Researching Lift-Assists: Nebulous Complexity

In this issue of Prehospital Emergency Care, Leggatt et al.¹ describe patient morbidity and mortality following lift-assist calls in London, Ontario. Within 14 days of an initial lift-assist call, 21% of patients had a subsequent emergency department (ED) visit; nearly 12% required hospital admission; and 1% died. Only one previous study specifically explored lift-assist calls: Cone et al.² estimated 44% of lift-assist patients had a repeat EMS encounter within 30 days, with just more than half of those repeat encounters resulting in transport to hospital. Other studies have examined simple falls in elderly patients, finding the decision whether to transport is complex;^3,4 the patients have substantial rates of subsequent health care system contact, hospital admission and death;^5,6 and EMS-initiated referrals to falls prevention programs are feasible and can be effective.^7-9 This newest work adds to our understanding of lift-assist calls, while raising important questions and also highlighting the difficulties of studying the sometimes nebulous yet complex situations encountered by prehospital providers.

Dr. Leggatt and her colleagues defined a lift-assist call as when an “…individual is assisted up to a more mobile position from the ground by paramedics, but not treated or brought to hospital for further medical attention.”¹ Although not explicitly stated, “…from the ground…” suggests these were largely simple falls. In contrast, Cone et al.² used a broader definition, including patients who “…are unable to move from an undesirable position to a preferred one….” This is the first nebulous complexity: What constitutes a lift-assist call? What differentiates a lift-assist from a non-transport or patient refusal after a simple fall? Is it necessary to distinguish between these circumstances—for either clinical or research purposes? If it is necessary to separate lift-assists from other non-transport situations, can that be done accurately using current EMS data sources and definitions? Consistency in how a lift-assist is defined, and consensus on whether and how lift-assists differ from simple falls and other non-transports, are necessary to advance lift-assist-related research, policies and practices.

Since simple falls or other mobility impairments might be an indicator of some underlying acute disease process, or a precursor to general functional decline, a thorough EMS assessment would seem appropriate. On the other hand, many lift-assist patients do not desire medical evaluation, “[t]hey simply want responders to physically help them back to a bed, chair or wheelchair.”² In Dr. Leggatt's study, 14% percent of the ambulance call reports for lift-assist patients were missing at least one vital sign—primarily temperature—and more than a quarter of patients with a history of diabetes did not have a documented blood glucose level.¹ One interpretation of this finding is that many patients were inadequately assessed, with the presumption that better assessment might have identified patients at risk and prevented some of the subsequent ED visits, hospitalizations and/or deaths. Another possibility is that some patients simply refused detailed assessment once the lift-assist was complete. Either way, this is the second nebulous complexity: What is the appropriate level of assessment for a true lift-assist call? Does that vary for lift-assist calls with a minor mechanism and no symptoms (e.g., “I just stumbled onto my knees, but I can't get myself up without help”) versus minor symptoms (e.g., “I felt light-headed so I sat right down on the floor, but now I can't get myself up”) versus no mechanism (e.g., “I just need help getting from sofa to the bedroom”)? Should any single abnormal assessment finding trigger transport? Or, are there thresholds beyond which an abnormal finding should trigger transport? Further, should the necessary level of assessment and threshold for transport vary for patients who live alone versus those who live with other family members versus those who live in assisted living facilities, and so forth? Although it seems intuitive that abnormal vital signs or other assessment findings would be associated with increased risk of subsequent health system contacts, hospital admission or death, none of the studies to date have explicitly linked abnormal assessment findings with adverse outcomes. One also must wonder whether policies requiring complete assessments for all lift-assist patients, and transport for patients with any abnormal finding, would simply convert lift-assist calls into patient refusal calls.

Selecting meaningful outcome measures is a third nebulous complexity that has long challenged researchers of EMS non-transports. Leggatt et al.¹ used 14-day ED visits, hospital admissions and deaths as study endpoints. Cone et al.² used 30-day repeat EMS encounters and transports to hospital. In their study of elderly patients with simple falls, Snooks et al.⁵ considered repeat EMS encounters, hospital admissions, primary care (“general practitioner”) office visits, and deaths within 14 days. For their study of falls at assisted living facilities, Williams et al.⁴ considered “time sensitive injuries” that required ED or hospital care within 48 to 72 hours. There is currently no consensus on the most appropriate measures of adverse outcomes after non-transport, nor on the relevant follow-up time. Interestingly, the issue of follow-up time doesn't only concern how long the clock runs, but also when the clock starts. For example, is an ED visit that occurs 45 minutes after a lift-assist call an indication that the lift-assist should have resulted in transport? Or, is it an indication that the lift-assist call facilitated an ED visit by private transport? Also, whatever timeframes and endpoints one considers, a common limitation of these studies is the inability to identify events that occur in competing or outside health systems. Strategies for ensuring that all relevant outcomes are assessed, over an appropriate time period, even if/when they occur in health systems that are not under the purview of the investigators, are critical for accurate studies of lift-assist outcomes.

The two lift-assist studies to date have described patient presentations and outcomes without any reference or comparison group. In both studies, approximately one in five lift-assist patients had a subsequent ED visit,^1,2 and this newest study reported one in eight lift-assist patients were subsequently hospitalized.¹ What we don't know is how the rates of repeat EMS calls, ED visits and hospitalizations in the lift-assist population compare to those of simple fall patients who are transported to hospital, or to those of other non-transport patients, or to those of the elderly population in general. In the absence of a randomized controlled trial in which lift-assist patients are allocated to transport and non-transport cohorts, what is the most appropriate comparison group for evaluating outcomes of lift-assist patients?

Whatever control group is used, the analytic methods also should address one final nebulous complexity: many lift-assist patients have repeated lift-assist calls. That is, there is potentially a person-level repeated measures factor in the data. One way to address this issue is to limit the analysis to only the first lift-assist encounter during the study period, with data from any subsequent EMS encounters used solely as outcome measures for that first encounter. Statistical procedures, such as the use of generalized estimating equations and other panel data techniques, can also address person-level repeated measures, but those typically require advanced analytical expertise. In the absence of those or similar controls, one must be concerned about the extent to which data for individual patients who have multiple lift-assist encounters with persistently good or persistently bad outcomes might bias the reported results.

All of these concerns were raised during the peer review of Dr. Leggatt's paper, and the authors have done their best to address these issues in their analysis, the presentation of the results, or the discussion of their findings. These are not “fatal flaws”; they represent the reality of conducting research about a relatively unexplored aspect of EMS. As more lift-assist studies are completed, and as more researchers become engaged on this topic, consensus on most of these nebulous complexities will surely emerge. In the meantime, authors of lift-assist studies should be as transparent as they can about these limitations in their work (as Dr. Leggatt's team has been), and readers of lift-assist papers must cautiously—but reasonably—consider how these issues might influence the findings, and what that might mean for the generalizability of the results.