After England, Taiwanese government is the second in the world to reimburse smoking cessation. Smoking cessation services in England do not require specific training in order to receive reimbursement. The advent of smoking cessation treatment services resulted in a rapid proliferation of training courses in smoking cessation over the first few years of implementation. An audit of smoking cessation training suggested that training quality was variable . Smoking cessation training in Taiwan is centralized and sponsored by the government. It is mandatory for all physicians who want to receive reimbursement. The results of this research represent our efforts at promoting quality in the training program.
There was not a standard how a smoking cessation training program should be organized or how many hours physicians should be trained. We complied with the suggestion of US PHS guideline that the physicians provide pharmacotherapy and brief intervention less than 10 minutes in each encounter. Therefore Pharmacotherapy of smoking cessation was an essential topic. The trans-theoretical model and the clinical skills for smoking cessation were the theoretical and practical components of counseling training. We included an hour of clinical case discussion to simulate real world cases and enhanced the interaction between lecturers and audience. The strategy of tobacco control illustrated the global tobacco issues and the policy aspects of tobacco control. Two additional topics were added in 2007: nicotine addiction and withdrawal, and the risks of smoking and the benefits of quitting. We believed the two new topics could enhance the physician's ability in encouraging smokers to quit. People may concern a six-hour training was not enough due to the complexity of behavior-changing counseling. The purpose of our counseling training was to provide the brief intervention during physician's daily practice, not the cognitive-behavior therapy which is usually provided by a psychotherapist. It costs too much to train a physician to provide cognitive-behavior therapy and such training was not suggested by smoking cessation guidelines.
It is not surprising that our program generated high levels of satisfaction and had significant pre- and post-test differences in trainee knowledge. Most of the Taiwanese physicians had little if any previous knowledge on smoking cessation information and skills. There were no courses teaching smoking cessation in medical schools and very few continuing medical education courses focused on smoking cessation. Initially, the National Health Insurance would not cover smoking cessation costs so very few physicians were willing to provide services. The situation changed in 2002 after the barrier of payment was removed. As a result, one in six physicians in Taiwan attended the training program and smoking cessation became a popular service. Critics may argue that one in six is not a large portion of physicians in comparison with other popular training such as hypertension or diabetes. However, the rapid expansion of smoking cessation services already exceeded the expectation of Taiwanese government. The budget was not enough to cover reimbursement in 2008 and the government has raised cigarette tax since June, 2009.
Long term evaluation results revealed that trainee confidence in providing services declined after a period of time. More trainees answered "neutral" or "unconfident" when asked about providing services compared to the short term evaluation. Because there was a significant difference in the confidence between the trainees contracting or not contracting with BHP, we inferred the decline in confidence as possible being due to trainees not providing services. Knowledge and skills of trainees not providing services would decay. A study revealed a similar observation. Physicians actively assisting patients to quit had higher levels of performance and self-efficacy . It is also possible that the trainees under estimated the complexity of smoking cessation service at the time of the short term evaluation. This may have contributed to the decline in confidence at follow-up. This may indicate an initial underestimation of the relative complexity of behavior change support. Similar results have been reported from other studies assessing tobacco prevention work in general practice .
The percentages of trainees practicing 5As guideline were higher than those in studies from the US [11–14], China , and Hong Kong . However, the high adherence to the guideline was probably due to the payment policy. To remove the influence of payment, we further analyzed the data by extracting the trainees not contracting with BHP from the total trainees. The trainees not contracting with BHP practiced 5As as follows: 75% for ask, 84% for advise, 66% for assess, 38% for assist, and 35% for arrange. These data represented the adherence to 5As guideline without the incentive of payment. The results were still higher than studies in the US (67% for ask, 74% for advise, 35% for assist, 8% for arrange) , China (48% for ask, 64% for advise)  and Hong Kong (77% for ask, 74% for advise, 20% for arrange) . The adherence levels in our study might be overestimated because respondents may have tried to choose the "right" answer [17, 18]. Collecting chart reports would better estimate adherence. The National Ambulatory Medical Care Survey (NAMCS) in the US was an example . In this study, 32% of patient charts did not include information about tobacco use, 81% of smokers did not receive assistance, and less than 2% received a prescription for pharmacotherapy. However, physicians might forget to record this information in charts and the results underestimated actual practice . There is not an ideal method to estimate guideline adherence except through direct observation of encounters, which is difficult to carry out .
Physicians with greater self-confidence, female, in private offices, and younger, showed greater compliance with smoking cessation guidelines [13, 14, 19–21]. However, another study revealed age, gender, practice site (e.g., HMO, solo practice), and pediatric subspecialty were not related to guideline adherence . One study of primary care physicians, who were older, with academic positions, with trained staff for counseling, and with higher confidence, showed they were more likely to advise patients to quit . In our study, physicians contracting with the BHP were more likely to have higher confidence in their ability to provide smoking cessation services and better adherence to the practice guideline. The significant correlation between confidence and adherence could be explained by the confounding effect of payment. Nevertheless, there was a possible causal relationship in that physicians with higher confidence were more adherent to the practice guideline [11, 20, 22, 23]. Internal medicine physicians were less confident than family physicians or psychiatrists in this study. The reason was probably less training in counseling skills. Block et al. reported that specialists had fewer counseling skills in smoking cessation than primary care physicians . Adherence to the guideline correlated to physicians' age in this study. The older the physician, the better the adherence to the guideline; however, age was excluded in the multiple regression models. We did not find differences in gender, practice settings (primary care, hospitals, medical centers), stakeholders (private, public), or training years (2002 to 2006) as having association with the adherence to the practice guideline.
The response rate of our long tern evaluation questionnaire (38%) was lower than other studies [11, 12, 19–23]. It has been well recognized that physicians in Taiwan are reluctant to answer questionnaires. For example, a study investigating physicians' attitudes toward DNR of terminally ill cancer patients in Taiwan had a response rate of 18% . Another study investigating obstetricians' willingness to practice collaboratively with midwives had a response rate of 16% . These were national studies and published in an international journal. To our knowledge, the response rate of our study was high compared to other Taiwanese studies [27–30].
The remaining question was whether the respondents represented a valid sample of the study population. There was no difference in gender, age, and training year between respondents and non-respondents. Another estimation of representativeness was the percentage of the trainees contracting with BHP. There were 70% trainees contracted with BHP in the study population while 70% in our respondents, too. The above analysis revealed that there were not obvious biases in our sample. However, we could not exclude the possibility that physicians who were better in providing smoking cessation services were more likely to respond to the questionnaire.
Evidence suggested the implementation of a smoking cessation practice guideline could decrease smoking rates. Katz et al. conducted a non-randomized, before-after trial at family practice settings. After a two-month period of guideline-derived intervention, the test sites had higher two-month and six-month quit rates compared to rates before the intervention while the controlled sites had no difference . Katz and his colleagues conducted another randomized controlled trial with similar design after two years and confirmed the effectiveness of implementing practice guideline . Ward et al. analyzed the data from 138 Veterans Administration medical centers suggesting the implementation of the Agency for Health Care Policy and Research (AHCPR) smoking cessation guideline increased smoking cessation counseling and decreased smoking rates . These studies suggested that the adherence to a practice guideline could be a good quality indicator of smoking cessation training programs.
There were several limitations to this study. There was not a control group or baseline data of guideline adherence to verify the effectiveness of training. We compared our data with other studies that were not conducted in Taiwan. The Chinese culture possible made people tend to choose the "right" answer and over-estimated the adherence. It was not a problem when comparing our data with the studies in China and Hong Kong, but could be a problem when comparing our data with studies in other non-Chinese countries. Although the representativeness of respondents in the long term evaluation was acceptable, there were greater numbers of family physicians and fewer psychiatrists. The difference was significant. The response rate to the long term evaluation questionnaire was relatively low, although it was higher when compared to other Taiwanese studies. The low response rate due to Taiwanese culture was understandable, however, it could compromise the quality of the study. The outcome of this study was adherence to the 5As guideline, which served as a surrogate outcome. The evidence of an increased quit rate after the training was not available.
It would be interesting to explore the reasons why 30% of the trainees were not contracting with BHP and not getting reimbursement. Less than 1% of trainees answered "unconfident" or "very unconfident" in providing smoking cessation services in the short term evaluation. The discrepancy between the confidence in providing services, and the practice of providing the services requires further study.
We suggest smoking cessation training should be one part of undergraduate course in medical school. It costs a lot of resources to implement post-graduate training in large scale. There are some medical schools in Taiwan which teach smoking cessation knowledge and technique in undergraduate curriculum. We anticipate more medical schools implementing this policy in the future.