Chun Fan Lee

Clinical effectiveness of school screening for adolescent idiopathic scoliosis: A large population-based retrospective cohort study

Study Design. Retrospective cohort study. Objective. To evaluate the clinical effectiveness of school scoliosis screening using a large and long–term-followed cohort of students in Hong Kong. Summary of Background Data. School screening for adolescent idiopathic scoliosis has been criticized as resulting in over-referrals for radiography and having low predictive values. Indeed, all but one previous retrospective cohort studies had no follow-up assessments of students until their skeletal maturity, leaving any late-developed curves undetected. The one study that completed this follow-up was well conducted but had low precisions due to its small sample size. Methods. A total of 157,444 students were eligible for a biennial scoliosis screening, and their screening results and medical records up to 19 years of age were available. Students first had forward bending test and angle of trunk rotation (ATR) performed. Those with ATR between 5° and 14° or signs of adolescent idiopathic scoliosis were assessed by moiré topography regularly. Students with an ATR ≥15°, ≥2 moiré lines, or significant clinical signs were referred for radiography and had their Cobb angle measured. Results. Of the 115,190 screened students in the cohort, 3228 (2.8%, 95% confidence interval [CI] = 2.7%–2.9%) were referred for radiography. At the final follow-up, the positive predictive values were 43.6% (41.8%–45.3%) for a Cobb angle ≥20° and 9.4% (8.4%–10.5%) for needing treatment, while the sensitivities were 88.1% (86.4%–89.6%) and 80.0% (75.6%–83.9%), respectively. Conclusion. This is the largest study that has demonstrated that school scoliosis screening in Hong Kong is predictive and sensitive with a low referral rate. Screening should thus be continued in order to facilitate early administration of conservative treatments.

A meta-analysis of the clinical effectiveness of school scoliosis screening

Study Design. A meta-analysis that systematically reviewed the evaluation studies of a scoliosis screening program reported in the literature. Objective. To evaluate the best current evidence on the clinical effectiveness of school screening for adolescent idiopathic scoliosis. Summary of Background Data. The use of school scoliosis screening is controversial, and its clinical effectiveness has been diversely reported. Methods. Data sources included 3 databases, namely, PubMed, Google scholar, CINAHL database, and the references from identified reviews and studies. Studies were included if: (1) they adopted a retrospective cohort design; (2) were screened using either the forward bending test (FBT), angle of trunk rotation, or Moiré topography; (3) reported results of screening tests and radiographic assessments; (4) screened adolescents only; (5) reported the incidence of curves with a minimum Cobb angle of 10° or greater; and (6) reported the number of referrals for radiography. Reviews, comments, case studies, and editorials were excluded. Results. Thirty-six studies, including 34 from the 775 initially identified studies and 2 from the references, met the selection criteria. The pooled referral rate for radiography was 5.0%, and the pooled positive predictive values for detecting curves ≥10°, curves ≥20°, and treatment were 28.0%, 5.6%, and 2.6%, respectively. There was substantial heterogeneity across studies. Meta-regression showed that programs using the FBT alone reported a higher referral rate (odds ratio [OR] = 2.91) and lower positive predictive values for curves ≥10° (OR = 0.49) and curves ≥20° (OR = 0.34) than programs using other tests. Only one small study followed students until skeletal maturity and reported the sensitivity of screening; however, the specificity was not reported. No severe publication bias was noted. Conclusion. The use of the FBT alone in school scoliosis screening is insufficient. We need large, retrospective cohort studies with sufficient follow-up to properly assess the clinical effectiveness of school scoliosis screening.

Measurement properties of the English and Chinese versions of the Functional Assessment of Cancer Therapy—Breast (FACT-B) in Asian breast cancer patients

The objective of the study was to examine the measurement properties of and comparability between the English and Chinese versions of the Functional Assessment of Cancer Therapy—Breast (FACT-B) in breast cancer patients in Singapore. This is an observational study of 271 Singaporean breast cancer patients. The known-group validity of FACT-B total score and Trial Outcome Index (TOI) were assessed in relation to performance status, evidence of disease, and treatment status cross-sectionally; responsiveness to change was assessed in relation to change in performance status longitudinally. Internal consistency and test–retest reliability were evaluated by the Cronbach’s alpha and intraclass correlation coefficient (ICC), respectively. Multiple regression analyses were performed to compare the scores on the two language versions, adjusting for covariates. The FACT-B total score and TOI demonstrated known-group validity in differentiating patients with different clinical status. They showed high internal consistency and test–retest reliability, with Cronbach’s alpha ranging from 0.87 to 0.91 and ICC ranging from 0.82 to 0.89. The English version was responsive to the change in performance status. The Chinese version was shown to be responsive to decline in performance status but the sample size of Chinese-speaking patients who improved in performance status was too small (Nxa0=xa06) for conclusive analysis about responsiveness to improvement. Two items concerning sexuality had a high item non-response rate (50.2 and 14.4%). No practically significant difference was found in the total score and TOI between the two language versions despite minor differences in two of the 37 items. The English and Chinese versions of the FACT-B are valid, responsive, and reliable instruments in assessing health-related quality of life in breast cancer patients in Singapore. Data collected from the English and Chinese versions can be pooled and either version could be used for bilingual patients.

Comparison of the measurement properties between a short and generic instrument, the 5-level EuroQoL Group’s 5-dimension (EQ-5D-5L) questionnaire, and a longer and disease-specific instrument, the Functional Assessment of Cancer Therapy—Breast (FACT-B), in Asian breast cancer patients

PurposeTo compare the measurement precision and related properties between the 5-level EuroQoL Group’s 5-dimension (EQ-5D-5L) questionnaire and the Functional Assessment of Cancer Therapy—Breast (FACT-B) questionnaire in assessing breast cancer patients.MethodsAn observational study of 269 Singaporean breast cancer patients. To compare discriminative ability and responsiveness, the effect sizes (in standard deviation) of the EQ-5D-5L and the FACT-B in relation to health indicators and the change in performance status or quality of life were estimated. Test–retest reliability was examined using the intraclass correlation (ICC).ResultsUsing performance status, evidence of disease, and treatment status as the criteria, the differences (FACT-B minus EQ-5D-5L) in the effect size for discriminative ability were negative or closed to zero, and the 90xa0% confidence intervals totally fell within the zone that indicated the non-inferiority of the EQ-5D-5L. For responsiveness and test–retest reliability, the confidence intervals of the differences in effect size and ICC overlapped the non-inferiority margin; thus, non-inferiority in these two aspects could neither be confirmed nor rejected.ConclusionsThe EQ-5D-5L was non-inferior to the FACT-B in discriminating breast cancer patients with different health conditions cross-sectionally. The EQ-5D-5L serves as a reasonable alternative or supplementary instrument to the FACT-B in assessing breast cancer patients’ health outcomes.

Costs of school scoliosis screening: a large, population-based study.

Study Design. This was a population-based retrospective study. Objective. To fully estimate the costs of the Hong Kong scoliosis screening program through a large, population-based study. Summary of Background Data. School screening for scoliosis has often been criticized for having high costs. In fact, the screening cost that has reported varied widely, from less than 1 to more than 30 US dollars (USD) per child screened. This variation is mainly due to the incomplete inclusion of cost items. Methods. We examined the screening and medical histories of a cohort of 115,190 screened students who were in Grade 5 in 1995/96 or 1996/97. The average costs spent on screening, diagnosing, following, and treating this cohort of students were calculated. Results. The total expenses in the screening centers increased steadily from USD 380,930 in 1995/96 to USD 2,417,824 in 2005/06. Based on the 115,190 students who were followed up until they were 19 years old or they left school, the costs of screening and diagnosing 1 student during adolescence were USD 17.94 and USD 2.08, respectively. Of the 1311 referrals who attended the specialist hospitals for diagnosis, 264 and 39 had been braced and operated on, respectively. The medical care cost averaged USD 34.61 per student screened. The cost of finding 1 student with a curvature ≥20° and 1 treated case were USD 4475.67 and USD 20,768.29 respectively. Conclusion. This was the largest study that has evaluated school scoliosis screening on students who were followed during their adolescence and accounted for all relevant costs. The cost per student screened in the scoliosis screening program in Hong Kong was comparable to that in Rochester, which had a similar protocol and was evaluated in a similar manner. The estimated costs can help the policy makers when they allocate healthcare resources.

Mapping the Functional Assessment of Cancer Therapy - Breast (FACT-B) to the 5-level EuroQoL group’s 5-dimension questionnaire (EQ-5D-5L) utility index in a Multi-ethnic Asian population

PurposeTo develop an algorithm for mapping the Functional Assessment of Cancer Therapy – Breast (FACT-B) to the 5-level EuroQoL Group’s 5-dimension questionnaire (EQ-5D-5L) utility index.MethodsA survey of 238 breast cancer patients in Singapore was conducted. Models using various regression methods with or without recognizing the upper boundary of utility values at 1 were fitted to predict the EQ-5D-5L utility index based on the five subscale scores of the FACT-B. Data from a follow-up survey of these patients were used to validate the results.ResultsA model that maps the physical, emotional, functional well-being and the breast cancer concerns subscales of the FACT-B to the EQ-5D-5L utility index was derived. The social well-being subscale was not associated to the utility index. Although theoretical assumptions may not be valid, ordinary least square outperformed other regression methods. The mean predicted utility index within each performance status level at follow-up deviated from the observed mean less than the minimally important difference of EQ-5D for cancer patients.ConclusionsThe mapping algorithm converts the FACT-B to the EQ-5D utility index. This enables oncologists, clinical researchers and policy makers to obtain a quantitative utility summary of a patient’s health status when only the FACT-B is assessed.

Eyes or subjects: Are ophthalmic randomized controlled trials properly designed and analyzed?

OBJECTIVEnIn ophthalmic randomized controlled trials (RCTs), each subject may have 2 potential data points (i.e., eyes) contributing to the clinical trial. Hence, various study designs may arise requiring different statistical tools. This study aimed to assess the appropriateness of study design, statistical tools used, and reporting of results in ophthalmic RCTs.nnnDESIGNnA systematic review of 69 ophthalmic RCTs.nnnMETHODSnThe study design, sample size calculation, statistical analysis, and reporting methodology of all RCTs published in 4 major general clinical ophthalmology journals in 2009 were assessed.nnnMAIN OUTCOME MEASURESnThe study design of each article under review is evaluated.nnnRESULTSnThe most common study design was a one-eye design (48%). Within this group, only half described the method of selecting the study eye, among which 5 chose the study eye by random selection. In the remaining trials, there were paired-eye design (13%), subject design (19%), and two-eye design (19%). Among the 13 two-eye design studies, 4 allocated both eyes of the subject to the same group, 4 allocated the eyes to different groups, and 4 did not restrict the allocation. None of these studies adjusted for the clustering effect in sample size calculation. Only 5 studies used statistical methods adjusting for nonindependence.nnnCONCLUSIONSnThere is currently substantial heterogeneity in the quality among published ophthalmic RCTs in terms of proper use of study design, sample size calculation, randomization method, and statistical tools. Future ophthalmic researchers are suggested to consult a statistician and to follow some guidelines such as the CONSORT statement when performing an RCT to improve further the quality of clinical trial.

A population-based cohort study of 394,401 children followed for 10 years exhibits sustained effectiveness of scoliosis screening

BACKGROUND CONTEXTnThe value of scoliosis screening has been recently shown in a multicenter randomized controlled trial. However, the long-term sustainability of the clinical effectiveness of scoliosis screening as a routine health service remains unknown.nnnPURPOSEnThe aim of this study was to assess the sustainability of the clinical effectiveness of school scoliosis screening.nnnSTUDY DESIGN/SETTINGnA large population-based cohort study with a 10-year follow-up was conducted.nnnPATIENT SAMPLEnA total of 394,401 students who were in the fifth grade during the five academic years from 1995/1996 to 1999/2000 formed five consecutive annual cohorts. The students were eligible for the Hong Kong scoliosis screening program, with their screening history and medical records until their nineteenth birthdays being assessed.nnnOUTCOME MEASURESnThe outcome measures considered in the study were development of adolescent idiopathic scoliosis by the 19 years of age and the Cobb angle.nnnMETHODSnThe clinical effectiveness of scoliosis screening was assessed by referral rate for radiographic diagnosis, sensitivity, specificity, and predictive values.nnnRESULTSnA total of 306,144 students (78%) participated in scoliosis screening, which used a two-tier system. The prevalence of curves of 20° or greater was 1.8% (95% confidence interval [CI], 1.7-1.8%), whereas the referral rate for radiography, the sensitivity, and the positive predictive value (PPV) for curves of 20° or greater were 4.1% (95% CI, 4.0-4.2%), 91% (95% CI, 90-92%), and 40% (95% CI, 39-41%), respectively. Across the five consecutive annual cohorts, the prevalence and sensitivity for curves of 20° or greater increased by 0.23% (95% CI, 0.21-0.25%; p<.001) and 0.76% (95% CI, 0.43-1.04%; p<.001) per year, respectively; however, the PPV was reduced by 1.71% (95% CI, 1.09-2.33%; p<.001) per year.nnnCONCLUSIONSnThis report describes the first large population-based study with a long-term follow-up indicating that a scoliosis screening program can have sustained clinical effectiveness in identifying patients with adolescent idiopathic scoliosis needing clinical observation. As the prevalence of adolescent idiopathic scoliosis increases, scoliosis screening should be continued as a routine health service in schools or by general practitioners if there is no scoliosis screening policy.

Analyzing Binary Outcome Data with Small Clusters: A Simulation Study

Binary outcome data with small clusters often arise in medical studies and the size of clusters might be informative of the outcome. The authors conducted a simulation study to examine the performance of a range of statistical methods. The simulation results showed that all methods performed mostly comparable in the estimation of covariate effects. However, the standard logistic regression approach that ignores the clustering encountered an undercoverage problem when the degree of clustering was nontrivial. The performance of random-effects logistic regression approach tended to be affected by low disease prevalence, relatively small cluster size, or informative cluster size.

Mapping and direct valuation: do they give equivalent EQ-5D-5L index scores?

ObjectiveUtility values of health states defined by health-related quality of life instruments can be derived from either direct valuation (‘valuation-derived’) or mapping (‘mapping-derived’). This study aimed to compare the utility-based EQ-5D-5L index scores derived from the two approaches as a means to validating the mapping function developed by van Hout et al for the EQ-5D-5L instrument.MethodsThis was an observational study of 269 breast cancer patients whose EQ-5D-5L index scores were derived from both methods. For comparing discriminatory ability and responsiveness to change, multivariable regression models were used to estimate the effect sizes of various health indicators on the index scores. Agreement and test-retest reliability were examined using intraclass correlation coefficient (ICC). Whenever appropriate, the 90xa0% confidence intervals (90xa0% CI) were compared to predefined equivalence margins.ResultsThe mean difference in and ICC between the valuation- and mapping-derived EQ-5D-5L index scores were 0.015 (90xa0% CIu2009=u20090.006 to 0.024) and 0.915, respectively. Discriminatory ability and responsiveness of the two indices were equivalent in 13 of 15 regression analyses. However, the mapping-derived index score was lower than the valuation-derived index score in patients experiencing extreme health problems, and the test-retest reliability of the former was lower than the latter, for example, their ICCs differed by 0.121 (90xa0% CIu2009=u20090.051 to 0.198) in patients who reported no change in performance status in the follow-up survey.ConclusionThis study provided the first evidence supporting the validity of the mapping function for converting EQ-5D-5L profile data into a utility-based index score.