The risk-adjusted Bernoulli cumulative sum (CUSUM) chart developed by Steiner et al. (2000) is an increasingly popular tool for monitoring clinical and surgical performance. In practice, however, use of a fixed control limit for the chart leads to quite variable in-control average run length (ARL) performance for patient populations with different risk score distributions. To overcome this problem, the simulation-based dynamic probability control limits (DPCLs) patient-by-patient for the risk-adjusted Bernoulli CUSUM charts is determined in this study. By maintaining the probability of a false alarm at a constant level conditional on no false alarm for previous observations, the risk-adjusted CUSUM charts with DPCLs have consistent in-control performance at the desired level with approximately geometrically distributed run lengths. Simulation results demonstrate that the proposed method does not rely on any information or assumptions about the patients' risk distributions. The use of DPCLs for risk-adjusted Bernoulli CUSUM charts allows each chart to be designed for the corresponding particular sequence of patients for a surgeon or hospital. The effect of estimation error on performance of risk-adjusted Bernoulli CUSUM chart with DPCLs is also examined. Our simulation results show that the in-control performance of risk-adjusted Bernoulli CUSUM chart with DPCLs is affected by the estimation error. The most influential factors are the specified desired in-control average run length, the Phase I sample size and the overall adverse event rate. However, the effect of estimation error is uniformly smaller for the risk-adjusted Bernoulli CUSUM chart with DPCLs than for the corresponding chart with a constant control limit under various realistic scenarios. In addition, there is a substantial reduction in the standard deviation of the in-control run length when DPCLs are used. Therefore, use of DPCLs has yet another advantage when designing a risk-adjusted Bernoulli CUSUM chart. These researches are results of joint work with Dr. William H. Woodall (Department of Statistics, Virginia Tech). Moreover, DPCLs are adapted to design the risk-adjusted CUSUM charts for multiresponses developed by Tang et al. (2015). It is shown that the in-control performance of the charts with DPCLs can be controlled for different patient populations because these limits are determined for each specific sequence of patients. Thus, the risk-adjusted CUSUM chart for multiresponses with DPCLs is more practical and should be applied to effectively monitor surgical performance by hospitals and healthcare practitioners. This research is a result of joint work with Dr. William H. Woodall (Department of Statistics, Virginia Tech) and Mr. Justin Loda (Department of Statistics, Virginia Tech). / Ph. D.

[pt] Os gráficos de controle de S(2) são ferramentas fundamentais amplamente utilizados para monitoramento da dispersão do processo em aplicações de CEP. O desempenho na Fase II de diferentes tipos de gráficos de controle, incluindo o gráfico de S(2), com parâmetros desconhecidos pode ser significativamente diferente do desempenho nominal por causa do efeito da estimação de parâmetros. Nos anos mais recentes, este efeito tem sido abordado predominantemente sob a perspectiva condicional, que considera a variabilidade das estimativas de parâmetros obtidas a partir de diferentes amostras de referência da Fase I em vez das típicas medidas de desempenho baseadas na distribuição marginal (incondicional) do número de amostras até o sinal (Run Length-RL), como sua média. À luz dessa nova perspectiva condicional, a análise do desempenho da Fase II e do projeto de gráficos de controle é frequentemente realizada usando o Exceedance Probability Criterion para a média da distribuição condicional do RL (CARL 0), isto é, o critério que garante uma alta probabilidade de que CARL 0 seja pelo menos um valor mínimo tolerado e especificado. Intervalos de tolerância para variâncias amostrais são úteis quando o maior interesse está focado na precisão dos valores de uma característica de qualidade, e podem ser usados na tomada de decisões sobre a aceitação de lotes por amostragem. Motivado pelo fato de que estes intervalos, especificamente no caso dos intervalos bilaterais, não foram abordados na literatura, limites bilaterais de tolerância exatos e aproximados de variâncias amostrais são derivados e apresentados neste trabalho. A relação matemática-estatística entre o intervalo de tolerância para a variância amostral e o Exceedance Probability (função de sobrevivência) da CARL 0 do gráfico de S(2) com parâmetro estimado é reconhecida, destacada e usada neste trabalho de tal forma que o estudo do desempenho na Fase II e o projeto desse gráfico pode ser baseado no intervalo de tolerância para a variância amostral, e vice-versa. Os trabalhos sobre o desempenho e projeto do gráfico de S(2) com parâmetro estimado focaram-se em apenas uma perspectiva (incondicional ou condicional) e consideraram somente um tipo de gráfico (unilateral superior ou bilateral). A existência de duas perspectivas e dois tipos de gráficos poderia ser confusa para os usuários. Por esse motivo, o desempenho e o projeto do gráfico de S(2) de acordo com essas duas perspectivas são comparados, considerando cada tipo de gráfico. Da mesma forma, esses dois tipos de gráficos também são comparados para cada perspectiva. Alguns resultados importantes relacionados ao projeto do gráfico de S(2), que ainda não estão disponíveis na literatura, foram necessários e obtidos neste trabalho para fornecer um estudo comparativo completo que permita aos usuários estarem cientes das diferenças significativas entre as duas perspectivas e os dois tipos de gráficos para tomar decisões informadas sobre a escolha do projeto do gráfico de S(2). Além disso, dado que a distribuição condicional do RL é em geral fortemente enviesada à direita, a mediana e alguns quantis extremos desta distribuição são propostos como medidas de desempenho complementares à sua tradicional média (CARL 0). Finalmente, algumas recomendações práticas são oferecidas. / [en] The S(2) control charts are fundamental tools widely used to monitor the process dispersion in applications of Statistical Process Monitoring and Control. Phase II performance of different types of control charts, including the S(2) chart, with unknown process parameters may be significantly different from the nominal performance due to the effect of parameter estimation. In the last few years, this effect has been addressed predominantly under the conditional perspective, which considers the variability of parameter estimates obtained from different Phase I reference samples instead of the traditional unconditional performance measures based on the marginal (unconditional) run length (RL) distribution, such as the unconditional average run length. In light of this new conditional perspective, the analysis of the Phase II performance and design of control charts is frequently undertaken using the Exceedance Probability Criterion for the conditional (given the parameter estimates) in-control average run length (CARL 0), that is, the criterion that ensures a high probability that the CARL 0 is at least a specified minimum tolerated value. Tolerance intervals for sample variances are useful when the main concern is the precision of the values of the quality characteristic and then they can be used in decision-making on lot acceptance sampling. Motivated by the fact that these tolerance intervals, specifically in the case of the two-sided intervals, have not been addressed in the literature so far, exact and approximate two-sided tolerance limits for the population of sample variances are derived and presented in this work. The mathematical-statistical relationship between tolerance interval for the sample variance and the exceedance probability (survival probability) of the CARL 0 for the S(2) control chart with estimated parameter is recognized, highlighted and used in this work in such a way that the study of the Phase II performance and design of this chart can be based on tolerance interval for the sample variance, and vice versa. Works on performance and design of S(2) chart with estimated parameter generally focused on only one perspective (either unconditional or conditional) and considered only one type of chart (either upper one-sided chart or two-sided chart). The existence of both perspectives and two types of charts may be confusing for practitioners. For that reason, the performance and design of S(2) control chart according to these two perspectives are compared, considering each type of chart. Similarly, these two types of charts are also compared for each perspective. Some important results related to the S(2) chart design, which are not yet available in the literature, were required and obtained in this work to provide a comprehensive comparative study that enables practitioners to be aware of the significant differences between these two perspectives and the two types of charts so that proper informed decisions about the chart design to choose can be made. Furthermore, because the conditional RL distribution is usually highly rightskewed, the median and some extreme quantiles of the conditional RL distribution are proposed as complementary performance measures to the customary mean (CARL 0). Finally, some practical recommendations are offered.