SEARCH

SEARCH BY CITATION

Keywords:

  • Low back pain;
  • Health-related quality of life;
  • Psychosocial factors

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES

Objective

To investigate the respective contribution of various biologic and psychosocial factors, especially initial health-related quality of life (HRQOL), in the natural history of acute low back pain (LBP) and to evaluate the impact of this condition on HRQOL.

Methods

For 3 months, we assessed 113 patients consulting for nonspecific acute LBP of <72 hours duration at inclusion and treated with acetaminophen. Endpoints included pain, disability assessed by the Roland Disability Questionnaire, and HRQOL assessed by the Short Form 36 health survey (SF-36).

Results

Seventy-three percent of patients recovered within 2 weeks and 5% of patients developed chronic LBP. Prior low back surgery, higher initial disability questionnaire score, lower SF-36 score, and temporary compensation status were independently associated with delayed recovery. The impact of the acute LBP episode on HRQOL was brief and moderate, except for patients with comorbidity, psychiatric disorders, those of foreign origin, unemployed, or with job dissatisfaction. The impact of compensation status, sick leave, and bed rest was more profound and lasting.

Conclusions

This study highlights the large contribution of work-related factors, but also initial HRQOL, to the prognosis of LBP. It also suggests that LBP impairs HRQOL mainly through compensation and inappropriate medical care, and that, in turn, impaired HRQOL favors the condition becoming chronic. These findings have implications for future research into the management of LBP.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES

Low back pain (LBP) affects 80% of the population at some time (1), and is one of the most frequent reasons both for consulting a primary care physician and for taking time off work (2). The LBP “epidemic” (3), observed in most industrialized countries, and the huge resulting costs led to substantial research starting in the 1960s concerning determinants, preventive maneuvers, and treatments. However, this research mainly focused on biomechanical determinants and was largely sterile. In the meantime, “practice variations, treatment fads and rising disability” were observed (4). The recent US and UK guidelines for management, which propose a de-escalation in diagnostic and therapeutic approaches to this condition, illustrate the failure of the traditional, biomechanical paradigm for conceptualization and management of LBP (5). The role of psychosocial factors in the development of chronic disability and the demand for health care and financial compensation has increasingly become recognized (3, 6–11) and a biopsychosocial analysis of LBP has been suggested (12).

It would therefore be useful to explore health-related quality of life (HRQOL) in relation to LBP (13, 14). HRQOL measurement instruments have been developed over the past 20 years to assess self-perceived health status and its components, such as physical functioning (ability to carry out activities of daily life), psychological functioning (emotional and mental well-being), social functioning (relationships with others and participation in social activities), perception of health status, and pain (15). Measurements of self-perceived health status have been widely used to evaluate the broad impact of various diseases on patients and the effectiveness of interventions (15). In contexts other than LBP, these measurements have recently been shown to predict the outcome of surgical procedures and to be related to the extent of seeking and satisfaction with care associated with various conditions (16–18). This approach may be useful in understanding the natural history of LBP, and HRQOL measurements could help improve the clinical management of patients by extending the assessment process beyond traditional, and clearly insufficient, clinical and functional disability variables.

HRQOL has not previously been evaluated as a predictor of recovery following acute LBP. It is possible that self-reported HRQOL at initial presentation may be a valuable tool for assessing the prognosis of an acute LBP episode. Also, the time course of the impact of the acute LBP episode on HRQOL and its components has not been documented using standard generic instruments, such as the Short Form 36 health survey (SF-36). The identification of factors predictive of a higher impact of LBP on HRQOL would be of great value for defining management strategies.

We report a study with 2 goals: 1) to determine the contributions of biologic and several psychosocial factors, especially HRQOL variables, to the natural history of acute LBP and recovery; and 2) to evaluate the impact of acute LBP on HRQOL. To avoid the main biases of cohort studies (19), we assembled, and for 3 months carefully followed, an inception cohort of patients with acute LBP treated symptomatically and exclusively with acetaminophen.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES

Patients.

All consecutive patients aged 18 years and older, self referring to (i.e., first consultation with) participating general practitioners (n = 40) or rheumatologists (n = 7) for a primary complaint of LBP between April 1, 1999 and December 31, 1999 were eligible. Only patients with pain duration <72 hours and without radiation below the gluteal fold were included. Patients with malignancy, infection, spondylarthropathy, vertebral fracture, neurologic signs, or an LBP episode during the previous 3 months were excluded, as were patients who did not speak French or were illiterate. The resulting study group consisted of 113 subjects. No additional information was recorded from the eligible but not included subject group that mostly consisted of patients with specific LBP. All included patients gave written informed consent for the study, which had received ethical committee approval.

Baseline assessment.

Sociodemographic and occupational characteristics gathered at the first visit included whether or not of French origin (Caucasian subjects with French parents), temporary compensation status (which is systematically given in France for any LBP occurring in the workplace even without injury; this status allows for free health care and for wages to be maintained in the case of absence from work), job difficulty (heavy labor), and job satisfaction (the patient was asked a simple binary question, “Are you satisfied with your job?”). Medical data collected included medical and surgical histories (comorbidities currently treated were recorded), pain intensity (on a 100-mm visual analog scale [VAS]), pain duration and aggravating and relieving factors, standardized assessment of lumbar movements, and the straight leg raising test (7).

During the first consultation, the patients completed the validated French translation of the Roland Disability Questionnaire (RDQ) (20, 21), a self-administered questionnaire evaluating functional disability (score 0–24, with 0 corresponding to no limitation), and the standard SF-36 questionnaire (22). The SF-36 is a widely used generic HRQOL instrument that measures 8 health constructs: physical functioning, 10 items; role limitation due to physical problems, 4 items; pain, 2 items; general health, 5 items; vitality, 4 items; social functioning, 2 items; role limitation due to emotional problems, 3 items; and mental health, 5 items. All items refer to the subject's functioning or status during the past 4 weeks, except those of physical functioning and general health, which consider the moment of filling in. Scores range from 0 to 100, with a higher score indicating better HRQOL. We used the validated French version (23) and the standard scoring system proposed by Ware et al (22).

Management.

To optimize the description of the natural course of the LBP episode, general practitioners only prescribed oral analgesics containing acetaminophen. Other drugs, such as nonsteroidal antiinflammatory drugs; acupuncture; and physical treatments were not authorized. The prescription of bed rest and sick leave was left to the discretion of the physicians, and was recorded.

Followup and endpoints.

Patients completed a diary every evening from day 1 to day 14. It included a VAS evaluating the average back pain for the day and the RDQ. Followup visits were scheduled on day 8 and, if pain or disability persisted, on days 15, 30, 60, and 90. The data collected at each visit included the patient's evaluation of pain and disability. In the absence of a consensual definition of recovery, we defined it, according to our previous publication (7), as the (near) disappearance of both pain (pain ≤ 20 mm) and disability (RDQ score ≤ 3). The date of recovery was determined from diaries and clinical visit data. The followup ended at 3 months, the duration that consensually defines chronic LBP (24).

The SF-36 questionnaire was also completed on day 30 to explore HRQOL during and immediately after the LBP episode, and again on day 90 to explore it at a distance after the LBP episode (for most subjects).

Statistical analysis.

To help interpretation, the scores of SF-36 scales are presented as absolute and standard deviation scores (SDS), calculated by dividing the difference between the patient's score and the mean of the general population of the same sex and age group by the standard deviation of the general population group. French general population reference values for age (10-year interval groups) and sex were used (25).

The rate of recovery was analyzed by life-table analysis. Cox's proportional hazards models were used to study the prognostic value for recovery of each clinical and psychosocial factors and initial HRQOL variables, adjusting for the potential confounding effect of the delay between time of onset of symptoms and time of entry into the study. The final prognostic model of recovery was constructed in several stages. First, the model had to predict outcome from demographic and occupational information, and from medical and surgical histories. In the second stage, the model had to predict outcome from clinical information, comorbidities, RDQ, and SF-36 scores; and in the third, the medical prescription (bed rest, sick leave) was added to the significant independent predictors of the preceding stage (these variables were tested on an intention-to-treat basis). The last stage was to test interactions between the predictive variables of the final model. The delay between time of onset of symptoms and time of entry into the study was systematically included in the models.

Multiple regression models were constructed to predict changes in SF-36 scale (absolute) scores between baseline and days 30 and 90, from sociodemographic, LBP episode, and care variables. These models systematically included sex, age, and the delay between time of onset of symptoms and time of entry into the study (and the value of scale score at baseline to control for the regression to the mean in models predicting change). Some of the scores of the SF-36 scales were skewed and residual plots from the corresponding regression analyses were checked for departure from normality.

To construct these (exploratory) multivariate models of HRQOL scales, candidate variables were tested and entered into the model in a manual forward stepwise procedure (using the standard enter P = 0.05 and the remove P = 0.10).

All computations were performed with SAS software (26).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES

Baseline characteristics of the 113 subjects are shown in Table 1. Final evaluation (ascertainment of complete recovery or persisting pain at 3 months) was obtained for 111 patients (98%; 1 patient was lost to followup on day 7 and 1 on day 8). Complete SF-36 questionnaires were available for 108 patients for day 30, and for 106 for day 90. The median duration of the LBP episode was 8 days and 87% of patients had recovered by day 30 (95% confidence interval [95% CI] 79–95%). Six patients had not recovered by day 90, i.e., 5% (95% CI 0–9%) developed chronic LBP. The mean (SD) values of VAS and RDQ scores for these patients were 27 (29) mm and 8.0 (3.8), respectively.

Table 1. Baseline characteristics of the study subjects (n = 113) at entry into the study*
VariableScoreSDS
  • *

    Data are no. (%) unless otherwise noted. LBP = low back pain; VAS = visual analog scale; SF-36 = short form 36.

  • SDS = standard deviation scores from reference values.

Sociodemographic variables  
 Age, mean (SD) years44.3 (13.7) 
 Male sex55 (49) 
 French origin107 (95) 
 Manual work (blue collar worker)33 (29) 
 Employed at entry92 (81) 
Back pain history  
 Previous acute episodes, >074 (66) 
 Previous chronic LBP episode, >3 months4 (3) 
 Prior low back surgery5 (4) 
 Duration of index episode, mean (SD) days1.1 (0.8) 
Pain and disability variables  
 Initial VAS score, mean (SD) mm68 (14) 
 Permanent pain at night16 (15) 
 Pain increased by coughing42 (37) 
 Pain increased by back movement103 (91) 
 Pain worse when standing78 (69) 
 Pain worse when lying down36 (32) 
 Initial disability questionnaire score, mean (SD)12.7 (5.6) 
Physical examination  
 Pain on passive movement103 (91) 
 Catch49 (43) 
Work-related variables  
 Temporary compensation status15 (13) 
 Job difficulty (heavy labor)25 (24) 
 Job dissatisfaction37 (33) 
Comorbidities  
 Psychiatric disorder7 (6) 
 Other comorbidity35 (31) 
Care variables  
 Self referred to a general practitioner103 (91) 
 Prescription of sick leave47 (42) 
 Prescription of bed rest25 (22) 
SF-36 scale scores, mean (SD)  
 Physical functioning51.7 (30.6)−2.7 (2.6)
 Role limitations–physical problems58.2 (41.0)−1.0 (1.5)
 Bodily pain59.5 (26.0)−0.8 (1.2)
 General health68.8 (19.4)−0.1 (1.1)
 Vitality55.9 (21.1)−0.3 (1.2)
 Social functioning74.1 (23.8)−0.5 (1.2)
 Role limitations–emotional problems69.0 (39.0)−0.6 (1.3)
 Mental health63.8 (19.9)−0.3 (1.1)

HRQOL as a prognostic factor of recovery.

Several HRQOL variables, some pain characteristics, disability variables, and temporary compensation status were predictive of recovery as defined above (see Table 2). The final prognostic multivariate recovery model is shown in Table 3. Factors with an independent predictive value were prior low back surgery, RDQ score, temporary compensation status, and the SF-36 general health scale (a 10-point lower score, approximately half an SD, was associated with a 10% reduction in the probability of recovery).

Table 2. Prognostic factors for recovery among subjects with acute low back pain (n = 113)*
VariableHazard ratio (95% CI)P
  • *

    Variables are coded as shown in Table 1. The hazard ratio may be interpreted as the relative risk of recovery at any time within 3 months. A hazard ratio greater than 1.0 indicates that a higher percentage of patients with the characteristic than without recovered. A hazard ratio less than 1.0 indicates that a lower percentage of patients with the characteristic than without recovered. Hazard ratios and P values are adjusted for the delay (hours) between the onset of the low back pain attack and inclusion in the study in a Cox's model. 95% CI = 95% confidence interval; LBP = low back pain; SF-36 = short form 36.

  • Hazard ratio associated with one year older or 1-point higher score.

  • Hazard ratio associated with ten-point lower score.

Age0.99 (0.98, 1.01)0.46
Male sex0.94 (0.64, 1.38)0.76
French origin0.92 (0.40, 2.10)0.85
Manual work0.97 (0.63, 1.49)0.89
Employment at entry0.95 (0.58, 1.58)0.85
Previous acute episodes0.84 (0.55, 1.24)0.37
Previous chronic LBP episode0.40 (0.12, 1.28)0.12
Prior back surgery0.82 (0.65, 1.05)0.10
Pain intensity at entry0.99 (0.98, 1.01)0.31
Permanent pain at night1.16 (0.70, 1.90)0.57
Pain increased by coughing0.63 (0.42, 0.95)0.03
Pain increased by back movement0.81 (0.42, 1.58)0.09
Pain worse when standing0.76 (0.50, 1.15)0.19
Pain worse when lying down0.66 (0.44, 1.00)0.05
Disability questionnaire0.96 (0.93, 0.99)0.02
Pain on passive movements0.72 (0.29, 1.79)0.48
Catch1.04 (0.71, 1.54)0.83
Temporary compensation status0.51 (0.28, 0.97)0.04
Job difficulty0.93 (0.58, 1.50)0.76
Job dissatisfaction1.00 (0.68, 1.49)0.98
Psychiatric disorder0.98 (0.43, 2.24)0.96
Other comorbidity0.72 (0.47, 1.11)0.13
Self referred to a general practitioner1.10 (0.57, 2.13)0.78
Prescription of sick leave0.81 (0.54, 1.22)0.32
Prescription of bed rest0.91 (0.57, 1.42)0.69
SF-36 physical functioning0.95 (0.90, 1.02)0.15
SF-36 role limitations–physical problems0.96 (0.92, 1.01)0.09
SF-36 bodily pain0.93 (0.87, 1.00)0.04
SF-36 general health0.85 (0.77, 0.94)0.001
SF-36 vitality0.93 (0.85, 1.01)0.09
SF-36 social functioning0.95 (0.88, 1.03)0.18
SF-36 role limitations–emotional problems0.98 (0.93, 1.03)0.33
SF-36 mental health0.92 (0.83, 1.01)0.07
Table 3. Final prognostic model for episode recovery: hazard ratios by proportional hazards model*
VariableHazard ratio (95% CI)P
  • *

    The hazard ratio may be interpreted as the relative risk of recovery at any time within 3 months. A hazard ratio greater than 1.0 indicates that a higher percentage of patients with the characteristic than without recovered. A hazard ratio less than 1.0 indicates that a lower percentage of patients with the characteristic than without recovered. The final model included all listed variables and the delay (hours) between the onset of the low back pain attack and inclusion in the study, and the employment status. 95% CI = 95% confidence interval; SF-36 = short form 36.

  • Hazard ratio associated with one-point higher score.

  • Hazard ratio associated with ten-point lower score.

Prior back surgery0.36 (0.11, 1.17)0.09
Disability questionnaire0.97 (0.93, 1.00)0.05
Temporary compensation status0.57 (0.29, 1.09)0.08
SF-36 general health0.89 (0.80, 0.99)0.03

LBP and its impact on HRQOL.

Figure 1 presents the SF-36 scale scores at baseline, day 30, and day 90. At baseline, the mean score for the physical functioning scale was low, unlike the score for general health (both reflect current HRQOL). Interestingly, scores for other scales, which reflect the HRQOL during the month before the LBP episode, were also lower than those of the reference population.

thumbnail image

Figure 1. Short Form 36 scales scores at baseline (closed circle), day 30 (shaded circle), and day 90 (open circle). A, absolute scores. B, standard scores. Scores expected from population reference values are shown with a dashed line. PF = physical functioning; RP = role limitations due to physical problems; BP = bodily pain; GH = general health; VT = vitality; SF = social functioning; RE = role limitations due to emotional problems; MH = mental health.

Download figure to PowerPoint

On day 30, scores of the physical functioning scale, which measured current status (i.e., after the recovery from the episode for most subjects), showed persistence of moderate impairment. Scales other than the general health scale assess HRQOL during the past 4 weeks, i.e., during the LBP episode. The mean impact of LBP was small (a reduction of no more than 6 points or 0.3 SDS from baseline scores). On day 90, i.e., at a distance after the LBP episode except for 6 subjects who had not recovered, HRQOL scores were between −0.2 and −0.3 SDS for physical functioning, role limitation-physical, bodily pain, and general health scales, and about −0.05 SDS for the other scales. The HRQOL scores on day 90 for the 73% of patients who recovered within 15 days were similar to those of the reference population. Scores for patients who did not recover within 15 days were about 1 SDS lower for all scales at all times (Figure 2).

thumbnail image

Figure 2. Short Form 36 scale scores at baseline (A), day 30 (B), and day 90 (C) according to recovery from low back pain episode within 15 days. Thin line represents patients who recovered; thick line represents patients who did not recover. PF = physical functioning; RP = role limitations due to physical problems; BP = bodily pain; GH = general health; VT = vitality; SF = social functioning; RE = role limitations due to emotional problems; MH = mental health.

Download figure to PowerPoint

Predictors, measured at baseline, of change in SF-36 scores between baseline and day 30 are shown in Table 4. The variables directly related to LBP (duration of the episode, baseline RDQ score) only predict small changes on 1 or 2 scales. In contrast, larger changes on more scales were predicted by being of foreign origin, by receiving temporary compensation or sick leave, and to a lesser extent, by being unemployed, being a manual worker, having job dissatisfaction, or having psychiatric comorbidity. Predictors of changes between baseline and day 90 are shown in Table 5. Duration of the episode, but not the baseline RDQ score, predicted only small changes on 1 scale (bodily pain). Stronger predictors were foreign origin, receiving temporary compensation, bed rest, and comorbidity other than psychiatric.

Table 4. Final multiple regression models for changes in SF-36 scales, absolute scores between baseline and day 30*
 Physical functioningRole limitations–physical problemsBodily painGeneral healthVitalitySocial functioningRole limitations–emotional problemsMental health
  • *

    All models include age, sex, and the delay between the onset of symptoms and inclusion in the study. SF-36 = short form 36; LBP = low back pain; RDQ = Roland Disability Questionnaire.

  • The regression coefficient represents the difference in absolute score on the scale between the presence and the absence of the predictor; its 95% confidence interval is given in parentheses.

  • The regression coefficient represents the change in absolute score of the scale per unit change in the predictor; its 95% confidence interval is given in parentheses.

French origin17.3 (3.2, 31.3) 14.2 (−6.5, 34.9)10.4 (0.3, 20.6)19.5 (6.6, 32.4)22.5 (2.8, 42.1)  
Employed9.7 (1.1, 18.4)20.7 (−2.6, 44.0)      
Manual work−10.0 (−16.9, −3.3)       
Temporary compensation status −21.3 (−45.1, 2.4)−11.4 (−25.3, 2.6) −9.5 (−18.2, −0.8) −26.6 (−49.6, −3.5)−7.0 (−15.4, 1.4)
Job dissatisfaction −17.0 (−33, −0.7) −4.2 (−8.7, −0.3)    
LBP episode duration−0.2 (−0.3, −0.1)    −0.2 (−0.4, −0.1)  
RDQ score −2.1 (−3.5, −0.7)      
Psychiatric comorbidity−13.7 (25.7, −1.8)     −31.7 (−62.7, −0.7) 
Other comorbidity        
Bedrest        
Sick leave −20.8 (−38.9, −2.7)−16.1 (−25.8, −6.3) −6.7 (−12.8, −0.7)−9.4 (−18.2, −0.6)−20.7 (−37.3, 4.1) 
Table 5. Final multiple regression models for changes in SF-36 scales, absolute scores between baseline and day 90*
 Physical functioningRole limitations–physical problemsBodily painGeneral healthVitalitySocial functioningRole limitations–emotional problemsMental health
  • *

    All models include age, sex, and the delay between the onset of symptoms and inclusion in the study. SF-36 = short form 36; LBP = low back pain; RDQ = Roland Disability Questionnaire.

  • The regression coefficient represents the difference in absolute score on the scale between the presence and the absence of the predictor; its 95% confidence interval is given in parentheses.

  • The regression coefficient represents the change in absolute score of the scale per unit change in the predictor; its 95% confidence interval is given in parentheses.

French origin 24.4 (−4.3, 52.8)20.5 (2.2, 38.9)11.7 (−1.3, 24.8)13.6 (−0.1, 27.2)   
Employed 20.6 (1.6, 39.5)      
Temporary compensation status −18.0 (−36.2, 0.3)    −22.1 (−40.1, −4.1) 
Job dissatisfaction−9.8 (−18.7, −0.9)       
LBP episode duration  −0.2 (−0.3, 0.0)     
RDQ score        
Psychiatric comorbidity        
Other comorbidity−8.7 (−19.1, 1.8)−19.0 (−33.7, −4.3)      
Bed rest−8.9 (−18.9, 1.2) −8.0 (−17.2, 1.3)     
Sick leave    −6.4 (−12.7, −0.1)   

Results according to the type of physician.

Patients self referring to rheumatologists were similar to those self referring to primary care physicians with the exception of initial HRQOL variables: in particular, mental health and role limitation–emotional components were less impaired in patients self-referring to rheumatologists. However, the outcome in terms of recovery and of day 30 and day 90 HRQOL scores was similar. This absence of difference may possibly be explained by the fact that patients referred to rheumatologists were more frequently prescribed bed rest (50% versus 20%; P = 0.04).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES

This inception cohort study of acute LBP with complete followup yielded 2 major findings. First, it highlights the large contribution of several psychosocial factors, including work-related factors and especially HRQOL, to the prognosis of LBP. Second, it suggests that LBP impairs HRQOL, especially through temporary compensation and inappropriate biomechanical medical care and that, in turn, impaired HRQOL predisposes the condition to become chronic.

As in a previous study using the same inception cohort design (17), the recovery rate among patients with acute LBP was high: 90% in 2–4 weeks. This is mainly because these studies, unlike others, included truly incident cases and excluded exacerbated chronic LBP and sciatica, which are distinct conditions (24). Acute isolated LBP appears as a condition where the biopathologic course of the natural history is spontaneous recovery. Note that this biologic core of the condition is best assessed by a specific disability measure, such as the RDQ (which has a better predictive value than the SF-36 physical functioning scale and better than pain and clinical symptoms/signs assessment) (7, 27).

Previous back surgery is also a prognostic factor of the LBP episode. This is consistent with certain individuals being highly prone to develop chronic LBP, as previously suggested (7, 27–30). It may also indicate a delayed risk of chronic LBP following back surgery and a iatrogenic component to the process of becoming chronic.

Temporary compensation status appeared in this study, as in many others (7, 24, 28, 31, 32), to hamper recovery substantially. This result is consistent with the extensive scientific evidence against the concept of work injury or accident (33). Our finding is original that HRQOL, and especially the general health scale of the SF-36, is more predictive than objective measures of health, such as treated comorbidities. It suggests that the psychological perception of health makes a major contribution to whether LBP becomes chronic. Self-rated health, with psychological distress, was described as an important predictor of persistent disabling LBP in a previous study (9).

HRQOL also appears a determinant of consultation for LBP and therefore of its inclusion as a medical problem. Our finding of lower than expected scores for several HRQOL components during the month preceding the LBP episode is consistent with several other studies that showed that consultation for LBP may be determined by factors other than LBP symptoms and disability (34–36).

Note that none of the many physical symptoms or signs recorded in this study was predictive of recovery. This result suggests that the identification of the underlying mechanical cause of the pain, i.e., discogenic or facet disease, may be of little interest in this context.

The direct impact on HRQOL of the morbidity of the acute LBP episode, the biologic core itself, appeared short and moderate. Conversely, more profound and lasting was the impact of compensation status and biomechanically oriented medical prescription (sick leave, bed rest). The uselessness and even the negative effect of sick leave and bed rest (compassionate but iatrogenic) (37, 38) and temporary compensation has been demonstrated in acute LBP. These factors contribute to the memorization (33) or stigmatization (39) of LBP and appear to be highly deleterious, as impaired HRQOL was found to be predictive of delayed recovery. Also, the impact of LBP on HRQOL was more profound in medically frail (comorbidity, psychiatric disorders) and culturally or socially marginalized (foreign, unemployed, job dissatisfied) patients.

This study has several limitations. First, the sample studied cannot be considered representative of the general population of acute LBP patients. These unselected self-referred patients were only included if this episode was their first contact with the health system for LBP. Therefore, they constitute the least biased population of LBP patients entering the French health care system for LBP. Nevertheless, all the subjects sought medical care, and this may cause bias for various socioeconomic factors (40). In particular, patients with compensation or work issues were probably overrepresented in the sample, and this could make the apparent prognosis worse than it really is. In contrast, the exclusion criteria, including illiteracy, have led to a clear underrepresentation of poorly educated and foreign-origin patients, who may have poorer prognoses than studied subjects.

Second, the patients received no treatment other than acetaminophen and no recommendation was given to the practitioners regarding bed rest and sick leave. This choice was consistent with our research aim (to study the natural history and test the influence of these 2 factors), but limits the generalization of our results: many adjuvant treatments (of doubtful efficacy) are still widely prescribed, and recommendations for active strategies (to avoid bed rest and encourage early return to work) are increasingly applied (41).

Third, the validity of scores for several scales of the SF-36 need to be considered. Indeed, patients who at entry had recently suffered from LBP or recovered during the followup may have had difficulties rating their status during the 4 weeks preceding the time of evaluation and may have been influenced by their perceptions at the time of assessment. However, there is no indication in the published literature regarding the direction or the amplitude of this potential bias.

Fourth, many statistical tests were performed, either for the construction of the predictive recovery model or for the development of exploratory models predicting SF-36 scale scores. Thus, some associations may have been purely incidental.

Finally, our sample was small in view of the low prevalence of some exposure factors and the number of chronic LBP patients (n = 6). It is indeed difficult to gather an inception cohort of very recent LBP because most patients only consult for LBP after several days (36, 42). However, the inception cohort design is necessary to avoid left truncation biases (19), which, in most previous studies, made prognosis of LBP appear much worse than it actually is.

This study has implications for future research in patient management. It suggests that the de-escalation of the medical care of LBP initiated during the last decade (5) should be pursued. Few diagnostic or therapeutic actions have any proven value and indeed many others are iatrogenic (3–6, 41). Avoiding inappropriate medical treatment of acute LBP appears necessary. Primary care physicians may be in a better position than specialists to care for LBP patients because their activity is less based on biomechanical considerations. Medicalization itself is probably of no value for a large subgroup of subjects (young patients and those with LBP of short duration). For others, it has been suggested that yellow flags indicating psychosocial barriers to recovery should be carefully considered by physicians, as are red flags indicating signs of serious disease (43). In addition to the classic work-related factors, impaired HRQOL (evidenced with instruments such as the SF-36), appears to be a useful yellow flag worthy of further investigation.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES