Predicting recovery in patients with acute low back pain: A Clinical Prediction Model.

BACKGROUND: There is substantial variability in the prognosis of acute low back pain (LBP). The ability to identify the probability of individual patients recovering by key time points would be valuable in making informed decisions about the amount and type of treatment to provide. Predicting recovery based on presentation 1-week after initially seeking care is clinically important and may be more accurate than predictions made at initial presentation. The aim of this study was to predict the probability of recovery at 1-week, 1-month and 3-months after 1-week review in patients who still have LBP 1-week after initially seeking care. METHODS: The study sample comprised 1070 patients with acute LBP, with a pain score of ≥2 1-week after initially seeking care. The primary outcome measure was days to recovery from pain. Ten potential prognostic factors were considered for inclusion in a multivariable Cox regression model. RESULTS: The final model included duration of current episode, number of previous episodes, depressive symptoms, intensity of pain at 1-week, and change in pain over the first week after seeking care. Depending on values of the predictor variables, the probability of recovery at 1-week, 1-month and 3-months after 1-week review ranged from 4% to 59%, 19% to 91% and 30% to 97%, respectively. The model had good discrimination (C = 0.758) and calibration. CONCLUSIONS: This study found that a model based on five easily collected variables could predict the probability of recovery at key time points in people who still have LBP 1-week after seeking care. SIGNIFICANCE: A clinical prediction model based on five easily collected variables was able to predict the likelihood of recovery from an episode of acute LBP at three key time points. The model had good discrimination (C = 0.758) and calibration.

Beyond the treatment effect: Evaluating the effects of patient preferences in randomised trials.

The treatments under comparison in a randomised trial should ideally have equal value and acceptability - a position of equipoise - to study participants. However, it is unlikely that true equipoise exists in practice, because at least some participants may have preferences for one treatment or the other, for a variety of reasons. These preferences may be related to study outcomes, and hence affect the estimation of the treatment effect. Furthermore, the effects of preferences can sometimes be substantial, and may even be larger than the direct effect of treatment. Preference effects are of interest in their own right, but they cannot be assessed in the standard parallel group design for a randomised trial. In this paper, we describe a model to represent the impact of preferences on trial outcomes, in addition to the usual treatment effect. In particular, we describe how outcomes might differ between participants who would choose one treatment or the other, if they were free to do so. Additionally, we investigate the difference in outcomes depending on whether or not a participant receives his or her preferred treatment, which we characterise through a so-called preference effect. We then discuss several study designs that have been proposed to measure and exploit data on preferences, and which constitute alternatives to the conventional parallel group design. Based on the model framework, we determine which of the various preference effects can or cannot be estimated with each design. We also illustrate these ideas with some examples of preference designs from the literature.

Obesity and hypertension in Australian young people: results from the Australian Health Survey 2011-2012.

BACKGROUND: Few studies have focused on the prevalence of obesity and hypertension among young people (ages 15-24). AIM: To characterise the prevalence of obesity and systolic hypertension in young people aged 15-24 years across Australia. METHODS: Using data from the 2011-2012 Australian Health Survey, a national cross-sectional population-based survey, we included 2163 young people aged 15-24 years. Risk factors were estimated using multinomial logistic regression. RESULTS: The prevalence of obesity increased from 8% to 15% through the ages of 15-24 among males, but the prevalence of overweight and obesity were both 14% for females across all age groups. Low levels of physical activity were a strong risk factor for obesity for both males (odds ratio (OR) 5.95, 95% confidence intervals (CI)1.83-19.36) and females (OR 3.20 95% CI 0.69-14.87). Low socioeconomic status was associated with obesity among females only (first quintile OR 4.65, 95% CI 1.97-10.99). Although the prevalence of hypertension was low (4% males, 3% females), the prevalence of high normal blood pressure was substantial, especially among males (28% males, 14% females). CONCLUSIONS: Overweight, obesity and high normal blood pressure were highly prevalent among Australian young people. Low levels of physical activity were identified as a risk factor for obesity for both male and females. Programmes targeting physical activity participation may need to be tailored differently for males and females, with a focus on females during early adolescence but early adult life for males.

The differential effect of socio-economic status, birth weight and gender on body mass index in Australian Aboriginal Children.

BACKGROUND: Adult Aboriginal Australians have 1.5-fold higher risk of obesity, but the trajectory of body mass index (BMI) through childhood and adolescence and the contribution of socio-economic factors remain unclear. Our objective was to determine the changes in BMI in Australian Aboriginal children relative to non-Aboriginal children as they move through adolescence into young adulthood, and to identify risk factors for higher BMI. METHODS: A prospective cohort study of Aboriginal and non-Aboriginal school children commenced in 2002 across 15 different screening areas across urban, regional and remote New South Wales, Australia. Socio-economic status was recorded at study enrolment and participants' BMI was measured every 2 years. We fitted a series of mixed linear regression models adjusting for age, birth weight and socio-economic status for boys and girls. RESULTS: In all, 3418 (1949 Aboriginal) participants were screened over a total of 11 387 participant years of follow-up. The prevalence of obesity was higher among Aboriginal children from mean age 11 years at baseline (11.6 vs 7.6%) to 16 years at 8 years follow-up (18.6 vs 12.3%). The mean BMI increased with age and was significantly higher among Aboriginal girls compared with non-Aboriginal girls (P<0.01). Girls born of low birth weight had a lower BMI than girls born of normal birth weight (P<0.001). Socio-economic status and low birth weight had a differential effect on BMI for Aboriginal boys compared with non-Aboriginal boys (P for interaction=0.01). Aboriginal boys of highest socio-economic status, unlike those of lower socio-economic status, had a higher BMI compared with non-Aboriginal boys. Non-Aboriginal boys of low birth weight were heavier than Aboriginal boys. CONCLUSIONS: Socio-economic status and birth weight have differential effects on BMI among Aboriginal boys, and Aboriginal girls had a higher mean BMI than non-Aboriginal girls through childhood and adolescence. Intervention programs need to recognise the differential risk for obesity for Aboriginal and non-Aboriginal boys and girls to maximise their impact.

Meta-analysis of agreement between MRI and pathologic breast tumour size after neoadjuvant chemotherapy.

BACKGROUND: Magnetic resonance imaging (MRI) has been proposed to guide breast cancer surgery by measuring residual tumour after neoadjuvant chemotherapy. This study-level meta-analysis examines MRI's agreement with pathology, compares MRI with alternative tests and investigates consistency between different measures of agreement. METHODS: A systematic literature search was undertaken. Mean differences (MDs) in tumour size between MRI or comparator tests and pathology were pooled by assuming a fixed effect. Limits of agreement (LOA) were estimated from a pooled variance by assuming equal variance of the differences across studies. RESULTS: Data were extracted from 19 studies (958 patients). The pooled MD between MRI and pathology from six studies was 0.1 cm (95% LOA: -4.2 to 4.4 cm). Similar overestimation for MRI (MD: 0.1 cm) and ultrasound (US) (MD: 0.1 cm) was observed, with comparable LOA (two studies). Overestimation was lower for MRI (MD: 0.1 cm) than mammography (MD: 0.4 cm; two studies). Overestimation by MRI (MD: 0.1 cm) was smaller than underestimation by clinical examination (MD: -0.3 cm). The LOA for mammography and clinical examination were wider than that for MRI. Percentage agreement between MRI and pathology was greater than that of comparator tests (six studies). The range of Pearson's/Spearman's correlations was wide (0.21-0.92; 16 studies). Inconsistencies between MDs, percentage agreement and correlations were common. CONCLUSION: Magnetic resonance imaging appears to slightly overestimate pathologic size, but measurement errors may be large enough to be clinically significant. Comparable performance by US was observed, but agreement with pathology was poorer for mammography and clinical examination. Percentage agreement can provide supplementary information to MDs and LOA, but Pearson's/Spearman's correlation does not provide evidence of agreement and should be avoided. Further comparisons of MRI and other tests using the recommended methods are warranted.

Early prediction of pathologic response to neoadjuvant therapy in breast cancer: systematic review of the accuracy of MRI.

Magnetic resonance imaging (MRI) has been proposed to have a role in predicting final pathologic response when undertaken early during neoadjuvant chemotherapy (NAC) in breast cancer. This paper examines the evidence for MRI's accuracy in early response prediction. A systematic literature search (to February 2011) was performed to identify studies reporting the accuracy of MRI during NAC in predicting pathologic response, including searches of MEDLINE, PREMEDLINE, EMBASE, and Cochrane databases. 13 studies were eligible (total 605 subjects, range 16-188). Dynamic contrast-enhanced (DCE) MRI was typically performed after 1-2 cycles of anthracycline-based or anthracycline/taxane-based NAC, and compared to a pre-NAC baseline scan. MRI parameters measured included changes in uni- or bidimensional tumour size, three-dimensional volume, quantitative dynamic contrast measurements (volume transfer constant [Ktrans], exchange rate constant [k(ep)], early contrast uptake [ECU]), and descriptive patterns of tumour reduction. Thresholds for identifying response varied across studies. Definitions of response included pathologic complete response (pCR), near-pCR, and residual tumour with evidence of NAC effect (range of response 0-58%). Heterogeneity across MRI parameters and the outcome definition precluded statistical meta-analysis. Based on descriptive presentation of the data, sensitivity/specificity pairs for prediction of pathologic response were highest in studies measuring reductions in Ktrans (near-pCR), ECU (pCR, but not near-pCR) and tumour volume (pCR or near-pCR), at high thresholds (typically >50%); lower sensitivity/specificity pairs were evident in studies measuring reductions in uni- or bidimensional tumour size. However, limitations in study methodology and data reporting preclude definitive conclusions. Methods proposed to address these limitations include: statistical comparison between MRI parameters, and MRI vs other tests (particularly ultrasound and clinical examination); standardising MRI thresholds and pCR definitions; and reporting changes in NAC based on test results. Further studies adopting these methods are warranted.

Optimal allocation of participants for the estimation of selection, preference and treatment effects in the two-stage randomised trial design.

Outcomes in clinical trials may be affected by the choice of treatment that participants might make, if they were indeed allowed to choose (a so-called selection effect), and by whether they actually receive their preferred treatment (a preference effect). Selection and preference effects can be important, but they cannot be estimated in the conventional trial design. An alternative approach is the two-stage randomised trial, in which participants are first randomly divided into two subgroups. In one subgroup, participants are randomly assigned to treatments, while in the other, participants are allowed to choose their own treatment. This approach yields estimates of the direct treatment effect, and of the preference and selection effects. The latter two provide insight that goes considerably beyond what is possible in the standard randomised trial. In this paper, we determine the optimal proportion of participants who should be allocated to the choice subgroup. The precision of the estimated selection, preference and treatment effects are functions of: the total sample size; the proportion of participants allocated to choose their treatment; the variances of the outcome; the proportions of participants who select each treatment in the choice group; and the selection, preference and treatment effects themselves. We develop general expressions for the optimum proportion of participants in the choice group, depending on which effects are of primary interest. We illustrate the results with trial data comparing alternative clinical management strategies for women with abnormal results on cervical screening.

Effect of dependent errors in the assessment of diagnostic or screening test accuracy when the reference standard is imperfect.

When no gold standard is available to evaluate a diagnostic or screening test, as is often the case, an imperfect reference standard test must be used instead. Furthermore, the errors of the test and its reference standard may not be independent. Some authors have opined that positively dependent errors will lead to overestimation of test performance. Although positive dependence does increase agreement between the test and the reference standard, it is not clear if test accuracy will necessarily be overestimated in this situation, and the case of negatively associated test errors is even less clear. To examine this issue in more detail, we derive the apparent sensitivity, specificity, and overall accuracy of a test relative to an imperfect reference standard and the bias in these parameters. We demonstrate that either positive or negative bias can occur if the reference standard is imperfect. The type and magnitude of bias depend on several components: the disease prevalence, the true test sensitivity and specificity, the covariance between the false-negative test errors among the true disease cases, and the covariance between the false-positive test errors among the true noncases. If, for example, sensitivity and specificity are 0.8 for both the test and reference standard and the errors have a moderate positive dependence, test sensitivity is then underestimated at low prevalence but overestimated at high prevalence, while the opposite occurs for specificity. We illustrate these ideas through general numerical calculations and an empirical example of screening for breast cancer with magnetic resonance imaging and mammography.

Who needs colonoscopy to identify colorectal cancer? Bowel symptoms do not add substantially to age and other medical history.

BACKGROUND: Many bodies advise that people with bowel symptoms undergo colonoscopy to detect colorectal cancer. AIM: To determine which bowel symptoms predict cancer on colonoscopy. METHODS: Information was collected on symptoms, demographics and medical history from patients subsequently undergoing colonoscopy. Multiple logistic regression modelling was used to identify predictors of colorectal cancer. An ROC curve was estimated for each model, and the area under the curve (AUC) was computed. RESULTS: Cancer was found in 159 patients and no cancer or adenoma in 7577 patients. Bowel symptoms that predicted cancer were rectal bleeding, change in bowel habit and rectal mucus. Prediction was the strongest in patients who had symptoms at least weekly and commencing within the previous 12 months; abdominal pain was predictive only in such patients. The odds ratios never exceeded 4.27. A model based on age, gender, and medical history was highly predictive (AUC = 0.79). Adding symptoms to this model increased the AUC to 0.85. CONCLUSIONS: This model predicts patients in whom colonoscopy will have the highest yield. Conversely, colonoscopy can be avoided in people at low risk: in our study, 95% of cancers could have been detected by doing only 60% of the colonoscopies.

Dermoscopy compared with naked eye examination for the diagnosis of primary melanoma: a meta-analysis of studies performed in a clinical setting.

BACKGROUND: Dermoscopy is a noninvasive technique that enables the clinician to perform direct microscopic examination of diagnostic features, not seen by the naked eye, in pigmented skin lesions. Diagnostic accuracy of dermoscopy has previously been assessed in meta-analyses including studies performed in experimental and clinical settings. OBJECTIVES: To assess the diagnostic accuracy of dermoscopy for the diagnosis of melanoma compared with naked eye examination by performing a meta-analysis exclusively on studies performed in a clinical setting. METHODS: We searched for publications from 1987 to January 2008 and found nine eligible studies. The selected studies compare diagnostic accuracy of dermoscopy with naked eye examination using a valid reference test on consecutive patients with a defined clinical presentation, performed in a clinical setting. Hierarchical summary receiver operator curve analysis was used to estimate the relative diagnostic accuracy for clinical examination with, and without, the use of dermoscopy. RESULTS: We found the relative diagnostic odds ratio for melanoma, for dermoscopy compared with naked eye examination, to be 15.6 [95% confidence interval (CI) 2.9-83.7, P = 0.016]; removal of two outlier studies changed this to 9.0 (95% CI 1.5-54.6, P = 0.03). CONCLUSIONS: Dermoscopy is more accurate than naked eye examination for the diagnosis of cutaneous melanoma in suspicious skin lesions when performed in the clinical setting.

Long-term prognosis of multiple sclerosis in Australia.

Long-term survival and progression of disability in multiple sclerosis have not been studied previously in Australia. We report the findings in a cohort of 159 patients from Newcastle. Median survival time from onset of symptoms to death was 42 years. When expected survival rates are compared with those of the Australian population, there is approximately a 10% reduction in survival time in multiple sclerosis patients, after 20 years or more from disease onset. The expected time to reach DSS 3 and DSS 6 was 7 years and 27 years respectively. Survival time of multiple sclerosis patients and rate of progression of the disease are similar in the Northern and Southern Hemispheres.

Meta-analysis of sentinel node imprint cytology in breast cancer.

BACKGROUND: Intraoperative diagnosis of breast cancer metastases in axillary sentinel nodes is desirable to avoid a second operation for lymphadenectomy. Imprint or touch-preparation cytology is a popular technique that has high specificity and a wide range of sensitivity. METHODS: A systematic search of electronic databases was performed. Included articles were assessed for methodological and reporting quality. Random-effects model pooled estimates of sensitivity and specificity were calculated. Single-variable and multivariable meta-regression analyses were performed for predictors of sensitivity. RESULTS: Thirty-one studies were included; all were of good methodological quality but reporting quality varied. Pooled sensitivity of imprint cytology was 63 (95 per cent confidence interval (c.i.) 57 to 69) per cent and specificity was 99 (95 per cent c.i. 98 to 99) per cent. Pooled sensitivity for macrometastases was 81 per cent and that for micrometastases 22 per cent. Mean or median primary tumour size (P = 0.004), the prevalence of metastases (P = 0.103) and the proportion of micrometastases (P = 0.022) were significant risk factors in single-variable meta-regression analysis. Only the proportion of micrometastases remained significant in multivariable analysis. Frozen sectioning had better sensitivity than imprint cytology in three of four direct comparisons. CONCLUSION: Imprint cytology is simple and rapid, and has good sensitivity for macrometastases. The significance of poor sensitivity for micrometastases will be determined by trials investigating their natural history.

Progressive increase in incidence and prevalence of multiple sclerosis in Newcastle, Australia: a 35-year study.

The prevalence of multiple sclerosis (MS) in Newcastle, Australia increased significantly between 1961 and 1981 and the incidence of the disease also increased between the decades 1950-1959 and 1971-1981. The present study sought to determine whether there has been a further increase in the frequency of MS in the subsequent 15 years, and to examine the potential factors underlying this change. The incidence, prevalence and clinical profile of multiple sclerosis were therefore re-examined in Newcastle, Australia in 1996 using comparable diagnostic criteria and methods to those employed in studies in the same region in 1961 and 1981. There has been a significant progressive increase in prevalence from 19.6 to 59.1 per 100,000 population and a significant increase in incidence from 1.2 to 2.4 per 100,000 population from 1961 to 1996. The most pronounced increase in prevalence was in females and in the age-group over 60 years, and there was also an increased incidence in females aged 20-29 years. There was little change in the age of disease onset, but duration of disease in females had increased substantially. The significant increase in prevalence is attributed to increased incidence, particularly in females; and to increased survival. Although such trends in prevalence have been observed in the Northern Hemisphere, this is the first such study in the Southern Hemisphere to show a longitudinal increase in prevalence and incidence over a period of this duration.

The comparative sensitivity of mammography and ultrasound in women with breast symptoms: an age-specific analysis.

Despite its importance for clinical policy, there is little research on the age at which ultrasound or mammography is the preferred initial diagnostic test in women with breast symptoms. We analysed data from a series of 3799 consecutively presenting cancers. The overall sensitivity was 77.6% for mammography and 80.5% for ultrasound. Sensitivity increased with age in both mammography and ultrasound, though more steeply for mammography. In the 975 women who had both tests, ultrasound had a higher sensitivity than mammography in women younger than 62 years of age, whereas mammography had a higher sensitivity than ultrasound in women older than 62 years. However, if the test results of the 2393 women who had mammography only are compared with the ultrasound results of the 975 women who had both tests, the 'cross-over' age at which the sensitivity of the two tests is equal occurs at the earlier age of 48 years. The presence of a palpable finding increases the sensitivity of ultrasound but does not influence the sensitivity of mammography.

Evidence relevant to the investigation of breast symptoms: the triple test.

This paper reviews and summarises evidence on the investigation of breast symptoms based on triple testing. The triple test (where any component is positive) has a sensitivity of 99.6%, a specificity of 62%, a likelihood ratio of 2.62 for a positive result, and a likelihood ratio of 0.006 for a negative result. We present measures of test accuracy for the components of the triple test that allow easy estimation of the probability of breast cancer in symptomatic women, given any possible combination of results.

Sensitivity and specificity of tests to detect eye disease in an older population.

OBJECTIVE: To compare the ability of tests of visual function to detect the presence of eye disease. DESIGN: Cross-sectional study. PARTICIPANTS: Three thousand six hundred fifty-four of 4433 (82.4%) eligible residents of an area near Sydney aged 49 years and older had a detailed eye examination, including retinal and lens photography and subsequent grading of eye disease, tests of presenting and corrected visual acuity, contrast sensitivity, screening visual field and intraocular pressure. MAIN OUTCOME MEASURES: Receiver operator characteristic (ROC) curves were created and area under the curve compared for each vision test. Sensitivity and specificity were calculated for each test. RESULTS: No single vision test predicted the presence of eye disease with any consistency. Best-corrected visual acuity or contrast sensitivity had the highest area under the ROC curve for most eye diseases examined but had poor sensitivity and specificity. For glaucoma and diabetic retinopathy there was no difference in area under the curve for any of the tests examined, and no test had a good balance of sensitivity and specificity. Screening tests (performed with presenting correction) did not perform as well as nonscreening tests (those carried out after refraction with best correction). CONCLUSIONS: Current vision tests are not particularly good at detecting eye disease compared with the "gold standard" of a full eye examination performed by an ophthalmologist. Further work in this area should be carried out before vision screening programs can be recommended for implementation among older people.

A comparison of methods to detect publication bias in meta-analysis.

Meta-analyses are subject to bias for many of reasons, including publication bias. Asymmetry in a funnel plot of study size against treatment effect is often used to identify such bias. We compare the performance of three simple methods of testing for bias: the rank correlation method; a simple linear regression of the standardized estimate of treatment effect on the precision of the estimate; and a regression of the treatment effect on sample size. The tests are applied to simulated meta-analyses in the presence and absence of publication bias. Both one-sided and two-sided censoring of studies based on statistical significance was used. The results indicate that none of the tests performs consistently well. Test performance varied with the magnitude of the true treatment effect, distribution of study size and whether a one- or two-tailed significance test was employed. Overall, the power of the tests was low when the number of studies per meta-analysis was close to that often observed in practice. Tests that showed the highest power also had type I error rates higher than the nominal level. Based on the empirical type I error rates, a regression of treatment effect on sample size, weighted by the inverse of the variance of the logit of the pooled proportion (using the marginal total) is the preferred method.

Should we use carbohydrate-deficient transferrin instead of gamma-glutamyltransferase for detecting problem drinkers? A systematic review and metaanalysis.

BACKGROUND: Carbohydrate-deficient transferrin (CDT) has been used as a test for excessive alcohol consumption in research, clinical, and medico-legal settings, but there remain conflicting data on its accuracy, with sensitivities ranging from <20% to 100%. We examined evidence of its benefit over a conventional and less expensive test, gamma-glutamyltransferase (GGT), and compared the accuracy of different CDT assay methods. METHODS: We performed a systematic review using summary ROC analysis of 110 studies prior to June 1998 on the use of CDT in the detection of alcohol dependence or hazardous/harmful alcohol use. RESULTS: We identified several potential sources of bias in studies. In studies examining CDT and GGT in the same subjects, subject characteristics were less likely to influence the comparison. In such paired studies, the original Pharmacia CDT assay was significantly more accurate than GGT, but the modified CDTect assay did not perform as well as the original and was not significantly better than GGT. The accuracy of the AXIS %CDT assay was statistically indistinguishable from modified CDTect. Several CDT assay methods appeared promising, in particular, liquid chromatography (chromatofocusing, HPLC, fast protein liquid chromatography) and isoelectric focusing, but there were insufficient paired studies from which to draw firm conclusions. CONCLUSIONS: In studies published before June 1998, the results obtained with commercially available CDT assays were not significantly better than GGT as markers of excessive alcohol use in paired studies. Further high-quality studies comparing CDTect (modified) and other CDT assays with GGT in the same subjects are needed.

Multiple sclerosis in Australia: prognostic factors.

In order to determine the influence of age of onset, sex, onset symptoms, clinical course and interval from onset to first relapse on the subsequent outcome of multiple sclerosis (MS), data from 2934 cases of MS documented in a large population based study undertaken in Australia have been analysed. Disability on prevalence day (30 June 1981) was defined on the Kurtzke disability scale as mild (DSS 0-3), moderate (DSS 4-6) and severe (DSS 7-9). Prognostic factors associated with mild vs moderate/severe, and moderate vs severe disability on prevalence day were identified by logistic regression analysis. A worse prognosis was significantly associated with older age of onset, progressive disease course, onset symptoms that were multiple, pyramidal or cerebellar, and a short interval between onset and first relapse.

Prevalence of chronic inflammatory demyelinating polyneuropathy in New South Wales, Australia.

A prevalence study of chronic inflammatory demyelinating polyneuropathy (CIDP) was performed in New South Wales (NSW), Australia, with a prevalence day of August 6, 1996, which coincided with a national census. The population of NSW was 5,995,544, and the crude prevalence of CIDP was 1.9 per 100,000 population. It was higher in male patients than in female patients, and the age-specific prevalence reached a maximum of 6.7 per 100,000 population in the 70- to 79-year-old age group. The prevalence in the city of Newcastle, with a population of 448,663, was 2.0 per 100,000 population and is representative of the whole of NSW. The estimated crude annual incidence was 0.15 per 100,000 population. The mean age of onset was 47.6 years (median, 53.5 years), 51% of patients had a relapsing-remitting course, the mean duration on prevalence day was 7.1 years (median, 5 years), and 87% of patients were able to walk without walking aids or other assistance.