## Risk Stratification: A Practical Guide for CliniciansRisk stratification is a statistical process by which quality of care can be assessed independently of patient case mix. The evaluation of risk-adjusted patient outcome has become an important part of managed care contracting in some markets, and risk-adjusted outcome rates for hospitals are being reported more frequently in the popular press and on the Internet. This book, written by a statistician and two surgeons for a clinical audience, is a practical guide to the process of risk stratification and does not require or assume an extensive mathematical background. It describes the rationale and assumptions for risk stratification, and provides information on evaluating the quality of various published risk-stratification studies. Numerous practical examples using real clinical data help to illustrate risk stratification in health care. The volume also serves as a step-by-step guide to the production and dissemination of risk-adjusted outcome results for local programs. |

### What people are saying - Write a review

We haven't found any reviews in the usual places.

### Contents

Risk | 7 |

Probability and risk | 9 |

Risk and odds | 11 |

Risk and a single risk factor | 15 |

Risk and multiple risk factors | 17 |

Selection of variables into logistic models | 25 |

Model error and classification accuracy | 26 |

Generalizing the results | 28 |

Multivariable risk stratification | 74 |

Calculation of expected risk for a study sample | 78 |

The observedexpected ratio | 82 |

Interpretation issues | 85 |

Conclusion | 87 |

Interpreting risk models | 88 |

Bias | 89 |

Missing data | 95 |

Conclusion | 29 |

Collecting data | 30 |

Identifying a question | 33 |

Identification of variables | 35 |

Case definition | 36 |

Case ascertainment | 40 |

Planning data collection | 41 |

Selecting data collection software | 43 |

Data entry | 44 |

Pilot testing | 45 |

Quality control | 46 |

Source documentation | 48 |

Conclusion | 49 |

Risk and published studies | 50 |

Evaluating the quality of a study | 51 |

Classical clinical research | 54 |

Risk studies | 61 |

Determining the appropriateness of a reference population | 64 |

Studies that can be used for risk measurements | 67 |

Conclusion | 68 |

References | 69 |

Applying published risk estimates to local data | 70 |

Presentation of results | 97 |

Administrative versus clinical data | 103 |

Some guidelines for application and interpretation of the methods | 105 |

Conclusion | 108 |

109 | |

Advanced issues | 110 |

Design issues | 112 |

Collecting data | 118 |

Univariate analysis | 120 |

Multivariate analysis | 137 |

Statistical significance | 140 |

Evaluating model prediction | 141 |

Variable coding | 144 |

Fit testing | 149 |

Plotting | 152 |

Interpretation | 155 |

Conclusion | 161 |

References | 162 |

Appendices | 164 |

Appendix 2 | 166 |

Appendix 3 | 167 |

### Common terms and phrases

actually adjusted appropriate ascertainment associated calculated case—control studies Chapter chi-square clinical research clinical trial cohort studies comparisons compute confidence interval confounding considered contingency table COPD coronary bypass surgery data collection data entry database described disease distribution effects ejection fraction equation error evaluated example Figure following coronary bypass hospital mortality important indicator variable intercept interpretation length=15 cm Line logistic model logistic regression logistic regression model mathematical measure missing data multiple multivariate neurological deficit null hypothesis O/E ratio observed occur odds ratio operation paraplegia patients percent plot population PPE2 predictor variable prevalence problems random regression coefficients relative frequency risk estimates risk factor variables risk factors risk models risk studies risk-stratification studies SAS code SAS System selection standard statistical power statistical significance stratification studies study design study sample t-test TAAAII techniques tion treatment univariate valve variance York model zero

### Popular passages

Page 3 - As in any other application the models are only as good as the data on which they are based.