Essays on the drivers of physician and patient behaviour in the context of medical incidents. Empirical analysis of defensive medicine and defensive communication.

Abstract

With the introduction of the Patients’ Rights Act twenty years ago, the Belgian government attempted to encourage physicians to administer high-quality care and to provide extensive information to patients, (1) ex ante about their clinical situation, treatment considerations and potential accompanying complications and (2) ex post about the care administered and all procedures that may have or have damaged patients’ condition (i.e., medical incidents). Physicians’ defensive behaviours out of fear of medical malpractice claims, such as conducting more tests and procedures than strictly medically necessary and avoiding high-risk patients and procedures (i.e., defensive medicine), or communicating defensively about medical incidents (i.e., defensive communication), may violate these patients’ rights. Empirically addressing the drivers and impacts of these behaviours, this dissertation contributes to a behavioural economics perspective on how to optimise levels of care, medical malpractice law, and the physicianpatient relationship. Literature on defensive medicine mainly focuses on the US, where malpractice claims are more common and the health care sector is organised quite differently than, for example, in European countries. Surveying 90 physician-specialists, we show, however, that the medical liability system in Belgium also drives physicians to perform more acts than strictly medically needed and avoid certain patients and acts predominantly to reduce their personal malpractice risk. In particular, 13 per cent of all tests and procedures are performed due to malpractice concerns. Our results indicate that the so-called assurance behaviour (i.e., administering more care) is more common than avoidance behaviour (i.e., avoiding high-risk patients and procedures). Given that physicians’ defensive behaviour is also prevalent in Belgium, the question raises how this behaviour impacts patients’ feelings and behaviours. Using a virtual reality experiment, we examine how physicians’ open versus defensive communication about a possible medical neglect impacts how patients feel and intend to behave. More specifically, we show 140 economics, medicine and physiotherapy students three videos of hypothetical medical incident conversations, vary the openness of physicians’ verbal communication, and ask the participants about their feelings and behavioural intentions. Keeping constant all other factors besides the treatment, e.g., physicians’ non-verbal behaviour, patients’ communication, and clinical condition, allows us to uncover the causal effects of physicians’ verbal communication style on patients. Comparing the results of the ones who are randomly assigned to the open disclosure videos and those to the defensive counterparts demonstrates that open verbal communication about a medical incident leads to higher patient intentions to take further steps against the treating physician (e.g., sending formal complaints to the hospital and contacting a lawyer to discuss options) and higher feelings of blame. At the same time, respondents watching the open communication videos rate the physician’s communication skills and general impression better compared to those who are confronted with a defensive physician. We find no significant differences between the groups regarding the trust in the physician and his competence, the perceived incident severity, and the likelihood of changing physician and filing suit. Next, we conduct another video-experiment, but this time with practising physicians as respondents to examine how patients’ behaviour, in turn, drives physicians’ treatment behaviour. In particular, we randomly assign 85 gynaecologists/obstetricians and orthopaedics to four hypothetical consultations with patients with either a critical attitude (i.e., getting ahead of the facts, showing distrust) or a non-critical attitude (i.e., displaying more neutral questions and expressions). After each video, we ask the physicians what treatment they would prescribe and how likely they believe the patient would take further steps in case of a medical incident (i.e., measure of perceived liability risk). Given that we keep constant patients’ clinical condition, preferences, and non-verbal communication and what the physician says, we are able to assess to what extent patients’ critical attitude (the only manipulation in our design) drives physicians’ perceived liability risk and consequent defensive behaviour. Our results show that physicians prescribe 17 percentage points more defensive treatment when experiencing a high liability risk (e.g., unnecessary C-sections, scans, surgeries) among patients with a different critical attitude. Besides patients’ critical attitude, one might expect that patients’ expertise level may be associated with different (defensive) treatment styles. Using detailed patient-level data from Statistics Netherlands, we assess whether expert patients (i.e., patients who enjoyed relevant medical training) have a different chance than semi-expert patients of receiving one of the most discussed defensive practices, namely a C-section. As common in existing literature, we consider expert patients as physicians and midwives. We define semi-expert patients as nurses and physiotherapists because of their experience in health care, though relatively low expertise level regarding gynaecology/obstetrics compared to physicians and midwives. Our results show that expert patients receive significantly fewer Csections (3.1 percentage points) than semi-expert patients. Using a difference-indifferences approach in which semi-expert patients constitute the treatment group and expert patients the control group, we demonstrate that this treatment gap diminishes after a Dutch law reform with the aim of (1) strengthening patients’ position in health care processes and (2) increasing quality of care through more openness about clinical decisions and medical incidents. In particular, semi-expert patients receive eight percentage points fewer C-sections after the reform compared to what happens to expert patients. Our preferred explanation for this finding is that the reform causes semi-expert patients to get more informed, lowering the level of information asymmetry among this group of patients so that they are less susceptible to physician-induced demand (i.e., agency discrimination). However, we are not able to exclude statistical discrimination in obstetrics. That is, expert patients may be better at communicating symptoms and expressing their preferences. Overall, this dissertation points out that the medical liability system induces a vicious circle in the physician-patient relationship: confusing signals about physicians’ true liability risk drive physicians to practise defensive medicine and defensive communication, which in turn influence patients’ feelings and behaviours and consequently physicians’ perceived liability risk and defensive behaviour again. This leads to an inefficient use of public resources and concerns about patient safety and rights. Policy makers and future research should focus more on educating physicians in informing patients ex ante and ex post medical treatment and treating patients with a different background with respect to patients’ rights. Furthermore, it is important to conduct more in-depth research on how to optimise levels of care and medical malpractice law.