Use of No-Suicide Contracts by Psychiatrists in Minnesota

Confronted with the necessity of assessing a patient’s risk of a serious suicide attempt within the next 24–48 hours, the clinician is forced to make a decision that cannot be based on accurate prediction (1). The known risk factors, although important in the assessment process, fall short of a mandate regarding which action to take in each situation (2). Action must be taken, however, to prevent suicide.

The no-suicide contract, in which a patient formally agrees to inform a relative or health care provider of suicidal intent rather than acting on the intent, is widely recommended as an evaluative and therapeutic tool. A MEDLINE search produced few references, consisting primarily of opinions, recommendations, and cautionary notes. In the psychiatric literature, no-suicide contracts receive brief mention as one component within a large list of recommended management strategies. The thoughtful review by Stanford et al. (3) of the history, use, and misuse of no-suicide contracts pointed out that these contracts have no empirically demonstrated effectiveness, that no medical-legal protection is conferred, and that no standardized form or technique exists for entering into such a contract with a suicidal patient.

The no-suicide contract is often advocated as a method of building a therapeutic alliance, although it may actually be efficacious only when a therapeutic alliance is already in place. Attempts to pressure a suicidal patient whom one barely knows into making a no-suicide contract could be interpreted by the patient as a clinical retreat into legalisms rather than an expression of genuine concern.

Furthermore, by agreeing to such a contract, a truly suicidal patient may lull the clinician into decreasing the level of safety measures, and a refusal to agree to a contract by a truly nonsuicidal patient might lead the clinician into instituting excessive safety measures. Despite these problems, the no-suicide contract has gained widespread acceptance as a useful transaction to initiate with a suicidal patient, and this risk management strategy is almost obligatory. Given the absence of evidence indicating whether it is helpful or harmful, this study examines the standard of care in a psychiatric community in regard to use of no-suicide contracts.

Method

A cover letter and enclosed postcard response card listing nine questions were mailed to 514 psychiatrists practicing in Minnesota. The names of the psychiatrists were drawn from the membership list of the Minnesota Psychiatric Society and from telephone directories and other sources. The postcard contained three demographic questions and six questions regarding the use of no-suicide contracts. Replies were anonymous. Data were analyzed by using chi-square tests for differences in proportions of variables as well as logistic regression models to examine the multivariate relationships among and relative importance of potential predictors of use of no-suicide contracts.

Results

There were 267 responses to the 514 questionnaires, a response rate of 52%. One hundred ninety-three (72%) of the respondents were men and 73 (27%) were women; one person did not indicate gender. One hundred sixty-eight (63%) of the respondents had an outpatient practice only, 43 (16%) an inpatient practice only, 43 (16%) a combined inpatient and outpatient practice, and 13 (5%) did not provide practice information. Fifty-one (19%) of the respondents had been out of residency for 0–5 years, 56 (21%) for 6–10 years, 69 (26%) for 11–20 years, and 83 (31%) for more than 20 years; eight (3%) did not provide this information. Forty-five (42%) of the 107 psychiatrists who had been out of residency for 10 years or less, compared with 24 (16%) of the 152 psychiatrists who had been out of residency for more than 11 years, were women (Yates-corrected χ²=21.61, df=1, p<0.001; Fisher’s exact p=0.0001).

No-suicide contracts were used by 152 (57%) of the respondents. Of those who used no-suicide contracts, 94 (62%) used verbal contracts only and 58 (38%) used written and verbal contracts. One hundred seventeen (77%) reported that they used no-suicide contracts because they thought it was helpful; 35 (23%) used such contracts although they did not think it helpful. Of those who used no-suicide contracts, 44 (29%) reported that they used such contracts with all patients whom they assessed as at risk for suicide, but 108 (71%) did so only with selected patients.

There was an inverse relationship between the use of no-suicide contracts and length of time in psychiatric practice: 39 (76%) of the psychiatrists who had been out of residency 5 years or less used no-suicide contracts, compared with 34 (61%) of those who had been out of residence for 6–10 years and 76 (50%) of those with 11 or more years of psychiatric practice. A logistic regression model in which use versus nonuse of the no-suicide contract was the response variable and sex, number of years out of residency (0–5 versus 6 or more years), and type of practice were the predictor variables indicated that number of years in practice was the only significant predictor of use of no-suicide contracts (odds ratio=2.28, 95% confidence interval [CI]=1.07–4.83, p=0.03).

Sixty-two (41%) of the psychiatrists who used no-suicide contracts reported that they had patients who had completed suicide or made serious attempts after entering into the contract. Fifty-two (25%) of the psychiatrists with 6 or more years of experience, compared with nine (18%) of the psychiatrists with 0–5 years of practice experience, reported that they had a patient make a serious suicide attempt after signing a no-suicide contract. A logistic regression that included sex, number of years out of residency training (0–5 years versus 6 or more years), and type of practice again revealed that years out of residency training was the only significant predictor of whether a psychiatrist reported having a patient make a suicide attempt after signing a no-suicide contract (odds ratio=0.31, 95% CI=0.13–0.75, p=0.009).

One limitation in this study is the potential sampling bias in that the data come only from responders. The decision to maintain the anonymity of responders, based on the assumption that this would improve the percentage of returns, precluded the possibility of a second mailing to those who failed to return a postcard or comparing the characteristics of responders with those of nonresponders. The sex distribution of the responders, however, was almost identical to that of the membership of the Minnesota Psychiatric Society (Yates-corrected χ²=0.003, df=1, p=0.96). A second limitation is that we do not have patient suicide data for psychiatrists who did not use no-suicide contracts.

Discussion

A survey of the use of no-suicide contracts among 267 Minnesota psychiatrists revealed that they were equally divided between those who did and those who did not use no-suicide contracts in their clinical work. This finding indicates that there is no community consensus or standard of care in regard to use of no-suicide contracts as an intervention with suicidal patients. Although experienced psychiatrists used no-suicide contracts less frequently, a higher percentage of experienced psychiatrists reported serious or completed suicide attempts in patients who had entered into no-suicide contracts with them. It is most likely that this finding reflects the greater number of years at risk for having suicide attempters in one’s practice. Nevertheless, the finding that 41% of the psychiatrists who used no-suicide contracts in their work with suicidal patients had patients who committed suicide or made serious attempts after signing a no-suicide contract underscores the tenuousness of counting on the contract as an effective suicide prevention tool. This finding does not imply that discussing a commitment to contact the physician or others as an alternative to suicide may not be of therapeutic value to a clinician and suicidal patient, but it is clear that the no-suicide contract has, at best, limited efficacy in general. There is a need for a randomized clinical trial to determine the utility (or lack thereof) of this suicide management tool.

Received Jan. 11, 1999; revisions received April 26 and June 7, 1999, and March 29, 2000; accepted April 5, 2000. From the Department of Psychiatry, University of Minnesota Medical School. Address reprint requests to Dr. Kroll, University of Minnesota Medical School, F282/2A West, 2450 Riverside Ave., Minneapolis, MN 55454-1495; [email protected] (e-mail).The author thanks Jennifer Carey and Mary DeWitt for logistical support and Joel Hechtner, Ph.D., for statistical analyses.