> ## Documentation Index > Fetch the complete documentation index at: https://docs.getrivet.ca/llms.txt > Use this file to discover all available pages before exploring further. # Minimal clinically important difference, by measure > What counts as real improvement rather than noise on PHQ-9, GAD-7, the Insomnia Severity Index and the rest — the point-change thresholds, per instrument. A raw score change from week to week tells you direction. A clinically meaningful change tells you whether the change is large enough to act on. What follows is a quick reference for the change thresholds on the most-used measures in the Rivet library. Rivet shows the raw score and the severity band; the interpretation of whether a change is clinically meaningful is yours. ## What "clinically meaningful change" means A change in score is clinically meaningful when it's larger than the measurement noise of the instrument — i.e. larger than what you'd expect from session-to-session variation alone. Different research traditions operationalize this differently: * **Reliable Change Index (RCI)** — the change a client would need to show for you to be statistically confident the change isn't just measurement noise (Jacobson & Truax 1991). * **Minimal Clinically Important Difference (MCID)** — the smallest change a client or clinician would notice as actually meaningful. * **Severity band crossing** — a drop from "moderately severe" to "mild" is meaningful regardless of the absolute change. For most published scales there's a single working number — that's the number below. ## Per-measure thresholds ### PHQ-9 (depression) * **5-point change** is the working threshold for clinically significant improvement (Kroenke 2010 follow-up). * **Severity band crossings** matter independently: moving from "moderately severe" (15–19) to "moderate" (10–14) is meaningful even if the absolute drop is smaller. * **Score ≥ 10** is the screening threshold for major depression. Citation: Kroenke, K., Spitzer, R. L., & Williams, J. B. W. (2001). *J Gen Intern Med,* 16(9): 606–613. ### GAD-7 (generalized anxiety) * **4-point change** is the working threshold for meaningful improvement. * **Score ≥ 10** is the threshold for probable generalized anxiety disorder and the trigger for further assessment. Citation: Spitzer, R. L., Kroenke, K., Williams, J. B. W., & Löwe, B. (2006). *Archives of Internal Medicine,* 166(10): 1092–1097. ### PCL-5 (PTSD symptoms) * **5- to 10-point change** is the working threshold for reliable change (Bovin 2016). The narrower number reflects research-grade reliable change; the wider one is the clinically meaningful threshold most commonly cited. * **Score ≥ 31–33** is the cutoff for probable PTSD. * Cluster-specific change matters: an overall drop driven entirely by Cluster D (cognition / mood) without movement in Cluster B (intrusion) suggests something different than a uniform reduction. Citation: Bovin, M. J., et al. (2016). *Psychological Assessment,* 28(11): 1379–1391. ### K10 (psychological distress) * **5-point change** is the practical threshold; K10's bands are 5-point wide (low / mild / moderate / severe) so a band crossing and an MCID roughly coincide. Citation: Kessler, R. C., et al. (2002). *Psychological Medicine,* 32(6): 959–976. ### DASS-21 (depression, anxiety, stress) Each subscale has its own bands and thresholds (the 21-item version sums each subscale × 2 to align with the 42-item norms — Rivet handles the multiplier automatically). A useful working threshold: * **Depression subscale:** \~9-point change at the multiplied score * **Anxiety subscale:** \~7-point change * **Stress subscale:** \~9-point change The above are rough working numbers commonly cited in DASS application literature; the original Lovibond & Lovibond (1995) manual focuses on severity bands rather than a single MCID. ### PSS-10 (perceived stress) * **3- to 5-point change** is a commonly-cited working threshold. * No formal clinical cutoff — descriptive measure with bands (low / moderate / high stress). Citation: Cohen, S., Kamarck, T., & Mermelstein, R. (1983). *Journal of Health and Social Behavior,* 24(4): 385–396. ### ISI (insomnia) * **6-point change** is the published meaningful improvement threshold for CBT-I outcomes. * **Score ≥ 15** is the cutoff for clinical insomnia. Citation: Bastien, C. H., Vallières, A., & Morin, C. M. (2001). *Sleep Medicine,* 2(4): 297–307. ### AUDIT (alcohol use) * **Score ≥ 8** is the WHO cutoff for hazardous drinking (≥ 7 for women in some adaptations). * **Score ≥ 16** indicates harmful drinking. * **Score ≥ 20** suggests possible alcohol dependence and warrants further evaluation. * Change-from-baseline matters more than a single MCID for AUDIT; the WHO scoring rubric focuses on bands. Citation: Babor, T. F., Higgins-Biddle, J. C., Saunders, J. B., & Monteiro, M. G. (2001). *AUDIT: The Alcohol Use Disorders Identification Test* (2nd ed.). WHO. ### WHO-5 (well-being) * **10-point change** on the 0–100 percentage scale is the working threshold (the 0–25 raw score is multiplied by 4 to produce the percentage; Rivet applies the multiplier automatically). * **≤ 28%** indicates a need for further depression screening. Citation: Topp, C. W., Østergaard, S. D., Søndergaard, S., & Bech, P. (2015). *Psychotherapy and Psychosomatics,* 84(3): 167–176. ## What Rivet shows; what's yours Rivet shows you: * The raw subscale score * The multiplied score where the algorithm requires it (DASS-21 × 2, WHO-5 × 4) * The matching severity band * Whether any cutoff fired (e.g. "probable PTSD", "clinical insomnia") Rivet does not: * Compute a Reliable Change Index from prior sessions (no longitudinal data; that's your chart's job) * Tell you whether a specific change is "clinically significant" — that's a judgment that takes the trajectory, the client's life context, and treatment phase into account * Diagnose anything — cutoffs surface a screening signal, not a diagnosis The numeric thresholds above are practitioner shorthand. The interpretation of what to do about a score change is the practice of psychotherapy, which remains yours. ## Related articles What Rivet does — and doesn't — when a high-risk item fires. How session-to-session scores land in your chart today.