File Name: statistical process control and quality improvement .zip
Statistical process control SPC is a method of quality control which employs statistical methods to monitor and control a process.
This study reports on a diabetes QI project in rural Guatemala whose primary aim was to improve glycemic control of a panel of adult diabetes patients. Formative research suggested multiple areas for programmatic improvement in ambulatory diabetes care. This project utilized the Model for Improvement and Agile Global Health, our organization's complementary healthcare implementation framework. A bundle of improvement activities were implemented at the home, clinic and institutional level.
Control charts of mean hemoglobin A1C HbA1C and proportion of patients meeting target HbA1C showed improvement as special cause variation was identified 3 months after the intervention began. Control charts for secondary process measures offered insights into the value of different components of the intervention. Intensity of home-based diabetes education emerged as an important driver of panel glycemic control. Statistical process control charts are a promising methodology for use with panels or registries of diabetes patients.
Quality improvement QI is a key element of strengthening health systems and improving diabetes care in low- and middle-income countries LMICs [ 1 , 2 ]. Statistical process control SPC is a set of powerful QI methods—which include the use of control charts—that can detect statistical changes in a healthcare process earlier than traditional research methodologies, including during the testing phase of an intervention [ 3 ].
There have been limited published applications of control charts in resource-limited settings or with diabetes panels [ 4 ]. In this study, we report on a small primary care QI project in rural Guatemala whose primary aim was to improve glycemic control in patients with Type 2 diabetes. Wuqu' Kawoq Maya Health Alliance is a non-governmental organization delivering primary healthcare in indigenous Maya areas in rural Guatemala.
Our clinical program is described in detail elsewhere [ 7 ]. In , we conducted a needs assessment of diabetes care in rural Guatemala [ 8 ], which suggested several areas for improvement. First, diabetes patients generally had limited health literacy on topics including the natural history of the disease, the role of diet and lifestyle, and prevention of common complications.
Second, due to economic difficulties and health system shortcomings, diabetes patients often sought care at multiple health facilities and lacked care continuity. Fourth, patients expressed strong preference for Mayan-language care delivery.
This project utilized the Model for Improvement [ 9 ] and Agile Global Health, [ 10 ] our organization's complementary healthcare implementation framework that emphasizes program flexibility, field experience for program managers, patient input into design and implementation elements, and cycles of improvement analogous to Plan, Do, Study, Act cycles. The theory of change for the intervention was visualized in a key driver diagram [ 9 ] in Fig. Key driver diagram for diabetes QI in rural Guatemala.
Primary drivers for the diabetes QI project were conceptualized at the Home, Clinic and Organizational levels. Improvement activities were structured to address each level. In October , a diabetes improvement team was formed consisting of the institution's chief medical officer, diabetes coordinator, clinical diabetes nurse, and diabetes educators.
The QI intervention, which began in October , addressed key drivers drawn from formative research. The primary activity at the home level was the implementation of a diabetes education program for patients and family members.
We previously had adapted a well-known Latino chronic disease curriculum for use in Mayan-speaking populations [ 11 ], and in October we initiated group-based education sessions. However, despite formative research on the feasibility of group education, actual attendance at group sessions was very low. In May , therefore, we re-launched our education intervention as a home-visit program delivered by a nurse educator consisting of eight sessions over 1 year.
The group-based curriculum was revised to encourage family participation, emphasize dietary and lifestyle aspects, and incorporate motivational interviewing techniques. In October , we created and filled the position of diabetes coordinator, who managed the panel, expedited subspecialty referrals, and delivered training to diabetes nurses.
Next, in June , the institution's diabetes protocol was revised to place a stronger emphasis on insulin-based therapy and to promote monthly clinic visits for improved continuity of care for complex or uncontrolled patients [ 7 ]. At this time, language policies were also formalized to mandate care delivery in Mayan languages when applicable. Finally, retention of diabetes ground-level staff clinical nurses and diabetes educators became a potentially critical problem due to staff burnout.
In September , we implemented a package of retention strategies including salary increases, flexible work arrangements and formal professional development opportunities.
Three activities were most important at this level. First, in March , an electronic diabetes registry was created to track individual patient data in real time, display trends in cohort-level control, and provide clinical reminders to providers. Simultaneously, we instituted monthly diabetes panel reviews in which members of the QI team discussed each patient's data and care plan.
In July , our institution hired a pharmaceutical procurement specialist and implemented a new electronic drug distribution system. The new procurement officer and distribution system together were intended to reduce supply chain failures—especially for insulin—and free up clinical staff from responsibilities relating to medicine distribution. This QI project was carried out with all adult type 2 diabetes patients served in the diabetes program. This sampling decision permitted evaluation of the impact of the QI project on long-term patients under active management for whom sufficient baseline clinical data were available.
Among the eight patients in the eligible sample who were dropped from the analysis due to lack of longitudinal data, two had abandoned care in the 12 months prior to the launch of the QI program and six were lost to follow-up after the launch. There were no statistically significant differences in gender, age or community location between those who dropped out and those retained; however, baseline HbA1C was higher in patients who dropped out 9.
The primary outcome measure was glycemic control as defined by mean HbA1C and proportion of patients meeting target HbA1C. We justify our target as appropriate to our context given the high risk of severe hypoglycemia in rural Guatemalan towns with limited access to emergency services.
Secondary measures were process indicators we considered proxies for each key driver Fig. At the home and clinic level, we measured visit intensity as defined by the mean number of visits in each setting per patient quarter.
At the organization level, we considered insulin use to be a proxy for overall institutional capacity as well as, given the psychological and cultural barriers to insulin use [ 13 ], a marker for the acceptability to patients of offered services , and we assessed the proportion of patients with insulin prescriptions. We used Stata version 13 College Station, TX to generate descriptive statistics for baseline patient characteristics. Control charts are the primary tool of SPC and permit analysis of process data [ 14 ].
Control charts have been used in a variety of health settings [ 15 ]. X-bar and S and P charts permit underlying data of variable subgroup sizes, which is an important feature of longitudinal diabetes program evaluation, as HbA1C values are typically not available for all patients during each period.
The CUSUM chart was selected to assess for cumulative improvement of panel performance as each data point incorporates information from prior time periods [ 14 ]. Secondary process measures were assessed via X-bar and S charts for mean home and clinic visits per quarter and a P chart for the proportion of patients receiving insulin prescriptions each quarter.
For all of the control charts, we followed methods well established in the literature for identifying special cause variation. First, we used the first 12 data points to generate the baseline mean. For our data set, this included 10 quarters of pre-intervention data and two data points that overlapped with the intervention Q4 and Q1.
We applied the following criteria to identify special cause variation [ 14 ]: i a single point outside control limits, ii a run of eight or more points in a row above or below the baseline, iii six consecutive points increasing or decreasing and iv two of three consecutive points near a control limit. A challenge of assessing improvement in panel quarterly mean HbA1C with control charts is that individuals with better or worse disease control may be monitored at a different frequency in the intervention period relative to the baseline period.
If differential monitoring occurs, the detection of special cause in a control chart may not reflect true changes in underlying panel glycemic control.
We, therefore, conducted a sensitivity analysis using the full-rank cohort by carrying forward HbA1C values from prior quarters, and replacing them as new values became available [ 14 ]. Following methods described in the literature, control limits were adjusted to account for autocorrelation [ 14 ].
Baseline demographic and clinical characteristics of the patient sample Table 1 included a female preponderance among clinic patients, low levels of formal education and high proportion of patients with abnormal body mass index BMI. Mean age was Most patients expressed preference for the local Mayan language over Spanish. Table 1 Baseline demographic and clinical characteristics of diabetes patient sample.
For continuous variables with non-normal distribution, median and interquartile range IQR are displayed. Baseline demographic and clinical characteristics of diabetes patient sample. In the intervention period, special cause was noted in Q3 when the mean decreased from the baseline mean of 8. During the baseline period, Qualitatively, both control charts appear to show maximal improvements in both mean HbA1C and proportion of patients meeting target in Q3 and Q4 with subsequent waning of improvements over time.
The plot representing lower CUSUM statistic—meaning the cumulative number of individuals with HbA1C dropping below the baseline mean HbA1C for the cohort shows special cause beginning in Q4 when the line dropped below the lower control limit.
Control charts for HbA1C targets. A X-bar control chart of mean HbA1C. A corresponding S chart did not show any special cause variation and is omitted. B P chart for proportion of patients meeting HbA1C goal.
In all charts, the vertical line at Q3 represents the start of improvement activities, the dashed horizontal lines at Q1 represents the time at which baseline limits are extended to the future, and ellipses indicate special cause.
Control charts for secondary process measures are portrayed in Figs 3 — 5. Mean home visits per quarter increased rapidly in the intervention period, then declined after a peak in Q2 as individual patients began completing the home-based curriculum Fig. Special cause occurred in and was sustained after Q1 when the upper control limit was exceeded.
Mean clinic visits per quarter also increased from baseline during the intervention period Fig. Finally, a P chart of the proportion of patients receiving insulin therapy showed progressive increases from the baseline through the intervention period Fig. X-bar and S chart of mean home visits per quarter. The solid vertical line at Q3 represents the start of improvement activities, and the dashed horizontal lines at Q1 represents the time at which baseline limits are extended to the future.
Ellipses indicate special cause during the intervention period. X-bar and S chart of mean clinical visits per quarter. P chart of insulin prescriptions per quarter. The sensitivity analysis of mean HbA1C using the full-rank data approach yielded findings consistent with those from the main analyses. The special cause criterion of crossing control limits was not met, however, as control limits were widened to account for autocorrelation in the full-rank data; similarly, with autocorrelated data, other special cause criteria cannot be formally applied [ 14 ].
Taken together, we conclude that the improvements observed in Fig. This study describes a diabetes QI project in a resource-limited setting that utilized SPC charts to show evidence of improved diabetes care and panel glycemic control. The primary aim of the QI project of improving mean HbA1C and the proportion of patients meeting target HbA1C were achieved as special cause was noted on control charts of the primary outcomes Fig.
In addition to offering statistical evidence of improvement, the control charts offered insights into our QI intervention that might not have been discerned with traditional pre- and post-test statistical analysis.
First, the control charts detected improvements early in the course of the intervention—qualitatively, after just two quarters, and formally with the detection of special cause after four quarters of the intervention period. Early detection of improvement or lack of improvement informed intervention refinement and built morale for ongoing QI efforts by the clinical team. Second, the control charts help indicate at which points in time the interventions worked best, and when the primary outcome and secondary process charts are considered together, they offer a window into how the interventions are working.
For example, the control charts of mean HbA1C Fig. Examination of the control charts of the secondary process measures Figs.
Quality Glossary Definition: Statistical process control. Statistical process control SPC is defined as the use of statistical techniques to control a process or production method. SPC tools and procedures can help you monitor process behavior, discover issues in internal systems, and find solutions for production issues. Statistical process control is often used interchangeably with statistical quality control SQC. A control chart helps one record data and lets you see when an unusual event, such as a very high or low observation compared with "typical" process performance, occurs.
This study reports on a diabetes QI project in rural Guatemala whose primary aim was to improve glycemic control of a panel of adult diabetes patients. Formative research suggested multiple areas for programmatic improvement in ambulatory diabetes care. This project utilized the Model for Improvement and Agile Global Health, our organization's complementary healthcare implementation framework. A bundle of improvement activities were implemented at the home, clinic and institutional level. Control charts of mean hemoglobin A1C HbA1C and proportion of patients meeting target HbA1C showed improvement as special cause variation was identified 3 months after the intervention began. Control charts for secondary process measures offered insights into the value of different components of the intervention.
Skip to search form Skip to main content You are currently offline. Some features of the site may not work correctly.
Бринкерхофф кивнул и положил трубку. - Стратмор отрицает, что ТРАНСТЕКСТ бьется над каким-то файлом восемнадцать часов. - Он был крайне со мной любезен, - просияв, сказал Бринкерхофф, довольный тем, что ему удалось остаться в живых после телефонного разговора. - Он заверил меня, что ТРАНСТЕКСТ в полной исправности.
Это была та же информация, которую получил Стратмор, когда сам запустил Следопыта. Тогда они оба подумали, что он где-то допустил ошибку, но сейчас-то она знала, что действовала правильно. Тем не менее информация на экране казалась невероятной: NDAKOTA ETDOSHISHA. EDU - ЕТ? - спросила Сьюзан.
О чем. - Квадрат Цезаря, - просияла Сьюзан. - Читается сверху .
Заручившись поддержкой федеральных фондов и получив карт-бланш на все необходимые меры для решения проблемы, АНБ приступило к созданию невозможного - первой универсальной машины для вскрытия шифров.
PDF | Article deals with the application of selected tools of statistical process control, through which we can achieve continuous quality.