Measurement of factors that negatively influence the outcome of quitting smoking among patients with COPD: psychometric analyses of the Try To Quit Smoking instrument

Abstract Aims To test internal consistency and factor structure of a brief instrument called Trying to Quit smoking. Background The most effective treatment for patients with chronic obstructive pulmonary disease is to quit smoking. Constant thoughts about quitting and repeated quit attempts can generate destructive feelings and make it more difficult to quit. Design Development and psychometric testing of the Trying to Quit smoking scale. Methods The Trying to Quit smoking, an instrument designed to assess pressure‐filled states of mind and corresponding pressure‐relief strategies, was tested among 63 Swedish patients with chronic obstructive pulmonary disease. Among these, the psychometric properties of the instrument were analysed by Exploratory Factor Analyses. Results Fourteen items were included in the factor analyses, loading on three factors labelled: (1) development of pressure‐filled mental states; (2) use of destructive pressure‐relief strategies; and (3) ambivalent thoughts when trying to quit smoking. These three factors accounted for more than 80% of the variance, performed well on the Kaiser‐Meyer‐Olkin (KMO) test and had high internal consistency.


Introduction
The prevalence of Chronic Obstructive Pulmonary Disease (COPD) is usually related to tobacco smoking and a result of a cumulative exposure over decades. In western countries (or in Sweden), it has been estimated that 10% of the population may suffer from COPD and 50% of the elderly smokers (>45 years) (Lundback et al. 2003). Therefore, COPD is still the leading cause of morbidity and mortality (GOLD 2013). In clinical settings, there is a need for a valid instrument to predict failures during attempts to quit smoking among patients with COPD. Such an instrument may enable health professionals to tailor their support on the patients' needs. In fact, almost half of the patients with COPD continue to smoke after the diagnosis (Tashkin & Murray 2009), with consequent worsening of symptoms and prognosis of COPD (Tashkin & Murray 2009, Vozoris & Stanbrook 2011. Failure in quitting smoking among patients with COPD may also induce feelings of frustration and powerlessness among healthcare providers (Lundh et al. 2006), possibly leading to professional disinvestment towards these patients.

Background
There is a wide range of instruments that can be employed by healthcare professionals to predict the outcome of a quit attempt, exploring dimensions such as nicotine dependence (Heatherton et al. 1991), cigarette dependence (Etter et al. 2003), abstinence-related expectancies (Hendricks et al. 2011), self-efficacy beliefs regarding smoking cessation (Rash & Copeland 2008, Speak et al. 2013) and of smoking outcome expectancies (Rash & Copeland 2008). None of these instruments seems to fully explore the difficulties patients with COPD deal with when they try to quit smoking. Patients with COPD and especially women seem to find it harder to quit smoking than patients without this diagnosis (Tashkin & Murray 2009, Vozoris & Stanbrook 2011, although the reasons of this difference are not clear. To increase success rates in smoking cessation among patients with COPD, programmes with different characteristics and intensity have been used. Recent systematic reviews suggested that the most effective smoking cessation strategy for this patient group would be pharmacological therapy in combination with intensive counselling (Warnier et al. 2013) and face-to-face contacts yielding higher quit rates than telephone contacts (Stead & Lancaster 2012). In primary health care, where most patients with COPD are treated, low-intensity support in smoking cessation is usually employed, such as simple advice to quit, minimal intervention or telephone counselling, alone or in combination with medications (nicotine replacement therapy, bupropion or varenicline (Cahill et al. 2013). To redirect and use efficiently the limited resources of healthcare providers in outpatients' settings, it would be very useful to identify those COPD patients who might develop pressure-filled mental states, have high risk of failure during a quit attempt or even give up any plan to quit smoking after an unsuccessful attempt.
In a recent qualitative study conducted by our group, some patients with COPD developed pressure-filled mental states when trying to quit smoking. These mental states include feeling fearful (worries about health), feeling criticized (from self and or others), feeling pressure (continuous thoughts about quitting) and worthlessness (e.g. being a weak person). Besides having a negative impact on mental well-being, these feelings may constitute risk factors for loss of motivation and for giving up further quit attempts (Lundh et al. 2012). Low self-efficacy (Baldwin et al. 2006) and feeling of guilt and shame (Arne et al. 2007, Wilson et al. 2011) seem to be common among patients with COPD. Constant thoughts about quitting and repeated quit attempts, not accompanied by confidence in success, can generate very destructive feelings and the very same decision to quit may appear unbearable (Jonsdottir & Jonsdottir 2007). To date, there is no instrument that tackles these mental processes.
The scope of this study is to test the internal consistency and factor structure of a brief instrument denominated TTQ, designed for processes that negatively influence the occurrence or the outcome of quit attempts among patients with COPD.

Methods Design
A survey and psychometric approach was used to test the initial reliability and validity of the TTQ scale.

Development of TTQ
A theoretical model was used to develop the TTQ questionnaire (Lundh et al. 2012). The model describes the process of trying to quit smoking in patients with COPD and the reasons behind successful and unsuccessful quit attempts, resignation and continued smoking. The model identifies as specific cue to continued smoking pressure-filled mental states when trying to quit, their causes and strategies commonly used to achieve pressure relief. Ideally, the TTQ was meant to cover the following aspects: (I) decision about smoking; (II) development of pressure-filled mental states; (III) strategies to manage pressure-filled mental states; and (IV) degree of hope and meaningfulness in quitting smoking. Based on the theoretical model, three dimensions and 19 items were identified to describe the process of trying to quit smoking.
The TTQ was developed through several steps and the content and face validity were assessed both by nurses and by smokers diagnosed with COPD : (1) first, four nurses responsible for patients with pulmonary diseases (experts) were asked to judge the comprehension of each item, the relevance of the questions and of the response alternatives and whether the TTQ appeared to measure target variables, i.e. factors influencing an attempt to quit smoking. After this assessment, some statements were rephrased; (2) second, 20 smokers diagnosed with COPD were asked to rate their understanding of each TTQ item. They also rated the applicability and completeness of the questionnaire as a whole. All items except three were rated as easy to understand. These three items were rephrased; (3) third, the TTQ was pretested among 20 smokers with COPD. Fifteen items were analysed separately for each dimension of the theoretical model. This analysis led to a preliminary grouping of the questionnaire items in themes following the theoretical model. Some adjustments were also made in the wording.
The final version of TTQ is a 19-item instrument covering factors likely to describe the mental processes of COPD patients before and during a quit attempt. For the (Kottner et al. 2011) purpose of the present study, some questions on demographic, severity of COPD and smoking history were added to the instrument. For all items the response is given on a 4-point Likert scale ranging from 1 (do not agree at all)-4 (do agree completely). In the final three-factor model, four items covering decisions about smoking were excluded and two positively worded items were reversed before analysis.

Data collection
We aimed to enrol in the study 250 consecutive smoking patients with COPD attending a regular clinical visit at Primary Healthcare Centres (PHCC). Enrolment and data collection took place in 29 primary healthcare facilities in the Stockholm County. At most PHCC, one or two nurses are in charge for nurse-led programme for pulmonary rehabilitation. In total, 171 nurses with such responsibility were invited to take part in the study, of which 31 accepted. They were each asked to recruit five patients with a diagnosis of COPD, i.e. a total of 155 patients. The patients were eligible if they currently smoked at least weekly and were able to speak Swedish. Informed consent to inclusion in the study was provided orally during the patients' regular consultations with the corresponding nurse from April 2011-November 2012. Prior to consent, the nurses provided a covering letter about the study and pamphlet with information about smoking cessation. In a Table 1. Demographic and behavioural characteristics of the participant.

Characteristic
All respondents (n = 102) Respondents trying to quit smoking (n = 63) 2-week period, each nurse recruited a mean of 3 (range 1-5) consecutive smoking patients with COPD during the patients' regular clinic visits, a total of 102 patients. Lack of time was the only reason for some nurses not being able to recruit all expected five patients.

Statistical analysis
If not otherwise stated, demographic and smoking history characteristics are presented as mean and standard deviation, for numerical continuous variables and as frequencies and percentages for categorical variables. Comparison of continuous variables between groups was done by a two-tailed t-test, where normality assumption was met. For categorical variables, the chi-square test was used when expected values were more than five; otherwise, the Fisher exact test was used.
Internal consistency (Chronbach's alpha) was evaluated for reliability. The acceptable value for minimum reliability coefficient was >0Á70 (Kottner et al. 2011). To explore the number of dimensions in TTQ, exploratory factor analyses (EFA) were conducted, using unweighted least squares (ULS) method based on polychoric correlation matrix followed by orthogonal varimax rotation. Kaiser-Meyer-Olkin (KMO) was performed to confirm adequate number of factors in data. The goodness of fit with three factors was confirmed after rotation. The Likelihood ratio test (chisquare) was used to compare the chosen three-factor model against the saturated factors model with conventional statistical significance indicated by P < 0Á05. Critical selection of items in the factor analysis was: no cross-loading; load-ing value over 0Á30; uniqueness over 0Á60. Bartlett's test of Sphericity was used to test the hypothesis that correlation was an identity matrix. No missing data were reported. All statistical analyses were performed using SAS version 9.3 (SAS Institutet Inc., Cary, NC). A significance level of 5% was chosen.

Research Ethics Committe approval
The committee of Ethics Research at Karolinska Institute in Stockholm approved the study (2008/1929-31/5).

Results
Participants' characteristics are described in Table 1. Women constituted the majority of the respondents (69%). Patients' mean age was 65 years (SD 8). Most of the participants had only attended the elementary schools, especially among men. Half of the participants were married or living with a partner. The majority (73%) had been diagnosed with moderate COPD [forced expiratory volume (FVC) at 1 second <60%] less than 3 years before. Participants had smoked for an average of 45 years and had an average of three past quit attempts (SD 4). Daily smokers consumed an average of 15 cigarettes per day (SD 7). All of the 63 respondents who were in the process of quitting (right-hand side of Table 1) were included in the present assessment of TTQ. The demographic characteristics of this group reflected those of the whole sample except for education and for FEV 1 %.

Exploratory factor analysis (EFA)
Sixty-three respondents actively trying to quit smoking answered the TTQ's 17 items intended to explore the mental states arising in the process of quitting (Tables 2  and 3) and were included in the EFA. Three factors accounted together for more than 80% of the variance. The Kaiser-Meyer-Olkin (KMO) was 60% and Cronbach's alpha >70% (Table 3). After rotated exploratory factor analysis, three items were excluded because of low loadings (below 0Á30) and high uniqueness (>0Á70) ( Table 2). KMO estimate 60% of the remaining 14 items and the Bartlett test was statistically significant (P < 0Á00). Analyses resulted in a three-factor solution including 14 items that accounted together for more than 80% of the variance (Table 3). Latent variables were labelled: 1. Developing pressurefilled mental states (F1) while trying to quit smoking, i.e. perception of being criticized, harbouring constant thoughts about quitting and feelings of worthlessness (5 items), (score 11Á83 AE 0Á63 max score 20 and min 5). 2. Use of destructive pressure-relief strategies (F2) included items indicating the quest for alibi not to proceed to action, such as endorsement of quitting being meaningless because of high age; or cutting down the daily consumption as sufficiently effective strategy (4 items) (score 5Á41 AE 0Á30 max 16 and min 4). 3. Ambivalent thoughts (F3), such as perceiving lack of support in the social environment, fear of physical reactions following abstinence, feeling obliged to quit (positively loading), accepting information about new methods in smoking cessation and refusing information on the progression of the disease (negatively loading) (5 items); (score 11Á21 AE 0Á40 max 20 min 5) (Table 4).
Items included in factor 1 developing pressure-filled mental states scores from mean 1Á70-3Á00 (max 4Á00). There was tendency of a floor effect in the response profile in factor 2 using destructive pressure-relief strategies, mean score from 1Á13-1Á35 (max 4Á00). For factor 3 ambivalent thoughts the item score from mean 1Á52-2Á84 (max 4Á00) (Table 4).
We also found that women's score on pressure-filled mental states was significantly higher than men, 13Á30 (3Á9) vs. 10Á05 (3Á5), P value <0Á00. There were no significant Table 3. Exploratory factor analysis varimax rotation factor loading (n = 63).

Rotated factor pattern (1)
Revised factor pattern (2) Bold marked values represent the chosen items in each factor based on uniqueness Gray marked values represent abs ( differences between women and men in use of destructive relief strategies 4Á89 (1Á9) vs. 5Á3 (1Á7), P value <0Á46 and ambivalent thoughts 10Á65 (2Á2) vs. 10Á79 (2Á4), P value <0Á82.

Main findings
This paper reports on the psychometric properties of a novel instrument, the TTQ questionnaire, based on a theoretical model applied to COPD smokers planning a quit attempt. Of the original 17 items included in the scale, 14 items were retained in the factor analyses and loaded on three factors, indicating: (1) development of pressure-filled mental states (5 items explaining different feelings associated with low self-efficacy and efforts to conceal smoking when not succeeding); (2) use of destructive pressure-relief strategies (4 items explaining strategies of rationalizing); and (3) ambivalent thoughts (5 items related to contrasting attitudes towards quitting). These items performed well on test of KMO and had high internal consistency. An exploratory factor analysis indicated that these three factors explained over 80% of variance in the data. The face validity of the TTQ (i.e. adherence to the underlying theoretical model) seemed partial (Lundh et al. 2012). Two of the factors identified by factor analysis corresponded rather well with those hypothesized by the theoretical model, that postulated pressure-filled mental states and destructive pressure-relief strategies factors, leading to hopelessness and resignation, therefore to continued smoking (Lundh et al. 2012). However, two of the originally hypothesized items did load on other factors, i.e. 'concealing smoking status', that loaded on the factor indicating 'pressure-filled mental state' and fear of getting information on disease progression, that loaded on a third factor identified as 'ambivalent thoughts'. This latter factor was not identified as a predictor of smoking cessation in the theoretical model. Items loading on this factor were originally used to indicate either pressure-filled mental states or use of constructive pressure-relief strategies. This discrepancy may indicate insufficiency of the brief assessment instrument TTQ towards predictors of both unsuccessful and successful quitting (constructive strategies) and/or failure of the theoretical model to include important mental processes such as ambivalent cognition. It was also found that women were more prone to develop pressure-filled mental states than men. This is in agreement with studies reporting that women experience poorer health-related quality of life than men with COPD and also show more frequent occurrence of anxiety and depression (Naberan et al. 2012). Therefore, TTQ can be a useful guidance in tailoring counselling on women's need.

Strengths and limitations of this study
The purpose of this study was to investigate if the novel theory-based instrument TTQ could reliably measure Table 4. Response distribution of TTQ items among COPD patients trying to quit smoking (n =63*). mental processes involved in smoking cessation among people with a diagnosis of COPD. Strengths of the study include the usual care setting that allowed the simultaneous exploration of feasibility, management and performance of the instrument. Several limitations may have impaired the ability to detect a deeper correspondence between the theoretical model underpinning the instrument and the empirical findings, above all the small sample size. Further, we were not able to study construct validity of the TTQ instrument at this stage. To do this, besides a larger sample, one would need an external reference (e.g. correlation with scales measuring the same factors) that was lacking in this study.
In line with the natural history of the disease, participants were mostly older, moderate to heavy smokers, with an average duration of smoking of 45 years. Furthermore, the sample consisted mostly of women, in agreement with the higher prevalence of COPD among Swedish women than men (GOLD 2013, National Board of Health & Welfare 2013). Other selection factors (i.e. at the clinic level) may also have been in place. Therefore, caution is needed when generalizing findings and conclusions to the whole population of smokers with COPD. However, as the nurses did not report any refusal to participate, participants should be fairly representative of the patient population served by primary care facilities in Sweden.
The TTQ was developed to identify pressure-filled mental states and destructive strategies when trying to quit smoking specifically among patients with COPD; therefore, its usefulness among smokers with other diseases or in non-clinical settings cannot be assumed. Also, satisfactory content and face validity do not imply actual usefulness of the instrument in predicting the risk of failure in a given quit attempt among COPD patients, an aspect that remains to be investigated in the frame of a longitudinal study.

Interpretations of findings in relation to previously published work
Other studies have found that anticipated negative selfevaluative emotions, defined as being dissatisfied with own smoking, feeling stupid, fed up, guilty, ashamed, in relation to behaviour change were connected with higher risk of relapses into smoking (Dijkstra & den Dijker 2005, Menninga et al. 2011. Therefore, it can be of importance to identify persons with high feeling of pressure-filled mental states and use of destructive pressure-relief strategies. Most of the destructive pressure-relief strategies described in the theoretical model also loaded on the same factor in the EFA. Perceiving ambivalence is an unavoidable component of all behavioural changes, as a part of the discrepancy between pros and cons in a decision, such as quitting smoking. Low ambivalence can indicate high motivation to quit smoking (Prochaska et al. 1992). On the other hand, if ambivalence is still present after decision about quit smoking is made, there is an increased risk for relapse (Baillie et al. 1995, Menninga et al. 2011. The relation of perceived ambivalence with relapse can also be mediated by triggers connected to situations where smoking was habitual in the past (Menninga et al. 2011). This indicates that patients' ambivalence may be an important aspect to be considered in relapse prevention. If so, understanding ambivalence would probably be major cue in motivational counselling (Balmford & Borland 2008).
The usefulness of TTQ rests on its ability to measure determinants of protracted smoking not necessarily corresponding to a tobacco dependence syndrome. Indeed, the instrument deals with feelings of pressure-filled mental states, for instance negative criticism and worthlessness. Destructive relief strategies, leading to giving up further quit attempts, specific to patients with COPD, are also tackled by the instrument, something that was not assessed or evaluated in previous studies.
Other instruments have attempted to include emotional aspects of the process of quitting smoking. For instance, a well-known clinical instrument, the 55-item Smoking Abstinence Questionnaire (SAQ), aims to measure the role of abstinence-related expectancies, such as withdrawal, social improvement/non-smoker identity, adverse outcomes, common reasons to quit smoking, optimistic outcomes (Hendricks et al. 2011). The 12-item Cigarette dependence scale (CDS) measures both nicotine dependence and some emotional aspects of the process of quitting smoking (5).
On the same track, the Smoking Abstinence Self-Efficacy Questionnaire (SASEQ) is a 6-item questionnaire measuring two dimensions describing social and emotional situations in risk of relapse. People who quit smoking had significantly higher scores of SASEQ before quitting than those who were not motivated to quit smoking (Speak et al. 2013).
Finally, the Brief Smoking Consequences Questionnaire -Adult (BSCQ-A) was developed to assess smoking outcome in adult heavy smokers according to their expectancies towards smoking or smoking cessation. Outcome expectancies are assessed on 10 scales named negative affect reduction, stimulation/state enhancement, health risk, taste/sensorimotor manipulation, social facilitation, weight control, craving/addiction, negative physical feelings, boredom reduction and negative social impression (Rash & Copeland 2008).
Some of these scales are of interest because they can be easily combined with TTQ for in-depth assessment of need for support, as they cover both overlapping and different constructs. For instance, items in the CDS (Etter et al. 2003), such as the feeling to be prisoner of cigarettes, seem to relate well to pressure-filled mental states assessed by the TTQ. Items in the SAQ (Hendricks et al. 2011) assessing expectations about withdrawal and adverse reactions when quitting are closely related to TTQ's pressure-filled mental states and ambivalent thoughts. As TTQ assesses feelings and thoughts not associated with self-efficacy, it can be a useful complement to the SASEQ scale (Speak et al. 2013) in predicting successful smoking cessation. Negative social impressions in the BSCQ-A (Rash & Copeland 2008) can be related to pressure-filled mental states and feelings of criticism.

Conclusions
The new assessment instrument 'Trying to quit smoking' (TTQ) can be useful to predict factors that have a negative impact on attempts to quit smoking among patients with COPD and guide professional support to smoking cessation.

Relevance to clinical practice
The TTQ instrument reliably identifies factors such as high degree of pressure-filled mental states, use of destructive pressure-relief strategies and feelings of ambivalent thoughts among patients with COPD. Patients with high scores in these domains, i.e. high feelings of pressure-filled mental states, may need tailored individual advice to succeed in their attempts to quit smoking. TTQ is a short instrument that easily identifies these patients in a clinical setting, allowing among other things a more efficient use of the nurses' time. In conclusion, the TTQ instrument can represent a useful tool to healthcare providers when counselling COPD patients for smoking cessation.