Response to ‘The association between D‐dimer in COVID‐19 patients and mortality remains beset of uncertainties’

We appreciate the opportunity to respond Dr Gris and colleagues who still have doubts about the predictive value of D‐dimer in COVID‐19.1., 2. We totally understand the concerns from Dr Gris and colleagues who just want to manage their patients better. We are trying to explain the potential influence factors here, hoping to provide any useful information for managing the COVID‐19 patients.

The mortalities differed greatly among studies, hospitals, or even countries. Up to 8 June 2020, the overall mortality of COVID‐19 in China was 5.5% (4638/84191),³ which was 6.6% for Hubei (containing Wuhan, 4512/68135), China.³ During the outbreak, our hospital was designated to admit the laboratory‐confirmed COVID‐19 patients who were moderate type (84.0%, 288/343), severe type (11.7%, 40/343), or critical severe type (4.4%, 15/343) according to the Chinese clinical guidance for COVID‐19 pneumonia diagnosis and treatment.⁴ Mild cases who had no or mild clinical symptoms, and no sign of pneumonia on chest imaging, were mainly admitted to mobile cabin hospitals.⁵ As a whole, the distribution of patients existing in our study was basically consistent with the epidemiological characteristics of COVID‐19 in Hubei, China. Thus, due to small sample sizes (191 cases, 99 cases, respectively) of the two retrospective studies,6., 7. the mortalities from the two studies should be unrepresentative, and comparison of the mortalities might be inappropriate.

Meanwhile, according to the distribution of C‐reactive protein (CRP; 67‐126, 1.3‐480) provided by Dr Gris et al and CRP (median: 3.22 mg/L, IQR: 0.34‐22.5 mg/L) in our original article,⁸ it could be roughly inferred that the study populations were significantly different between the two studies. It is certain that the severe/critical cases suffered much higher mortalities compared to those mild or moderate cases. Thus, these might contribute to the difference of mortalities.

Another unignorable point we have mentioned as a limitation was the difference of length from symptom onset to admission.⁸ Due to differences in patient number and medical resources in different areas, the lengths from illness onset to admission might be hugely varying. For example, the median of length from illness onset to admission in our study was 10 days (interquartile range [IQR]: 7‐15 days), which might also contribute to the difference between the data from Dr Gris and ours. This was why we suggested that dynamic measurement of D‐dimer could provide more information.

Actually, when we first analyzed the data, we were surprised by the striking difference of mortalities between two group (12 deaths versus 1 deaths). Then we cross‐checked the original data several times and it was true. There were unexpectedly no significant differences observed in Dr Gris's data (8/77 versus 17/93),² which might be attributed to the above reasons. Furthermore, because of the limited data Dr Gris et al provided, we also do not know whether there were differences in the baseline characteristics. In this case, it would be better to establish a suitable cutoff value and evaluate the predictive value of D‐dimer in COVID‐19 based on their own population.

The potential use of D‐dimer in COVID‐19 based on recent evidence has been simply summarized in our previous response. There are still many uncertainties and potential uses of D‐dimer in COVID‐19, such as whether it can be used to guide anticoagulation adjustment, initiate mechanical ventilation, and de‐escalating critical care support,⁹ all of which are worthwhile to expect the further studies to describe in more detail.

CONFLICTS OF INTEREST

The authors declare that they have no conflicts of interest regarding this article.