How accurate are diagnoses for rheumatoid arthritis and juvenile idiopathic arthritis in the general practice research database?

Risks of smoking and benefits of smoking cessation on hospitalisations for cardiovascular events and respiratory infection in patients with rheumatoid arthritis: a retrospective cohort study using the Clinical Practice Research Datalink

Objectives

To investigate the associations between smoking status, smoking cessation and hospitalisations for cardiovascular events (CVE) and respiratory tract infections (RTI) in an inception cohort of patients with rheumatoid arthritis (RA).

Methods

The study was set within UK primary care electronic health records (the Clinical Practice Research Datalink) linked to hospital inpatient data (Hospital Episode Statistics). Patients with RA were followed from diagnosis to hospitalisation with a record of CVE or RTI, leaving their general practice, death, or 10 January 2012, whichever was earliest. Smoking status (never, current, former) was defined using primary care data and could vary over time. Survival analysis was performed using Cox regression (first event) and conditional risk set models (multiple RTIs).

Results

5677 patients were included in the cohort: 68% female, median age 61 years. The age-adjusted and sex-adjusted risks of hospitalisation for CVE or RTI were more than twice as high in current vs never smokers (CVE HR (95% CI) 2.19 (1.44 to 3.31); RTI 2.18 (1.71 to 2.78)). The risks for both outcomes were significantly higher in current compared with former smokers (CVE 1.51 (1.04 to 2.19), RTI 1.29 (1.04 to 1.61)). For each additional year of smoking cessation, the risk of first CVE and RTI hospitalisation fell significantly, approximately 25% and 15% respectively in the adjusted models.

Conclusions

Patients with RA who smoke have an increased risk of hospitalisation with CVE or RTI compared with never and former smokers. The risk decreases for each additional year of smoking cessation. Patients with RA who smoke should be advised to stop smoking.

Association between NICE guidance on biologic therapies with rates of hip and knee replacement among rheumatoid arthritis patients in England and Wales: An interrupted time-series analysis

OBJECTIVE

To estimate the impact of NICE approval of tumor necrosis factor inhibitor (TNFi) therapies on the incidence of total hip replacement (THR) and total knee replacement (TKR) among rheumatoid arthritis (RA) patients in England and Wales.

METHODS

Primary care data [Clinical Practice Research Datalink (CPRD)] for the study period (1995-2014) were used to identify incident adult RA patients. The age and sex-standardised 5-year incidence of THR and TKR was calculated separately for RA patients diagnosed in each six-months between 1995-2009. We took a natural experimental approach, using segmented linear regression to estimate changes in level and trend following the publication of NICE TA 36 in March 2002, incorporating a 1-year lag. Regression coefficients were used to calculate average change in rates, adjusted for prior level and trend.

RESULTS

We identified 17,505 incident RA patients of whom 465 and 650 underwent THR and TKR surgery, respectively. The modeled average incidence of THR and TKR over the biologic-era was 6.57/1000 person years (PYs) and 8.51/1000 PYs, respectively, with projected (had pre-NICE TA 36 level and trend continued uninterrupted) figures of 5.63/1000 PYs and 12.92 PYs, respectively. NICE guidance was associated with a significant average decrease in TKR incidence of -4.41/1000 PYs (95% C.I. -6.88 to -1.94), equating to a relative 34% reduction. Overall, no effect was seen on THR rates.

CONCLUSIONS

Among incident RA patients in England and Wales, NICE guidance on TNFi therapies for RA management was temporally associated with reduced rates of TKR but not THR.

Supplementing electronic health records through sample collection and patient diaries: A study set within a primary care research database

PURPOSE

To describe a novel observational study that supplemented primary care electronic health record (EHR) data with sample collection and patient diaries.

METHODS

The study was set in primary care in England. A list of 3974 potentially eligible patients was compiled using data from the Clinical Practice Research Datalink. Interested general practices opted into the study then confirmed patient suitability and sent out postal invitations. Participants completed a drug-use diary and provided saliva samples to the research team to combine with EHR data.

RESULTS

Of 252 practices contacted to participate, 66 (26%) mailed invitations to patients. Of the 3974 potentially eligible patients, 859 (22%) were at participating practices, and 526 (13%) were sent invitations. Of those invited, 117 (22%) consented to participate of whom 86 (74%) completed the study.

CONCLUSIONS

We have confirmed the feasibility of supplementing EHR with data collected directly from patients. Although the present study successfully collected essential data from patients, it also underlined the requirement for improved engagement with both patients and general practitioners to support similar studies.

Comparison of cardiovascular risk algorithms in patients with vs without rheumatoid arthritis and the role of C-reactive protein in predicting cardiovascular outcomes in rheumatoid arthritis

Objectives

The aims were to compare the performance of cardiovascular risk calculators, Framingham Risk Score (FRS) and QRISK2, in RA and matched non-RA patients and to evaluate whether their performance could be enhanced by the addition of CRP.

Methods

We conducted a retrospective analysis, using a clinical practice data set linked to Hospital Episode Statistics (HES) data from the UK. Patients presenting with at least one RA diagnosis code and no prior cardiovascular events were matched to non-RA patients using disease risk scores. The overall performance of the FRS and QRISK2 was compared between cohorts, and assessed with and without CRP in the RA cohort using C-Index, Akaike Information Criterion (AIC) and the net reclassification index (NRI).

Results

Four thousand seven hundred and eighty RA patients met the inclusion criteria and were followed for a mean of 3.8 years. The C-Index for the FRS in the non-RA and RA cohort was 0.783 and 0.754 (P < 0.001) and that of the QRISK2 was 0.770 and 0.744 (P < 0.001), respectively. Log[CRP] was positively associated with cardiovascular events, but improvements in the FRS and QRISK2 C-Indices as a result of inclusion of CRP were small, from 0.764 to 0.767 (P = 0.026) for FRS and from 0.764 to 0.765 (P = 0.250) for QRISK2. The NRI was 3.2% (95% CI: -2.8, 5.7%) for FRS and -2.0% (95% CI: -5.8, 4.5%) for QRISK2.

Conclusion

The C-Index for the FRS and QRISK2 was significantly better in the non-RA compared with RA patients. The addition of CRP in both equations was not associated with a significant improvement in reclassification based on NRI.

Oral glucocorticoid use and osteonecrosis in children and adults with chronic inflammatory diseases: a population-based cohort study

OBJECTIVES

We studied oral glucocorticoids and osteonecrosis, a rare but serious bone disease, in individuals with various chronic inflammatory diseases. We hypothesised that we would find stronger associations in adults versus children and in people with autoimmune diseases.

DESIGN

Retrospective cohort study.

SETTING

Population-representative data (1994-2013) from general practices in the UK.

PARTICIPANTS

Children and adults diagnosed with asthma; inflammatory bowel disease; juvenile, psoriatic or rheumatoid arthritis; psoriasis; or systemic lupus.

EXPOSURES

Oral glucocorticoid patterns.

PRIMARY AND SECONDARY OUTCOME MEASURES

Diagnosed osteonecrosis (primary) and osteonecrosis plus clinical features (eg, symptoms, pain medication, surgical repair) (secondary). Discrete time failure models estimated the adjusted hazard ratio (aHR) of incident osteonecrosis following oral glucocorticoid exposure. Hypothesis testing was one sided (with corresponding 90% CI) since glucocorticoids were unlikely protective.

RESULTS

After adjusting for demographic, disease-related and health utilisation factors, glucocorticoid exposure was associated with osteonecrosis in adults (ages 18-49, aHR 2.1 (90% CI 1.5 to 2.9); ages ≥50, aHR 1.3 (90% CI 1.01 to 1.7)). However, low-dose glucocorticoids, corresponding to average doses <7.5 mg prednisolone daily and maximum doses <30 mg daily, were not associated with osteonecrosis in adults. Furthermore, even at high glucocorticoid doses, there was no evidence of increased osteonecrosis among glucocorticoid-exposed children (p=0.04 for interaction by age) (any glucocorticoid exposure, ages 2-9: aHR 1.1 (90% CI 0.7 to 1.7); ages 10-17: aHR 0.6 (90% CI 0.3 to 1.6)). Arthritis, inflammatory bowel disease and lupus were independently associated with osteonecrosis, but there was a similar dose relationship between glucocorticoids and osteonecrosis among adults with low-risk and high-risk diseases.

CONCLUSIONS

Glucocorticoid use was clearly associated with osteonecrosis in a dose-related fashion in adults, especially young adults, but this risk was not detectable in children. The absolute risk of glucocorticoid-associated osteonecrosis in the general paediatric population and in adults taking low glucocorticoid doses is at most extremely small.

Risk of Incident Diabetes Mellitus Associated With the Dosage and Duration of Oral Glucocorticoid Therapy in Patients With Rheumatoid Arthritis

Objective

To quantify the risk of incident diabetes mellitus (DM) associated with the dosage, duration, and timing of glucocorticoid (GC) use in patients with rheumatoid arthritis (RA).

Methods

We undertook a cohort study using 2 databases: a UK primary care database (the Clinical Practice Research Datalink [CPRD]) including 21,962 RA patients (1992–2009) and the US National Data Bank for Rheumatic Diseases (NDB) including 12,657 RA patients (1998–2013). Information on the dosage and timing of GC use was extracted. DM in the CPRD was defined using Read codes, at least 2 prescriptions for oral antidiabetic medication, or abnormal blood test results. DM in the NDB was defined through patient self‐reports. Data were analyzed using time‐dependent Cox models and a novel weighted cumulative dose (WCD) model that accounts for dosage, duration, and timing of treatment.

Results

The hazard ratio (HR) was 1.30 (95% confidence interval [95% CI] 1.17–1.45) and 1.61 (95% CI 1.37–1.89) in current GC users compared to nonusers in the CPRD and the NDB, respectively. A range of conventional statistical models consistently confirmed increases in risk with the GC dosage and duration. The WCD model showed that recent GC use contributed the most to the current risk of DM, while doses taken >6 months previously did not influence current risk. In the CPRD, 5 mg of prednisolone equivalent dose for the last 1, 3, and 6 months was significantly associated with HRs of 1.20, 1.43, and 1.48, respectively, compared to nonusers.

Conclusion

GC use is a clinically important and quantifiable risk factor for DM. Risk is influenced by the dosage and treatment duration, although only for GC use within the last 6 months.

The effect of anti-rheumatic medications for coronary artery diseases risk in patients with rheumatoid arthritis might be changed over time: A nationwide population-based cohort study

Objectives

To determine whether anti-rheumatic drug usage is associated with risk of coronary artery diseases (CAD) in incident Rheumatoid Arthritis (RA) patients.

Methods

Data were obtained from the Taiwan National Health Insurance Research Database. The study cohort comprised 6260 patients who were newly diagnosed with RA between 2001–2010. The study endpoint was occurrence of CAD according to the ICD-9-CM codes. We used the WHO Defined Daily Dose (DDD) as a tool to assess the drugs exposure. The Cox proportional hazards regression model was used to estimate the hazard ratio (HR) of disease after controlling for demographic and other co-morbidities. When the proportionality assumption is violated, a spline curve of the Scaled Schoenfeld residuals is fitted to demonstrate the estimated effect on CAD over time for drug usage.

Results

Among RA patients, use of celecoxib, and etoricoxib was associated with significantly decreased incidence of CAD. The adjusted HR(95% CI) of CAD for low-dose celecoxib (DDD≦1) and high-dose user were 0.47(0.34, 0.65) and 0.37(0.24, 0.58) during the 4 year follow-up time; however, it became 0.98(0.70, 1.37) and1.29(0.85, 1.95). Adjusted HR(95% CI) of CAD for etoricoxib users remained 0.47(0.26, 0.84).

Conclusions

This study revealed association of decreased CAD risk in RA patients taking 2 different kinds of COX-2i in comparison with nonusers. The effect might be changed over time, after about 4 years.

Clinical code set engineering for reusing EHR data for research: A review

INTRODUCTION

The construction of reliable, reusable clinical code sets is essential when re-using Electronic Health Record (EHR) data for research. Yet code set definitions are rarely transparent and their sharing is almost non-existent. There is a lack of methodological standards for the management (construction, sharing, revision and reuse) of clinical code sets which needs to be addressed to ensure the reliability and credibility of studies which use code sets.

OBJECTIVE

To review methodological literature on the management of sets of clinical codes used in research on clinical databases and to provide a list of best practice recommendations for future studies and software tools.

METHODS

We performed an exhaustive search for methodological papers about clinical code set engineering for re-using EHR data in research. This was supplemented with papers identified by snowball sampling. In addition, a list of e-phenotyping systems was constructed by merging references from several systematic reviews on this topic, and the processes adopted by those systems for code set management was reviewed.

RESULTS

Thirty methodological papers were reviewed. Common approaches included: creating an initial list of synonyms for the condition of interest (n=20); making use of the hierarchical nature of coding terminologies during searching (n=23); reviewing sets with clinician input (n=20); and reusing and updating an existing code set (n=20). Several open source software tools (n=3) were discovered.

DISCUSSION

There is a need for software tools that enable users to easily and quickly create, revise, extend, review and share code sets and we provide a list of recommendations for their design and implementation.

CONCLUSION

Research re-using EHR data could be improved through the further development, more widespread use and routine reporting of the methods by which clinical codes were selected.

Primary Prevention of Myocardial Infarction in Rheumatoid Arthritis Using Aspirin: A Case-crossover Study and a Propensity Score–matched Cohort Study

Objective.

Subjects with rheumatoid arthritis (RA) are at higher risk of developing cardiovascular disease, which is their leading cause of death. Conflicting evidence exists regarding the efficacy of aspirin (ASA) as primary prevention. We evaluated whether a protective association exists between ASA and myocardial infarction (MI) in RA subjects.

Methods.

In the United Kingdom, persons age ≥ 60 years receive free ASA by prescription and 75% of use is by prescription. Subjects ≥ 60 years with RA in the population-based The Health Improvement Network database constituted our study population. We excluded patients with history of MI, angina, stroke, peripheral vascular disease, or coronary artery procedures. Our main outcome was the occurrence of fatal and nonfatal MI. We performed a case-crossover study with each subject contributing a hazard period and a control period 90 days prior to the MI. In addition, to minimize confounding by indication, a propensity score (PS)–matched cohort study was performed, considering all patients with RA with an incident prescription of low-dose ASA as our exposed group.

Results.

We did not find a protective effect in the case-crossover study (OR 1.83, 95% CI 0.71–4.71), with 55 subjects exposed in the hazard period and 44 in the control period. Similarly, among 1836 subjects included in the PS-matched cohort study (918 ASA users and 918 ASA non-users), we did not find a protective effect of low ASA on MI (HR 1.39, 95% CI 0.87–2.23).

Conclusion.

We did not find a protective effect of ASA on MI in patients with RA when used as primary prophylaxis.

Opportunities and Challenges in Developing a Cohort of Patients with Type 2 Diabetes Mellitus Using Electronic Primary Care Data

Purpose

To develop a cohort of patients with T2DM treated with insulin using CPRD to obtain an accurate diagnosis date. This was used to analyse time from T2DM diagnosis to first ever insulin prescription between 01/01/2000 and 30/06/2012, for patients in England and Wales.

Methods

Patients aged 18 years and over at diagnosis, were included if prescribed an anti-diabetic drug and were excluded if first diagnosis-specific code was inconsistent with a T2DM diagnosis. Diagnosis codes were split into 8 categories based on whether they related to specific T2DM or non-specific diabetes codes. Patients were excluded if they had non-specific diagnosis codes and were prescribed insulin as their first-ever treatment for diabetes. Descriptive statistics for time from T2DM diagnosis to insulin initiation were calculated.

Results

Two hundred and fifty-six codes were identified which were consistent with a first-ever diagnosis of T2DM. 7 codes were considered to clearly define a diagnosis of T2DM, which were reported for 64% of patients. The final cohort comprised 11,917 patients and the median time to first insulin prescription from the date of diagnosis was 4.4 years.

Conclusions

A clear definition of cohort development is required to compare and interpret results from studies. Use of diagnosis and product codes is essential when examining use of drugs such as insulin, where competing diagnoses need to be considered separately.

The Use of a Bayesian Hierarchy to Develop and Validate a Co-Morbidity Score to Predict Mortality for Linked Primary and Secondary Care Data from the NHS in England

Background

We have assessed whether the linkage between routine primary and secondary care records provided an opportunity to develop an improved population based co-morbidity score with the combined information on co-morbidities from both health care settings.

Methods

We extracted all people older than 20 years at the start of 2005 within the linkage between the Hospital Episodes Statistics, Clinical Practice Research Datalink, and Office for National Statistics death register in England. A random 50% sample was used to identify relevant diagnostic codes using a Bayesian hierarchy to share information between similar Read and ICD 10 code groupings. Internal validation of the score was performed in the remaining 50% and discrimination was assessed using Harrell’s C statistic. Comparisons were made over time, age, and consultation rate with the Charlson and Elixhauser indexes.

Results

657,264 people were followed up from the 1st January 2005. 98 groupings of codes were derived from the Bayesian hierarchy, and 37 had an adjusted weighting of greater than zero in the Cox proportional hazards model. 11 of these groupings had a different weighting dependent on whether they were coded from hospital or primary care. The C statistic reduced from 0.88 (95% confidence interval 0.88–0.88) in the first year of follow up, to 0.85 (0.85–0.85) including all 5 years. When we stratified the linked score by consultation rate the association with mortality remained consistent, but there was a significant interaction with age, with improved discrimination and fit in those under 50 years old (C = 0.85, 0.83–0.87) compared to the Charlson (C = 0.79, 0.77–0.82) or Elixhauser index (C = 0.81, 0.79–0.83).

Conclusions

The use of linked population based primary and secondary care data developed a co-morbidity score that had improved discrimination, particularly in younger age groups, and had a greater effect when adjusting for co-morbidity than existing scores.

Smoking-Related Mortality in Patients With Early Rheumatoid Arthritis: A Retrospective Cohort Study Using the Clinical Practice Research Datalink

Objective

To investigate the association between smoking status and smoking cessation with mortality in patients with rheumatoid arthritis (RA).

Methods

An incident cohort of patients with RA was identified using the Clinical Practice Research Datalink, a database of UK primary care electronic medical records. Time‐varying smoking status, years of cessation, and amount smoked were determined from patients' medical records. The date and underlying cause of death were identified by linkage with Office for National Statistics records. The associations between smoking status and smoking cessation with all‐cause and cause‐specific mortality (circulatory disease, all cancers, lung cancer, respiratory disease, and respiratory infection) were investigated using adjusted Cox (all‐cause mortality) and Fine‐Gray (cause‐specific mortality) regression.

Results

The cohort comprised 5,677 patients (median age 61.4 years, 68% women), with 40% as never smokers, 34% former smokers, and 26% current smokers at baseline. Compared to never smoking, current smoking was associated with an increased risk of all‐cause mortality (hazard ratio 1.98 [95% confidence interval (95% CI) 1.56, 2.53]), and mortality due to circulatory disease (subdistribution hazard ratio [SHR] 1.96 [95% CI 1.33, 2.90]) and lung cancer (SHR 23.2 [95% CI 5.15, 105]). Each year of smoking cessation was associated with a decreased risk of all‐cause mortality (former heavy smokers SHR 0.85 [95% CI 0.77, 0.94], former light smokers SHR 0.90 [95% CI 0.84, 0.97]).

Conclusion

Current smoking is associated with an increased risk of all‐cause, cardiovascular, and lung cancer mortality in patients with RA. Each year of cessation is associated with a reduced risk of all‐cause mortality. This information may prove helpful in smoking cessation programs for patients with RA.

What evidence is there for a delay in diagnostic coding of RA in UK general practice records? An observational study of free text

OBJECTIVES

Much research with electronic health records (EHRs) uses coded or structured data only; important information captured in the free text remains unused. One dimension of EHR data quality assessment is 'currency' or timeliness, that is, data are representative of the patient state at the time of measurement. We explored the use of free text in UK general practice patient records to evaluate delays in recording of rheumatoid arthritis (RA) diagnosis. We also aimed to locate and quantify disease and diagnostic information recorded only in text.

SETTING

UK general practice patient records from the Clinical Practice Research Datalink.

PARTICIPANTS

294 individuals with incident diagnosis of RA between 2005 and 2008; 204 women and 85 men, median age 63 years.

PRIMARY AND SECONDARY OUTCOME MEASURES

Assessment of (1) quantity and timing of text entries for disease-modifying antirheumatic drugs (DMARDs) as a proxy for the RA disease code, and (2) quantity, location and timing of free text information relating to RA onset and diagnosis.

RESULTS

Inflammatory markers, pain and DMARDs were the most common categories of disease information in text prior to RA diagnostic code; 10-37% of patients had such information only in text. Read codes associated with RA-related text included correspondence, general consultation and arthritis codes. 64 patients (22%) had DMARD text entries >14 days prior to RA code; these patients had more and earlier referrals to rheumatology, tests, swelling, pain and DMARD prescriptions, suggestive of an earlier implicit diagnosis than was recorded by the diagnostic code.

CONCLUSIONS

RA-related symptoms, tests, referrals and prescriptions were recorded in free text with 22% of patients showing strong evidence of delay in coding of diagnosis. Researchers using EHRs may need to mitigate for delayed codes by incorporating text into their case-ascertainment strategies. Natural language processing techniques have the capability to do this at scale.

Improved survival in rheumatoid arthritis: a general population-based cohort study

OBJECTIVE

Mortality trends of rheumatoid arthritis (RA) are largely unknown over the past decade when new drugs and management strategies have been adopted to effectively treat RA.

METHODS

Using The Health Improvement Network, an electronic medical record database representative of the UK general population, we identified patients with incident RA and up to five individuals without RA matched for age, sex and year of diagnosis between 1999 and 2014. The RA cohort was divided in two sub-cohorts based on the year of RA diagnosis: the early cohort (1999-2006) and the late cohort (2007-2014). We compared mortality rates, HRs (using a Cox proportional hazard model) and rate differences (using an additive hazard model) between RA and non-RA cohorts adjusting for potential confounders.

RESULTS

Patients with RA diagnosed between 1999 and 2006 had a considerably higher mortality rate than their comparison cohort (ie, 29.1 vs 18.0 deaths/1000 person-years), as compared with a moderate difference in patients with RA diagnosed between 2007 and 2014 and their comparison cohort (17.0 vs 12.9 deaths/1000 years). The corresponding absolute mortality rate differences were 9.5 deaths/1000 person-years (95% CIs 7.5 to 11.6) and 3.1 deaths/1000 person-years (95% CI 1.5 to 4.6) and the mortality HRs were 1.56 (95% CI 1.44 to 1.69) and 1.29 (95% CI 1.17 to 1.42), respectively (both p values for interaction <0.01).

CONCLUSION

This general population-based cohort study indicates that the survival of patients with RA has improved over the past decade to a greater degree than in the general population. Improved management of RA and its associated comorbidities over recent years may be providing a survival benefit.

Oral glucocorticoid therapy and all-cause and cause-specific mortality in patients with rheumatoid arthritis: a retrospective cohort study

Previous studies of glucocorticoid (GC) therapy and mortality have had inconsistent results and have not considered possible perimortal bias—a type of protopathic bias where illness in the latter stages of life influences GC exposure, and might affect the observed relationship between GC use and death. This study aimed to investigate all-cause and cause-specific mortality in association with GC therapy in patients with rheumatoid arthritis (RA), and explore possible perimortal bias. A retrospective cohort study using the primary care electronic medical records. Oral GC exposure was identified from prescriptions. Mortality data were obtained from the UK Office for National Statistics. Multivariable Cox proportional hazards regression models assessed the association between GC use models and death. Several methods to explore perimortal bias were examined. The cohort included 16,762 patients. For ever GC use there was an adjusted hazard ratio for all-cause mortality of 1.97 (95 % CI 1.81–2.15). Current GC dose of below 5 mg per day (prednisolone equivalent dose) was not associated with an increased risk of death, but a dose–response association was seen for higher dose categories. The association between ever GC use and all-cause mortality was partly explained by perimortal bias. GC therapy was associated with an increased risk of mortality for all specific causes considered, albeit to a lesser extent for cardiovascular causes. GC use was associated with an increased risk of death in RA, at least partially explained by perimortal bias. Importantly, GC doses below 5 mg were not associated with an increased risk of death.

Defining Disease Phenotypes in Primary Care Electronic Health Records by a Machine Learning Approach: A Case Study in Identifying Rheumatoid Arthritis

OBJECTIVES

1) To use data-driven method to examine clinical codes (risk factors) of a medical condition in primary care electronic health records (EHRs) that can accurately predict a diagnosis of the condition in secondary care EHRs. 2) To develop and validate a disease phenotyping algorithm for rheumatoid arthritis using primary care EHRs.

METHODS

This study linked routine primary and secondary care EHRs in Wales, UK. A machine learning based scheme was used to identify patients with rheumatoid arthritis from primary care EHRs via the following steps: i) selection of variables by comparing relative frequencies of Read codes in the primary care dataset associated with disease case compared to non-disease control (disease/non-disease based on the secondary care diagnosis); ii) reduction of predictors/associated variables using a Random Forest method, iii) induction of decision rules from decision tree model. The proposed method was then extensively validated on an independent dataset, and compared for performance with two existing deterministic algorithms for RA which had been developed using expert clinical knowledge.

RESULTS

Primary care EHRs were available for 2,238,360 patients over the age of 16 and of these 20,667 were also linked in the secondary care rheumatology clinical system. In the linked dataset, 900 predictors (out of a total of 43,100 variables) in the primary care record were discovered more frequently in those with versus those without RA. These variables were reduced to 37 groups of related clinical codes, which were used to develop a decision tree model. The final algorithm identified 8 predictors related to diagnostic codes for RA, medication codes, such as those for disease modifying anti-rheumatic drugs, and absence of alternative diagnoses such as psoriatic arthritis. The proposed data-driven method performed as well as the expert clinical knowledge based methods.

CONCLUSION

Data-driven scheme, such as ensemble machine learning methods, has the potential of identifying the most informative predictors in a cost-effective and rapid way to accurately and reliably classify rheumatoid arthritis or other complex medical conditions in primary care EHRs.

Heart Disease, Hypertension, Gestational Diabetes Mellitus, and Preeclampsia/Eclampsia in Mothers With Juvenile Arthritis: A Nested Case-Control Study

OBJECTIVE

To determine whether women with a history of juvenile arthritis are at higher risk for heart disease and hypertension and for developing adverse maternal outcomes: gestational diabetes mellitus, maternal hypertension, and preeclampsia/eclampsia.

METHODS

We designed a nested case-control study from a cohort of first-time mothers with prior physician billing codes suggesting juvenile arthritis, and a matched comparison group without juvenile arthritis. For the nested case-control design, we selected 3 controls for each case for the outcomes of heart disease (n = 403), prepregnancy hypertension (n = 66), gestational diabetes mellitus (n = 285), maternal hypertension (n = 561), and preeclampsia/eclampsia (n = 236). We used conditional logistic regression, adjusting for maternal age and education.

RESULTS

Having juvenile arthritis was associated with heart disease (odds ratio [OR] 2.44 [95% confidence interval (95% CI) 1.15-5.15]) but not with gestational hypertension, diabetes mellitus, or preeclampsia/eclampsia. All 66 cases of prepregnancy hypertension had juvenile arthritis. Having prepregnancy hypertension was strongly associated with preeclampsia/eclampsia (OR 8.05 [95% CI 2.69-24.07]).

CONCLUSION

Women with a history of juvenile arthritis had a higher risk of heart disease. This risk signals the potential importance of cardiac prevention strategies in juvenile arthritis. As this was a retrospective study, it was not possible to correct for some relevant potential confounders. Further studies should assess the impact of medications, disease severity, and other factors (e.g., obesity) on cardiac outcomes in juvenile arthritis.

Influenza and Pneumococcal Vaccination Uptake in Patients with Rheumatoid Arthritis Treated with Immunosuppressive Therapy in the UK: A Retrospective Cohort Study Using Data from the Clinical Practice Research Datalink

INTRODUCTION

Guidelines for the management of rheumatoid arthritis (RA) recommend using influenza and pneumococcal vaccinations to mitigate infection risk. The level of adherence to these guidelines is not well known in the UK. The aims of this study were to describe the uptake of influenza and pneumococcal vaccinations in patients with RA in the UK, to compare the characteristics of those vaccinated to those not vaccinated and to compare vaccination rates across regions of the UK.

METHODS

A retrospective cohort study of adults diagnosed with incident RA and treated with non-biologic immunosuppressive therapy, using data from a large primary care database. For the influenza vaccination, patients were considered unvaccinated on 1st September each year and upon vaccination their status changed to vaccinated. For pneumococcal vaccination, patients were considered vaccinated after their first vaccination until the end of follow-up. Patients were stratified by age 65 at the start of follow-up, given differences in vaccination guidelines for the general population.

RESULTS

Overall (N=15,724), 80% patients received at least one influenza vaccination, and 50% patients received a pneumococcal vaccination, during follow-up (mean 5.3 years). Of those aged below 65 years (N=9,969), 73% patients had received at least one influenza vaccination, and 43% patients received at least one pneumococcal vaccination. Of those aged over 65 years (N=5,755), 91% patients received at least one influenza vaccination, and 61% patients had received at least one pneumococcal vaccination. Those vaccinated were older, had more comorbidity and visited the GP more often. Regional differences in vaccination rates were seen with the highest rates in Northern Ireland, and the lowest rates in London.

CONCLUSIONS

One in five patients received no influenza vaccinations and one in two patients received no pneumonia vaccine over five years of follow-up. There remains significant scope to improve uptake of vaccinations in patients with RA.

Predicting the 10-year risk of hip and major osteoporotic fracture in rheumatoid arthritis and in the general population: an independent validation and update of UK FRAX without bone mineral density

Objectives

FRAX incorporates rheumatoid arthritis (RA) as a dichotomous predictor for predicting the 10-year risk of hip and major osteoporotic fracture (MOF). However, fracture risk may deviate with disease severity, duration or treatment. Aims were to validate, and if needed to update, UK FRAX for patients with RA and to compare predictive performance with the general population (GP).

Methods

Cohort study within UK Clinical Practice Research Datalink (CPRD) (RA: n=11 582, GP: n=38 755), also linked to hospital admissions for hip fracture (CPRD-Hospital Episode Statistics, HES) (RA: n=7221, GP: n=24 227). Predictive performance of UK FRAX without bone mineral density was assessed by discrimination and calibration. Updating methods included recalibration and extension. Differences in predictive performance were assessed by the C-statistic and Net Reclassification Improvement (NRI) using the UK National Osteoporosis Guideline Group intervention thresholds.

Results

UK FRAX significantly overestimated fracture risk in patients with RA, both for MOF (mean predicted vs observed 10-year risk: 13.3% vs 8.4%) and hip fracture (CPRD: 5.5% vs 3.1%, CPRD-HES: 5.5% vs 4.1%). Calibration was good for hip fracture in the GP (CPRD-HES: 2.7% vs 2.4%). Discrimination was good for hip fracture (RA: 0.78, GP: 0.83) and moderate for MOF (RA: 0.69, GP: 0.71). Extension of the recalibrated UK FRAX using CPRD-HES with duration of RA disease, glucocorticoids (>7.5 mg/day) and secondary osteoporosis did not improve the NRI (0.01, 95% CI −0.04 to 0.05) or C-statistic (0.78).

Conclusions

UK FRAX overestimated fracture risk in RA, but performed well for hip fracture in the GP after linkage to hospitalisations. Extension of the recalibrated UK FRAX did not improve predictive performance.

Rheumatoid Arthritis and Incidence of Twelve Initial Presentations of Cardiovascular Disease: A Population Record-Linkage Cohort Study in England

INTRODUCTION

While rheumatoid arthritis is an established risk factor for cardiovascular disease (CVD), our knowledge of how the pattern of risk varies for different cardiovascular phenotypes is incomplete. The association between rheumatoid arthritis and the initial presentation of 12 types of CVDs were examined in a contemporary population of men and women of a wide age range.

METHODS

CALIBER data, which links primary care, hospital and mortality data in England, was analysed. A cohort of people aged ≥18 years and without history of CVD was assembled and included all patients with prospectively recorded rheumatoid arthritis from January 1997, until March 2010, matched with up to ten people without rheumatoid arthritis by age, sex and general practice. The associations between rheumatoid arthritis and the initial presentation of 12 types of CVDs were estimated using multivariable random effects Poisson regression models.

RESULTS

The analysis included 12,120 individuals with rheumatoid arthritis and 121,191 comparators. Of these, 2,525 patients with and 18,146 without rheumatoid arthritis developed CVDs during a median of 4.2 years of follow-up. Patients with rheumatoid arthritis had higher rates of myocardial infarction (adjusted incidence ratio [IRR] = 1.43, 95%CI 1.21-1.70), unheralded coronary death (IRR = 1.60, 95%CI 1.18-2.18), heart failure (IRR = 1.61, 95%CI 1.43-1.83), cardiac arrest (HR = 2.26, 95%CI 1.69-3.02) and peripheral arterial disease (HR = 1.36, 95%CI 1.14-1.62); and lower rates of stable angina (HR = 0.83, 95%CI 0.73-0.95). There was no evidence of association with cerebrovascular diseases, abdominal aortic aneurysm or unstable angina, or of interactions with sex or age.

CONCLUSIONS

The observed associations with some but not all types of CVDs inform both clinical practice and the selection of cardiovascular endpoints for trials and for the development of prognostic models for patients with rheumatoid arthritis.

Cardiovascular risk factor management in patients with RA compared to matched non-RA patients

Objective. RA is associated with a 50–60% increase in risk of cardiovascular (CV) death. This study aimed to compare management of CV risk factors in RA and matched non-RA patients.

Methods. A retrospective cohort study was conducted using UK clinical practice data. Patients presenting with an incident RA diagnosis were matched 1:4 to non-RA patients based on a propensity score for RA, entry year, CV risk category and treatment received at index date (date of RA diagnosis). Patients tested and treated for CV risk factors as well as those attaining CV risk factor management goals were evaluated in both groups.

Results. Between 1987 and 2010, 24 859 RA patients were identified and matched to 87 304 non-RA patients. At index date, groups had similar baseline characteristics. Annual blood pressure, lipids and diabetes-related testing were similar in both groups, although CRP and ESR were higher in RA patients at diagnosis and decreased over time. RA patients prescribed antihypertensives increased from 38.2% at diagnosis to 45.7% at 5 years, from 14.0 to 20.6% for lipid-lowering treatments and from 5.1 to 6.4% for antidiabetics. Similar treatment percentages were observed in non-RA patients, although slightly lower for antihypertensives. Modest (2%) but significantly lower attainment of lipid and diabetes goals at 1 year was observed in RA patients.

Conclusion. There were no differences between groups in the frequency of testing and treatment of CV risk factors. Higher CV risk in RA patients seems unlikely to be driven by differences in traditional CV risk factor management.

Half of U.K. patients with rheumatoid arthritis are prescribed oral glucocorticoid therapy in primary care: a retrospective drug utilisation study

BACKGROUND

Patients with rheumatoid arthritis (RA) have shared care between rheumatologists and general practitioners (GPs). Rheumatologists guide immunosuppressive therapy, whilst GPs rely on analgesia and glucocorticoid (GC) therapy to manage active disease. The objective of this study was to describe patterns of GC prescribing for patients with RA in primary care and to determine the influence of patient characteristics and prescriber.

METHODS

Incident RA patients were identified within the Clinical Practice Research Datalink, a United Kingdom (UK) primary care research database. Descriptive statistics identified patterns of oral GC prescribing. Prescribers were categorised by their tendency to prescribe GCs (high/low). Logistic regression was used to identify baseline characteristics associated with GC prescriptions during follow-up and to examine whether baseline characteristics influenced prescribing differently in high versus low prescribers.

RESULTS

A total of 7777 patients (47%) received ≥1 GC prescription during follow-up. The average daily dose was 7.5 mg (IQR 5-15.3 mg). Of those who received GCs, >50% were prescribed >10 mg/day and 20 % >30 mg/day. The median proportion of time spent on GCs was 26.3% (IQR 3.8-70.0%). Age and cardiovascular disease (CVD) were associated with increased likelihood of receiving GCs. High prescribers more commonly prescribed GC therapy in older patients and patients with hypertension.

CONCLUSIONS

Half of patients with incident RA received GCs in primary care. Average GC use was 7.5 mg for 25% of the time, perhaps higher usage than rheumatologists and GPs might expect. GCs were prescribed more commonly in certain high-risk populations, including older patients and those with CVD.

An algorithm to identify rheumatoid arthritis in primary care: a Clinical Practice Research Datalink study

OBJECTIVE

Rheumatoid arthritis (RA) is a multisystem, inflammatory disorder associated with increased levels of morbidity and mortality. While much research into the condition is conducted in the secondary care setting, routinely collected primary care databases provide an important source of research data. This study aimed to update an algorithm to define RA that was previously developed and validated in the General Practice Research Database (GPRD).

METHODS

The original algorithm consisted of two criteria. Individuals meeting at least one were considered to have RA. Criterion 1: ≥ 1 RA Read code and a disease modifying antirheumatic drug (DMARD) without an alternative indication. Criterion 2: ≥ 2 RA Read codes, with at least one 'strong' code and no alternative diagnoses. Lists of codes for consultations and prescriptions were obtained from the authors of the original algorithm where these were available, or compiled based on the original description and clinical knowledge. 4161 people with a first Read code for RA between 1 January 2010 and 31 December 2012 were selected from the Clinical Practice Research Datalink (CPRD, successor to the GPRD), and the criteria applied.

RESULTS

Code lists were updated for the introduction of new Read codes and biological DMARDs. 3577/4161 (86%) of people met the updated algorithm for RA, compared to 61% in the original development study. 62.8% of people fulfilled both Criterion 1 and Criterion 2.

CONCLUSIONS

Those wishing to define RA in the CPRD, should consider using this updated algorithm, rather than a single RA code, if they wish to identify only those who are most likely to have RA.

Statin use and mortality in rheumatoid arthritis: a general population-based cohort study

BACKGROUND

Dual lipid-lowering and anti-inflammatory properties of statins may lead to survival benefits in patients with rheumatoid arthritis (RA). However, data on this topic are limited, and the role of statins in RA remains unclear.

OBJECTIVES

To examine the association of statin use with overall mortality among patients with RA in a general population context.

METHODS

We conducted an incident user cohort study with time-stratified propensity score matching using a UK general population database. The study population included individuals aged ≥20 years who had a diagnosis of RA and had used at least one disease-modifying antirheumatic drug (DMARD) between January 2000 and December 2012. To closely account for potential confounders, we compared propensity score matched cohorts of statin initiators and comparators (non-initiators) within 1-year cohort accrual blocks.

RESULTS

432 deaths occurred during follow-up (mean 4.51 years) of the 2943 statin initiators for an incidence rate of 32.6/1000 person-years (PY), while the 513 deaths among 2943 matched comparators resulted in an incidence rate of 40.6/1000 PY. Baseline characteristics were well-balanced across the two groups. Statin initiation was associated with a 21% lower risk of all-cause mortality (HR=0.79, 95% CI 0.68 to 0.91). When we defined RA by its diagnosis code alone (not requiring DMARD use), the corresponding HR was 0.81 (95% CI 0.74 to 0.90).

CONCLUSIONS

Statin initiation is associated with a lower risk of mortality among patients with RA. The magnitude of association is similar to that seen in previous randomised trials among the general population.

Can the publication of guidelines change the management of early rheumatoid arthritis? An interrupted time series analysis from the United Kingdom

OBJECTIVE

To assess whether publication of national treatment guidelines improved the management of early RA in the UK.

METHODS

Incident diagnoses of RA in persons aged over 18 years from 1995 to 2010 were identified from the Clinical Practice Research Datalink. Using a natural experimental study design, interrupted time series analysis was used to assess whether trends in the proportion of patients receiving DMARDs, within 3 and 12 months of diagnosis, changed following publication of British Society for Rheumatology guidelines in 2006.

RESULTS

Between 1995 and 2010, 11 772 incident cases of RA were identified. There was a progressive increase in the proportion of patients prescribed any DMARD within 12 months from 43.3% in 1995 to 78.5% in 2010. After publication of the British Society for Rheumatology guidelines, the proportion of patients prescribed any DMARD within 12 months increased by 4.2% (P = 0.053). Prior to the guidance, prescribing was increasing by 1.64% per year, compared with 3.55% per year after publication (P < 0.001).

CONCLUSION

Guidelines published by a national body can improve the proportion of patients receiving DMARD treatment in the first year after diagnosis of RA.

Antibiotic Exposure and Juvenile Idiopathic Arthritis: A Case-Control Study

BACKGROUND AND OBJECTIVE

Recent evidence has linked childhood antibiotic use and microbiome disturbance to autoimmune conditions. This study tested the hypothesis that antibiotic exposure was associated with newly diagnosed juvenile idiopathic arthritis (JIA).

METHODS

We performed a nested case-control study in a population-representative medical records database from the United Kingdom. Children with newly diagnosed JIA were compared with age- and gender-matched control subjects randomly selected from general practices containing at least 1 case, excluding those with inflammatory bowel disease, immunodeficiency, or other systemic rheumatic diseases. Conditional logistic regression was used to examine the association between antibacterial antibiotics (including number of antibiotic courses and timing) and JIA after adjusting for significant confounders.

RESULTS

Any antibiotic exposure was associated with an increased rate of developing JIA (adjusted odds ratio: 2.1 [95% confidence interval: 1.2-3.5]). This relationship was dose dependent (adjusted odds ratio over 5 antibiotic courses: 3.0 [95% confidence interval: 1.6-5.6]), strongest for exposures within 1 year of diagnosis, and did not substantively change when adjusting for number or type of infections. In contrast, nonbacterial antimicrobial agents (eg, antifungal, antiviral) were not associated with JIA. In addition, antibiotic-treated upper respiratory tract infections were more strongly associated with JIA than untreated upper respiratory tract infections.

CONCLUSIONS

Antibiotics were associated with newly diagnosed JIA in a dose- and time-dependent fashion in a large pediatric population. Antibiotic exposure may play a role in JIA pathogenesis, perhaps mediated through alterations in the microbiome.

A validation study of a new classification algorithm to identify rheumatoid arthritis using administrative health databases: case-control and cohort diagnostic accuracy studies. Results from the RECord linkage On Rheumatic Diseases study of the Italian Society for Rheumatology

OBJECTIVES

To develop and validate a new algorithm to identify patients with rheumatoid arthritis (RA) and estimate disease prevalence using administrative health databases (AHDs) of the Italian Lombardy region.

DESIGN

Case-control and cohort diagnostic accuracy study.

METHODS

In a randomly selected sample of 827 patients drawn from a tertiary rheumatology centre (training set), clinically validated diagnoses were linked to administrative data including diagnostic codes and drug prescriptions. An algorithm in steps of decreasing specificity was developed and its accuracy assessed calculating sensitivity/specificity, positive predictive value (PPV)/negative predictive value, with corresponding CIs. The algorithm was applied to two validating sets: 106 patients from a secondary rheumatology centre and 6087 participants from the primary care. Alternative algorithms were developed to increase PPV at population level. Crude and adjusted prevalence estimates taking into account algorithm misclassification rates were obtained for the Lombardy region.

RESULTS

The algorithms included: RA certification by a rheumatologist, certification for other autoimmune diseases by specialists, RA code in the hospital discharge form, prescription of disease-modifying antirheumatic drugs and oral glucocorticoids. In the training set, a four-step algorithm identified clinically diagnosed RA cases with a sensitivity of 96.3 (95% CI 93.6 to 98.2) and a specificity of 90.3 (87.4 to 92.7). Both external validations showed highly consistent results. More specific algorithms achieved >80% PPV at the population level. The crude RA prevalence in Lombardy was 0.52%, and estimates adjusted for misclassification ranged from 0.31% (95% CI 0.14% to 0.42%) to 0.37% (0.25% to 0.47%).

CONCLUSIONS

AHDs are valuable tools for the identification of RA cases at the population level, and allow estimation of disease prevalence and to select retrospective cohorts.

Validation of the rheumatoid arthritis diagnosis in the Swedish National Patient Register: a cohort study from Stockholm County

Background

The Swedish National Patient Register offers unique possibilities for identification of large cohorts, such as patients with rheumatoid arthritis (RA). Although the overall diagnostic validity in the register has been reported as good, the aims of this study were to a) specifically validate the RA diagnosis from contemporary outpatient specialist care in this register, and b) assess the proportion of patients identified via algorithms to define incident RA in the register who in clinical practice also have new-onset disease.

Methods

211 individuals with prevalent or incident RA in the National Patient Register were included. By extracting diagnosis-related parameters from their medical records, we determined if the patient fulfilled the 2010 ACR/EULAR- and the 1987 ACR-classification criteria for RA. We also determined whether clinical diagnosis was synchronous with disease onset as defined through register-based algorithms.

Results

For 91% of the prevalent patients, the RA diagnosis in the National Patient Register fulfilled classification criteria or clinical diagnosis for RA. Among individuals identified with incident RA using a strict algorithm for new-onset disease, the RA diagnosis was substantiated in 91%, of whom 92% also represented new-onset disease.

Conclusions

The validity of the RA diagnosis in the National Patient Register was high and, by using specific algorithms, new-onset RA can be defined. These findings strengthen the notion that the National Patient Register may be used to define RA populations with high validity to allow for high-quality epidemiological studies.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2474-15-432) contains supplementary material, which is available to authorized users.

The risk of major and any (non-hip) fragility fracture after hip fracture in the United Kingdom: 2000-2010

The risk of a subsequent major or any fracture after a hip fracture and secular trends herein were examined. Within 1 year, 2.7 and 8.4% of patients sustained a major or any (non-hip) fracture, which increased to 14.7 and 32.5% after 5 years. Subsequent fracture rates increased during the study period both for major and any (non-hip) fracture.

INTRODUCTION

Hip fractures are associated with subsequent fractures, particularly in the year following initial fracture. Age-adjusted hip fracture rates have stabilised in many developed countries, but secular trends in subsequent fracture remain poorly documented. We thus evaluated secular trends (2000-2010) and determinants for the risk of a subsequent major (humerus, vertebral, or forearm) and any (non-hip) fracture after hip fracture.

METHODS

Patients ≥50 years with a hip fracture between 2000 and 2010 were extracted from the UK Clinical Practice Research Datalink (n = 30,516). Incidence rates, cumulative incidence probabilities, and adjusted hazard ratios (aHRs) were calculated.

RESULTS

Within 1 year following hip fracture, 2.7 and 8.4% of patients sustained a major or any (non-hip) fracture, which increased to 14.7 and 32.5% after 5 years, respectively. The most important risk factors for a subsequent major fracture within 1 year were the female gender [aHR 1.90, 95% confidence interval (CI) 1.51-2.40] and a history of secondary osteoporosis (aHR 1.54, 95% CI 1.17-2.02). The annual risk increased during the study period for both subsequent major (2009-2010 vs. 2000-2002: aHR 1.44, 95% CI 1.12-1.83) and any (non-hip) facture (2009-2010 vs. 2000-2002: aHR 1.80, 95% CI 1.58-2.06).

CONCLUSION

The risk of sustaining a major or any (non-hip) fracture after hip fracture is small in the first year. However, given the recent rise in secondary fracture rates and the substantial risk of subsequent fracture in the longer term, fracture prevention is clearly indicated for patients who have sustained a hip fracture.

The descriptive epidemiology of rheumatoid arthritis in Catalonia: a retrospective study using routinely collected data

Information on the epidemiology of rheumatoid arthritis (RA) in Southern Europe is scarce. We estimated the age- and gender-adjusted incidence and prevalence of RA in Catalonia using routinely collected primary care records. We identified incident (2009-2012) and prevalent (on 31 December 2012) cases of RA in the SIDIAP database using ICD-10 codes. SIDIAP contains anonymized data from computerized primary care records for about five million adults (>80 % of the population). We estimated age- (5-year groups) and gender-specific, and directly standardized incidence and prevalence of RA and confidence intervals (95% CIs) assuming a Poisson distribution. A total of 20,091 prevalent (among whom 5,796 incident) cases of RA were identified among 4,796,498 study participants observed for up to 4 years. Rates of RA increased with age in both genders, peaking at the age of 65-70 years. Age- and gender-standardized incidence and prevalence rates were 0.20/1,000 person-years (95% CI 0.19-0.20) and 4.17/1,000 (4.11-4.23) respectively. Rheumatoid factor was positive (≥10 IU/mL) in 1,833 (73.9 %) of 2,482 cases tested in primary care. The incidence and prevalence of RA in Catalonia are similar to those of other Southern European regions, and lower than those of northern areas. This data will inform health care planning and resource allocation.

What is the risk of having a total hip or knee replacement for patients with lupus?

Individuals with systemic lupus erythematosus (SLE) frequently have arthralgia but joint damage leading to surgery is thought to be less common. In addition to inflammatory damage, other reasons like avascular necrosis (AVN), which is often associated with steroid use, excessive alcohol intake and antiphospholipid syndrome (APS), may increase the likelihood of large joint failure. In this study we aimed to determine the likelihood of having a total hip replacement (THR) or total knee replacement (TKR) for individuals with SLE compared to those without lupus, by performing a retrospective matched case control study of all THRs and TKRs that were performed between 1991 and 2011 and recorded in the General Practice Research Database (GPRD). Individuals with inflammatory arthritis due to any other causes were excluded and the results were adjusted for steroid use, alcohol consumption (drinking status) and APS. The results show that patients with lupus who had a THR or TKR were younger than their peers without lupus. In addition, they appeared to have a significantly increased risk of TKR but the increased risk of THR did not remain after adjustment for steroid use, alcohol consumption and APS.

An administrative data validation study of the accuracy of algorithms for identifying rheumatoid arthritis: the influence of the reference standard on algorithm performance

Background

We have previously validated administrative data algorithms to identify patients with rheumatoid arthritis (RA) using rheumatology clinic records as the reference standard. Here we reassessed the accuracy of the algorithms using primary care records as the reference standard.

Methods

We performed a retrospective chart abstraction study using a random sample of 7500 adult patients under the care of 83 family physicians contributing to the Electronic Medical Record Administrative data Linked Database (EMRALD) in Ontario, Canada. Using physician-reported diagnoses as the reference standard, we computed and compared the sensitivity, specificity, and predictive values for over 100 administrative data algorithms for RA case ascertainment.

Results

We identified 69 patients with RA for a lifetime RA prevalence of 0.9%. All algorithms had excellent specificity (>97%). However, sensitivity varied (75-90%) among physician billing algorithms. Despite the low prevalence of RA, most algorithms had adequate positive predictive value (PPV; 51-83%). The algorithm of “[1 hospitalization RA diagnosis code] or [3 physician RA diagnosis codes with ≥1 by a specialist over 2 years]” had a sensitivity of 78% (95% CI 69–88), specificity of 100% (95% CI 100–100), PPV of 78% (95% CI 69–88) and NPV of 100% (95% CI 100–100).

Conclusions

Administrative data algorithms for detecting RA patients achieved a high degree of accuracy amongst the general population. However, results varied slightly from our previous report, which can be attributed to differences in the reference standards with respect to disease prevalence, spectrum of disease, and type of comparator group.

Temporal growth and geographic variation in the use of laboratory tests by NHS general practices: using routine data to identify research priorities

BACKGROUND

Laboratory tests are extensively used for diagnosis and monitoring in UK primary care. Test usage by GPs, and associated costs, have grown substantially in recent years.

AIM

This study aimed to quantify temporal growth and geographic variation in utilisation of laboratory tests.

DESIGN AND SETTING

Retrospective cohort study using data from general practices in the UK.

METHOD

Data from the General Practice Research Database, including patient demographics, clinical details, and laboratory test results, were used to estimate rates of change in utilisation between 2005 and 2009, and identify tests with greatest inter-regional variation, by fitting random-effects Poisson regression models. The study also investigated indications for test requests, using diagnoses and symptoms recorded in the 2 weeks before each test.

RESULTS

Around 660 000 tests were recorded in 230 000 person-years of follow-up. Test use increased by 24.2%, from 23 872 to 29 644 tests per 10 000 person-years, between 2005 and 2009. Tests with the largest increases were faecal occult blood (121%) and C-reactive protein (86%). There was substantial geographic variation in test utilisation; GPs in some regions requested tests such as plasma viscosity and cardiac enzymes at a rate more than three times the national average.

CONCLUSION

Increases in the use of laboratory tests have substantial resource implications. Rapid increases in particular tests may be supported by evidence-based guidelines, but these are often vague about who should be tested, how often, and for how long. Substantial regional variation in test use may reflect uncertainty about diagnostic accuracy and appropriate indications for the laboratory test. There is a need for further research on the diagnostic accuracy, therapeutic impact, and effect on patient health outcomes of the most rapidly increasing and geographically variable tests.

Diabetes incidence in psoriatic arthritis, psoriasis and rheumatoid arthritis: a UK population-based cohort study

OBJECTIVE

The objective of this study was to evaluate the incidence of diabetes among patients with PsA and RA in the general population.

METHODS

We conducted a cohort study using an electronic medical records database representative of the UK general population (1986-2010). We estimated hazard ratios (HRs) for incident diabetes in PsA, psoriasis and RA cohorts compared with age- and sex-matched comparison cohorts without the corresponding conditions, adjusting for BMI, smoking, alcohol use, co-morbidities and glucocorticoids at baseline.

RESULTS

Cohorts included 4196 persons with PsA, 59 281 with psoriasis and 11 158 with RA, with mean follow-up times of 5.9, 5.8 and 5.5 years, respectively. Incidence rates for diabetes were 7.3, 6.4 and 6.3 cases per 1000 person-years among individuals with PsA, psoriasis and RA, respectively. Age- and sex-matched HRs for diabetes were 1.72 (95% CI 1.46, 2.02) in PsA, 1.39 (95% CI 1.32, 1.45) in psoriasis and 1.12 (95% CI 1.01, 1.25) in RA. After adjustment for BMI, smoking and alcohol, the HRs were attenuated substantially (1.43, 1.24 and 1.00, respectively). With further adjustment for baseline glucocorticoid use and co-morbidities, the HRs were 1.33 (1.09, 1.61) in PsA, 1.21 (1.15, 1.27) in psoriasis and 0.94 (0.84, 1.06) in RA.

CONCLUSION

This general population study suggests an increased incidence of diabetes in PsA and RA, which is substantially explained by obesity and lifestyle factors. These findings support the importance of managing such factors in PsA and RA patients.

Cardiovascular screening in rheumatoid arthritis: a cross-sectional primary care database study

Background

Patients with rheumatoid arthritis (RA) are known to be at increased risk of vascular disease. It is not known whether screening for vascular risk factors occurs in primary care. The aim of this study was to determine whether guidance advocating cardiovascular screening in RA patients is being implemented in primary care.

Methods

This study was undertaken in a UK primary care consultation database. All patients with a diagnosis of RA between 2000 and 2008, and still registered with the GP practice in 2009 were matched by age, gender and GP practice to three non-RA patients. Evidence of screening for five traditional vascular risk factors (blood pressure, lipids, glucose, weight, smoking) was compared in those with and without RA using logistic regression models. A comparison was also made with diabetes.

Results

401 RA patients were identified and matched to 1198 non-RA patients. No differences in the overall rates of screening were found (all five risk factors: RA 24.9% vs no RA 25.6%), but RA patients were more likely to have a smoking status recorded (67% versus 62%). In contrast, those with diabetes were up to 12 times as likely to receive vascular screening.

Conclusions

Despite the excess risk of vascular disease in patients with RA being of a similar magnitude to that seen in diabetes, patients with RA did not receive additional CVD screening in primary care, although this was achieved in patients with diabetes. More emphasis needs to be placed on ensuring those with RA are actively screened for cardiovascular disease in primary care.

Optimising use of electronic health records to describe the presentation of rheumatoid arthritis in primary care: a strategy for developing code lists

Background

Research using electronic health records (EHRs) relies heavily on coded clinical data. Due to variation in coding practices, it can be difficult to aggregate the codes for a condition in order to define cases. This paper describes a methodology to develop ‘indicator markers’ found in patients with early rheumatoid arthritis (RA); these are a broader range of codes which may allow a probabilistic case definition to use in cases where no diagnostic code is yet recorded.

Methods

We examined EHRs of 5,843 patients in the General Practice Research Database, aged ≥30y, with a first coded diagnosis of RA between 2005 and 2008. Lists of indicator markers for RA were developed initially by panels of clinicians drawing up code-lists and then modified based on scrutiny of available data. The prevalence of indicator markers, and their temporal relationship to RA codes, was examined in patients from 3y before to 14d after recorded RA diagnosis.

Findings

Indicator markers were common throughout EHRs of RA patients, with 83.5% having 2 or more markers. 34% of patients received a disease-specific prescription before RA was coded; 42% had a referral to rheumatology, and 63% had a test for rheumatoid factor. 65% had at least one joint symptom or sign recorded and in 44% this was at least 6-months before recorded RA diagnosis.

Conclusion

Indicator markers of RA may be valuable for case definition in cases which do not yet have a diagnostic code. The clinical diagnosis of RA is likely to occur some months before it is coded, shown by markers frequently occurring ≥6 months before recorded diagnosis. It is difficult to differentiate delay in diagnosis from delay in recording. Information concealed in free text may be required for the accurate identification of patients and to assess the quality of care in general practice.

A retrospective cohort study: 10-year trend of disease-modifying antirheumatic drugs and biological agents use in patients with rheumatoid arthritis at Veteran Affairs Medical Centers

OBJECTIVES

To evaluate the trends in patterns of disease-modifying antirheumatic drugs (DMARDs) and biological agents use from 1999 to 2009 and to identify patient characteristics associated with different patterns of their use in a national sample of Veterans with rheumatoid arthritis (RA).

DESIGN

A retrospective cohort study.

SETTINGS

Administrative databases of the USA Department of Veterans Affairs.

PARTICIPANTS

An incident cohort of 13 254 patients with newly diagnosed RA was identified.

PRIMARY OUTCOME MEASURES

Trends and choice of DMARDs and biological agents' usage, and time intervals between RA diagnosis and treatment

RESULTS

Methotrexate use as first-line agent increased from 39.9% to 57.2% over the study period (p<0.001). Although biological dispensations increased over other DMARDs and biological agents, from 3.4% to 25% from 1999 to 2009, the percentage of RA patients diagnosed between 1999 and 2007 who had biologics dispensations remained steady at 23.3-26.7%. Compared with Caucasian, African Americans were less likely to receive biologics (HR 0.71, 95% CI 0.63 to 0.81). Patients aged 75 and older were less likely to receive biologics than those younger than 45 (HR 0.29, 95% CI 0.23 to 0.36). The time interval between RA diagnosis and treatment with DMARDs and biological agents decreased significantly over time (median: 51 days in 1999-2001 to 28 days in 2006-2007).

CONCLUSIONS

Methotrexate use increased as it became the preferred first-line agent, while other traditional agents declined. Dispensation of biologics increased significantly, but the proportion of RA patients eventually given biologics stabilised below 30%. A significant shorter time between RA diagnosis and DMARD or biological agent initiation in recent years suggests improvements in quality of care. There were disproportionately lower use of biologics in certain age and ethnic groups, and further studies will be needed to elucidate these observations.

The risk of pulmonary embolism and deep vein thrombosis in rheumatoid arthritis: a UK population-based outpatient cohort study

Background

Recent hospital-based studies have suggested a sixfold increased risk of pulmonary embolism (PE) in rheumatoid arthritis (RA) in the year following admission. We evaluated the risk of PE and deep vein thrombosis (DVT) and associated time trend among RA patients (84.5% without a history of hospitalisation during the past year) derived from the general population.

Methods

We conducted a cohort study using an electronic medical records database representative of the UK general population, collected from 1986 to 2010. Primary definitions of the RA cohort (exposure) and PE/DVT outcomes required physician diagnoses followed by corresponding treatments. We estimated relative risks (RRs) of PE and DVT compared with a matched non-RA comparison cohort, adjusting for age, sex, smoking, body mass index, comorbidities and hospitalisations.

Results

Among 9589 individuals with RA (69% female, mean age of 58 years), 82 developed PE and 110 developed DVT (incidence rates, 1.5 and 2.1 per 1000 person-years). Compared with non-RA individuals (N=95 776), the age-, sex- and entry-time-matched RRs were 2.23 (95% CI 1.75 to 2.86) for PE and 2.20 (CI 1.78 to 2.71) for DVT. Adjusting for other covariates, the corresponding RRs were 2.16 (CI 1.68 to 2.79) and 2.16 (CI 1.74 to 2.69). The time-specific RRs for PE were 3.27, 1.88 and 2.35 for follow-up times of <1 year, 1–4.9 years, and ≥5 years, and corresponding RRs for DVT were 3.16, 1.82 and 2.32.

Conclusions

This population-based study indicates an increased risk of PE and DVT in RA, supporting increased monitoring of venous-thromboembolic complications and risk factors in RA, regardless of hospitalisation.

Regional and temporal variation in the treatment of rheumatoid arthritis across the UK: a descriptive register-based cohort study

OBJECTIVES

To describe current disease-modifying antirheumatic drugs (DMARDs) prescription in rheumatoid arthritis (RA) with reference to best practice and to identify temporal and regional trends in the UK.

DESIGN

Descriptive, register-based cohort study.

PARTICIPANTS

Permanently registered patients aged ≥18 years with a recorded diagnosis of RA between 1 January 1995 and 31 March 2010 and matched controls. Participants with RA were identified through screening of all patients in the General Practice Research Database (GPRD) with a clinical or referral record for RA and at least 1 day of follow-up.

SETTING

639 general practices in the UK supplying data to the GPRD.

MAIN OUTCOME MEASURES

Medication prescribing between 3 and 12 months of RA diagnosis by region and time period (1995-1999, 2000-2005 and 2006-April 2010).

RESULTS

Of the 35 911 patients in the full RA cohort, 15 259 patients (42%) had incident RA. Analysis of prescribing in incident RA patients demonstrated that between 1995 (baseline) and 2010 there was a substantial increase in DMARD, and specifically methotrexate, prescribing across all regions with a less marked increase in combination DMARD prescribing. Taking 12-month prescribing as a snapshot: DMARD prescribing was 19-49% at baseline increasing to 45-74% by 2006-April 2010; methotrexate prescribing was 4-16% at baseline increasing to 32-60%; combination DMARD prescribing was 0-8% at baseline increasing to 3-17%. However, there was marked regional variation in the proportion of RA patients receiving DMARD regardless of time period.

CONCLUSIONS

There has been a substantial increase in prescribing of DMARDs for RA since 1995; however, regional variation persists across the UK with relative undertreatment, according to established best practice. Improved implementation of evidence-based best clinical practice to facilitate removal of treatment variation is warranted. This may occur as a result of the implementation of published national guidance.

Potential application of item-response theory to interpretation of medical codes in electronic patient records

Background

Electronic patient records are generally coded using extensive sets of codes but the significance of the utilisation of individual codes may be unclear. Item response theory (IRT) models are used to characterise the psychometric properties of items included in tests and questionnaires. This study asked whether the properties of medical codes in electronic patient records may be characterised through the application of item response theory models.

Methods

Data were provided by a cohort of 47,845 participants from 414 family practices in the UK General Practice Research Database (GPRD) with a first stroke between 1997 and 2006. Each eligible stroke code, out of a set of 202 OXMIS and Read codes, was coded as either recorded or not recorded for each participant. A two parameter IRT model was fitted using marginal maximum likelihood estimation. Estimated parameters from the model were considered to characterise each code with respect to the latent trait of stroke diagnosis. The location parameter is referred to as a calibration parameter, while the slope parameter is referred to as a discrimination parameter.

Results

There were 79,874 stroke code occurrences available for analysis. Utilisation of codes varied between family practices with intraclass correlation coefficients of up to 0.25 for the most frequently used codes. IRT analyses were restricted to 110 Read codes. Calibration and discrimination parameters were estimated for 77 (70%) codes that were endorsed for 1,942 stroke patients. Parameters were not estimated for the remaining more frequently used codes. Discrimination parameter values ranged from 0.67 to 2.78, while calibration parameters values ranged from 4.47 to 11.58. The two parameter model gave a better fit to the data than either the one- or three-parameter models. However, high chi-square values for about a fifth of the stroke codes were suggestive of poor item fit.

Conclusion

The application of item response theory models to coded electronic patient records might potentially contribute to identifying medical codes that offer poor discrimination or low calibration. This might indicate the need for improved coding sets or a requirement for improved clinical coding practice. However, in this study estimates were only obtained for a small proportion of participants and there was some evidence of poor model fit. There was also evidence of variation in the utilisation of codes between family practices raising the possibility that, in practice, properties of codes may vary for different coders.

Does the presence of connective tissue disease modify survival in patients with pulmonary fibrosis?

OBJECTIVES

Previous studies into the survival differences between individuals with idiopathic pulmonary fibrosis and those with connective tissue disease associated pulmonary fibrosis (CTD-PF) have yielded mixed results. The aim of this study is to compare the survival of individuals with CTD-PF to those with idiopathic pulmonary fibrosis clinical syndrome (IPF-CS) using data derived from The Health Improvement network, a large primary care database in the UK.

METHODS

Incident cases of CTD-PF and IPF-CS between the years 2000-2009 were identified. Survival analysis was performed using Kaplan-Meier methods, stratified by type of connective tissue disease. Cox regression was then used to compare mortality rates between the groups, adjusting for age, gender and year of diagnosis.

RESULTS

A total of 324 cases of CTD-PF and 2209 cases of IPF-CS were followed up over a mean period of 2.3 years. During this period, 113 (34.9%) cases of CTD-PF and 1073 (48.6%) cases of IPF-CS died. The mortality rates for cases with CTD-PF and IPF-CS were 123.6 per 1000 person years (95%CI: 102.8-148.9) and 229.8 per 1000 person years (95% CI: 216.4-244.0) respectively. After adjusting for age, sex and year of diagnosis, cases with CTD-PF had a better prognosis compared to those with IPF-CS (HR 0.76,95%CI: 0.62-0.92).

CONCLUSION

The prognosis of individuals with CTD-PF appears to be significantly better than those with IPF-CS, but remains an important cause of death in patients with connective tissue disease, and requires more effective treatment options.

Prevalence of long-term oral glucocorticoid prescriptions in the UK over the past 20 years

OBJECTIVE

To assess trends in long-term (i.e. ≥3 months) oral glucocorticoid (GC) prescriptions over the past 20 years.

METHODS

Data of UK adult patients registered between January 1989 and December 2008 with general practices contributing to The Health Improvement Network (THIN) database were obtained. The annual prevalence of long-term oral GC prescriptions was assessed in the whole population and specifically in people with RA, PMR/GCA, asthma, chronic obstructive pulmonary disease (COPD), Crohn's disease and ulcerative colitis (UC). Trends over the 20-year period were estimated using sex- and age-adjusted Poisson regression models.

RESULTS

During the 26 035 154 person-years of follow-up, an average of 0.75% (95% CI 0.74, 0.75) of the study population was prescribed long-term oral GC therapy at any time point. This rose from 0.59% (0.52, 0.67) in 1989 to 0.79% (0.78, 0.80) in 2008. Long-term prescriptions significantly increased in patients with RA [from 10.3% (8.7, 11.9) to 13.6% (12.9, 14.2)] and PMR/GCA [from 57.6% (53.3, 62.0) to 66.5% (65.2, 67.7)], decreased in patients with asthma, COPD and Crohn's disease and remained stable in patients with UC. However, when only incident cases were considered, we found a decreased use of GCs in patients with RA and UC [odds ratio 0.97 (95% CI 0.96, 0.97) and 0.94 (95% CI 0.93, 0.96) per increasing year, respectively].

CONCLUSION

Over the past 20 years, long-term oral GC prescriptions have increased by 34%. Patients newly diagnosed with RA, Crohn's disease or UC are, however, less likely to receive long-term GC prescriptions than patients with a long past medical history of the disease, suggesting changes in physicians' practice.

Validation of psoriatic arthritis diagnoses in electronic medical records using natural language processing

OBJECTIVES

To test whether data extracted from full text patient visit notes from an electronic medical record would improve the classification of psoriatic arthritis (PsA) compared with an algorithm based on codified data.

METHODS

From the >1,350,000 adults in a large academic electronic medical record, all 2318 patients with a billing code for PsA were extracted and 550 were randomly selected for chart review and algorithm training. Using codified data and phrases extracted from narrative data using natural language processing, 31 predictors were extracted and 3 random forest algorithms were trained using coded, narrative, and combined predictors. The receiver operator curve was used to identify the optimal algorithm and a cut-point was chosen to achieve the maximum sensitivity possible at a 90% positive predictive value (PPV). The algorithm was then used to classify the remaining 1768 charts and finally validated in a random sample of 300 cases predicted to have PsA.

RESULTS

The PPV of a single PsA code was 57% (95% CI 55%-58%). Using a combination of coded data and natural language processing (NLP), the random forest algorithm reached a PPV of 90% (95% CI 86%-93%) at a sensitivity of 87% (95% CI 83%-91%) in the training data. The PPV was 93% (95% CI 89%-96%) in the validation set. Adding NLP predictors to codified data increased the area under the receiver operator curve (P < 0.001).

CONCLUSIONS

Using NLP with text notes from electronic medical records improved the performance of the prediction algorithm significantly. Random forests were a useful tool to accurately classify psoriatic arthritis cases to enable epidemiological research.