During the last two decades a steady increase in the annual number of patients with spinal fractures related to AS was found in Sweden. Although it has been suggested that through better medical treatment of AS the risk of suffering an unstable fracture should be reduced, this has not been found in this analysis of this registry data. The data from the SNHDR does not allow similar conclusions as a recently published study in 758 patients with AS by Vosse et al.  presenting reduced fracture risk if the patients were receiving medical therapy (OR 0.65). Swedish healthcare resource utilisation in patients with AS did not differ from other developed countries with 6–7 physician visits annually, thus improved medical therapy according to international guidelines should be expected during the last decades . Thus either the effect of medical therapy has not reached epidemiological significance, yet, or other underlying factors have to be put into consideration.
The first and possibly most obvious explanation would be an observational error, meaning a systematic bias during data collection. A recently published review on the SNHDR found high validity especially for surgical diagnoses . Validation of the SNHDR using national quality registries demonstrated that impressive 94% of all knee arthroplasty cases, 93% of all hip arthroplasty cases, and 95% of all hip fracture cases were correctly identified within the SNHDR. Similar numbers can be assumed for other orthopaedic diagnoses as vertebral fractures and AS. Furthermore in Sweden full reimbursement for treatment requires registration of codes for diagnosis and treatment in the National Hospital Discharge Registry. Therefore it can be assumed that coding errors are minimal.
A second possible reason for increased numbers of spinal fractures related to AS is an improved survival of patients suffering from unstable vertebral fractures. The metaanalysis of Westerveld et al.  found an overall mortality of 17.7% within the first three months after a vertebral fracture in AS, being 6.4% in the operatively treated and 11.3% in the non-operatively treated subgroup (n.s.). A recently published survival study by Schoenfeld et al.  found an increased mortality in patients with ankylosing spondylitis compared to controls even at time points up to 2 years after fracture. Optimised acute treatment during the last decades may have led to improved survival directly after injury, leading to more hospital admissions being registered in the National Hospital Discharge Registry. Unfortunately there is no data in the SNHDR on the mortality during the immediate post-injury phase.
A further explanation for the increase of spinal fractures in AS is an increasing level of activity with a reduced safety profile and greater risk for injuries. Multiple medical treatment strategies as well as physiotherapy interventions have been found to improve function and reduce stiffness in AS [8, 9]. It can be assumed that a new generation of patients with AS appeared during the last decades, being early under medical treatment and receiving optimised physiotherapy. These patients very likely have a quality of life similar to healthy individuals and experience ankylosis much later in life and at a much lower degree of kyphosis . Unfortunately once the biomechanical flexibility of the spine is declining, these still very active patients are prone to injuries, possibly leading to a greater number of spinal fractures .
Beyond that it is possible that patients with AS have nowadays a prolonged life span due to improved therapy. This would cause an increasing population of patients with AS due to reduced mortality. This hypothesis is supported by the finding that the number of registered patients with AS increased during the observed years (r = 0.67) (Table 1).
Finally, improved diagnostics identify earlier and more accurately AS. In the National Hospital Discharge Registry a relatively unchanged annual number of spinal fractures related to ankylosing spondylitis was seen until 2000. From 2001 onward a significant linear increase was found, suggesting either a greater spread of the disease, or - rather more likely – improved diagnosis of AS. The greater implementation of the New York-criteria for diagnosis of AS allowed a more standardised and homogenous identification of the disease . Possibly a certain number of patients with AS and vertebral fractures did not enter the study because the disease was not identified, yet.
The observed increase in total numbers of vertebral fractures in all regions of the spine is accompanied by a relative decline in cervical and lumbar fractures in favour of thoracic fractures (Table 1, Figure 1). In the last years there has been a positive trend towards the use of computed tomography instead of conventional radiographs especially in patients with AS . A decreased number of missed fractures could explain the relative increase in thoracic fractures, which can be hard to visualise on plain radiographs. Other possible explanations as a change in trauma patterns, and reduced ankylosis of cervical and lumbar spine are only hypothetical and lack any supporting evidence.
With regard to activity recommendations there exist diverging opinions on restrictions, but most authors agree that following general safety precautions are valid [17, 19, 20]: 1. The excessive use of alcohol should be avoided. 2. Contact sports (i.e. rugby, martial arts, ice hockey) should be avoided. 3. High impact sports (i.e. tennis, soccer) are not recommended during an acute inflammatory phase and in protracted stages of the disease. 4. Seat belts and car seat headrests should be used at all times while driving. Due to the increasing ankylosis and secondary osteoporosis restrictions for physical activity weigh stronger the older the patient is, and the longer he is suffering from AS.