For more than a generation, researchers have called upon the "biopsychosocial model" to explain disease in general [1]. In therapeutic circles, most back patients are probably, at least initially, considered from a pathoanatomical aspect, and in particular unsuitable ergonomic circumstances and excessive workload are considered relevant obstacles to recovery. Increasingly, however, psychosocial factors have come to outweigh biological factors in explanations of back pain, in particular for persistent back pain [2]. Even so, the relative importance of particular bio-psycho-social factors requires further studies. For example, smoking [3] and obesity [4] have been shown in many studies to be positively associated with back problems but it is not known whether this link is causal or an expression of some other underlying biological or psychological variable, which is linked with both smoking/obesity and back problems.

One factor that could be suspected of having such a dual link, coping, has attracted attention. People vary in their ability to cope with stress and their coping strategies differ as well [5]. Research in this area is extensive but has not resulted in a useful classification system, and the clinical validity of the concept is as yet unclear [6]. Coping is variously defined in the literature, and the literature on it is voluminous. Nevertheless, the classical definition of "coping", one at the core of most others, is "purposeful efforts to manage or modify the negative impact of stress" [7]. Comprehensive reviews have shown that coping in its various guises is related to the onset of chronic pain and outcomes of treatment for it [18]. Specifically regarding back problems, however, the literature is both sparse and conflicting [9–11]. We were interested in examining whether coping was associated with back problems and whether coping could explain some associations between various potential risk factors and back problems often reported in the literature.

Social class is another factor that is often implicated in ill health. Intellectual capacity and type of education can be considered proxy measures of social class, because both these would lead to greater or lesser societal opportunities and occupational exposures. In order to establish a possible link between these factors and back problems, it would be necessary to study a young population that has not yet been subjected to occupational exposures. Again, it would be interesting to see if coping plays an independent role in relation to back problems when studied in relation to these two basic determinants of social class.

Another issue to be taken into account in the explanation of back problems is co-morbidity. Studies of adults [12] as well as of children have shown that those with back problems are more likely also to suffer from other disorders [13]. This repeated finding raises another critical question: Are back problems but one of the possible phenotypic expressions of an inherent weakness? In other words, are the circumstances that bring about the onset of illness in general as well as back problems more or less the same?

Back problems have been shown to commence early in life [14] and early onset predicts further problems [15–17]. It is therefore important to concentrate the search for risk factors in the younger population so as to eventually be able to develop a primary prevention strategy.

We obtained a data set from the Swedish Army that offered an unprecedented opportunity to conduct a preliminary inquiry into such questions. It consists of cross-sectional data on an almost complete cohort of 18-year old men who had been summoned to serve in the military. About 48,000 individual cases were represented in the data for the year we analyzed (1998), which is an ample enough study population to take into consideration a variety of factors. Selection bias in these data is also minimized because military service is obligatory in Sweden and the full spectrum of society is called to serve. It is also important to note that data were collected prior to the selection process (accept/reject) had taken place, hence eliminating selection bias. Furthermore, almost all in the study population (91%) were still in school at the time they were summoned, which allows analysis of possible factors implicated in back problems prior to exposure to heavy physical workloads entailed in some occupations.

In short, the primary aim in this investigation was to explore how different types of factors are associated with the dependent variables back problems and ill health ("outcome") prior to entry into the labor force. Accordingly, we tested the effect on outcome of five independent variables that comprised two anthropometric factors (height and body mass index), two variables that we defined as psychological (intellectual capacity and coping in relation to stress), and one social variable (type of education). On the basis of these analyses, we then investigated which type of variables was most strongly associated with back problems. In particular, we investigated whether coping in our analysis has so much influence that other factors become negligible. We also compared the pattern of associations for the two outcome variables, back problems and ill health, to indicate whether each is a distinct condition or whether they both might have a common pathogenesis.

In many previous studies, where associations between potential risk factors and back pain were tested and in cases where control for extraneous factors was undertaken, it is difficult to understand how variables entered into the analyses were managed. For example, in most studies relating obesity to low back pain, there is not even a definition of the cut-point for obesity [4]. In our study we decided to follow the recommendations of Zahr [18], to look at the spread of data before deciding on the method of analysis which is most clinically relevant. For example, it would not be suitable to simply divide an independent variable into high and low or to use mean values, if the association is U-shaped or unevenly distributed. The large sample size in our study made it possible to include relevantly sub-classified variables in the analyses.

Some of our results were similar to those found in numerous other epidemiologic studies, namely some weak to moderately strong associations between a number of independent variables and BP. Our study, however, is exceptional in certain respects and has produced findings not commonly reported in the epidemiologic literature.

First, we noted that coping is strongly associated with BP and that this is an association that withstands the effect of the other variables, whereas the other variables that we tested typically became weakened when challenged with coping. We thus surmise that coping may not be a mere risk indicator, whereas the others most certainly are.

Second, we found that the statistical pattern is very similar for BP and IH in general. This finding suggests common aspects in the pathogenesis of both. In fact, it may imply that BP is but one of the expressions of IH in general. Coping, with an exceptionally high odds ratio, covers a large area under ROC for IH but only a small area in relation to BP. This, again, indicates that young men with BP may constitute a subpopulation nested within the group with IH.

These findings raise the question of why inability to cope is associated with BP and IH IH. We suggest three possibilities, with one not necessarily precluding the other. Regarding the first possibility, poor coping skills may serve to impede social and educational achievement. Early tracking and negative experiences of this nature would be a basis for a pervasive sense of powerlessness in social relations and a sense of not being in control of one's destiny–in short, self-recognition of being in the lowest social stratum, reliant upon the will of others. Studies from the literature on general health and illness have shown that a sense of powerlessness is associated with wear-and-tear on the body, diverse types of ill health, and self-reported poor health (22,23). The other possibility is that ill health may have a deleterious effect on coping, which becomes particularly apparent when examining young men's beliefs on their own ability to cope in war-like situations. The third possibility is that inability to cope and ill health may have a common origin, such as a genetic predisposition toward general frailty. Within the wider population, a certain subgroup may, consequently, be more prone toward diseases of various types, including back problems.

Methodologically, our study is unique in that it consists of a large and almost entire cohort, which excludes selection bias and allows for meaningful multi variable analyses. In addition, our cohort consists of young men, and data from them were collected prior to their entry into the work force, which enabled us to avoid the confusing element of occupational exposure. Finally, as stated earlier, the various variables in the study were collected independently of each other, and, it is important to note, the psychologists who assessed the inability to cope were blinded to data on health, including back problems. For this reason, their findings would have remained uncontaminated by the pre-existence of BP and IH.

Bias might be suspected in a study in the military sector. Conceivably, some of those called to serve may resist induction, and, in doing so, they may over-report disease and otherwise attempt to subvert induction tests. For the most part, however, we found dose-response relationships between the outcome and independent variables, which would be unlikely, had willfully poor performances subverted test scores to a substantial extent. Also, the supply of young men at enrolment is far larger than the demand for manpower, with only less than half of those who went through the enrolment procedures in 1998 actually later being inducted into the military. For this reason, over-reporting of ill health in order to avoid military or civil service is hardly necessary. Willingness to serve in the military varies over time, but military forces considered it to be quite high when data for this study were collected. Under-reporting of ill health by some young men intent on a military career may therefore be more likely.

This study, however, has two notable weaknesses. First and obviously, our cross-sectional data collected at one point in time cannot be used to test a cause-effect but mere associations (24). Second, although the most important independent variable, coping in relation to stress, has been found to be a valid predictor of soldiers' performance during their military service (20) its exact contents and definition are unknown to us. If the Swedish Army defines coping in relation to physical stamina in war-like situations, then the presence of certain diseases, such as asthma, might induce a non-coping response. In other words, ill health and poor coping may be but expressions of the same phenomenon. This might explain the very high association between coping and IH.

Interesting as our results may be, they cannot really be taken seriously, until the coping variable is properly defined and described but, unfortunately, the Swedish Army is not able to divulge its interview questions. Obviously, the link between coping and BP/IH must be tested with a known coping instrument. It is also always necessary to challenge the results of any study in other settings. We therefore proceeded to conducting a prospective observational study of the development of back problems in military recruits, who were investigated for coping ability prior to the first three months of military service, using a well known coping questionnaire. The results of this second study, which has been completed, will be reported separately.

Our conclusions, at this time, are:

• Coping, as defined by the Swedish Army, is strongly associated with back problems in young Swedish men.

• In this cohort, coping, as defined by the Swedish Army, is very strongly associated with ill health.

• In this cohort, coping, as defined by the Swedish Army, has a strong effect on the association between height, body mass index, intellectual capacity, type of education and the outcome variables back pain and ill health.

• In this cohort, back pain and ill health, as defined by the Swedish Army, have a similar pattern of associations.

• The "black box" of coping, as defined by the Swedish Army, must be defined and described.

• These findings need to be reproduced in prospective studies, in order to test for causality.