This article does a pretty good job outlining the problems and the potential relationships between symptoms:
The most common understanding of this comorbidity is that having an anxiety disorder predisposes one to develop an alcohol or substance use disorder via self-medication—using alcohol or drugs to modulate anxiety and negative affect. However, substance use disorder experts have argued that the social, occupational, and physiologic effects of substance use can generate new anxiety symptoms in vulnerable individuals. In other words, physiologic and/or environmental disruptions from chronic alcohol or substance use could promote conditions and circumstances in which anxiety symptoms are more likely to emerge or worsen.
Although DSM-IV-TR does not delve into the causes of mental disorders, it states that substance use can cause or “induce” an anxiety syndrome with symptoms that resemble or are identical to those of the various anxiety syndromes that are not related to substance disorder.
Alternatively, the idea that a third factor can serve as a common cause for both conditions fits with the view that substance use disorder and anxiety disorder can be phenotypic expressions of a common underlying genetic/physiologic liability.
Finally, these models are not mutually exclusive. Anxiety symptoms or substance use could cause or aggravate the other.
It also covers the possibilities in the sequence of symptoms (with a little psychiatric bias):
Anxiety disorder typically begins before a substance use disorder in comorbid cases, although some studies have reported the opposite pattern or roughly simultaneous onset of both disorders.
Using a prospective method in college students, we found that the risk of developing alcohol dependence for the first time as a junior or senior more than tripled among students who had an anxiety diagnosis as a freshman. We also found, however, that students who were alcohol dependent as freshman were 4 times more likely than other freshman to develop an anxiety disorder for the first time within the next 6 years.
In short, having either an anxiety or alcohol disorder earlier in life appears to increase the probability of developing the other later. This finding supports the idea that the types of associations that link pathologic anxiety and substance use vary among individuals and, perhaps, within individuals over time.
All of this seems to really come down to good, conservative diagnostic practice.