Effects of stress on pain in females using a mobile health app in the Russia-Ukraine conflict

The study investigates whether acute stress during the onset of the Russia-Ukraine war produced stress-induced analgesia (SIA)—a reduction in pain perception associated with stress—under real-world ecological conditions. Prior work shows stress can facilitate or impair cognition depending on context, and laboratory research consistently demonstrates SIA following exposure to stressors. However, systematic measurement of stress and pain during extreme real-life events like armed conflict is scarce. The authors leverage a large-scale menstrual tracking app (Flo) to examine temporal dynamics between self-reported stress and pain among women in Ukraine before and after the invasion began on February 24, 2022. They hypothesized an inverse relationship between stress and pain at the start of the conflict, with effects diminishing over subsequent weeks, and with smaller effects in neighboring countries compared with Ukraine as the epicenter.

Foundational work documented reduced pain reports among wounded soldiers in wartime contexts, suggesting contextual modulation of pain by stress. Extensive lab-based research has replicated SIA and delineated neuroanatomical, neurochemical, and molecular mechanisms, including roles for endogenous opioids and supra-spinal circuits. Beyond pain, stress impacts cognition and memory, sometimes enhancing encoding of stress-relevant information. War and armed conflict are associated with profound long-term health consequences, including elevated rates of depression, anxiety, and PTSD among civilians, with women often exhibiting heightened vulnerability due to increased responsibilities and exposure to stressors. Mobile and wearable technologies now enable ecological monitoring of psychological and physiological states, and menstrual tracking apps provide longitudinal symptom logs (e.g., stress, pain, mood), offering an opportunity to assess SIA outside the lab during acute societal stressors such as the Russia-Ukraine conflict.

Design and data source: Observational ecological study using self-tracked symptom data from the Flo mobile app. Users can log daily symptoms (e.g., pain types, mood states, stress) once per day per symptom. Population and time window: Included Flo users whose smartphone location settings were set to Ukraine at last login as of 2022-04-06 and who were active between 2022-01-25 and 2022-03-26 (30 days before and 30 days after the invasion on 2022-02-24). Total Ukrainian users registered: 3,732,111. Users logging symptoms: 201,572 before and 196,311 after the start of the war. Total symptoms logged: 1,844,546 before and 1,807,209 after. Analytic sample: To mitigate selection and collider biases, analyses focused on users who logged at least two symptoms across the window and at least two symptoms on any given day. This yielded 87,315 users, with 1,234,133 symptoms logged before and 1,176,613 after. Age (optional, potentially error-prone) was cleaned by excluding 5,487 missing/implausible entries; mean age 21.1 years (SD 6.5; range 11–39; n=81,828). Ethics: Users provided informed electronic consent via app terms; protocol approved by WCG IRB. Measures: Physiological symptoms (e.g., cramps, headache, abdominal pain, perineum pain, backache, tender breasts, swelling, fatigue, acne, bloating, disease/injury, nausea, cravings, diarrhea, insomnia), mood-related symptoms (e.g., calm, frisky, depressed, irritated, mood swings, happy, apathetic, guilty, energetic, sad, obsessive thoughts, anxious, confused, very self-critical), and stress. Pain outcome aggregated from pain-related symptoms (e.g., headache, backache, abdominal pain, cramps, perineum pain). Statistical analysis: Primary model was a logit with a binary dependent variable indicating whether a pain symptom was logged on a given day (1) vs not (0). Independent variables included binary indicators for stress and 26 other symptoms (controls), time fixed effects (daily dummies), and interactions between stress and time fixed effects to allow time-varying stress effects. Heteroskedasticity-robust (HC1) standard errors were used. The marginal effect of stress on pain at/after 2022-02-24 was interpreted as reflecting acute stress strength. Specificity analyses replaced stress with other symptoms in the interaction term (26 alternative models) and, separately, replaced pain with other symptoms as dependent variables to examine how stress affected them (26 symptoms overall; subset of 10 physiological symptoms). Geographical analysis: Assessed regional specificity by estimating, for 38 European countries neighboring/surrounding Ukraine, the Z-statistic for the stress × time interaction on 2022-02-24, with Ukraine hypothesized to show the strongest effect. Operational considerations: Restricting to users logging at least two symptoms per day ensured users had opportunity to log pain but chose otherwise, reducing collider bias. Analyses also restricted to users active both before and after the invasion to mitigate attrition bias.

- Symptom prevalence shifts (30 days before vs after 2022-02-24): Stress logging jumped from 3% (Feb 23) to 16% (Feb 24). Grouped pain prevalence (headache, backache, abdominal pain, cramps, perineum pain) decreased from 16% to 12% immediately after war onset. Stress prevalence had been stable for over a year pre-war. - Time-varying marginal effects of stress on pain (logit model): The day before the war (Feb 23), marginal effect of stress on pain was −0.161 (SE 0.077). On the day the war started (Feb 24), it decreased to −0.572 (SE 0.053) and remained low and statistically significant for 15 consecutive days until March 11, 2022, gradually returning toward pre-war levels in subsequent weeks. - Specificity analyses: Substituting stress with other symptoms showed that, while some symptoms (e.g., acne) exhibited temporal changes around war onset, stress had the most profound and sustained effect in reducing likelihood of logging pain. When using other symptoms as outcomes, stress showed attenuation for some mood-related symptoms (e.g., mood swings, feeling calm), but pain was the most affected physiological symptom and the second most affected overall after mood swings. - Geographic patterns: The strongest country-level effect (Z-statistic for stress × time interaction on 2022-02-24) was in Ukraine (−5.46). Significant effects were also observed in nearby countries including Poland (−3.01), Czechia (−2.37), and Moldova (−2.13), with attenuation by distance from Ukraine. Overall, higher self-reported stress was associated with a lower likelihood of reporting pain immediately following the invasion, consistent with stress-induced analgesia, with effects diminishing over time.

Findings support the hypothesized real-world manifestation of stress-induced analgesia during an acute extreme stressor: users who logged stress were significantly less likely to log pain symptoms after the invasion’s onset. The effect was temporally bound—strongest immediately after 2022-02-24 and returning toward baseline within weeks—aligning with the view that SIA is an acute, adaptive response prioritizing immediate survival. The effect displayed specificity: stress more strongly reduced pain than other physiological symptoms, and no other symptom mirrored the magnitude of stress’s impact on pain; mood-related symptoms also shifted as expected under stress. Geographic analyses localized the strongest effect to Ukraine, with spillover to neighboring countries, consistent with differential exposure to conflict-related stress. While acute SIA may be adaptive, the discussion underscores well-documented long-term adverse mental and physical health consequences of war-related stress (e.g., PTSD, chronic pain comorbidity), highlighting a need for trauma-informed, multi-level interventions and healthcare system preparedness. The observed time course complements laboratory constraints and motivates further real-world psychophysiology research. Integrating concurrent autonomic and neural measures via wearable technology could elucidate mechanisms (e.g., cardiovascular gating of nociception, amygdala/insula/pons involvement) underpinning SIA in ecological settings.

Using large-scale app-based symptom tracking, the study documents a real-life instance of stress-induced analgesia during the onset of the Russia-Ukraine war: Ukrainian users reporting stress were less likely to report pain, with effects strongest immediately post-invasion and diminishing over subsequent weeks. This contributes ecological evidence to a phenomenon previously characterized mainly in laboratories, demonstrating specificity to pain and geographic localization to the conflict epicenter. Future research should integrate wearable-derived autonomic and neural measures to clarify mechanisms and trajectories of SIA, examine whether acute SIA magnitude predicts later comorbidity (e.g., chronic pain and PTSD), and extend analyses across populations and stressor types to inform targeted, trauma-informed interventions.

- Self-tracked, optional symptom logging may not perfectly reflect true psychological/physiological states and can be influenced by users’ tracking behaviors and perceived symptom importance. - Location was assessed at data download’s initial timepoint and not dynamically across the conflict, potentially misclassifying users who moved. - Potential sample selection issues: attrition bias (fewer users logging after war onset) and collider bias (users preferentially logging stress). The analytic design mitigated but did not eliminate these risks by focusing on users active both before and after the invasion and requiring at least two symptoms logged per day. - Age data were optional and potentially inaccurate; implausible/missing entries were excluded. - App-based measures lack symptom intensity and rely on binary daily logging, limiting granularity.