Prevalence and Predictors of Smoking by Inpatients During A Hospital Stay

Abstract

Background

Accredited U.S. hospitals prohibit smoking inside hospital buildings. Patients are expected to abstain from smoking throughout their hospitalization but how many do is unclear. Smoking by inpatients may compromise patient safety, hospital inefficiency, and clinical care.

Method

We conducted an observational study of adult cigarette smokers visited while hospitalized by a tobacco counselor and reached for telephone follow-up in the 2 weeks after discharge. We assessed smoking during the hospital stay at the time of counseling for all patients and at follow-up for those reached. We used generalized linear models to estimate adjusted relative risk (ARR) for smoking while hospitalized adjusted by patient and admission characteristics.

Results

Over the 3 year study period (2007-2010), 5,399 smokers were visited by counselors, of whom 14.9% had smoked between admission and the visit. Of 3549 eligible and consented to follow-up, 2185 were reached. Smoking at any time during the hospitalization, was reported by 18.4%, less often during winter months than the rest of the year (14% vs. 20%, p=0.007). Smoking at any time while hospitalized was less common among those aged ≥50 (ARR: 0.72 ; 95% CI: 0.61-0.86), admitted to a cardiac floor (ARR: 0.61; 95% CI: 0.48-0.77), and intending to quit after discharge (ARR: 0.60; 95% CI: 0.50-0.72), and more common among those with longer stays (ARR: 1.32 ; 95% CI: 1.11-1.57), and experiencing cigarette cravings (slight: ARR: 1.37; 95% CI: 1.10-1.72; moderate: ARR: 2.20; 95% CI: 1.79-2.70). NRT ordered the day of admission was associated with less smoking before the counseling visit (ARR: 0.86 ; 95% CI: 0.75-0.99), but not for the entire hospital stay.

Conclusions

Nearly one fifth of smokers admitted to a smoke-free hospital smoked during their hospital stay. Ordering NRT routinely at admission and routinely, monitoring patients’ cigarette cravings might reduce smoking among admitted patients.

Introduction

The Joint Commission requires accredited U.S. hospitals to have a policy prohibiting smoking in hospital buildings, and hospitals have nearly universally adopted smoke-free building policies.^1,2 However, the requirement does not extend to the hospital campus, and in the majority of hospitals where a campus-wide smoking ban is not in place³, hospitalized smokers may travel outside the hospital if they wish to smoke. The fact that patients may go outdoors to smoke, especially without supervision or in inclement weather, raises safety concerns. Quality of care and hospital efficiency could be compromised if patients are not available for examinations or treatments.⁴ Patients who smoke during a hospitalization may experience direct negative consequences such as delayed recovery from cardiopulmonary disease or impaired wound healing,⁵ and certainly they deprive themselves of an opportunity to initiate a quit attempt in a supportive, smoke-free environment.

Few recent studies have evaluated patients’ behavior under smoke-free building policies except in hospitals outside of the U.S. or among special populations.^6-9 Two studies in U.S. hospitals, both completed over 10 years ago, reported that a quarter of smokers admitted to a general hospital smoked during their stay. ^10,11 Since then, the proportion of smokers who smoke during a hospital stay may have fallen as a result of declining population smoking prevalence,¹² clinical practice guidelines recommending the use of nicotine replacement therapy (NRT) to relieve nicotine withdrawal symptoms,¹³ and increasingly common hospital smoking bans.

We conducted an observational study of smokers admitted to a large urban teaching hospital with a smoke-free building policy. Our aim was to assess the proportion of smokers who continue to smoke during their admission and to identify patient and admission characteristics, including NRT use, associated with smoking while hospitalized.

Methods

Setting and Participants

The study was conducted at Massachusetts General Hospital (MGH), a 900-bed urban teaching hospital affiliated with Harvard Medical School and was approved by the Internal Review Board of Partners Healthcare System of which MGH is a member. MGH bans smoking in all indoor areas and on the outdoor campus except in two outdoor shelters which patients can use. Patients’ smoking status is routinely recorded via admission and post-operative order templates in the computerized physician order entry system used by the medical, surgical, neurology and gynecology services. The templates also facilitate the ordering of NRT. The system generates an automatic electronic referral to the MGH Tobacco Treatment Service (TTS). In addition, staff may make referrals at any time during an admission electronically or by telephone. TTS counselors attempt to visit all smokers at the bedside to assist in managing nicotine withdrawal. Pharmacotherapy, typically nicotine replacement therapy, is available to admitted smokers to manage cravings. All patients receive brief advice to quit (≤5 minutes). Those considering quitting receive cessation counseling (averaging 20 minutes) employing motivational interviewing and covering behavioral strategies and use of medication to remain quit. Counselors may recommend varenicline and bupropion to patients who are committed to remaining quit after discharge. Use of these medications may be initiated during the admission but, unlike NRT, their effect on cravings is typically delayed and unlikely to reduce the urge to smoke during the stay.

Patients were enrolled in the study over the three year period from May 2007 through April 2010. Enrolled patients were contacted by phone in the two weeks after discharge to determine if they had smoked during their admission. Patients were eligible for the study if they received counseling from a TTS counselor during their stay, reported having smoked cigarettes within the month prior to admission, had a telephone where they could be reached, were discharged to home, had no communication barriers (e.g. deafness, non-English-speaking) or health issues precluding participation (e.g. metastatic disease, palliative care). Counselors asked eligible patients to consent to participation at the end of the counseling session.

Measures

TTS counselors recorded patients’ average daily cigarette consumption in the month prior to admission when smoking normally (i.e. not too ill to smoke or cutting down temporarily for a planned admission, cig/day), elapsed time since last cigarette smoked, cigarette craving (categorized as none, slight, mild, moderate or severe) and intention to smoke after discharge (“I will stay quit,” “I will try to quit,” “I do not know,” or “I do not plan to quit”) at the time of their inpatient visit. At the post-discharge call, patients were asked if they had smoked any cigarettes during their hospital stay. Patients’ age, sex, date of admission, admitting service, length of stay (LOS), and pharmacy orders for NRT were obtained from hospital records.

The outcome of interest, smoking while hospitalized (SWH), was determined from data collected at the TTS counselor visit and at the post-discharge telephone call. Smoking early in the hospitalization (early SWH) was assessed at the TTS visit by comparing the patient's report of the time elapsed since their last cigarette to the time elapsed between the visit and hospital admission. Patients who last smoked on the day of admission were not considered to have smoked while hospitalized because it could not be determined whether smoking occurred after the admission began. Smoking at any time during hospitalization was assessed in follow-up telephone survey at two weeks of discharge. Patients were defined as having smoked at any time during the admission if (1) they reported during their inpatient assessment that their last cigarette was smoked on day two or later of their hospital stay or (2) they reported smoking while hospitalized at the follow-up assessment.

Analysis

We calculated the prevalence and 95% confidence intervals (CI) of early SWH for all patients counseled by the TTS and smoking at any time during the admission for those reached for follow up. Logistic regression was used to test for linear trends in prevalence by study year. Characteristics of patients who did and did not smoke in the hospital were compared in bivariate analyses using t-tests and Wilcoxon rank-sum tests for continuous variables and chi-squared tests for categorical variables. We used generalized linear models to explore associations between patient and admission characteristics and early SWH and smoking at any time during the admission. The models employed a Poisson distribution, log link function and robust standard errors, and included age (<50 years old vs. ≥50 years old), sex, baseline smoking level (<10 cig/day vs. ≥10 cig/day), cigarette craving (categorized slight, mild moderate or severe with none as the reference group), admission service (cardiac vs. non-cardiac), intention to quit after discharge (“I will quit” , “I will try to quit” , or “I don't know” with “I do not plan to quit” as the reference group) and NRT ordered on admission. We included a term for the winter months (December, January and February) to assess whether colder outdoor temperatures might discourage patients from venturing outside to smoke. The models also included terms for duration of hospitalization at the time of assessing SWH, dichotomized at the median. In the model predicting smoking at any time during the stay, this term was based on total LOS (<5 vs. ≥ 5 days). The model predicting early SWH included a term for days between admission and the visit (<3 vs. ≥ 3 days). We present relative risks adjusted for all terms in the models (ARR) and 95% CI. Results were evaluated using the p<0.05 criterion for statistical significance. All analyses were conducted using Stata statistical software (StataCorp. 2007. Stata Statistical Software: Release 10. College Station, TX: StataCorp LP.)

Results

During the study period, 171,879 were admitted for an overnight stay, of whom 63,899 were admitted using a template prompting smoking status identification. Smoking status was identified for 54,166 (85%), of whom 11,439 (21%) were smokers and were automatically referred to the TTS. A total of 12,771 referrals were received by the TTS, including 1,332 consult requests. Counselors were able to see 8,358 (65%) patients: 2,361 received withdrawal assistance and brief advice only; 598 were not current cigarette smokers; and 5,399 received cessation counseling (Figure 1). Of the smokers counseled by the TTS, 80% (4,295/5,399) were eligible for follow-up, and 83% (3,555/4,295) consented. The follow up response rate was 61% (2,185/3,555) of those consenting to follow up and 40% (2,185/5,399) of those counseled. Patients had a mean age of 53 years and the majority (58%) were male (Table 1). The median LOS was 5 days, and the median day of stay for the TTS visit was day 3. Most patients (62%) received NRT during hospitalization and a third received it on the first day of their stay. NRT orders were more common among those smoking ≥10 cig/day on admission (36% vs. 22%, p<0.001) and throughout the stay (68% vs. 43%, p<0.001).

Figure 1.

Study Flow Diagram

Characteristic	Counseled during admission				Reached for follow up 2 weeks after discharge
	Total	Smoked before counseling session		P	Total	Smoked during admission		P
	N=5399	Yes N=807	No N=4592		N=2185	Yes N=403	No N=1782
Age, mean (SD)	53 (14.0)	48 (13.6)	54 (13.9)	<0.001	54(13.0)	50 (12.9)	55 (12.9)	<0.001
Male, n (%)	3145 (58.2)	456 (56.5)	2689 (58.6)	0.28	1211 (55.4)	154 (55.8)	1057 (55.4)	0.89
Cig/day, mean, median (IQR)	18, 20 (10-20)	20, 20 (10-20)	18, 20 (10-20)	<0.001	18, 20 (10-20)	20, 20 (10-25)	18, 20 (10-20)	<0.001
Craving severity, n (%)				<0.001				<0.001
None	2941 (54.5)	312 (38.7)	2629 (57.2)		1275 (58.4)	170 (42)	1105 (62)
Slight/mild	1377 (25.5)	214 (26.5)	1163 (25.3)		548 (25.1)	107 (26.6)	441 (24.8)
Moderate/severe	1079 (20.0)	280 (34.7)	799 (17.4)		362 (16.6)	126 (31.3)	236 (13.2)
Intention to quit after discharge, n (%)				<0.001				<0.001
No or don't know	1273 (23.6)	284 (35.2)	989 (21.5)		310 (14.2)	100 (24.8)	210 (11.8)
Yes, or will try	3997 (74.0)	496 (61.5)	3501 (76.2)		1852 (84.8)	297 (73.7)	1555 (87.3)
Unknown	129 (2.4)	27 (3.3)	102 (2.2)		23 (1.0)	6 (1.5)	17 (1.0)
Length of stay, days, mean, median(IQR)	6.9, 5 (3-8)	8.3, 6 (4-10)	6.7, 5 (3-8)	<0.001	6.0, 5 (3-7)	6.8, 5 (3-8)	5.8, 4 (3-7)	<0.001
Admitted to cardiac service, n (%)	1461 (27.1)	133 (16.5)	1328 (28.9)	<0.001	757 (34.6)	83 (20.6)	674 (37.8)	<0.001
NRT ordered, n (%)
On admission	1773 (32.8)	243 (30.1)	1530 (33.3)	0.07	716 (32.8)	130 (32.3)	586 (32.9)	0.81
At any time during stay	3370 (62.4)	549 (68.0)	2821 (61.4)	<0.001	1346 (61.6)	272 (67.5)	1074 (60.3)	0.007

At the time of TTS visit, 14.9% (95% CI: 14.0%-15.9%) of patients reported smoking since hospital admission. Among those reached at follow up, 18.4% (95% CI: 16.8-20.0%) reported smoking at any time during their admission. Over the three year study period, prevalence of smoking before the TTS visit declined slightly from 16.0% (95% CI: 14.3-17.7%) to 14.9% (95% CI: 13.3-16.5%) in year 2 and 13.8% (95% CI: 12.1-15.5%) in year 3, but the trend did not reach statistical significance (p=0.08), and no similar reduction was observed among the patients reached for follow up (Figure 1). Patients admitted during winter months (December, January and February) were less likely to smoke at any time during their admission than those admitted during the rest of the year (14% vs. 20%, p=0.007).

Table 1 presents characteristics of patients who had and had not smoked by the time of the TTS visit and of patients who were reached for follow up who did and did not smoke at any time during the hospitalization. Those who reported smoking early in the admission were younger, smoked more cigarettes per day, reported more severe cravings, were less likely to report they will quit or try to quit on discharge, had longer hospital stays, were less likely to be admitted to the cardiac service, and more likely to receive NRT during the stay than those who had abstained up to the TTS visit. Among those eligible for follow up, patients who were reached for follow up were less likely to have reported smoking at the time of the counseling session than telephone nonrespondents (13% vs. 16%, p<0.001). All characteristics associated with having smoked at the time of the TTS visit were similarly associated with smoking at any time during the stay.

The generalized linear models are presented in Figure 2. Four percent of patients (N=207) were missing a data element and were excluded from these analyses. Early SWH (Figure 2, left panel) was more common among heavier smokers (ARR: 1.43 ; 95% CI: 1.17-1.73), and those with stronger cravings (moderate: ARR: 1.22; 95% CI: 1.15-1.29; severe: ARR: 1.15; 95% CI: 1.09-1.21), and less common those 50 or older (ARR: 0.68; 95% CI: 0.60-0.78), intending to quit (will try to quit: ARR: 0.85; 95% CI: 0.78-0.92; will quit: ARR: 0.46; 95% CI: 0.37-0.58), admitted to the cardiac service (ARR: 0.80; 95% CI: 0.67-0.96) and receiving NRT the day of admission (ARR: 0.83 ; 95% CI: 0.72-0.96). Patients who had reached day 3 or later of their stay before being seen by the TTS (longer stay) were more than twice as likely to report having smoked than those seen earlier in the stay (ARR: 2.39; 95% CI: 2.06-2.78).

Figure 2.

Patient and Admission Characteristics Associated with Smoking While Hospitalized

Note: *Longer stay is defined as counseled on day 3 or later of hospitalization (left panel) or total length of stay ≥ 5 days (right panel)

SWH at any point during hospitalization (Figure 2, right panel) was more frequent in those with stronger cravings (moderate: ARR: 1.23; 95% CI: 1.14-1.33; moderate: ARR: 1.25; 95% CI: 1.18-1.34), and longer stays (ARR: 1.36 ; 95% CI: 1.14-1.62), and less frequent in older patients (ARR: 0.74 ; 95% CI: 0.62-0.88), those who intend to quit (will try to quit: ARR: 0.81; 95% CI: 0.71-0.91; will quit: ARR: 0.46;95% CI: 0.34-0.63) and those admitted to a cardiac service (ARR: 0.64; 95% CI: 0.51-0.81). In contrast to early SWH, SWH at any point was less frequent during the winter months (ARR: 0.76; 95% CI: 0.61-0.96) but was not associated with receiving NRT on admission or heavy smoking.

Discussion

In a large sample of smokers admitted to a smoke-free hospital from 2007 to 2010, 18% reported having smoked during their hospital stay. The majority of smokers received NRT. In multivariate models, patients were more likely to report having smoked while hospitalized if they were younger, had more severe cigarette cravings, did not report planning to quit, had been in the hospital longer, and were not admitted to a cardiac unit. Smoking during the first few days of the stay was more common among heavy smokers in both unadjusted and adjusted analyses. Receipt of NRT on admission was associated with a lower rate of smoking early in the stay, however this effect was not seen in smoking at any point during the hospitalization.

Our results indicate a decrease in the rate of SWH when compared to a previous study¹¹ of inpatient smokers at the same hospital which found 25% smoked during their stay in 1995. The rate of NRT use among inpatients at this hospital rose dramatically over this period from 5% in 1995¹⁴ to 62% in the present study, and this change may be responsible in part for the decline in SWH. An order for NRT on admission was associated with less smoking when assessed at the time of the counselor visit, indicating that initiating NRT use early might reduce patient desire to leave the hospital to smoke. However, this effect was not seen when patients were reached at follow up and could report on their entire hospitalization, suggesting that the benefit of NRT might be only to delay but not prevent smoking. Patients with longer stays might benefit from increasing nicotine dose or supplementation of patch with shorter-acting forms of NRT (e.g. gum or lozenge).

Admitting staff were more likely to order NRT on admission for heavier smokers, which may reflect a reasonable expectation that these patients were especially likely to experience withdrawal. We found that the intensity of cigarette cravings was more strongly associated with smoking while hospitalized than the number of cigarettes smoked per day. It may be difficult to predict the intensity of cravings during an admission from pre-admission smoking level due to individual variability in response to illness and the hospital environment.¹⁵ Routine ongoing assessment of cigarette craving, while more time-consuming, might be a more effective means of identifying patients who will have difficulty remaining abstinent during their hospital stay. It could also be helpful in adjusting the dose for patients already receiving NRT.

Extending hospital smoking bans to encompass all outdoor areas may discourage inpatients from smoking and such bans are increasingly common. As of February 2008, 45% of hospitals nationwide reported implementing campus-wide smoke-free policies.¹⁰ However, campus-wide bans alone, in the absence of a clinical policy that prohibits inpatients from leaving the hospital to smoke, have not been sufficient to prevent smoking by inpatients.³

There are several limitations to this study. It was an observational study conducted in a single urban hospital in the state of Massachusetts. The results may be influenced by the state's low prevalence of smoking or its weather patterns, or the physical setting of the hospital and its outdoor smoking shelters. Because discharge diagnoses were not available, we were not able to account for seasonal changes in case-mix that might account for the observed reduction in smoking during the winter. Our data on NRT orders after admission do not allow us to determine reliably whether NRT was first offered before or after the patient left to smoke. The estimated rates of smoking during hospitalization may be biased downward for 2 reasons. First, we based our estimate on patient self-report which is subject to patient reluctance to disclose or failure to recall smoking as an inpatient. Second, our estimate of smoking at any time during hospitalization relied in part on telephone follow up. The follow up had a relatively low response rate (40% of those counseled were eligible, consented and responded) and may have been subject to nonresponse bias: respondents differed from nonrespondents in that they were less likely to have reported smoking by the time of the TTS counseling visit. Because our multivariate analysis was exploratory and well powered by our large sample size, the results should be interpreted with caution.

Conclusion

The 1992 Joint Commission regulations requiring hospitals to prohibit smoking in indoor areas led to the adoption of smoke-free policies in hospitals nationwide. This in turn encouraged temporary abstinence among a large number of smokers. Nonetheless, in 2007-2010, nearly one in five smokers admitted to one smoke-free hospital smoked during their hospital stay. This proportion is not much lower than that reported a decade ago, despite a hospital policy that now limits smoking to two outdoor shelters and the substantial use of NRT to treat nicotine withdrawal symptoms in hospitalized smokers. Assessment of cigarette cravings, especially among younger smokers and those who do not plan to quit after discharge, could identify high-risk patients. The routine order of NRT on admission and the expansion of smoke-free policies to cover the entire hospital campus are two strategies that might decrease the proportion of smokers who smoke while hospitalized. This could improve patient safety, hospital efficiency, and clinical outcomes for hospitalized smokers.