The Effectiveness of a Telephone Smoking Cessation Program in Mental Health Clinic Patients by Level of Mental Well-Being and Functioning: A Secondary Data Analysis of a Randomized Clinical Trial

Background Few studies have examined the effectiveness of telephone smoking cessation interventions by severity of behavioral health symptoms. Using data from a telephone counseling study, we examined whether abstinence rates varied by level of behavioral health symptoms. Methods The parent study recruited adults who smoke cigarettes (N = 577) referred by mental health providers at six Veterans Health Administration facilities. Participants were randomized to specialized telephone counseling (intervention) or state Quitline referral (control). Participants completed assessments at baseline and 6 months, including the BASIS-24, a self-report measure of behavioral health symptoms and functioning. We used the BASIS-24 median to dichotomize participants as having high or low scores. The primary outcome was 30-day self-reported abstinence at 6 months. We compared groups on outcomes by logistic regression and performed an interaction effect analysis between treatment assignment and groups. Results At baseline, those with high behavioral health symptoms scores reported heavier nicotine dependence and more sedative and/or antidepressant use. At 6 months, participants with low behavioral health symptoms scores in the intervention reported higher rates of 30-day abstinence compared to those in the control arm (26% vs 13%, OR = 2.3, 95% CI = 1.8, 2.9). People with high behavioral health symptoms scores reported no difference in 30-day abstinence between the treatment assignments at 6 months (12% vs. 13%, OR = 1.1, 95% CI = 0.6, 2.0). Conclusions Only participants with low behavioral health symptoms scores reported higher abstinence rates in the intervention compared to the state Quitline. Future research can examine alternative approaches for people with worse mental well-being and functioning. Trial registration The parent study is registered at www.clinicaltrials.govNCT00724308.


INTRODUCTION
Smoking remains a leading cause of preventable death.People with mental illness smoke 40% of all cigarettes in the United States and experience excess smoking-related mortality and morbidity. 1,2servational research has established that people with serious mental health problems, including those with serious mental illness (SMI, e.g.4][5][6][7] Heavier nicotine dependence, 8 lower distress tolerance, 9 and acute psychological distress reactivity 10 have been proposed mechanisms underlying di culty quitting and higher rates of relapse among people with mental illness who smoke.
Contrary to prevalent beliefs among health professionals, people with mental health disorders can and are motivated to quit. 11Cessation likely does not negatively impact mental health outcomes and may in fact improve psychological symptoms. 12,13Thus, determining effective cessation interventions for people who smoke with worse mental well-being and functioning should be a high priority for clinicians and public health researchers.
9][20][21] However, the question remains whether the characteristics of the telephone treatment received may impact abstinence outcomes differently depending on the extent of a person's reported mental well-being and functioning.Are all forms of telephone counseling equally effective for all mental health patients?This secondary data analysis of a telephone care coordination program for people who smoke receiving mental health services examines whether abstinence outcomes differed by Behavior and Symptom Identi cation Scale (BASIS-24) scores, which measure levels of mental well-being and functioning among populations in mental health clinics, as well as others. 22

METHODS
The TeleQuit Mental Health study tested a specialized, multi-session telephone smoking cessation program for mental health patients who smoke (N = 577) referred by mental health providers at six Veterans Health Administration (VA) facilities in the Northeast. 23,24Any patient seen in the mental health clinics was eligible for referral via electronic medical record (EMR) consult, a 10-15 second process linked to a tobacco use clinical reminder.Providers were encouraged to refer all people who smoke regardless of desire to quit.Other eligibility criteria included having access to a telephone, having a mailing address, and having smoked cigarettes in the past 30 days.Referred patients were mailed a welcome packet with information about the study.Study staff called patients up to 5 times to screen and enroll participants.Participants were randomized to receive either a) multi-session telephone cessation counseling designed for patients with a mental health diagnosis or b) warm transfer via three-way call to the Quitline in their state.All participants were offered cessation medications (nicotine replacement therapy [NRT] or buproprion) and mailed a self-help educational packet.
Participants completed assessments at baseline, 2 months, and 6 months.6][27][28] BASIS-24 is not a direct measure of either serious mental illness or psychological/mental distress.BASIS-24 is a measure of mental well-being and functioning among populations in mental health clinics.It broadly measures behavioral health and functioning across six subscales: Depression/functioning, interpersonal relationships, psychosis, substance abuse, emotional lability, and self-harm.While there are no cutoff scores or quartiles used to categorize patients using BASIS-24 scores in the literature, having higher scores likely indicates worse mental well-being and functioning within a clinical patient population. 29us, we categorized participants scoring at or above the overall BASIS-24 score median as having worse overall mental health well-being and functioning or a high behavioral health symptoms score (n = 264) and those scoring below the median as having better overall mental health well-being and functioning or a low behavioral health symptoms score (n = 263).As psychotic disorders partially comprise serious mental illnesses, we additionally ran analyses using the BASIS-24 psychosis subscale median as the cutoff score and found similar results for primary and secondary outcomes (data not shown).We chose the analyses using the overall BASIS-24 score as the cutoff with the understanding that smoking abstinence outcomes for participants with high BASIS-24 scores may reasonably re ect outcomes for patients with serious mental health problems.
We examined 30-day self-reported smoking abstinence at 6-months as our primary outcome.Nonrespondents at 6 months were treated as smokers.We had three secondary outcomes: engagement with cessation treatment at 6 months, which included receipt of telephone counseling, self-reported use of cessation medications, and a 24-hour quit attempt.For secondary outcomes, we treated non-respondents as not having engaged with cessation treatment at 6 months (i.e., no telephone counseling, not actively using cessation medications, and no ≥ 24 hour quit attempts made).
We compared demographic, health, and smoking characteristics at baseline between the two groups, de ned as high vs. low scores of behavioral health symptoms.Wilcoxon's rank-sum test was used for continuous variables and Chi-square or Fisher's exact test for categorical variables.Next, logistic regression was used to compare groups on primary and secondary outcomes, adjusting for baseline cigarettes per day and clustering by site.For each outcome, we tested for an interaction effect between treatment assignment and behavioral health symptoms score category (high vs. low).Finally, we performed multivariable logistic regression to determine the factors associated with high behavioral health symptoms and functioning scores, and report ndings as adjusted odds ratios (AOR) with 95% con dence intervals (CIs).All statistical tests were 2 sided with P < .05indicating statistical signi cance.Analyses were performed with SAS, version 9.4 (SAS Institute Inc).

RESULTS
We describe demographic, health, and smoking characteristics by behavioral health symptoms score category at baseline in Table 1 (see end of document).Participants with high vs. low scores of behavioral health symptoms were similar in most demographic characteristics.The study population across both groups was predominantly male, mostly Black or white, and not Hispanic/Latino, with a mean age of 53 years old.In both groups, participants smoked an average of 16 cigarettes per day and were "very motivated" to quit smoking.However, compared to those with low behavioral health symptoms scores (i.e., better mental well-being and functioning), those with high scores were signi cantly more likely to have heavier nicotine dependence (% reporting smoking within 5 minutes of waking in the morning; 40% vs. 27%), were more likely to report current sedative/sleeping pill use (18% vs. 8%, p < 0.01) and current antidepressant use (53% vs. 35%, p < 0.0001).Regarding comorbidities, we found that participants reporting high behavioral health symptoms scores were more likely to have angina/prior heart attack (19% vs. 8%, p < 0.01) or seizures, epilepsy, or convulsions (8% vs. 3%, p = 0.02).At 6 months, participants with low behavioral health symptoms scores reported long-term abstinence rates that were not signi cantly different from those with high behavioral health symptoms scores (20% vs 13%), as shown in Table 2a.However, we found an interaction effect between behavioral health symptoms scores and treatment assignment.At 6 months, participants with low behavioral health symptoms scores in the specialized, multi-session telephone smoking cessation program were more likely to report 30-day abstinence compared to participants with low behavioral health symptoms scores referred to a state Quitline (26% vs 13%, OR = 2.3, 95% CI = 1.8, 2.9).Participants with low behavioral health symptoms scores in the intervention arm were also more likely to have made a quit attempt longer than 24 hours at 6 months compared to those who received the state Quitline referral (62% vs 49%, OR = 1.7, 95% CI = 1.2, 2.5) (Table 2b).*P-value for interaction term.Participants were considered abstinent at 6-month follow-up if they reported not having smoked any cigarettes in the prior 30 days.NRT = nicotine replacement therapy.We categorized participants as high vs. low behavioral health symptoms scores based on whether they were below or above the median score on the BASIS-24 at baseline.By contrast, we observed no signi cant difference in 30-day abstinence among participants with high behavioral health symptoms scores between the treatment groups at 6 months (12% vs. 13%, OR = 1.1, 95% CI = 0.6, 2.0).There was also no difference between treatment arms among people with high behavioral health symptoms scores in the percent of individuals making at least one ≥ 24 hour quit attempts at 6 months (56% vs. 59%, OR = 0.9, 95% CI = 0.6,1.3).Notably, people with high behavioral health symptoms scores in the intervention arm made greater use of telephone counseling than those in the control arm (51% vs. 33%, OR = 2.3, 95% CI [1.3, 4.3].We did not observe an interaction effect in the relationship between treatment condition and use of telephone counseling or NRT/bupropion at 6 months (Table 2b).
On multivariable analysis (Table 3), we found that Hispanic/Latino ethnicity (OR 1.

DISCUSSION
This study examines the effectiveness of an intensive telephone intervention for smoking cessation compared to referral to the state Quitline among mental health patients who smoke referred by mental health providers at six VA facilities.We found few differences in demographic characteristics at baseline.
Regarding other characteristics assessed at baseline, people with high behavioral health symptoms scores reported heavier nicotine dependence and were more likely to report current use of sedatives, sleeping pills, and/or antidepressants.They were also more likely to report a history of seizures, epilepsy, or convulsions and/or angina or heart attack.We found a signi cant interaction effect between behavioral health symptoms scores and treatment assignment.People with low behavioral health symptoms scores in the specialized counseling arm were signi cantly more likely to report 30-day abstinence at 6 months and to have made a quit attempt compared to people with low behavioral health symptoms scores in the state Quitline counseling arm.People with high behavioral health symptoms scores did not have signi cantly different abstinence outcomes and did not differ in the likelihood of having made a quit attempt at 6 months based on treatment assignment.
There were a few differences in baseline characteristics between the high and low behavioral health symptoms score groups.However, differences in long-term abstinence outcomes at 6 months are unlikely to be attributable to self-reported race, ethnicity, income, educational attainment, motivation to quit, nor number of cigarettes smoked per day.The multivariable analysis suggests that factors associated with having higher behavioral health symptoms and functioning scores may include Hispanic/Latino ethnicity, current antidepressant, cannabis, and/or sedative/sleeping pill use, and a history of heart attack/angina.
[32] The observed signi cant interaction effect between behavioral health symptoms score level and treatment assignment has several implications.It adds nuance to existing research that suggests mental health patients bene t from intensive, specialized telephone interventions. 33,34In our study, we found that particularly people who smoke with low behavioral health symptoms scores bene ted signi cantly more from an approach that included multiple sessions, relapse-sensitive timing, and customization for mental health patients.People with high behavioral health symptoms scores had similarly low abstinence rates regardless of telephone treatment assignment.This suggests people who smoke with high behavioral health symptoms scores may need a different approach.Integrated care, which involves delivering treatment for tobacco use and psychiatric care in a single clinical setting, has been found to be effective for patients with PTSD and may be a model for people who smoke with high behavioral health symptoms scores as well. 35Understanding the nuances in cessation patterns among people who smoke with mental health diagnoses will inform cost-effective, evidence-based public health decisions.Future research can evaluate potential smoking harm reduction strategies such as e-cigarettes among the hard-to-treat population of mental health patients reporting higher behavior health symptoms scores or people with serious mental illness.
There are several limitations to consider.A secondary data analysis is by nature limited by the data collection from the parent study, which relied on self-reported abstinence.BASIS-24 is well-regarded as a measure of behavioral health symptoms and functioning, but a cutoff point for de ning categorically a high behavioral health symptoms score has not been validated in the literature.Thus, our use of the overall BASIS-24 score median as a cutoff score to compare patient groups by their level of mental wellbeing and functioning requires further evaluation longitudinally, and in other clinical and research settings.In addition, only approximately 10% of eligible patients were referred to the parent study, thus raising the question of selection bias among referring providers in favor of people whom they may perceive to be more likely to quit.Finally, the study examined a veteran population that was overwhelmingly male, so the results may not be generalizable to non-veterans and/or females.

Further research to explain
Abbreviations BASIS-24 24-item Behavior and Symptom Identi cation Scale OR Odds ratio CI Con dence interval SMI Serious mental illness PTSD Post-traumatic stress disorder VA Veterans Association EMR Electronic medical record NRT Nicotine replacement therapy Declarations Disclaimer: The views expressed in this article are those of the authors and do not necessarily re ect the position or policy of the Department of Veterans Affairs or the United States government.

Table 1
Baseline Characteristics in Participants With High vs. Low Behavioral Health Symptoms Scores *We used Wilcoxon's rank-sum test for continuous variables and Chi-square or Fisher's exact test for categorical variables Standard Deviations: a 12, b 11, c 11, d 12 *We used Wilcoxon's rank-sum test for continuous variables and Chi-square or Fisher's exact test for categorical variables Standard Deviations: a 12, b 11, c 11, d 12 *We used Wilcoxon's rank-sum test for continuous variables and Chi-square or Fisher's exact test for categorical variables Standard Deviations: a 12, b 11, c 11, d 12

Table 2 a
: Self-Reported Abstinence at 6 months in Participants with High vs. Low Behavioral Health Symptoms Scores *P value for interaction term.Participants were considered abstinent at 6-month follow-up if they reported not having smoked any cigarettes in the prior 30 days.We categorized participants as high vs. low behavioral health symptoms scores based on whether they were below or above the median score on the BASIS-24 subscale at baseline.CI: Con dence Interval; OR: Odds Ratio (adjusting for baseline cigarettes per day and site clustering).