Smoking causes almost 90% of deaths from lung cancer, about 80% of deaths from bronchitis and emphysema, and approximately 17% of deaths from heart disease. It increases the risk of developing cancer of the mouth, lip, throat, bladder, kidney, stomach, liver and cervix. ¹

Implementing smoke-free policies in mental health settings presents a challenge. ^{Reference Campion, Lawn, Brownlie, Hunter, Gynther and Pols2} Smoking is more prevalent among people with mental health problems than the general population, ^{Reference Meltzer, Gill, Petticrew and Hinds3} particularly among those with severe mental illness. ^{Reference Coulthard, Farrell, Singleton and Meltzer4} Smokers with mental health problems are also more likely to smoke heavily and show signs of more severe nicotine dependence. ^{Reference Williams, Ziedonis, Abanyie, Steinberg, Foulds and Benowitz5}

Barriers to reduction of smoking in in-patient psychiatric settings include anxiety that smoking cessation could lead to deterioration in mental health and behaviour. There are high rates of smoking in mental healthcare professionals, ^{Reference Rowe and Clark6} some of whom have negative attitudes towards smoke-free policies, with a significant minority believing that there was some therapeutic benefit from patients being allowed to smoke. ^{Reference Dickens, Stubbs and Haw7}

The Health Act 2006 required that all indoor and substantially enclosed outdoor workplaces and public places in England and Wales became smoke-free by 1 July 2007. Residential mental health settings were given a temporary exemption for 1 year only.

Nottinghamshire Healthcare National Health Service (NHS) Trust introduced a total smoking ban for all mental health units within the Trust on 31 March 2007. This included Rampton Hospital, a high secure, long-stay psychiatric hospital for patients with complex mental health disorders who are a grave and immediate danger to the public or themselves (the majority have committed serious offences). Rampton Hospital was the first high secure hospital in the UK to become smoke-free.

A previous survey showed that approximately 70% of patients in Rampton Hospital were smokers and most were heavy smokers. ^{Reference Cormac, Ferriter, Benning and Saul8} Three months before the hospital became smoke-free, patients were informed about the new smoke-free policy and were offered help to stop smoking, including nicotine replacement therapy. Extensive preparations were implemented before the ban including staff training and information for both staff and patients. Individual support was provided for patients by nursing staff. It was made clear that the ban would be universal and apply to staff in the workplace as well as patients. These were among the salient factors that were later to be identified by Campion et al ⁹ as crucial for the successful implementation of a smoke-free policy.

When the smoke-free policy was introduced, tobacco and ignition sources became prohibited for patients, staff and visitors, and this was rigorously enforced in the hospital. On the weekend of the policy introduction, all wards were fully staffed and additional activities were provided as a distraction.

We report a retrospective evaluation of the impact of the smoking ban. The evaluation met the criteria for a service evaluation as defined by the National Patient Safety Agency ^{Reference Cormac, Ferriter, Benning and Saul8} as it was designed and conducted solely to define or judge current care. It involved analysis of existing data and the use of simple questionnaires. The aim was to evaluate the impact of a total smoking ban in buildings and grounds in a high secure psychiatric hospital.

Method

The total ban in Rampton Hospital was implemented from midnight on 31 March 2007. We collected data before and after the ban on:

1. • survey of staff and patients

2. • untoward incidents and seclusions

3. • changes in psychotropic medication

4. • nicotine replacement therapy.

Results

There were 340 resident patients in the hospital during the 8-month evaluation period. Of these, 42 patients were transferred elsewhere or were admitted to the hospital during this period. A total of 298 patients were resident in the hospital for the whole 8-month evaluation period, 217 of whom were smokers before the ban (72.8%). Only seven patients quit smoking in anticipation of the implementation of the smoke-free policy. Patients either made unsuccessful quit attempts or made no quit attempts before the policy was introduced.

Untoward incidents

Table 1 shows the number of incidents for the four time periods, for smokers, non-smokers and all patients. Table 2 shows the number of seclusions as a result of severe untoward incidents.

Table 1 Violent incidents in pre-ban smokers and non-smokers before and after the ban

	December 2006	March 2007	April 2007	July 2007
Pre-ban smokers
Self-harm	48	61	61	60
Verbal abuse	84	95	85	99
Physical aggression	25	22	30	34
Damage to property	1	2	2	5
Total	158	180	178	198
Pre-ban non-smokers
Self-harm	27	21	20	26
Verbal abuse	46	32	32	33
Physical aggression	38	56	58	22
Damage to property	1	1	2	2
Total	112	110	112	83
All patients
Self-harm	75	82	81	86
Verbal abuse	130	127	117	132
Physical aggression	63	78	88	56
Damage to property	2	3	4	7
Total	270	290	290	281

Table 2 Episodes of seclusion of pre-ban smokers and non-smokers before and after the ban

	December 2006	March 2007	April 2007	July 2007
Pre-ban smokers
Threatening behaviour	20	17	26	40
Attacking staff	4	6	7	8
Attacking fellow patient	2	4	5	6
Total	26	27	38	54
Pre-ban non-smokers
Threatening behaviour	9	23	19	6
Attacking staff	3	5	6	8
Attacking fellow patient	0	1	1	1
Total	12	29	26	15
All patients
Threatening behaviour	29	40	45	40
Attacking staff	7	11	13	16
Attacking fellow patient	2	5	6	7
Total	38	56	64	63

Analyses were carried out using χ²-test of paired results, comparing the months of March and April 2007, December 2006 and July 2007, for both pre-ban smokers and non-smokers. Only one result was significant - the comparison of number of incidents for December and July. This showed significantly more incidents for pre-ban smokers in July than in December (P = 0.01, d.f. = 1). There were no significant results for comparisons of numbers of seclusions between pre-ban smokers or non-smokers for either time period comparison.

The hospital security department reported finding contraband tobacco and ignition sources on seven occasions after the ban. However, a full search of all patient areas was not carried out at the point of introduction of the ban and the majority of patients were simply asked to voluntarily surrender their smoking materials.

An analysis of the fire alarm record showed no incidents attributable to illicit smoking between December 2006 and July 2007 inclusive. There were no incidents of major indiscipline such as riots, hostage-taking or rooftop protests in the 4 months following the ban. Three patients sought to challenge the legality of the ban in the courts but this challenge has been rejected.

Changes in psychotropic medication

A comparison was made of pre-ban smokers' and non-smokers' use of four classes of psychotropic medication: regular antipsychotic medication, regular benzodiazepines, PRN antipsychotic medication, and PRN benzodiazepines. We compared March with April and December with July, using the t-test. Table 3 shows the results of these analyses in terms of mean doses for smokers and Table 4 shows the results for non-smokers. As described earlier, the unit of calculation is the dose of medication expressed as a percentage of the maximum BNF limit. The only statistically significant result (P=0.025) was a decline in mean dose of regular antipsychotic medication in smokers from March to April.

Table 3 Pre-ban smokers: comparison of mean dose of medication in the two study periods

	Patients on medication n	Mean (s.d.)	s.e.	95% CI	t	d.f.	P a
Regular antipsychotics
March	166	64.06 (39.38)	3.06	0.37 to 5.42	2.27	165	0.025
April	166	61.16 (37.37)	2.90
Regular benzodiazepines
March	8	66.79 (80.37)	28.41	−12.57 to 16.08	0.29	7	0.78
April	8	65.03 (89.62)	31.68
PRN antipsychotics
March	27	2.76 (2.70)	0.52	−0.68 to 0.98	0.37	26	0.71
April	27	2.61 (2.16)	0.41
PRN benzodiazepines
March	21	8.55 (7.84)	1.71	−10.34 to 1.28	−1.63	20	0.12
April	21	13.08 (16.77)	3.66
Regular antipsychotics
December	158	64.96 (38.99)	3.10	−0.8 to 7.86	1.93	157	0.055
July	158	61.07 (36.86)	2.93
Regular benzodiazepines
December	8	51.38 (46.65)	16.49	−51.73 to 18.90	−1.10	7	0.31
July	8	67.80 (84.94)	30.03
PRN antipsychotics
December	14	1.83 (1.26)	0.33	−1.74 to 0.81	−0.79	13	0.44
July	14	2.29 (2.28)	0.61
PRN benzodiazepines
December	9	10.78 (7.43)	2.48	−14.70 to 8.64	−0.60	8	0.57
July	9	13.81 (14.41)	4.80

Table 4 Pre-ban non-smokers: comparison of mean dose of medication in the two study periods

	Patients on medication n	Mean (s.d.)	s.e.	95% CI	t	d.f.	P a
Regular antipsychotics
March	56	62.90 (40.21)	5.37	−0.81 to 0.72	−0.13	55	0.90
April	56	62.95 (40.37)	5.39
Regular benzodiazepines
March	3	52.04 (41.87)	24.17	−3.32 to 5.34	1.00	2	0.42
April	3	51.03 (42.94)	24.79
PRN antipsychotics
March	4	1.18 (1.01)	0.50	−7.66 to 0.68	−2.66	3	0.08
April	4	4.67 (2.78)	1.39
PRN benzodiazepines
March	6	9.32 (8.08)	3.30	−7.57 to 0.56	−2.22	5	0.08
April	6	12.82 (9.86)	4.02
Regular antipsychotics
December	53	63.32 (38.57)	5.30	−5.07 to 1.16	−1.26	52	0.21
July	53	65.28 (41.01)	5.63
Regular benzodiazepines
December	3	56.06 (38.06)	21.97	−26.99 to 43.34	1.00	2	0.42
July	3	47.88 (46.68)	26.89
PRN antipsychotics
December	3	3.40 (4.07)	2.35	−10.62 to 16.85	0.71	2	0.55
July	3	1.29 (1.69)	0.98
PRN benzodiazepines
December	4	11.49 (9.73)	4.86	−9.69 to 23.40	1.32	3	0.28
July	4	4.64 (1.37)	0.69

Discussion

The implementation of a smoke-free policy in Rampton Hospital provided a rare opportunity to evaluate the impact of a total smoking ban in a highly controlled environment. Unlike other health facilities in the UK, it was not possible for patients to go outside the buildings or grounds to smoke, because of security restrictions. As a high secure hospital, effective search policies and procedures were already present to prevent contraband items entering the hospital and this was extended to tobacco and smoking-related materials.

However, owing to the opportunistic nature of this evaluation, there were limits to the data that were available for evaluation; moreover, data were available only for four time periods. The statistically significant result for the comparison of December and July incidents may be an artefact of a potentially seasonal drop in incidents in the period before Christmas.

We are not in a position to say whether any patients were transferred or discharged during the study period for reasons connected with the smoking ban but we consider the number to be few, if any. Wishing to smoke tobacco or deterioration in behaviour would not normally be considered a sufficient reason for leaving the hospital.

Staff concerns about patients' aggression, self-harm and increased need for medication decreased after the ban, but were consistently higher among nurses. A major problem with these data is that because the questionnaires were anonymous it was not possible to link the pre-ban responses to the post-ban responses for either patients or staff. It is possible that those staff members and patients who responded to the pre-ban questionnaire represented biased samples (i.e. of those who were more hostile to the ban), whereas those who responded to the post-ban questionnaire were more pro-ban. This potential source of bias could account for any changes in favour of the ban.

Despite the fears of staff and patients, the transition from a smoking environment to a non-smoking environment went smoothly. Our results do not show a marked increase in use of psychotropic medication, self-harm or behavioural disturbance as a result of the smoking ban. This evaluation replicates the findings of Hempel et al, ^{Reference Hempel, Kownacki, Malin, Ozone and Cormack11} who reported a similar uneventful transition to a non-smoking environment in a maximum security hospital in the USA. However, in another paper ^{Reference Cormac, Brown, Creasey, Ferriter and Huckstep12} we describe one of the few potential problems of abrupt smoking cessation, which was an increase in clozapine plasma levels for individuals who quit smoking.

Cormac & McNally ^{Reference Cormac and McNally13} have described the strategy used to implement the total ban on smoking, which was applied at Rampton Hospital. This strategy could be applied in other psychiatric settings. What is now required is a long-term evaluation of the health benefits of smoke-free environments to patients in Rampton Hospital and other long-stay NHS facilities.

This evaluation supports taking a robust approach to tackling smoking in psychiatric settings and should give confidence to those considering the implementation of a total rather than partial smoking ban in their psychiatric service.