Explore chapters and articles related to this topic
Designing and Delivering a DTx Clinical Research Program: No Need to Re-invent the Wheel
Published in Oleksandr Sverdlov, Joris van Dam, Digital Therapeutics, 2023
Colin A. Espie, Alasdair L. Henry
Case-control studies permit two existing but differing groups to be compared, for instance, a group of patients (cases) compared to those without the condition (controls). Such studies allow differences in treatment outcomes between two groups that differ on a critical characteristic. As individuals are not randomized in case-control studies, there is an opportunity for bias. Matching controls to cases is a way to reduce bias using methods including propensity score matching (Rosenbaum, 2020). However, it can be challenging to identify appropriate control groups (Grimes and Schulz, 2002), and these may be historical controls (Grimes and Schulz, 2002). Case-control studies are also open to influence from confounding factors that prevent any causal associations from being inferred.
Causal Inference for Observational Studies/Real-World Data
Published in Harry Yang, Binbing Yu, Real-World Evidence in Drug Development and Evaluation, 2021
Matching refers to classifying subjects into small groups so that there are both treated and untreated subjects in each group and they are similar in terms of matching variables. Matching on propensity scores closely resembles a block randomized design, where the treatment assignment is random within each subgroup. Generally, propensity score matching is considered advantageous in the following perspectives. (1) It is more robust in the sense that it uses a nonparametric way to balance the covariate distributions between treated and untreated groups, which does not rely on parametric outcome models. (2) It resembles the randomization design, which is easily interpretable to a general audience. (3) It is more objective in the sense that the causal effect inference is conducted only after good matches are established and the outcome variable never enters the matching process. (4) Matching-based sensitivity analysis is well developed to assess the impact of hidden bias.
Big Data Analytics
Published in Shein-Chung Chow, Innovative Statistics in Regulatory Science, 2019
where . Propensity score matching is a powerful tool reducing the bias due to possible confounding effects. Sometimes, propensity score matching is also considered as post-study randomization as compared to a randomized clinical trial for reducing bias. For propensity score matching, it should be noted that data available for matching will decrease as the number of matching factors (potential confounding factors) increase.
Association between smoking and glycemic control in men with newly diagnosed type 2 diabetes: a retrospective matched cohort study
Published in Annals of Medicine, 2022
Hon-Ke Sia, Chew-Teng Kor, Shih-Te Tu, Pei-Yung Liao, Jiun-Yi Wang
Meanwhile, this study has a few limitations. First, selection bias might occur because health behaviours and characteristics between smokers and non-smokers could be different. In the study, propensity score matching has been used to reduce such a potential bias. Second, smokers might have a lower economic status and thus not be able to afford newer or additional medications. This could affect the findings. However, the National Health Insurance in Taiwan covers almost 100% of the population and provides easy access to medical services. Therefore, the treatment or change in medication during the follow-up between smokers and non-smokers was less affected by socioeconomic status. Third, our study included only men with T2DM owing to a very low prevalence of female smokers, which is attributable to the country's cultural background. Therefore, the generalisability of our findings to the whole population should be with caution. Fourth, as a retrospective study, the causal interpretation of this study was limited. Moreover, although propensity score matching was used to improve the comparability of participant characteristics and to minimise selection bias, other unmeasured factors such as dietary habits might affect the selection of controls.
Opioid-related adverse drug events in surgical patients: risk factors and association with clinical outcomes
Published in Expert Opinion on Drug Safety, 2022
Chin Hang Yiu, Danijela Gnjidic, Asad Patanwala, Ian Fong, David Begley, Kok Eng Khor, Joanne Rimington, Bernadette Bugeja, Jonathan Penm
This was the first study that identified risk factors for ORADEs in surgical patients, which also accounted for concurrent use of medications and opioid dosage. Specific risk factors for gastrointestinal and respiratory ORADEs were also identified. Propensity score matching was performed to identify the associations between ORADEs and clinical outcomes, which included LOS and 28-day readmission rate. However, as propensity score matching could be considered as a quasi-experimental design, randomization is limited [29]. Thus, the ability to conclude a causal association between ORADEs and clinical outcomes is lacking. In addition, ICD-10-AM codes were used to define ORADEs, it was possible that these codes were not related to the use of opioids and therefore led to an over-estimation of the prevalence of ORADEs. On the contrary, accurate documentations of ICD-10-AM codes by clinicians require adequate training and thus omissions in codes might sometimes occur [30], which under-represent patients experiencing ORADEs. The use of ICD-10-AM codes to define comorbidities is also a limitation as comorbidities were potentially under-reported [31].
Effects of foam rolling on hip pain in patients with hip osteoarthritis: a retrospective propensity-matched cohort study
Published in Physiotherapy Theory and Practice, 2022
Hisashi Ikutomo, Koutatsu Nagai, Keiichi Tagomori, Namika Miura, Kenichi Okamura, Takato Okuno, Norikazu Nakagawa, Kensaku Masuhara
The 115 patients in the FR and non-FR groups were retrospectively matched in a 1:1 ratio using a propensity-score algorithm to adjust for baseline differences between the two groups. Propensity-score matching is a well-validated statistical technique that creates comparable groups and allows for the accurate assessment of treatment effect (Stukel et al., 2007). We estimated the scores using a multivariable logistic regression model. Patients were matched for age, sex, body mass index (BMI), Kellgren and Lawrence grade in the involved limb, and hip pain VAS at baseline. Baseline characteristics and outcome measurements between the FR and non-FR groups were compared using the Mann-Whitney U test for continuous variables and the chi-squared test for categorical variables. Effects of interventions on outcome measurements in the groups were compared using split-plot design variance analysis (ANOVA). If the ANOVA showed significant interactions, Bonferroni post-hoc test was used to identify the mean differences. Additionally, each matched group was classified as a high or low group according to the radiographic progression of hip osteoarthritis. We defined high groups as baseline Kellgren and Lawrence grade of 2–4 in the involved limb. The changes in hip pain VAS were compared after the intervention in the four groups using Kruskal-Wallis and Bonferroni post-hoc tests. Statistical significance was set at the level of p < .05. All analyzes were performed using IBM SPSS software (IBM Japan, Tokyo, Japan).