The phrase primary literature refers to the actual journal articles that scientists publish -- hopefully in peer-reviewed journals. These articles describe not only the outcomes of studies but provide detailed descriptions of how the research was performed. Reading such an article with a critical eye can allow an informed science journalist to identify possible sources of bias and misinterpretation in the research and also frame the study in an appropriate light. For example, a case report describing one person with an obscure illness should not be given the same amount of attention and importance as a randomized controlled trial of hundreds or thousands of people with that same disease.
Reporting on science based on newspaper coverage or press releases is a potentially dangerous exercise. Although sometimes individuals may find it necessary -- particularly for studies that were only presented at conferences that the author did not attend -- it is essential for a science writer to always make the source they are using clear. The informed reader should realize that a study described "as reported by" another author is being filtered through two layers of possible misinterpretation. Making sources clear also means that there are good explanations for what has happened when a research study is found to be misreported by a large number of writers. The public has a legitimate interest in knowing whether such mistakes are due to something that the scientists said or due to misinterpretations in the early news coverage that have being passed along.
Example: In January 2011, an early newspaper article reported, incorrectly, that 10 percent of abstinent teens had STDs. This understandably newsworthy statement was picked up by a variety of other media outlets and quickly spread across the web. However, what the study had actually found was that 10 percent of teens with STDs had claimed to be abstinent -- a very different, although still interesting, result.
Although there are certainly mistakes in peer-reviewed articles, the primary source material is almost always the highest quality basis for science writing. Relying on newspaper reports and press releases is far too unreliable, because even the people who should know better sometimes get it wrong. Their mistakes may be the result of ignorance, lack of preparation, or simply typographical errors, but why risk repeating them when it is possible to go directly to the source? There is a reasonable basis to assume that a journal article has been written by experts in the field, but a press release could have been written by a publicist who has no background in the topic they are describing and who didn't understand what he was being paid to write.
In the case of the inaccurate press release I describe in the introduction, I suspect the mistake was simply due to a writer who was looking for a good headline and did not understand the substance of the research being promoted. Hopefully the release will soon be retracted and any incorrect media coverage of the topic will be appropriately corrected. In addition, I hope that this serves as an important reminder to everyone that information which makes you scratch your head and say "Really?" needs to be verified. Even though I knew that chlamydia was a curable disease, the first thing I did after reading the press release was find the study it was describing to make certain that it was not showing that there were hidden reservoirs of C. trachomatis that were consistently resistant to treatment. There weren't.
Although walled-off infections are sometimes an issue in women's with pelvic inflammatory disease, who need to have such abscesses surgically drained, the original paper agreed that chlamydia is easily treatable with antibiotics... but only if it can be detected. The real issue described by the research wasn't that chlamydia can't be cured once it's treated. It was that people don't get treated unless they get screened -- and that many people are unaware of their asymptomatic infections.