Using Latent Class Analysis to Create Allostatic Load Profiles to Investigate the Effects of Occupational and Perceived Stress in Firefighters

Abstract

Background: Allostatic load is a construct used to assess the sum of the effects of physiological stress across multiple body systems over time. Allostatic load is typically measured using the allostatic load index (ALI); however, this measure does not fully capture the multivariate nature of allostatic load. Using allostatic load profiles as opposed to the commonly used ALI, we hope to explore the multivariate nature of allostatic load and understand its association with perceived stress, since the existing literature exploring the association between perceived and physiological stress is inconclusive.

Objectives: The objectives of this thesis are to 1) develop allostatic load profiles using latent class analysis of both study-created and clinical-based thresholds for biomarkers of stress; 2) examine the association between allostatic load profiles and perceived stress in firefighters; 3) assess whether study-based or clinical-based thresholds are more suitable when measuring stress in the firefighters in question.

Methods: Using available biomarker data from a sample of 57 male firefighters in Waterloo Fire Rescue, we developed allostatic load profiles using latent class analysis. Biomarkers included systolic blood pressure (SBP), diastolic blood pressure (DBP), hemoglobin A1c (HbA1c), low-density lipoprotein (LDL), high-density lipoprotein (HDL), heart rate variability (HRV), cortisol, waist to hip ratio (WHR) and body mass index (BMI). We then employed logistic regression to assess the association between allostatic load profiles and perceived stress in these firefighters.

Results: Our results demonstrated that the use of allostatic load profiles (ALPs), created with study-based thresholds, showed how different biomarkers contribute to elevated or non-elevated physiological stress profiles. In regression models of ALP on Perceived Stress Scale (PSS-10) scores, we saw a consistent positive association such that an increase of 1 unit in PSS-10 score increased the odds of being in the elevated ALP group anywhere between 13.6% (OR = 1.136, 95% C.I. = 1.012, 1.299) in a model with PSS-10 only to 19.2% (OR = 1.192, 95% C.I. = 1.039, 1.400) in a model with PSS-10, length of service (LOS), and the following behavioural confounders: smoking, sleep hours, exercise and alcohol intake frequency.

Conclusions: Allostatic load profiles captured the multivariate nature of allostatic load and demonstrated a significant association with the PSS-10, whereas ALI was unable to significantly demonstrate a relationship with perceived stress. The results of this thesis also demonstrated the need for using study-based or study-specific thresholds when examining unique populations with different fitness levels than the general population. Further research can benefit from the use of allostatic load profiles in conjunction with study-based thresholds to accurately and completely address the needs of persons working in high-stress, dangerous occupations.