APA Recommendations to the USPSTF on Suicide Risk for Adolescents and Older Adults

May 20, 2013

Virginia Moyer, MD, MPH
Chair, U.S. Preventive Services Task Force
540 Gaither Road
Rockville, MD 20850

Dear Dr. Moyer:

The American Psychological Association would like to thank the United States Preventive Services Task Force (USPSTF) Topic Group for Stakeholders (TOPS) for revisiting the 2004 recommendation on screening for suicide risk in adolescents, adults, and older adults in primary care settings.  We are providing comments related to two groups - adolescents and older adults.

Our first set of comments focuses on conclusions concerning screening of older adults for risk of suicide. The I Statement (indicating insufficient evidence to assess the balance of benefits and harms) appears to be warranted, in light of the state of the literature on screening for suicide risk among older adults in primary care. Furthermore, the Suggestions for Practice provide clear and appropriate guidance regarding assessing for suicide risk in patients who have a psychiatric disorder or other risk factors. Nonetheless, there are several important ways in which the recommendation statement could be improved, to include adding context regarding the base rate of suicidal behavior, including additional studies of older adults, and emphasizing the need for systems in place to refer or treat older adults at risk of suicide. These issues are discussed below, along with a review of each of the key questions addressed in the recommendation statement.

An overarching theme in these remarks is the suggestion that the recommendation statement should be worded in a way that does not discourage screening by clinicians. Although there may not (yet) be sufficient evidence to support recommending screening, there is also no evidence of serious adverse events related to screening older adults, and there are reasons to believe that screening for suicidal ideation in primary care may represent a unique opportunity, particularly for older adults. As noted in the USPSTF recommendation statement, many individuals who die by suicide are seen in primary care within a short period before the death (Luoma, Martin, & Pearson, 2002), and this is particularly true for older adults. Moreover, screening instruments can improve detection up to 600% compared to waiting for the patient to mention suicidal thoughts to the physician (Bryan, Corso, Rudd, & Cordero, 2008). Thus, even though it may not be possible to recommend screening, we urge the USPSTF to consider the suggestions below to avoid unduly discouraging clinicians from suicide risk screening in primary care.

A limitation of the recommendation statement is that it does not adequately convey the implications of the fact that suicidal behavior is a low base-rate event, and may thus lead readers to misinterpret findings. Suicide is a statistically rare (albeit catastrophic) event in all age groups (McIntosh & Drapeau, 2012). Furthermore, among older adults, suicide attempts are far less frequent than in other age groups. The ratio of attempts to deaths is estimated at 100-200 to 1 among youth, whereas it is estimated at 4 to 1 among older adults (McIntosh & Drapeau, 2012). These facts have implications for interpretation of evidence regarding detection as well as the benefits of screening, as described in more detail below. Providing context for the readers may aid in interpreting the evidence.

In considering benefits of detection (Key Question 1: whether screening results in improved health outcomes), it should be noted that trials examining the clinical utility of screening with sufficient power to detect death by suicide are virtually infeasible, due to the low base-rate. The only study cited as addressing Key Question 1 had a very short-term follow-up (two weeks) and a relatively small sample, and thus may not have had sufficient power to detect changes even in suicide ideation or attempts. Similarly, very large sample sizes would be needed to examine the effects of screening on suicide attempts in older adults, given the very low rates of this behavior. Readers should be cautioned not to interpret lack of evidence as lack of benefit from screening.

The evidence regarding accuracy of screening instruments in detection of suicide risk (Key Question 2) should also be considered in light of the difficulty of predicting low base-rate events. Readers should be advised that it is not reasonable to expect a high positive predictive value (PPV) for suicide outcomes, or for suicide attempts among older adults. The low PPV is an implication of the low base rate of suicide in general and the low rate of suicide attempts in older adults. It should be noted that each of the studies reviewed in regard to this question had acceptable sensitivity and specificity. In addition, in evaluating detection of suicide risk in older adults, it may be helpful to consider the study conducted by Fujiwasa et al. (2005). Although the screening was conducted in a community sample, these investigators tested a 5-item screening instrument that was designed for a primary care setting and found reasonable results (sensitivity = .70, specificity = .69, PPV = .32, NPV = .93).

With respect to evaluating harms of screening (Key Question 3), it may be helpful to include the study by Cukrowicz, Smith, and Poindexter (2010). These investigators found no increased harms in older adults one month or three months after a suicide risk screening conducted as part of a research protocol. Furthermore, the emphasis in this section may be misleading. Although the recommendation statement correctly indicates that “the possibility of short-term harms from participation in suicide screening itself cannot be dismissed,” it would also be accurate to state that no actual harms (and certainly no serious adverse events) were documented in any of the screening studies that were cited. A danger of overstating the possibility of harms, in absence of evidence of any harms, is that clinicians may be reluctant to ask patients about suicide even when warranted.

Several issues merit additional attention with respect to the recommendation statement regarding effects (or harms) of treatment on suicidal behavior or ideation (Key Questions 4, 5, and 6). Considering the large differences in the epidemiology and risk factors for suicidal ideation and behavior in older adults compared to other age groups, it would seem important to evaluate results separately for older adults (65+) and adults of other ages.

In addition, several important reports regarding interventions with older adults were omitted. It would be helpful to include follow-up results from the PROSPECT study (Alexopoulos et al., 2009), one of the treatment studies in an older adult primary care sample. In addition, a study that was not mentioned but seems relevant is the study by Oyama et al. (2006). Oyama and colleagues screened a community sample of older adults, but the intervention involved follow-up by general practitioners. Another report that may be relevant (although not specific to older adults) is the study by Manber et al. (2011), which showed that cognitive behavioral therapy for insomnia (CBT-I) was associated with reductions in suicidal ideation.

Perhaps the most important deficit in the discussion of interventions, however, is the virtual absence of any mention of integrated psychological services within primary care, although it has been supported by research examining depression (Unützer et al., 2002) as well as suicidal ideation (Alexopoulos et al., 2009; Bruce et al., 2004). In the section on Treatment, the statement indicates that patients can be referred for psychotherapy. It would be helpful to mention that integrating psychological services into the primary care setting may improve outcomes relative to referral to outside practices.

On a related note, in the Useful Resources section, it is suggested that physicians screen for depression "when appropriate systems are in place to ensure adequate diagnosis, treatment, and followup." Although it would clearly be inadvisable to screen for depression (or suicide risk) without a system in place to follow up on detected cases, it is also inadvisable (and should be viewed as unacceptable) to offer primary care services without such a follow-up system in place. It is suggested that the importance of follow-up systems of care be highlighted in this report, to include mention of the value of integrated psychological care within primary care settings.

Finally, the wording of the plain-language version of the recommendation statement appears to be inaccurate in places. In describing the potential benefits and harms, the following statement was made: “Unfortunately, the Task Force found that, for people not already known to be at increased risk, there are not good ways for primary care professionals to identify those who are likely to try to take their own lives.” This statement confuses lack of evidence of accurate screening devices with evidence of lack of accuracy of screening devices. Similarly, saying that “these studies did not show that treatment was very effective in preventing suicide” does not accurately portray the fact that methodological challenges make it difficult to determine whether or not interventions are effective in preventing suicide.

In summary, the USPSTF draft recommendation statement on screening for suicide risk accurately reports the limitations in the research literature pertaining to screening for suicide risk in primary care, but the wording of the report could be improved in order to avoid misleading readers.

Our second set of comments focuses on conclusions concerning screening of adolescents for risk of suicide, primarily on the data reviewed concerning adolescents who may be at risk for suicide.

Before examining the key questions, a few issues concerning the selection of studies should be addressed. First, if the point is to review studies relevant to primary care, it is unclear why studies conducted in specialty mental health populations were included in the review. These are two very different samples of patients, and screening instruments to detect risk amongst these groups differ tremendously. In the primary care population, the focus should be on detecting individuals with potential risk who could then benefit from treatment. In specialty mental health care, and particularly in treatment studies of suicidal individuals, the question is often on the level of risk. Thus, risk associated with inaccurate detection errors varies tremendously between these two groups. And further, even if one is to review the psychiatric treatment data, as was the case, it is unclear why only studies focused on depression, substance misuse, posttraumatic stress disorder, and borderline personality disorder were included. While each of these syndromes may heighten one’s risk for suicide, there are ample data now to show other psychiatric problems, such as anxiety, also increase one’s risk for suicide (2). It is concerning that the inclusion criteria may only perpetuate beliefs in primary care providers (PCPs) that suicide risk is only associated with a small subset of mental health problems. Finally, suicide risk has shown to be elevated in well over 20 different medical conditions (3).

The first key question facing the USPSTF was whether or not screening could detect individuals at risk for suicide and result in improved outcomes. Rightly so, the review concluded that there is presently no data to support improvement in health-related outcomes in adolescents. Research is this area is still emerging, but to date, no firm conclusions have been drawn. In fact, the National Institute of Mental Health recently issued a request for proposals to develop a validated screening tool for the pediatric emergency department that could serve this very purpose. The question, however, is whether or not this is the appropriate question to be asking. Is the expectation that PCPs screen and improve outcomes, or is the more probable behavior for PCPs to screen for suicide risk and have improved linkages with community-based mental health providers who then hold the responsibility for improving outcomes?

The second key question focused on the accuracy of screening detection. Perhaps the most troubling issue with the report is that only four of 89 possible studies were included in the recommendation. Two of these studies included adolescent samples, yet both were of high-risk psychiatric populations, thus limiting the generalizability of the findings. Two studies were inexplicably absent from the review. In a 2010 study, Diamond and colleagues (4) reported on the Behavioral Health Screen in a pediatric primary care population. The screening tool was found to have strong psychometric properties, including a suicide screen with 83% sensitivity and 87% specificity. More recently, Horowitz and colleagues (2012) (5) reported on the Ask Suicide-Screening Questions, a 4-item screening instrument in the pediatric emergency department with 96.9% sensitivity and 87.6% specificity. Negative predictive value was also high in this sample of both psychiatric and medical/surgical patients. Both of these studies suggest that the generalizability of the accuracy of screening for suicide risk in adolescents may extend beyond psychiatric patients.

The third key question resulted in findings of no identified serious adverse effects of screening, although it is interesting to note that the pediatric studies reviewed were all conducted in school settings and not primary care or specialty mental health.

The fourth, fifth, and sixth key questions focused on treatment. An examination of the treatment literature, almost none of which is conducted in primary care, seems to counter the main purpose of the review, that is: “it is critical to know whether primary care-based screening is likely to help reduce suicide in the United States by identifying patients in need of treatment and referring them to appropriate care” (1, pp. 14-15). While it is perhaps understandable to review psychopharmacological studies, as PCPs may engage in this intervention in practice, it is unclear why treatment studies conducted in specialty mental health were reviewed. Very few, if any, PCPs are performing psychotherapy with their patients. Thus, this section of the report simply shows PCPs what is available to their patients outside of the primary care office.

Before dissecting the findings, a few points should be made. First, most psychotherapy treatment studies are conducted in university-based research laboratories. And while these trials often involve a comparison group in the community, increasingly this approach is less likely. Further work is needed to disseminate these therapies into the community, where PCPs are most likely to make referrals. Second, it is widely considered unethical to conduct randomized treatment studies on suicidal individuals with a no-treatment control. In medicine, it may be fairly common for early phase clinical trials to compare the efficacy of a specific medication against an inactive placebo. This simply does not happen in psychotherapy research with this high-risk group, where the presence of suicidal ideation and/or behavior is often a requirement of entering the treatment study. Thus, the studies reviewed by the USPSTF were largely comparing a specific psychotherapy to a non-specific psychotherapy (i.e., usual care) being conducted in the community. This is a critical point in the context of the findings. Throughout the O’Connor, Gaynes, Burda, Soh, and Whitlock (2013) report (1), the authors noted that “most trials reported reduced suicidal ideation in both intervention and control groups” (p. 11), and “both the psychotherapy and usual care groups generally showed substantial improvements” (p. 12). These findings point out the challenge in psychotherapy research with suicidal youth. That is, most youth who enter treatment (whether it be the specialized treatment condition or a community-based control condition) tend to get better. Therefore the conclusion that “there [is] little evidence on the accuracy of screening in adolescence, and what little data are available showed that treatment did not demonstrate a positive effect” (p. 18) is absolutely false.

It is clear that more research is needed to definitively examine the impact of suicide screening on short- and long-term health outcomes. Until that time, a more careful examination of the present data seems to suggest that (a) screening instruments may detect youth who are at risk for suicidal ideation and possibly behavior, (b) there is no harm associated with conducting screening, and (c) screening adolescents in primary care settings may result in the detection and referral of at-risk youth into treatment where, by and large, most studies have shown that youth who engage in treatment tend to get better.

The fear, of course, with the current state of the USPSTF draft recommendation is that PCPs will only hear about the conclusion — that there is insufficient evidence to support screening for suicide risk and treatment may not work — and not have the opportunity, time, or research background to more carefully review the findings. Thus, while the report does not say that screening is harmful and should not be conducted, in the extraordinarily busy and demanding world of primary care medicine, the “insufficient evidence” designation may be enough to deter PCPs from engaging in a simple practice that could detect individuals at high risk for suicide. Screening may have the ability to refer these individuals into treatments early in the course of their pain and suffering and forever change the course of their life trajectory.

References for Older Adults

Alexopoulos, G. S., Reynolds, C. R., III, Bruce, M. L., Katz, I. R., Raue, P. J., Mulsant, B. H., Oslin, D. W., & Ten Have, T. (2009). Reducing suicidal ideation and depression in older primary care patients: 24-month outcomes of the PROSPECT Study. American Journal of Psychiatry, 166, 882-890.

Bruce, M. L., Ten Have, T. R., Reynolds, C. F., III, Katz, I. R., Schulberg, H. C., et al. (2004). Reducing suicidal ideation and depressive symptoms in depressed older primary care patients: A randomized controlled trial. Journal of the American Medical Association, 291, 1081-1091.

Bryan, C. J., Corso, K. A., Rudd, M. D., & Cordero, L. (2008). Improving identification of suicidal patients in primary care through routine screening. Primary Care and Community Psychiatry, 13, 143-147.

Cukrowicz, K., Smith, P., & Poindexter, E. (2010). The Effect of Participating in Suicide Research: Does Participating in a Research Protocol on Suicide and Psychiatric Symptoms Increase Suicide Ideation and Attempts? Suicide and Life-Threatening Behavior, 40, 535-543.

Fujiwasa, D., Tanaka, E., Sakamoto, S., Neichi, K., Nakagawa, A., & Ono, Y. (2005). The development of a brief screening instrument for depression and suicidal ideation for elderly: The Depression and Suicide Screen. Psychiatry and Clinical Neurosciences, 59, 634-638.

Luomo, J.B., Martin, C.E., & Pearson, J.L. (2002). Contact with mental health and primary care providers before suicide: A review of the evidence. American Journal of Psychiatry, 159, 909-916.

Manber, R., Bernert, R. A., Suh, S., Nawoakowski, S., Siebern, A. T., & Ong, J. C. (2011). CBT for insomnia in patients with high and low depressive symptom severity: Adherence and clinical outcomes. Journal of Clinical Sleep Medicine, 7, 645–652.

McIntosh, J. L., & Drapeau, C. W. (for the American Association of Suicidology). (2012). U.S.A. suicide 2010: Official final data. Washington, DC: American Association of Suicidology, dated September 20, 2012, downloaded from http://www.suicidology.org.

Oyama, H., Fujita, M., Goto, M., Shibuya, H., & Sakashita, T. (2006). Outcomes of community-based screening for depression and suicide prevention among Japanese elders. The Gerontologist, 46, 821-826.

Unützer, J., Katon, W., Callahan, C. M., Williams, J. W., Jr., Hunkeler, E., et al. (2002). Collaborative care management of late-life depression in the primary care setting: A randomized controlled trial. Journal of the American Medical Association, 288, 2836-2845.

References for Adolescents

Diamond, G., Levy, S., Bevans, K.B., Fein, J. A., Wintersteen, M. B., Tien, A., & Creed, T. A. (2010). Development, validation, and utility of internet-based behavioral health screen for adolescents. Pediatrics, 126, e163-e170.

Horowitz, L. M., Bridge, J. A., Teach, S. J., Ballard, E., Klima, J., Rosenstein, D. L., et al. (2012). Ask suicide-screening questions (ASQ): A brief instrument for the pediatric emergency department. Archives of Pediatric Adolescent Medicine, 166, 1170-1176.

O’Connor, E., Gaynes, B. N., Burda, B. U., Soh, C., & Whitlock, E. P. (2013). Screening for and treatment of suicide risk relevant to primary care: A systematic review for the U.S. Preventive Services Task Force. Annals of Internal Medicine, 158, 741-754.