk were matched with their results from screening tests. The characteristics of the cases among this population were compared with those of the cases in the remainder of the population. Patterns of referral and age at referral were studied in both groups. SETTING--The study was conducted within Oxfordshire Health District. SUBJECTS--433 Children at high risk born in 1984 to mothers living in the health district at delivery and who either weighed less than 2000 g or weighed 2000 g and over and required admission to a special care nursery for longer than 24 hours. The low risk population (6254) were infants without these characteristics who were resident in the health district at the time of referral. INTERVENTIONS--Screening tests for vision or ocular defects already routinely used were applied by health visitors at 8 and 18 months to the children at high risk. MAIN OUTCOME MEASURE--Comparison of results of screening tests with vision and ocular defects detected by the age of 2. RESULTS--Screening tests in current use for vision loss and squint in this age group were insensitive and had a low positive predictive value when applied to a high risk population. Defects that were not apparent on direct inspection were unlikely to be detected by these tests. In the high risk group the relative risk of having a defect was 2.8 (95% confidence interval 1.8 to 4.5) but 85% of all cases detected by the age of 2 were in children at low risk. Referral patterns and age of referral differed in the two groups. CONCLUSIONS--Screening by health visitors of high risk populations contributes little to the detection of vision and ocular defects. This type of evaluation needs to be applied also to low risk populations, who have different referral patterns and contribute most of the cases.     

Hospital work load produced by breast-cancer screening programme run by trained non-medical staff.

In a feasibility study of mass population screening for breast cancer by annual clinical examination and mammography the findings of non-medical staff (nurses and radiographers) were used to estimate the hospital work load generated by such a programme. Among 2490 women who attended for the first time by invitation the rate of referral for a surgical opinion based on the findings of the non-medical staff was 7.9% and the biopsy rate 2.5%. In the second and third years referral rates fell to 4.3% and 2.7% respectively and the biopsy rates to 1.1% and 1.4%. The rates of referral and biopsy among 1203 women who referred themselves for screening were higher, but many self-referred women were symptomatic; those without symptoms had rates of referral and biopsy similar to those of the invited women. Extrapolation of these findings to a population of 200,000 in a typical health district showed that the hospital work load would be high in the first year of screening with 44 outpatient referrals per week and 14 biopsies. By the third year, however, only seven referrals and four biopsies a week could be expected. The work load would be reduced by a third if screening were confined to women over the age of 50.     

Influence of Royal College of Radiologists' guidelines on referral from general practice. Royal College of Radiologists Working Party.

OBJECTIVE--To measure the effect on general practitioner referrals for radiography of introducing guidelines of good practice together with monitoring and peer review. DESIGN--Collection of referral data during 1 January 1989 to 31 December 1990. Guidelines were introduced on 1 January 1990. SETTING--Open access radiology services provided by one non-teaching district in England. SUBJECTS--144614 registered patients from 22 practices. MAIN OUTCOME MEASURES--Number of referrals per 1000 registered patients for radiography of the chest, skull, spine, abdomen, limbs, and joints and for barium investigation and excretion urography. RESULTS--Overall referrals fell from 88.4/1000 registered patients to 77.2/1000 after the guidelines were introduced. The commonest reasons for referral were for examination of the chest, spine, and limbs and joints and referrals for these fell by 9.4%, 17.5%, and 13.5% respectively. Referrals for skull radiography fell by 30% (from 241 to 168). CONCLUSIONS--By helping general practitioners to be more selective in their use of diagnostic radiology, the guidelines reduced the rate of referral and thus patients'' exposure to radiation.     

Referral patterns to an ophthalmic outpatient clinic by general practitioners and ophthalmic opticians and the role of these professionals in screening for ocular disease.

Case notes of 1113 consecutive new patients referred to a consultant ophthalmologist at a district general hospital were reviewed to determine the source and efficacy of referrals and the current screening practices of general practitioners and ophthalmic opticians. General practitioners initiated referral in 546 cases (49%) and ophthalmic opticians referral in 439 (39%). Visual loss or visual disturbance was the most important single reason for referral (345 cases; 31%), followed by suspected glaucoma (145 cases; 13%), abnormalities of binocular vision (140; 12.5%), disorders of eyelids or ocular adnexa (127; 11%), and red eye (86; 8%). General practitioners referred many more patients with disorders of the eyelids and adnexa and ophthalmic opticians many more patients with suspected glaucoma. Ophthalmic opticians were far more likely than general practitioners to refer patients with suspected glaucoma correctly. A total of 180 patients (16%) were referred from ocular screening, in 149 cases by ophthalmic opticians and in 10 by general practitioners. Seventy patients had glaucoma or incomplete features of glaucoma, all of them referred by ophthalmic opticians. Of eight diabetic patients referred by ophthalmic opticians, three had asymptomatic disease and in two diabetes was diagnosed as a result of ocular screening. No patient was referred for asymptomatic diabetic retinopathy from screening by general practitioners. Ophthalmic opticians were more likely than general practitioners to diagnose retinopathy requiring photocoagulation. Use of a community based service to screen for glaucoma could save unnecessary consultant outpatient appointments. A similar service could facilitate detection of diabetic retinopathy at a stage when treatment is most effective.     

National Health Service breast screening programme results for 1991-2.

OBJECTIVES--To report the results of the NHS breast screening programme for the year March 1991 to April 1992. DESIGN--A report of statistics was derived from Körner (K62) returns and from the radiology quality assurance programme. MAIN OUTCOME MEASURES--Detection rates for breast cancer and small (< or = 10 mm diameter) invasive cancer, benign biopsy rates, and recall and acceptance rates. RESULTS--The acceptance rate for screening across the United Kingdom was 71.3%. The referral rate for further investigation was 6.2% (regional 4.3-9.0%). The breast cancer detection rate was 6.2 cancers per 1000 women screened (5.1-9.0) and the detection rate of invasive cancers < or = 10 mm was 1.4/1000 (1.0-2.3). 72% of screening programmes reached the target 70% acceptance rate, and 95% of programmes achieved a recall rate of less than 10%. 75% of programmes had a cancer detection rate of more than 5/1000, but only 32% had a detection rate for invasive cancers < or = 10 mm of more than 1.5/1000. CONCLUSIONS--Overall, the results of the screening programme for the year 1991-2 can be regarded as extremely satisfactory, given the size and complexity of the operation.     

The development of the referral outcome diagram and an analysis of laboratory cancer detection rates in the English NHS cervical screening programme – is there an optimum level of detection of CIN 1 and CIN 2 lesions?

Objective: To use routine annual data from the English cervical screening laboratories (KC61 returns) to evaluate individual laboratory return characteristics with particular reference to factors associated with sensitivity and specificity. Methods: A graphical technique has been developed using data on referral to colposcopy and histological outcomes called a referral outcome (ROUT) diagram. The average grade of cervical intraepithelial neoplasia (CIN) detected (the mean CIN score, MCS) is plotted against the odds of a false‐positive referral. Further analysis has been conducted to examine the relationship between the MCS and screen‐detected invasive cancer rate. Results: There are large variations in ROUT diagram positions of individual laboratories and the diagram can be used to identify laboratories for further investigation. These variations are strongly influenced by substantial differences in the rate of low‐grade referrals and the MCS (and positive predictive value) are inversely related to the referral rate for low‐grade cytology (P < 0.001). There is a strong association between high MCS values and increased screen‐detected cancer rates (P < 0.001) particularly above an MCS of 2.2. The data can be re‐formulated in terms of CIN 2 and CIN 3 only where it can be shown that the invasive cancer rate rapidly increases if the numbers of CIN 2 lesions detected drops below 50% of the number of CIN 3 lesions. Given the complexity of cervical screening this may best be viewed as a hypothesis generating observation, best tested by interventional studies. Conclusions: The ROUT diagram represents a new and potentially interesting way of presenting annual return data. The national programme in England needs to balance the prevention of cancer against too many unnecessary referrals to colposcopy and the ROUT diagram, and associated data given in this paper may help toward this. Further research is required including examining the role of referral policy and threshold criteria in influencing low‐grade referrals and the relationship between MCS and cancer detection rate.     

Evaluating perinatal mortality rates: effects of referral and case mix.

OBJECTIVE--To evaluate perinatal mortality rates as a method of auditing obstetric and neonatal care after account had been taken of transfer between hospitals during pregnancy and case mix. DESIGN--Case-control study of perinatal deaths. SETTING--Leicestershire health district. SUBJECTS--1179 singleton perinatal deaths and their selected live born controls among 114,362 singleton births to women whose place of residence was Leicestershire during 1978-87. MAIN OUTCOME MEASURE--Crude perinatal mortality rates and rates adjusted for case mix. RESULTS--An estimated 11,701 of the 28,750 women booked for delivery in general practitioner maternity units were transferred to consultant units during their pregnancy. These 11,701 women had a high perinatal mortality rate (16.8/1000 deliveries). Perinatal mortality rates by place of booking showed little difference between general practitioner units (8.8/1000) and consultant units (9.3-11.7/1000). Perinatal mortality rates by place of delivery, however, showed substantial differences between general practitioner units (3.3/1000) and consultant units (9.4-12.6/1000) because of the selective referral of high risk women from general practitioner units to consultant units. Adjustment for risk factors made little difference to the rates except when the subset of deaths due to immaturity was adjusted for birth weight. CONCLUSION--Perinatal mortality rates should be adjusted for case mix and referral patterns to get a meaningful result. Even when this is done it is difficult to compare the effectiveness of hospital units with perinatal mortality rates because of the increasingly small subset of perinatal deaths that are amenable to medical intervention.     

Role of non-medical staff in screening for breast cancer.

In a feasibility study of mass population screening for breast cancer by clinical examination and mammography the ability of non-medical staff (nurses and radiographers) to act as primary screeners was compared with that of medical staff (surgeons and radiologists). In 240 women with cancer the rate of detection of the disease by the non-medical staff was comparable to that of the medical staff, although the non-medical staff detected more cancers by mammography alone than did the medical staff. The rate of detection by surgeons and particularly nurses was lower in women without symptoms than in those with symptoms, whereas the incidence of detection by radiologists and radiographers was lower in women aged under 50 than in those over 50. The rate of detection by all groups of staff significantly increased with increases in tumour size. The results suggest that non-medical staff can act effectively as primary screeners, but that for the detection of cancer in asymptomatic women, particularly those over 50, mammography is probably more effective than clinical examination.     

A Mixed Methods Study of a Health Worker Training Intervention to Increase Syndromic Referral for Gambiense Human African Trypanosomiasis in South Sudan

Background

Active screening by mobile teams is considered the most effective method for detecting gambiense-type human African trypanosomiasis (HAT) but constrained funding in many post-conflict countries limits this approach. Non-specialist health care workers (HCWs) in peripheral health facilities could be trained to identify potential cases for testing based on symptoms. We tested a training intervention for HCWs in peripheral facilities in Nimule, South Sudan to increase knowledge of HAT symptomatology and the rate of syndromic referrals to a central screening and treatment centre.

Methodology/Principal Findings

We trained 108 HCWs from 61/74 of the public, private and military peripheral health facilities in the county during six one-day workshops and assessed behaviour change using quantitative and qualitative methods. In four months prior to training, only 2/562 people passively screened for HAT were referred from a peripheral HCW (0 cases detected) compared to 13/352 (2 cases detected) in the four months after, a 6.5-fold increase in the referral rate observed by the hospital. Modest increases in absolute referrals received, however, concealed higher levels of referral activity in the periphery. HCWs in 71.4% of facilities followed-up had made referrals, incorporating new and pre-existing ideas about HAT case detection into referral practice. HCW knowledge scores of HAT symptoms improved across all demographic sub-groups. Of 71 HAT referrals made, two-thirds were from new referrers. Only 11 patients completed the referral, largely because of difficulties patients in remote areas faced accessing transportation.

Conclusions/Significance

The training increased knowledge and this led to more widespread appropriate HAT referrals from a low base. Many referrals were not completed, however. Increasing access to screening and/or diagnostic tests in the periphery will be needed for greater impact on case-detection in this context. These data suggest it may be possible for peripheral HCWs to target the use of rapid diagnostic tests for HAT.     

Effect of NHS reforms on general practitioners' referral patterns.

OBJECTIVE--To compare outpatient referral patterns in fundholding and non-fundholding practices before and after the implementation of the NHS reforms in April 1991. DESIGN--Prospective collection of data on general practitioners'' referrals to specialist outpatient clinics between June 1990 and March 1992 and detailed comparison of two time periods: October 1990 to March 1991 (phase 1) and October 1991 to March 1992 (phase 2). SETTING--10 fundholding practices and six non-fundholding practices in the Oxford region. SUBJECTS--Patients referred to consultant outpatient clinics. RESULTS--After implementation of the NHS reforms there was no change in the proportion of referrals from the two groups of practices which crossed district boundaries. Both groups of practices increased their referral rates in phase 2 of the study, the fundholders from 107.3 per 1000 patients per annum (95% confidence interval 106 to 109) to 111.4 (110 to 113) and the non-fundholders from 95.0 (93 to 97) to 112.0 (110 to 114). In phase 2 there was no difference in overall standardised referral rates between fundholders and non-fundholders. Just over 20% of referrals went to private clinics in phase 1. By phase 2 this proportion had reduced by 2.2% (1.0% to 3.4%) among the fundholders and by 2.7% (1.2% to 4.2%) among the non-fundholders. CONCLUSIONS--Referral patterns among fundholders and non-fundholders were strikingly similar after the implementation of the NHS reforms. There was no evidence that fundholding was encouraging a shift from specialist to general practice care or that budgetary pressures were affecting general practitioners'' referral behaviour.     

The effect of deprivation on variations in general practitioners' referral rates: a cross sectional study of computerised data on new medical and surgical outpatient referrals in Nottinghamshire.

OBJECTIVE: To determine the effect of deprivation on variations in general practitioners'' referral rates using the Jarman underprivileged area (UPA(8)) score as a proxy measure. DESIGN: Cross sectional survey of new medical and surgical referrals from general practices to hospitals (determined from hospital activity data). SETTING: All of the 183 general practices in Nottinghamshire and all of the 19 hospitals in Trent region. MAIN OUTCOME MEASURES: The relation between the referral rates per 1000 registered patients and the practice population''s UPA(8) score (calculated on the basis of electoral ward), with adjustment for the number of partners, percentage of patients aged over 65 years, and fundholding status of each practice. RESULTS: There was a significant independent association between deprivation, as measured by the UPA(8) score, and high total referral rates and high medical referral rates (P < 0.0001). The UPA(8) score alone explained 23% of the total variation in total referral rates and 32% of the variation in medical referral rates. On multivariate analysis, where partnership size, fundholding status, and percentage of men and women aged over 65 years were included, the UPA(8) score explained 29% and 35% of the variation in total and medical referral rates respectively. CONCLUSION: Of the variables studied, the UPA(8) score was the strongest predictor of variations in referral rates. This association is most likely to be through a link with morbidity, although it could reflect differences in patients'' perceptions, doctors'' behaviour, or the use and provision of services.     

Understanding variation in rates of referral among general practitioners: are inappropriate referrals important and would guidelines help to reduce rates?

OBJECTIVES--To determine the extent to which variation in rates of referral among general practitioners may be explained by inappropriate referrals and to estimate the effect of implementing referral guidelines. SETTING--Practices within Cambridge Health Authority and Addenbrooke''s Hospital, Cambridge. MAIN OUTCOME MEASURES--Data on practice referral rates from hospital computers, inappropriate referrals as judged by hospital consultants, and inappropriate referrals as judged against referral guidelines which had been developed locally between general practitioners and specialists. Effect of referral guidelines on referral patterns as judged by general practitioners using the guidelines in clinical practice. RESULTS--There was 2.5-fold variation in referral rates among general practices. According to the specialists, 9.6% (95% confidence interval 6.4% to 12.9%) of referrals by general practitioners and 8.9% (2.6% to 15.2%) of referrals from other specialists were judged possibly or definitely inappropriate. Against locally determined referral guidelines 15.9% of referrals by general practitioners were judged possibly inappropriate (11.8% to 20.0%). Elimination of all possibly inappropriate referrals could reduce variation in practice referral rates only from 2.5-fold to 2.1-fold. An estimate of the effect of using referral guidelines for 60 common conditions in routine general practice suggested that application of guidelines would have been unlikely to reduce rates of referral in hospital (95% confidence interval -4.5% to 8.6% of consultations resulting in referral). CONCLUSION--The variation in referral rates among general practitioners in Cambridge could not be explained by inappropriate referrals. Application of referral guidelines would be unlikely to reduce the number of patients referred to hospital.     

Balancing sensitivity and specificity: sixteen year's of experience from the mammography screening programme in Copenhagen, Denmark

Aim: To report on sensitivity and specificity from 7 invitation rounds of the organised, population-based mammography screening programme started in Copenhagen, Denmark, in 1991, and offered biennially to women aged 50–69. Changes over time were related to organisation and technology. Methods: Individualized data were retrieved on outcome of screening mammography, assessment, surgery, and interval cancers. European Guideline performance indicators were calculated, supplemented with false positive and interval cancer rates per 1000 screens. False positive tests were divided into those sorted out at assessment (Type 1) and at surgery (Type 2). Results: In total, 1392 invasive breast cancers/ductal carcinoma in situ cases (DCIS) were diagnosed, giving an overall detection rate of 7.6 per 1000 screens. Of 5178 false positive tests, 4666 were Type 1 and 512 Type 2. The 468 interval cancers constituted 25% of all breast cancers (=screen detected + interval cancer). Almost all outcome measures were well within the desirable level of the European Guidelines. Risk of Type 2 false positive tests was positively associated with detection rate especially at initial screen, and interval cancer rate was negatively associated with detection rate. This association was decoupled after introduction of high resolution ultrasound and stereotactic breast biopsies, resulting in a Benign-to-Malignant-Ratio (BMR) of 1:11.40. Conclusion: Mammography screening is a delicate balance between benefits and risks. Increase in detection rate came at cost of increase in risk of benign biopsies. Introduction of new technologies broke this pattern and a slight increase in detection rate coincided with an unprecedentedly low BMR.     

Genome-wide detection of additional fetal chromosomal abnormalities by cell-free DNA testing of 15,626 consecutive pregnant women

Cell-free DNA(cfDNA) testing for common fetal trisomies(T21, T18, T13) is highly effective. However, the usefulness of cfDNA testing in detecting other chromosomal abnormalities is unclear. We evaluated the performance of cfDNA testing for genome-wide abnormalities, and analyzed the incremental yield by reporting extra abnormalities. We performed genome-wide cfDNA testing in 15,626 consecutive pregnancies prospectively enrolled in this study. cfDNA testing results were reported and counseling was given depending on the presence of extra chromosomal abnormalities. cfDNA testing identified 190 cases(1.2%)of chromosomal abnormalities including 100 common trisomies and 90 additional abnormalities. By expanding the cfDNA reporting range to genome-wide abnormalities, the false positive rate increased to 0.39%(P0.001) and positive predictive value(PPV) was reduced to 65.58%(P=0.42). However, the detection yield increased from 0.44% to 0.65%(P=0.014), and cfDNA testing detected 38.61%(39/101) additional abnormalities with no ultrasound and biochemical screening findings. cfDNA testing outperformed biochemical screening by showing 60 times higher true positive rate and fewer false negative results.Genome-wide cfDNA testing significantly increased the diagnostic yield by detecting extra abnormalities, especially those without diagnostic indications. Genome-wide cfDNA testing has fewer false positive and false negative results compared with biochemical screening.     

Deficient colour vision and interpretation of histopathology slides: cross sectional study.

OBJECTIVE: To determine whether histopathologists with deficient colour vision make more errors in slide interpretation than those with normal colour vision. DESIGN: Examination of projected transparencies of histopathological slides under standardised conditions by subjects whose colour discriminating ability was accurately assessed. SETTING: Departments of histopathology in 45 hospitals in the United Kingdom. SUBJECTS: 270 male histopathologists and medical laboratory scientific officers. MAIN OUTCOME MEASURES: Number of slides correctly identified by subjects whose colour vision was measured on the Ishihara, City University, and Farnsworth-Munsell 100 hue tests. RESULTS: Mean (SD) scores (out of 10) for doctors with colour deficient vision were 9.4 (0.7) v 9.9 (0.4) for controls (P < 0.01) and 7.5 (1.6) v 9.4 (0.7) for scientific officers (P < 0.001). When subjects with colour deficient vision were categorised into severe, moderate, or mild, there was a significant trend towards those with severe deficiency making more mistakes (P < 0.001). CONCLUSIONS: Histopathologists and medical laboratory scientific officers should have their colour vision tested; if they are found to have a severe protan or deutan deficiency, they should be advised to adopt a safe system of working.     

Influence of the Royal College of Radiologists' guidelines on hospital practice: a multicentre study. Royal College of Radiologists Working Party.

OBJECTIVE--To measure the effect on hospital radiology referral practice of introducing a strategy for change involving guidelines of good practice, monitoring, and peer review. DESIGN--Prospective data collection over a continuous 21-24 month period at each centre some time between January 1987 and December 1990. SETTING--Five district general hospitals and one district health authority. SUBJECTS--314,663 inpatient discharges, deaths, and day cases and 1,706,781 outpatient attendances under the care of 722 consultants from 25 clinical specialties. MAIN OUTCOME MEASURES--Number of referrals for x ray examination per 100 inpatient discharges, deaths, and day cases and per 100 new outpatient attenders. RESULTS--Most doctors were prepared to accept standards of clinical practice set by peers and also the monitoring and review of their practice with respect to these standards by local colleagues. 18% of firms were identified before guidelines were instituted as having persistently high referral rates. Appreciable, and often dramatic reductions in referral rates for individual x ray examinations were recorded by a substantial number of firms in every centre and in every specialty after guidelines were instituted. The major part of this reduction was achieved by some of the firms whose initial practice did not meet "high referral" criteria. Important variations in compliance with agreed standards of good practice were observed. CONCLUSIONS--The study offers strong experimental evidence to support a recent suggestion that at least a fifth of radiological examinations carried out in NHS hospitals are clinically unhelpful. The problem of how to assure compliance with agreed standards of practice needs to be resolved. Until this happens medical audit alone is unlikely to translate good practice into common practice.     

Posters. P11 Positive predictive values for high and low‐grade cytological abnormalities as surrogates for specificity and sensitivity

Introduction Positive predictive value (PPV), measuring the percentage of moderate dyskaryosis or worse confirmed as CIN2 or worse, is used as a measure of accuracy in cervical screening. However, it relates more to specificity than sensitivity because the denominator includes false positives rather than false negatives. Low values reflect over‐reporting of high‐grade dyskaryosis but high values may reflect under‐reporting. Sensitivity is impossible to measure from correlation of cytology with outcome because women with negative cytology are rarely referred for colposcopy. Rates of CIN3 resulting from referrals for low‐grade cytology may be used as a surrogate for sensitivity, as high values may reflect under‐reporting (ref). Study design Outcome of colposcopy referrals was monitored during a period of 4 years, using a fail‐safe database. Results PPV at Guy's & St Thomas rose from 54% in 1998/1999 to 69% in 2001/2002. The former was below the NHSCSP recommended range. During the same period of time CIN1 rates for moderate dyskaryosis fell from 37% to 24%, reflecting the main source of discrepancy. While specificity increased (as reflected by increasing PPV) sensitivity remained constant in that CIN3 rates for mild dyskaryosis and borderline remained below 6%: average rates in England have fallen over the last 3 years and were 7.4% in 2000/2001 (ref). CIN2 rates for mild dyskaryosis also remained constant at 11% to 12%. Conclusion Correlation of biopsy results with high‐ and low‐grade cytological abnormalities is a useful method of monitoring accuracy of cytology reporting, and can be used to measure over‐ and under‐reporting as surrogates for specificity and sensitivity.