Operative procedures designed to result in a localized fusion of portions of the lumbar spine are done for various reasons, including infection, fracture, scoliosis, intervertebral disc disease, and mechanical instability. One of the most common complications is pseudarthrosis at one level or more, but the rates reported in the literature have ranged widely (from 8 per cent of 200 patients to four of six patients who had had a two-level arthrodesis)3,5,9,18,20,21,27,28,30,31,33. Some authors have contended that a pseudarthrosis after an attempted arthrodesis cannot automatically be assumed to be the principal cause of continuing symptoms and that operative repair therefore is not always mandatory5,6,29,37. However, it has been our experience, as well as that of others10,15,16,21,28,32,35, that a pseudarthrosis after an attempted arthrodesis in the lumbar spine does result in continuing pain and impairment and that patients may be helped by repair10,11,16. Pseudarthrosis can also lead to complications more serious than pain16,24.
An attempt at the operative repair of a pseudarthrosis is considered to be a salvage procedure in a patient who has already had at least one previous attempt at spinal arthrodesis. The reported results of such repairs have not always been encouraging, but most authors have believed that some improvement can be achieved5,6,11,15,16,24. The rate of successful repair has ranged from 49 per cent of forty-three pseudarthroses, to thirteen of seventeen patients who had a repair for a failed interbody arthrodesis, to twelve of fourteen patients who had a repair for a failed posterior arthrodesis5,15,29,30,38,39. Whether or not a solid fusion is achieved after repair of a pseudarthrosis has been reported to depend on many factors, such as smoking, metabolic disease, previous operations on the back, and the use of instrumentation3,14,15,29,30,33,36,38. However, we are aware of few reports in which the functional outcome of repair of a pseudarthrosis in the lumbar spine has been assessed from the viewpoint of the patient11,15,19, so we believed that a detailed study of functional outcomes was needed.
The purpose of this retrospective review was to assess the functional outcomes, with use of a detailed questionnaire, for seventy-two patients who had had at least one attempt at repair of a pseudarthrosis in the lumbar spine.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
†Read in part at the Annual Meeting of The American Academy of Orthopaedic Surgeons, New Orleans, Louisiana, February 28, 1994.
‡Madigan Army Medical Center, Fort Lewis, Washington 98431-5000.
§Orthopaedic Physicians, Incorporated, 1229 Madison Street, Suite 1600, Seattle, Washington 98104.
Eighty-six patients had a total of eighty-eight primary attempts at repair of a pseudarthrosis in the lumbar spine between January 1, 1986, and July 1, 1990. The diagnosis of pseudarthrosis was suspected because of continuing pain, was investigated with imaging studies2,4,7,12,26, and was confirmed operatively at the time of the repair. The pseudarthrosis was not classified according to type8, and three-dimensional computerized tomography was not used 13. The reason for the index attempt at lumbar arthrodesis was continuing low-back pain after a previous laminectomy for removal of a herniated disc or decompression in thirty-seven patients, painful degenerative spondylolisthesis in fourteen, pain associated with a positive discogram in fourteen, painful isthmic spondylolisthesis in thirteen, painful degenerative osteoarthrosis of the spine in ten, low-back pain associated with spinal stenosis in seven, traumatic spondylolisthesis in four, and disc herniation, scoliosis, and fracture in one patient each. (Some patients had more than one diagnosis.) All patients subsequently had a lumbar pseudarthrosis identified at one level or more. The preoperative symptoms, as determined from the questionnaire, were mostly back pain in thirteen patients, mostly pain in the lower limbs in two, and pain in both the back and the lower limbs in fifty-seven.
Seventy-six patients had a posterior repair and instrumentation, five had bone-grafting without instrumentation, five had an anterior and posterior repair during the same anesthesia session, and two had only an anterior repair. (Two patients had two procedures each.)
There were thirty-seven women and forty-nine men; the mean age was forty-five years (range, eighteen to seventy-five years). The patients had had a mean of 2.2 (range, one to four) previous operations on the lumbar spine at the same level as the repair of the pseudarthrosis. Detailed radiographs, including oblique radiographs and lateral radiographs made with the spine in flexion and extension, were available for eighty-six of the eighty-eight repairs. Three-dimensional computerized tomography was not used in this study.
A detailed written questionnaire, which included sections on pain, functional status, progress after the operation, satisfaction of the patient, smoking history, and work history, was completed by seventy-two (84 per cent) of the eighty-six patients at a mean of fifty-one months (range, twenty-five to seventy-eight months) after the operation. The portions of the questionnaire that were used to assess pain and functional status have been described previously17,22,34,35,38. A 100-point grading system was used, with 21 points assigned to pain; 40 points, to functional status (according to the disability index of Roland and Morris22); 18 points, to progress since the operation; and 21 points, to satisfaction of the patient (Table I). The Roland and Morris disability index is a series of twenty-four questions regarding activities of daily living. The result was graded as excellent (a score of 90 to 100 points), good (70 to 89 points), fair (50 to 69 points), or poor (less than 50 points).
Forty-two (58 per cent) of the seventy-two patients who completed the questionnaire had been receiving Workers' Compensation at the time of the repair of the pseudarthrosis.
Rate of Fusion
Radiographic data were available after eighty-six (98 per cent) of the eighty-eight repairs. There was radiographic evidence of fusion after seventy-one (83 per cent) of the index repairs (Figs. 1-A and 1-B). Subsequent reoperations on the fifteen persistent pseudarthroses eventually resulted in fusion in ten. Thus, eighty-one (94 per cent) of the index repairs for which radiographic data were available were ultimately successful. This is important, as the questionnaire, which was administered at a mean of fifty-one months after the first attempt at repair of the pseudarthrosis, also reflects the outcome in the patients who subsequently had at least one more attempt at repair. Of the ten patients who had a successful result after additional attempts, six had one attempt, two had two attempts, and two had three attempts before fusion was achieved (Figs. 2-A and 2-B).
Five patients never had a fusion after the initial and subsequent attempts to repair the pseudarthrosis. Four of these patients refused to consider any additional attempt at repair. The fifth patient had an additional attempt but still had a pseudarthrosis at the time of writing (Figs. 3-A, 3-B, 3-C through 3-D).
All of the patients in whom the initial attempt at repair consisted of an anterior or a sequential anterior and posterior repair had a solid fusion. With the numbers available, we found no significant association between a successful initial or subsequent repair and the patient's age, gender, body-mass index, return to work, or functional status. The likelihood of failure of the first attempt at repair to achieve a solid fusion increased as the levels that had been repaired increased (p = 0.02). This finding is consistent with those of previous studies of arthrodesis in the lumbar spine9,18,36 and of repair of pseudarthroses5.
Questionnaire
The scores of the seventy-two patients who completed the questionnaire ranged from 17 to 98 points. Seven patients (10 per cent) who scored between 90 and 100 points were classified as having an excellent result; twelve (17 per cent) who scored between 70 and 89 points, a good result; fourteen (19 per cent) who scored between 50 and 69 points, a fair result; and thirty-nine (54 per cent) who scored less than 50 points, a poor result. Four of the five patients who did not have a successful repair reported a poor result and the fifth, a fair result. The remaining thirty-five patients who reported a poor result had radiographic evidence of a solid repair. The outcome scores were plotted against the patient's age and body-mass index. With use of regression plot analysis, no significant association with obesity or age was found with the numbers available.
Fifty-one (71 per cent) of the seventy-two patients said that the operation had resulted in some improvement, and fifty-five (76 per cent) said that they would have the operation again if the circumstances were similar to those before the repair. Only twelve patients (17 per cent) said that the operation had made the symptoms worse. The patients who had a solid fusion at the latest follow-up evaluation had a score of 52 ± 24.6 points (mean and standard deviation), compared with 39 ± 13.9 points for those who still had a pseudarthrosis. However, because of the small number of patients who had a pseudarthrosis at the latest follow-up evaluation and the over-all large number of low scores, the difference between these scores was not significant. The five patients who had a persistent pseudarthrosis used more narcotics, as reported on the questionnaire, than those who had a solid fusion (2.5 points compared with 0.4 point; p = 0.04).
Work Status
Fifty (69 per cent) of the seventy-two patients stated that they had been working before the operation; a few had been receiving Workers' Compensation benefits or had had a job other than the one that they had postoperatively. Twenty-five patients reported that their previous job had involved strenuous labor (frequent lifting of thirty pounds [13.6 kilograms] or more, as defined on the questionnaire).
Only twenty (28 per cent) of the seventy-two patients reported that they were working at the time of the latest follow-up evaluation, despite having a solid fusion. Thirty of the remaining patients reported that they had retired because of the back symptoms; of these, twenty-three were still receiving disability compensation. Three patients had retired for reasons other than the continuing back symptoms, three were trying to reopen their Workers' Compensation claim, six were still being paid wages for their previous job even though they were not working (active time-loss), two were in retraining, and eight did not work outside of the home. None of the five patients in whom the repair had failed had returned to work. The patients who had had a job preoperatively that had required frequent lifting of thirty pounds (13.6 kilograms) or more were less likely to have returned to work (p = 0.03).
The patient's work status was not assigned points in the calculation of the outcome score. The mean outcome score for the twenty patients who had returned to work was 71 ± 22.6 points, compared with 43 ± 20.5 points for the fifty-two patients who had not returned to work or who had not worked outside of the home preoperatively. The thirty patients who had retired because of continuing back pain had a mean score of 39 ± 16.9 points, compared with 61 ± 24.9 points for the forty-two patients who had not retired for that reason. These differences were significant (p < 0.001 for both). The patients who had returned to work had a higher mean score on the portion of the questionnaire dealing with whether the operation had decreased the symptoms than those who had not returned to work. These patients also tended to have a higher score on the patient-satisfaction portion of the questionnaire, but, with the numbers available, the values for the two groups were not significantly different. The patients who had returned to work had had a mean of 1.7 ± 0.90 previous operations on the back at the same level as the repair, whereas those who had not returned to work had had a mean of 2.4 ± 1.14 such operations (p = 0.02).
Only nine (21 per cent) of forty-two patients who had been receiving Workers' Compensation at the time of the index operation returned to work outside of the home, compared with eleven (50 per cent) of twenty-two patients who had not been receiving Workers' Compensation (p = 0.007). The patients who had been receiving Workers' Compensation were also more likely to have retired because of the back symptoms (p = 0.004).
As a group, the patients who had been receiving Workers' Compensation did poorly on the scored portions of the questionnaire, which dealt with pain, functional status, progress since the operation, and patient satisfaction. They were more likely to have lower outcome scores, to check more responses on functional limitations, to give a higher rating to the worst pain that they had had, to have been a smoker at the time of the operation, and to be less satisfied with the outcome (Table II). However, with the numbers available, we found no significant association between the patient's Workers' Compensation status and the radiographic success of the repair of the pseudarthrosis. The patients who had been receiving Workers' Compensation were not more likely to be using narcotics.
Smoking History
Thirty-four (47 per cent) of the seventy-two patients who completed the questionnaire indicated that they had been smokers and thirty-eight (53 per cent), that they had not been smokers at the time of the index operation. The thirty-four who had been smokers had a smoking history of four to eighty-eight pack-years. (A pack-year is defined as a mean of one pack of cigarettes a day for one year. If three packs a day are smoked over a one-year period, that counts as three pack-years.) The mean outcome score for the thirty-eight non-smokers was 59 ± 25.1 points, compared with 48 ± 21.2 points for the twelve patients who had smoked less than one pack a day. The mean score for the seventeen patients who had smoked between one and two packs a day was 43 ± 23.9 points, compared with 35 ± 14.0 points for the five patients who had smoked more than two packs a day. This trend for the smokers to have lower scores than the non-smokers was significant (p = 0.048), according to analysis of variance. With use of regression plot analysis, the number of pack-years was plotted against the smokers' scores. Analysis of variance for the regression coefficients showed a definite negative linear association between the score and the number of pack-years (p = 0.02).
On the patient-satisfaction portion of the questionnaire, the smokers had a mean score of 11 ± 6.4 points (of a possible 21 points), compared with a mean of 14 ± 5.7 points for the non-smokers (p = 0.04). The non-smokers also had a higher score with regard to the extent of improvement after the operation (p = 0.05). There was also a difference in the number of pack-years between the patients who eventually had a solid fusion and those who did not. Four of the five repairs that failed were in patients who had been smokers at the time of the operation. These four patients had a history of a mean of fifty-one pack-years, compared with thirty pack-years for the thirty-four patients who had smoked and had a solid fusion; despite the small numbers of pseudarthroses, this difference was significant (p = 0.04).
Seven patients had stopped smoking before the operation. These patients had a mean outcome score of 65 ± 30.4 points, compared with 45 ± 20.5 points for those who had not stopped smoking at all or who had stopped only after the operation (p = 0.03). These seven patients also had a better mean score with regard to functional status (14 compared with 12 points; p = 0.03), were less likely to have been receiving Workers' Compensation at the time of the operation (p = 0.002), were more likely to have returned to work full time (p < 0.001), and were less likely to have retired because of continuing back symptoms (p = 0.048). Six of these seven patients had returned to full-time employment. None of the patients who had stopped smoking before the operation had a pseudarthrosis at the latest follow-up evaluation. The patients who had not stopped smoking until after the operation did not score differently than those who had not stopped smoking at all.
The patients who had been receiving Workers' Compensation at the time of the operation had important findings related to their smoking history. Twenty-seven (64 per cent) of forty-two patients who had been receiving Workers' Compensation had been smokers, compared with seven (23 per cent) of thirty patients who had not been receiving such compensation and for whom complete data were available (p < 0.001, chi-square test). In other words, patients who had been receiving Workers' Compensation were much more likely to have been smokers at the time of the operation than were those who had not been receiving such compensation (Table II).
Repair of a pseudarthrosis is a challenging task in terms of both technical difficulties and selection of patients. As not all pseudarthroses cause symptoms5,11,21,37, it is difficult to decide whether the patient's current symptoms are due to the pseudarthrosis or to some other problem. Proving the presence of a pseudarthrosis is not easy. Although several imaging techniques are available to assist in the diagnosis, none are unfailingly accurate2,4,7,8,13,36. We chose to include in this study only patients who had an operatively confirmed pseudarthrosis. Investigators who perform future randomized, controlled studies of outcomes after repair of pseudarthroses will have to use proved imaging techniques if a non-operatively managed group of patients is to be of any value as a control.
Eventually, there was radiographic evidence of a successful repair of eighty-one (94 per cent) of the eighty-six pseudarthroses for which radiographic data were available. All of the anterior and combined anterior and posterior repairs were successful, but there were too few for statistical comparison with the posterior repairs. Cleveland et al. reported the results for patients who had had a pseudarthrosis, 119 of whom had had a repair5. Thirty-five of the 119 patients had a reoperation for a persistent pseudarthrosis after the initial attempt at repair. The patients who had a solid fusion were generally free of disability, although these authors did not use a detailed functional-outcome questionnaire. Kim and Michelsen reported on twenty-nine patients who had had a repair of a pseudarthrosis; thirteen (45 per cent) had an unsatisfactory result, and eight (28 per cent) did not have a solid fusion10. Rothman and Booth assessed the results for thirty-nine patients who had had a repair of a pseudarthrosis; thirty-two said that the operation had been worthwhile23. This is in agreement with our finding that, in general, patients are satisfied with this procedure; however, those authors also did not use a detailed functional-outcome assessment23. Waddell et al. evaluated patients after a reoperation for a failed lumbar disc operation and found that those who had a pseudarthrosis had a poor result35.
Brown et al., in a study of 100 arthrodeses from the fourth lumbar to the first sacral vertebra, reported that the fifty patients who had smoked had a rate of pseudarthrosis of 40 per cent, whereas the fifty who had not smoked had a rate of only 8 per cent3. We also found a relationship between smoking and persistent pseudarthrosis with regard to the number of pack-years. Smoking was shown to have other negative effects, including a negative linear association with the outcome score. These data seem to indicate a dose-dependent relationship between smoking and a poor outcome. Recent experimental results of spinal arthrodesis in rabbits indicated a very strong relationship between pseudarthrosis and infusion of nicotine25. Our patients who were still smoking at the time of the operation had lower over-all outcome scores, reported less improvement after the operation, and were less satisfied. A small group of seven patients who had quit smoking before the operation had higher outcome scores, had fewer functional limitations, were more likely to have returned to work full time, and were less likely to have retired or to be receiving Workers' Compensation for the back symptoms. In summary, smoking negatively affected the functional outcome and the rate of fusion in our study. These findings are consistent with those of previous studies of primary arthrodesis3,30. Cessation of smoking preoperatively appeared to contribute to a better over-all outcome.
The patient-outcome portion of the questionnaire was designed to evaluate important measures of outcome, some of which have been used in the past to assess failures of spinal arthrodesis17,23,35,38. Thirty-nine (54 per cent) of our seventy-two patients who completed the questionnaire had a score of less than 50 points, which we consider to be a poor result, but we did not have preoperative ratings for comparison. However, fifty-one patients (71 per cent) reported that the operation had resulted in improvement, and fifty-five (76 per cent) said that they would have the operation again if the circumstances were similar. This relatively high rate of satisfaction probably indicates the severity of the condition before the repair of the pseudarthrosis. It also calls attention to the need for a controlled, prospective, randomized study, based on a functional outcome questionnaire that is administered both before and after the operation. We have developed such a preoperative questionnaire, which can be scored similarly to our postoperative questionnaire. The use of these questionnaires in prospective studies could provide important data on the effectiveness of repair of a pseudarthrosis. The postoperative questionnaire, which also includes assessment of subsequent operations, admissions to the hospital, physical-therapy visits, emergency-room visits, work data, and other variables, could also be used to perform a cost-benefit analysis of repair of pseudarthroses, as has been done for spinal arthrodesis32.
Work data are not a reliable indication of outcome, as they are affected by many other factors1,15,17,32,35,38. These patients have usually been managed for many years for the back symptoms and have had multiple operations; many consider themselves disabled, and very few are working. Our work data were obtained through the questionnaire, which asked: "How would you rate your job prior to surgery?" We thought that it was possible that more of our patients were disabled before the index operation than was reported on the questionnaire because it is probable that some patients considered their "job prior to surgery" to be the job that they had held before the original attempt at arthrodesis or before they had begun to receive Workers' Compensation, as opposed to the job that they had had immediately before the repair of the pseudarthrosis. This may have introduced a bias into the study. The questionnaire has been modified to clarify this question. Nevertheless, we do know that only twenty patients returned to work, and this is discouraging. Investigators in future studies must look carefully at this, but we believe that most of these twenty patients would not have returned to work without the repair of the pseudarthrosis, so the repair was worthwhile for at least these patients. Furthermore, each patient must be assessed carefully by the spinal surgeon, with not only the findings on radiographs and physical examination taken into account but also the patient's attitude, work status, smoking status, and motivation. Then, a reasonable decision can be made as to whether or not the pseudarthrosis should be repaired.
In summary, eighty-one (94 per cent) of the eighty-six repairs for which radiographic data were available led to a solid fusion, and most patients thought that the operation had been beneficial. The outcome, however, at least as measured with use of our detailed functional-outcome questionnaire, revealed that these patients were still quite limited in their functional abilities and had a noticeable amount of pain. Patients who had been smokers or who had been receiving Workers' Compensation at the time of the index operation had a lower rate of satisfactory results. The patients who had stopped smoking before the operation had much better results, and their status improved after the repair. The patients who had worked in a job that required heavy lifting were unlikely to have returned to work. Caution, therefore, should be used when counseling a patient preoperatively, and intensive rehabilitation and conditioning certainly should be attempted before this salvage procedure.
NOTE: The authors thank Troy Patience for the statistical analysis.