The prevalence of serious pathology in musculoskeletal physiotherapy patients – a nationwide register-based cohort study

Background: Musculoskeletal conditions are the single largest contributor to years lived with disability worldwide. Most musculoskeletal conditions can be managed and treated in primary care, but for a small proportion of these patients the symptoms are caused by serious pathology. Although the general practitioner usually performs initial screening for serious pathology, evaluation and treatment by physiotherapists are often part of the treatment pathway. It is however unclear, how many patients in primary care physiotherapy have symptoms caused by a serious pathology. Historically the prevalence of serious pathology in primary care has been investigated in small populations with spine specific conditions, thus a more general prevalence in the group of patients with musculoskeletal conditions is yet to be estimated. Therefore, the aim of this study was to estimate the prevalence of neoplasm, cauda equina syndrome, spinal fracture, infection and inflammatory pathology among patients referred for musculoskeletal physiotherapy evaluation and treatment. Methods: The study was a prospective nationwide register-based cohort study. We identified all referrals for primary care musculoskeletal physiotherapy in the Danish National Health Insurance Service Register between 1 January 2014 and 31 December 2017. Records of hospital contacts were extracted from the Danish National Patient Register within 180 days from first physiotherapy contact, identifying all diagnoses of serious pathology. Period prevalence proportions with 95%CI of the serious pathology categories were estimated. Results: A total of 1,568,704 courses of treatment were included in the analysis. The overall prevalence of serious pathology was 2.30%.The prevalence of neoplasm was 2.11%, of which 1.13% was malignant neoplasms. The prevalence of cauda equina syndrome was 0.01%, fractures 0.13%, infections 0.01% and inflammatory pathology of

3 the spine 0.06%. Higher prevalence's were observed among patients with a previous history of serious pathology, aged above 50 and more comorbidity.

Conclusions:
Although serious pathology among musculoskeletal physiotherapy patients is rare, the present study found an overall prevalence of serious pathology which exceeded the guideline endorsed prevalence estimates of serious pathology of 1%.

Background
Musculoskeletal conditions are the single largest contributor to years lived with disability worldwide (1). These conditions are typically characterised by pain and disability, which may have substantial consequences for the affected individual causing reduced ability to work or limited participation in social activities (2). Most of the musculoskeletal conditions are considered benign and non-specific. However, a small proportion of patients with musculoskeletal conditions have an undiscovered serious pathology causing their symptoms (3). Previously, serious pathology among patients with musculoskeletal disorders have mainly focused on spine specific pathologies, such as spinal malignant neoplasms, fractures, cauda equine syndrome, spinal infections and axial spondylarthritis (4,5). Early identification of these serious pathologies is of great importance because they necessitate timely and correct diagnosis and treatment, which cannot be provided in primary care settings (3).
Initial screening for these serious pathologies is usually performed by the General Practitioner (GP), and although the screening is primarily described in spine specific guidelines (3)(4)(5), it is important for all musculoskeletal conditions. It is commonly acknowledged, that serious pathology in the group of patients with musculoskeletal conditions is rare. Historically the prevalence of serious pathology has been investigated in populations with spine specific conditions, and a more general prevalence in the group of patients with musculoskeletal conditions is yet to be estimated. European guidelines 4 suggests that 1% of LBP patients in primary care have a serious pathology causing their musculoskeletal symptoms (3,4). Of these serious pathologies, fractures and malignancy are most common, while cauda equine syndrome and spinal infections are less common with an estimated prevalence of 0.04% and 0.01% respectively (6).
Although the initial screening for serious pathology in primary care is usually performed by the GP, other healthcare providers also play a central role in the treatment pathway. In Denmark, GP's acts as gatekeepers into the healthcare system, meaning most patients with musculoskeletal conditions seek the GP, who then examine and refer the patient to appropriate treatment. Often this treatment will include primary care physiotherapy. While screening for signs and symptoms of serious pathologies is also part of the physiotherapy guidelines, no studies have yet investigated how many patients diagnosed with serious pathology have been treated in primary care physiotherapy. Thus, the aim of this study was to estimate the prevalence of neoplasm, cauda equina syndrome, spinal fractures, infections and inflammatory pathology in patients referred for musculoskeletal physiotherapy treatment.

Design and registers
The study was a prospective nationwide register-based cohort study. The present study builds on data from two healthcare registries; 1) The Danish National Health Insurance Service Register (NHSR), which contains daily information on physiotherapy interventions received in private primary care since 1990 with the exception of self-paid therapy without reimbursement (7), and 2) the Danish National Patient Register (NPR) (8), which includes information on hospital diagnoses and contact dates for all in-and outpatient contacts in public and private somatic hospitals in Denmark. In NPR there is a primary diagnose and up to several secondary diagnoses describing each patient's individual 5 course of treatment. Diagnoses are coded using the International Classification of Diseases and Related Health Problems (ICD-10) system (9). The study was reported as recommended in the RECORD checklist (extended STROBE checklist) (10).

Population
In Denmark, national healthcare registries provide individual-based records of contact to the healthcare system for the entire population (11). In the present study all records of referrals for musculoskeletal physiotherapy treatment between 1.

Serious pathology
All records of primary hospital diagnose within 180 days from first contact with the physiotherapist was obtained from NPR, identifying all diagnoses of serious pathology in the study population. Each diagnose represented a case and patients could potentially be diagnosed with more than one serious pathology in the study period. We included five categories of serious pathology (table 1), which not only represent spine specific conditions, but also more general serious pathologies such as benign neoplasms. 6

Statistical analysis
The flow of observations was described and descriptive characteristics of the cohort were presented. Period (180 days) prevalence proportions of serious pathology were calculated and presented as prevalence estimates with 95% confidence interval (95% CI). Sensitivity analysis was performed on the prevalence proportions by including both primary and secondary ICD-10 diagnoses codes from DNPR. Prevalence estimates were calculated only including each patient's first course of treatment, thereby changing the cohort from observations to individual patients. For the categories neoplasm, fracture and inflammatory pathologies, prevalence estimates were calculated and presented stratified into previously diagnosed with a similar pathology or not. For each of these categories, the following characteristics were presented: 1) Gender, 2) age divided into <50 / ≥50 years of age at first contact to physiotherapist and 3) comorbidity based on the revised Charlson comorbidity index (13,14) using ICD-10 diagnoses from the DNPR the last 10 years. The original scale from 0-24 were divided into 0 (no comorbidity) and >0 (comorbidity).
All statistical analyses were performed using STATA version 15  The prevalence of neoplasm was 2.11%, of which 1.13% was malignant neoplasms. The 8 prevalence of cauda equina syndrome was 0.01%, fractures 0.13%, infections 0.01% and inflammatory diseases of the spine 0.06%. Changing the included diagnoses to both primary and secondary diagnoses had little impact on the estimated prevalence, changing the any serious pathology estimate to 2.60% (data not shown). Only including the first course of treatment for each patient did not change the estimated prevalence (n = 1,101,948). Table 3 presents prevalence estimates of neoplasms, fracture and inflammatory pathology. In all of the pathology categories, there was a lower prevalence among patients who had not been diagnosed with the same pathology previously. Among those not previously diagnosed, patients over the age of 50 or patients with co-morbidity had higher prevalence estimates in the neoplasm and fracture pathology categories. In the inflammatory pathology category only minor differences in prevalence estimates were detected. and inflammatory pathology of the spine was 0.01%, 0.13%, 0.01% and 0.06% respectively. When previously diagnosed patients were excluded, the prevalence of malign neoplasm, benign neoplasm, fracture and inflammatory pathology was 0.64%, 0.80%, 0.10% and 0.05% respectively.

Strengths and limitations
A major strength of the study is, that the included cohort represents all patients seen in primary care physiotherapy because of musculoskeletal conditions, thus no bias due to selection was present. However, patients who died or migrated within 180 days from their first physiotherapy contact were excluded. Because the study estimated prevalence of serious (and possibly fatal) pathologies, patients might have died of for example neoplasms which means the estimated prevalence could be underestimated. Also, we have no information of possible diagnose among migrated patients. Nevertheless, taking into account the relatively few patients that were excluded the possible underestimation would probably be small. Also misclassifications of serious pathology could have occurred.
Although the NPR is based on ICD-10 diagnoses, which enables transparent categorizations of serious pathology, patients could initially present with a suspected serious pathology, but eventually be diagnosed with another condition. If the initial diagnose is not changed correctly afterwards, there is a risk of misclassifications which would result in an overestimated prevalence. This potential misclassification is however thought to be small. To avoid such misclassifications in the neoplasm category, we could have used the Danish Cancer Registry (DCR) (15). In the DCR only verified diagnoses of cancer (malignant neoplasms) are recorded. However, the DCR does not contain information on benign neoplasms. We chose to include both benign and malign neoplasms, because the signs and symptoms of benign neoplasm are often the same as malignant neoplasm, thereby making the distinction of symptoms very difficult. This possible overestimation would, however, still be based on a suspicion of serious pathology which means the patient should be referred to further evaluation in secondary care.
Furthermore, as benign neoplasms also can severely affect the patients' general health status the detection of overall symptoms of neoplasms and timely referral still seems important. If the physiotherapist has even a vague suspicion that the patient might have a serious pathology the physiotherapist should send the patient back to the GP for further investigation.
In the NHSR it is not possible to extract reasons for referrals, meaning we cannot categorize the musculoskeletal conditions into specific diagnose groups. It however seems plausible, that the prevalence of spine specific serious pathologies, such as fractures and cauda equina syndrome, are higher among patients with spine specific conditions.
Unfortunately, this hypothesis could not be investigated in the present study.
Interpretation of the results Generalisability of the results

13
The external validity of the study is considered excellent, as the study was based on Danish national healthcare registries, which covers the total Danish population. Because of the study power and completeness of the Danish healthcare registries, the prevalence of different categories of serious pathologies form a very robust and accurate estimation in the group of patients with musculoskeletal conditions treated in primary care physiotherapy.

Conclusion
The prevalence of serious pathology among musculoskeletal physiotherapy patients was 2.3%. This means, that although serious pathology is rare, it is more frequent than the guideline endorsed prevalence estimates suggests.