Low serum calcium is significantly associated with higher all-cause mortality in the timeframe 28–2500 days after AMI compared to the reference group (normal-high calcium). There is no significant difference between the other calcium groups in this time period. All four calcium groups did not differ from each other in the timeframe > 2500 days after the event. The division of the follow-up time in those two time-periods was done post-hoc based on Kaplan–Meier-curve and Scheonfeld analysis.

Prior studies examined in-hospital and short-term mortality for admission calcium levels after AMI. They found increased mortality especially for decreased admission serum calcium^3,4,5, but also for hypercalcemia⁴. We excluded patients, who died within the first 28 days after the event in order to concentrate on long term mortality exclusively. Because our data was obtained from a multicenter survey and the admission serum calcium levels were measured by different laboratories, we decided to build quartiles to categorize the cases into four calcium groups (low, normal-low, normal-high and high), as suggested by other studies. For the present study, calcium status was determined by total serum calcium measurements, a method which is widely used to assess calcium status of patients in daily clinical practice. About 50% of total serum calcium is present in form of ionized calcium¹⁰; only ionized and free, that is not albumin-bound serum calcium is biologically active¹⁰. Thus, assessment of the calcium status can also be based on the measurement of ionized calcium. Particularly in patients with suspected calcium disorders, calcium status estimations can differ from each other when either based on total serum calcium or ionized serum calcium¹¹. This especially applies to certain conditions like hypoalbuminemia, which is why total serum calcium levels can be corrected for albumin concentrations¹². Nevertheless, recent studies questioned, whether albumin adjustment of total calcium does improve the estimation of calcium status^13,14,15. Some prior studies on association of calcium levels and outcome of cardiovascular diseases based their analyses on ionized calcium concentrations which causes a limited comparability with this study^16,17. However, the majority of the previous studies also based their results on total serum calcium leading to a good comparability with the results of the present study.

Total serum calcium levels can be influenced by many parameters. One important factor is impaired renal function, which can affect serum calcium^18,19. Certain medications like diuretics and calcium channels blockers might influence serum calcium levels as well^20,21,22. Therefore, they were considered as potential covariates and were kept in the model if they reached the significance criterion. Accordingly, the final model was adjusted for renal function and intake of diuretics and calcium channel blockers before the event.

Serum calcium is known to be a predictor of long-term mortality in different cardiovascular diseases^23,24. There are previous studies, which examined associations between serum calcium levels and long-term mortality in patients with coronary heart diseases and after AMI^25,26,27,28. Jiang et al. investigated 192 patients with ST-Elevation AMI and divided them into a normal calcium group and a hypocalcemia group. In a multivariate logistic regression for survival after 150 days, they found the hypocalcemia group to have significantly worse mid-term survival than the normal calcium group²⁵. A further study found similar results²⁶. Xingbo Gu et al. compared mid-term mortality of 2594 patients with acute coronary syndrome according to serum calcium. They split up the cases in quartiles as in the present study. Multivariate Cox-regression with a median follow-up period of 21.8 months showed significantly higher mortality among the lowest calcium group and no significant differences for the other three groups. This is very close to what we found for the first time period after the event (28–2500 days). In our study, this effect starts to weaken over time and no more significant differences can be found for the later observation time. It seems plausible, that a laboratory value has a more reliable predictive value in the nearer future and that its predictive value weakens as time goes on.

Researchers from a Chinese study presented similar results for patients with established coronary heart diseases²⁷. The quartile of patients with the lowest serum calcium levels had the highest risk of long-term mortality (median follow-up: 4.9 years). Higher serum calcium levels did not correlate with higher mortality.

Another study from 2012 examined associations between baseline calcium levels and risk of cardiovascular and all-cause mortality in a population with stable coronary heart disease. 1206 patients were followed up for 8 years. The quartile of patients with the highest serum calcium levels had the highest risk of all-cause mortality compared to the other calcium groups²⁸. This is in a way contrary to what we found. Though, comparability is limited due to different time points of measurement of serum calcium levels (at time at AMI versus a stable state of coronary heart disease).

Nevertheless, the exact mechanisms are unclear how decreased admission calcium levels lead to a higher long-term mortality after AMI. Blood calcium levels and intracellular calcium homoeostasis are regulated precisely and even small deviations can lead to organic malfunction especially in cardiac electrophysiological processes^29,30,31. Several studies found associations between decreased serum calcium levels and some cardiovascular risk factors such as hypertension^32,33, diabetes mellitus type 2³⁴, smoking³⁵ or left ventricular systolic dysfunction³⁶. This might contribute to a higher cardiovascular and all-cause mortality in the low calcium group. Nevertheless, we adjusted the final COX regression model for those comorbidities and low serum calcium was still a significant risk factor for mortality in the time between 28 and 2500 days after AMI. Nonetheless we must consider insufficient information on these comorbidities and therefore the possibility of residual confounding. Furthermore, we might not have considered all relevant confounders which can be associated with decrease serum calcium. An important possible confounder in this regard might be the presence of a malignant disease. A study from China with 25,000 cancer patients revealed that 26.7% of them had hypocalcemia³⁷. The presence of hypocalcemia was also associated with a higher in-hospital-mortality compared to cancer patients without any electrolyte disorders³⁷. Thongprayoon et al. examined the association of long-term mortality among hospitalized patients with various admission serum ionized calcium levels and found that hypocalcemia was significantly associated with higher long-term mortality even after multivariate adjustment³⁸. They suggested, that ionized serum calcium could be viewed as a sick index or marker of disease severity.

Furthermore, Thongprayoon et al. found that patients with low serum ionized calcium carried an increased risk of ventricular arrhythmia³⁸. It is also known, that low admission serum calcium levels are independently associated with an increased risk of sudden cardiac arrest³⁹. One decisive factor for this could be a prolonged QT interval, which is known to be a risk factor for sudden cardiac death⁴⁰. Some studies indeed found an association between low serum calcium and prolonged QT interval time^41,42. Furthermore, calcium supplementation seems to be effective for shortening repolarization intervals⁴². Nevertheless, some limitations to our study in this regard must be considered. Firstly, no information on patients QT interval time was available, so it is not possible to determine, weather low serum calcium is indeed associated with prolonged QT intervals in patients included in this study. Secondly, since all-cause mortality was the primary outcome of the study and no further information on the cause of death was available, we are no able to differentiate between mortality caused by sudden cardiac arrests, mortality caused by other cardiovascular diseases and non-cardiovascular mortality caused by other diseases.

As a consequence, we can only speculate, on whether hypocalcemic patients with AMI would benefit from calcium supplementation in a long term. More research is needed in this regard.

Strengths and limitations

There are several strengths of this study. First, the high number of cases from a population-based registry with consecutive enrollment avoids selection bias. The large amount of collected data on relevant covariates such as sociodemographic information, risk factors, comorbidities and information on in-hospital complications and treatment provides the opportunity for extensive adjustments. With a median follow-up time of 6.0 years the observation period after the event is quite long.

Nevertheless, the following limitations to our study must be mentioned. Since only patients up to 74 years were included, results cannot necessarily be applied to older patients. Moreover, the results may not be generalized to all ethnic groups since no information on ethnicity was available. Since this is a multicenter study, admission serum calcium levels were measured by different laboratories, which may cause some bias. Moreover, no information on ionized calcium concentrations and serum albumin concentrations was available. As only free and not albumin-bound calcium is physiologically active, this must be considered as a further limitation of this study. Furthermore, the outcome of the study was all-cause death and no information on the cause of death was available for this study. Also, no data was collected on treatment of abnormal serum calcium levels. Finally, we might not have considered all relevant confounders and cannot exclude possible reverse causation.