12 CHAPTER 1 Geriatric Physical Therapy in the 21st Century enough of the variance to be clinically useful.Under s are ular co for prognosis ange in di typically statistical analyses to evaluate the relative impact of one GRC scale during an exercise intervention.Thus,using the pts 器罗 be both s fndings (6MWT to he ne for the outhat are deemed starisicallye and pro community-dwelling older adult. the e many tools that do not have an establishe re dence that would be criti appraising a stud fo of ch ange represof ch deemed statisticall significant should be further evalu- one group versus amount of change in the comparison group)and make a clinical judgment, ased or expe ng mus onstrate statistically understanding of the co ition,abo ces d tha al assume clinical importance.An outcome deemed to Systematic Reviews.The purpose of a systematic repres ent a statistically significant improvement may, review is to aggregate the finding across studies to nonetheless a sma ct on the patien provide bout the streng amount nty the 0P (MCID A rep ents the smallest amount of cha deemed being reviewed in the svstematic review is based on the quality level of each included article as well as thee me tools,an ize nitud of the change the be cores res). ize may be cal A common approach for establishing a tool's MCID sized by the authors. or quantitatively ted link the patient's reported staten ent of outcome through a meta-analysis into with change in a too rived effect size across all studies. or a vana an er ect s ary es.The GRC is a 15-point rank-ordered scale A commonly applied rule of thumb is that with-7representing"a very great deal worse"0 repre ang no hang epresenting "a very grea ；and more than 0.8,a large effect n o calc izes likel been used to link the amount of change on the 6-minute across the p walk test (6MWT)and patient-reported outcomes of change;in community dwe e adults,a 20-m the strength of the recommendations can increase in distance ring the 6Mw I represents B0X1-3 Tool,as a very great deal better almost the same,hardly any worse at all deal bette 十4 a little hett -6 =almost the same,hardly any better at all -7=a very great deal worse =no change12 CHAPTER 1 Geriatric Physical Therapy in the 21st Century Common Descriptors Used for Each of 15 Possible Responses to Patient-Reported Outcomes Using a Global Rating of Change Tool, as Described by Jaeschke et al.21 enough of the variance to be clinically useful. Under￾powered studies are of particular concern for prognosis studies of adults aged 75 years and older. Therapy Studies. Therapy studies typically use statistical analyses to evaluate the relative impact of one or more interventions within or across groups of sub￾jects. The concepts of statistical significance and clinical importance both need to be examined in assessing the findings of a study. Differences between or among groups that are deemed statistically significant are con￾sidered real, that is, not occurring by chance, and pro￾vide a reasonable level of confidence that similar out￾comes would be obtained for comparable groups receiving comparable interventions. Only findings deemed statistically significant should be further evalu￾ated for clinical importance. Although a finding must demonstrate statistically sig￾nificant differences to be further evaluated for clinical importance, statistical significance alone does not assume clinical importance. An outcome deemed to represent a statistically significant improvement may, nonetheless, have such a small impact on the patient that the amount of change is clinically unimportant. The term minimum clinically important difference (MCID) represents the smallest amount of change deemed clinically important for the patient. An MCID has been established for many commonly used outcome tools, and the number of tools with established MCID scores is growing annually. A common approach for establishing a tool’s MCID is to link the patient’s reported statement of outcome with the amount of change obtained in a tool. The Global Rating of Change (GRC) tool,21 or a varia￾tion of it, is often used as an anchor for patient-reported outcomes. The GRC is a 15-point rank-ordered scale, with –7 representing “a very great deal worse”; 0 repre￾senting “no change”; and 17 representing “a very great deal better.” Box 1-3 lists all descriptors commonly used as labels across this scale. For example, this tool has been used to link the amount of change on the 6-minute walk test (6MWT) and patient-reported outcomes of change; in community-dwelling older adults, a 20-m increase in distance walked during the 6MWT represents a small but clinically meaningful improvement.22 This MCID was established from the average change in dis￾tance walked for patients who reported their improve￾ments as 2 (a little better) or 3 (somewhat better) on the GRC scale during an exercise intervention. Thus, using the MCID of 20 m on the 6MWT as an example, the finding of a study must be both statistically significant AND demonstrate a change of at least 20 m on the (6MWT) to be deemed clinically important for the community-dwelling older adult. For the many tools that do not have an established MCID, the person critically appraising a study would simply identify the amount of change represented in the study (pretest to posttest change; or amount of change in one group versus amount of change in the comparison group) and make a clinical judgment, based on experi￾ence and an understanding of the condition, about the likelihood that the amount of reported change would be clinically meaningful to the patient. Systematic Reviews. The purpose of a systematic review is to aggregate the findings across studies to provide a recommendation about the “strength” (cer￾tainty) of the body of evidence on a given topic. The strength of the recommendation for each outcome being reviewed in the systematic review is based on the quality level of each included article as well as the effect size (magnitude of the change or the correlation of scores). Effect size may be calculated for each individ￾ual article and then descriptively discussed and synthe￾sized by the authors, or quantitatively aggregated through a meta-analysis into one mathematically de￾rived effect size across all studies. The specific meta￾analysis used to calculate an effect size will vary based on the statistical analyses performed in the original studies. A commonly applied rule of thumb is that an effect size of at least 0.2 represents a small effect; 0.5, a medium effect; and more than 0.8, a large effect. A confidence interval is also calculated with the meta￾analysis, which provides a range of effect sizes likely across the population. Many grading schemes are available to categorize the strength of the recommendations that one can draw from a systematic review. Some are fairly elaborate 17 5 a very great deal better 16 5 a great deal better 15 5 a good deal better 14 5 moderately better 13 5 somewhat better 12 5 a little better 11 5 almost the same, hardly any better at all 0 5 no change 21 5 almost the same, hardly any worse at all 22 5 a little worse 23 5 somewhat worse 24 5 moderately worse 25 5 a good deal worse 26 5 a great deal worse 27 5 a very great deal worse BOX 1-3