Resilient responses to stressful events among chiropractic students: A cross-sectional survey

Setting and Participants

The study setting was the Palmer College of Chiropractic in Davenport, Iowa, USA. The doctor of chiropractic (DC) program is completed across 10 trimesters, spanning 3.3 years, with the option for individualized curricular schedules for students who prefer reduced class loads. A baccalaureate (BS) track completed in tandem with the DC program offers students the ability to practice in states that require a BS degree.

We did not perform a formal sample size calculation for this pilot project. Rather, we used a convenience sample of chiropractic students who were willing to complete this survey to decrease coverage error by including all chiropractic students (n = 873) enrolled in the DC program in March 2019.

Questionnaires

Demographic variables included self-reported current trimester, time in the DC program, age, marital status, gender, veteran status, previous careers, highest degree earned, and grade point average. In this study, the gender demographic represented the multidimensional construct of gender identity among participants rather than the sex assigned at birth.^29,30 Self-reported options for the gender variable included male, female, and other. For this survey, male and female were used for gender to reflect the spectrum gender may represent.³¹ Other options on the survey consisted of nonbinary, transgender, and those who preferred not to identify.³¹ Gender and associated options were based on culturally accepted and supported evidence.^29,30 See Appendix A (available online accompanying the article) for details on survey content.

The RSES served as our measure of resilience, or the cognitive, behavioral, and emotional coping responses of chiropractic students.^13,14 The 22-item RSES is rated on a 5-point scale ranging from 0 = not at all like me to 4 = exactly like me. Items are calculated in the positive direction for a total summed score in which “higher scores indicate higher levels of protective responses to stressful life events.”^14(p162) The RSES can be interpreted as a flat score or by its 5 domains (Fig. 1)³²: Meaning Making and Restoration (domain 1), Active Coping (domain 2), Cognitive Flexibility (domain 3), Spirituality (domain 4), and Self-Efficacy (domain 5).¹⁴ The RSES was developed and resulting data validated with predominately male samples of active-duty military personnel of the same age as health professions students and in veterans, with internal consistency (α = .91–.93) and test-retest reliability (r = .87) demonstrated.¹⁴ Although developed by the miliary for use in active-duty personnel, the RSES has been used in civilian settings, such as with emergency first responders, although there is limited research in this area.^33,34

View Figure

• Download as Powerpoint
• Download Figure

The Perceived Stress Scale (PSS)^35–37 measured student perceptions of the intensity of stress levels over the past month. The PSS is a 10-item instrument, scored on a 5-point Likert-type scale (0–4), with answers ranging from never to very often, with higher total scores suggesting greater perceived stress. A psychometric review of data from the PSS measures suggests adequate levels of validity and internal consistency of about α = .7.³⁷ The PSS is commonly used with students in chiropractic training programs internationally, with most studies indicating moderate to high levels of perceived stress, particularly among female chiropractic students.^1,2,17,38

The World Health Organization Quality of Life (WHOQOL-BREF)^39–41 provided a snapshot of student perceptions of their general QOL. While studies have evaluated QOL among health care students, we did not find studies that had used this concept or specific measure in chiropractic students.^42–45 The WHOQOL-BREF, an abbreviated version of the 100-item WHOQOL-100, contains 26 items, including 2 benchmarks for QOL and general health and 24 individual items broken down into 4 domains (Fig, 1): domain 1, Physical Health; domain 2, Psychological Health; domain 3, Social Relationships; and domain 4, Environment. Each item used an adaptation of a 5-point (0–4) Likert-type scale, asking how much, how completely, how often, how good, or how satisfied. Raw scores are calculated using the 5-point interval scale, with domain scores omitted if >20% of a domain, or 1 of 3 questions of domain 3, is unanswered. To account for an unequal number of items in each domain, scores are transformed to a 0–100 scale for comparison between domains. The internal consistency of data resulting from the WHOQOL-BREF is reported as [original: all domains α = .65–.93; international: domains 1, 2, 4 α > .7, domain 3 α = .51–.77].^40,41

Data Collection

Data were collected using SurveyMonkey (SurveyMonkey Enterprise, San Mateo, CA, USA), an online survey platform, from March 13 to April 2, 2019. The Department of Institutional Research and Effectiveness (IRE) distributed an email survey link to all DC students. All students, regardless of their survey completion status, received multiple communications from IRE over a 1-month deployment period. An email sent 1 week prior to survey release contained a brief study description, contents of the survey, and important disclosures regarding the anonymous and voluntary nature of participation and ability to skip questions. Before entering the survey, students consented to participation using a check box. Email-linked surveys and encouragements were sent for 3 consecutive weeks following study launch. The 4th email included a final encouragement and restated the importance of student participation. Institutional research personnel collected, stored, and analyzed the resulting data. No participation incentives were offered.

Data Analysis

Data were exported from the survey platform into Microsoft Excel (Microsoft Corp, Redmond, WA, USA), then uploaded into IBM SPSS Statistics, v.25 (IBM Corp, Armonk, NY, USA) for analysis. Descriptive and inferential statistics were calculated, and a data summary was generated on SPSS (Tables 1–7). Internal consistency reliability was examined using Cronbach’s alpha, with scores >.7 indicating adequate reliability.

Table 1 Respondent Demographics

View Fullscreen

Table 2 RSES Flat Score Hierarchical Multivariate Regression Model

View Fullscreen

Table 3 RSES Domain 1 (Meaning Making) Hierarchical Multivariate Regression Model

View Fullscreen

Table 4 RSES Domain 2 (Active Coping) Hierarchical Multivariate Regression Model

View Fullscreen

Table 5 RSES Domain 3 (Cognitive Flexibility) Hierarchical Multivariate Regression Model

View Fullscreen

Table 6 RSES Domain 4 (Spirituality) Hierarchical Multivariate Regression Model

View Fullscreen

Table 7 RSES Domain 1 (Self-Efficacy) Hierarchical Multivariate Regression Model

View Fullscreen

Hierarchical multivariate regression (HMR) assessed both the RSES flat score and 5 RSES domain outcomes against 12 predictors: “How would you rate your QOL (single-item QOL),” “How satisfied are you with your health,” WHOQOL domains (Physical Health, Psychological Health, Social Relationships, and Environment), PSS, and gender, age, trimester, highest degree earned, and year at the college.

Mean imputation was used if ≤20% of the answers of any 1 domain were left unanswered per participant, and participants were omitted from the analysis if more than 1 domain score was missing. If only 1 domain score was missing, the mean imputation was used on that single domain to ensure completion across all domains.

Variables were entered across 4 steps in accordance with theoretical or empirically known impact on RSES scaled score: step 1, gender; step 2, single-item QOL, How satisfied are you with your health, and the 4 WHOQOL domains; step 3, age, trimester, highest degree earned, and year at the college; and step 4, PSS.

Interpretation of HMR results occurred primarily using the standardized beta (β), R², R² change, and semipartial squared (sr²), where β indicates the direction (ie, positive or inverse), importance (ie, which variable contributes more), and strength of the effect (ie, a higher absolute value indicates a stronger effect) for each contributing variable. The R² and R² change statistics indicate goodness-of-fit (ie, accounted for variance) for the regression model, where the higher the value the better the fit, and sr² indicates the unique variance contribution of individual variables. The alpha value chosen for this pilot study was .05.

Assumption Testing

Data collection resulted in a robust sample size with almost double the recommended cases needed for HMR analysis (ie, at least 10 cases per indicator). Inspection of bivariate correlations indicated singularity (ie, the absence of extreme correlations ±1.0). Although the WHOQOL Psychological Health exhibited high correlations (≥.7) with the single-item QOL (r = .724) and the PSS (r = −.703), all variables met conventionally acceptable multicollinearity thresholds for variance inflation factor (<5; range: 1.000 to 4.145).⁴⁷ Examination of z-scores and Mahalanobis distance occurred for univariate and multivariate outliers. Two multivariate outliers were indicated but retained for data analysis after the data were further inspected. Residual and scatter plots indicated that the assumptions of normality and linearity were all satisfied.

Internal Consistency Reliability

The PSS (k = 10) exhibited acceptable reliability, α = .886, scale mean = 17.91, and σ2 = 43.02. Three of 4 WHOQOL domains exhibited lower than expected reliability: Physical Health (k = 7; α = .371, scale mean = 22.62, σ2 = 7.64), Psychological Health (k = 6; α = .459, scale mean = 20.24, σ2 = 9.25), Social Relationships (k = 3; α = .681, scale mean = 10.81, σ2 = 6.45), and Environment (k = 8; α = .739, scale mean = 29.91, σ2 = 19.55). Given the presence of domains within the RSES,¹⁴ calculating Cronbach’s alpha on the entire instrument would violate the assumption of tau equivalence. The RSES domains exhibited acceptable reliability: Meaning Making (k = 9; α = .868, scale mean = 24.22, σ2 = 42.06), Active Coping (k = 8; α = .796, scale mean = 22.23, σ2 = 25.27), Cognitive Flexibility (k = 3; α = .690, scale mean = 7.48, σ2 = 6.22), Spirituality (k = 2; α = .884, scale mean = 4.05, σ2 = 8.57), and Self-Efficacy (k = 2; α = .786, scale mean = 5.95, σ2 = 3.08).

Regression: RSES Flat Score

Explanatory variables in step 1, F(1, 219) = 9.931, p < .002, R² = .043, and Step 2, F(7, 213) = 20.991, p < .001; R² = .408, were significantly predictive of the outcome RSES flat score. The linear combinations of variables in step 2 predicted variance in the model over and above that of steps 1, 3, and 4 (ΔR² = .365, p < .001; Table 2).

Three variables contributed to 41% of the variance of RSES. Gender (β = −.181, t[219] = −3.328, p < .001; sr² = .031), single-item QOL (β = .213, t[219] = 2.556, p < .011; sr² = .018), and WHOQOL Psychological Health (β = .505, t[219] = 5.509, p < .001; sr² = .084) significantly predicted RSES flat score. Examination of the RSES flat score suggested that males who report higher ratings on the single-item QOL and higher ratings for WHOQOL Psychological Health also exhibit higher resilience to stressful events. Psychological Health was the most important model predictor within the linear combination of variables concerning the RSES flat score.

Regres sions: RSES Domains 1–5

Study Limitations

Although the study did achieve high survey coverage of the population, it experienced a degree of survey nonresponse error. The study did, however, yield a robust sample size with more than double the number of cases needed to adequately complete the HLM quantitative analyses.

Despite the adequate representation of age and degree demographics in our resulting sample, the current study’s findings are not generalizable to chiropractic student populations broadly due to its nature as a convenience sample at 1 US chiropractic college, the lower response rate, and the low number of transgender, nonbinary, or gender nonconforming students who took the survey. The disproportionate number of declines by male students, paired with the unbalanced proportion of female to male and other responders, may skew the results, since males make up a higher proportion of chiropractic students at this institution and male gender was a strong predictor of RSES scores across most domains. That male students were more likely to decline to complete the survey raises the question as to why. Are male chiropractic students less interested in biopsychosocial concepts than female students are? Do male and female students prioritize their time differently and, if so, how and why? Do other reasons drive these gender differences? And how might students who do not identify as male or female experience the stressors of chiropractic school? All of these questions offer much in the way of future research opportunities. Future studies may consider incentives to encourage participation and increase sample sizes.

Of those students who completed the questionnaire, an equal proportion of males and females omitted data. However, most of those omitting amounts of data disqualifying their survey were female. The reasons students omitted items are unknown. Perhaps students hesitated to self-disclose unfavorable answers, despite guarantees of anonymity, in courses taught by known faculty members. Such hesitancy to disclosure negative self-perceptions might also explain our low sample size. Perhaps nonparticipants or those who omitted items did not want to appear less resilient. Maybe questions about mental health were upsetting. Refusal to participate may also represent a coping strategy, such as protecting one’s time for other activities. Future studies with more representative samples and additional chiropractic educational institutions could increase generalizability to other chiropractic student populations. In addition, data collection for this survey occurred in the spring of 2019, well before the start of the COVID pandemic. Studies exploring the resilience of chiropractic students either during or postpandemic may report different results than those reported here.

While this study relied on established questionnaires, in some cases the current study found lagging validity and reliability evidence for chiropractic students. As such, utility may be less robust than originally anticipated during the study design phase of this project. Validation and reliability testing of the RSES in larger chiropractic student cohorts is recommended. We believed similarities existed between chiropractic students and active military personnel in our choice to use the RSES, as these populations are similar in age. However, the RSES was designed and validated in male-dominated samples (88%–99%) with significant combat experience, including receiving hostile fire (88%) and joining battle missions (83%).¹⁴ Resilience under combat circumstances may differ appreciably from resilience required in demanding classroom settings. Thus, the content validity of RSES items should be established with subject matter experts (ie, chiropractic students). Such validation would bolster our understanding of the unique stressors and resiliencies of this population and is recommended by the authors of the RSES, who considered the instrument as potentially appropriate for civilian populations for whom stress or trauma were likely.¹⁴

This study is among the first to use RSES to report gender differences in resiliency scores, which makes this study a unique contribution to the literature, underscoring the need for the chiropractic profession to develop and test an RSES instrument to measure the resilience experiences unique to chiropractic students. Perhaps the RSES does not fully capture key facets of resilience in student populations, particularly for female students. Innes² reported that females are more likely to choose emotion-based coping styles, although this trait did not fully explain differences in resilience scores in their sample. Therefore, construct validity might be explored, especially in relation to personal qualities that, like resilience, educators might seek to instill in chiropractic students, such as hardiness and grit, and for which there may be some conceptual overlap.

Qualitative research could allow a deeper look into the reasons for the trends we are seeing in our results regarding resilience scores and gender as well as that factors that play into the discrepancies seen between those identifying as male, female, and nonconforming or transgender. Finally, establishing the predictive validity of the RSES would be useful for studies evaluating evidence-based interventions for coping and increased resilience of chiropractic students over the course of their academic career.