Article Text
Abstract
Objectives To estimate prevalence, incidence and mortality rates, and annual healthcare costs of primary Sjögren’s syndrome (pSS) and SS associated with other autoimmune disorders (SS+AID) in France.
Methods French national healthcare claims-based study within the prospective Système National des Données de Santé database that includes the majority of the French population. An algorithm was developed to identify patients with SS and SS-related healthcare claims were analysed between 2011 and 2018.
Results Overall, 23 848 patients with pSS and 14 809 with SS+AID were identified. From 2011 to 2018, the prevalence rate increased slightly for pSS (23–32 per 100000) and SS+AID (16–20 per 100 000), with females comprising 90%–91% and 92%–93% of cases, respectively. The incidence rate of SS per 100 000 persons decreased from 2012 (pSS: 4.3; SS+AID: 2.0) to 2017 (pSS: 0.7; SS+AID: 0.3). Mortality rates per 100 000 persons increased from 2012 to 2018 in patients with pSS (0.2–0.8) or SS+AID (0.1–0.5); mean age of death also increased. Artificial tears and hydroxychloroquine were the most common drug reimbursements. Less than half of patients received annual specialist care from a dentist or ophthalmologist. Healthcare costs associated with SS increased from 2011 to 2018 and exceeded the national estimate of expected costs for chronic diseases.
Conclusion In this large French population database study, the low prevalence of pSS confirms that it is an orphan disease. SS is clinically and economically burdensome; these findings may help clinicians better understand routine healthcare received by patients.
- autoimmune diseases
- epidemiology
- Sjogren's syndrome
- prevalence
Data availability statement
Data are available on reasonable request. Bristol Myers Squibb policy on data sharing may be found at https://www.bms.com/researchers-and-partners/independent-research/data-sharing-request-process.html.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
The true prevalence of primary Sjögren’s syndrome (pSS) in the general population has not been precisely defined, with reports in the literature varying from 0.01% to 0.3%.
WHAT THIS STUDY ADDS
In this large claims-based study, including almost the entire French population, estimated nationwide pSS prevalence was 0.023% to 0.032%.
This low prevalence is consistent with that of orphan diseases but might also reflect the underdiagnosis of this disease.
Healthcare costs associated with SS increased between 2011 and 2018 and surpassed estimates of other chronic diseases.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
These findings may help clinicians understand epidemiological healthcare and treatment trends in patients with SS and will help to increase awareness of this rare but burdensome disease.
Introduction
Sjögren’s syndrome (SS) is a chronic, systemic autoimmune disease characterised by oral and ocular dryness related to lymphocytic infiltration of exocrine glands, which can present as primary SS (pSS) or as SS associated with other autoimmune disorders (SS+AID).1–3 Sicca symptoms are the most common manifestations in patients with SS which, although not life-threatening, significantly impact patients’ daily activities and quality of life compared with the general population.4–6 Moreover, 30%–70% of patients with pSS7–9 (depending on ethnicity/age)3 10–12 might experience extraglandular manifestations. The association between sicca symptoms and stereotypical pSS systemic manifestations such as cutaneous, pulmonary or nervous system involvement and/or cryoglobulinaemic vasculitis, facilitates the diagnosis in most cases.2 Conversely, existence of SS+AID could be underevaluated, as sicca symptoms might be undermined in the set of multiple systemic manifestations.
Demographic and epidemiological studies on SS+AID are lacking in contrast to pSS, which is known to predominantly affect females (female:male ratio=9:1) at an average age of 52–56 years.11 13 14 Although the true prevalence of pSS in the general population has not been precisely defined, epidemiological reports range from 0.1 to 3 per 1000 persons (0.01%–0.3%).13–16 In a systematic review and meta-analysis, overall global prevalence rates were estimated to be 61 per 100 000 persons (0.06%), although there was considerable geographic variation.14 An analysis of three epidemiological studies estimated the prevalence in Europe to be 39 per 100 000 persons (0.04%).17 In the French general population, the calculated prevalence was previously reported to range from 1.02 to 1.52 cases per 10 000 persons (0.010%–0.015%); however, despite best efforts, the capturing of cases may have been incomplete as it did not rely on a medico-administrative database.3
Despite an increase in investigational therapies for pSS over the last 10 years, most trials have failed to meet their primary end points,18–21 except for recent phase II trials, which have shown encouraging efficacy of disease-modifying antirheumatic drugs (DMARDs) or biological therapies for systemic manifestations in patients with pSS with the EULAR Sjögren Syndrome Disease Activity Index score as the primary end point.22–26 Current practice, as supported by the EULAR SS Task Force recommendations, is to alleviate dryness with symptomatic treatments and empirically treat systemic manifestations with DMARDs, immunosuppressants, glucocorticoids and sometimes, biological therapies.27
The primary objective of this study was to estimate the prevalence of pSS and SS+AID using data from the French national health insurance claims database. Secondary objectives were to estimate the incidence, mortality rate and annual healthcare costs associated with pSS and SS+AID.
Methods
Data source
This was a claims-based study using data recorded prospectively in the Système National des Données de Santé (SNDS). The SNDS comprised data from almost all of the French population (~67 million people). Data available include sociodemographic data, outpatient health resource use (consultations, procedures, drug dispensation) covered by insurance and sick days. Inpatient data included discharge summaries, including admission reason and comorbidities described using the International Classification of Diseases, Tenth Revision (ICD-10) codes, procedures and highly expensive drugs dispensed during inpatient stays (such as biologics).28 All SNDS data are pseudonymised. For this study, data from 1 January 2011 to 31 December 2018 were extracted.
Development of the patient identification algorithm
This study used a two-step approach to identify patients with pSS and SS+AID. First, a 1/97 random sample of the SNDS, called the Echantillon généraliste de bénéficiaires (EGB), was used to obtain a representative sample of healthcare beneficiaries.28 In collaboration with data scientists and SS specialists, an algorithm was developed within the EGB database to identify patients with pSS and SS+AID, which was subsequently adapted for this study.29 A 10-fold cross-validation was performed using a logistic regression analysis to estimate the accuracy of 15 different algorithm models. A receiver operating characteristic curve determined the most efficient algorithm. A full description of the algorithm development is available in the online supplemental methods.
Supplemental material
The adapted algorithm identified patients with pSS with an M35.0 ICD-10 code and at least two prescriptions of a treatment of interest (full list in online supplemental table S1) before or after the first M35.0 code. Patients with SS+AID fulfilled these same criteria, and had an ICD-10 code for another associated connective tissue disease: rheumatoid arthritis (RA), juvenile arthritis, systemic lupus erythematosus (SLE), systemic sclerosis or other overlap syndromes (see online supplemental methods).
pSS and SS+AID identification in the SNDS
Individuals were identified in the SNDS database with the adapted algorithm. Patients with a history of amyloidosis, sarcoidosis, head and neck cancer with radiotherapy, other specified congenital myopathies, bone marrow transplant rejection, diseases of the blood and blood-forming organs and certain immune disorders including HIV/AIDS and autoimmune myopathies, as defined by the ICD-10, were excluded. Patients were included only if they had been in the SNDS for at least 1 year prior to diagnosis. This minimal 1-year look-back period was necessary to estimate incidence by excluding patients with prevalent SS.
Study outcomes and analyses
Disease onset was defined as the earliest date among the following events: first occurrence of the M35.0 ICD-10 code for SS, first reimbursement of a drug of interest or an antinuclear antibody assay, or first procedure on salivary glands or Schirmer’s test. Prevalence rates were estimated for the years 2011–2018. Denominators used for estimating prevalence rates were derived from French population census data.
Incidence rates were estimated for the years 2012–2017 by including patients with at least two prescriptions of a treatment of interest (full list in online supplemental table S1) between 2012 and 2018. To ensure health insurance coverage and prevent counting patients who switched healthcare providers as separate incident cases, we included in the incidence rate estimate only patients with any type of reimbursement in the database during the 10–16 months before the earliest date of disease onset. Differences in age at time of death for the pSS and SS+AID cohorts were determined by Student’s t-test.
Drugs and healthcare consumption estimates
Healthcare cost estimates were based on French healthcare insurance rates.30 Frequencies of use for the drugs of interest were calculated for 2017; this year was selected because it had the highest total number of patients with SS. The cost analysis was based on data from healthcare insurance plans, and therefore cannot be directly attributed to the disease. We included direct and indirect costs in this analysis; these included medical resource consumption such as drug dispensation, radiology and biology tests, medical consults and hospitalisations as well as the amount of sick leave. Day care hospital stays were defined as hospital stays for diagnosis/therapeutic procedures not requiring overnight stay (<1 day).
Results
Identification of patients with pSS or SS+AID
Using the prespecified algorithms, 23 848 patients with pSS and 14 809 patients with SS+AID were identified during the study period (2011–2018). At disease onset, the mean (SD; IQR) age was 60.2 years (15.4; 50–72) for pSS and 56.6 years (15.1; 47–67) for SS+AID (table 1). Female patients constituted 90%–91% of pSS cases and 92%–93% of SS+AID cases. The majority of ICD-10 M35.0 diagnostic codes were identified in patients during day care hospital stays (75%, n=17 956 (pSS); 81%, n=12 031 (SS+AID); figure 1). In patients with SS+AID, RA was the most frequently associated AID (9070 patients (53%)) followed by SLE (4731 patients (28%); table 2).
Prevalence, incidence and mortality rate of patients with pSS or SS+AID
The estimated prevalence of pSS increased early in the study period and stabilised between 2014 and 2018. In 2011, the prevalence of pSS was 23 cases per 100 000 persons and in 2018, prevalence increased to 32 cases per 100 000 persons. The prevalence of SS+AID was numerically lower than pSS (16 cases per 100 000 persons in 2011; 20 cases per 100 000 persons in 2018; figure 2A). The median time between two hospital inpatient visits, or two hospital outpatient visits related to SS, was 52 months in patients with pSS and 56 months in patients with SS+AID.
The estimated annual incidence rate of pSS decreased from 4.3 cases per 100 000 persons in 2012 to 0.7 cases per 100 000 persons in 2017. A similar pattern was demonstrated for the estimated annual incidence rate of SS+AID, which decreased from 2.0 cases per 100 000 persons in 2012 to 0.3 cases per 100 000 persons in 2017 (figure 2B). Between 2012 and 2017, female patients comprised similar proportions of incident cases of pSS and SS+AID, ranging from 85%–87% to 85%–90%, respectively. The mean (SD) age at which patients were counted as a pSS incident case was 62 (15) years and 57 (16) years for females and males, respectively; for SS+AID, the mean age was 58 (15) and 53 (16) years for females and males, respectively.
From 2012 to 2018, the mortality rate of patients with pSS increased from 0.2 deaths to 0.8 deaths per 100 000 persons, and for patients with SS+AID, 0.1 deaths to 0.5 deaths per 100 000 persons (online supplemental table S2). Among patients who died, the proportion who were female ranged from 77% in 2013 to 83% in 2017 for patients with pSS, and from 84% in 2013 to 90% in 2012 for patients with SS+AID, which was lower than the aforementioned prevalence sex ratio. However, from 2012 to 2018, the mean age at the time of death varied from 74 to 82 years in patients with pSS and 72 to 78 years in patients with SS+AID. In patients with pSS, mean age at time of death was higher each year for female versus male patients, but in patients with SS+AID, it was generally similar for female and male patients.
Drug consumption
From 2014 to 2018, for patients with pSS, approximately 30%–31% of male patients and 22%–23% of female patients did not receive any reimbursement for drugs related to SS, compared with approximately 16%–19% of male patients and 14%–15% of female patients with SS+AID during the same time frame. The mean number of prescriptions dispensed from 2011 to 2018 per patient was 40 for pSS and 55 for SS+AID (online supplemental table S3). Detailed analyses for 2017 (the timepoint with the highest total number of patients with SS) revealed that among immunosuppressants, methotrexate was the most commonly reimbursed drug, but was reimbursed for a higher proportion of patients with SS+AID (25%) than patients with pSS (4%) (table 3). Male patients with pSS were reimbursed for azathioprine, cyclophosphamide, mycophenolic acid and rituximab more frequently than female patients with pSS, while reimbursement for methotrexate and pilocarpine were similar between both sexes, although reimbursements for these drugs were infrequent (<4%). Among other types of medication reimbursed in 2017, artificial tears were the most frequently reimbursed drug (58% (pSS), 53% (SS+AID)), followed by hydroxychloroquine (17% (pSS), 33% (SS+AID)) and pilocarpine magistral preparation (13% (pSS), 12% (SS+AID); table 3).
Healthcare utilisation and costs
The proportion of patients with pSS or SS+AID who had consultations with either a dentist or an ophthalmologist ranged from approximately 23% in 2011 to 50% in 2018. Patients in each cohort (pSS or SS+AID) had a mean of two dentist and one ophthalmologist reimbursements per year (online supplemental table S3). Mean annual healthcare costs increased over the study period and were higher for patients with SS+AID than for patients with pSS (figure 3). Within the period of 2011–2018, the median (mean) annual costs per patient ranged from €2290 (€5780) to €4261 (€9710) in patients with pSS and €4035 (€8819) to €7472 (€13,273) for those with SS+AID.
Discussion
Using data from the nationwide French SNDS database, this study aimed to estimate the prevalence of pSS and SS+AID, as well as the associated incidence, mortality rate and annual healthcare costs. A total of 23 848 patients with pSS and 14 809 patients with SS+AID were identified, the majority of whom were female. Over the study period, decreases were observed in the incidence rates for pSS and SS+AID while mortality rates increased. Healthcare costs were greater in patients with SS+AID compared with pSS; overall mean costs were greater than French healthcare insurance projections for chronic diseases.30
After an initial increase, the prevalence of pSS was stable over time and consistent with previous reports.3 Our finding that the prevalence of pSS ranged from 23 to 32 per 100 000 persons confirms that pSS may be considered an orphan disease (defined by the European Commission as affecting fewer than 1 in 2000 persons).14 Orphan disease status can be an incentive for new therapeutic advances as drugs specifically patented for these indications are endowed with an advantageous status by regulatory health authorities. This might be of particular interest in the case of pSS, for which scant therapeutic options are available. Research and care in the field benefit from specific funding and the help of healthcare networks for rare diseases at both the national level31 and European level.32 These networks promote the development and dissemination of useful tools for patient care and education, such as educational programmes, and national diagnosis and treatment guidelines.33
The estimated annual incidence rate and age at diagnosis were numerically higher in patients with pSS compared with patients with SS+AID. In males, the average age at disease onset was numerically higher compared with females. We observed some variability in incidence rates (per 100 000 persons), which were highest in the first year (2012; pSS: 4.3 and SS+AID: 2.0) and lowest in the last year (2017; pSS: 0.7 and SS+AID: 0.3) of the study period. This decrease in incidence may be related to an improvement in diagnostic procedures and classification criteria, along with better dissemination of their use in clinical practice, leading to fewer false diagnoses.34 The high initial incidence rate could be explained by a relatively short look-back period, given the long time interval between the two hospital visits (52–56 months), which was the main source of patient identification by ICD-10 codes. Thus, it is probable that some patients who were already diagnosed with pSS, but had not yet had a hospital visit during the look-back period, were recorded as incident cases in the first years of this study. Furthermore, algorithms used to detect pSS relied on specific drug prescriptions which could appear after the first ICD-10 code. As this study did not include a look-after period, potential patients at the end of the study period may have been missed, explaining the decreased incidence in the last years of the study.
Unlike the incidence rate, the mortality rate of patients with pSS or SS+AID trended upwards; the mortality rate in patients with pSS was 0.2 deaths per 100 000 persons in 2012 and 0.8 deaths per 100 000 persons in 2018, and in patients with SS+AID, it was 0.1 deaths per 100 000 persons in 2012 and 0.5 deaths per 100 000 persons in 2018. Despite the incidence rate decreasing each year and the mortality rate increasing, the prevalence of pSS and SS+AID still increased by a small amount each year, as the number of new cases remained greater than the number of deaths. This may be partly due to patients with SS living longer, as demonstrated by an increasing age of death during our observation period. The observation of a higher mortality rate in patients with pSS versus patients with SS+AID was unexpected and may be related to their older age (mean 60.2 vs 56.6 years, respectively); this should be explored further in future studies.
Mortality rates by sex were marginally higher for males with pSS and SS+AID, given the overall female:male ratio of approximately 9:1. pSS has been previously reported as being more severe in males than females.7 A recently published systematic review and meta-analysis reported mortality risk factors in patients with pSS which included older age, male sex, hypocomplementaemia, vasculitis, interstitial lung disease, positive anti-La/SS type B and cryoglobulinaemia.28 However, in this study, the higher mortality rate in male patients with SS might be related to other causes, as a previous systematic review and meta-analysis found that overall mortality rates between patients with SS and the general population appear similar.35 Other factors such as cardiovascular disease or cancer might explain this difference in mortality, as they are common causes of greater and earlier mortality in males, especially given the older age of male patients with SS compared with female patients in our study.
Historically, SS+AID has been termed ‘secondary’ SS, compared with pSS, and subsequently has been much less researched and excluded from clinical trials.36 Thus, the overall prevalence of SS+AID has been poorly studied, since almost all prevalence studies used pSS classification criteria. However, the frequency of SS associated with other AID has been previously reported. SS is estimated to be prevalent in 7%–31% of RA cases,37–42 and 6.5%–19% of SLE cases.1 43 Therefore, it is unsurprising, given its higher prevalence in the general population, that RA is the most prevalent AID in patients with SS+AID, followed by SLE. However, it is interesting that, in around a quarter of cases, the diagnosis of SS is made after the diagnosis of the associated AID, often within the following 5 years. This suggests either a lack of screening for SS at the time of diagnosis of the associated AID or that SS became clinically apparent only after the AID was diagnosed. This potential underestimation may explain why the prevalence of SS+AID was lower than pSS; if the other AID is the predominant disorder, the ICD-10 code for SS may not have been reported, and/or the patient may not have been screened for SS. Taking the example of RA, which has prevalence estimates within the SNDS database of around 350 cases per 100 000 persons,44 a higher prevalence of SS+RA would have been expected. Inversely, in a significant portion of patients with SS+AID, the diagnosis of SS is made >10 years before the AID. The late development of SLE after SS diagnosis has been described by Manoussakis et al, which gives distinct clinical features in these patients from those in patients with pSS, specifically those at a younger age at SS diagnosis.45 Overall, these results underscore the importance of appropriate SS screening when diagnosing other AIDs, as well as diligent follow-up of patients with pSS, including monitoring for the onset of systemic symptoms, especially when the diagnosis is made at a young age.7
The most common drug reimbursement for patients with SS was artificial tears, highlighting both the high burden of sicca symptoms and the lack of validated systemic treatment. It is worth noting that 24% of patients with pSS and 14% with SS+AID received no drug reimbursements. These patients could either have moderate dryness and not require symptomatic treatment, or (self-)medicate with over-the-counter or non-reimbursed artificial tear drops and therefore might attend follow-up consultations less frequently. Each year, <50% of patients with pSS or SS+AID consulted with a dentist; the same proportion consulted with an ophthalmologist, underscoring the multidisciplinary care needs of patients with SS. We observed very few reimbursements for SS diagnostic procedures, which is probably explained by most patients with SS being identified during a day care hospital stay where individual procedures performed during this visit are not usually coded, as they are not associated with any additional funding. Therefore, these procedures are likely to be widely used in practice, but not sensible for SS identification in this claims database.
Overall healthcare costs for patients with SS were higher than estimates for other inflammatory diseases.30 These costs were higher than the French healthcare insurance projection of a mean cost of €5158, based on French universal healthcare reports from 2016; however, this projection incorporates a wide variety of diagnoses, such as inflammatory or rare diseases or HIV/AIDS.30 The higher-than-expected costs provide further evidence to support pSS as a clinically and economically burdensome disease, as recently reported in a Swedish retrospective cohort study.46 The authors of this study reported that healthcare costs such as symptom relief, dental care and outpatient care were significantly higher in patients with pSS compared with the general population. Also, for each year of our study period, the healthcare costs in patients with SS+AID exceeded that of patients with pSS. This was partly linked to a higher consumption of costly immunosuppressive and biological therapies by patients with SS+AID. While there was an overall increase in healthcare costs over the study period, a small proportion may reflect the 7% inflation in France between 2011 and 2018.47
The main study limitations included non-medically validated diagnoses based on healthcare insurance claims and potential patient identification biases including favouring patients referred to hospital, which may have missed patients with less severe disease, therefore artificially inflating cost estimates and underestimating disease prevalence. However, the long study period enabled a stabilisation of prevalence. Nonetheless, the very low prevalence observed here encourages increasing awareness of this condition to avoid its underdiagnosis. As cases were identified based on disease-specific drug reimbursements, we cannot exclude a potential bias favouring identification of patients who required or accepted treatments, which may have increased healthcare cost estimates. We also acknowledge the use of a relatively short look-back period for defining incident cases is a potential limitation and the use of a longer duration, for example, the median interval between two hospital visits, might have led to more stable estimates. Additionally, it should be noted that healthcare cost estimates cannot be directly attributed to SS, as the cost analysis used data from healthcare insurance plans which likely included other medical treatment; therefore, our estimates may represent the cost of SS and the cost of associated diseases, especially for SS+AID. Study strengths include use of the large SNDS database, which records claims data of almost all of the French population and is thus highly representative of the French population. Also, use of a validated algorithm for identifying patients with pSS and SS+AID, to our knowledge, has never been done at such a large scale.
In this claims-based database study of the French population, prevalence estimates confirm that pSS is an orphan disease. Such observations provide insights regarding the clinical and economic burden of rare but often severe AID and show that SS associated with other AID is probably underdiagnosed, or at least under-recorded, emphasising the need to improve awareness of this association among physicians. These results may also help inform and optimise care for patients with SS in France. Additionally, these results reveal trends in pSS and SS+AID, which may facilitate public health planning in France.
Data availability statement
Data are available on reasonable request. Bristol Myers Squibb policy on data sharing may be found at https://www.bms.com/researchers-and-partners/independent-research/data-sharing-request-process.html.
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
Acknowledgments
Professional medical writing and editorial assistance was provided by Brooke Bouza, PhD, at Caudex, a division of IPG Health Medical Communications, and was funded by Bristol Myers Squibb. Portions of the data from this study were presented at the EULAR 2021 Virtual Congress.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors RS, LC, MB, GD, JZ, VV-M, VD-P has met the authorship criteria established by the International Committee of Medical Journal Editors. All authors contributed to the study design and data interpretation. Statistical analysis was performed by GD. All authors contributed to writing and reviewing the manuscript, and all authors approved the final manuscript. RS accepts responsibility for the overall content as guarantor.
Funding This work was supported by Bristol Myers Squibb.
Competing interests RS is a consultant for Amgen, Boehringer Ingelheim, Bristol Myers Squibb, Fresenius Kabi, GlaxoSmithKline, Janssen, Pfizer and Roche. LC is a consultant for Bristol Myers Squibb. GD is a consultant for Bristol Myers Squibb and a consultant with no fees for Roche. JZ and VV-M are employees of and shareholders in Bristol Myers Squibb. VD-P is a consultant for AbbVie, Bristol Myers Squibb and Novartis, and has received grant/research support from Bristol Myers Squibb, Eli Lilly and Roche-Chugai. MB declares no conflicts of interest.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.