Visualizing Reliable Change in Psychotherapy

Reliable Change in Psychotherapy Over Sessions

What This Chart Shows

The chart above presents a visualization of reliable change in psychotherapy over the course of treatment. Rather than showing the traditional Kaplan-Meier survival curve (where higher values indicate patients who have not yet experienced reliable change), we’ve inverted the data to show the cumulative proportion of patients who have experienced reliable change over time. This offers several key insights:

  • Early Progress: Only a small percentage of patients (approximately 5-10%) experience reliable change within the first 3 sessions
  • Steady Growth Phase: Between sessions 5-10, we see a consistent upward trajectory, with approximately 50% of patients experiencing reliable change by session 10
  • Intermediate Plateau: There’s a brief plateau around sessions 10-11, which may represent a consolidation phase in therapy
  • Second Growth Phase: Between sessions 13-16, the percentage of patients achieving reliable change accelerates again
  • Late-Stage Gains: By sessions 20-21, over 80% of patients have experienced reliable change

This pattern reflects the dose-response relationship in psychotherapy: initial sessions provide foundation-building, middle sessions address core issues, and later sessions consolidate gains. The visualization helps clinicians understand expected outcomes at different points in therapy and can inform treatment planning and policy decisions about optimal treatment length.

Technical Details: How We Created This Visualization

This reliable change curve was converted from a Kaplan-Meier (KM) survival curve originally published in:

Harnett, Paul, Analise O’Donovan, and Michael J. Lambert. “The dose response relationship in psychotherapy: Implications for social policy.” Clinical Psychologist 14.2 (2010): 39-44.

The original KM curve showed the “survival” of patients who had not yet experienced reliable change over the course of therapy sessions (where higher values indicate patients without reliable change). To create a more intuitive visualization, we transformed this into a “reliable change achieved” curve using the following process:

  1. Data Extraction: We used WebPlotDigitizer to extract the exact data points from Figure 1 in the Harnett et al. paper. This tool allowed us to capture the precise coordinates of each point on the original KM curve.

  2. Transformation: We applied the mathematical transformation: Reliable Change = 1 - Survival to each data point, effectively inverting the curve to show the cumulative proportion of patients who have achieved reliable change rather than those who haven’t.

  3. Data Cleaning: Some points required minor adjustments to ensure the trajectory was monotonically increasing (as the proportion of patients achieving reliable change shouldn’t decrease over time in an averaged dataset).

  4. Visualization: Using SVG, we created a responsive chart that clearly shows the reliable change trajectory, with grid lines to help clinicians gauge percentages at each session number.

This transformation makes the data more accessible and meaningful. While survival curves are standard in research literature, they can be counterintuitive when communicating about positive therapeutic outcomes. Our reliable change curve presents the same data in a format that emphasizes treatment effectiveness, while remaining scientifically accurate.

Literature

This visualization is based on data from the following publication:

Harnett, P., O’Donovan, A., & Lambert, M. J. (2010). The dose response relationship in psychotherapy: Implications for social policy. Clinical Psychologist, 14(2), 39-44. https://doi.org/10.1080/13284207.2010.500309

The original Kaplan-Meier survival curve appears in Figure 1 of the paper. Our transformation of this data into a reliable change curve is intended for educational purposes and falls under fair use. The authors define reliable change using Jacobson and Truax’s (1991) criteria, representing a statistically significant improvement in standardized outcome measures. We encourage readers interested in the complete research and its methodological details to consult the original publication.