The Journal of Behavioral Health Services & Research

, Volume 34, Issue 1, pp 43–55

Effects of a Tobacco Ban on Long-term Psychiatric Patients


    • Research DepartmentPenetanguishene Mental Health Centre
  • Daniel Parle
    • Research DepartmentPenetanguishene Mental Health Centre
  • Joseph Gagné
    • Research DepartmentPenetanguishene Mental Health Centre
Regular Article

DOI: 10.1007/s11414-006-9043-1

Cite this article as:
Harris, G.T., Parle, D. & Gagné, J. J Behav Health Serv Res (2007) 34: 43. doi:10.1007/s11414-006-9043-1


A total ban on all tobacco products was implemented in a diverse psychiatric institution. A post hoc evaluation examined the effect of the ban on long-term patients by comparing their characteristics the year before the ban to the year after. Several variables measuring physical health, psychiatric symptomatology, feelings of well-being, and interpersonal conflict were coded with very high reliability from health records. For the majority of patients who were in the maximum security forensic division, the tobacco ban was associated with almost no detectable ill effects with some clear benefits. Among the remainder of the long-term patients, the ban might have been associated with a temporary increase in physical aggression towards staff members. It was concluded that successful implementation, and the avoidance of ill effects, depended entirely on the success staff members had in actually preventing patient access to tobacco.


policy evaluationtobacco controlinpatientsmental healthpsychiatric hospitals

Annually, more than 45,000 Canadians die prematurely from tobacco use. Smoking is related to more than 24 illnesses1 and is the leading preventable cause of death and disease.2 Employers could reduce operating costs, provide healthier environments for clients and employees, and increase employee satisfaction by adopting smoke-free policies3 which could be facilitated by effective cessation techniques.46

The highest rates of tobacco addiction are found among persons with major mental illness, especially schizophrenia.7,8 Evins and colleagues9 concluded that smoking rates in patients with serious mental disorders ranged from 75 to 85% with heavier and “efficient” smoking.10,11 Among persons with addictions, Currie et al.12 reported that smoking was three times that in the general population (estimated at 23%). Substance abusers are more likely to die from tobacco related illnesses than from alcohol related illnesses.13,14 Death due to smoking-related illnesses is two to six times greater in patients with major mental illness than in the general population.914 It seems likely that smoking in these vulnerable populations has often been ignored, in part because of the belief that persons with mental illness benefit from smoking and that they lack the capacity to quit. Some clinicians fear that quitting will lead to worsening psychiatric symptoms and increased aggression.15 For persons being treated for addiction, counselors are often concerned that recovery is jeopardized if clients are obliged to abandon all addictions simultaneously. These concerns appear unfounded. Individuals with mental illnesses are interested in cessation and benefit from cessation efforts.916 Patients in recovery from addictions are only slightly less inclined to try quitting tobacco than the general population and quitting can improve long-term abstinence from other drugs.1721

Where total tobacco proscriptions (banning all tobacco products) have been instituted, common experiences include greater ease of enforcing tobacco prohibition than in partial bans, elimination of environmental tobacco smoke, no rise in aggression, and gradual acceptance of the ban.8,2225 Some facilities have reported eventual acceptance of partial bans.26,27 Elsewhere, however, clandestine smoking, pressure on staff members to escort patients to smoke breaks, erosion of therapeutic relationships, and cigarette black markets have led some facilities to retreat from smoke-free efforts.2830

In summary, health, therapeutics, and ethics all demand that mental health facilities consider ways to reduce or eliminate clients’ smoking. The literature suggests that total bans can be at least as effective as partial bans or the relegation of smoking to outdoors. Clinicians are particularly poised to provide such cessation support methods as short-term counseling, smoke free zones, and individualized cessation programs.9,3134

Background to the Present Overall Approach

In deciding what to measure in evaluating the effects of tobacco control efforts in psychiatric settings, the available literature suggests the following. Both men and women smokers report concerns about weight gain when quitting,3537 and these probably undermine decisions to quit and remain abstinent. Average weight gain by heavy smokers has been reported to be 8.9 kg.38,39 Although it is often thought to help in coping with stress, smoking is empirically associated with mood lability, heightened stress, and depression. Cessation might lead to temporary worsening of mood, but cessation beyond 3 months is associated with improved mood.40,41 Interestingly, empirical studies have not supported the expectation of increased disruptive and assaultive behavior among patients obliged to quit smoking.8,22,23,25,26,42

Addington31 reviewed the literature on nicotine and schizophrenia hypothesizing inverse links between nicotine and negative symptoms, Parkinsonism, and the sedative effects of antipsychotic drugs. Hughes43 postulated that stimulant effects of nicotine counteract the sedative side effects of many psychoactive drugs. Williams and Ziedonis14 suggested that smoking by persons with mental illness or addiction to other substances is a form of “self-medication” for depression, anxiety, boredom and loneliness. They hypothesized that smoking by persons with schizophrenia produces transient normalization of abnormal sensory activity and is associated with higher levels of positive symptoms and decreased negative symptoms. They14 also cited studies linking smoking to long-term lessening of depressive symptoms, but increased depression during cessation. Smith and colleagues,44 however, evaluated the short-term effect of admission to a smoke-free acute facility reporting that nicotine withdrawal did not aggravate psychiatric symptoms. Smoking increases the metabolism of some psychoactive drugs and can reduce serum levels as much as 40%.14 For clozapine, smoking cessation results in such large increases in serum levels that precautions are required to avoid toxicity.31,45

The Present Study

The Mental Health Centre Penetanguishene had a history of facilitating smoking. Tobacco was sold at subsidized prices in canteens; ventilated smoking rooms were built in each ward; and clinicians were explicitly assigned to manage tobacco, escort patients outdoors to smoke and light cigarettes. Cigarettes and smoke breaks were used as informal rewards, and until 10 years ago, as earned reinforcers in behavior modification.8 The 296-bed facility is 150 km from Canada’s largest city in a small town with a population of 8,500. It includes a 140-bed maximum-security unit housing male mentally disordered offenders, almost all of whom have a history of serious interpersonal violence. Elsewhere on the grounds is a 26-bed program for acute short-term civil patients and a 38-bed program for long-term patients. A 26-bed unit houses elderly psychiatric patients and a 25-bed program is for individuals with developmental disability and mental illness. A 12-bed program provides residential care for substance abuse and mental illness, and a 20-bed minimum-security ward houses forensic patients.

Despite predictable difficulties in enforcing a tobacco ban in such a diverse institution, the administration announced such policy to take effect on May 6, 2003.46 This announcement met with much surprise and dissatisfaction from some clients and staff members. Indeed, even many nonsmokers questioned its wisdom, especially for the long-term, maximum security male patients with histories of serious violent crimes. After the ban was in place for several months, a methodologically sound evaluation was attempted. A preferable design might have entailed quantification of patients’ pre-ban smoking habits with prospective data collection on the circumstances (i.e., tobacco-related or not) of aggression and interpersonal conflict. Psychiatric symptoms thought especially sensitive to tobacco withdrawal might have been assessed specifically. In the event, however, the present evaluation was designed after implementation and relied on information already routinely gathered. Nevertheless, the authors attempted to ascertain the clinical effects of the ban in as many domains as possible.



A search identified inpatients present for the entire period from May 6, 2002 to May 9, 2004 (1 year before and after the implementation of the tobacco ban). It was also possible to distinguish patients who had smoked before the ban from those who did not. Thus, pre–post differences unique to the smokers can reasonably be attributed to the ban. For all variables, research assistants extracted weekly raw counts or averages from personal health information files (clinical records). For a random sample of ten, two research assistants independently recorded all variables to permit an assessment of inter-rater reliability. For each variable, inter-rater reliability was assessed by intra-class correlation coefficients and exceeded 0.90, p < 0.01.


Average daily dose of clozapine and olanzapine, and clozapine blood levels were recorded for each week. The clinical staff routinely made daily detailed observational notes, sometimes using structured formats, and these were used to count for each patient the number of instances each week of physical aggression defined as forceful and/or aggressive physical contact (biting, punching, hitting, slapping, pushing, spitting, kicking, and throwing anything) separately for patient victims, staff victims, and objects. The number of instances of verbal aggression (defined as uttering threats, name-calling, insults, directed yelling, and directed profanity) were recorded separately for patient and staff victims. Restraint or seclusion was also recorded, as was each administration of pro re nata (prn) drugs for temporary relief of anxiety or agitation.

Researchers used the daily notes to count, for each of the 104 study weeks, each time a patient was noted to be responding to internal stimuli (i.e., experiencing a hallucination or thought disorder) as indicated by talking to oneself, staring for long periods, responding oddly, showing delayed responses, or being severely preoccupied. Any mention that the patient was exhibiting self-isolation or lack of activity was also recorded. Research assistants counted each instance in which the patient was noted to exhibit good mood (i.e., clinical observations using such terms as cheerful, happy, bright affect, upbeat, smiling, joking, optimistic, outgoing, enthusiastic, pleasant, confident, content, jovial, positive, cooperative, etc.). Similarly, they also recorded each instance of poor mood (i.e., clinical observations using such terms as angry, sad, down, discouraged, miserable, withdrawn, depressed, disturbed, afraid, annoyed, negative, hopeless, pessimistic, irritable, suicidal, flat affect, etc.).

Finally, two measures were recorded less frequently: using records of annual physicals, research assistants recorded evidence of chronic obstructive pulmonary disorder in the two study years and recorded the result of monthly weighings.


Throughout the 2-year period, 119 inpatients remained—83 in the maximum security division and 32 in the open wards (the home for four patients was not recorded). The patients (89% male) had a mean age of 46.8 (SD = 11.1) years, with primary diagnoses of schizophrenia (47%), affective and other psychoses (14%), personality disorder (17%), mental retardation (12%), and unspecified and other disorders (10%).

Pre–post comparisons

Among 23 smokers rated as having signs of compromised cardio-pulmonary health at their annual medical checkups in the year before the tobacco ban, 17 received a clear/healthy assessment at their annual physical examinations in the year after (p < 0.05, Fisher’s exact test). For all results that follow, inferences may rely on the comparison of confidence intervals;47 a mean falling outside the 95% CI of another may be taken as a statistically significant difference. For most measures, Table 1 shows the average scores for the year before the tobacco ban and the year after. Some measures that showed no differences on any factor (lack of activity, responding to internal stimuli, verbal aggression, use of prn medications) are not shown. Among the comparisons in Table 1, some showed overall differences probably having little to do with the tobacco ban. For example, patients in the open wards, especially nonsmokers, exhibited more physical aggression than patients in maximum security. This was probably due to selection—only very violent civil patients would be expected to remain in the hospital for 2 years or more, while many forensic patients remain for years, nonviolent behavior notwithstanding.48
Table 1

All statistically significant differencesa obtained




Year 1

Year 2

Year 1

Year 2

Mn (SD)

95% CI

Mn (SD)

95% CI

Mn (SD)

95% CI

Mn (SD)

95% CI

Good mood (all wards)

41.3 (70.6)

25.2 to 57.3

29.9 (32.0)

22.6 to 37.2

56.2 (104)

23.0 to 89.3

57.2 (107)

21.6 to 92.8

Clozapine dose (all wards)

504 (150)

443 to 565

401 (169)

333 to 470

391 (81.3)

262 to 520

466 (162)

63.3 to 868

Clozapine blood levels (all wards)

414 (145)

356 to 472

590 (182)

515 to 666

212 (90.2)

0 to 435

527 (431)

0 to 1,597

Weight (kg) (all wards)

88.7 (20.7)

84.0 to 93.4

93.5 (22.2)

88.2 to 98.5

88.8 (22.2)

81.8 to 95.8

89.3 (24.2)

81.7 to 96.8

Self-isolation (maximum security)

7.04 (10.9)

4.12 to 9.95

5.82 (9.42)

3.27 to 8.36

2.88 (3.83)

1.34 to 4.43

2.24 (3.02)

0.99 to 3.49

Bad mood (maximum security)

13.2 (16.5)

8.76 to 17.6

10.3 (13.2)

6.71 to 13.9

18.6 (38.6)

3.35 to 33.9

19.3 (38.5)

3.44 to 35.2

Good mood (maximum security)

38.2 (77.8)

17.6 to 58.2

21.9 (22.3)

15.9 to 27.9

37.9 (50.2)

17.6 to 58.2

29.0 (16.9)

21.8 to 36.3

Restraint and seclusion (maximum security)

6.29 (12.8)

2.87 to 9.70

6.78 (19.5)

1.51 to 12.1

9.74 (26.3)

0 to 20.16

9.08 (23.5)

0 to 18.8

Physical aggression patients (maximum)a

0.91 (2.11)

0.34 to 1.48

0.93 (2.10)

0.36 to 1.49

0.96 (1.74)

0.27 to 1.65

0.92 (1.93)

0.12 to 1.72

Physical aggression to staff (maximum)

0.63 (1.53)

0.21 to 1.04

0.80 (1.81)

0.31 to 1.29

1.89 (4.00)

0.31 to 3.47

2.56 (7.24)

0 to 5.55

Self-isolation (open wards)

5.12 (9.60)

0.18 to 10.1

3.24 (7.0)

0 to 6.84

3.57 (5.60)

0.34 to 6.80

0.50 (1.16)

0 to 1.17

Bad mood (open wards)a

36.7 (35.0)

18.7 to 54.7

31.7 (31.6)

15.5 to 47.9

32.5 (39.6)

8.64 to 56.4

30.4 (26.6)

15.1 to 47.9

Good mood (open wards)

53.8 (51.3)

27.4 to 80.2

46.2 (43.1)

24.1 to 68.4

90.1 (160)

0 to 182

104 (165)

8.01 to 199

Restraint and seclusion (open wards)

5.18 (8.23)

0.94 to 9.41

3.65 (7.75)

0 to 7.63

33.2 (92)

0 to 86.3

71.4 (159)

0 to 163

Physical aggression to patients (open)

7.76 (26.7)

0 to 21.5

1.82 (5.80)

0 to 4.81

34.4 (101)

0 to 92.6

29.9 (75.3)

0 to 73.4

Physical aggression to staff (open)

2.76 (6.75)

0 to 6.24

7.35 (26.5)

0 to 21.0

85.6 (283)

0 to 249

81.1 (251)

0 to 226

Italics indicates significant differences associated with the tobacco ban. Statistical significance was evaluated with ANOVA, but may also be inferred from 95% confidence intervals as explained in the text.

aIndicates a variable that exhibited no statistically significant differences, but is included here for comparison and completeness.

Statistically significant differences obtained among smokers without parallel differences among nonsmokers are the ones that might be uniquely attributed to the tobacco ban. These are indicated in italics in Table 1. Thus, in open wards, the ban appeared to be associated with an increase in aggression towards staff members and a similar decrease in aggression towards patients. In addition, across all wards, the tobacco ban appeared to be uniquely associated with patients being less often observed to be in a “good mood,” and an average weight gain of almost 5 kg, a decrease in daily clozapine dose of 100 mg accompanied by a similar mean increase in serum levels.

More conservative null hypothesis significance testing supported the same conclusions. Analyses of variance with one within-subjects factor (before vs after) and two between-subjects factors (smoker vs nonsmoker and maximum security vs open wards) were conducted. For physical aggression towards staff, there was a main effect of location, F(1, 106) = 4.65, p < 0.05, and smoker, F(1, 106) = 4.09, p < 0.05, both of which were qualified by the three-way interaction, F(1, 106) = 4.33, p < 0.05—the tobacco ban was associated with an increase in physical aggression towards staff only by smokers in the open wards. Similarly, the ban was associated with a decrease in clozapine dose, F(1, 19) = 2.01, p < 0.10, one-tailed, accompanied by an increase in serum clozapine levels only among smokers, F(1, 19) = 6.86, p < 0.05. Weight increased especially in maximum security, F(1, 107) = 4.47, p < 0.05, and especially among smokers, F(1, 107) = 5.82, p < 0.05. There were no differences between analyses of variance and decisions based on confidence limits in conclusions about statistical significance, with two exceptions: analyses of variance identified no significant effects for good mood and no significant interactions for physical aggression towards patients. As discussed next, the smoking ban was unlikely to have affected these measures.

Plotting weekly totals

For measures that appeared to show a unique effect of the tobacco ban, weekly data were plotted (Figs. 1, 2, and 3). These indicated that the apparent differences for physical aggression towards patients (top of Fig. 1) and good mood (top of Fig. 2) were unlikely to have anything to do with the ban because any apparent differences are attributable to spikes in the measures early in the year1 before the ban. The difference in physical aggression towards staff members (bottom of Fig. 1) is not entirely clear because it appears to have occurred at some delay. Changes in weight (bottom of Fig. 2) and clozapine dose and blood levels (Fig. 3) are more clearly attributable to the ban. The ban appeared to increase clozapine serum levels, accompanied by less variability (bottom of Fig. 3), at lower drug doses (top of Fig. 3). Finally, the tobacco ban was associated with an increase in olanzapine dose just for the patients who had been smokers (Fig. 4).
Figure 1

Physical aggression by smokers on nonforensic wards directed towards fellow patients (top) and staff members (bottom). The vertical line indicates the day the tobacco ban was implemented.
Figure 2

“Good mood” scores (top) and mean body weight (bottom) for smokers in all wards. The vertical line indicates the day the tobacco ban was implemented.
Figure 3

Mean daily clozapine dose in mg/day (top) and serum levels (bottom) for smokers. The vertical line indicates the day the tobacco ban was implemented.
Figure 4

Mean daily olanzapine dose (mg) for smokers ( and nonsmokers ( for the year before the tobacco ban (Pre) and the year after (Post).


The results indicated that a total ban on tobacco in a multiprogram psychiatric facility had few negative and some positive effects on long-term inpatients. There appeared to be an increase in patient–staff physical aggression in the nonforensic areas, and there were significant weight gains among those smokers who were obliged to quit. The increase in aggression appeared to return to pre-ban levels within a year, while the weight gains did not. Weight gains were no greater than expected based on the literature.

Meanwhile, the ban was associated with a significant decrease in pulmonary difficulties among smokers. For patients receiving clozapine, the ban was also associated with a decrease in the drug dose and an increase in serum drug levels accompanied by more stability. There was an unexplained2 increase in the average dose of olanzapine. Although, many aspects of psychiatric symptomatology and overall psychological well-being were examined, there was no evidence of any other adverse effects of the ban3.

Implications for behavioral health

One result for the maximum security forensic wards (comprising the majority of the patients studied) was especially notable. Before the ban, there were dire predictions from staff members and patients that it would surely cause mayhem. The prediction was that, for lifelong, severely mentally ill smokers who also had a serious history of violent crime (usually including homicide), the compulsory tobacco ban would produce an explosion in violence. The results clearly did not support this; in fact, no evidence of even the most minor troubles was found in maximum security wards. The study showed that the tobacco ban was perhaps close to a nonevent in maximum security because nicotine withdrawal was quick and complete for most smokers. The results are consistent with others’ evaluation of a smoking ban in a maximum security psychiatric facility.22

The results for the minority of long-term patients in other wards were more mixed and suggested that the ban produced an increase in physical aggression directed towards staff members. Interestingly, this occurred in the absence of any detectable worsening of psychiatric symptoms. A key question raised by the findings is, “Why did physical aggression increase in other wards and not in maximum security where almost all patients had much more severe histories of violence?” One likely explanation is that the nonforensic staff were less successful in stopping the use of tobacco because patients in nonforensic wards had more opportunities to obtain it (from visitors, and while off the grounds for work or recreation). It seems likely that many tobacco-addicted nonforensic patients spent the year after the ban in partial and intermittent nicotine withdrawal.

Other research suggests that clinical staff are sometimes pessimistic about the value of smoking reduction efforts and that their reluctance to enforce antismoking policies can be responsible for lack of success.8,49,50 Had the nonforensic staff in the present study been better able to prevent access to tobacco, the results there might have mirrored those in maximum security. Indeed, redoubled efforts at the end of the second year to eliminate tobacco might have been partly responsible for the return of aggression to pre-ban levels.

As a rigorous evaluation, the present study had limitations. Data were gathered retrospectively from routine, but very detailed, clinical records. No measures hypothesized to be specifically related to nicotine withdrawal were gathered. The present evaluation concentrated on long-term patients (especially those with a history of interpersonal violence) where it was anticipated that problems would be worst. Nevertheless, the evaluation data were recorded with very high reliability and gave little indication of any ill effects of the tobacco ban among the inpatients of greatest concern—maximum security forensic patients. The results indicated benefits. The results also suggested that more whole-hearted implementation of the tobacco ban throughout the facility would have achieved similar benefits for all patients. The results support tobacco ban policies in all residential mental health facilities to limit the harm to patients. The authors encourage combining effective antitobacco policies with psychosocial interventions to help mentally ill smokers quit.33,34 The advantages of building rigorous empirical evaluations into all health policy initiatives from the outset were also clear.


A 7-week strike by frontline clinical staff and support workers ended exactly 1 year before the implementation of the tobacco ban. Perhaps the spikes in these measures were related to the disruptions caused by the strike and its immediate aftermath.


Several hypotheses might be entertained—perhaps, dose increased because patients’ weight increased, or perhaps physicians increased doses in anticipation of the tobacco ban because they predicted that patients would experience withdrawal-related agitation, even though the present data indicate that this increase in agitation did not occur. Future studies are encouraged to test these possibilities.


A subsidiary examination for short-term patients coded every incident report on the nonforensic ward with the most acute and transient patients 1 year before and 1 year after the tobacco ban. There were fewer incidents recorded in the year after than in the year before, even though the year after had more patient-days. No trend indicated that the ban was associated with an increase in incidents. Incidents involving assaultive behavior declined, and no incident category increased. Incidents attributed to tobacco increased, but in the context of a greater drop in incidents involving all other causes. No evidence supported the hypothesis that the ban caused a worsening of staff–patient conflict on this short-term ward. More details about this subsidiary test are available from the authors.



The authors gratefully acknowledge the support and assistance of Jennifer Brodeur and Cheryl Kingsmill who collected all the data and the administrators of our institution, especially Russell Fleming. Nancy Bell, Paul Greenall, Martha McDonald, Marc Simpson, Allison Thoms, and Len Wood also provided valuable help and guidance.

Copyright information

© National Council for Community Behavioral Healthcare 2006