RESUMO
El diseño ideal para estimar los efectos de intervenciones es un ensayo clínico con asignación aleatoria (AA), en los que esta AA a la intervención, equilibra características observadas y no observadas de los sujetos que componen los grupos en estudio. Si no se puede realizar AA, existe una alternativa para controlar sesgos, conocida como puntajes de propensión (PP). Los PP son útiles para estimar el efecto de una intervención o exposición sobre un resultado en estudios observacionales, especialmente cuando se estima la posibilidad que existan sesgos de confusión y selección debido a una asignación de tratamiento no aleatoria. Son una técnica de emparejamiento avanzada de variables de confusión como edad, sexo, etc., permitiendo que se puedan comparar pacientes de edad similar en grupos de intervención y de comparación. Esto, se puede complejizar si se decide agregar muchas variables al proceso de emparejamiento (edad, sexo, etnia, nivel educacional), puesto que será difícil encontrar coincidencias exactas para los sujetos. Entonces, los PP resuelven este problema de dimensionalidad comprimiendo factores relevantes en un solo puntaje, así, los pacientes con PP similares se comparan entre los grupos de intervención y comparación. El objetivo de este manuscrito fue generar un documento de estudio referente al uso de los PP y su rol en investigación quirúrgica.
The ideal design to estimate the effects of interventions, is a randomized clinical trial (RCT), in which the random allocation (RA) to the intervention allows balancing the observed and unobserved characteristics of the subjects that make up the study groups. In situations where RA cannot be performed, there is an alternative to bias control, which is known as propensity score (PS). This tool is useful for estimating the effect of an intervention or exposure on an outcome in observational studies, especially when it is estimated that confounding and selection biases may exist due to an unintended intervention RA. Is an advanced matching technique for confounding variables such as age, sex, etc., which allows patients of similar age to be compared in the intervention and comparison groups. This can become more complex if it is decided to add many variables to the matching process (age, sex, ethnicity, educational level), since it will be difficult to find exact matches for the subjects under study. PS then solves this dimensionality problem by compressing the relevant factors into a single score, such that patients with similar PS are compared between the intervention and comparison groups. The aim of this manuscript was to generate a study document referring to some issues of the use of PS in surgical research.
RESUMO
@#Objective To explore the evolving strategies and compare perioperative and long-term outcomes of tricuspid valve replacement (TVR) in recent 20 years in our hospital. Methods Between 1998 and 2018, the clinical data of 608 patients who underwent TVR at the Department of Cardiac Surgery, Guangdong Provincial People’s Hospital were retrospectively analyzed. There were 201 males and 407 females, with a median age of 47.0 (36.0, 57.0) years. Patients were divided into a biological tricuspid valve (BTV, n=427) group and a mechanical tricuspid valve (MTV, n=181) group. Propensity score matching was used to balance the baseline difference. Surgical strategy evolving, postoperative and long-term outcomes were analyzed between the two groups. Results Since 2008, the usage ratio of biological valves was significantly higher than that of mechanical valves. Seventy-nine (13.0%) patients died in hospital after TVR. Before propensity score matching, the postoperative mortality of the BTV group was higher than that of the MTV group (15.2% vs. 7.7%, P=0.012), and there was no statistical difference between the two groups after matching (10.4% vs. 7.2%, P=0.372). The duration of postoperative ventilator support in the BTV group was longer than that in the MTV group [22.0 (15.0, 37.0) h vs. 19.0 (11.0, 27.0) h, P=0.003], and the incidence of postoperative dialysis and re-thoracotomy exploring for bleeding was higher in the BTV group (8.9 % vs. 2.8%, 9.4% vs. 6.6%, respectively). However, there was no statistical difference in mortality after matching. The median follow-up time of discharged patients was 101.0 (65.0, 147.0) months, ranged from 1 to 265 months, and the follow-up rate was 82.2%. During the follow-up period, there were 101 deaths (19.1%) of whom 68 were from the BTV group and 33 from the MTV group. The survival rates at 1 year, 5 years, 10 years, 15 years and 20 years of all patients were 85.0% (95%CI 82.2-87.9), 78.9% (95%CI 75.7-82.4), 71.1% (95%CI 67.3-75.3), 59.7% (95%CI 54.2-65.6) and 51.7% (95%CI 43.3-60.7), and there was no statistical difference between the two groups after matching (P=0.46). The median time of tricuspid valve failure was 84.0 (54.0-111.0) months. Conclusion TVR is associated with high perioperative risks. There is no statistical difference in early mortality and long-term survival between biological and mechanical valve, while patients with mechanical valve has higher risk of re-operation for valve failure. Therefore, the type of prosthetic valve can be selected according to the patients' age, physical condition and the doctors’ experience.
RESUMO
PURPOSE: The aim of this study was to compare ESD only and subsequent Laparoscopy-assisted gastrectomy (LAG) patients for EGC through propensity score matching analysis. METHODS: This study was a retrospective review of the records of 46 consecutive patients with gastric cancer who underwent LAG after ESD from 2009, September to 2014, September, and propensity matching analysis was performed with 92 patients who underwent LAG without ESD as a control group. Subgroup analysis was performed with the interval of subsequent laparoscopic gastrectomy and endoscopic resection (within 2 weeks, 2~4 weeks, above 4 weeks). RESULTS: There were no significant differences in age, gender, body mass index, comorbidity, previous abdominal surgery, and location of the lesion or clinical stage between the two groups. Early postoperative outcomes including postoperative complications and postoperative hospital stay were not significantly different between the two groups. According to subgroup analysis with the interval of laparoscopic gastrectomy and endoscopic resection, there were no significant differences in early outcomes in three groups. However, only early postoperative complication rate was significantly higher in patients who received LAG more than 4 weeks after ESD (0 versus 4.76 versus 40 per cent; p=0.0032). CONCLUSION: We analyzed the influence of ESD on subsequent LAG using propensity score matching to reduce the bias. However, we found that ESD might induce inflammation for a significant duration, but ESD had little influence on early postoperative outcome of LAG.