Hyperglycemia is common among people admitted to the hospital even in the absence of a history of diabetes. Although admission hyperglycemia can reflect unrecognized diabetes, it mostly occurs on the background of normal glucose homeostasis that is abruptly perturbed by an acute disease. Being due to extensive counterregulation, it has been called stress hyperglycemia (SH) (1). Undiagnosed diabetes and SH can be distinguished by measuring HbA1c, but SH can occur even in patients with diabetes. To account for this continuum, in 2015, Roberts et al. (2) introduced the stress hyperglycemia ratio (SHR). SHR is the ratio between the glucose value recorded upon hospital admission and the chronic glycemic burden in the prior 8–12 weeks, estimated from HbA1c. Simply put, SHR is >1.0 when admission glucose is higher than prior estimated mean glucose, while SHR is <1.0 when admission glucose is lower than prior estimated mean glucose (Fig. 1A). In the original report, among 2,290 patients, SHR in the 4th to 5th quintiles was associated with critical illness (intensive care unit admittance or death), more so than glucose and HbA1c individually. Subsequently, a small study described a linear relation between SHR with poor outcomes shortly after myocardial infarction (MI) (3). Of note, admission hyperglycemia has been described as a negative prognostic factor in patients with coronavirus disease 2019 (COVID-19), especially among those without diabetes (4), supporting a role for SH in determining COVID-19 course (5).
In this issue of Diabetes Care, Yang et al. (6) evaluated the impact of SHR on 2-year outcomes among 5,562 patients admitted for an acute coronary syndrome (ACS) requiring percutaneous intervention. Patients with SHR in the 4th and 5th quintiles displayed higher rates of major adverse cardiovascular and cerebrovascular events (MACCE) and major adverse cardiovascular events (MACE). Results were consistent and robust in several subanalyses by type of outcome, diabetes status, and type of ACS.
This finding is in line with prior evidence. A study of 6,287 patients admitted for MI showed higher 5-year rates of heart failure and all-cause death in those with SHR in the highest quartile (7). In another study, among 4,362 patients admitted for ACS, SHR in the highest quartile was associated with significantly worse rates of MACCE after 2.5 years (7).
The added value of the observations by Yang et al. resides in the analysis of continuous SHR values under nonlinear assumptions. In biology and medicine, relationships between observations are often assumed to be linear, but no-threshold linearity rarely holds true in nature. Of note, the authors found that the relation between SHR and rates of MACCE or MACE had a lowest risk point at an SHR value of ∼0.8. Patients with markedly low SHR (e.g., 0.6) had higher rates of poor outcomes that were similar to those of patients with very high SHR (e.g., 1.5), depicting a U-shaped relation. Mirroring the definition of SH, we could define SHR <1.0 as “stress hypoglycemia” or “relative hypoglycemia.” Therefore, perturbations of prior glycemic homeostasis in both directions identified patients with long-term adverse outcomes after ACS.
Over the years, it remained controversial as to why SH is associated with adverse outcomes. SH may mark a more severe disease with enhanced inflammation and neurohormonal activation. On the other hand, acute hyperglycemia can affect immune cell function and coagulation (8), thereby worsening short-term outcomes. Why SHR projects its prognostic effect long-term remains unclear, and it is puzzling that relative hypoglycemia can exert the same effect.
In patients with diabetes, the low-SHR group may include cases of iatrogenic hypoglycemia, which is well known to be associated with short-term (9) and long-term (10) adverse events. Remarkably, the U shape was more prominent for cerebrovascular events, and it would be of interest to verify how many hypoglycemic events were in fact misdiagnosed as cerebrovascular events.
In patients without diabetes, for whom the curve was more U-shaped, low SHR may result from impaired stress-induced counterregulation. Release of stress hormones is a physiologic response prompting cellular programs instrumental to the organism’s resilience to dangerous conditions. Indeed, moderate SH may even be protective under some acute conditions (11). Admission hypoglycemia was previously found to be associated with 30-day mortality after MI (12), but it is challenging to understand why such an abnormal response to an acute condition can project its prognostic effect long-term.
Notably, low SHR can occur in patients with HbA1c in the prediabetes range. For example, an admission glucose of 5 mmol/L with HbA1c of 46 mmol/mol yields an SHR of 0.66. Thus, the poor outcome of the low-SHR group could be due at least in part to the risk conveyed by prediabetes. Furthermore, low SHR may also be observed among so-called fast glycators, who have higher HbA1c values for each average glucose concentration (13). Accumulation of advanced glycation end products in these individuals may be one mechanism driving long-term risk.
Some other points deserve attention. First, why was the J- or U-shaped relation between SHR and outcome not identified before? While a J-shaped trend was in fact present in the original description, Roberts et al. (2) included onlypatients with admission glucose of 5.5 mmol/L or higher, and few had very low SHR values building the left-ascending branch of the U. As HbA1c is not always measured in all hospitalized patients, a selection bias inevitably occurs in retrospective studies using data not purportedly collected to calculate SHR.
Second, one would expect the lowest risk point to be set at SHR values around 1.0. Instead, even in the study by Roberts et al., the reference group with the lowest proportion of critical illness had an average SHR of ∼0.8. This may be explained partly by treatments administered just before or upon admission to control glycemia. More importantly, the SHR equation is still based on the relation between HbA1c and mean glucose described by Nathan et al. in 2008 (14). Incorporating the new and more reliable equation of glucose management indicator (15) would rebalance SHR toward higher values, thereby moving the lowest risk point closer to 1.0. Since SHR around 1.0 can be observed with hyperglycemia or normoglycemia, it is more likely that SHR is a sensitive indicator or metabolic stress response than a causal mechanism determining the outcome.
Finally, one limitation of SHR is its reliance on a single glycemic value recorded upon admission, which can depend on several factors, including timing of the last meal and therapies (e.g., glucocorticoids). Indeed, mean glucose during the hospital stay is a better outcome predictor than admission glycemia (16). While SH is mostly transient (17), whether such patients are at risk for diabetes development is unclear, which remains a concern during the COVID-19 pandemic (18). Thus, we suggest that future studies incorporate mean in-hospital glycemia into SHR and look at the future patient’s metabolic history to refine the clinical meaning of SHR.
In conclusion, new data illustrate how deviations from an individual’s prior glucose homeostasis in both directions are associated with adverse outcomes after ACS. Mammalian physiology makes great efforts to keep glycemia under strict control. Thus, we argue that patients with abnormal SHR may be more susceptible to metabolic perturbations during subsequent years, including a mixture of impaired counterregulation, hyperglycemic spikes, enhanced glycation, prediabetes, and progression to diabetes (Fig. 1B). All of these would make patients less resilient to stress and prone to adverse outcomes. Unveiling the different pathophysiological meaning and clinical implications of SH and relative hypoglycemia deserves future investigation.
See accompanying article, p. 947.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.