1.3.0.0 Grad­ing of Sci­en­tific Ev­i­dence

Since the ADA first began pub­lish­ing prac­tice guide­lines, there has been con­sid­er­able evo­lu­tion in the eval­u­a­tion of sci­en­tific ev­i­dence and in the de­vel­op­ment of ev­i­dence-‍based guide­lines. In 2002, the ADA de­vel­oped a clas­sification sys­tem to grade the qual­i­ty of sci­en­tific ev­i­dence sup­porting ADA rec­om­men­da­tions. A 2015 anal­y­sis of the ev­i­dence cited in the Stan­dards of Care found steady im­provement in qual­i­ty over the pre­vi­ous 10 years, with the 2014 Stan­dards of Care for the first time hav­ing the ma­jor­i­ty of bul­let­ed rec­om­men­da­tions sup­ported by A- or B-‍level ev­i­dence (4). A grad­ing sys­tem (Table 1) de­vel­oped by the ADA and mo­de­led after ex­ist­ing meth­ods was used to clar­i­fy and cod­i­fy the ev­i­dence that forms the basis for the rec­om­men­da­tions. ADA rec­om­men­da­tions are as­signed rat­ings of A, B, or C, de­pend­ing on the qual­i­ty of ev­i­dence. Ex­pert opin­ion E is a sep­a­rate cat­e­go­ry for rec­om­men­da­tions in which there is no ev­i­dence from clin­i­cal tri­als, in which clin­i­cal tri­als may be im­prac­ti­cal, or in which there is conflict­ing ev­i­dence. Rec­om­men­da­tions with an A rat­ing are based on large well-‍de­signed clin­i­cal tri­als or well-‍done meta-‍anal­y­ses. Gen­er­al­ly, these rec­om­men­da­tions have the best chance of im­prov­ing out­comes when ap­plied to the pop­u­la­tion to which they are ap­pro­pri­ate. Rec­om­men­da­tions with lower lev­els of ev­i­dence may be equal­ly im­por­tant but are not as well sup­ported.

Of course, ev­i­dence is only one com­po­nent of clin­i­cal de­ci­sion mak­ing. Clin­i­cians care for pa­tients, not pop­u­la­tions; guide­lines must al­ways be in­ter­pret­ed with the in­di­vid­u­al pa­tient in mind. In­di­vid­u­al cir­cum­stances, such as co­mor­bid and coex­ist­ing dis­eases, age, ed­u­ca­tion, dis­abil­i­ty, and, above all, pa­tients’ val­ues and pref­er­ences, must be con­sid­ered and may lead to dif­fer­ent treat­ment tar­gets and strate­gies. Fur­ther­more, con­ven­tion­al ev­i­dence hi­er­ar­chies, such as the one adapt­ed by the ADA, may miss nu­ances im­por­tant in di­a­betes care. For ex­am­ple, al­though there is ex­cel­lent ev­i­dence from clin­i­cal tri­als sup­porting the im­por­tance of achiev­ing mul­ti­ple risk fac­tor con­trol, the op­ti­mal way to achieve this re­sult is less clear. It is difficult to as­sess each com­po­nent of such a com­plex in­ter­ven­tion.

Table 1—ADA ev­i­dence-‍grad­ing sys­tem for “Stan­dards of Med­i­cal Care in Di­a­betes”