Intended for healthcare professionals

General Practice

Community survey of factors associated with consultation for low back pain

BMJ 1998; 317 doi: (Published 05 December 1998) Cite this as: BMJ 1998;317:1564

This article has a correction. Please see:

  1. Robin Waxman, research officer,
  2. Alan Tennant, principal research fellow,
  3. Philip Helliwell, senior lecturer in rheumatology
  1. Rheumatology and Rehabilitation Research Unit, University of Leeds Research School of Medicine, Leeds, LS2 9NZ
  1. Correspondence to: Ms Waxman
  • Accepted 5 December 1998


Objective: To investigate the psychosocial factors associated with consultation for low back pain.

Design: Two phase cross sectional postal survey.

Setting: Bradford Metropolitan Health District.

Subjects: 1813 adults responding to the phase 1 questionnaire. 540 of the 782 with an episode of low back pain in the past 12 months completed the second questionnaire.

Main outcome measures: Six psychosocial constructs.

Results: 406 (52%) of the respondents reporting back pain in the past 12 months had not consulted a health professional. Logistic regression showed that consultation was associated with externalised beliefs regarding pain management (odds ratio 3.6; 95% confidence interval 2.1 to 6.0). Duration of pain affected the factors associated with consultation. Consultation for episodes lasting less than two weeks (n=290) was associated with greater than median pain (3.0; 1.7 to 5.5), consultation for episodes over two weeks (n=243) was associated with increased disability (3.7; 1.5 to 9.0), and consultation for episodes over three months (n=143) with increased depression (3.9; 1.3 to 11.8).

Conclusions: The results support a role for psychosocial factors in consultation for low back pain and suggest that the reasons for consultation vary with duration of pain. Duration of the episode may be a useful guide to management of non-specific low back pain.

Key messages

  • Only half of those who experience low back pain consult a health service professional

  • Most of those who consult have had low back pain for more than two weeks

  • The severity of pain may influence consulting behaviour at the acute stage, but after the acute period those who consult are more likely to have increased disability, externalised beliefs about pain management, and depressive symptoms.

  • Management of low back pain may need to vary according to the duration of pain


A community survey of low back pain and consultation behaviour in Bradford Metropolitan Health District in 1994 found that half of those who reported an episode of low back pain in the past year had not consulted a health professional for the problem.1 Severe pain was reported by respondents who did and did not consult, which suggested psychosocial factors might have a role. This hypothesis is supported by a Swedish study that found few clinical or radiographic differences between patients who did and did not consult for back pain.2Most studies that have investigated care seeking for low back pain have examined only physical and social factors.37 Wright et al, however, found that consultation was associated with general health questionnaire scores above the threshold for psychiatric morbidity.8

We conducted a further survey of the Bradford population to examine the non-physical aspects of consulting behaviour including cognitive strategies, health beliefs, and psychological wellbeing. Knowledge of the precise reasons for consultation with low back pain should help inform primary care management and indicate the mix of services required to cope with the increasing numbers of patients disabled by low back pain.

Subjects and methods

With the approval of the local ethics committee we recontacted the 3184 valid respondents to the 1994 Bradford study.1 The original study was based on an age stratified random sample of Bradford residents aged 25-64 derived from the family health services authority population register. Full details of the methods have been published.1According to 1997 Yorkshire Regional Health Authority records, about 18% of the sample had changed address, resulting in a valid sample frame of 2598.

We used a two phase postal questionnaire. Phase 1 directed recipients towards a picture of the back with a shaded area between the lowest rib and the gluteal folds and asked the following questions: have you ever had back pain in the shaded area which lasted for more than a day? and, if yes, have you had this back pain in the past 12 months?

Respondents who answered yes to these questions were then asked to tick all health professionals they had visited for their back pain in the past 12 months from a list. We classified respondents as consulting if they reported visiting their general practitioner, hospital doctor, workplace doctor or nurse, pain clinic, or accident and emergency department. Access to physiotherapy in the NHS is usually by referral from primary or secondary care medicine. Respondents who ticked only a physiotherapist, osteopath, or chiropractor were considered to have self referred and classified as non-consulting within the NHS.

Respondents who reported having had an episode of low back pain in the past year were sent a phase 2 questionnaire which obtained information on characteristics of low back pain, tests and treatments, work and benefits, health and social support, plus measures of six psychosocial constructs and one measure of disability. The information gathered from this questionnaire formed the basis of this study.

We assessed the prevalence of back pain in three ways based on responses to the phase 1 questionnaire: lifetime prevalence (number of respondents who had ever experienced back pain lasting more than a day per 100 population aged 25-64 in 1994), annual prevalence (number who had had back pain in past 12 months), and point prevalence (number who had back pain on the day they completed the second questionnaire).

Psychosocial measurements

We selected instruments that had been validated in populations with low back pain and that identified characteristics of coping and health beliefs that might distinguish thosewho did and did not consult. We used the Roland and Morris disability questionnaire, the fear avoidance beliefs questionnaire, the coping strategies questionnaire (cognitive coping and helplessness dimensions only), the pain locus of control questionnaire, the modified somatic perceptions questionnaire, the modified Zung depression index, and the back beliefs questionnaire.915


We used logistic regression to model the association between prognostic variables and consulting. All variables were entered into the model in a single step, allowing a maximum of 10 cases per variable category. 16 17All variables were grouped into predisposing, perceived need, and psychosocial factors based loosely on Andersen's behavioural model of health services use (box).18Predisposing and perceived need factors were modelled separately, and those variables that contained at least one category with a significance of P<0.1 were combined and analysed with the psychosocial factors. Prognostic index variables were derived for predisposing and perceived need factors and used to model subgroups based on duration of pain, and the index variables were then combined with psychosocial variables.16 The final model contains only variables with a significance <0.05. Age group and sex were retained throughout regardless of significance.

View this table:


We had a 70% (n=1813) response rate to the phase 1 screening questionnaire. This showed a 62% (95% confidence interval 59% to 64%) lifetime prevalence, 43% (41% to 45%) annual prevalence, and 21% (19% to 23%) point prevalence of low back pain among the adult population of Bradford (table 1). Of the 782 respondents to the phase 1 questionnaire who reported having had low back pain in the past 12 months, 406 (52% (49% to 56%)) reported that they had not consulted an NHS professional. One hundred and eleven (29%) of those who had consulted had also visited a physiotherapist, osteopath, or chiropractor. Only 37 (<5%) people reported consulting only a physiotherapist, osteopath, or chiropractor.

Table 1.

Population prevalences (95% confidence intervals) for back pain in Bradford (rate per 100 people aged 25-64 in 1994)

View this table:

We received 540 valid responses (69%) to the phase 2 questionnaire. We found no significant difference between respondents and non-respondents in terms of consulting for low back pain, but women and those aged over 35 were significantly more likely to respond, resulting in proportionally more women and fewer people under the age of 35 compared with the 1997 adult population of Bradford.

Among the 540 respondents, the odds of consulting were significantly increased by having ever had the cause of low back pain diagnosed, being unemployed or retired, having first ever episode of low back pain in the previous year, having greater than median worst pain, and having externalised locus of control for pain management (table 2).

Table 2.

Odds ratios for consultation for low back pain for 540 respondents*

View this table:

Table 3 shows the results of univariate analyses by duration of low back pain. Because these groups displayed different prognostic indicators for consultation behaviour we decided to analyse acute (duration ≤two weeks), non-acute (>two weeks), and chronic cases (>three months) separately. Non-acute cases comprise subacute plus chronic cases.

Table 3.

Effect of severity and periodicity of pain on consultation for low back pain according to duration of episode

View this table:

Consultation for episodes of less than two weeks was associated with having ever been given a diagnosis for low back pain, being unemployed or retired, first ever low back pain episode in the previous year, and greater than median worst pain (table 4). Psychosocial factors were not significant. Consultation for episodes lasting more than two weeks was associated with having ever been given a diagnosis for low back pain, externalised locus of control for pain management, and a greater than median Roland and Morris disability score. Predisposing factors were not significant. Consultation for episodes of more than three months was associated with external locus of control for pain management, and a greater than median Zung depression score. Predisposing and perceived need factors were not significant.

Table 4.

Significant odds ratios for consulting with low back pain according to duration of episode*

View this table:


Half of those who had experienced low back pain in the past year in Bradford consulted a health service professional. Horal in Sweden, and Wright et al and Dodd in the United Kingdom found similar rates of consultation for low back pain. 2 8 9 However, many more people with acute low back pain (71% in this survey) do not consult. The factors associated with not consulting in this group are intuitively sensible—less severe pain, employment, and previous experience of back pain which may have resolved spontaneously.

Consultation for chronic low back pain, however, may be governed by depressive symptoms rather than pain characteristics, which is important for management. Furthermore, studies have shown that depression is more strongly associated with impaired function and the failure to resume normal activity than intensity of pain. 20 21 Our results support this hypothesis.

The Clinical Standards Advisory Group's guidelines for back pain emphasise a biopsychosocial approach to management.22 Primary care is identified as having a pivotal role in control of pain and prevention of disability, and our results would support this for episodes lasting less than three months. However, the guidelines also state that if patients have not returned to work within three months primary care management has failed and chronic pain and disability are likely. Our results suggest that assessment of patients should include duration of low back pain. Patients who present with low back pain lasting more than three months may require referral to psychological support services. Half of all serious cases of depression go unrecognised in primary care, especially those where patients present with physical symptoms. 23 24 In addition, patients often focus on physical symptoms rather than the psychosocial consequences of their illness. 25 26

Significant associations found in epidemiological research may not be causal, partly because the time sequence of events cannot be established. This is especially true in observational studies where the results of modelling of complex relations between aetiology and outcome may be due to the effects of chance associations or unidentified confounding factors. Logistic regression allows us to control for a number of factors simultaneously, but the logistic regression model also has a weakness in that complex matters are presented as a simple equation. We controlled for several possible prognostic factors in our modelling but restricted our analysis to six psychosocial constructs in addition to several clinical and demographic variables.

Although not fully representative of consultation patterns for low back pain, our model of consulting behaviour gives a useful insight into an unexplored dimension of low back pain. Future research might explore the efficacy of treating the factors we found to be associated with consultation for acute, non-acute, and chronic low back pain.


  • Contributors PH conceived the study and together with AT and G Rajaratnam received the grant that funded it. RW was responsible for planning and coordination, data collection, data analysis, and writing the paper with the supervision of PH and AT. A Vail gave statistical and methodological advice and M Tennant helped with statistical analysis. PH, AT, and M A Chamberlain advised on the finished paper. The patients and staff of the Ling House Surgery, Keighley, provided a pilot setting. Members of the Bradford Branch of the National Back Pain Association gave us their views on the questionnaire and the results.

  • Funding Northern and Yorkshire Regional Health Authority.

  • Conflict of interest None declared.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.