Journal of Pediatric , Vol. 25, No. 4, 2000, pp. 219–224

Commentary: Treatments for : Criteria, Mechanisms, and Health Care Policy

Arthur C. Houts, PhD University of Memphis

Mellon and McGrath (this issue) have done an ad- tive treatment that costs considerably more than a mirable job of applying the modified Chambless cri- cure? teria (Task Force on Promotion and Dissemination of Psychological Procedures, 1995; Chambless et al., 1996), as suggested by Journal of Pediatric Psychology History and Current Context of editors, to the outcome literature on psychological Psychological Treatments for Enuresis treatments for medically uncomplicated nocturnal enuresis in children. This brief commentary ad- We are fortunate to have the case of treatments for dresses three issues. First, what can we learn from enuresis to teach a historical lesson about the pit- the past? Specifically, how can we place psychologi- falls of our current enthusiasm for empirically sup- cal treatment for enuresis in historical context and ported treatments. The treatment of enuresis has a also in the current context of a general movement long and very colorful history dating as far back as within to respond to a managed 1550 B.C. Glicklich (1951) recounted two instruc- care environment by identifying empirically sup- tive examples of “treatments.” In West Africa, chil- ported psychological treatments? Second, and in dren who wet the bed were “treated” by attaching light of that history of psychological treatments, a large frog to their waist, and this apparently why should we take seriously the conclusion that frightened them into being dry. Among the Navaho “psycho” therapies, as contrasted with behavior tribe, one preferred treatment was a ritual that re- therapies, hold any promise for the treatment of quired enuretic children to stand naked over a children’s bedwetting? In other words, there is burning bird’s nest, and this was believed to pro- something very mistaken with a criterion that has duce a cure of bedwetting because birds did not soil resulted in a blessing even so mild as “promising” their nests. We may snicker at these practices of the for such “psycho” therapies. What might be added past, but the laugh is really on us. These practices to a criteria set to correct this type of mistake? worked, and they worked on a variable interval Third, and finally, how has it happened that the schedule of reinforcement because they were occa- most effective treatment for a problem that affects sionally followed by the spontaneous cessation of 10% of secondary school-age children has remained bedwetting, something that happens for about 16 so underutilized for over half a century? Something out of every 100 children within a 12-month pe- is seriously broken with our health care system and riod. It is no wonder then that so many peculiar our culture of caring for children who wet the bed. treatments have been tried, and so many odd prac- What kind of health care system routinely foregoes tices have persisted. In some ways what we really curing bedwetting and opts instead for mere pallia- need to know about treatments for bedwetting is which treatments definitely do not work. All correspondence should be sent to Arthur C. Houts, Department of Such a Popperian approach to the issue of iden- Psychology, University of Memphis, Memphis, Tennessee 38152. E-mail: [email protected]. tifying empirically supported treatments is counter-

᭧ 2000 Society of Pediatric Psychology 220 Houts

intuitive for most of us who have viewed science expectancy that one is receiving medication as op- as an inductive process. Ollendick (1999) recently posed to a nonmedication treatment. A procedure expressed the logic of current thinking regarding believable as a psychological treatment but known the empirically supported treatment movement. to be ineffective would provide the much needed parallel placebo condition for various psychologi- Surely “treatments that work” are desirable and cal treatments. their development and promulgation should be The history of the treatment of childhood en- encouraged; after all, to argue the converse, that uresis has demonstrated that verbal psychothera- “treatments that do not work” should be devel- pies have not produced outcomes that were much oped and disseminated hardly seems tenable better than pill placebo controls (Houts, Berman, & and makes little sense for a profession commit- Abramson, 1994). Some research in the 1960s in- ted to the welfare of those whom we serve (Ol- cluded treatments such as supportive counseling lendick, 1999, p. 1). and psychotherapy (De Leon & Mandell, 1966; Werry & Cohrssen, 1965). However, by the mid- Quite the contrary and from a methodological 1970s, most psychological research had moved on point of view, I believe that what we need most ur- and was devoted either to developing alternative gently is to identify treatments that do not work. In behavioral procedures based on operant condition- the case of much positive psychotherapy outcome ing (e.g., Azrin, Sneed, & Foxx, 1974) or to improv- research for various mental disorders, we already ing urine alarm treatments within Lovibond’s (1963) know that people get better with the passage of time, avoidance learning formulation (e.g., Young & Mor- and we believe some inert treatments are effective. gan, 1972). Mellon and McGrath have presented Whenever we study a problematic condition such as much of that history in their review, but they have bedwetting with a natural course toward remission, not been nearly so shocked as I have been to find we need to be particularly attentive to the frog effect the return of verbal psychotherapy as “promising.” and to the burning bird’s nest effect. Otherwise, our It is important to reiterate that this judgment was cumulative wisdom will include all manner of proce- fairly enough required of Mellon and McGrath, be- dures that have “worked” because we may mistake cause they duly applied the modified criteria for applying the procedure for the natural develop- empirically supported treatments. mental resolution of the problem. In the interest of returning psychotherapy for The process of identifying treatments that defi- enuresis back to the realm of the repressed, I offer nitely do not work could benefit us in two ways. the following analysis of the verbal psychotherapy The obvious benefit is to rule out ineffective proce- studies reviewed by Mellon and McGrath. This anal- dures. Knowing what definitely does not work ysis also leads to discussion of the problems with would clear the air and purchase a small degree of criteria for declaring some intervention “prom- progress even if we could not as yet identify a proce- ising.” dure that did in fact work. One of the great difficul- ties with developmental problems such as bedwet- ting is that at any given time, many different things Repressing Psychotherapy for Enuresis can appear to work. The scientific task is to weed out procedures that may appear to work from those Mellon and McGrath have included four reports on procedures that can be trusted to work reliably, not the use of hypnosis to treat enuresis and one report only because they have passed muster in numerous that included a cognitive self-control treatment. In randomized clinical trials but also because we have the analysis that follows, I have confined my re- verified knowledge about the mechanism through marks to the two controlled studies that used ran- which they produced positive outcomes. dom assignment to conditions, because there is no Second, identification of treatments that do not point in considering less rigorous designs in the work could provide much needed psychological case of such alternative treatments for enuresis. The placebo control groups. For quite some time, treat- fact is that we have over 50 randomized trials show- ment outcome researchers have needed a nonpill ing the effectiveness of urine alarm treatment, so any placebo condition. If a psychological treatment is new treatment has to meet a rather high standard. compared to a pill placebo, the comparison is never Hypnosis apparently qualified as promising on exactly parallel because the placebo condition con- the basis of one controlled study (Edwards & Van sists not only of an inert treatment but also of the Der Spuy, 1985), which was also accompanied by Treatments for Enuresis 221

two case series and one quasi experiment. The con- for bedwetting. These investigators compared a trolled study provided a good illustration about urine alarm protocol without any relapse preven- how research reports can be misleading. Kenneth tion component to their new cognitive behavioral Spence was alleged to have said that he could ad- intervention that consisted of some combination of vance psychology immediately if someone would self-monitoring, self-reinforcement, and learning just give him enough money to pay other people self-guided talk. They also included a token econ- not to conduct and publish research. Edwards and omy intervention and a no-treatment control group Van Der Spuy (1985) compared hypnotic trance in their study. The outcomes from this trial were induction and suggestions with trance alone, sug- also reported in two other publications (Ronen, Ra- gestions alone, and no treatment. The statistical re- hav, & Wozner, 1995; Ronen & Wozner, 1995), porting in this research report was completely where the same mistakes from the original publica- inadequate. The investigators did not assess pre- tion were repeated. The authors established a crite- treatment differences between the four groups, and ria of 3 consecutive weeks of dry nights for the they did not provide the data for a reviewer to do designation of “dry,” but they reported statistical so. This was important because there appeared to be analyses that were inappropriate. They displayed a differences in the groups at baseline with respect to 4 Group (Urine Alarm, Token Economy, Cognitive wetting frequency. The main analyses were con- Therapy, No Treatment Control) ϫ 3 Outcome (Dry, ducted as repeated measures analyses of variance on Improved, Dropout) table of the percentage of chil- Z scores of weekly wetting frequency during the 6- dren in each outcome category for each of the four week treatment period, but it was impossible to re- groups as their Table 2. The text repeatedly referred construct the actual data because no standard devia- to this table as Table 1, which in fact showed aver- tions (SDs)were ever reported. The authors stated age wetting frequency for the four groups. The au- that both suggestion conditions resulted in a reduc- thors reported a chi-square statistical analysis with tion of wet nights that was greater than the no- nine degrees of freedom on the data presented in treatment controls in a 6-week treatment period, their Table 2, and this resulted in a chi-square value but it is impossible to tell if the analyses were done with p Յ001. The appropriate degrees of freedom correctly because the authors do not report degrees for that omnibus test should have been six, not of freedom with their F tests. In the main analysis, nine. Nevertheless, that reported chi-square test no mention was ever made about the number of showed that at least one of the several possible one children who became dry during the treatment pe- degree of freedom tests within the 4 ϫ 3 table was riod for each of the four groups. What was later re- statistically significant. In fact, the chi-square test ported was that after a 6-month follow-up, only for the comparison of the 63.2% cure rate for the 19.4% of the children from the three hypnosis urine alarm with the 75% cure rate for the cognitive treatment groups had completely ceased bedwet- intervention was never reported, and when I calcu- ting. That outcome speaks for itself. Hypnosis was lated it, the chi-square value was less than one (.64) not effective except when it was compared to out- and associated with p ϭ .42. Nevertheless, the au- come for no-treatment controls. The authors did thors stated in their abstract that “cognitive inter- not even make that comparison using their own no vention was the most effective treatment method, treatment control data, but they relied instead on as evidenced by the highest rate of success and the the no-treatment control results from another pre- lowest rate of drop out or relapse” (Ronen, viously published study conducted by completely Wozner, & Rahav, 1992, p. 1). The one degree of different investigators and published 20 years ear- freedom test for the relative difference in rate of lier. The results from this clinical trial were only dropout was never reported by the authors, and marginally better than what could be expected from when I computed the correct test, there was no dif- investigations of spontaneous cessation of bedwet- ference in dropout between the cognitive group and ting due to maturation and the passage of time. the urine alarm group. This erroneous result regard- There is nothing whatsoever promising about hyp- ing rate of dropout was reported again by Ronen notic suggestion to treat bedwetting despite the fact and Wozner (1995). As for the authors’ claims re- that the literature contains one adequately de- garding differences in relapse rate, it was impossible signed, if poorly executed, study. to discern how they calculated a nine degree of free- Similar problems occurred in the report by Ro- dom chi-square from a 4 Group ϫ 2 Outcome (Re- nen, Wozner, and Rahav (1992), where the authors mained Dry vs. Relapsed) table. When I calculated reported outcomes for their new cognitive therapy the appropriate test, the comparison of relapse rate 222 Houts

between the cognitive intervention (15%) and the parents and professionals talking to children about urine alarm treatment (60%) was statistically sig- stopping bedwetting with little or no results? The nificant. At least one of the three conclusions stated one treatment that we know to work reliably, the in the abstract was actually supported by the data urine alarm, does not require cognitive control and presented. The authors of this report did not ad- most likely works via a process of active avoidance dress the question of how their results compared to conditioning (Houts, 1991, 1995). The issue in this so much previous research that has shown quite dif- case is about burden of proof. When we have very ferent outcomes. For example, if nothing is done to strong evidence for a dominant effective treatment, prevent relapse with the urine alarm, we have known as we do in the history of treatment for enuresis, for over 25 years that the relapse rate is likely to be and when we have such clear evidence for failure 40% or greater. This was never addressed. Why did of cognitively mediated treatments, the burden of the authors use a urine alarm intervention without proof is on those who propose a new variation of some relapse prevention method? the failed treatment and a mechanism that has been What was even more surprising was the finding shown to fail in the past. Why did such a treatment that 15 out of 20 children who received the cogni- appear to work in this particular trial? Where is the tive intervention in this study attained the dryness evidence that the comparison treatments were criterion. How could that have happened in light of faithfully conducted? the history of repeated failure of verbal psychother- Scientific reasoning occurs in a web of belief, apies for bedwetting? The authors did not comment and that reasoning is inherently conservative. We on that issue. In the introduction to their article, change our beliefs so as to produce minimal adjust- they spoke about the role of conscious learning of ment in the overall web of belief. In this sense, the daytime control of bladder function, but they never criterion of declaring an intervention promising on addressed the issue of how changing what children the basis of at least one well-controlled study is mis- said to themselves during the day could produce a leading. One well-controlled study is not enough, change in physical control over their bladder func- as the case of cognitive treatments for bedwetting tion during sleep. In a separate publication that ap- shows. The history of investigation in a domain peared 3 years later, the authors noted that children needs to be considered, and the identification of in the cognitive intervention also practiced reten- treatments that do not work can be relevant in as- tion control training, and they practiced urine re- sessing claims for “new and improved” treatments. tention during the day (Ronen & Wozner, 1995). It is disturbing enough to think that talking therapy The children were also told that “bedwetting does for enuresis may make a comeback, but it is even not depend on bad luck or illness but, rather, is a more disturbing to consider, as Mellon and McGrath function of motivation and willpower” (Ronen & have indicated, that the most effective treatment Wozner, 1995, p. 10). Apparently, the children in known for children’s bedwetting is rarely promul- the cognitive intervention group received rather ex- gated by our current health care system. tensive coaching and therapy sessions from the in- vestigators, whereas the urine alarm and the token economy interventions were conducted by parents. Enuresis Treatment and the Health All of the children in the no-treatment control Care System group were lost to follow-up. In sum, there is con- siderable reason to doubt the care and methodologi- As Mellon and McGrath have noted in their review, cal rigor with which this study was conducted and the available evidence regarding what treatments reported. children receive for bedwetting suggests that the In light of the extensive history of failure of cog- front line service providers, pediatricians and family nitive types of intervention for bedwetting, the physicians, have traditionally favored medication outcome from Ronen, Wozner, and Rahav (1992) treatments and have only rarely recommended should be considered an anomaly. The fundamental urine alarm treatment. Although we do not have idea that children willfully control their thought more recent direct surveys of medical practitioners, processes during sleep and the proposition that there is every reason to believe that the situation such cognitive control accounts for bedwetting are has worsened due to massive advertising by the simply not plausible. If such propositions were true, pharmaceutical industry, to typical procedures of then why do we have a history of 3,500 years of managed care organizations, and to the failure Treatments for Enuresis 223

of health care professionals actually to practice insurance policies provide benefits according to evidence-based health care. what is determined to be medically necessary by a Over the past 10 years, millions of dollars have physician, and those insurance policies also typi- been spent promoting synthetic vasopressin cally have different reimbursement policies for pro- (DDAVP) as a treatment for bedwetting. First ap- viders of psychological services, who are most likely proved in the United States in 1989, this treatment competent to deliver or supervise urine alarm treat- quickly replaced the leading medication treatment ments. A medical doctor may judge that urine alarm for bedwetting, imipramine, because DDAVP had treatment is medically necessary, but the treatment fewer side effects and did not carry the mortality may not be reimbursable because of limitations of risk due to cardiac failure associated with imipra- the policy on nonmedical providers. Similar scenar- mine. When it was first introduced, DDAVP was ios can and do occur in managed care organizations available as a nasal spray. That delivery system was where enrollee children are restricted to certain pro- required because the synthesis of the drug was said vider lists, and many of those organizations relegate to be too costly to be affordable in oral tablets, a mental health services to master’s level providers route of administration that required wasting a lot without psychological training in how to use the of the compound to achieve adequate blood levels. urine alarm protocol. Managed care organizations In the past 3 years, the drug has been made avail- commonly do not know about treatment outcome able in tablet form. The monthly cost of treating evidence in the case of bedwetting, despite the fact a child with DDAVP is approximately $150. What that they are interested in reducing costs and pro- should be remembered is that DDAVP rarely stops viding evidence-based treatments. As a result of bedwetting, and when the child stops taking the these processes, children who are diagnosed with medication, the child reverts to wetting (Moffatt, functional nocturnal enuresis are treated with what Harlos, Kirshen, & Burd, 1993). Given the mecha- is reimbursable and what is familiar, namely, medi- nism by which DDAVP works, one should expect cations. complete relapse once the medication is with- In addition to not having any easy channel of drawn. referral for urine alarm treatment, most pediatri- In order to maintain a child on DDAVP for 1 cians and family physicians are visited regularly by year, the cost would be approximately $1,800. A 12- representatives of the various pharmaceutical com- week course of urine alarm treatment that included panies that manufacture and advertise medication all supporting materials and regular consultations treatments for bedwetting. The culture of profes- with a doctoral-level psychologist would cost only sional medicine and the culture of many parents $500. In other words, maintaining a child on inef- support medication solutions to children’s prob- fective drug treatment for a year costs over three lems. As someone who has followed these cultures times what it costs to cure bedwetting. Why then surrounding the problem of enuresis for the past 20 do most children still receive some type of pharma- years, I am not surprised that urine alarm treatment ceutical treatment when even leading medical au- remains so underutilized. Perhaps it will take an- thorities (Moffatt, 1997) have clearly recommended other 20 years for a treatment that was devised over urine alarm treatment as the treatment of choice 60 years ago finally to be provided on a routine ba- based on current outcome evidence? sis to the almost 7 million children affected by en- Currently, third-party payers such as insurance uresis in the United States. companies and managed care organizations along with medical professionals determine what hap- Received August 23, 1999; accepted August 25, 1999 pens to the majority of bedwetting children. Most

References

Azrin, N. H., Sneed, T. J., & Foxx, R. M. (1974). Dry-bed Johnson, B., Woody, S. R., Sue, S., Beutler, L., Wil- training: Rapid elimination of childhood enuresis. Be- liams, D. A., & McCurry, S. (1996). An update on em- haviour Research and Therapy, 12, 147–156. pirically validated therapies. Clinical Psychologist, 49, Chambless, D. L., Sanderson, W. C., Shoham, V., Bennett 5–18. Johnson, S., Pope, K. S., Crits-Christoph, P., Baker, M., De Leon, G., & Mandell, W. (1966). A comparison of con- 224 Houts

ditioning and psychotherapy in the treatment of (1993). Desmopressin acetate and nocturnal enuresis: functional enuresis. Journal of Clinical Psychology, 22, How much do we know? , 92(3), 420–425. 326–330. Ollendick, T. H. (1999). Clinical science and clinical prac- Edwards, S. D., & Van Der Spuy, H. I. J. (1985). Hypnother- tice: Where to from here? The Clinical Psychologist, apy as a treatment for enuresis. Journal of Child Psychol- 52, 1–3. ogy and Psychiatry, 26, 161–170. Ronen, T., Rahav, G., & Wozner, Y. (1995). Self-control and Glicklich, L. B. (1951). An historical account of enuresis. enuresis. Journal of Cognitive Psychotherapy, 9, 249–258. Pediatrics, 8, 859–876. Ronen, T., & Wozner, Y. (1995). A self-control intervention Houts, A. C. (1991). Nocturnal enuresis as a biobehavioral package for the treatment of primary nocturnal enure- problem. Behavior Therapy, 22, 133–151. sis. Child and Family Behavior Therapy, 17, 1–20. Houts, A. C. (1995). Behavioural treatment for enuresis. Ronen, T., Wozner, Y., & Rahav, G. (1992). Cognitive inter- Scandinavian Journal of Urology and Nephrology, suppl vention in enuresis. Child and Family Behavior Therapy, 173, 83–87. 14, 1–14. Houts, A. C., Berman, J. S., & Abramson, H. A. (1994). The Task Force on Promotion and Dissemination of Psycholog- effectiveness of psychological and pharmacological ical Procedures. (1995). Training in and dissemination treatments for nocturnal enuresis. Journal of Consulting of empirically-validated psychological treatments: Re- and Clinical Psychology, 62, 737–745. port and recommendations. Clinical Psychologist, 48, Lovibond, S. H. (1963). The mechanism of conditioning 3–23. treatment of enuresis. Behaviour Research and Therapy, Werry, J. S., & Cohrssen, J. (1965). Enuresis: An etiologic 1, 17–21. and therapeutic study. Journal of Pediatrics, 6, 423–431. Moffatt, M. E. K. (1997). Nocturnal enuresis: A review of Young, G. C., & Morgan, R. T. T. (1972). Overlearning in the efficacy of treatments and practical advice for cli- the conditioning treatment of enuresis: A long-term nicians. Developmental and Behavioral Pediatrics, 18(1), follow-up study. Behaviour Research and Therapy, 10, 49–56. 419–420. Moffatt, M. E. K., Harlos, S., Kirshen, A. J., & Burd, L.