Estimating medical risk in human spaceflight.

NASA and commercial spaceflight companies will soon be retuning humans to the Moon and then eventually sending them on to Mars. These distant planetary destinations will pose new risks-in particular for the health of the astronaut crews. The bulk of the evidence characterizing human health and performance in spaceflight has come from missions in Low Earth Orbit. As missions last longer and travel farther from Earth, medical risk is expected to contribute an increasing proportion of total mission risk. To date, there have been no reliable estimates of how much. The Integrated Medical Model (IMM) is a Probabilistic Risk Assessment (PRA) Monte-Carlo simulation tool developed by NASA for medical risk assessment. This paper uses the IMM to provide an evidence-based, quantified medical risk estimate comparison across different spaceflight mission durations. We discuss model limitations and unimplemented capabilities providing insight into the complexity of medical risk estimation for human spaceflight. The results enable prioritization of medical needs in the context of other mission risks. These findings provide a reasonable bounding estimate for medical risk in missions to the Moon and Mars and hold value for risk managers and mission planners in performing cost-benefit trades for mission capability and research investments.

Introduction

With the exception of the U.S. Apollo missions, the history of human spaceflight has exclusively occurred in low Earth orbit (LEO). The longest lunar mission performed to date was Apollo 17; three astronauts spent 12 days traveling to and from the moon; two astronauts spent 75 h on the lunar surface[1]. Because of astronauts’ experience on long-duration spaceflight in LEO, the deconditioning effects of missions out to 6 months are generally well characterized. Significant unknowns remain, however, regarding the resilience of the human system during sustained lunar and Mars mission durations.

With the bulk of prior operational experience in LEO, it is easy for mission planners to make faulty assumptions regarding the magnitude of human system risk. In LEO, the risk from human system failure, or medical risk, has historically been small compared to the risk involved in getting to and from space[2]. As mission duration extends for sustained lunar and Mars missions, this risk balance shifts. Medical risk in a mission is strongly correlated to both distance from Earth and mission duration. This paper is designed to improve understanding and communication of medical risk in human spaceflight. Characterizing medical risk is part of addressing the question of whether astronauts will become ill or injured during a given mission and, by extension, whether they will be able to perform the jobs they are asked to do in those missions. This paper is the first publication of this type of medical probabilistic risk assessment (PRA) comparing evidence-based quantitative estimates of spaceflight medical risk across the spectrum of mission durations that NASA uses for planning purposes. As both government and commercial spaceflight consider LEO, lunar, and Mars missions over the next decade, an understanding of the changing risk to human health and performance associated with increased distance from Earth and mission duration is critical to mission success.

LEO operations include specific advantages in the medical risk arena that contribute to a low medical risk posture when compared to missions beyond LEO. As mission duration increases, deconditioning effects on the human body become more pronounced. In addition, distance from Earth creates three key operational changes that increase medical risk[3]:

Communication: Medical support in LEO depends on real-time communications to enable flight surgeons to provide telemedical evaluation and recommendations quickly and accurately. When real-time communications are no longer possible, the operational paradigm shifts to store-and-forward telemedicine. In this paradigm, the crew must act autonomously during the initial phases of any medical issue[4,5].

Resupply: Medical support also depends on a robust consumables resupply chain that exists in LEO. This resupply chain will be strained for lunar missions and non-existent for consumables on a Mars mission. This does not imply that pre-positioning of needed resources will not occur, but degradable resources like food and medications are not amenable to that model and this can alter the medical risk posture[6-9].

Evacuation: The option for evacuation of an ill or injured crewmember to definitive care is feasible in LEO in a reasonable timeframe. In lunar missions, evacuation times are longer which implies that a different set of medical capabilities may be required to stabilize an ill or injured crewmember for a much longer transfer time. In the case of a Mars mission, evacuation to definitive care is unavailable. These differences suggest that the current operating paradigm for medical support will need to change to mitigate increased risk[3]. The effective design and implementation of a Crew Health and Performance (CHP) System that anticipates these changes is important to mission success.

NASA has used the Integrated Medical Model (IMM) to inform International Space Station and other probabilistic risk assessment (PRA) based mission analyses[10]. However, estimates of medical risks for exploration spaceflight and comparison with past risk acceptance are lacking in the literature. This is in part because of unique challenges associated with modeling medical outcomes in human spaceflight. With fewer than 600 people having flown in space under a wide range of conditions and non-standardized monitoring and medical tracking across those missions, the evidence base needed to inform medical risk analysis is still evolving. NASA uses Design Reference Missions (DRMs) that provide a baseline set of mission assumptions to enable common risk assessment information[11]. Existing medical PRA approaches that were developed for LEO have limitations when applied to future missions beyond LEO. However, these types of PRA applications can still provide value in helping mission planners approach a reasonable order of magnitude and bounding for risk. IMM is used here to estimate medical risk differences between DRMs that represent past, current, and future NASA and commercial missions under consideration. The results are then considered in the context of evidence surrounding the effects of the operational changes from LEO discussed above and the effects of long-duration spaceflight on the deconditioning of humans.

Methods

The IMM was developed at NASA starting in 2008 and was transitioned to operations in 2017[12]. It is used here to provide quantitative risk assessment for each of the DRMs considered below. IMM models 100 medical conditions and includes the capability to assess the impact of resource limitation or depletion on successful treatment of medical conditions. Full description of the model and its validation is provided elsewhere in the literature[10,12-14].

Table 1 shows the DRMs considered for analysis including mission duration and number of crew. These are chosen based on the average durations and crew complements for existing missions or current planning for future missions. These durations and crew complements may change in future mission design iterations and so should be considered an approximation here. The model simulates two DRMs that reflect previous mission types including Space Shuttle (DRM 1) and the International Space Station (DRM 4). The remaining DRMs approximate potential future missions for which NASA needs risk assessment. DRMs 2 and 3 approximate short-duration LEO or lunar missions. DRM 5 approximates longer-duration LEO or lunar missions and Mars preparatory missions. DRMs 6 and 7 approximate initial and sustaining Mars missions.

Table 1

Design reference mission attributes.

DRM	Mission duration (days)	Number of crew
1	14	7
2	21	4
3	42	4
4	180	6
5	365	4
6	730	4
7	1195	4

The IMM incorporates evidence from all ISS missions as well as data from Apollo, Skylab, Mir, and Space Shuttle programs, but the resource table is baselined to the ISS medical system resources[10]. Medical capability here is defined as the complete set of resources that enable the crew to perform medical monitoring, diagnosis, and treatment for medical conditions that occur in spaceflight. Resources needed for treatment are explicit in the IMM resource table, but resources such as monitoring capabilities, diagnostic capabilities, and crew capability implicitly reflect the ISS operating paradigm in the outcomes data that feed the model. The ISS medical capability is comprehensive and rigorously scrutinized; it is specifically intended to provide options for minimizing LEO spaceflight medical risk[10]. The ISS medical capability is assumed here to represent a reasonable upper bound on the benefits that medical resources are likely to bring to the risk posture. This is because of increasing mass, volume, power, and data bandwidth restrictions that are expected to limit medical resources in missions that occur beyond LEO. This assumption could be challenged by advances in autonomous medical capability or propulsion technology that may address mass and volume limits, for example, future enhanced capabilities not currently available may result in more risk reduction for less mass and volume.

For the purpose of comparing the effectiveness of the fielded medical capability for each of the DRMs, we model in units of ‘ISS Medical Capability’[10] described below:

No Medical Capability—this case approximates a mission scenario where there is completely ineffective matching of medical capability to medical need within the mission or no resources available. In this case all conditions go untreated.

Unlimited ISS Medical Capability—this case approximates the best possible matching between medical capability and medical need that could be expected, based on the historic ISS Medical Capability. In Unlimited ISS Medical Capability, all conditions are modeled as fully treated.

Limited ISS Medical Capability—this case represents a tailorable example case of no resupply, i.e., an ISS medical capability that can run out of medications. Medical conditions for which there are insufficient resources to fully treat are modeled as partially treated or untreated.

This paper evaluates several mission level outcomes: Total medical events (TME); loss of crew life (LOCL), likelihood of reaching consideration of crewmember evacuation criteria (EVAC), and crew health index (CHI). (Note: The one hundred medical conditions modeled by IMM are shown in Extended Data in Table 2). The incidence of IMM medical events occurring during simulated missions is based on historical mission and cohort data contained within the Integrated Medical Evidence Database (iMED)[14,15]. For each condition, the probable percentage of occurrence of “best case” and “worst case” scenarios are specified, as well as a defined set of medical resources that are used to treat the condition. Best case conditions are those that present in the mild-moderate end of the clinical spectrum. Worst cases are those that present on the severe end of the clinical spectrum[10]. The iMED entry for each condition details the specific medical resources and quantity necessary for treatment in both the best and worst-case scenarios. In-flight medical treatment is assumed to follow a specified protocol or clinical practice guidelines for each medical condition and is constrained by resource availability on the ISS (i.e., the ISS Medical Capabilities)[10].

The IMM modeling process is shown in Fig. 1 below. It is described in detail elsewhere and is briefly summarized here[10,12-14].

Fig. 1

The IMM 4.1 Monte-Carlo simulation incorporates a mission timeline, progression-path assessment, treatment-path assessment, and event-outcome evaluation.

These are summed across all occurring conditions to provide a trial outcome and across all simulation runs to provide the simulation outcomes.

IMM uses stochastic processes via Monte-Carlo simulation. First mission and crew characteristics are specified for a given DRM. Each IMM Monte-Carlo iteration includes medical condition occurrence along the Mission Timeframe that progress to a Best or Worst Case scenario in the Progression Path Assessment Step. Best and worst-case scenarios for medical conditions are generated based on probability distributions. Incidence of medical conditions is held constant. During the Treatment Path Assessment step, available resources are queried and either Full, Partial, or No treatment is applied to the specific condition. Medical events, treatments, and outcomes are randomly generated based on probability distributions. Event Outcomes are evaluated and tallied to provide a trial outcome. Summation of the health and medical outcomes across all simulation iteration steps provides the primary outcomes[10]. Primary outcomes for this study include TME, CHI, EVAC, LOCL, and required resources.

CHI is a calculated percentage using the quality-adjusted mission time lost (QAMTL) due to in-flight medical events and resources available to treat those conditions.

For a given condition, QAMTL is determined by summing the product of functional impairment and duration for three clinical phases of that condition: (1) diagnosis and initial treatment, (2) ongoing treatment, and (3) end-state. Functional impairment (FI) is an estimated measure of the affected crewmember’s health and performance ability. An average for the entire crew and the mission are reported. CHI percent values range from 0–100, where zero represents complete crew impairment and 100 represents a completely functional crew. Functional impairments in the IMM are estimated using the AMA Guides to the Evaluation of Permanent Impairment’s general principles and rules[16]. Note that the AMA Guides estimate permanent impairment based on terrestrial norms. Application of these assessments in estimating functional impairment in a spaceflight environment likely overpredict impairment in some outcomes (i.e., lower limbs) and underpredict impairment in other instances (i.e., eyes, hands) Given that comprehensive space environment functional impairment data has not been collected, the AMA Guides are considered a sufficiently robust evidence source for making relative medical risk assessments.

EVAC in the context of IMM means that medical evacuation from the ISS is considered for definitive treatment of the afflicted crewmember. When this outcome is reached, it effectively ends the mission for that crewmember. Here ‘definitive’ treatment is defined as the best possible treatment available at a US tertiary care hospital on Earth. EVAC is considered an end-state result if any of the following criteria are met: (1) potential LOCL; (2) potential significant permanent impairment; or (3) potential intractable pain. When an EVAC state is reached during an IMM simulation, the availability of a return vehicle or the likelihood of a successful clinical outcome should a return vehicle be available is assumed to exist.

LOCL in the context of IMM is interpreted to mean that the clinical scenario resulted in death of the affected crewmember(s). During the simulation, the rarity of the LOCL and EVAC end-states results in crewmembers only assigned one end-state of EVAC or LOCL in most simulation trials. In rare instances, EVAC or LOCL for a second condition is reached prior to being reached for an earlier condition. Both events are recorded as EVAC or LOCL outcomes.

Required resources for treatment are tracked for each simulated mission trial. The initial quantities for medical resources are baselined to a fully stocked ISS medical kit. The quantity of each resource used to treat a given condition is decremented from available resources. If more of any particular resource is used than is available, then that resource is considered “depleted” in the Limited ISS Medical Capability scenario. Events that require a “depleted” resource are considered “partially treated” if only a portion of the required resources are available. If there are sufficient available alternate resources in the Single ISS Medical Capability scenario, the event is considered “fully treated.” If there are no resources available to treat the event, the event is considered “untreated.” When an event is “partially treated”, outcomes are based on a weighted average of the treated and untreated values. For example, if a condition has a 0% chance of going to EVAC in the best case treated scenario and a 0–100% chance of going to EVAC in the best case untreated scenario, then under best case partial treatment, it is possible the condition will go to EVAC.

Each IMM simulation consists of 100,000 Monte-Carlo trials, where each trial is considered a unique mission simulation. Convergence of each simulation is evaluated by confirming a <5% change in the average standard deviation of the CHI, EVAC, and LOCL model outcomes in the last 2 sets of 1000 simulation mission trials.

CHI is calculated as a percentage, with 95% confidence intervals (CIs) for the associated distributions. EVAC and LOCL are probabilities, with 95% CI of the mean for EVAC and LOCL obtained using bootstrap resampling of the simulation output.

Definition of crew attributes allow for tailoring crew-dependent variations by defining a limited set of individual factors that affect medical illness. These include sex, coronary artery calcium score, dental crowns, contact lens use, and prior abdominal surgery. The model does not consider most crew attributes that are already attenuated by the astronaut selection standards and flight certification standards[17]. Environmental Injury likelihood is attributed in part to individual crewmembers engaging in extravehicular activity, where decompression sickness, paresthesias, and fingernail delamination are medical conditions linked to EVA. Note that Tables 3–5 documenting the assigned individual attributes by crewmember used for each DRM are shown in Extended Data.