On the Influence of Apologies on the Likelihood of Lawsuits in Cases of Perceived Medical Negligence: Analysis of Archival and Experimental Data (Preprint)

<sec> <title>BACKGROUND</title> Disappointing medical care (DMC) encompasses cases of medical failures, malpractice, or errors. Literature suggests that individuals’ perceptions of harm resulting from medical procedures influence their intention to seek legal recourse and that apologies may mitigate the inclination for legal action. </sec> <sec> <title>OBJECTIVE</title> Here, we aim to scrutinize and potentially challenge this prevailing notion. </sec> <sec> <title>METHODS</title> We conducted 4 studies using a dataset of social media posts detailing possible DMC incidents to which we linked a proxy for legal action, specifically, future posts related to legal action. Study 1 used a machine learning model to predict a proxy for an intent to file a lawsuit based on the content of 3815 posts. Two preregistered crowdsourcing studies (N=1115) assessed the impact of different apology types on intention-to-sue in 10 diverse medical scenarios with 4 apology conditions. Finally, study 4 aimed to test whether the predictors of legal intent identified in the crowdsourced studies and modeled as a function of case attributes can generalize to an actual subset of 165 Reddit (Reddit, Inc) posts. </sec> <sec> <title>RESULTS</title> Results show that apologies are rarely mentioned in descriptions of DMCs and that the descriptions of DMCs predict the proxy of legal action (area under the curve [AUC]=0.78). Crowdsourcing studies reinforce these findings: people agree on which cases are worthy of legal action (interclass agreement: 0.96 in study 1 and 0.71 in study 2), and our results demonstrate that physical and emotional damage are independently the strongest predictors of intention to file a lawsuit, together accounting for 43%-48% of model variance. Apologies are not statistically significant predictors for the intent to file a lawsuit (&lt;i&gt;P&lt;/i&gt;&amp;gt;.05), both separately and in interaction with physical and emotional damage. A model developed in the crowdsourcing study and based on the attributes of cases, when applied to large-scale data, reached an AUC of 0.67-0.70. However, these attributes did not capture the entire range of behaviors, as a model that was based on the words in the cases reached a significantly higher AUC of 0.79. </sec> <sec> <title>CONCLUSIONS</title> Text-based apologies appear to exert little influence on individuals’ intention to file a lawsuit in DMC cases, while physical and emotional damage are the primary motivators. This suggests that medical providers aiming to mitigate legal risks must explore alternative interventions beyond apologies. </sec>