This is, we believe, the first study from a low-income country concerning depression in a large sample of orthopaedic outpatients. The prevalence of depression in our sample was high with an SRQ score of 9 or more, which indicates probable depressive disorder, occurring in 45.6% of men and 76.1% of women. These proportions are almost identical to those we recorded in a similar consecutive sample of patients attending medical clinics at the same hospital (47% of men and 63% of women)  and are much higher than those we previously found in a population-based sample in rural Pakistan (17.5% of men and 44.1% of women) . In the latter study all but one of the people who scored nine or more on the SRQ had depressive disorder ascertained by research interview. Thus this cut-off point appears to indicate probable depressive disorder rather than emotional distress.
Our results are consistent with studies from the West, which have reported that the prevalence of psychological disorder following physical trauma in a variety of settings is 23% to 41% [9, 11, 12, 15, 25]. Such proportions are much higher than those found in the general population and somewhat higher than those reported in some medical clinics in the West . As a result of methodological differences our results cannot be directly compared with those reported from the west but it is notable that in both USA and Pakistan trauma patients have a very high prevalence of probable depressive disorder.
Our primary hypothesis that the prevalence of depression would be higher in patients with medically unexplained musculoskeletal symptoms was disproved. Although patients with medically unexplained musculoskeletal pains experienced great psychological distress the level was even higher in patients with arthritis, backache and other bone pathology. This is different from studies in high-income countries , and our own previous study in medical clinics in Pakistan , which indicated that depression is often associated with 'medically unexplained symptoms'. The patients included in this study are clearly very distressed with a high proportion entertaining suicidal ideas: 24% compared to 5% in our previous population study . Thus we anticipate that many of these patients need treatment for depression.
It is apparent that the usual sociodemographic factors are clearly associated with a high SRQ score (female, little education, financial difficulties, lack of social support, divorced, separated or widowed status, numerous children) . It is probable that many of these people may have been depressed before their fracture or illness because of their life circumstances. It is impressive, therefore, that several orthopaedic diagnoses added additional variance to the explanation of SRQ score in multiple regression. This indicates that the presence of such illness is associated with even greater distress than can be accounted for by sociodemographic circumstances. This may be understandable in terms of the further hardship such illness brings in a setting of poverty and hardship. These orthopaedic diagnoses will probably have prevented these patients from earning a living or doing their usual household and other tasks.
The main limitation of this study is the fact that we did not perform second stage interviews to ascertain definite psychiatric diagnoses. It is possible that physical symptoms could have inflated the SRQ score but we controlled for severity of pain and level of physical disability in the multiple regression analysis and still we found that the diagnoses of arthritis, other bone pathology, backache/pelvic inflammatory disease (PID) and major fracture were significantly associated with increased SRQ score. Our previous work indicates that the cut-off of 8/9 is the most appropriate to detect psychiatric disorder in primary care clinics in Pakistan. The high proportion with suicidal ideas testifies to the very high level of distress, and probable depressive disorder experienced by many of these patients. A further limitation of only using the SRQ as a measure of depression is the gender differences in the psychometric properties of the SRQ. In our earlier work a cut-off score of 5/6 was found to be better for males and, had we used this cut-off in the present study the prevalence figure for men would be higher. Since our main findings in this study are concerned with the total SRQ score, the cut-off is not affecting our main results concerned with the relationship between total SRQ and correlated features. We used the conventional cut-off so that our prevalence figures are consistent with all the previous studies from Pakistan, which have used the same cut-off scores for both men and women.
The main strength of the present study is the high response rate and we believe that our sample is not biased towards those patients who might have been prepared to undergo a more detailed research interview at the clinic visit, which would have been required to establish psychiatric diagnosis with certainty. As this is a cross-sectional study we cannot establish for certainty the level of distress that these patients experienced before their current illness, although we know the level of depression in the population is high . It is possible that they were depressed prior to their injury and actually experienced injury as a result of their depression. We were not made aware of any participants who had actually injured themselves deliberately; accidents are more common while someone is depressed and preoccupied. It is very likely that the injury and illnesses exacerbated pre-existing depressive illness.