On beaches around the world characterized by wave breaking activity across surf zones (herein referred to as “surf beaches”), it is well established that the primary cause of rescues conducted by lifeguards, as well as fatal drownings, is rip currents (e.g., Klein et al., 2003; Gensini and Ashley, 2010a; Brighton et al., 2013; Brander and Scott, 2016). Rip currents are strong and concentrated flows of water moving away from the shoreline that are driven by alongshore variability in wave breaking and energy dissipation (Castelle et al., 2016). They are complex and variable features that are manifest as diverse types, which can be both persistent and transient in occurrence and location, may occupy deeper channels between shallower sand banks, or may lack any morphologic expression at all and can occur along open stretches of beaches, both oceanic and lacustrine, or against hard structures such as headlands or piers (Castelle et al., 2016).

Typical rip currents are on the order of 5–50 m wide and extend to the seaward limit of the surf zone, where they may recirculate, or extend past the surf zone variable distances offshore (Castelle et al., 2016). Mean rip current flow speeds over sustained periods (hours) are on the order of 0.3–0.5 ms-1, but rips can experience short-lived pulsations of 2 ms-1 or more (MacMahan et al., 2006) making them a significant hazard to swimmers or waders of all swimming abilities who may find themselves caught in them. Inexperienced surfers and bodyboarders can also be imperiled by rip currents (Attard et al., 2015).

There has been a significant and recent increase in research relating to both physical and social aspects associated with the rip current hazard (e.g., Hatfield et al., 2012; Brannstrom et al., 2014; McCarroll et al., 2014; Scott et al., 2014; Castelle et al., 2016b; Houser et al., 2017). However, an ongoing challenge in addressing the actual societal and economic impact of the rip current hazard for beach safety practitioners, governments, and scientists alike is obtaining accurate values of the number of rip-current-related lifeguard rescues and fatal rip current drownings. In terms of the latter, two key factors make it impossible to determine the number of deaths caused by rip currents with complete accuracy.

First, it is well established that the majority of fatal rip current drownings occur on beaches unpatrolled by lifeguards or outside of seasonal or daily beach patrol times (Branche and Stewart, 2001; Brander and Scott, 2016; SLSA, 2017). In some of these incidents, there are simply no eyewitness accounts available to help determine the cause of drowning. In others, drowning deaths are observed, but by people lacking necessary awareness and understanding to correctly attribute the role (if any) of a rip current in a drowning.

Second, in many countries there are no national requirements for reporting the causal factors (such as rip currents) in coastal drowning deaths. Even in countries that do, such as Australia (Brighton et al., 2013) and Costa Rica (Arozarena et al., 2015), the documented number of rip current fatalities is likely underestimated for the reasons previously noted. For example, while Brighton et al. (2013) determined an average of 21 rip-current-related fatalities on Australian beaches per year, they emphasized that this value was an underestimate as it was based only on confirmed rip-current-related drowning deaths.

The United States, with thousands of kilometers of coastline affected by rip currents and hundreds of millions of beachgoers each year, presents a challenge in accurately determining the number of rip-current-related drownings that occur. There are five distinct coastal regions characterized by different wave climates and physical characteristics, such as geologic setting and beach type: (i) the continental Pacific West Coast; (ii) the Atlantic East Coast; (iii) the Gulf Coast; (iv) the coastlines of the Great Lakes; and (v) the Hawaiian Islands. Air and water temperature differences, as well as beach user demographics and beach usage, can also vary greatly among these regions, creating variable swimming seasons throughout the country

The complex forcing mechanisms associated with rip current formation, type, and location both within and between these regions not only lead to exposure to the rip current hazard being extremely variable spatially and temporally, but also make it difficult and laborious to provide a gross estimate of the occurrence and location of rip currents on United States beaches at any given time. Similarly, although some coastal US National Weather Service (NWS) offices receive daily reports on rip current activity from lifeguards to assist in evaluating and disseminating their public rip current hazard advisory (Houser et al., 2017; Moulton et al., 2017), these reports do not typically include the specific type, location, or number of rip currents.

Perhaps most importantly, as in other countries, the presence of lifeguards on US beaches is temporally and spatially variable. While some beaches have lifeguard beach patrols year round, and two (Los Angeles County and the city of San Diego) staff lifeguards 24 h a day, year round, others are staffed seasonally or are completely unstaffed (not patrolled). As such, there are many periods of time and beaches where lifeguards are absent. The breadth of services provided by US lifeguard agencies also varies tremendously. Some are staffed and funded as primary providers of public safety, with a variety of advanced training and equipment, such as oceangoing rescue vessels, 9-1-1 answering points, and advanced medical training. Others provide more basic services with limited technology (Brewster, 2017).

Despite these challenges, several attempts have been made to quantify the number of rip-current-related fatalities on US beaches. Lushine (1991) combined documented rip current drowning fatalities in Florida, North Carolina, and Alabama with various nationwide drowning statistic databases to estimate that 150 rip-current-related fatalities occur each year nationally. Gensini and Ashley (2010a) used LexisNexis, an online archive of newspaper articles sourced from local and national newspapers, combined with the National Climatic Data Center's (NCDC) Storm Data (now known as Storm Events) database (which uses a wide variety of sources from emergency management officials to newspaper clipping services), to conclude that on average 35 people die from rip currents in the US each year. In contrast the United States Lifesaving Association (USLA) has estimated that rip current fatalities in the US can exceed 100 per year.

The USLA estimate was arrived at internally in 2004 through a two-step process outlined in documentation submitted to the National Weather Service that is provided here as Supplement. First, the number of deaths each year at surf beaches was estimated based on several published studies. Second, the USLA theorized that the percentage of rescues from drowning due to rip currents, based on reports by lifeguards at surf beaches (then found to be over 80 %), is a proxy for the relative proportion of surf drowning fatalities due to rip currents (in the absence of rescue) and applied that percentage to the total number of estimated surf beach deaths (see Supplement). The discrepancies among these three estimates bear further evaluation.

Since 1966, the USLA has been soliciting annual data from beach lifeguard agencies and other water rescue agencies around the country, including the number of rescues from drowning, the cause of those rescues, the number of medical aids provided, drowning fatalities, estimated attendance, and many other data points. Lifeguard agencies are managed independently of the USLA, which sets recommended operational guidelines. These agencies are only obligated to report annual statistics to the USLA if they are certified (accredited) by the USLA, although they are welcome to report regardless of certification status. The USLA is the only national group collecting these data. Most, though not all, water rescue agencies reporting data to the USLA serve surf beaches where rip currents are present. In 2016, the final year of data included in this study, there were 150 USLA-certified agencies nationwide, varying in size from Los Angeles County and California State Parks on the large side (over 700 lifeguards each) to very small agencies with as few as 10 lifeguards. There are many other US water rescue agencies (the specific number is unknown) that do not report data to the USLA.

As noted, one of the data points collected by the USLA is rescues from drowning, including those from rip currents. Those reporting are surf lifeguards trained to identify and rescue people from distress in rip currents. As noted earlier, the USLA, based on an evaluation of the data it collects, has consistently reported over many years that the primary cause of over 80 % of rescues from drowning by lifeguards at surf beaches is rip currents and that in some areas this proportion is higher. However, two independent published studies have reviewed USLA data and come to different conclusions from the USLA regarding the percent of rip-current-related rescues. Gensini and Ashley (2010b) reviewed the USLA data from 2000 to 2009 and concluded that roughly 36.5 % of rescues reported to the USLA in those years were due to rip currents. Brighton et al. (2013) reviewed the USLA data from 2005 to 2011 and concluded that 53.7 % of the rescues reported to the USLA were due to rip currents. Thus, three sources, reviewing similar data, although during different time periods, have come to widely varying conclusions about what the data collected and reported by the USLA show (Brewster, 2010; Brewster and Gould, 2014).

Rescues from rip currents at beaches where lifeguards are present and report their data can provide insight into the magnitude of the hazard and may be useful as a proxy for the percent of drowning deaths at surf beaches. The primary aim of this study is, therefore, to accurately evaluate and report the percentage of rescues from rip currents by lifeguards reporting to the USLA. An additional aim is to determine why researchers have come to vastly different conclusions as to what the USLA data show and comment on the USLA estimate that rip-current-related drowning fatalities in the US exceed 100 per year.

The USLA refers to itself as “America's nonprofit professional association of beach lifeguards and open water rescuers” (USLA, https://www.usla.org, last access: 14 February 2018). The USLA does not directly train or certify beach lifeguards, but rather promulgates training standards and certifies (accredits) lifeguard providers (agencies) that choose to apply and are found to meet USLA requirements. These lifeguard agencies are typically funded by federal, state, and local governments, as well as a few private entities, some working as contractors to governments.

Many public and private beach lifeguard agencies in the United States record work output and beach observations in a manner similar to that of police and fire agencies. The resulting data offer measures of the services provided and help guide staffing and budgeting decisions. Each year many lifeguard agencies report these data to the USLA. In the final year of this study (2016), 148 agencies reported. These rescue reports vary in magnitude from Los Angeles County, which reported 12 956 rescues from drowning that year, to much smaller agencies that reported as few as 1 rescue (USLA, 2018).

The USLA has suggested a variety of metrics that should be used by beach lifeguard agencies to encourage overall consistency of reporting. These metrics include actual work output, such as rescues from drowning and medical aids performed, drowning deaths, and many other data points. They also include estimates of beach attendance. Annual summaries and the underlying data provided to the USLA are published and made freely available at https://www.usla.org/page/statistics (last access: 14 February 2018).

One of the key data points reported to the USLA is the number of rescues from drowning. For purposes of reporting, the USLA defines rescues as, “Total persons who are judged to be in imminent peril and brought to safety by a lifeguard. Usually involves physical contact. Does not include people who are given oral instructions to move to a safer location.” (USLA, 2018).

The USLA also encourages agencies to document and report the primary cause of distress that led to the rescue. The primary cause reporting options for rescues include “surf”, “rip current”, “scuba”, and “swiftwater”. Agencies may choose none of these if they do not categorize the primary cause of rescue or if none of these categories apply to a given rescue. Surf refers to rescues in response to people who find themselves in distress due to the action of breaking waves or being out of depth. Rip current refers to rescues in response to people caught in rip currents. Scuba refers to rescues involving scuba divers. Swiftwater refers to people in distress in inland areas due to, for example, river flooding, and are therefore not rip current related.

Data on rescues are typically tabulated in rescue reports by the lifeguards who effect the rescues. USLA training materials include extensive information on identifying rip currents and rescuing people in peril from rip currents (Brewster, 2017). The rescue reports are compiled by the agencies and subsequently reported annually, via an online reporting system, to the USLA. Prior to the initiation of an online reporting system, reports were submitted manually via mail or email. The transition to electronic reporting occurred gradually, beginning in the late 1990s.

One of the challenges for reviewers of data reported to the USLA is that reporting lifeguard agencies are under no obligation to tabulate or report the primary cause of distress that led to the rescue. For example, in a given year one agency might report 50 rescues broken down by primary cause, but another agency may simply only report 50 rescues (no primary cause). If the total number of reported rescues for the year is compared to the total number in which rip currents were identified as the primary cause, without factoring out those agencies that failed to report a primary cause, then the actual proportion of rescues related to rip currents (or other primary causes) is diluted.

A second challenge for reviewers of USLA data is that some reporting agencies are solely responsible for inland areas, such as reservoirs and rivers, where surf and rip currents are not present (the Great Lakes, where rip currents can occur, are an exception). Nevertheless, these agencies' total rescue numbers are included in the total number of rescues in any given year. For reasons similar to primary cause reporting, if the total number of reported rescues for a given year is compared to the total number in which rip currents were identified as the primary cause, without factoring out those agencies that serve beaches without rip currents, then the proportion of rescues related to rip currents is further diluted.

A third challenge for reviewers of the USLA data is that some agencies oversee both surf and inland areas, but report totals of all rescues at both venues (and the underlying causes). One example is the city of San Diego, which reports thousands of rescues each year including some (albeit a small number) that occur in Mission Bay, which is a low-energy estuarine environment with no surf conditions or rip currents. Similarly, California State Parks oversee lifeguards at both surf beaches and inland lakes (including reservoirs), including them all in a total number of rescues (and underlying causes).

In determining the percent of rescues attributable to rip currents at surf beaches, it is necessary to exclude rescue reports from agencies that do not identify the primary cause of the rescue and to exclude, to the greatest extent possible, rescue reports from inland areas where rip currents are not present. If these steps are not taken in data evaluation, the percent of rip-current-related rescues will be misrepresented. Avoiding this misrepresentation requires both an in-depth review of the data and knowledge of which reporting agencies serve only inland areas. Even then, for the hybrid agencies that cover both inland and surf, it is not possible to exclude the inland rescue data because they are not separately reported. A goal of this study is to attempt to eliminate factors in the USLA rescue dataset that artificially underrepresent the impact of rip currents on rescues and drowning.

Analysis of USLA rescue data was restricted to the most recent 20 years of compiled data from 1997 to 2016. Data were first checked and corrected for any errors and anomalies. For example, there were several isolated examples where data from one agency appeared twice in a given year, and in a few other cases, the total addition of yearly rescues was found to be mathematically incorrect. These turned out to be minor and did not affect the overall data outputs significantly. As a typical example, a double reporting of data by an agency in 2002 increased the total number of rescues by 10, but this was only 0.021 % of the total number of rescues in the year.

The dataset was then culled using objective decision rules. Specifically, as the purpose was to examine rip current rescues on surf beaches, rescue data from any agency overseeing a body of water that did not include surf beaches were removed. While the Great Lakes represent one of the five coastal regions in the US and are subject to physical forcing mechanisms that can generate rip currents, they were not included in further analysis since, with one minor exception, rescue data from the Great Lakes do not include primary cause of rescue.

Lifeguard agencies in other coastal regions that did not report a primary cause of rescues were also removed. This, unfortunately, resulted in removal of the entire dataset of Los Angeles County, which normally reports the largest number of rescues of any beach agency. It was found that in a typical year this is more than 15 % of all rescues reported to the USLA. However, a random sampling of agencies reporting in Orange County (to the immediate south of Los Angeles County) found rip currents to be the primary cause in 83 % of rescues from drowning. This is comparable to all West Coast agencies, so it appears likely that if Los Angeles County were to report, it would report similar values.

Figure 1 shows the total number of agencies reporting for each year and the excluded agencies (those with no primary cause being reported or non-surf beach agencies). Agencies with both surf and non-surf beaches were included if they reported a primary cause, despite the inevitable, unknown degree of overall dilution of rip currents as a primary cause. Any reports of rescues due to the swiftwater rescue cause were removed from consideration since, by definition, they do not occur at surf beaches. In general, the number of included surf agencies that report primary cause has increased over time, while the number of excluded agencies has remained relatively constant.

Where scuba was listed as the primary cause, the rescues were included, as these rescues can and do take place in surf environments. In these cases, as in others, the primary cause is up to the determination of the reporting rescuer. That is, for example, a scuba diver may be rescued due to complications from scuba diving, from being caught in a rip current, or both. The primary cause is what is to be reported and what we rely on here.

Primary causes of surf beach rescues conducted for the period 1997–2016 for all included reporting agencies in the US were geographically separated into East, West, and Gulf coasts, as well as the Hawaiian Islands (Table 1). As described previously, the Great Lakes were not included because, with one minor exception, no agency from the Great Lakes reported a primary cause. In general, the percent of rescues caused by distress due to rip currents ranged from 75.3 % (East Coast) to 84.7 % (West Coast), with a long-term average across all regions of 81.9 %.

Figure 2a shows the gross reporting of the primary cause of rescues for included agencies during the period 1997–2016, and while the number of rescues for all primary causes clearly fluctuates temporally, as evident in Fig. 2b, this is largely due to the increase in reporting agencies over this time. As is also evident in Fig. 2b, the percentage of total rip current rescues as the primary cause of all rescues nationally varies annually from 75.7 % (2005) to 85.1 % (1999) with no clear temporal trend apparent. There are many factors involved that can impact the number of rip current rescues that occur in a given year, including weather conditions, surf conditions, number of rip currents present, and beach visitation numbers. However, overall, even if the rip rescue data are normalized by the number of reporting lifeguard agencies, the number of surf rescues attributable to rip currents does not vary greatly over time.

An examination of rescue data reported by surf lifeguards in the United States to the United States Lifesaving Association has shown that rip currents are the primary cause of between 75.3 % and 84.7 % of all surf rescues on regional American beaches, with a 20-year average of 81.9 %, a significantly higher estimate than previously reported in the scientific literature. Using the percentage of rip current rescues as a proxy to estimate the number of annual drowning deaths attributable to rip currents in the US suggests a value of 90 solely within the limited jurisdictions of surf rescue agencies reporting to the USLA. Thus, an annual figure of over 100 nationwide is not unreasonable.

Considering the number of US lifeguard agencies that fail to report a primary cause of rescue, it is recommended that the United States Lifesaving Association communicates with these lifeguard agencies to endeavor to increase the level of reporting of surf-related rescues by primary cause. It would also be desirable for a range of consistent and comprehensive data, involving both physical, environmental, and beach conditions and demographic beachgoer characteristics, to be reported by lifeguards. However, it is well established that data collection for beach lifeguards is difficult (Williamson et al., 2006; Harada et al., 2011; Morgan et al., 2013) for a variety of logistical and personal factors and is the fundamental challenge in balancing the tasks of providing water safety vigilance, rescue capability, and data collection, the former of which should not be compromised.

Nevertheless, it is vital to continue to work toward developing increasingly accurate estimates of both rip-current-related rescues and drowning deaths so that local governments, public policymakers, tourism authorities, public health professionals, and funders of mitigation measures understand that rip currents are by far the greatest health hazard related to those entering the water at surf beaches. Through this awareness, appropriate resources such as the provision of additional lifeguard services and development of public education programs can be justified and implemented to assist in drowning prevention.