Outcome of Surgery for Sciatica - a Comparison of Data From Three National Quality Registries

This is an observational register based study, with prospective data registration and a retrospective study design. Included are individuals treated surgically for a lumbar disc herniation between the ages of 18 through 65 years, without a history of previous lumbar spine surgery. Surgery has been performed in Denmark, Norway or Sweden during 2011, 2012 or 2013. Data will be presented according to the STROBE criteria.

The registers

All registries have the aim of studying outcome after spine surgery. All departments and patients participate voluntarily. At the time of admission, the patient reports data consisting of information on social factors, comorbidity and previous surgery. After surgery, the surgeon records diagnosis and type of surgery performed.

The Swespine Register has included individuals treated with surgery for lumbar disc herniation since 1993. During the last decade, the number of departments participating in the registry has varied between 35 and 41 of the 42 to 45 departments providing spinal surgery services in Sweden. Coverage is approximately 90%. The completeness (number of patients reported to Swespine at the time of surgery) is approximately 80%.

The Norwegian Spine register, NORspine, is based on experiences from the Swespine register and previous validation studies from a local clinical registry, and was founded in 2007. In total 36 of 40 centers performing lumbar spine surgery in Norway report to NORspine. Coverage is approximately 90%. The completeness is approximately 65%.

The Danish spine register, DaneSpine, is based on Swespine and was acquired by the Danish Spine Society from the Swedish Society of Spinal Surgeons in 2009 and has successively been implemented. In total 17 of 19 centers performing lumbar spine surgery in Denmark report to DaneSpine. Coverage is approximately 90%. The completeness is approximately 60%.

Quality assurance

As for all studies, there is a risk that loss to follow-up may bias the results. Solberg et al. (2011) studied 633 patients, who were operated on for degenerative disorders of the lumbar spine in Norway, and found that a loss to follow-up of 22% would not bias conclusions about overall treatment effects. There were no indications of worse outcomes in the non-responders group. In a similar one-center study of the DaneSpine. Højmark et al. (2016) found that a loss to follow-up of 12% at did not seem to bias the conclusions that can be drawn from DaneSpine at that center. Preliminary data indicates that predictors of outcome after lumbar disc herniation surgery are comparable with data in a study with a very high follow-up rate and with the Swespine register.

Data handling

Anonymized individual level data from all three registers will be pooled in one database. The cohort will be divided by country for comparisons.

Missing data and out of range data

In case of missing data case exclusion analysis by analysis, will be used. Out of range data will be deleted.

Analysis

The data will be cleaned by excluding patients with missing or incorrect date of surgery, missing date for follow-up, previous lumbar spine surgery and surgery other than discectomy only.

After data cleaning, we aim to perform blinded statistical analyses, in which the independent statistician performing the analyses is unaware of group belonging (i.e. country). The code will not be broken until the analyses and interpretations have been performed.

Comparisons of indications for surgery

Analysis of baseline data will include age at surgery, sex, anthropometrics, number of smokers, Oswestry Disability Index, numeric rating scale leg pain, numeric rating scale back pain, EQ-5D, number on sick leave, employment status, and duration of leg and back pain and presented as mean (SD), mean (95% confidence interval), or number (%).

Variables will be analyzed by analysis of variance, Chi-square or logistic regression tests. Data will be presented as crude (unadjusted) data to elucidate any differences between the countries.

Comparisons of outcome

Comparisons of the change of the outcome variables from baseline to 1 year, as well as comparisons of the actual value at 1 year will be performed. Analysis of covariance (ANCOVA), Chi-square or logistic regression tests will be used and the crude (unadjusted) data will be presented.

The baseline variables will be used as covariates in the ANCOVA and the adjusted data presented. Additionally, the results from the ANCOVA will be compared with results from propensity score matching or ANCOVA with the propensity score as a covariate.

Non-response analysis

A non-response analysis will be performed comparing all available baseline variables between those that responded to the 1 year follow-up with those that did not respond.

Sample size

A study of similar character has never been performed before. Due to the nature of the study, the sample size is not formulated in the guise of power, risk level, or clinical difference. The number of patients participating in the study is estimated to 7500. The sample is so large that differences in the Oswestry Disability Index of as low as 2 points may be detected (power 90%, significance level 5%), but in the interpretation the minimal important difference of 10-15 points in the Oswestry Disability Index often referred to has to be taken into account.