J Infect Dis. 1991 May;163(5):951-8.
A total of 747 febrile granulocytopenic patients with cancer were randomized to receive ceftazidime plus amikacin (CA) with or without vancomycin (V) as initial empirical therapy. Single gram-positive bacteremias responded in 29 (43%) of 68 patients treated with CA and in 48 (72%) of 67 treated with CAV (P = .001). For single gram-negative bacteremias and clinically documented and possible infections the response rates of CA and CAV were 80% and 63% (P = .17), 55% and 75% (P = .009), and 74% and 81% (P = .16), respectively. However, for patients with gram-positive bacteremia and for all other patients, there were no differences by treatment regimens in the proportion of febrile patients on each trial day (P = .85, P = .82, respectively) or in the duration of fever (P = .22, P = .93, respectively). Moreover, no patient with gram-positive bacteremia died during the first 3 days of true empirical therapy. Antibiotic-associated nephrotoxicity was more frequent in patients treated with vancomycin (6% vs. 2%, P = .02). These results do not support the empirical addition of vancomycin to initial antibiotic therapy in cancer patients with fever and granulocytopenia.
共有747例癌症发热性粒细胞减少患者被随机分组，接受头孢他啶加丁胺卡那霉素（CA），初始经验性治疗中或加或不加万古霉素（V）。68例接受CA治疗的患者中，29例（43%）的单一革兰氏阳性菌血症得到缓解；67例接受CAV治疗的患者中，48例（72%）得到缓解（P = 0.001）。对于单一革兰氏阴性菌血症以及临床记录的和可能的感染，CA和CAV的缓解率分别为80%和63%（P = 0.17）、55%和75%（P = 0.009）、74%和81%（P = 0.16）。然而，对于革兰氏阳性菌血症患者和所有其他患者，各试验日发热患者比例（分别为P = 0.85、P = 0.82）或发热持续时间（分别为P = 0.22、P = 0.93）在治疗方案上并无差异。此外，在真正经验性治疗的头3天内，没有革兰氏阳性菌血症患者死亡。接受万古霉素治疗的患者中抗生素相关性肾毒性更常见（6%对2%，P = 0.02）。这些结果不支持在癌症发热和粒细胞减少患者的初始抗生素治疗中经验性添加万古霉素。
