Add To Cart

Prevention of Medical Errors for Social Workers, Counselors & MFTs
Section 2
Avoiding Errors in Clinical Prediction: Part II

Question 2 | Test | Table of Contents

Sources of Error
What causes mistakes in clinical prediction? Although the general public tends to attribute errors to practitioner incompetence, it appears that errors stem more from the social and psychological constraints imposed on practitioners than from practitioner ignorance or carelessness (Bosk, 1989).

Social Constraints  

The social context in which practitioners operate can have a profound effect on the practitioner's predictive capacities. The following social constraints are typically cited: lack of necessary information; administrative pressures (for example, deadlines, excessive paperwork)) high caseloads; fear of legal reprisal; and lack of necessary training, experience, or both (Fishman, 1991).

In addition, the confusion and uncertainty that pervade clinical work often undermine the predictive capacities of clinicians (Davis, 1972). Clients' problems are generally complex and involve multiple actors, interests, and issues; clear-cut facts or answers are often difficult to find; causes are generally elusive; and clients' motivations are frequently impossible to understand or explain. This kind of informational morass often incapacitates the predictive capabilities of even the most skilled practitioner.

Reducing Prediction Errors
Current research on clinical prediction has identified some basic steps practitioners can take to predict more accurately. Though many of these suggestions will seem elementary to social work practitioners, it is important to discuss them because in the rush of events they can easily be neglected by even the most experienced clinician.

Look for Historical Patterns  

Assuming that some accurate history is known, it is important to be aware of the behavioral patterns within that history when attempting to make predictions (Gruppen, Wolf, & Billi, 1991). For example, a history of assault is often a strong indicator of future violent behavior (Yesavage, 1983).

Likewise, Hirschfeld and Davidson (1988) observed that a history of suicide attempts is often highly relevant in treatment planning because "a history of suicide attempts significantly increases the likeliness of subsequent suicide" (p. 632). Lastly, Johnson and L'Esperance (1984) demonstrated that, among other factors, a history of abuse during childhood was a significant predictor in forecasting a recurrence of child abuse by the parents they studied.

Know the Demographics

Specific demographic factors such as age, gender, race, diagnosis, and marital status have repeatedly been found to be statistically correlated with behavioral outcomes such as assaultiveness (Yesavage, 1983), suicide (Hirschfeld & Davidson, 1988), child abuse (Johnson & L'Esperance, 1984), recidivism in foster care (Turner, 1984), and discharges from psychiatric hospitals against medical advice (Louks, Mason, & Backus, 1989).

Such data, which can function as useful indicators or predictors, have been compiled in actuarial or statistical tables for use by practitioners in some clinical judgment situations (Monahan, 1984). Indeed, there is evidence that practitioners often use an informal schema involving such predictors when making decisions about hospitalization for psychiatric treatment (Segal, Watson, & Nelson, 1985). Formal attempts to use demographic data have emerged in efforts to develop criteria for identifying high-risk hospital patients in discharge planning (Berkman, Rehr, & Rosenberg, 1980) and in referral for social work services (Becker & Becker, 1986).

Know the Current Stressors

Considerable research has accumulated since the 1960s on the impact of certain stressful life events on individual adjustment and behavior (Holmes & Rahe, 1967; Lin, Simeone, Ensel, & Kuo, 1979). Lately this information has been used to improve prediction by practitioners required to identify and provide services to high risk populations (Becker & Becker, 1986).

 In addition, current research is beginning to point more specifically to the relationship of certain stressors to specific behavioral outcomes such as violence (Wilson & Kneisl, 1988) and suicide (Hirschfeld & Davidson, 1988). Because the information generated by this research enlightens practitioners about the individual's social situation, it promises to become an increasingly important factor in all clinical decision making (Monahan, 1981).

Be Aware of Significant Cues

A whole area of research has recently opened up regarding "cue utilization" by practitioners in clinical decision making (Engel, Wigton, LaDuca, & Blacklow, 1990). Cues are defined as "clinical pieces of information" useful in "making judgments . . . about the clinical state" (Engel et al., 1990, p.64). Most of this work is based on social judgment theory derived from the work of Brunswick (1952). Theorists are quick to point out that cues differ in meaning and value depending on the social context in which they occur. Cue utilization in making predictions is therefore highly contextual.

Make Short-Term Predictions

Several studies have demonstrated that accuracy is greatly increased by shortening the time span covered by predictive judgments (Monahan, 1984). Though the reasons for this are intuitively obvious (for example, I can predict with reasonable certainty where I will be five minutes from now, but I can predict with much less certainty where I will be one year from now), this procedure tends to be ignored in the push to make decisive forecasts that will justify present courses of action. This tendency often undercuts attempts to achieve greater accuracy in prediction.

Although a number of authors (Cocozza & Steadman, 1978; Kahneman et al., 1982) have tried to document the prognostic incapacity of clinicians, the courts and the general public increasingly expect practitioners, particularly in fields such as mental health and child welfare, to officially and accurately predict individual behavior. These expectations often become entrenched in administrative law or agency regulations, which add further stress to practitioner involvement in case situations. In addition, these expectations carry with them an ever-growing risk of tort liability (Besharov, 1985).

Given these pressures, it becomes even more necessary to practice skillfully what some have called "defensive social work" (Besharov, 1985, p. 136)--that is, the ability to intervene prudently with minimal risk of harm to the client and minimal future legal consequences for the practitioner. When making predictions, defensive practitioners would use the steps advocated earlier and would realize that the results of even the best attempts at prediction are often no better than chance (Cocozza & Steadman, 1978).

Forecasting, by its nature, is at most an uncertain art. When used with proper care, however, it can provide clients, colleagues, and the community with at least general guidelines on which to base future decisions and interventions.

The suggestions discussed here, although by no means original or exhaustive, illustrate some of the more frequently recommended procedures for eliminating errors that often occur in forecasting. Although their use cannot guarantee accuracy, it can at least help practitioners alleviate some of the more common errors in prediction and improve reliability.

--Murdach, A. D. (1994). Avoiding Errors in Clinical Prediction. Social Work, 39(4), 381.

Personal Reflection Exercise #2
The preceding section contained information about errors in clinical prediction.  Write case study example regarding how you might use the content of this section in your practice.

Peer-Reviewed Journal Article References:
Garb, H. N., Wood, J. M., & Baker, M. (2018). The Lackland Behavioral Questionnaire: The use of biographical data and statistical prediction rules for public safety screening. Psychological Assessment, 30(8), 1039–1048.

Garb, H. N., & Wood, J. M. (2019). Methodological advances in statistical prediction. Psychological Assessment, 31(12), 1456–1466.

Karon, B. P. (2000). The clinical interpretation of the Thematic Apperception Test, Rorschach, and other clinical data: A reexamination of statistical versus clinical prediction. Professional Psychology: Research and Practice, 31(2), 230–233.

Nahum, L., Barcellona-Lehmann, S., Morand, S., Sander, D., & Schnider, A. (2012). Intrinsic emotional relevance of outcomes and prediction error: Their influence on early processing of subsequent stimulus during reversal learning. Journal of Psychophysiology, 26(1), 42–50.

Ruscio, J. (2000). The role of complex thought in clinical prediction: Social accountability and the need for cognition. Journal of Consulting and Clinical Psychology, 68(1), 145–154.

Siddaway, A. P., Quinlivan, L., Kapur, N., O'Connor, R. C., & de Beurs, D. (2020). Cautions, concerns, and future directions for using machine learning in relation to mental health problems and clinical and forensic risks: A brief comment on “Model complexity improves the prediction of nonsuicidal self-injury” (Fox et al., 2019). Journal of Consulting and Clinical Psychology, 88(4), 384–387.

What are five ways to reducing prediction errors?
To select and enter your answer go to Test.

Section 3
Table of Contents