Introduction: Identifying the potential effective factors of rhabdomyolysis-induced acute kidney injury (AKI) is of major importance for both treatment and logistic concerns. The present study aimed to evaluate the value of creatine kinase (CK) in predicting the risk of rhabdomyolysis-induced AKI through meta-analysis.
Methods: Two reviewers searched the electronic databases of Medline, EMBASE, Cochrane library, Scopus, and Google Scholar. Data regarding study design, patient characteristics, number of cases, mean and screening characteristics of CK, and final patient outcome were extracted from relevant studies. Pooled measures of standardized mean difference, OR, and diagnostic accuracy were calculated using STATA version 11.0.
Result: 5997 non-redundant studies were found (143 potentially relevant). 27 articles met the inclusion criteria but 9 were excluded due to lack of data. The correlation between serum CK and AKI occurrence was stronger in traumatic cases (SMD = 1.34, 95 % CI = 1.25-1.42, I(2) = 94 %; p < 0.001). This correlation was more prominent in crush-induced AKI (adjusted OR = 14.7, 95 % CI = 7.63-28.52, I(2) = 0.0 %; p = 0.001). Area under the ROC curve of CK in predicting AKI occurrence was 0.75 (95 % CI = 0.71-0.79).
Conclusion: The results of this meta-analysis declared the significant role of rhabdomyolysis etiology (traumatic/non-traumatic) in predictive performance of CK. There was a significant correlation between mean CK level and risk of crush-induced AKI. The pooled OR of CK was considerable, but its screening performance characteristics were not desirable.
Keywords: Acute kidney injury; Creatine kinase; Decision support techniques; Meta-analysis; Rhabdomyolysis.