Lumbar spinal stenosis, assessing failure and worsening after surgery. Identifying predictive factors with critical use of data from a national spine registry (NORspine)
Abstract
Norsk sammendrag
Spinal stenose er en vanlig lidelse som skyldes trang ryggmargskanal og karakteriseres av smerter i rygg og bein og redusert gangfunksjon. Operasjon er ofte nødvendig og spinal stenose er den hyppigste årsaken til ryggkirurgi i Norge (3, 23). Resultatene etter kirurgi er noe sprikende: de fleste blir bedre, noen blir ikke bedre, og enkelte blir verre (5, 6, 7).
Det er utfordrende å måle resultat etter behandling for smertetilstander fordi det ikke finnes klare konkrete endepunkt. Pasient-rapporterte resultater er sentrale, man kan bruke smerteskalar eller spørreskjema på funksjon og livskvalitet. Fortolkning av svar i skala-form kan være krevende, det fordres en viss endring på skalaene for at endringen skal være klinisk relevant. Man kan lette fortolkningen med å lage kategorier der pasientene klassifiseres som enten bedre, uendret eller verre.
Vi har brukt data fra Norsk kvalitetsregister for ryggkirurgi og analysert pasienter operert for spinal stenose. Registeret inneholder data om pasientforhold og plager før operasjon, operasjonstekniske forhold og resultater 3-og 12 måneder etter operasjon. Gjennom dette har vi tilegnet oss ny kunnskap om spinal stenose pasienter.
Registerdata er beheftet med flere usikkerhetsområder, mange pasienter faller fra og svarer ikke på oppfølgingene, og data kan i tillegg bli feilregistrert. Vi har derfor undersøkt kvaliteten på registerdata som ble brukt i denne doktorgradsavhandlingen.
Det er kjent at ikke alle pasienter blir kvitt plagene etter kirurgi for spinal stenose, og vi fant at om lag 20% rapporterte at plagene var uendret eller verre etter kirurgi. Videre fant vi de grenseverdiene som definerte mislykket kirurgi (uendret eller verre) og forverring på de mest brukte skalaene med størst nøyaktighet.
Vi testet samsvar av registerdata ved å kontrollere opp mot journaldata og fant at datakvaliteten i registeret var vekslende. Pasientrapporterte data og operasjonstekniske faktorer hadde høyt samsvar, mens andre helseforhold og komplikasjoner hadde dårligere samsvar med pasientjournalen. Pasienter som ikke svarte på oppfølgingsskjemaer fra registeret skilte seg noe fra de som svarte; de var litt yngre og oftere røykere. Resultatene etter operasjon var like i de to gruppene.
Den vanligste komplikasjonen til kirurgi for spinal stenose er rift på nervehinnen, dette medfører lekkasje av spinalvæske, eksponering av nervetråder og noen ganger behov for reoperasjon og forlenger sengeleie. Vi fant noe dårligere resultater etter operasjon hos pasienter som fikk rift på nervehinnen.
Det kan være vanskelig å beslutte om kirurgi er riktig for den enkelte pasient. Vi identifiserte noen faktorer som øker risikoen for mislykket kirurgi og forverring (alder over 70 år, tidligere ryggkirurgi og ryggsmerter over 12 mnd., samt noen sosioøkonomiske variabler). Disse faktorene kan bidra til bedre pasient informasjon og slik gi støtte til beslutning om operasjon eller ikke operasjon.
Vi håper våre resultater er nyttige for klinikere og at de bidrar til bedre informasjon til pasienter samt gode behandlingsvalg. Vi håper også resultatene kan gi grunnlag for videre forskning på ryggkirurgi. English abstract
Background
Results after surgery for lumbar spinal stenosis (LSS) vary; most patients improve, but some do not, and some even worsen. Some patients also suffer from complications. Previous studies have identified certain factors that may predict outcomes after surgery for LSS. Development in surgical technique may have reached a ceiling because new techniques fail to prove better; this emphasizes focus on careful patient selection to improve the overall results.
National medical registries collect a large number of data and reflect daily practice. Because of the large number of participants, registry studies are optimal for studying complications of surgery. However, registry data are vulnerable to wrong recordings and loss of follow-up. Hence, registry data should be assessed for bias before conclusions are drawn.
Methods
We reviewed patients operated on for LSS in Norway for ten years (2007-2017). Prospectively collected data from the NORspine registry was the foundation of the observational studies included in the thesis. We also supplemented registry data with data from patient records and performed a cross-sectional study.
We included patients treated over two years from four hospitals to assess data accuracy. Data was re-captured from electronic patient records, and we assessed the agreement between the two data sources using kappa statistics.
To assess potential bias due to loss to follow-up, we compared baseline variables between patients completing follow-up and those who did not. We also contacted patients lost to follow-up to see if they reported different clinical outcomes. We used simple descriptive statistics and compared baseline data and clinical outcomes between the groups with student T-tests.
We defined criteria for failure and worsening using a transition scale (Global Perceived Effect (GPE)) as an external anchor and receiver operating characteristic (ROC) curve analyses to identify the best cut-offs on PROMs commonly used to assess the effect of spine surgery. We also studied if a dural tear affected the clinical outcome, defined as failure or worsening, using logistic regression analyses and adjusting for possible confounding factors.
Finally, we tried to identify variables that could predict failure and worsening using multiple logistic regression analyses with the cut-offs identified earlier in our project. We selected baseline variables with acceptable accuracy according to an early part of our project.
Results
The study population comprised 11873 patients, and 8919 (75%) completed 12 months of follow-up. We reviewed 474 patient records to assess NORspine accuracy and the impact of loss to follow-up.
Patient-recorded variables and surgeon-reported surgical details displayed moderate to good accuracy; however, surgeon-reported complications and comorbidity were underreported. Patients lost to follow-up were younger and, more often, were smokers. However, there were no statistically significant differences in clinical outcomes. The following PROM cut-offs most accurately defined patient-reported failure (and
worsening): ODI final score of more than 31 (39), ODI percentage improvement of less than 20% (9%) and ODI improvement of less than 8 (4) points. These cut-offs had good to excellent accuracies (AUC= 0.86-0.91).
Dural tears occurred in nearly 5%. Patients who suffered a dural tear increased the odds of failure (and worsening) with an odds ratio of 1,45 (1,50).
After LSS surgery, a proportion of 33 % was defined as failure and 22 % as worse. Age over 70 years, previous spinal surgery, and duration of back pain over 12 months were essential baseline variables associated with failure and worsening (Odds ratio 1,85 – 2,21); socioeconomic factors also affected the odds for failure and worsening (OR 1,26 – 1,67).
Conclusions
There are concerns regarding data quality in the spine registry; data should be used and interpreted with care. Patients lost to follow-up reported similar clinical outcomes as those who completed follow-up, and missing data from loss to follow-up can most likely be treated as missing at random. Cut-offs for failure and worsening are accurate and can be used in future research and clinical work. LSS patients over 70 years, with previous spine surgery and duration of back pain over 12 months, had increased odds for failure and worsening; this could aid in patient selection.
Has parts
Paper 1: Alhaug, Ole Kristian; Kaur, Simran; Dolatowski, Filip Celestyn; Småstuen, Milada Cvancarova; Solberg, Tore; Lønne, Greger. Accuracy and agreement of national spine register data for 474 patients compared to corresponding electronic patient records. European spine journal 2022 ;Volum 31. s. 801-811 https://doi.org/10.1007/s00586-021-07093-8 This article is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0)Paper 2: Kaur, Simran; Alhaug, Ole Kristian; Dolatowski, Filip Celestyn; Solberg, Tore; Lønne, Greger. Characteristics and outcomes of patients who did not respond to a national spine surgery registry. BMC Musculoskeletal Disorders 2023 ;Volum 24.(1) s. 1-7 https://doi.org/10.1186/s12891-023-06267-3 This article is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0)
Paper 3: Alhaug, Ole Kristian; Dolatowski, Filip Celestyn; Solberg, Tore K.; Lønne, Greger. Criteria for failure and worsening after surgery for lumbar spinal stenosis: a prospective national spine registry observational study. The spine journal 2021 ;Volum 21.(9) s. 1489-1496 https://doi.org/10.1016/j.spinee.2021.04.008 This article is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0)
Paper 4: Alhaug, Ole Kristian; Dolatowski, Filip Celestyn; Austevoll, Ivar Magne; Mjønes, Sverre Markussen; Lønne, Greger. Incidental dural tears associated with worse clinical outcomes in patients operated for lumbar spinal stenosis. Acta Neurochirurgica 2022 ;Volum 165.(1) s. 99-106 https://doi.org/10.1007/s00701-022-05421-5 This article is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0)
Paper 5: Alhaug, Ole Kristian; Dolatowski, Filip Celestyn; Solberg, Tore K.; Lønne, Greger. Predictors for failure after surgery for lumbar spinal stenosis: a prospective observational study. The spine journal 2022 ;Volum 23.(2) s. 261-270 https://doi.org/10.1016/j.spinee.2022.10.010 This article is available under the Creative Commons CC-BY-NC-ND