DEI is limited to a maximum of 72 h in Germany [25]. However, some German eye banks allow DEI up to a maximum of 48 h according to their own regulations. This could be one of the reasons for the discrepancy between the average German discard rate of 33% and 44% at the Eye Bank in Rhineland-Palatinate. Nevertheless, German regulations are very liberal compared with other countries. Although no restrictions regarding DEI are specified in the guidelines and regulations, the actual postmortem times in many European countries and in the USA are significantly lower than the German time frame [26]. For example, the Cornea Donor Study (USA) included only corneas with a time from death to preservation of less than 12 h (refrigeration/cooling of body/eyes) or less than 8 h (no refrigeration) [27]. In the UK, guidelines for blood transfusion services recommend limiting DEI to 24 h [28]. Our study confirms that a longer DEI leads to a lower ECD. Relative to a mean DEI of 30.7 h in our cohort, this reduction is 224 cells/mm² after this time, making it a matter of debate whether such ECD loss should be considered clinically significant. With a DEI between 24 and 48 h, 55.3% of the harvested corneas had sufficient quality, and with a DEI between 48 and 72 h, as many as 38.0% were still good enough to be transplanted. Similar to our study, Boehringer et al. and to a lesser extent Linke at al. found a negative linear effect of DEI on ECD after penetrating keratoplasty ^[23][29][23,29]. Several other studies could not find a relevant correlation between DEI and ECD changes ^[16][21][24][16,21,24]. One reason for these results could be the much stricter regulations regarding the allowed DEI. Since the allowed collection time in the Eye Bank of Rhineland-Palatinate is up to 72 h, we were able to detect effects that were not detectable with shorter DEI. To our knowledge, this is the first study to investigate the influence of such long DEI on the quality of donor corneas. We think that the high demand for donor corneas allows corneas to be excised for up to 72 h, even at the risk of the ECD perhaps being lower and the discard rate higher. At least half of the corneas excised after 24 h were still suitable for transplantation. On the other hand, DEI is a factor influenced by conditions such as distance to donor or availability of trained personnel at the time of donor death.

In our study, 35.8% of all corneas were from pseudophakic eyes. In the donor age groups of 80–89 years and 90–99 years, 53% and 69% of the eyes were pseudophakic. The mean ECD of pseudophakic eyes was significantly lower than that of phakic eyes (1687 ± 619 cells/mm² vs. 2152 ± 523 cells/mm2). The mean ECD values in this study are lower than those obtained in other studies such as Schaub et al. (phakic 2936 ± 262 cells/mm², pseudophakic 2645 ± 200 cells/mm²), but the donor population differs in terms of the lower proportion of pseudophakic eyes and the higher proportion of female donors compared with our study. Compared with other studies showing a negative influence of pseudophakia on corneal ECD ^[13][30][13,30], the negative influence determined in our study seems rather strong (378 cells/mm² less in pseudophakic eyes), although a direct comparison with other studies is not possible because of different study designs.

Differences in endothelial cell loss may be due to different surgical techniques. The effect of ocular surgical trauma on the cornea is well known [31], but it is not an exclusion criterion for corneal donation. In our retrospective study design, the ophthalmologic history of the donors was mostly unknown. Our collective consists of a high proportion of eyes that underwent cataract surgery and thus have a lower ECD. The result of our study confirms that lens status is an important factor for corneal quality. It must be considered whether the additional effort to determine the donor’s lens status before corneal harvesting by taking a medical history from the relatives or the treating ophthalmologist is reasonable.

Age is certainly the best studied factor influencing corneal quality. In Germany, there is no age limit for corneal donation. We found that age has a significant negative impact on ECD results. Interestingly, our results show a lower ECD and a wider standard deviation compared with other studies. Gain et al. reported a mean ECD of 2135 cells/mm² by donors older than 85 years [8]. Gavrilov et al. found a mean ECD of 2059 ± 313 cells/mm² in donors older than 80 years [7]. Our study revealed a mean ECD of 1846 ± 627 cells/mm² in donors aged between 80 and 89 years (n = 613) and a mean ECD of 1839 ± 586 cells/mm² in donors aged 90 years and older (n = 147). The donor age-related ECD loss in our study (6.2 cells/mm²/year) is comparable with the results reported by McGlumphy et al. (5.2 cells/mm²/year). It needs to be discussed whether an age restriction should be applied to corneal donors because of the negative influence of aging on ECD. In our department, as in many others, corneas from old donors are considered necessary to meet the high demand for grafts. Remarkably, as many as 45% of corneas harvested from donors older than 80 years and 49% of corneas harvested from donors older than 90 years were suitable for transplantation. In addition, there are other positive aspects of older donors that should considered, such as the easier preparation and intraoperative handling of the Descemet endothelial complex for DMEK.

In our study, donor sex was found to have a significant effect on ECD, with corneas from male donors having an average of 189 cells/mm² less than those from female donors, as also found in other studies ^[15][16][15,16]. However, a number of studies found no association between female sex and higher ECD under in vivo conditions ^[10][17][18][10,17,18], so the reasons for the difference in ECD between males and females in donor corneas are not yet understood. Other authors found only minor differences in ECD between male and female corneas in living subjects ^[15][32][15,32].

To the best of our knowledge, the possible influence of CPR on corneal ECD and quality has not been investigated in previous studies. Interestingly, the ECD of corneas from donors who received CPR was significantly higher than that of donors who did not receive CPR. However, this difference could be explained by age because the mean age of all donors who received CPR was 70.5 years, whereas the mean age of all donors who did not receive CPR was 74.7 years. It is conceivable that CPR is more likely to be performed in younger donors with higher ECD than in older, multimorbid patients, some of whom may decline CPR (for example by living will).

In summary, the aim of our analysis was to identify factors that significantly influence the suitability of donor corneas for transplantation. Age, male sex, longer DEI, and pseudophakic lens status negatively influenced ECD and thus graft usability. Cause of death did not appear to be relevant to donor quality. To meet the growing demand for donor tissue, we do not recommend limiting the donor age or collection time to 24 h or excluding oncologic or septic donors, as is done in many countries. We propose a model that can be used to predict the “usability” of the donor based on all known parameters to reduce the number of unnecessary declarations and to economize this process.

ECD= 2919 − 6 × age[years] − 189 [if male] − 7 × DEI[hours] − 378[if pseudophakic]

This model takes into account the factors that most influence the quality of the corneas harvested, such as donor age, sex, DEI, and lens status. This model provides an estimate of ECD based on factors that can be determined prior to corneal harvest. However, the donor itself and the general random effect also influence the ECD. Nevertheless, our model predicts a trend for expected ECD. In practice, this may be useful when a qualitative consideration becomes necessary. Occasionally, the situation arises where corneas could be collected from two donors simultaneously, but human or organizational resources only allow corneas to be collected from one donor in the allowed period. In this case, the model could be used to predict corneas with a higher probability of high ECD based on donor-dependent factors.