Temporal trends and practice variation of paediatric diagnostic tests in primary care

Introduction

It has been reported that 70% of clinical decisions involve the use of diagnostic tests.[1] Paediatric test use has not previously been characterised in large-scale population studies, especially in primary care, where most paediatric health contacts occur. Compared with adults, children more commonly present with undifferentiated symptoms, and non-specific complaints such as abdominal pain, headaches, and fatigue often have no identifiable underlying cause.[2] While tests are one of the many diagnostic strategies available to clinicians, they must balance the risks of over-investigation and unnecessary referrals with missing or delaying a diagnosis. It is difficult to achieve the right balance of care, and the threshold to test, treat or refer varies among clinicians [3], which can lead to substantial variation in the health care delivered to children.[4, 5]

Measuring variation in testing can identify tests that are potentially overused or underused, with both over and under use having potentially harmful consequences.[6] For the patient, overuse can result in testing cascades, which cause harm through unnecessary treatment, physical, and emotional trauma. [7] For the health care system, overuse generates additional workload for clinicians and can lead to unnecessary referrals, healthcare contacts, and health spending in an already overburdened system.[8, 9] Underuse of tests can result in missed or delayed diagnosis and treatment with potentially serious physical, emotional, and financial consequences for patients, families, and clinicians.

Studies analysing test use in adults have reported that test use in primary care increased by 8.5% annually[10, 11]. We previously published a study that described temporal trends in paediatric blood tests in Oxfordshire from 2005 to 2019 and found that test use increased in outpatient (specialty and general practice) settings compared with inpatient services where test rates remained stable. [12] The overall aim of this study was therefore to quantify and analyse temporal change and variation in paediatric diagnostic tests in primary care across England. The specific objectives were to 1) quantify how paediatric diagnostic test use has changed over time and varied by primary care practice and 2) determine how demographic and socioeconomic factors impact test use.

Methods

Results

Temporal trends in overall test use

The age-adjusted rate of total test use increased from 399 tests per 1,000 child years in 2007 to 608 tests per 1,000 child years in 2019, an average annual percentage increase of 3.6% per year (AAPC 95% CI 3.4 to 3.8%), see Figure 1a, Appendix Table 1. Test rates initially increased by 5.1% per year (APC 95% CI 4.7 to 5.6%) between 2007 to 2014, then increased by 1.6% per year (APC 95% CI 0.9 to 2.1%) between 2014 and 2019. Figure 1b shows temporal changes in test use stratified by test type. The greatest increase was observed for blood tests which increased by 8.0% per year (AAPC 95% 7.7 to 8.4%). Rates of test use by gender and age group are shown in Figure 1c and 1d. Test rates were consistently higher in females compared with males. When stratified by age group, the rates of change were similar for both genders and age groups, except for 11-15-year-olds, where testing increased by 4.0% per year (AAPC 95% CI 3.7 to 4.3%) among females and slightly less, 3.4% per year among males (AAPC 95% CI 3.0 to 3.9%, p=0.02 for the difference between groups). Figure 1e demonstrates test use by IMD deprivation quintile. Test rates were highest in the most deprived cohort (quintile 5) and increased the most, with an AAPC of 4.4% (95% CI 4.1 to 4.8%), compared with those from the lower quintiles of deprivation (Appendix Table 1).

• Download figure
• Open in new tab

Figure 1. Temporal trends in paediatric test use in English primary care practices from 2007 to 2019

Temporal trends in specific tests

The average annual percentage change (AAPC) for each test is presented in Figure 2. The three greatest increases were for tests with zero or negligible use at the beginning of the study period. Fecal calprotectin testing was subject to the greatest average annual change, increasing from 0 tests/1,000 child-years in 2007 to 1.8 tests/1,000 child-years by 2019, equivalent to 105.5% per year (AAPC 95% CI 97.5 to 122.2%). This was followed by fractional exhaled nitric oxide (FeNO) tests which increased by 40.3% per year (AAPC 95% CI 26.7 to 64.7%) from 0 tests/1,000 child-years in 2007 to 0.2 tests/1,000 child-years in 2019, then vitamin D tests which increased by 27.0% per year (AAPC 95% CI 25.5 to 30.4%) from 0.5 tests/1,000 child-years in 2007 to 8.5 tests/1,000 child-years in 2019. The following tests increased by greater than 10% per year, in descending order: folate, vitamin B12, coeliac testing, helicobacter testing, iron studies, HbA1c, immunoglobulins, C reactive protein, MRI brain, bone profile, and allergen-specific IgE (see Appendix Table 2). Tests that decreased in use included urine microscopy/culture/sensitivities, hearing tests, spirometry, CT head, peak flow measurements, renal ultrasound, and monospot testing for glandular fever. The changes were largely consistent by gender, age group (see Appendix Figure 1), and deprivation quintile (see Appendix Figure 2).

• Download figure
• Open in new tab

Figure 2. Temporal change in specific tests for children aged 0 to 15 years in English primary care from 2007 to 2019

Rank order of practice variability of specific tests

Figure 3 shows the rank order of the tests from highest to lowest practice variability (CoV). FeNO was subject to the greatest practice variability, with an adjusted CoV of 123.7% (95% CI 123.6 to 123.9%). This was followed by hearing tests (CoV 51.6%, 95% CI 51.4% to 51.7%) and Vitamin D tests (CoV 38.1%, 95% CI 38.0 to 38.3%). Tests with higher rates of use (represented by larger bubbles in Figure 3) were, on average, subject to lower practice variability.

• Download figure
• Open in new tab

Figure 3. Rank order of between-practice variability of tests in 2019; adjusted for gender, age, and deprivation

Figure 4 plots the adjusted coefficient of variation of each of the 35 tests against their test rate. The median coefficient of variation of test use was 16.5% (IQR 12.1 to 21.3%) and the median rate of test use was 6.9 tests/1000 child-years (IQR 2.4 to 19.3%). Most tests were either classified as low-test rate-high variability (37%, 13 out of 35) or high test rate-low variability (37%, 13 out of 35), see Appendix Table 3. The following five tests were classified as high test rate – high variability: Iron studies, Coeliac testing, Vitamin B₁₂, Folate, and Vitamin D.

• Download figure
• Open in new tab

Figure 4. Test rate and degree of practice variability for specific tests in 2019

Discussion

To our knowledge, this study represents the first nationwide analysis of temporal trends and practice variation in paediatric test use between 2007 and 2019. We analysed 14 million tests over 27.8 million child-years of observation across nearly 1500 primary care practices and found that test use increased at a rate of 4% per year. Blood tests increased by the highest margin, and females aged 11-15 experienced the greatest increase, as well as children from practices in more deprived areas. Tests with the biggest temporal increases included: fecal calprotectin, FeNO, vitamin D, folate, vitamin B₁₂, coeliac, and helicobacter tests. Tests subject to the largest practice variation included FeNO, hearing tests, vitamin D, helicobacter testing, and monospot testing for glandular fever. We also identified tests with high rates of test use and practice variability: iron studies, vitamin B₁₂, coeliac test, folate, and vitamin D.

The increasing use of both calprotectin and FeNO tests reflect the implementation of these tests in UK primary care during the study period. These two tests serve as interesting case studies: in 2019 there was low practice variability in calprotectin testing but high variability in FeNO testing. Several factors may explain this discrepancy in practice variability in 2019, including the ease with which both test samples are obtained and analysed. Additionally, it could be due to the lack of equipment, time, access, and funding for FeNO testing in primary care despite evidence of their feasibility and acceptability, and national guidance recommending FeNO testing for the diagnostic workup of childhood asthma. [23–25]

The rise in vitamin D tests is consistent with the 42-fold increase observed for children in Minnesota from 2002 to 2017, though the odds of detecting low levels remained small, suggesting that tests were overused.[26] Similar trends were observed in Australia and Canada.[27, 28] There are no specific guidelines for the paediatric population, however, the US Preventive Services Task Force and the American Society for Clinical Pathology guidelines for adults do not recommend screening for vitamin D deficiency [29]. Therefore, increased vitamin D testing is likely to represent overuse in children.

The increases observed in requests for hematinic tests (vitamin B₁₂, folate, iron) were surprising, and indicate growing clinician concern for nutritional deficiencies and anaemia. Similar test increases were observed in adults in UK primary care [11] which could suggest a creep of adult diagnostic practices into paediatric care.[18] Haematinic tests, in addition to testing for coeliac disease, helicobacter pylori, vitamin D, CRP, and HbA1c may also be requested for more non-specific symptoms including fatigue, musculoskeletal issues or abdominal symptoms. Further research is needed to understand the implications of the observed increases in testing and examine their appropriateness. This could be achieved by examining the test results and disease sequelae (similar to the Vitamin D studies described earlier[26, 28]) or determining whether testing indications were concordant with evidence-based guidelines.

Previous research on paediatric diagnostic test use focused on specific tests, were undertaken in smaller geographic areas, or in hospital settings. Quality improvement studies were conducted in paediatric intensive care units to reduce blood cultures in febrile infants; which drive unnecessary antibiotic treatment and hospital-acquired infections.[30] Potassium testing was also examined in a retrospective study which revealed that increased potassium testing did not influence the need for potassium replacement in patients with status asthmaticus on continuous albuterol.[31] Increased use of certain tests can be appropriate in the right population and setting. For example, rapid diagnostic testing for respiratory infections in the emergency setting has been shown to prevent unnecessary antibiotic prescriptions. [32] It is, therefore, prudent to examine all test use to identify potential areas where testing may confer benefit as well as areas where testing causes harms.

Our study results concur with findings of increased blood test utilisation in Oxfordshire primary care practices; particularly for vitamin D, folate, vitamin B₁₂, iron studies, coeliac testing, HbA1c, bone profile, CRP, thyroid function tests, urea and electrolytes, and liver function tests.[12] Notably, test increases were more pronounced in our nationwide study compared with Oxfordshire. This may be the result of local policies discouraging testing in primary care or the referral of children to outpatient services for testing.

We specifically looked at test utilisation in the period preceding the pandemic, however, testing rates likely decreased substantially during the pandemic with decreased paediatric consultations in primary care.[33] Assuming test rates recovered, and test rates continued at the rate of growth since 2014 (APC 1.6%), then by 2024 test rates in primary care would be 656 tests/1,000 children per year. This has considerable cost implications. A Vitamin D test costs approximately £10 at the Oxford University Hospitals NHS Trust Laboratories. Applying the rates of vitamin D requests to the rest of the UK population in 2019, vitamin D tests requested for children in primary care cost £2.2 million in 2019. If rates continued to increase at the AAPC of 29.6% per year, then by 2024, the rate of vitamin D testing would be 42.3 tests/1,000 child-years, costing £5.4 million across UK primary care annually. While these are estimates (and assume testing rates recovered post-pandemic), they provide an indication of the potential scale of the financial consequences if testing rates continued to increase. Additional harms, including physical side effects of tests from radiation, psychological distress and anxiety associated with tests, and the potential consequences that include overdiagnosis, further tests, and unnecessary treatments are also important and merit further research.

This study had several limitations. First, CPRD data relies on the quality of the electronic data input by clinicians using electronic health record software which can be highly variable. Second, double counting could have occurred if both the request and the completed test were coded separately, however, this is not likely to have varied by patient demographic, practice, or calendar time, therefore, the metrics of temporal trends and practice variation would nevertheless be valid. Third, there was some subjectivity in how the test names were coded and grouped. To address this, the code list of tests (and their associated panels) was developed and cross-checked with 1) existing NHS trust laboratory test lists, 2) the test codes from a previous laboratory-based study based in an NHS trust corroborated by a consultant chemical pathologist[12] and 3) another clinician-researcher (CH). Grouped test codes (i.e. “Immunoglobulins (IgA, IgG, IgM)”) may have obscured more subtle trends in contrast with single condition-specific test codes, which can reveal more about the diagnostic strategies for a particular disease. For example, the decline in monospot testing for glandular fever may reflect shifts in clinicians’ preferences towards alternative diagnostic strategies such as serology tests or clinical diagnosis.

Conclusions

This study provides a broader picture of paediatric testing practices based on individual-level data across primary care in England. Increased testing rates can generate more clinical activity including more specialist referrals, and the potential cost implications are substantial. Future research should compare tests against clinical guideline standards and examine the test results to judge whether test increases are warranted and evaluate its downstream impact on patient outcomes and cost.

Study approval

The protocol was approved via the CPRD’s Research Data Governance process (study reference 22_001998) and is available on the Open Science Framework (osf.io/pwgtf).

Data sharing and availability

The study protocol was registered on the Open Science Framework (doi:10.17605/osf.io/gkncj). All test codes, R code used for data management, analysis and creating the figures is available on GitHub (https://github.com/elizabethtthomas/cprd-paediatric-tests).

Author contributions

ETT contributed to study conceptualisation, methodology, completed the data analyses and wrote the original draft. DRW reviewed the statistical aspects of the study and provided critical input on the original manuscript. PG provided supervisory input and comments and feedback on the manuscript. RP provided supervisory input, contributed to conceptualisation, methodology, reviewed the statistical aspects of the study and revision of the manuscript. CH also provided supervisory input, and contributed to conceptualisation, methodology, data interpretation and revision of the manuscript. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Declarations of interests

This project was funded by a grant from the NIHR SPCR grant (Award 624). The funders were not involved in the study design and conduct, data collection and interpretation, or manuscript preparation and approval to submit the manuscript for publication.

ETT was supported by a Clarendon scholarship to study for a Doctor of Philosophy (DPhil) at the University of Oxford (2020-23). PG has received grants from the Canadian Institutes of Health Research (CIHR), the Physicians Services Incorporated Foundation, and The Hospital for Sick Children. He has received nonfinancial support from the EBMLive Steering Committee (expenses reimbursed to attend conferences) and the CIHR Institute of Human Development, Child and Youth Health (as a member of the institute advisory board, expenses reimbursed to attend meetings), is a member of the CMAJ Open and BMJ Evidence Based Medicine Editorial Board. RP is partly supported by the NIHR Applied Research Collaboration (ARC) Oxford & Thames Valley, the NIHR Oxford BRC, the NIHR Oxford MedTech and In-Vitro Diagnostics Co-operative (MIC) and the Oxford Martin School. CJH receives funding support from the NIHR School of Primary Care Research. The funders had no role in study design, manuscript submission, or collection, management, analysis, or interpretation of study data. All other authors have no sources of funding to declare.