Monitoring cholesterol levels: measurement error or true change?

BACKGROUND

Cholesterol level monitoring is a common clinical activity, but the optimal monitoring interval is unknown and practice varies.

OBJECTIVE

To estimate, in patients receiving cholesterol-lowering medication, the variation in initial response to treatment, the long-term drift from initial response, and the detectability of long-term changes in on-treatment cholesterol level ("signal") given short-term, within-person variation ("noise").

DESIGN

Analysis of cholesterol measurement data in the LIPID (Long-Term Intervention with Pravastatin in Ischaemic Disease) study.

SETTING

Randomized, placebo-controlled trial in Australia and New Zealand (June 1990 to May 1997).

PATIENTS

9014 patients with past coronary heart disease who were randomly assigned to receive pravastatin or placebo.

MEASUREMENTS

Serial cholesterol concentrations at randomization, 6 months, and 12 months, and then annually to 5 years.

RESULTS

Both the placebo and pravastatin groups showed small increases in within-person variability over time. The estimated within-person SD increased from 0.40 mmol/L (15 mg/dL) (coefficient of variation, 7%) to 0.60 mmol/L (23 mg/dL) (coefficient of variation, 11%), but it took almost 4 years for the long-term variation to exceed the short-term variation. This slow increase in variation and the modest increase in mean cholesterol level, about 2% per year, suggest that most of the variation in the study is due to short-term biological and analytic variability. Our calculations suggest that, for patients with levels that are 0.5 mmol/L or more (> or =19 mg/dL) under target, monitoring is likely to detect many more false-positive results than true-positive results for at least the first 3 years after treatment has commenced.

LIMITATIONS

Patients may respond differently to agents other than pravastatin. Future values for nonadherent patients were imputed.

CONCLUSION

The signal-noise ratio in cholesterol level monitoring is weak. The signal of a small increase in cholesterol level is difficult to detect against the background of a short-term variability of 7%. In annual rechecks in adherent patients, many apparent increases in cholesterol level may be false positive. Independent of the office visit schedule, the interval for monitoring patients who are receiving stable cholesterol-lowering treatment could be lengthened.

HPV triage testing or repeat Pap smear for the management of atypical squamous cells (ASCUS) on Pap smear: is there evidence of process utility?

A two-stage standard gamble was used to evaluate women's preferences for alternative managements of atypical squamous cells of undermined significance (ASCUS) on Pap smear (repeat Pap smear compared with immediate HPV test), and to test for the evidence of process utility. Women's utilities for the health state scenarios were clustered towards the upper end of the 0-1 scale with considerable variability in women's preferences. There was evidence of process utility, with immediate human papillomavirus (HPV) testing strategies having lower valuations than repeat Pap smear, where the clinical outcome was the same. Mean (95% CI) utilities for HPV testing (negative test) followed by resolution were 0.9967 (0.9957-0.9978) compared with repeat Pap smear followed by resolution: 0.9972 (0.9964-0.9980). Mean (95% CI) utilities for immediate HPV testing (positive test), followed by colposcopy, biopsy and treatment were 0.9354 (0.8544-1.0) compared with repeat Pap smear followed by colposcopy, biopsy and treatment: 0.9656 (0.9081-1.0). Our results add to the existing evidence that the impact of healthcare interventions on well-being is not limited to the effect of the intervention on the health outcomes expected from the intervention; process of care can have quality of life implications for the individual. A modelled application of trial-based data will allow characterisation of the true population costs, benefits, risks and harms of alternative triage strategies and subsequent policy implications thereof.

A self administered reliable questionnaire to assess lower bowel symptoms

Background

Bowel symptoms are considered indicators of the presence of colorectal cancer and other bowel diseases. Self administered questionnaires that elicit information about lower bowel symptoms have not been assessed for reliability, although this has been done for upper bowel symptoms. Our aim was to develop a self administered questionnaire for eliciting the presence, nature and severity of lower bowel symptoms potentially related to colorectal cancer, and assess its reliability.

Methods

Immediately before consulting a gastroenterologist or colorectal surgeon, 263 patients likely to have a colonoscopy completed the questionnaire. Reliability was assessed in two ways: by assessing agreement between patient responses and (a) responses given by the doctor at the consultation; and (b) responses given by patients two weeks later.

Results

There was more than 75% agreement for 78% of the questions for the patient-doctor comparison and for 92% of the questions for the patient-patient comparison. Agreement for the length of time a symptom was present, its severity, duration, frequency of occurrence and whether or not medical consultation had been sought, all had agreement of greater than 70%. Over all questions, the chance corrected agreement for the patient-doctor comparison had a median kappa of 65% (which represents substantial agreement), interquartile range 57–72%. The patient-patient comparison also showed substantial agreement with a median kappa of 75%, interquartile range 68–81%.

Conclusion

This self administered questionnaire about lower bowel symptoms is a useful way of eliciting details of bowel symptoms. It is a reliable instrument that is acceptable to patients and easily completed. Its use could guide the clinical consultation, allowing a more efficient, comprehensive and useful interaction, ensuring that all symptoms are assessed. It will also be a useful tool in research studies on bowel symptoms and their predictive value for colorectal cancer and other diseases. Studies assessing whether bowel symptoms predict the presence of colorectal cancer should provide estimates of the reliability of the symptom elicitation.

Use of randomised trials to decide when to monitor response to new treatment

Is monitoring initial response to treatment always helpful in clinical management of patients? Bell and colleagues have developed a framework for deciding whether surrogate outcomes should be used to monitor initial response to treatment in chronic disease.

Effects of study methods and biases on estimates of invasive breast cancer overdetection with mammography screening: a systematic review

Estimates of breast-cancer overdetection, the detection with screening of cancer that would not have presented clinically during the woman's lifetime (and therefore would not be diagnosed in the absence of screening), vary widely. We systematically reviewed estimates of overdetection to assess the extent to which these might be biased by study methods. Primary research papers and reviews that estimated overdetection of invasive breast cancer were eligible for inclusion. For each paper we appraised the study design and methods to identify the extent and effect of bias. Two reviews and six primary studies were included. We categorised studies as being based on cumulative-incidence or incidence-rate methods. The least biased overdetection estimates range from -4% to 7.1% for women aged 40-49 years, 1.7% to 54% for women aged 50-59 years, and 7% to 21% for women aged 60-69 years. Studies consistently show that cancer overdetection occurs in screening for breast cancer; however, reported estimates are biased. Sensitivity of mammography for both cancers that will progress and for overdetected cancers may be increasing with time. New studies are urgently needed to quantify the true extent of overdetection in current mammography screening programmes. These studies should be designed to avoid the multiple sources of bias identified in this review.

GPs want tailored, user friendly evidence summaries--a cross sectional study in New South Wales

BACKGROUND

This study aimed to measure the use of, and barriers to, using evidence among general practitioners since computerisation of general practice; GP preference for patient involvement in health care decisions; and GPs' preferred strategies to increase the use of evidence.

METHODS

A cross sectional, open ended telephone survey was conducted with 107 (out of 155) New South Wales GPs randomly selected from the New South Wales Medical Board register. The survey sought self report to open ended questions about information sources informing decisions, perceived barriers to using evidence, and suggested strategies to improve clinical decisions, plus Degner scale for patient involvement.

RESULTS

Evidence based sources remained the least likely to be used for informing decisions about patient care (23.4%). Opinion based sources were most commonly used (50.5%), with industry sponsored sources second (27.1%). Rural GPs were more likely to use opinion based sources (OR=1.55, 95% CI: 1.00-2.40). The most common perceived barriers were 'a lack of time' (22.0%), 'a lack of evidence or conflicting evidence' (13.1%), 'not knowing where to look' (10.3%) and 'not being able to tailor evidence to individual patients' (9.3%). The majority of GPs (72.0%) preferred patients to have some role in decision making. The most common suggestions for improving decision making were 'simply formatted evidence summaries' (28.0%) and 'mechanisms for tailoring evidence with individual patients' (13.1%).

DISCUSSION

The use of evidence based sources for clinical decision making in general practice remains limited. Potential strategies to overcome this should focus on providing more user friendly evidence summaries, involving patients in evidence based decision making, and finding mechanisms to tailor evidence to individual patients.

Observed Differences in Diagnostic Test Accuracy between Patient Subgroups: Is It Real or Due to Reference Standard Misclassification?

Before a new test is introduced in clinical practice, its accuracy should be assessed. In the past decade, researchers have put an increased emphasis on exploring differences in test sensitivity and specificity between patient subgroups. If the reference standard is imperfect and the prevalence of the target condition differs among subgroups, apparent differences in test sensitivity and specificity between subgroups may be caused by reference standard misclassification. We provide guidance on how to determine whether observed differences may be explained by reference standard misclassification. Such misclassification may be ascertained by examining how the apparent sensitivity and specificity change with the prevalence of the target condition in the subgroups.

Patient Decision Aids to Support Clinical Decision Making: Evaluating the Decision or the Outcomes of the Decision

Decision aids (DAs) are tools to support patients make informed health decisions with their practitioner. They aim to improve patient knowledge of options, incorporate patient preferences and values, and increase patient involvement in health decision making. Increasingly, the debate about DAs concerns how they should be implemented in practice, with the view that DAs are superior to usual clinical care in facilitating health decisions. The authors challenge this view and suggest that DA research has focused on measures of decision process, leaving the effects on the outcome of the decision relatively unknown. It is still unclear in which conditions DAs are better for patient health and well-being than clinician-led decisions. The authors present a new randomized design to examine the effects of DA-supported patient choice on patient-centered outcomes to identify where DAs are best implemented in clinical practice. In this design, patients are randomized to 1 of 4 arms: intervention A, intervention B, choice of either intervention supported by a clinician, or choice of either intervention supported by a decision aid. Health and quality of life measured over the long term are presented as the primary outcomes. The authors propose that this design will allow the proper assessment of different modes of decision making.

Three methods to construct predictive models using logistic regression and likelihood ratios to facilitate adjustment for pretest probability give similar results

OBJECTIVE

To compare three predictive models based on logistic regression to estimate adjusted likelihood ratios allowing for interdependency between diagnostic variables (tests).

STUDY DESIGN AND SETTING

This study was a review of the theoretical basis, assumptions, and limitations of published models; and a statistical extension of methods and application to a case study of the diagnosis of obstructive airways disease based on history and clinical examination.

RESULTS

Albert's method includes an offset term to estimate an adjusted likelihood ratio for combinations of tests. Spiegelhalter and Knill-Jones method uses the unadjusted likelihood ratio for each test as a predictor and computes shrinkage factors to allow for interdependence. Knottnerus' method differs from the other methods because it requires sequencing of tests, which limits its application to situations where there are few tests and substantial data. Although parameter estimates differed between the models, predicted "posttest" probabilities were generally similar.

CONCLUSION

Construction of predictive models using logistic regression is preferred to the independence Bayes' approach when it is important to adjust for dependency of tests errors. Methods to estimate adjusted likelihood ratios from predictive models should be considered in preference to a standard logistic regression model to facilitate ease of interpretation and application. Albert's method provides the most straightforward approach.

Accuracy of reading liquid based cytology slides using the ThinPrep Imager compared with conventional cytology: prospective study

OBJECTIVE

To compare the accuracy of liquid based cytology using the computerised ThinPrep Imager with that of manually read conventional cytology.

DESIGN

Prospective study.

SETTING

Pathology laboratory in Sydney, Australia.

PARTICIPANTS

55,164 split sample pairs (liquid based sample collected after conventional sample from one collection) from consecutive samples of women choosing both types of cytology and whose specimens were examined between August 2004 and June 2005.

MAIN OUTCOME MEASURES

Primary outcome was accuracy of slides for detecting squamous lesions. Secondary outcomes were rate of unsatisfactory slides, distribution of squamous cytological classifications, and accuracy of detecting glandular lesions.

RESULTS

Fewer unsatisfactory slides were found for imager read cytology than for conventional cytology (1.8% v 3.1%; P<0.001). More slides were classified as abnormal by imager read cytology (7.4% v 6.0% overall and 2.8% v 2.2% for cervical intraepithelial neoplasia of grade 1 or higher). Among 550 patients in whom imager read cytology was cervical intraepithelial neoplasia grade 1 or higher and conventional cytology was less severe than grade 1, 133 of 380 biopsy samples taken were high grade histology. Among 294 patients in whom imager read cytology was less severe than cervical intraepithelial neoplasia grade 1 and conventional cytology was grade 1 or higher, 62 of 210 biopsy samples taken were high grade histology. Imager read cytology therefore detected 71 more cases of high grade histology than did conventional cytology, resulting from 170 more biopsies. Similar results were found when one pathologist reread the slides, masked to cytology results.

CONCLUSION

The ThinPrep Imager detects 1.29 more cases of histological high grade squamous disease per 1000 women screened than conventional cytology, with cervical intraepithelial neoplasia grade 1 as the threshold for referral to colposcopy. More imager read slides than conventional slides were satisfactory for examination and more contained low grade cytological abnormalities.

Lack of sphygmomanometer calibration causes over- and under-detection of hypertension: a computer simulation study

OBJECTIVE

To estimate the contribution of inadequate sphygmomanometer calibration to over- and under-detection of hypertension.

DESIGN

Monte Carlo simulation of the measurement of blood pressure (BP) of a population with calibrated and uncalibrated sphygmomanometers. Simulated BP measurements included systematic sphygmomanometer error and random variability.

MAIN OUTCOME MEASURES

The percentage of hypertensive adults (BP > 140/90 mmHg) not detected and the percentage of adults incorrectly classified hypertensive due to sphygmomanometer error. The percentage of the false positives and false negatives attributable to sphygmomanometer error. The number of additional visits patients need to make to obtain the same improvement in hypertension detection as is obtained by sphygmomanometer calibration.

RESULTS

After three visits, uncalibrated sphygmomanometer error causes 20 and 28% of all undetected adult systolic and diastolic hypertension, respectively, and 15 and 31% of all falsely detected adult systolic and diastolic hypertension. In some groups, under-detection is worse; for example, sphygmomanometer error causes 27% of all missed systolic hypertension in 35-44-year-old females. In some age groups, over-detection is worse; for example, after three visits, sphygmomanometer error causes 63 and 50% of falsely detected systolic and diastolic hypertension in 18-24-year-old females, respectively. In-service sphygmomanometer calibration achieves the same or greater improvement in hypertension detection as an additional two visits.

CONCLUSIONS

Uncalibrated sphygmomanometers are a preventable cause of clinically significant over- and under-detection of hypertension. Sphygmomanometers should be calibrated regularly by accredited organizations or technicians. Standards and guidelines governing sphygmomanometers in service should be revised. Sphygmomanometer calibration is a cost-effective way of improving hypertension detection.

Screening for colorectal cancer using the faecal occult blood test, Hemoccult

BACKGROUND

Colorectal cancer is a leading cause of morbidity and mortality, especially in the Western world. The human and financial costs of this disease have prompted considerable research efforts to evaluate the ability of screening tests to detect the cancer at an early curable stage. Tests that have been considered for population screening include variants of the faecal occult blood test, flexible sigmoidoscopy and colonoscopy. Reducing mortality from colorectal cancer (CRC) may be achieved by the introduction of population-based screening programmes.

OBJECTIVES

To determine whether screening for colorectal cancer using the faecal occult blood test (guaiac or immunochemical) reduces colorectal cancer mortality and to consider the benefits, harms and potential consequences of screening.

SEARCH STRATEGY

Published and unpublished data for this review were identified by: Reviewing studies included in the previous Cochrane review; Searching several electronic databases (Cochrane Library, Medline, Embase, CINAHL, PsychInfo, Amed, SIGLE, HMIC); and Writing to the principal investigators of potentially eligible trials.

SELECTION CRITERIA

We included in this review all randomised trials of screening for colorectal cancer that compared faecal occult blood test (guaiac or immunochemical) on more than one occasion with no screening and reported colorectal cancer mortality.

DATA COLLECTION AND ANALYSIS

Data from the eligible trials were independently extracted by two reviewers. The primary data analysis was performed using the group participants were originally randomised to ('intention to screen'), whether or not they attended screening; a secondary analysis adjusted for non-attendence. We calculated the relative risks and risk differences for each trial, and then overall, using fixed and random effects models (including testing for heterogeneity of effects). We identified nine articles concerning four randomised controlled trials and two controlled trials involving over 320,000 participants with follow-up ranging from 8 to 18 years.

MAIN RESULTS

Combined results from the 4 eligible randomised controlled trials shows that participants allocated to screening had a 16% reduction in the relative risk of colorectal cancer mortality (RR 0.84, CI: 0.78-0.90). In the 3 studies that used biennial screening (Funen, Minnesota, Nottingham) there was a 15% relative risk reduction (RR 0.85, CI: 0.78-0.92) in colorectal cancer mortality. When adjusted for screening attendance in the individual studies, there was a 25% relative risk reduction (RR 0.75, CI: 0.66 - 0.84) for those attending at least one round of screening using the faecal occult blood test.

AUTHORS' CONCLUSIONS

Benefits of screening include a modest reduction in colorectal cancer mortality, a possible reduction in cancer incidence through the detection and removal of colorectal adenomas, and potentially, the less invasive surgery that earlier treatment of colorectal cancers may involve. Harmful effects of screening include the psycho-social consequences of receiving a false-positive result, the potentially significant complications of colonoscopy or a false-negative result, the possibility of overdiagnosis (leading to unnecessary investigations or treatment) and the complications associated with treatment.

Cervical cytology reading times: A comparison between thinprep imager and conventional methods

We aimed to compare the times cytologists spend reviewing cervical cytology slides processed by the ThinPrep Imager (TPI) with times they spend examining conventional cytology (CC) slides. We also aimed to examine the effect of cytologists' experience on reading times. Using a cross-sectional analytical design, we analyzed routine laboratory data, collected retrospectively over 7 months, for 41 cytologists, including paired data for 20 who read both TPI and CC slides. For the 20 cytologists who read both types of cytology, the mean reading rate was 13.3 slides per hour for TPI slides and 6.1 slides per hour for CC slides. The mean within-reader difference between TPI and CC rates was 7.2 slides per hour (P < 0.001). For CC reading, mean times did not differ between those who were additionally trained to read TPI slides and those who only read CC. Slower readers had greater increases in speed when using the TPI compared with CC reading than did faster readers (P < 0.001). More experienced cytologists tended to read CC slides more quickly than did those less experienced, but experience did not affect TPI reading times or within-reader differences in reading times between cytology types. The TPI significantly reduced reading times compared with CC. This reduction was greater amongst slower readers, and was unrelated to experience.

Intraocular lens power in bilateral cataract surgery: Whether adjusting for error of predicted refraction in the first eye improves prediction in the second eye

PURPOSE

To assess whether the retrospectively calculated intraocular lens (IOL) position value in the first eye reduces the error of predicted refraction in the second.

SETTING

Prince of Wales Hospital, Sydney, Australia.

METHODS

One hundred twenty-one consecutive patients who had bilateral cataract surgery with the same IOL (SI-30NB, Advanced Medical Optics) were identified. The case-derived A-constant in the first eye was calculated from the postoperative refraction. This value was used to calculate the adjusted error of predicted refraction in the second eye and compared against the unadjusted error in that eye (calculated using manufacturer's A-constant).

RESULTS

Axial length (r = 0.97), corneal power (r = 0.97), and IOL power (r = 0.90) were strongly correlated between eyes with no statistically significant mean interocular difference. Although there was no significant interocular difference in the mean error of predicted refraction (SRK/T), there was only a moderate correlation between eyes (r = 0.40). Using the axial length vergence formula, the mean adjusted error of predicted refraction in the second eye (-0.66 diopter [D]) was significantly larger than the mean unadjusted error (-0.47 D) (P = .029). The standard deviation of the adjusted error of predicted refraction (SRK/T) in the second eye (0.85 D) was greater than the standard deviation of the unadjusted error (0.79). Similarly, the adjusted mean absolute error of predicted refraction (0.65 D) was greater than the unadjusted error (0.63 D).

CONCLUSION

Adjusting the IOL power in the second eye by the amount of overprediction or underprediction in the first eye did not improve prediction accuracy because the error of predicted refraction varied independently between the 2 eyes of an individual.

Impact of privacy legislation on the number and characteristics of people who are recruited for research: a randomised controlled trial

BACKGROUND

Privacy laws have recently created restrictions on how researchers can approach study participants.

METHOD

In a randomised trial of 152 patients, 50-74 years old, in a family practice, 60 were randomly selected to opt-out and 92 to opt-in methods. Patients were sent an introductory letter by their doctor in two phases, opt-out before and opt-in after introduction of the new Privacy Legislation in December 2001. Opt-out patients were contacted by researchers. Opt-in patients were contacted if patients responded by email, free telephone number or a reply-paid card.

RESULTS

Opt-in recruited fewer patients (47%; 43/92) after invitation compared with opt-out (67%; 40/60); (-20%; [-4% to -36%]). No proportional difference in recruitment was found between opt-in and opt-out groups varied by age, sex or socioeconomic status. The opt-in group had significantly more people in active decision-making roles (+30%; [10% to 50%]; p = 0.003). Non-significant trends were observed towards opt-in being less likely to include people with lower education (-11.8%; [-30% to 6.4%]; p = 0.13) and people who were not screened (-19.1%; [-40.1% to 1.9%]; p = 0.08). Opt-in was more likely to recruit people with a family history of colorectal cancer (+12.7%; [-2.8%, 28.2%]; p = 0.12).

CONCLUSIONS

The number of participants required to be approached was markedly increased in opt-in recruitment. Existing participants (eg, screening attendees) with a vested interest such as increased risk, and those preferring an active role in health decision making and with less education were likely to be recruited in opt-in. Research costs and generalisability are affected by implementing privacy legislation.

When is measuring sensitivity and specificity sufficient to evaluate a diagnostic test, and when do we need randomized trials?

The clinical value of using a new diagnostic test depends on whether it improves patient outcomes beyond the outcomes achieved using an old diagnostic test. When can studies of diagnostic test accuracy provide sufficient information to infer clinical value, and when do clinicians need to wait for results from randomized trials? The authors argue that accuracy studies suffice if a new diagnostic test is safer or more specific than, but of similar sensitivity to, an old test. However, if a new test is more sensitive than an old test, it leads to the detection of extra cases of disease. Results from treatment trials that enrolled only patients detected by the old test may not apply to these extra cases. Clinicians need to wait for results from randomized trials assessing treatment efficacy in cases detected by the new diagnostic test, unless they can be satisfied that the new test detects the same spectrum and subtype of disease as the old test or that treatment response is similar across the spectrum of disease.

Informed choice for screening: implications for evaluation

Evaluation of screening should reflect consumer priorities. We need to make more effort to find out what they really are

Comparative accuracy: assessing new tests against existing diagnostic pathways

Most studies of diagnostic accuracy only compare a test with the reference standard. Is this helpful?

Radiological surveillance of interval breast cancers in screening programmes

Interval breast cancers-those diagnosed after a negative mammographic screen and before the next scheduled screen-are an important indicator of the potential effectiveness of population screening for breast cancer. Although the incidence of interval cancers is usually monitored, radiological surveillance is not undertaken routinely in most screening programmes. Here, we describe radiological surveillance of interval breast cancers and discuss methodological difficulties in the radiological review process and in the categorisation of interval cancers as false-negative, true, or occult. Furthermore, we identify methods that affect whether an interval cancer is classified as a false-negative (missed) or a true interval cancer. For all radiological categories of interval cancers, we outline possible changes to screening programmes that might improve cancer detection. Standardised radiological surveillance of interval cancers might allow within-programme comparisons and has the potential to guide practice and improve quality.

New methods give better estimates of changes in diagnostic accuracy when prior information is provided

OBJECTIVE

Whether tests such as imaging should be read with or without access to prior clinical information is controversial. Naïve comparisons may suggest that the provision of prior information improves test accuracy, whereas in fact the opposite may be true. This is because provision of clinical background may actually bias test readers to over- or underinterpret relevant test findings, and they may suboptimally integrate the previous and current evidence. We propose comparing the combined accuracy of prior information and a test read (i) with or (ii) without knowledge of prior information. Analysis methods include simple decision rules and logistic regression.

STUDY DESIGN AND SETTING

A study of cancer detection in women presenting with breast symptoms, in whom ultrasound could be read with or without reviewing prior mammography.

RESULTS

Naïve analysis gave an area under the receiver operating characteristics curve (AUC) for ultrasound read with mammography on view that was 4.6% higher (P < .01) than without mammography on view. Our approach, comparing the combined accuracy of mammography and ultrasound read i) with and ii) without knowledge of mammographic findings, showed much smaller differences.

CONCLUSION

Our approach is more appropriate than naïve analyses. The particular choice of analytic method depends on the study size and the diagnostic accuracy of combinations of the prior information and the test reading.

Effect of study design and quality on unsatisfactory rates, cytology classifications, and accuracy in liquid-based versus conventional cervical cytology: a systematic review

BACKGROUND

Liquid-based cytology is reported to increase the sensitivity of cervical cytology and the proportion of slides that are satisfactory for assessment, in comparison with conventional cytology. Although some countries have changed to liquid-based cytology for cervical screening, controversy remains. We reviewed the published work to assess the performance of liquid-based cytology relative to conventional cytology in primary studies assessed to be of low, medium, or high methodological quality.

METHODS

56 primary studies were reviewed and assessed with strict methodological criteria. Liquid-based cytology and conventional cytology were compared in terms of the percentage of slides classified as unsatisfactory, the percentage of slides classified in each cytology category, and the accuracy of detection of high-grade disease. Data were examined for studies overall and in strata to examine the effect of study quality on results.

FINDINGS

The median difference in the percentage of unsatisfactory slides between liquid-based cytology and conventional cytology was 0.17%. Only one small study was a randomised controlled trial. The classification of high-grade squamous epithelial lesion varied according to study quality (p=0.04), with conventional cytology classifying more slides in this category than did liquid-based cytology in high-quality studies (n=3) only. In medium-quality (n=30) and high-quality studies, liquid-based cytology classified more slides as atypical squamous cells of unknown significance than did conventional cytology when compared with low-quality studies (n=17; p=0.05). Only four studies provided sufficient verified data to allow estimation of sensitivity and specificity and comparison of test accuracy.

INTERPRETATION

We saw no evidence that liquid-based cytology reduced the proportion of unsatisfactory slides, or detected more high-grade lesions in high-quality studies, than conventional cytology. This review does not lend support to claims of better performance by liquid-based cytology. Large randomised controlled trials are needed.