Validity of self-reported smoking in schizophrenia patients
Takeaki Takeuchi, MD, MPH, PhD, Department of Hygiene and Public Health, Teikyo University School of Medicine, 2-11-1 Kaga, Itabashi, Tokyo 173-8605, Japan. Email: firstname.lastname@example.org
Aims: It is important to evaluate smoking status among schizophrenia patients because such patients are highly inclined to take up smoking, but only a few studies have focused on the validity of self-reported smoking in relation to schizophrenia. The aim of the present study was therefore to determine the validity of self-reported smoking and to investigate whether self-reported smoking is biased in schizophrenia patients.
Methods: A total of 158 schizophrenia patients answered self-rated questions of smoking status, and the level of carbon monoxide (CO) in expired air was measured. The relationship of the self-reported smoking to the CO levels was determined, and interaction of the disease duration and education level on this relationship was evaluated using correlation and receiver operating characteristic analyses.
Results: The CO levels in the expired air were found to be positively related to the self-reported data (P < 0.01). The stratified data on the duration of schizophrenia indicated that the positive relationship between self-reported smoking and the CO concentration became less obvious with the increase in duration of schizophrenia (Ptrend < 0.01). In contrast, the stratified data on education level did not indicate any such significant modifying effect.
Conclusions: Although self-reporting is useful for evaluating smoking status, psychiatrists should use objective methods of measurement to assess the smoking status of chronic schizophrenia patients.
IT IS IMPORTANT TO evaluate smoking status among schizophrenia patients because such patients are highly inclined to take up smoking.1 Schizophrenia patients who smoke regularly are at a higher risk of developing coronary heart diseases and stroke than the general population.2 Previous studies have suggested that schizophrenia patients may smoke as a form of self-medication for reducing the cognitive deficit,3,4 and the antipsychotic drugs they take interact with the metabolism of nicotine.3,5,6 Although it is hasty to advise all schizophrenia patients to stop smoking for health reasons, it is reasonable to advise the patients to stop smoking after having evaluated their smoking status.
The smoking status has been evaluated on the basis of several biochemical markers such as cotinine, thiocyanate, and carbon monoxide (CO).6–8 In a clinical setting, the detection of the CO level in expired air is more practical and therefore more commonly used in evaluating the smoking status than the detection of the cotinine levels in blood or uric acid, but all these detection methods are frequently used in research. Moreover, the costs of both analyses are considerably different, ranging from less than $1 per sample for CO analysis to $20 per sample for cotinine analysis.7 Some researchers have evaluated smoking status using objective methods of measurement such as the detection of CO levels in expired air.6,9–11 Murray et al. reported that self-reported smoking status was highly consistent with that determined by CO analysis in a non-schizophrenia group.6 It is possible, however, that self-reported smoking status in schizophrenia patients may be more erroneous than that in non-patients because such self-reports are highly dependent on the patient's cognitive and/or mental condition.12 To confirm the validity of self-reported smoking status in schizophrenia patients, we need to compare the self-reported smoking status with that determined by more reliable methods.
Therefore, in order to determine the validity of self-reported smoking and to investigate whether self-reported smoking is biased in schizophrenia patients, we monitored the CO level in exhaled air; the use of CO as a biochemical marker is a cost-effective solution and is used in all types of clinical settings. Self-reported smoking status and the data obtained from the objective measurements were first descriptively analyzed, and the relationship of self-reported to objective measures were analyzed, after adjusting for the potentially confounding effects of age and sex. We also assessed the effect of modifications of disease duration and education level because these factors may influence the level of communication or cognitive impairment in schizophrenia patients.13
The participants were 158 Japanese schizophrenia patients who were recruited from a psychiatric clinic located in a suburban area of Tokyo. The selection criteria were as follows: (i) treatment by physicians-in-charge from November 2007 to April 2008; (ii) diagnosis of schizophrenia as per DSM-IV-TR during the aforementioned period of hospitalization; (iii) stable medical condition (chlorpromazine-equivalent dose, 1062.5 mg) and a drug regimen that remained unchanged for >1 month; (iv) absence of respiratory diseases that result in an increase in the levels of CO, such as chronic obstructive pulmonary disease (COPD); and (v) agreement to participate in the study by signing the written informed consent form describing in detail the purpose and nature of the study and the risks associated with all the procedures. Concerning criterion 1, the single psychiatrist, whose interrater variance (κ) was 0.953, assessed all the patients, and this psychiatrist had discussed the patients' diagnoses once a week in the hospital in order to confirm the reliability of diagnoses. Furthermore, in the process of data collection and analysis, three psychiatrists checked the diagnosis correctness of schizophrenia using medical records. None of the patients in the present study had regular jobs because they were unable to lead normal social lives owing to the disease conditions. Twenty-seven patients (17.1%) had a family history of schizophrenia, with the condition having affected third-degree relatives. This study was performed in accordance with the World Medical Association Declaration of Helsinki, and the protocol was approved by the Institutional Review Board of the Teikyo University School of Medicine.
Self-reported smoking status and CO measurement
The patients were required to answer self-rated questions about their smoking status (current smoker, former smoker, or non-smoker) and daily consumption of cigarettes. The CO level in the expired air was measured with the Micro Smokerlyzer (Bedfont Technical Instruments, Kent, UK) by a psychiatrist proficient in using the machine. The participants were instructed to first hold their breath for 20 s and then exhale into the instrument. The end-tidal reading for CO in the expired air and the carbon monoxide measurement procedure involved two successive trials after lunch (during the fixed relaxing time in the hospital). Subsequently the average of the two readings was calculated. Although usefulness of the CO level varies according to the p.p.m., we used 9 p.p.m. as a cut-off because that is commonly used in previous studies.7,14,15
Both crude and age- and sex-adjusted analyses were performed to determine the relationship between the levels of exhaled CO and the self-reported data on smoking (smoking status and daily consumption of cigarettes) using ordinal logistic regression. The interactions between the disease duration and education level were then evaluated on Spearman's correlation and receiver operating characteristic (ROC) in order to investigate the interactive influence of these two factors on the self-reported data on smoking and the CO levels. P ≤ 0.05 (two-tailed) was accepted as significant. Statistical analysis was done using STATA version 8.0 (StataCorp, College Station, TX, USA).
Patient characteristics are listed in Table 1. The prevalence of admitted smoking was 40.5% in men and 35.4% in women. For the 60 patients who reported that they were current smokers, the daily consumption of cigarettes ranged from three to 40 (mean ± SD, 19.2 ± 9.7 per day). The CO levels in the expired air were found to be positively related to the self-reported data on smoking status and daily consumption of cigarettes on both the crude and age- and sex-adjusted analyses (P < 0.01 for all the analyses). Among the stratified data by duration of schizophrenia in Table 2, area under the curve (AUC) and correlation coefficients (r) decreased in turn. The stratified data on the duration of schizophrenia indicated that the duration of the condition had significant modifying effects on the relation between the CO concentration and the self-reported data on smoking: the positive relationship between self-reported smoking and the CO concentration became less obvious with the increase in the duration of schizophrenia (Ptrend < 0.01 in the correlation analyses). This indicates that the longer the duration of schizophrenia, the lower the accuracy of smoking data. In contrast, the stratified data on education level did not indicate any such significant modifying effect (Ptrend > 0.05).
Table 1. Patient characteristics (n = 158)
|Age (years), mean ± SD, [range]||54.7 ± 15.2||[16–87]|| || |
|Ratio of women||79 (50.0)|| || || |
|Duration of schizophrenia (years), mean ± SD [range]||28.3 ± 15.3||[0–75]|| || |
| <10||19 (12.0)|| || || |
| 10–19||31 (19.6)|| || || |
| 20–29||30 (19.0)|| || || |
| ≥30||78 (49.4)|| || || |
|Educational attainment (years), mean ± SD, [range]||12.3 ± 2.5||[9–18]|| || |
| Middle school||41 (26.0)|| || || |
| High school||56 (35.4)|| || || |
| Vocational school/2-year college||34 (21.5)|| || || |
| University/higher||27 (17.1)|| || || |
|Self-reported conditions of smoking|| || ||Male||Female|
|Smoking status|| || || || |
| Current||60 (38.0)|| ||32 (40.5)||28 (35.5)|
| Past||27 (17.1)|| ||22 (27.9)||5 (6.3)|
| Never||71 (44.9)|| ||25 (31.6)||46 (58.2)|
|No. cigarettes (per day), mean ± SD, [range]||12.8 ± 22.5||[0–240]|| || |
|Carbon monoxide per person due to number of cigarette smoked (p.p.m.), mean ± SD, [range]||9.3 ± 9.6||[2–48]|| || |
| No. cigarettes smoked per day|| || || || |
| Never||3.6 ± 1.1||[2–8]|| || |
| 1–9||6.8 ± 4.2||[2–13]|| || |
| 10–19||14.1 ± 6.2||[5–26]|| || |
| ≥20||15.7 ± 12.1||[3–48]|| || |
Table 2. Self-reported smoking status vs duration of schizophrenia and educational attainment (n = 158)
|General analysis (n = 158)||0.96||0.76||<0.01||0.84||0.28||<0.01|
|Group analyses|| || || || || || |
|Duration of schizophrenia (years)|| || || || || || |
| <10 (n = 19)||1.00||0.94||<0.01||1.00||0.95||<0.01|
| 10–19 (n = 31)||1.00||0.86||<0.01||0.82||0.52||<0.01|
| 20–29 (N =30)||0.95||0.77||<0.01||0.81||0.47||<0.01|
| ≥30 (n = 78)||0.94||0.70||<0.01||0.81||0.14||0.22|
|Educational attainment|| || || || || || |
| Middle school ( n = 41)||0.91||0.69||<0.01||0.83||0.13||0.43|
| High school (n = 56)||0.97||0.82||<0.01||0.90||0.54||<0.01|
| Vocational school/2-year college (n = 34)||0.95||0.71||<0.01||0.86||0.55||<0.01|
| University/higher (n = 27)||1.00||0.81||<0.01||0.70||0.49||<0.01|
For decades, clinical and research studies have been conducted to determine the awareness of smoking status among schizophrenia patients, but only a few studies have focused on the validity of self-reported smoking in relation to schizophrenia. To the best of our knowledge, this is one of the first studies to investigate the validity of self-reported smoking among schizophrenia patients. Psychiatrists should use objective methods of measurement such as the detection of the CO level in expired air to assess the smoking status of chronic schizophrenia patients because the duration of the disease may affect the accuracy of their self-reported data. The interaction of disease duration on the data seems reasonable because the patient's baseline functioning progressively deteriorates after each relapse of the psychosis, and also because the older patients have deterioration of cognitive functioning.13
In contrast, in the present study education level was not found to significantly modify the relationship between self-reported smoking data and the CO level, which was surprising because Swanson et al. had reported that schizophrenia patients in the higher education group had higher cognitive functioning than those with lower levels of education.16 Unknown factors that were not adjusted for, however, may have skewed our results.
Several limitations of this study should be taken into consideration. First, the number of subjects was relatively small, and all the patients were selected from the same hospital. The present study, however, was conducted in a clinical setting, and the patients were diagnosed with schizophrenia as per the DSM-IV criteria. Second, the self-reported data could be biased toward responses that the patient considers socially desirable. But when patients replied that it is not socially acceptable to smoke, it indicated that they were not influenced by psychological defense mechanisms, suggesting that information error was probably low in this group. Murray et al. have reported that 1% of self-reported non-smokers in a non-patient group had high levels of CO in expired air.6 In the present study, however, none of the self-reported non-smokers had high levels of CO in expired air. Third, the patients did not have fixed time intervals between each cigarette smoked, but the time at which CO was measured was the main relaxing time for patients in the hospital; habitual smokers were the most likely to smoke during that time. The present CO measurements were performed in the latter part of the relaxing time to involve all smokers as bona fide smokers. Fourth, the dose of nicotine contained in different types of cigarettes had not been adjusted for in the present study. In the future we recommend that multicenter studies involving a higher number of participants be carried out; moreover, various influential factors should be taken into consideration.
Despite these limitations, the present study showed that self-reporting is still useful in evaluating smoking status of schizophrenia patients, and that the smoking status of chronic schizophrenia patients should be assessed using biological measurements; this would increase the accuracy of the data related to smoking among schizophrenia patients.
The authors reported no financial support or potential conflicts of interests.