Commentary on Baker et al. (2012): Assessing dependence—are specific criteria for nicotine enough or do we need to consider its forms of administration?


The paper by Baker et al. [1] discussing revised criteria for assessing dependence on tobacco by the DSM system is very timely. Hopefully it will be a springboard for a much needed discussion. There are some issues that merit comment.

  • 1Continuous versus discrete estimation of dependence.
  • 2Generic versus specific criteria for substance and administration forms.
  • 3Condensing the Fagerstrom Test for Cigarette Dependence, FTCD (formerly called Fagerstrom Test for Nicotine Dependence) [2] to two questions.
  • 4Including withdrawal symptoms (WS) in dependence assessment.

For many medical conditions in general a discrete diagnosis of whether the disease is present or not is all that is needed. However there are conditions that depend on severity for their definition. A continuous scale helps to inform treatment decisions and risk of negative consequences of tobacco use, identify those in need of treatment and classify research subjects. A qualitative—discrete assessment—might be sufficient when legal-medical issues need to be decided on. For example approval to have a treatment reimbursed or paid for by a third party. However such decisions should also be possible based on a continuous scale. One reason for the popularity of the FTCD over the DSM system is that FTCD estimates the degree of dependence.

Many would agree with Baker et al. that using generic criteria across drugs is a suboptimal way of assessing dependence. For tobacco, using specific criteria yields a more precise assessment. However one might take this a step further and to an even more detailed level. Substances vary considerably with regard to the form and context in which they are administered (e.g. smoked and smokeless tobacco). Thus there is a possibility that different administration forms may be associated with different dependence patterns. The Fagerstrom Test for Cigarette Dependence has been altered and adjusted to measure dependence on smokeless tobacco. There is little evidence that the modified FTCD has actually predicted difficulties in quitting smokeless tobacco or cotinine levels when using smokeless tobacco. In two recent trials the modified FTCD was found to be unrelated to quitting outcome [3, 4]. Others however have found modified scales to predict quitting [5]. It is obvious that modifying an existing scale used for cigarette smoking would have difficulties picking up dependence characteristics from water pipe smoking. A case can be made that in order to characterise dependence on products that have varying forms of administration, specific questions should be developed and validated.

Evidence seems to suggest that the Heaviness of Smoking Index (HSI; number of cigarettes/day and time to first cigarette after awakening) can play a role in assessing tobacco dependence. In a recent study with subjects from 10 varenicline trials, FTCD did significantly predict abstinence. But the two HSI questions alone did equally well at predicting cessation outcome [6]. Thus no information was lost by using just two questions, at least when long term quitting is used as a criterion for dependence. A possible limitation with the HSI can be that it was developed to differentiate smokers asking for help with stopping smoking. The distribution of scores is biased towards left- lower scores- and it may not be as good a tool for the more low dependent smokers, chippers and beginners [7]. However, a population level study in England also found that it performed at least as well as the FTCD in predicting success at quitting [8]. This is a potentially important area for future research.

Including craving and WS in the dependence assessment might be a good idea. I was struggling with whether to include craving or not when I developed the FTCD. There are however some problems with including WS. What seems most natural is to measure WS as a reaction to a quit attempt. For that to be possible the person must have made an attempt and be able to remember the symptoms and their intensity. The way around this problem, as suggested by Baker et al. is to ask about WS after just a few hours of non smoking. However that is not without problems. Some of the WS, like sadness and depression, may take longer to develop and we should also consider whether the impact and relevance of WS occurring more or less during normal smoking are the same as those resulting from a firm resolution to quit smoking forever? The serious resolution to quit smoking forever may by itself give rise to more grief, low mood and anxiety than the symptoms that can be identified during short periods of temporary abstinence. Ultimately how the rating of WS from temporary abstinence and quitting smoking long-term relates to dependence is an empirical question. Craving is a different matter. As Baker et al. note, there is now evidence that craving ratings made during a normal smoking day can predict success of subsequent attempts to quit [8].

Declaration of interest

The author receives money from consulting for pharmaceutical companies.