A Randomized Controlled Trial of Hypnotherapy for Smoking Cessation Robert Lambe, MD, Carl Osier, MD, Peter Franks, MD Plainville, Massachusetts, and Ballston Spa and Rochester, New York A randomized controlled study in a family practice setting was conducted on the use of hypnosis in helping people quit smoking. In the hypnosis group 21 percent of patients quit smoking by the three month follow-up compared with 6 percent in the control group. By six months there were no significant differences between the two groups, and at one year 22 percent in the hypnosis group and 20 percent in the control group had quit. The only significant predictor of success with quitting was having a college education. Cigarette smoking is one of the leading prevent­ success of hypnosis; follow-up supportive contact able causes of serious cardiac and pulmonary dis­ of patients; standardized follow-up of smokers for ease in the United States. Despite a multiplicity of a significant interval; and use of individualized techniques to help smokers quit, the magnitude of hypnotic suggestions. An additional major differ­ the problem remains enormous. One of the meth­ ence between existing studies and the one re­ ods that has been widely written about is hypno­ ported here is that this study was set in a primary therapy, yet the studies in the literature present a care setting, with the hypnosis performed by fam­ bewildering array of techniques and success rates. ily physicians. Furthermore, the protocol was one Reports range from 94 percent of smokers quit­ that could be followed easily by primary care ting1 to 20 percent.2 Most studies have used highly physicians having limited training in hypnosis and selected subjects and often no control groups, so limited time. that the interpretation of results for use in the pri­ mary care setting is difficult. In addition, many METHODS investigators have failed to provide information on The patient population at the Rochester Family long-term follow-up, despite known high recidi­ Medicine Program is representative of a cross- vism rates for cigarette smoking. section of Monroe County in upstate New York, In 1980 Holroyd3 summarized the experience which has been described in detail elsewhere.4 All with hypnosis and smoking in the 1970s and in­ patients aged over 18 years who did not have a cluded several suggestions for future research psychiatric diagnosis entering the Family Medi­ based on the weaknesses of prior studies. This cine Center for scheduled health care received a paper reports a randomized control trial of hypno­ screening questionnaire to determine eligibility for therapy that was carried out to help smokers quit the study. Patients were eligible if they wished to in which several of Holroyd’s suggestions were quit smoking and were willing to undergo hypno­ included: investigation of individual differences sis. These patients were given a second question­ among subjects, such as smoking histories and naire about their health history. demographic data to look for relationships with Questions included years of smoking, number of cigarettes smoked, presence of other smokers at home, presence of smokers at work, previous at­ From the Department of Preventive, Family, and Rehabilitation Medicine, University of Rochester School of Medicine and Flighland tempts at quitting, perceived stress, and education. Hospital, Rochester, New York. Requests for reprints should be ad­ Recruitment was planned to continue until 180 dressed to Dr. Peter Franks, Jacob W. Holler Family Medicine Cen­ ter, 885 South Avenue, Rochester, NY 14620. eligible patients could be identified. This number © 1986 Appleton-Century-Crofts THE JOURNAL OF FAMILY PRACTICE, VOL. 22, NO. 1: 61-65, 1986 61 HYPNOTHERAPY FOR SMOKING CESSATION was based on a one-tailed alpha error of 0.05 and a their originally assigned groups. Univariate com­ beta error of 0.1. It was assumed that the effec­ parisons were made using chi-square tests or t tiveness of antismoking advice would be 5 per­ tests as appropriate. Outcomes, which were cent5 and that to be clinically useful, smoking measured at 3, 6, and 12 months, were whether the would have to be at least 20 percent effective. This patient had quit and the number of cigarettes number also allowed for a 10 percent dropout smoked. All patients who could be contacted at rate.6 After recruitment patients were randomized each follow-up period were included in each anal­ to hypnosis and control groups using the Zelen ysis. To enable examination of the independent design.7 contribution of hypnotherapy to the outcomes of All control patients received a letter notifying interest, while controlling for baseline differences them that the physicians at the Family Medicine between the hypnosis and control groups, step­ Center hoped they would quit smoking. They also wise regression analyses were used. Study group received a copy of the National Institutes of and the factors thought to affect smoking (from the Health booklet Calling It Quits. Follow-up began second questionnaire) were entered into these with the date of this letter. A letter was mailed to analyses as independent variables in the order in all patients in the hypnosis group asking them which they accounted for most of the change in the to make their first appointment with one of the outcome variables. Logistic regression was used hypnotists (C.O. or R.L.). If the patient did not for the analyses of the dichotomous outcome respond, a second letter was mailed. Continued “ quitting,” and ordinary linear regression was failure to respond then led to a telephone call to used for the outcome “number of cigarettes encourage participation. If the patient still refused smoked.” All analyses were conducted using the entry, had already quit, or had moved and could SAS computer package.8 not be contacted, follow-up was begun at the time of the telephone call. The hypnotherapy consisted of two 40-minute RESULTS sessions, two weeks apart. At the first visit, after Two hundred forty-two patients who were obtaining informed consent, the hypnotists fol­ smokers (49 percent of all patients) were con­ lowed a standard protocol (available on request tacted, and 180 (74 percent) who were interested in from the authors). After the trance was termi­ hypnosis as a method of helping them quit were nated, the method of autohypnosis was explained included in the study for randomization. Because and a list of instructions given. An evaluation of the patient population in this study was highly depth of trance was noted by the hypnotists on a mobile, follow-up was a consistent problem. Sev­ standard form. At the time of the second session, a eral patients were temporarily lost to follow-up trance was induced again, and suggestions rein­ because of brief loss of telephone service or delays forced. During the trance, the subject was asked to in obtaining new addresses. Furthermore, three choose a quit date. Follow-up began on the date of patients were known to have died during the the second session. study, two from smoking-related diseases. Table 1 All patients in the hypnosis and control groups summarizes the number of patients available for were called three times in the four months after the follow-up throughout the study. In the hypnosis identified start date of follow-up. These calls were group, 45 patients underwent at least one hypnosis for several purposes: to ascertain the amount of session, 6 quit before hypnosis, 18 declined hyp­ smoking (number of cigarettes per day), to offer nosis, and 21 were lost to follow-up. encouragement, and to determine whether sub­ Baseline comparisons between the two groups jects in the hypnosis group were using self­ are displayed in Table 2. Patients in the hypnosis hypnosis. Subsequent to this contact, all subjects group tended to be younger, more educated, less were contacted again by telephone or question­ likely to have Medicaid, less likely to have other naire at six and 12 months after intervention to smokers at home, but more likely to have other determine the amount of their smoking. smokers at work. Using the intention-to-treat principle, all pa­ At the three-month follow-up contact, hypnosis tients were included in the analysis according to patients were significantly more successful in re- 62 THE JOURNAL OF FAMILY PRACTICE, VOL. 22, NO. 1, 1986 HYPNOTHERAPY FOR SMOKING CESSATION TABLE 1. FOLLOW-UP OF ALL RANDOMIZED TABLE 2. BASELINE COMPARISONS BETWEEN PATIENTS HYPNOSIS AND CONTROL GROUPS Hypnosis Control Hypnosis Control Total randomized 90 90 Age, years (mean) 32.4 38.8 First telephone call 69 68 Female (percent) 69 68 Third telephone call 57 58 College educated 36 24 (3 months) (percent) 6-month contact 66 64 Stress (percent) 62 64 12-month contact 60 60 Cigarettes smoked 25.7 26.6 per day (mean) Medicaid (percent) 14 24 Other smokers 55 61 at home (percent) Other smokers 72 57 at work (percent) Previous smoking 6 10 ducing their smoking consumption, but by the program (percent) six-month follow-up the two groups were not sig­ Smoking, years 12.7 11.3 nificantly different. These results are summarized (mean) in Table 3. The results were not significantly changed when only those subjects who actually underwent hypnosis were compared with the con­ trol group. Stepwise regression analysis revealed that the only consistent predictor of success for risk ratio = 4.3, 95 percent confidence interval = reduction of smoking was having a college educa­ 2.8-6.8) of successful quitting. The analysis re­ tion (R2 = 14 percent, F = 19.4, P = .0001 at the vealed that at the time of the three-month follow­ 12-month follow-up). up that smoking fewer cigarettes at entry into the At the first three-month follow-up contact, study (beta = - .07, standard error = .03), being in being in the hypnosis group and being exposed to the hypnosis group (adjusted risk ratio = 3.6, 95 other smokers at work were also significant, but percent confidence interval = 1.9-6.8), and having these variables became nonsignificant with longer a college education (adjusted risk ratio = 7.1, 95 follow-up.