Gastric dilatation volvulus in dogs: utility of lactate as a predictor of survival

[...]

Appraisal, application and reflection

Of the 15 studies appraised, many did not have the PICO question given above (association between plasma lactate concentration and outcome) as their primary objective. Rather, plasma lactate was measured as part of a broader study assessing other variables and / or outcomes. Whilst this did not affect results, it meant that, in some cases, statistical analysis was not as detailed or as focused on the PICO question of this Knowledge Summary as might otherwise have been the case (for example, some studies reported receiver operating characteristic (ROC) analysis and optimal sensitivity and specificity, while others did not). This explains why these studies did not have 'lactate' as a title keyword. However, the six studies of Rauserova-Lexmaulovaa et al. (2020), Grassato et al. (2020), Santoro Beer et al. (2013), Green et al. (2011), Zacher et al. (2010) and De Papp et al. (1999) did have the PICO question of the Knowledge Summary (association between survival and plasma lactate level) as the primary study focus.

Twelve of the 15 studies were retrospective, with the limitations inherent to retrospective studies, as described above, in particular there was variation both within and / or between studies in the presenting signs, characteristics of the study population, and analysis and presentation of data.

Median overall survival across all studies = 78.4% (range 64.1–88%).

Single pre-intervention plasma lactate measurements

Thirteen studies reported an initial (pre-intervention) lactate level in survivors vs non-survivors. In 11 papers this was reported as median with range for the two groups, in one paper (White et al., 2021) this was reported as mean with range for the two groups; while in three studies (O'Neill et al., 2017 Green et al., 2012; and De Papp et al., 1999) this was reported as number of cases above and below a pre-selected lactate cut-off value (> 4 mmol/L in the first and > 6 mmol/L in the latter two studies respectively).

Of the total 15 studies appraised 11 papers reported a significant difference in initial lactate concentration between survivors and non-survivors,4 papers reported no significant difference in initial lactate concentration between survivors and non-survivors.

In all studies, the range of lactate values reported for both survivors (range 0.1–20 mmol/L) and non-survivors (range 1.1 - 25.3 mmol/L) was very wide, with, in all cases, significant overlap in values between the two groups (summarised in the table below). In many of the studies, the sample size was small (as indicated above and below), and in no study was power analysis undertaken to detect a significant difference between lactate levels in survivors versus non-survivors. This raises the possibility that those studies reporting non-significance were under-powered to detect significance in plasma lactate levels between survivors and non-survivors.

Even in those studies where a significant difference was found between populations, the broad range of values (and overlap between survivors and non-survivors) has important implications for decision-making in individual animals. In practical terms, although clinicians may (correctly) infer that a 'high lactate’ (HIL) value is more commonly associated with non-survival compared to a 'lower lactate’ level, the range of lactate values within these two groups, and the substantial overlap between them, makes prediction at the level of individual cases difficult.

[…]