The authors stated that their aim was to examine the effects of 3-months' treatment with LMWH on cancer mortality. The inclusion criteria were clearly defined in terms of the treatment, comparator, outcomes and population; the population was defined by the condition under treatment (VTE) rather than by the target outcome (cancer-related mortality). Studies of patients with and without cancer were therefore included, in addition to studies which were exclusively of cancer patients. It could be argued that restricting the included studies to those conducted exclusively in cancer patients would provide a more relevant population. However, such an approach would inevitably substantially reduce the available data. In this case, a subgroup analysis using data from cancer patients alone showed no significant difference from the overall result, providing some justification for the authors' approach.
The search strategy employed was adequate and the search terms were reported in the article. Procedures undertaken for the study selection and data extraction were likely to have minimised reviewer error or bias, but the methodological quality of the primary studies did not appear to have been assessed. The potential impact of methodological weaknesses in the included studies upon the findings of the review cannot, therefore, be assessed.
The authors acknowledged that the studies included in the review were small open-label trials, most of which did not have cancer mortality as a primary end point. In addition, there was little information on participant characteristics (age, type and stage of cancer) and the included studies used different treatment regimens. Thus, the clinical appropriateness of combining the data remains uncertain, despite the absence of evidence of statistic heterogeneity.
Bearing these limitations in mind, readers should be cautious when interpreting the findings of this review.