As a populous developing country, China not only ranks highest in the number of drowning deaths worldwide  but also has a high drowning mortality rate of 4.4 per 100,000 people . Among drowning deaths in China, some would certainly be cases of victim-instead-of-rescuer syndrome, although the proportion is unclear as there is a lack of relevant research. Our data suggest an average of 0.2 (55/225) rescuer deaths per rescue incident, and a mortality rate of rescuers statistically indistinguishable from that of PDVs, which indicates that drowning rescue is a high-risk activity [18, 19]. Ensuring awareness of rescuers’ safety first  might help reduce the mortality.
Unfortunately, but perhaps unsurprisingly, children constituted a large proportion of drowning victims for both PDVs [6, 21] and rescuers. The underlying reasons could be diverse. On one hand, children usually have limited competence at flexible cognitive control, exhibit a low risk perception and an overestimation of their swimming ability, all of which could lead to increased immature decision-making and risk-taking behaviors [22–24]. This speculation may be supported by the fact that more non-swimmers were found among the children than among the adult rescuers in our current study. On the other hand, the ethics that children have learned could also drive them to take altruistic actions [14, 25]. Therefore, the appropriateness of firmly discouraging children who cannot swim from rescuing should be brought into public awareness.
Local geographic conditions accounted for most of the variability in drowning incidence . The top three provinces involved in our study, which were not the provinces with the largest populations, were all located within the Yangtze River Basin, which is known for its large number of various bodies of water. Consistent with other reports in China, most of the rescue incidents in our study occurred in fresh water and natural bodies of water [26, 27]. Our results indicated that drowning location had a critical impact on the rescuers’ survival, wherein lakes appeared to present lower risk than ponds, reservoirs and wells. Possible explanations for this location difference could be multifactorial. The lakes, whether artificial or natural, attract sightseers; therefore, multiple witnesses are typically available to perform rescue in a collaborative way. In contrast, ponds, reservoirs and wells in China often have steep and slippery water edges with considerable depth, resulting in lengthier stays in the water, along with greater difficulties for rescuers, all of which lead to detrimental outcomes in general . A study conducted by Moran  also revealed that it was difficult for persons to exit deep water with variable edges. We further found that age was associated with the drowning location ; i.e., ponds were particularly high-risk places for children. However, children as PDVs also seem particularly likely to be associated with rescuer fatality , although the reason for this is unclear. Possible explanations could be that children are incapable of providing good cooperation to the rescuers, and that their rescuer was likely to be a peer.
Rescue efforts in hazardous conditions were often highly challenging, especially when additional rescuers or tools were not readily available [18, 28]. Pearn et al. indicated that 50 % of rescuers in rescuer-who-drowns incidents were not familiar with the particular body of water and its hazards . Therefore, drowning location would be a key variable in drowning prevention for both PDVs  and rescuers. Good understanding of the water environment before rescue should be strongly emphasized, and plans to effectively implement rescue operations in likely drowning environments need to be in place. Additionally, placement of isolation fencing and simple life-saving instruments in open water areas presumed to pose a high drowning risk could not only prevent most of the drowning caused by falls  but also provide PDVs and rescuers with floatation objects to hold that would improve the chances of survival .
Although the causes of drowning among rescuers were complex, several preventive suggestions have been proposed [16, 17]. A successful rescue is technically challenging [18, 19]. Rescuers without prior rescue knowledge mostly blundered into dangers and encountered life-threatening events, including death . Our study showed that pulling PDVs directly in the water increased the likelihood of a drowning fatality for rescuers. This may be owing to the fact that rescuers standing in water are more likely to lose their balance and become entangled with PDVs; this typically occurs with a contact tow (a rescuer human chain) . Other practices such as swimming to the victim have also been found to be dangerous [14, 15]. Therefore, direct personal contact in an aquatic environment is extremely dangerous for rescuers. It has been proposed that rescue attempts from land or boat be the top priority . The employing of water-away and indirect contact (offering an object such as a pole, stick, rope or plank)—or even non-contact—was the secondary recommendation [30, 33, 34]. In addition, some professional rescue tools, such as rescue buoys, lifejackets, boats and lifelines, have proven essential for successful aquatic lifesaving [18, 19]. Unfortunately, few Chinese rescuers performed water rescues properly.
On-scene rescuers also assume the responsibility to perform on-site pre-hospital resuscitation after rescue . It has been reported that almost 30 % of PDVs who were rescued by bystanders needed cardiopulmonary resuscitation , which, if done properly, could effectively contribute to a positive outcome [36, 37]. Rates of on-scene resuscitation provided by bystanders vary among studies [15, 36, 38]. A Dutch study showed that approximately 31.8 % of victims received resuscitation aid from bystanders . Similarly, Topjian et al. found that one-third to one-half of rescued children in their study had received bystander cardiopulmonary resuscitation . A hospital-based prospective study in Japan reported that 92 % of the patients transported to hospital because of drowning had been given pre-hospital resuscitation . The rate of pre-hospital resuscitation among victims in China is lower than these figures, leaving substantial room for improvement. Therefore, educating the public in basic first aid skills could be an important step .
Although the present study offers a valuable data source for exploring drowning rescue events because most of the key variable information was contained in the reports [17, 40–42], the media is likely to underestimate the number of cases because of selective reporting . Therefore, it is impossible to provide population-based rates on either PDVs or rescuers, and without population data it is not possible to extrapolate this information to all drowning rescue attempts. The media articles also often lack information on potential confounds, such as characteristics of water flow, the distance from shore, alcohol consumption of the rescuers, life vest application, differences in estimated swimming capacity (e.g., weak, average or strong), relationship between victims and rescuers, whether victims were warned of the potential drowning peril, or how the call for help was made. Moreover, we could not rule out the possibility that rescued victims died later, because media reports typically reported on victims’ deaths at the scene but lacked post-rescue follow-ups. Additionally, missing data was another unavoidable problem: in cases where the rescuers left immediately after the rescue action, or where many rescuers performed rescue simultaneously, it would be difficult for the media to collect information on each of the rescuers at the desired level of detail.