As someone who occasionally (or sometimes more than occasionally) says the wrong thing in an attempt to be helpful or comforting of "fix" the situation, I was particularly terrified by a recent article in the Journal of Occupational Health Psychology. Beehr, Bowling, and Bennett's article titled "Occupational stress and failures of social support: When helping hurts" appears in this month's JOHP. In the interest of full disclosure (my personal motto), I'll admit that my university does not yet have full access to the article and my own personal hard copy has not arrived in the mail yet (one of the pitfalls of living in rural America), thus I can only provide a link to the abstract. But, from the abstract, it appears that these fairly well-known and well-respected scientists found that certain types of social support are actually harmful to coping with stressful situations.
Namely, the authors explored the effects of (1) interactions that make the person focus on how stressful the workplace is, (2) help that makes the recipient feel inadequate or incompetent, and (3) help that is unwanted. All three were related to worse rather than improved psychological and physical health, and interactions that drew the attention to the stress (the first type of help listed) was most harmful. Though the study appears to be done in an online survey format, this is still an alarming finding given the ways we currently think about and treat individuals in therapy and those who come to us for help and social support. 

I mention that we may be hopeful that these results are obtained via an online medium rather than in face-to-face interactions because we may still find (in another future study) that in face-to-face interactions, the benefit comes from the presence of another person and belief that he or she cares rather than the content of the advice or support. Perhaps it is simply the presence of another or the knowledge that the other person cares - even if they aren't adept at knowing what to say to make it all better. There are several types of findings that I believe could be used to support such a conclusion (and future studies to investigate the differences between internet and face-to-face social support interventions). 

One particular subfield or line of research concerns social facilitation wherein the mere presence of another person can facilitate action on particular types of tasks - those the individual is not learning and is more of an expert at, where the task is more routinized. Additionally, I'm considering findings that the presence of others may reduce levels of the stress hormone cortisol as support for this idea (e.g., Allen, Blascovich, Tomaka,&Kelsey, 1991). The reverse is true too - social ostracism has been linked to physical pain (e.g., Eisenberger & Lieberman, 2004; MacDonald & Leary, 2005). This finding would at least help those of us who worry about saying the wrong thing to troubled friends feel a bit better about our own attempts to aid, even if should still provide concern for the researcher and clinician. 

However, I find these results troubling because of what we know about interventions that have been shown to help people. Specifically, James Pennebaker has conducted research on the stress reducing effects of expressive writing - writing about stressful or traumatic events - for decades now (for a fairly good review/summary of findings despite its age, see Pennebaker, 1997). These types of interventions have helped survivors of natural disasters and rape as well as those experiencing less severe negative events, such as job loss (Spera, Buhrfeind, & Pennebaker, 1994). His research and that stemming from his discoveries have shown that writing about these events can reduce intrusive negative thoughts, depressive symptoms, physiological symptoms of stress, etc. (you can Google his name and see that almost any of his research demonstrates some part of this, which is why I'm not linking to specific articles). 

So we are forced to question how confronting the issue in writing might aid individuals in the coping process in some studies, but talking about the issue might be harmful? Neither situation really allows the individual to choose to address the issue or talk about it on their own terms, though perhaps the more recent internet-based intervention forces the individual to deal with the event more so than the writing because it requires the individual almost confide in others while the writing may remain private. But then, what does this say about the effect of social support and stress if the situation may be better handled by the individual alone through writing? And how can this be reconciled with other findings about the benefits of social support for stress relief?