Clinical prediction rules

A prognostic model is the mathematical combination of two or more patient or disease characteristics to predict outcome. Confusingly, prognostic models are also termed prognostic indexes, risk scores, probability models, risk stratification schemes or clinical prediction rules (Reilly and Evans 2006). To be useful, they must be shown to predict clinically relevant outcomes reliably. They must, therefore, be derived from a representative cohort in which outcome has been measured accurately. Next, they must be validated, not just in the data from which they were derived (internal validation) but also on data from independent cohorts (external validation) (Wyatt and Altman 1995 Justice et al. 1999 Altman and Royston 2000). Lastly, a model must be simple to use and have clinical credibility, otherwise it is unlikely to be taken up in routine clinical practice (Table 14.1). 

Laupacis A, Sekar N, Stiell IG (1997). Clinical prediction rules. A review and suggested modifications of methodological standards. Journal of the American Medical Association 277 488-494... 

The cardinal clinical sign of fracture is bony tenderness and careful palpation of the injured ankle will often differentiate between a fracture and a soft tissue injury. Clinical prediction rules have been shown to be useful in determining which children with ankle injuries require X-ray investigation (Dayan et al. 2004 Myers et al. 2005). Adherence to such rules may miss the diagnosis of a small proportion of fractures but these will generally be those of little clinical significance. 

Hess EP, Thiruganasambandamoorthy V, Wells GA, Erwin P, Jaffe AS, Hollander JE, Montori VM, Stiell IG. Diagnostic accuracy of clinical prediction rules to exclude acute coronary syndrome in the emergency department a systematic review. CJEM 2008 10 373-82. 

Hicks GE, Fritz JM, DeUtto A, McGill SM. Preliminary development of a clinical prediction rule for determining which patients with low back pain will respond to a stabilization exercise program. Arch Phys Med Rehabil. 2005 86 1753-62. 

The Stroke-Thrombolytic Predictive Instrument (Stroke-TPI) has recently been developed in order to provide patient-specific estimates of the probability of a more favorable outcome with rt-PA, and has been proposed as a decision-making aid to patient selection for rt-PA." The estimates from this tool should, however, be treated with caution. The prediction rule is dependent on post hoc mathematical modeling, uses clinical trial data from subjects randomized beyond 3 hours who are not rt-PA-eligible according to FDA labeling and current best practice, and has not been externally validated. It is, therefore, not appropriate to exclude patients from rt-PA treatment based solely on Stroke-TPI predictions. 

Narrow impact analysis of the prediction rule used as a decision rule (I.e. prospective demonstration in one setting that the use of the prediction rule Improves clinical decision making)... 

Predictive validity is the ability of a model to predict the effect that pharmacological or other manipulations will have on the condition being modeled. This criterion can present a real difficulty, in that drug development is often dictated by animal models. For example, if a given model only detects a subset of effective compounds (i.e. those belonging to a specific chemical class), then useful candidates will be discarded long before clinical trials, and the flaw in the model s predictive validity will not be discovered. Thus, the possibility that a model will yield false negatives cannot be ruled out. 

As a general rule, in vivo assays are more challenging than in vitro assays because the matrices for the samples are more complex. The most common use for in vivo assays is to measure the concentration of NCE dosed into a laboratory animal by collecting multiple sample time points, one can use the analytical results to plot the PK profile of the NCE and also obtain various PK parameters that help determine a test compound s PK properties. Preclinical PK parameters of a test compound are then used to predict its human PK parameters. Another use of in vivo assays is combining the results with pharmacodynamic (PD) observations to perform PK/PD modeling.77 82 PK/PD modeling is an important aspect of new drug discovery because it can be used to predict the exposures and durations required to determine clinical efficacy of a NCE. 

Based on the analysis of 2245 compounds from the WDI which were investigated in Phase II and later clinical trials, Lipinski s Rule-of-5 predicts... 

Lipinski s rules are designed to predict oral availability of compounds that passively diffuse across membranes. Lipinski s rules are based on observations of a database of approximately 2,500 drugs or compounds studied in clinical trials. In general, the compounds could be described structurally with limits on their molecular weight, number of hydrogen-bond donors... 

Attempts continue to predict metal sensitivity in the individual patient so that the choice of material can be made accordingly. In vitro tests for metal allergies have been developed on the basis of lymphokine (MIF) release from sensitized T lymphocytes exposed to metal-protein complexes (11). About 6% of patients without a previous metal implant had positive reactions to nickel, chromium, or cobalt. However, it is still not clear whether such a positive reaction is a reliable predictor of clinical problems. In practice few patients have either local or systemic reactions when symptoms occur and other causes are ruled out, the implant should be removed. Some workers recommend removal of an implant whenever there is both a positive MIF test and a positive skin test, even in the current absence of a serious reaction. Allergic dermatitis will clear up as soon as the metal has begun to be cleared from the tissue. The type of metal and the amount released into the tissue will affect the time taken for the disappearance of toxic dermatological phenomena. 

---