[Show abstract][Hide abstract] ABSTRACT: OBJECTIVE: To evaluate the benefit of mass eradication of Helicobacter pylori infection in reducing premalignant gastric lesions. DESIGN: Mass eradication of H pylori infection was started from 2004 for a Taiwanese population with prevalent H pylori infection, who were >30 years of age. Participants positive for the (13)C-urea breath test underwent endoscopic screening and 1-week clarithromycin-based triple therapy. For subjects whose initial treatment failed, 10-day levofloxacin-based triple therapy was administered. The main outcome measures were changes in the prevalence of H pylori infection and premalignant gastric lesions, and changes in the incidence of premalignant gastric lesions and gastric cancer before (1995-2003) and after (2004-2008) chemoprevention using various comparators. RESULTS: The reduction in H pylori infection was 78.7% (95% CI 76.8% to 80.7%), and the estimated incidence of re-infection/recrudescence was 1% (95% CI 0.6% to 1.4%) per person-year. The effectiveness of reducing the incidence of gastric atrophy resulting from chemoprevention was significant at 77.2% (95% CI 72.3% to 81.2%), while the reduction in intestinal metaplasia was not significant. Compared with the 5-year period before chemoprevention and in the absence of endoscopic screening, the effectiveness in reducing gastric cancer incidence during the chemoprevention period was 25% (rate ratio 0.753, 95% CI 0.372 to 1.524). The reduction in peptic ulcer disease was 67.4% (95% CI 52.2% to 77.8%), while the incidence of oesophagitis was 6% (95% CI 5.1% to 6.9%) after treatment. CONCLUSIONS: Population-based eradication of H pylori infection has led to a significant reduction in gastric atrophy at the expense of increased oesophagitis. The ultimate benefit in reducing gastric cancer incidence and its mortality should be validated by a further long-term follow-up. TRIAL REGISTRATION NUMBER: NCT00155389.
[Show abstract][Hide abstract] ABSTRACT: To estimate the long-term (29-year) effect of mammographic screening on breast cancer mortality in terms of both relative and absolute effects.
This study was carried out under the auspices of the Swedish National Board of Health and Welfare. The board determined that, because randomization was at a community level and was to invitation to screening, informed verbal consent could be given by the participants when they attended the screening examination. A total of 133 065 women aged 40-74 years residing in two Swedish counties were randomized into a group invited to mammographic screening and a control group receiving usual care. Case status and cause of death were determined by the local trial end point committees and, independently, by an external committee. Mortality analysis was performed by using negative binomial regression.
There was a highly significant reduction in breast cancer mortality in women invited to screening according to both local end point committee data (relative risk [RR] = 0.69; 95% confidence interval: 0.56, 0.84; P < .0001) and consensus data (RR = 0.73; 95% confidence interval: 0.59, 0.89; P = .002). At 29 years of follow-up, the number of women needed to undergo screening for 7 years to prevent one breast cancer death was 414 according to local data and 519 according to consensus data. Most prevented breast cancer deaths would have occurred (in the absence of screening) after the first 10 years of follow-up.
Invitation to mammographic screening results in a highly significant decrease in breast cancer-specific mortality. Evaluation of the full impact of screening, in particular estimates of absolute benefit and number needed to screen, requires follow-up times exceeding 20 years because the observed number of breast cancer deaths prevented increases with increasing time of follow-up.
[Show abstract][Hide abstract] ABSTRACT: Useful predictive models for identifying patients at high risk of bacteremia at the emergency department (ED) are lacking. This study attempted to provide useful predictive models for identifying patients at high risk of bacteremia at the ED.
A prospective cohort study was conducted at the ED of a tertiary care hospital from October 1 to November 30, 2004. Patients aged 15 years or older, who had at least two sets of blood culture, were recruited. Data were analyzed on selected covariates, including demographic characteristics, predisposing conditions, clinical presentations, laboratory tests, and presumptive diagnosis, at the ED. An iterative procedure was used to build up a logistic model, which was then simplified into a coefficient-based scoring system.
A total of 558 patients with 84 episodes of true bacteremia were enrolled. Predictors of bacteremia and their assigned scores were as follows: fever greater than or equal to 38.3°C [odds ratio (OR), 2.64], 1 point; tachycardia greater than or equal to 120/min (OR, 2.521), 1 point; lymphopenia less than 0.5×10(3)/μL (OR, 3.356), 2 points; aspartate transaminase greater than 40IU/L (OR, 2.355), 1 point; C-reactive protein greater than 10mg/dL (OR, 2.226), 1 point; procalcitonin greater than 0.5 ng/mL (OR, 3.147), 2 points; and presumptive diagnosis of respiratory tract infection (OR, 0.236), -2 points. The area under the receiver operating characteristic curves of the original logistic model and the simplified scoring model using the aforementioned seven predictors and their assigned scores were 0.854 (95% confidence interval, 0.806-0.902) and 0.845 (95% confidence interval, 0.798-0.894), respectively.
This simplified scoring system could rapidly identify high-risk patients of bacteremia at the ED.
Journal of microbiology, immunology, and infection = Wei mian yu gan ran za zhi 06/2011; 44(6):449-55. · 1.63 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: This study is to identify an optimal cut-off for two-stage breast cancer screening making allowance for variation of the baseline incidence rate and utility values between sensitivity and specificity.
We used data from a two-stage breast cancer screening of Taiwanese women aged 50-69 years for whom risk stratification was based on a composite risk score (conventional risk factors); subjects with a risk score greater than the cut-off score were screened using mammography. The Bayesian posterior risk for breast cancer was computed by incorporation of the baseline incidence rate and the risk score. Bayes' maximum utility decision rule was then developed to determine the optimal screening cut-off.
With a risk score of -9 applied to the current two-stage breast cancer screening programme, we could detect one breast cancer case for every 1406 women. Given different predetermined risks, the selected cut-offs were -9 for 1:1200, -8 for 1:800, -4 for 1:600, -1 for 1:400 and 3 for 1:200 for women aged 50-59 years. When the regret utility ratio of positive predictive value to negative predictive value was set at 1:10, the specificity and sensitivity were 58.5% and 70.4%, respectively, and the optimal cut-off was -5.5. When the ratio was set at 10:1, the sensitivity and specificity were 75.5% and 57.1%, respectively, and the optimal cut-off value was -7.5.
This study demonstrates that Bayes' maximum utility decision rule can be used to determine optimal cut-off values for two-stage breast cancer screening in countries or areas with lower or intermediate incidence of breast cancer.
Journal of Evaluation in Clinical Practice 12/2010; 16(6):1345-52. · 1.58 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Vaccination is an effective method for controlling avian influenza (AI), especially in countries with endemic infection. This study conducted a Bayesian meta-analysis to evaluate the efficacy of AI vaccines in chickens. We included both inactivated and recombinant fowlpox virus expressing H5 (rFPV-H5) vaccine studies that used specific-pathogen-free chickens where outcomes against the H5N1 or H5N2 AI viruses were measured. Vaccine efficacy was evaluated by protection from mortality, protection from morbidity, reductions in virus isolation from the respiratory tract, and reductions in virus isolation from the cloaca. The efficacies for homologous inactivated vaccines by those four outcomes were 92% (95% confidence interval 90%-95%), 94% (91%-96%), 54% (50%-58%), and 88% (84%-91%), respectively. Corresponding figures for heterologous inactivated vaccines were 68% (63%-73%), 78% (74%-81%), 24% (16%-31%), and 71% (64%-77%); and efficacies for rFPV-H5 vaccine were 97% (94%-99%), 93% (90%-94%), 21% (14%-27%), and 78% (72%-84%), respectively. Although those vaccines protect chickens from morbidity and mortality, virus shedding would be an important biosecurity issue for further AI endemic control.
[Show abstract][Hide abstract] ABSTRACT: Community-associated methicillin-resistant Staphylococcus aureus (MRSA) has become an important pathogen in community and nosocomial infections. The impact of these emerging MRSA strains on mortality in adult patients with community-onset S aureus bacteremia remains uncertain. We defined community strain MRSA (CoSt-MRSA) and hospital strain MRSA (HoSt-MRSA) according to the results of staphylococcus cassette chromosome mec (SCCmec) molecular typing: CoSt-MRSA isolates had SCCmec type IV or V genes, and HoSt-MRSA isolates had SCCmec type I, II, or III genes. We quantitatively evaluated the impact of the MRSA strain on mortality in patients with CoSt-MRSA or HoSt-MRSA bacteremia by comparison with mortality in patients with methicillin-susceptible S aureus (MSSA) bacteremia.We studied an observational cohort of 500 patients with MSSA bacteremia, 111 patients with CoSt-MRSA, and 133 patients with HoSt-MRSA bacteremia from January 1, 2001, through December 31, 2007. The 90-day cumulative probability of survival in patients with MSSA, CoSt-MRSA, and HoSt-MRSA bacteremia was 71%, 70%, and 55%, respectively (p = 0.014, by Wilcoxon rank-sum test).Compared to patients with MSSA bacteremia, patients with HoSt-MRSA bacteremia were associated with an increased risk of mortality in the first multivariate analysis model adjusting for all potential confounders (hazard ratio [HR], 1.525; 95% confidence interval [CI], 1.091-2.131), in the second model adjusting for all confounders except acute severity of bacteremia (HR, 1.489; 95% CI, 1.071-2.070), and in stratified analysis in patients with low Charlson comorbidity scores (score 0-2) (HR, 3.093; 95% CI, 1.507-6.350).Compared to patients with MSSA bacteremia, patients with CoSt-MRSA bacteremia did not show significant differences in mortality rate in the 2 multivariate analysis models (first model: HR, 1.106; 95% CI, 0.748-1.637; second model: HR, 1.028; 95% CI, 0.697-1.516) or in stratified analysis (HR, 1.092; 95% CI, 0.539-2.214).In conclusion, using MSSA as reference, traditional hospital strain MRSA had a higher impact on bacteremia mortality than community strain MRSA.
Medicine 09/2010; 89(5):285-94. · 4.35 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Evaluation of long-term effectiveness of population-based breast cancer service screening program in a small geographic area
may suffer from self-selection bias and small samples. Under a prospective cohort design with exposed and non-exposed groups
classified by whether women attended the screen upon invitation, we proposed a Bayesian acyclic graphic model for correcting
self-selection bias with or without incorporation of prior information derived from previous studies with an identical screening
program in Sweden by chronological order and applied it to an organized breast cancer service screening program in Pirkanmaa
center of Finland. The relative mortality rate of breast cancer was 0.27 (95% CI 0.12–0.61) for the exposed group versus the
non-exposed group without adjusting for self-selection bias. With adjustment for selection-bias, the adjusted relative mortality
rate without using previous data was 0.76 (95% CI 0.49–1.15), whereas a statistically significant result was achieved [0.73
(95% CI 0.57–0.93)] with incorporation of previous information. With the incorporation of external data sources from Sweden
in chronological order, adjusted relative mortality rate was 0.67 (0.55–0.80). We demonstrated how to apply a Bayesian acyclic
graphic model with self-selection bias adjustment to evaluating an organized but non-randomized breast cancer screening program
in a small geographic area with a significant 27% mortality reduction that is consistent with the previous result but more
precise. Around 33% mortality was estimated by taking previous randomized controlled data from Sweden.
KeywordsBreast cancer screening-Self-selection bias-Bayesian acyclic graphic model-Mortality reduction
Breast Cancer Research and Treatment 06/2010; 121(3):671-678. · 4.20 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: To investigate the prognostic effects of the serum total cholesterol (TC) levels on long-term functional outcomes in patients with first-time noncardioembolic ischemic stroke.
Patients (N=109) with first-time ischemic stroke.
Serial Barthel Index (BI) scores at onset; 2 weeks; and 1, 2, 4, and 6 months after stroke. We analyzed the impact of the serum TC level and other clinical factors on the repeated measurements of BI scores at these 6 time points by using a linear mixed regression model.
Taking correlation across repeated measurement of BI scores, the TC level, baseline BI, follow-up time, and infarct size were identified as significant predictors for serial BI scores. Higher TC levels correlated with better functional outcomes. A 1-unit (mmol/L) increase in the TC caused a 3.12 (95% confidence interval [CI], .79-5.46) increase in the BI score after controlling for other clinical factors such as age, baseline functional status, and size of infarct. An elevation of 1 unit of baseline BI led to a .49 increase (95% CI, .38-.59) per unit in subsequent BI scores. A small infarct (<1cm) had higher BI scores than larger infarct by 9.09 (95% CI, 2.03-16.16).
The serum TC level measured at the acute stage of noncardioembolic ischemic stroke is an independent predictor for long-term functional outcomes.
Archives of physical medicine and rehabilitation 06/2010; 91(6):913-8. · 2.18 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: To establish a predictive model for evaluating improvement in patients with idiopathic sudden sensorineural hearing loss (ISSHL).
Bayesian cure rate model.
Tertiary referral center.
Two hundred patients whose conditions were diagnosed as ISSHL from January 2001 to April 2007 were enrolled to build a model and to train relevant parameters for prediction.
The time to improvement and potential predictive factors were collected for analysis. Established factors and significant variables in the univariate analysis were included in the final model. A Bayesian approach with the WinBUGS program was applied to predict the median and 95% confidence intervals (CIs) for the time to improvement, long-term probability of improvement, and improvement probabilities at specific days.
The significant predictors in the final model include distortion product otoacoustic emission, auditory brainstem response, vestibular evoked myogenic potential, and audiometric types. The overall results predicted by the different combinations of covariates were summarized and organized in an Access program file that is convenient for clinical application. The results of area under the receiver operating characteristic curves at the 7th, 14th, and 30th days were 0.709 (95% CI, 0.692-0.717), 0.752 (95% CI, 0.737-0.753), and 0.807 (95% CI, 0.788-0.811), respectively. It showed that predictive validity, particularly at 1 month or so, is satisfactory.
By using the cure rate model under the context of a Bayesian survival analysis, we first identified auditory objective factors as significant predictors of improvement of ISSHL patients and further predicted the time to improvement with these correlates. The model showed a satisfactory predictive validity, particularly for 1-month individual prediction, which prompted one to make an individual prediction with an available Access program.
Otology & neurotology: official publication of the American Otological Society, American Neurotology Society [and] European Academy of Otology and Neurotology 04/2010; 31(3):385-93. · 1.44 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Previous studies using linear regression analysis have shown that age, weight, gender, and the site of operation affect intravenous patient-controlled analgesia (IVPCA) narcotic use. However, there are inconsistent observations in the literature. The authors postulate that patient variables could have different effects at various doses of narcotics. To test this hypothesis, the authors analyzed the effect of patient variables on increasing doses of IVPCA narcotic with quantile regression.
The authors collected retrospective data from 1,782 patients who received IVPCA for a minimum of 3 days after surgery. The authors used stepwise linear regression model to identify variables that significantly affected the total IVPCA requirements. Quantile regression model was further applied to assess the effects of selected variables on the ascending percentile of IVPCA narcotic use.
Gender, age, body weight, cancer, and surgical site were identified as significant predictors for IVPCA demand. Body weight had the most and cancer had the least significant effects on total IVPCA demands. The results of quantile regression model revealed that the determinants under consideration varied with different percentiles of IVPCA demand. The patient variables correlated with IVPCA narcotic use differently when the dose exceeded the seventieth to eightieth percentiles compared with other percentiles of narcotic use.
The authors' findings highlight the heterogeneous postoperative pain requirements among patients and the consequent complex process of efficiently managing postoperative pain.
[Show abstract][Hide abstract] ABSTRACT: The efficacy of acupressure in relieving pain has been documented; however, its effectiveness for chronic headache compared to the muscle relaxant medication has not yet been elucidated. To address this, a randomized, controlled clinical trial was conducted in a medical center in Southern Taiwan in 2003. Twenty-eight patients suffering chronic headache were randomly assigned to the acupressure group (n = 14) or the muscle relaxant medication group (n = 14). Outcome measures regarding self-appraised pain scores (measured on a visual analogue scale; VAS) and ratings of how headaches affected life quality were recorded at baseline, 1 month after treatment, and at a 6-month follow-up. Pain areas were recorded in order to establish trigger points. Results showed that mean scores on the VAS at post-treatment assessment were significantly lower in the acupressure group (32.9+/-26.0) than in the muscle relaxant medication group (55.7+/-28.7) (p = 0.047). The superiority of acupressure over muscle relaxant medication remained at 6-month follow-up assessments (p = 0.002). The quality of life ratings related to headache showed similar differences between the two groups in the post treatment and at six-month assessments. Trigger points BL2, GV20, GB20, TH21, and GB5 were used most commonly for etiological assessment. In conclusion, our study suggests that 1 month of acupressure treatment is more effective in reducing chronic headache than 1 month of muscle relaxant treatment, and that the effect remains 6 months after treatment. Trigger points help demonstrate the treatment technique recommended if a larger-scale study is conducted in the future.
The American Journal of Chinese Medicine 01/2010; 38(1):1-14. · 2.63 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: To investigate the prevalence and correlates of somatic and autonomic neuropathies concurrently in prediabetic and diabetic subjects in the general population.
Nerve conduction and heart rate variability studies were used for detecting both types of diabetic neuropathy. Of 1,576 inhabitants, 143 persons were identified as prediabetic or having diabetic mellitus. One hundred and thirty-three persons completed a nerve conduction test and 118 persons a valid heart rate variability test. Significant correlates responsible for diabetic neuropathies were also collected.
The prevalence rates of definite, probable and no somatic neuropathy were 9.0% (12/133), 20.3% (27/133) and 70.7% (94/133), respectively. The corresponding figures for autonomic neuropathy were 14.4% (17/118), 54.2% (64/118) and 31.4% (37/118). The prevalence rate for the presence of both neuropathies was 16.9%. In a univariate analysis, age, renal insufficiency, HbA1c and fasting glucose level were significantly associated with somatic neuropathy whereas only systolic blood pressure was statistically significantly associated with autonomic neuropathy. In a multivariate analysis, systolic blood pressure and fasting glucose level were positively associated with somatic neuropathy and systolic blood pressure remained statistically significant for autonomic neuropathy. Conclusions: The prevalence of autonomic neuropathy is twofold compared with somatic neuropathy. Concurrence of the two kinds of neuropathy was approximately half of sole somatic neuropathy and a quarter of sole autonomic neuropathy. Systolic blood pressure and fasting glucose level were related to somatic neuropathy whereas only systolic blood pressure was correlated with autonomic neuropathy.
[Show abstract][Hide abstract] ABSTRACT: Health policy makers are usually stranded by the complicated infrastructure and intensive computation related to economic evaluation.
It is therefore valuable to develop a computer-aided tool to help health personnel to perform economic evaluation with ease.
The infrastructure for economic evaluation was first designed. Markov process with micro-simulation was applied to model the disease natural history or lifetime sequale to project the effectiveness by comparing all possible decisions. All the essential elements of economic evaluation together with sensitivity analysis are encoded in this computer-aided software written with SAS Screen Control Language in user-defined menu style. ILLUSTRATION: Screening versus no screening for colorectal cancer was used as an example.
The computer-aided model for economic evaluation was developed in this study. It is anticipated that the flexibility and user-defined menu style facilitate the wide application of economic evaluation to health care intervention program.
Journal of Evaluation in Clinical Practice 10/2009; 15(5):797-803. · 1.58 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: To assess cost-effectiveness of hepatitis B virus (HBV) vaccination strategies from health care payer and societal perspectives, focusing on the long-term effect, in Taiwan where prevalence of HBV and Hepatitis B e Antigen (HBeAg) is high.
A decision analysis was performed to compare total costs and effectiveness between two vaccination strategies: universal vaccination and no-vaccination. The Markov process was defined as a series of states including acute HBV infection, asymptomatic carrier, chronic hepatitis, compensated and decompensated liver cirrhosis, hepatoma, and death. Direct and indirect costs were also imputed based on estimates. The incremental cost-effectiveness ratio (ICER) per life-year gained and quality-adjusted life years gained were calculated at a 3% discount rate. By assigning a series of specific distributions to each parameter, a probabilistic cost-effective analysis using Monte Carlo simulation was conducted to yield 5000 ICER replicates.
The effectiveness of a universal vaccination program for reducing hepatocellular carcinoma cases and deaths was approximately 86%. The average life years gained per subject as a result of such a universal vaccination was 3.9. The vaccination program dominated over a no-vaccination program (less cost and more effectiveness).
A universal vaccination program against hepatitis B infection is not only effective for reducing long-term sequelae but is also a cost-saving primary preventive strategy, which supports a universal infant immunization in endemic area with high prevalence of HBV and HBeAg.
[Show abstract][Hide abstract] ABSTRACT: We aimed to quantify the mortality reduction by which the early detection of Parkinson's disease (PD) within a community-based study could reduce the number of advanced cases.
Data used in this study were derived from two community-based surveys and from a clinical series of PD cases identified from a medical centre. The cumulative survival by Hoehn-Yahr (H-Y) scale was estimated and the mortality reduction derived from a community-based survey was predicted.
A total of 117 PD patients were detected through two community-based approaches. By comparing the H-Y stage distribution of screen-detected cases with those from the clinical series, a 22.5% excess in the number of early PD (H-Y stage I and stage II) were identified with screening. The risk ratios of being H-Y stage III or severe for community-based detected cases versus clinical series were 0.49 (95% confidence interval: 0.30-0.78). The total death rate adjusted by H-Y stage distribution was 21% and 28% for cases from community and clinical series, respectively.
The present study revealed that early detection of PD through a community-based survey may reduce 51% incidence of stage III or more severe PD at diagnosis, leading to a 25% reduction in mortality.
Journal of Evaluation in Clinical Practice 09/2009; 15(4):587-91. · 1.58 Impact Factor