The influence of population characteristics on variation in general practice based morbidity estimations

Trends in the Annual Consultation Incidence and Prevalence of Low Back Pain and Osteoarthritis in England from 2000 to 2019: Comparative Estimates from Two Clinical Practice Databases

Purpose

To compare estimates of annual person-consulting incidence and prevalence of low back pain (LBP) and osteoarthritis for two national English electronic health record databases (Clinical Practice Research Datalink (CPRD) Aurum and CPRD GOLD).

Patients and Methods

Retrospective, population-based, longitudinal cohort study. LBP and osteoarthritis cases were defined using established codelists in people aged ≥15 and ≥45 years, respectively. Incident cases were new recorded cases in a given calendar year with no relevant consultation in the previous 3 years (denominator = exact person-time in the same calendar year for the at-risk population). Prevalent cases were individuals with ≥1 consultation for the condition of interest recorded in a given calendar year, irrespective of prior consultations for the same condition (denominator = all patients with complete registration history in the previous 3 years). We estimated age-sex standardised incidence and annual (12-month period) prevalence for both conditions in 2000–2019, overall, and by sex, age group, and region.

Results

Standardised incidence and prevalence of LBP from Aurum were lower than those from GOLD until 2014, after which estimates were similar. Both databases showed recent declines in incidence and prevalence of LBP: declines began earlier in GOLD (after 2012–2014) than Aurum (after 2014–2015). Standardised incidence (after 2011) and prevalence of osteoarthritis (after 2003) were higher in Aurum than GOLD and showed different trends: incidence and prevalence were stable or increasing in Aurum, decreasing in GOLD. Stratified estimates in CPRD Aurum suggested consistently higher occurrence among women, older age groups, and those living in the north of England.

Conclusion

Comparative analyses of two English databases produced conflicting estimates and trends for two common musculoskeletal conditions. Aurum estimates appeared more consistent with external sources and may be useful for monitoring population musculoskeletal health and healthcare demand, but they remain sensitive to analytic decisions and data quality.

Calculating incidence rates and prevalence proportions: not as simple as it seems

BACKGROUND

Incidence rates and prevalence proportions are commonly used to express the populations health status. Since there are several methods used to calculate these epidemiological measures, good comparison between studies and countries is difficult. This study investigates the impact of different operational definitions of numerators and denominators on incidence rates and prevalence proportions.

METHODS

Data from routine electronic health records of general practices contributing to NIVEL Primary Care Database was used. Incidence rates were calculated using different denominators (person-years at-risk, person-years and midterm population). Three different prevalence proportions were determined: 1 year period prevalence proportions, point-prevalence proportions and contact prevalence proportions.

RESULTS

One year period prevalence proportions were substantially higher than point-prevalence (58.3 - 206.6%) for long-lasting diseases, and one year period prevalence proportions were higher than contact prevalence proportions (26.2 - 79.7%). For incidence rates, the use of different denominators resulted in small differences between the different calculation methods (-1.3 - 14.8%). Using person-years at-risk or a midterm population resulted in higher rates compared to using person-years.

CONCLUSIONS

All different operational definitions affect incidence rates and prevalence proportions to some extent. Therefore, it is important that the terminology and methodology is well described by sources reporting these epidemiological measures. When comparing incidence rates and prevalence proportions from different sources, it is important to be aware of the operational definitions applied and their impact.

Routine primary care data for scientific research, quality of care programs and educational purposes: the Julius General Practitioners' Network (JGPN)

BACKGROUND

General Practitioners (GPs) in the Netherlands routinely register all patient contacts electronically. These records include longitudinally gathered clinical information of the patient contacts in coded data and free text.

METHODS

Diagnoses are coded according to the International Coding of Primary Care (ICPC). Drug prescriptions are labelled with the Anatomical Therapeutic Chemical Classification (ATC), and letters of hospital specialists and paramedic health care professionals are linked or directly incorporated in the electronic medical files. A network of a large group of GPs collecting routine care data on an ongoing basis can be used for answering various research questions.

RESULTS

The Julius General Practitioners' Network (JGPN) database consists of routine care data from over ten years of a dynamic cohort of around 370,000 individuals registered with the participating GPs from the city of Utrecht and its vicinity. Health care data are extracted anonymously every quartile of a year and these data are used by researchers.

CONCLUSION

We describe the content and usability of our JGPN database, and how a wide variety of research questions could be answered, as illustrated with examples of published articles.

Population trends in the incidence and initial management of osteoarthritis: age-period-cohort analysis of the Clinical Practice Research Datalink, 1992-2013

Objective

To determine recent trends in the rate and management of new cases of OA presenting to primary healthcare using UK nationally representative data.

Methods

Using the Clinical Practice Research Datalink we identified new cases of diagnosed OA and clinical OA (including OA-relevant peripheral joint pain in those aged over 45 years) using established code lists. For both definitions we estimated annual incidence density using exact person-time, and undertook descriptive analysis and age-period-cohort modelling. Demographic characteristics and management were described for incident cases in each calendar year. Sensitivity analyses explored the robustness of the findings to key assumptions.

Results

Between 1992 and 2013 the annual age-sex standardized incidence rate for clinical OA increased from 29.2 to 40.5/1000 person-years. After controlling for period effects, the consultation incidence of clinical OA was higher for successive cohorts born after the mid-1950s, particularly women. In contrast, with the exception of hand OA, we observed no increase in the incidence of diagnosed OA: 8.6/1000 person-years in 2004 down to 6.3 in 2013. In 2013, 16.4% of clinical OA cases had an X-ray referral. While NSAID prescriptions fell from 2004, the proportion prescribed opioid analgesia rose markedly (0.1% of diagnosed OA in 1992 to 1.9% in 2013).

Conclusion

Rising rates of clinical OA, continued use of plain radiography and a shift towards opioid analgesic prescription are concerning. Our findings support the search for policies to tackle this common problem that promote joint pain prevention while avoiding excessive and inappropriate health care.

Estimating incidence and prevalence rates of chronic diseases using disease modeling

BACKGROUND

Morbidity estimates between different GP registration networks show large, unexplained variations. This research explores the potential of modeling differences between networks in distinguishing new (incident) cases from existing (prevalent) cases in obtaining more reliable estimates.

METHODS

Data from five Dutch GP registration networks and data on four chronic diseases (chronic obstructive pulmonary disease [COPD], diabetes, heart failure, and osteoarthritis of the knee) were used. A joint model (DisMod model) was fitted using all information on morbidity (incidence and prevalence) and mortality in each network, including a factor for misclassification of prevalent cases as incident cases.

RESULTS

The observed estimates vary considerably between networks. Using disease modeling including a misclassification term improved the consistency between prevalence and incidence rates, but did not systematically decrease the variation between networks. Osteoarthritis of the knee showed large modeled misclassifications, especially in episode of care-based registries.

CONCLUSION

Registries that code episodes of care rather than disease generally provide lower estimates of the prevalence of chronic diseases requiring low levels of health care such as osteoarthritis. For other diseases, modeling misclassification rates does not systematically decrease the variation between registration networks. Using disease modeling provides insight in the reliability of estimates.

Dementia incidence trend over 1992-2014 in the Netherlands: Analysis of primary care data

BACKGROUND

Recent reports have suggested declining age-specific incidence rates of dementia in high-income countries over time. Improved education and cardiovascular health in early age have been suggested to be bringing about this effect. The aim of this study was to estimate the age-specific dementia incidence trend in primary care records from a large population in the Netherlands.

METHODS AND FINDINGS

A dynamic cohort representative of the Dutch population was composed using primary care records from general practice registration networks (GPRNs) across the country. Data regarding dementia incidence were obtained using general-practitioner-recorded diagnosis of dementia within the electronic health records. Age-specific dementia incidence rates were calculated for all persons aged 60 y and over; negative binomial regression analysis was used to estimate the time trend. Nine out of eleven GPRNs provided data on more than 800,000 older people for the years 1992 to 2014, corresponding to over 4 million person-years and 23,186 incident dementia cases. The annual growth in dementia incidence rate was estimated to be 2.1% (95% CI 0.5% to 3.8%), and incidence rates were 1.08 (95% CI 1.04 to 1.13) times higher for women compared to men. Despite their relatively low numbers of person-years, the highest age groups contributed most to the increasing trend. There was no significant overall change in incidence rates since the start of a national dementia program in 2003 (-0.025; 95% CI -0.062 to 0.011). Increased awareness of dementia by patients and doctors in more recent years may have influenced dementia diagnosis by general practitioners in electronic health records, and needs to be taken into account when interpreting the data.

CONCLUSIONS

Within the clinical records of a large, representative sample of the Dutch population, we found no evidence for a declining incidence trend of dementia in the Netherlands. This could indicate true stability in incidence rates, or a balance between increased detection and a true reduction. Irrespective of the exact rates and mechanisms underlying these findings, they illustrate that the burden of work for physicians and nurses in general practice associated with newly diagnosed dementia has not been subject to substantial change in the past two decades. Hence, with the ageing of Western societies, we still need to anticipate a dramatic absolute increase in dementia occurrence over the years to come.

Inter-practice variation in polypharmacy prevalence amongst older patients in primary care

PURPOSE

Complex medication management in older people with multiple chronic conditions can introduce practice variation in polypharmacy prevalence. This study aimed to determine the inter-practice variation in polypharmacy prevalence and examine how this variation was influenced by patient and practice characteristics.

METHODS

This cohort study included 45,731 patients aged 55 years and older with at least one prescribed medication from 126 general practices that participated in NIVEL Primary Care Database in the Netherlands. Medication dispensing data of the year 2012 were used to determine polypharmacy. Polypharmacy was defined as the chronic and simultaneous use of at least five different medications. Multilevel logistic regression models were constructed to quantify the polypharmacy prevalence variation between practices. Patient characteristics (age, gender, socioeconomic status, number, and type of chronic conditions) and practice characteristics (practice location and practice population) were added to the models.

RESULTS

After accounting for differences in patient and practice characteristics, polypharmacy rates varied with a factor of 2.4 between practices (from 12.4% to 30.1%) and an overall mean of 19.8%. Age and type of conditions were highly positively associated with polypharmacy, and to a lesser extent a lower socioeconomic status.

CONCLUSIONS

Considerable variation in polypharmacy rates existed between general practices, even after accounting for patient and practice characteristics, which suggests that there is not much agreement concerning medication management in this complex patient group. Initiatives that could reduce inappropriate heterogeneity in medication management can add value to the care delivered to these patients. Copyright © 2016 John Wiley & Sons, Ltd.

Annual consultation incidence of osteoarthritis estimated from population-based health care data in England

OBJECTIVES

To estimate the consultation incidence of OA using population-based health care data in England and compare OA incidence figures with those derived in other countries.

METHODS

A population-based health care database (Consultations in Primary Care Archive) in England was used to derive the consultation incidence of OA (overall and by joint site) using the maximum available run-in period method. These estimates, and their distribution by age and sex, were compared with those published from population-based health care databases in Canada, the Netherlands and Spain. A novel age-stratified run-in period method was then used to investigate whether the consultation incidence has been increasing over time in younger adults.

RESULTS

The annual consultation incidence of OA (any joint) was 8.6/1000 persons ≥15 years of age (95% CI 7.9, 9.3) [6.3 (95% CI 5.5, 7.1) in men and 10.8 (95% CI 9.8, 12.0) in women]. Incidence increased sharply between 45 and 64 years of age, peaking at 75-84 years. The joint-specific incidence was 1.4 (95% CI 1.1, 1.7), 3.5 (95% CI 3.1, 3.9) and 1.3 (95% CI 1.1, 1.6) for hip OA, knee OA and hand OA, respectively. The estimates and their distribution by age and sex were broadly consistent with international estimates. Between 2003 and 2010, incidence in those aged 35-44 years increased from 0.3 to 2.0/1000 persons.

CONCLUSION

Newly diagnosed cases of OA in England occur in 9 in 1000 at-risk adults each year, similar to other international estimates. Although lower, the consultation incidence proportion in younger adults appears to have increased in the past decade.

Why do they keep coming back? Psychosocial etiology of persistence of frequent attendance in primary care: a prospective cohort study

BACKGROUND

Patients who visit their General Practitioner (GP) very frequently over extended periods of time often have multimorbidity and are costly in primary and specialist healthcare. We investigated the impact of patient-level psychosocial and GP-level factors on the persistence of frequent attendance (FA) in primary care.

METHODS

Two-year prospective cohort study in 623 incident adult frequent attenders (>90th attendance centile; age and sex-adjusted) in 2009. Information was collected through questionnaires (patients, GPs) and GPs' patient data. We used multilevel, ordinal logistic regression analysis, controlling for somatic illness and demographic factors with FA in 2010 and/or 2011 as the outcome.

RESULTS

Other anxiety (odds ratio (OR) 2.00; 95% confidence interval from 1.29 to 3.10) over 3years and the number of life events in 3years (OR 1.06; 1.01-1.10 per event; range of 0 to 12) and, at baseline, panic disorder (OR 5.40; 1.67-17.48), other anxiety (OR 2.78; 1.04-7.46), illness behavior (OR 1.13; 1.05-1.20 per point; 28-point scale) and lack of mastery (OR 1.08; 1.01-1.15 per point; 28-point scale) were associated with persistence of FA. We found no evidence of synergistic effects of somatic, psychological and social problems. We found no strong evidence of effects of GP characteristics.

CONCLUSION

Panic disorder, other anxiety, negative life events, illness behavior and lack of mastery are independently associated with persistence of frequent attendance. Effective intervention at these factors, apart from their intrinsic benefits to these patients, may reduce attendance rates, and healthcare expenditures in primary and specialist care.

Do practice characteristics explain differences in morbidity estimates between electronic health record based general practice registration networks?

Background

General practice based registration networks (GPRNs) provide information on population health derived from electronic health records (EHR). Morbidity estimates from different GPRNs reveal considerable, unexplained differences. Previous research showed that population characteristics could not explain this variation. In this study we investigate the influence of practice characteristics on the variation in incidence and prevalence figures between general practices and between GPRNs.

Methods

We analyzed the influence of eight practice characteristics, such as type of practice, percentage female general practitioners, and employment of a practice nurse, on the variation in morbidity estimates of twelve diseases between six Dutch GPRNs. We used multilevel logistic regression analysis and expressed the variation between practices and GPRNs in median odds ratios (MOR). Furthermore, we analyzed the influence of type of EHR software package and province within one large national GPRN.

Results

Hardly any practice characteristic showed an effect on morbidity estimates. Adjusting for the practice characteristics did also not alter the variation between practices or between GPRNs, as MORs remained stable. The EHR software package ‘Medicom’ and the province ‘Groningen’ showed significant effects on the prevalence figures of several diseases, but this hardly diminished the variation between practices.

Conclusion

Practice characteristics do not explain the differences in morbidity estimates between GPRNs.

Electronic supplementary material

The online version of this article (doi:10.1186/s12875-014-0176-7) contains supplementary material, which is available to authorized users.

Similar multimorbidity patterns in primary care patients from two European regions: results of a factor analysis

OBJECTIVE

To compare the similarities among the multimorbidity patterns identified in primary care patients from two European regions (Spain and the Netherlands) with similar organisational features of their primary care systems, using validated methodologies.

METHODOLOGY

This observational, retrospective, multicentre study analysed information from primary care electronic medical records. Multimorbidity patterns were assessed using exploratory factor analysis of the diagnostic information of patients over 14 years of age. The analysis was stratified by age groups and sex.

RESULTS

The analysis of Dutch data revealed a higher prevalence of multimorbidity which corresponds with the clustering of a higher number of diseases in each of the patterns. Relevant clinical similarities were found between both countries for three multimorbidity patterns that were previously identified in the original Spanish study: cardiometabolic, mechanical and psychiatric-substance abuse. In addition, the clinical evolution towards complexity of the cardiometabolic pattern with advancing age--already demonstrated in the original study--was corroborated in the Dutch context. A clear association between mechanical and psychosocial disorders was unique to the Dutch population, as well as the recurrent presentation of the psychiatric-substance abuse pattern in all age and sex groups.

CONCLUSIONS

The similarities found for the cardiometabolic, mechanical and psychiatric-substance abuse patterns in primary care patients from two different European countries could offer initial clues for the elaboration of clinical practice guidelines, if further evidenced in other contexts. This study also endorses the use of primary care electronic medical records for the epidemiologic characterization of multimorbidity.

Morbidity and doctor characteristics only partly explain the substantial healthcare expenditures of frequent attenders: a record linkage study between patient data and reimbursements data

Background

Frequently attending patients to primary care (FA) are likely to cost more in primary care than their non-frequently attending counterparts. But how much is spent on specialist care of FAs? We describe the healthcare expenditures of frequently attending patients during 1, 2 or 3 years and test the hypothesis that additional costs can be explained by FAs’ combined morbidity and primary care physicians’ characteristics.

Methods

Record linkage study. Pseudonymised clinical data from the medical records of 16 531 patients from 39 general practices were linked to healthcare insurer’s reimbursements data. Main outcome measures were all reimbursed primary and specialist healthcare costs between 2007 and 2009. Multilevel linear regression analysis was used to quantify the effects of the different durations of frequent attendance on three-year total healthcare expenditures in primary and specialist care, while adjusting for age, sex, morbidities and for primary care physicians characteristics. Primary care physicians’ characteristics were collected through administrative data and a questionnaire.

Results

Unadjusted mean 3-year expenditures were 5044 and 15 824 Euros for non-FAs and three-year-FAs, respectively. After adjustment for all other included confounders, costs both in primary and specialist care remained substantially higher and increased with longer duration of frequent attendance. As compared to non-FAs, adjusted mean expenditures were 1723 and 5293 Euros higher for one-year and three-year FAs, respectively.

Conclusions

FAs of primary care give rise to substantial costs not only in primary, but also in specialist care that cannot be explained by their multimorbidity. Primary care physicians’ working styles appear not to explain these excess costs. The mechanisms behind this excess expenditure remain to be elucidated.

Predictability of persistent frequent attendance in primary care: a temporal and geographical validation study

BACKGROUND

Frequent attenders are patients who visit their general practitioner exceptionally frequently. Frequent attendance is usually transitory, but some frequent attenders become persistent. Clinically, prediction of persistent frequent attendance is useful to target treatment at underlying diseases or problems. Scientifically it is useful for the selection of high-risk populations for trials. We previously developed a model to predict which frequent attenders become persistent.

AIM

To validate an existing prediction model for persistent frequent attendance that uses information solely from General Practitioners' electronic medical records.

METHODS

We applied the existing model (N = 3,045, 2003-2005) to a later time frame (2009-2011) in the original derivation network (N = 4,032, temporal validation) and to patients of another network (SMILE; 2007-2009, N = 5,462, temporal and geographical validation). Model improvement was studied by adding three new predictors (presence of medically unexplained problems, prescriptions of psychoactive drugs and antibiotics). Finally, we derived a model on the three data sets combined (N = 12,539). We expressed discrimination using histograms of the predicted values and the concordance-statistic (c-statistic) and calibration using the calibration slope (1 = ideal) and Hosmer-Lemeshow tests.

RESULTS

The existing model (c-statistic 0.67) discriminated moderately with predicted values between 7.5 and 50 percent and c-statistics of 0.62 and 0.63, for validation in the original network and SMILE network, respectively. Calibration (0.99 originally) was better in SMILE than in the original network (slopes 0.84 and 0.65, respectively). Adding information on the three new predictors did not importantly improve the model (c-statistics 0.64 and 0.63, respectively). Performance of the model based on the combined data was similar (c-statistic 0.65).

CONCLUSION

This external validation study showed that persistent frequent attenders can be prospectively identified moderately well using data solely from patients' electronic medical records.

International comparisons of the consultation prevalence of musculoskeletal conditions using population-based healthcare data from England and Sweden

Objectives

To assess the consultation prevalence of musculoskeletal (MSK) conditions as presented in different healthcare systems, and to determine the feasibility of comparing prevalence figures between nations.

Methods

The settings were an English regional database (Consultations in Primary Care Archive (CiPCA)) and the Swedish Skåne County Health Care Register. Case definitions, data extraction and analysis procedures were harmonised. The number of people consulting per 10 000 registered population in primary care, and in primary or secondary care, in the year 2010 (annual consultation prevalence) were determined for doctor-diagnosed osteoarthritis (OA), rheumatoid arthritis (RA), low back pain, and spondyloarthritis including psoriatic arthritis and ankylosing spondylitis (AS). Seven-year period consultation prevalences were also determined.

Results

Combining primary and secondary care, annual consultation prevalences of any MSK condition (2143 vs 1610/10 000) and low back pain (587 vs 294/10 000) were higher in England than in Sweden, but higher for RA, spondyloarthritis and psoriatic arthritis in Sweden. Annual primary care prevalence figures for OA (176 vs 196/10 000), RA (25 vs 26/10 000), spondyloarthritis (both 8/10 000) and psoriatic arthritis (5 vs 3/10 000) were similar between England and Sweden. AS was rarely recorded in Swedish primary care. These patterns were also observed for 7-year period consultation prevalences.

Conclusions

A rigorous methodological approach allowed feasible comparison of MSK consultation prevalence between England and Sweden. Differences in prevalence of inflammatory and unspecific pain conditions may be partially explained by known variations in healthcare systems and recording practice. Routine healthcare data offers potential for investigating variations in occurrence and outcome of MSK conditions between nations.