Identifying Predictors of Pressurized Submarine Escape Training (PSET) Attrition

Abstract

Pressurized Submarine Escape Training (PSET) physically prepares submariners to safely escape a submarine at depth. Failure to complete PSET is not a submarine service disqualification. Serious medical incidents are rare, but the safety record tradeoff has been low throughput. From 2009 to 2015, only 34% of students screened completed PSET. Students may be medically screened out of the training altogether (disqualified), or dropout during the physical training (attrite). Training records from 12,122 U.S. Navy students were used to identify factors contributing to training disqualification and attrition. Multivariate logistic regression model predictors included demographic and screening items. Association to PSET disqualification included cold/congestion/cough (OR_adj 12.34), limited duty status (OR_adj 4.29), Physical Readiness Test failure (OR_adj 3.37), pneumonia or bronchitis in last 2 years (OR_adj 3.17) and nervousness or anxiety in tight spaces (OR_adj 2.37). Basic Enlisted Submarine School students were more likely to be disqualified and attrite than other submariner groups, and black/African American (OR_adj 1.53) students were more likely to attrite than white students. Only cold/congestion/cough (OR_adj 1.52), trouble swimming (OR_adj 1.53), and screening during cold/flu season (OR_adj 1.28), were associated with training attrition. Recommendations to modify screening requirements are listed in conclusions.

INTRODUCTION

While unlikely, there is a possibility that a submarine may lose power or suffer catastrophic damage and become disabled (DISSUB). If this occurs, there are two possible courses of action to save lives: rescue (which is the preferred means of survival for the U.S. Navy) or escape. Rescue occurs when survivors evacuate the DISSUB into a submersible vehicle with external support. If conditions preclude rescue, escape is necessary. To escape, distressed submarine survivors individually exit the submarine through an escape trunk and ascend to surface with an escape suit and integrated raft. To prepare, the Navy’s Pressurized Submarine Escape Training (PSET) teaches Submarine Force officers and enlisted personnel safe escape from a disabled submarine at depth. Pressurized Submarine Escape Training also builds individual confidence in a planned escape, increases the chances that an escape will be attempted, and decreases anxiety while performing escape procedures.^1,2

Introducing submarine escape training methods at depth, the original escape tower in Groton, CT was built in 1930 and torn down in 1987. In the 1970s, the Navy ceased PSET due to degradation of facilities and raised health risk concerns.^3,4 Most notably, the escape equipment used consisted of an inflatable life vest and a plastic hood known as the Steinke Hood. In the event that a survivor made it to the surface, this equipment offered no environmental protection while awaiting rescue. After a few decades of non-pressurized escape training, the tragic sinking of the Russian submarine Kursk in 2000, and improvements in submarine escape and immersion equipment (SEIE) suits offering thermal protection, the Navy reviewed its escape training methods. As a result, construction for a new tower occurred in 2005 and in 2009 PSET was reinstated and officially resumed in Groton, CT.

Due to the dangers of pressurization and in-water training, the Navy classifies PSET as high-risk. Pressurized Submarine Escape Training has been associated with morbidity and mortality, particularly from diving related illnesses such as decompression sickness, and pulmonary over-inflation syndromes (POIS), including arterial gas embolism (AGE).^3,5,6 Further research has suggested those with lower-than-mean predicted spirometry measures may be at greater risk for PSET injury.⁷ Accounting for substantial hazards following water immersion and pressure, Weathersby et al estimated the medical risk incident rate associated with PSET hooded ascents at 1–6 per 10,000 escapes.⁵ Although these studies suggest some necessary medical screening components, the overall body of knowledge in this area has remained lacking. In the absence of compelling data, PSET medical screening requirements were developed to minimize risk and training accidents, through a consensus of subject matter experts in anticipation of PSET.⁸

Current PSET medical standards are more stringent than those required for U.S. Navy divers. Trainees at increased risk of gas embolism and barotrauma, and those with conditions potentially worsened within a hyperbaric environment are disqualified for PSET.⁸ Medical screening must occur within 72 hours prior to PSET and no physical standard may be waived. Failure to complete PSET is not a disqualification for submarine service. While other personnel may experience PSET, the only naval personnel regularly scheduled to attend the training are Basic Enlisted Submarine School (BESS) and Submarine Officer Basic Course (SOBC) students.

Since 2009, current screening methods and procedures have demonstrated an excellent record of safety (no permanent disability or death has occurred). Consistent with previous estimates,⁵ 1 case of POIS, which resulted in an AGE, occurred with successful treatment resulting.⁹ Unfortunately, the impressive safety record tradeoff has been a low throughput. A previous review found 41% of students were medically qualified for PSET at the time of screening, and only 32% of those screened successfully completed all phases of training.^10,11

Given the low PSET completion rate coupled with medical screening criteria developed with limited supporting data, the objectives of this study were to identify the specific practices and elements of medical screening and training associated with disqualifications and attrition, while also identifying groups of individuals most likely to dropout. Findings may provide evidence to foster change that could result in higher student completion rates without jeopardizing student safety. Increasing the number of students who qualify for and complete PSET should provide the submarine force with more sailors who are sufficiently prepared to escape from a disabled submarine.

METHODS

Pressurized Submarine Escape Screening and Training

Pressurized Submarine Escape Training lasts 2 days and includes medical screening, computer training, classroom instruction, surface survival, pressurized chamber test descent/ascent to 60 feet depth (18 .3 m, 2.8 atm), 15 feet depth (4.6 m. 1.45 atm abs) buoyant escape ascent, and pressurized ascent from 37 feet depth (11.4 m, 2.12 atm abs) for qualifying students using a SEIE suit. Medical screening includes demographics, medical history, medical record review, chest radiograph (CXR) review, physical exam, and pulmonary function test (PFT). The self-reported medical history form includes demographics and 38 questions used to assess the physical and psychological readiness of each student for pressurized training; including 2 catchall items for conditions or reasons not otherwise addressed (Table I). Each item renders a yes or no response. A confirmatory response on any screening item could potentially result in a PSET disqualification. In practice however, confirmatory responses are also used to alert the medical training staff of a specific physical condition that may need to be addressed prior to pressurization (e.g., wearing of contact lenses). Anytime during the 2-day course, students may be deemed unfit for training or voluntarily drop out.

Data Collection

Instructors enter screening and outcome data into a secure database, typically daily. To expedite data entry, most screening items default to a “no”, “not present” or “normal” response. Dive supervisor training logs are compared against database entries to ensure accuracy in training outcomes and database queries are employed to flag inaccuracies. Routine screening data collected on students and retained in the database includes basic demographics (i.e., age, gender, race or ethnicity, rank or rating, and current naval activity) and the aforementioned medical information. One of 10 broad disqualification codes is assigned to students not qualified for PSET. Those screened fit for PSET are tracked further through the 4 physical training phases (i.e., pressure test, surface survival, 15 feet ascent, and 37 feet ascent). Outcomes at each phase are recorded.

Following approval by the Naval Submarine Medical Research Laboratory (NSMRL) Institutional Review Board (Protocol NSMRL.2015.0006) in compliance with all applicable Federal regulations governing the protection of human subjects, access was granted to the Naval Submarine School’s (NAVSUBSCOL) PSET database. No intervention or interaction with subjects occurred, and no additional data were collected. Retrospective data from November 2009 to July 2015 were validated and assessed for systematic errors. If data were missing or contradictory, a catchall ‘‘notes” field was reviewed to correct input. However, very few data were missing because the database restricts data entry of the next medical screening item if the current item is skipped or invalid. Although rare, if a student was screened more than once, only their final screening information was included in the dataset.

Sample

During the reporting period, 12,122 students screened for PSET. Medical screening responses and training outcomes were recorded in the database for all students who arrived for medical screening. Paygrade was not recorded for 12 students; race or ethnicity was missing for 20 students; and age was missing for 13 students. Most students were male (99.4%) and attached to SUBSCHOL in BESS (75.9%), SOBC (17.0%), or Submarine Officer Advanced Course (SOAC) (3.9%). Other students (3.2%) included prospective PSET instructors, Undersea Medical Officer candidates, and other submariners attached to boats currently stationed in Groton, CT. All but 1 of the 67 females was an O1-O3 officer attending SOBC as an Undersea Medical Officer. The majority were in the E1–E3 pay grades (70.8%); 6.7% were E4-E6; and less than 1% were E7-E9. Junior officers (O1-O3) represented 18% of the sample. Warrant Officers (W1–W4) and other officers in the O4–O7 pay grades each comprised 0.1% of the remaining sample. The median age was 22 years (range 17–59). The distribution of race or ethnicity was; 74.1% white, 10.6% black or African American, 9.6% Hispanic or Latino, and 5.8% other (other included the self-reported categories American Indian or Alaska Native, Asian, Native Hawaiian, or other Pacific Islander, and other).

Screening and Training Outcomes

Students may have more than one disqualification reason. When this occurs, medical personnel select the disqualifying reason that best fits the current scenario. A careful review of the disqualifying medical screening items showed many of the “other” reasons were due to scheduling issues. To account for this in the analyses, an 11th not qualified code was added. The screening process renders three possible outcomes; qualified for PSET, qualified for unpressurized submarine escape training (USET); or not qualified for any physical training (NPQ). With only 1% of screenings resulting in NPQ, screening outcome was collapsed into a dichotomous outcome variable by combining USET and NPQ into 1 group, thereby rendering either a PSET qualified or disqualified screening outcome. Retaining the NPQ students in the model allows for best predictability in describing the subgroups of students who are not qualified for the 37 feet ascent, the ultimate goal in the training.

Not all PSET qualifiers complete the entire PSET evolution successfully. Regardless of screening qualification status and where students dropped in the training pipeline, only students completing both the 15 and 37 feet ascents receive PSET as their training outcome; otherwise, they receive USET as their training outcome. For those who qualify for PSET but do not complete training, attrition reasons and the phase of training where the dropout occurred are also tracked.

Predictor Selection

Physical exam, health assessment, and PFT values are only measured and recorded for students remaining in the screening/training pipeline after the self-administered medical history is assessed; therefore, these items were not included as possible predictors. Further, actual PFT measurements (i.e., force expiratory volume in 1 second and force vital capacity) were not collected for 48.4% of students; this included those who did not progress to this training stage, did not meet the criteria, and those who did pass, but were just entered as pass/fail. Therefore, PFT results were also excluded as possible predictors. Additionally, a missing or abnormal CXR or a PFT failure results in immediate PSET disqualification thereby eliminating these factors as possible predictors.

For screening and training predictor selection, bivariate logistic regression analysis was conducted on all 5 demographic variables (i.e., pay grade, race or ethnicity, age, gender, and activity group), 38 medical screening items, and 1 additional item based on whether or not the student was screened during the months of December, January, February, or March (i.e., cold and flu season). The 38 medical items included current and historic health indicators specific to pulmonary, cardiovascular, psychiatric, neurological, genitourinary, and ear, nose and throat diseases, symptoms, and disorders as well as other items to assess physical and psychological readiness (Table I). The cold/flu season item was added after an examination of data showed a high frequency of positive responses on the cold/congestion/cough question (item 25 in Table I) from December through March (Fig. 1) and the lowest qualification rates from January to March (Fig. 2). Six questions addressing Post Traumatic Stress Disorder, emotional problems, self-destructive behavior, sleeping less than 4 hours, eye impairments, and hypotension or hypoglycemia were added to the screening form in May 2010 potentially excluding 990 students from the multivariate prediction analyses. Because the O4–O7 and W1–W4 groups consisted of so few members, all officers were collapsed into 1 category, and Warrant Officers were collapsed with Chiefs (i.e., E7–E9).

Data Analysis

Data analyses were conducted using IBM SPSS Statistics, Version 23. Reasons for pipeline washout and total, and yearly trends in screening results by disqualifying factor were examined. Descriptive analyses characterizing students screened and their distributions were compared between the PSET qualifiers/disqualifiers and PSET completions/attritions.

Model building for logistic regression analyses was conducted to determine if the 5 demographic variables, 38 medical screening items, and the 1 seasonal item (44 possible predictors) could distinguish PSET qualifiers from disqualifiers and completions from attritions when accounting for other variables in the model. Initially, bivariate odds ratios (ORs) were calculated for all potential predictors. Two separate multivariate prediction models were then developed to predict screening (N = 12,122) and training (n = 5,261) outcomes. To be included in either multivariate analysis, the model building strategy required each potential covariate to have a bivariate OR p-value <0.25 and a minimum of 10 events for the qualifier/disqualifier groups and completion/attrition groups. Using backward elimination, factors meeting these criteria were entered using the p < 0.05 for entry and p > 0.05 removal selection criteria. Collinearity diagnostics were performed, and model results from several regression models were compared with and without suspected confounders. The final multivariate models included all predictors shown to be independently associated (p < 0.05) with the outcome variable. Adjusted odds ratios (OR_adj) were used to rank order medical screening predictors and seasonal predictor by strength of association with outcome.

RESULTS

During the first 6 years of PSET, 43% of the 12,122 students screened were deemed physically qualified for the pressurized training (Fig. 3) with the remaining 56% qualified for unpressurized training or unfit for either training thereby receiving no course credit (1%). However, only 34% of all students successfully completed all phases of the training evolution (79% of those deemed qualified for PSET). Hence, 21% of those initially qualified did not go on to complete the training successfully. This resulted in 65% completing USET. Figure 3 shows most students were screened out due to cold/congestion/cough (Table I, item 25) (16%); followed by difficulty clearing ears, no visualization of tympanic membrane (TM), or inability to Valsalva (12%); and PFT (spirometry) failure or unmeasurable PFT (12%).

Regarding the 130 students who were not physically qualified for PSET or USET, most (60%) were disqualified due to a skin condition (which, to avoid contamination, prohibits them from wearing the escape suit) (15%), limited duty status (15%), cold/congestion/cough (15%), or an unspecified reason (15%) that could not be further determined from record notes.

PSET Disqualification Predictors

Table IIA and IIB shows bivariate and multivariate associations between student demographics and medical screening items and PSET screening outcomes. The demographic distributions between the PSET outcome groups appear similar, however, bivariate ORs show significant differences (p < 0.05) for activity group, pay grade, race/ethnicity, and age. A slightly higher proportion of BESS students were deemed not qualified as were those with E1–E3 pay grades, black or African American students, and those aged 17–22 years. Due to collinearity between the activity group and pay grade predictors, few students in the E7–E9/warrant officer category in pay grade, and a greater association between activity group and qualification outcome than pay grade and qualification outcome, pay grade was excluded and activity group was entered in the multivariate model. Gender and 8 screening items were also excluded because of a bivariate OR p-value >0.25 or a minimum of 10 events in the qualifier or disqualifier group.

Including activity group, race, and age, 34 possible predictors entered the multivariate model. Activity group was the only demographic entered in the adjusted model found to be associated with PSET disqualification. When accounting for other variables in the model, BESS students were more likely to disqualify than SOBC, SOAC, and other submariner students. Although bivariate analysis showed race/ethnicity and age were associated with disqualification, neither variable was retained. Seventeen out of 30 medical screening items were retained in the final model as well as the seasonal cold/flu item. The screening items with the strongest association to PSET disqualification were cold/congestion/cough (OR_adj 12.34); current limited duty status (OR_adj 4.29); inability to pass Physical Readiness Test (PRT) (OR_adj 3.37); pneumonia or bronchitis in last 2 years (OR_adj 3.17); and nervous or anxious in tight spaces (OR_adj 2.37). Thirteen medical items with a minimum of 10 events for qualifier/disqualifier outcome and a significant (p < 0.05) association with disqualification for the bivariate analysis were not retained in final model.

Training Attrition Predictors

Regarding the 5,261 students qualified for PSET, 79% (34% of total screened) completed all phases of the training. For the 21% that did not complete, Table III shows the phase these qualified PSET students were dropped from training and disqualifying reason (n = 1,055). Most of these students (52%) were dropped during chamber pressurization and 82% of the drops during the pressurization phase were attributed to middle ear barotrauma (MEB) ranging from mild to severe. The most frequent other medical related reasons were inability to clear/difficulty clearing the ears and sinus squeeze. Most trainee related reasons were due to a trainee procedural issue, and mechanical incidents unrelated to the students were most commonly caused by a water visibility issue or an unspecified mechanical issue leading to cancellation of training. Given the small number of students who were PSET qualified and subsequently dropped due to mechanical, trainee, or medical reasons other than MEB, only the associations with medical screening items and MEB were examined. To identify associations with attrition due to MEB, responses to select medical screening items were compared between students who were dropped due to MEB (n = 581) to those who completed PSET (n = 4,144). Results show those who did suffer a MEB during the pressure test, buoyant ascent, or hooded ascent, were more likely to have trouble swimming, drank alcohol in last 12 hours, or currently smoked (p < 0.05).

Model building to predict training attrition with less than half of the students remaining in the pipeline revealed fewer significant associations. The bivariate logistic regressions for most student demographics showed significant association (p < 0.05) to training attrition (Table IVA and IVB). Other activity groups were less likely to attrite during training than BESS students, and officers were also less likely to attrite than the E1–E3 reference group. The sample’s mid-aged groups (23–34 years old) were also less likely to attrite than the youngest students. Gender and attrition were not associated, but black or African American groups were more likely to attrite than the white reference group. Hispanic or Latino and other race or ethnicity groups showed no difference compared to the white reference group.

As was the case for the disqualification multivariate regression, pay grade was dropped from the model due to collinearity with activity group. Thirty medical screening items and gender were also excluded because of a bivariate OR p-value >0.25 or a minimum of 10 events in the qualifier or disqualifier group. Ten medical screening items meeting entry criteria, and activity group, race/ethnicity, and age category entered in the multivariate model. Only activity group, race/ethnicity, cold/congestion/cough in last week (OR_adj 1.52), trouble swimming (OR_adj 1.53), and screening during cold/flu season (OR_adj 1.28), were retained in the final model (Table IVA and IVB). Table V lists the strongest medical screening factors and the added seasonal item associated with PSET disqualification and PSET training attrition.

Yearly Trends in Screening Outcomes

To determine if screening results have altered since training was reinstated, Figure 4 compares yearly screening results by broad disqualification category. Overall, there was an upward trend in the proportion of students qualified for PSET since the new facility opened in 2009. For years with 12 months of data (excludes 2015), 52% of students were physically qualified for PSET in 2014, in contrast to 36% who qualified in 2012. Across all years, the leading disqualification reason (i.e., cold/congestion/cough) has excluded a yearly minimum of 13% (2011) to upwards of 17% (2013) of all students screened. Non-specificity of disqualification reasons sharply increased in 2015, with 9% reported as “other” compared to previous years of 1–2% non-specified.

DISCUSSION

Pressurized Submarine Escape Training screening aims to identify health issues or conditions that may affect fitness for training. Retrospective training results identified factors associated with training disqualification and attrition. Potential predictors of screening and training outcomes included demographics, medical screening items, and a seasonal factor. Medical screenings deemed most students unfit for PSET. Of those physically qualified, the majority successfully completed PSET.

The following factors: cold/congestion/cough; trouble swimming; being screened during peak flu season; and being a BESS student were associated with both PSET disqualification and attrition. After adjusting for other predictors, students exhibiting cold/congestion/cough were 12 times more likely disqualified due to medical screening. The concern with cold/congestion/cough is the potential for barotrauma of middle ear or sinuses. A positive response to this screening item could indicate symptoms or a diagnosis, the severity of which is unclear without examination by medical personnel. Considering colds are a temporary health condition, it’s likely that many of these students would qualify at a later date once symptoms resolved. The same holds true for most other factors associated with disqualification, particularly: limited or light duty status; unable to pass PRT; dental problems; open cuts/sores, stitches, skin conditions, or new tattoos; and less than 4 hours sleep. Rather than relying solely on item response, history or presence of an illness could prompt a referral for examination by a physician, and a clinical decision on whether the student is physically qualified for PSET.

Due to holiday stand down and scheduled maintenance usually occurring in December, fewer screenings occur during the cold and flu season. However, with a 13% and 18% higher likelihood of disqualifications and attritions, respectively, occurring from December to March, attempts to rescreen during other timeframes warrants consideration.

After adjusting for other factors, BESS students who were the youngest activity group, had the least military experience, but were perhaps the healthiest, were more likely to disqualify and attrite than other activity groups. This could relate to survivorship bias or resiliency. Those who have progressed further in their career may be more confident or better equipped to handle high-risk training. Furthermore, PSET is not a requirement; students may drop out for any reason. Less experienced sailors may be skeptical of high-risk training or less confident in safety measures. The higher likelihood for BESS students to disqualify and attrite is concerning as they comprise 76% of students screened for training as their schedules have been designed to accommodate for the training. This is not true for their enlisted nuclear-trained counterparts who typically only receive USET at Naval Nuclear Power Training Command.

There was no significant association between race/ethnicity and medical disqualifications. However, black/African American students were more likely to attrite during training when compared to the white reference group. This contradicts studies examining causes of military attrition. Citing different motivations to join the service, minority groups were found less likely to attrite from military service than their white counterparts.^12,13 However, the nature of PSET (i.e., water immersion) may influence students to self-select out. Some research has found fear of drowning to be higher in African Americans than white counterparts,¹⁴ and African Americans and Hispanics or Latinos to have a higher “at-risk” swimming ability than white subjects.¹⁵ Additionally, in an Army setting, drowning victims were disproportionately black.¹⁶

Despite comprehensive medical screening, MEB led the reasons for training attritions (55%), usually occurring during the pressure test. Given the high number of MEB, albeit mostly mild and not severe, whether continuation of chamber pressurization to 60 feet is necessary preparation for a 37 feet escape warrants future discussion. Those who did have a MEB were significantly more likely to report difficulty swimming, alcohol use in last 12 hours, or current smoking behavior, compared to those who completed the training. This finding justifies the need to screen for these behaviors. A recent Canadian Forces study on pressurized escape training also found smokers more likely to experience otic barotrauma.¹⁷

While the number of disqualifications and attritions due to anxiety and other specific psychological causes cannot be made, 5 of the 18 factors found associated to PSET disqualifications specifically address mental health. Additional psychological considerations in PSET training could potentially increase throughput and increase safety. Future research in state anxiety, intervention and preparation, water orientation, and personality variables as they relate to PSET outcomes could provide data driven modifications to current training practices.¹⁸

During the 2014–2015 timeframe, escape trainer undersea medical officers reported implementing modified training to their medical screeners to assist with improved Valsalva detection and training to students to increase throughput (personal communication). Although a certain correlation cannot be made, 2014 and 2015 showed some of the fewest disqualifications due to ear issues that include inability to Valsalva or to detect Valsalva. Overall, since 2009, medical screening disqualifications have decreased. While it is difficult to ascertain the underlying causes of this, disqualifications due to cold/congestion/cough are more volatile across years. This could be due to changes in medical staff (approximately every 2–3 years), varying subjective measures of disqualifications criteria, and the severity of cold/flu seasons from year to year.

Results of this study are limited because physical exam, health assessment, and spirometry results could not be ruled out as predictors. These factors are only measured and recorded for students remaining in the screening/training pipeline after self-reported health and medical history are assessed. Future efforts to record these measures and spirometry values, as opposed to pass/fail, could further describe the trainee population and explain disqualification and attrition causes.

With only 67 females screened, no gender differences could be determined. Future analyses should aim at identifying differences in screening results and training outcomes between female and male students. This could result in additional opportunities for change in screening requirements and procedures.

CONCLUSIONS

• – The following medical practices should be the focus for modification review; congestion/cough/cold; difficulty clearing ears, inability to Valsalva or to visualize Valsalva; and spirometry failures.

• – Exploring opportunities to rescreen students who were medically disqualified for the following temporary reasons would improve throughput; missing physical exam, chest radiographs, or medical record; temporary health issues; limited or light duty status; inability to pass PRT; and open cuts, sores, or skin conditions.

• – Maintenance and other temporary closures of the escape trainer should occur during the December, January to March timeframe.

• – Expanding PSET opportunities to nuclear-trained enlisted personnel and submarine officers should be considered.

Previous Presentations

Presented as a poster at the 2017 Military Health System Research Symposium.

Funding

This study was supported by the Defense Medical Research and Development Program (DMRDP), work unit F1501. This supplement was sponsored by the Office of the Secretary of Defense for Health Affairs.

Acknowledgments

We thank the Naval Submarine School leadership and give special thanks to the former and current Undersea Medical Officers, medical staff, and instructors at the Submariner Escape Trainer, Groton, CT.

References

Author notes