Routine Health Datasets Research Articles

Maternal pre-pregnancy body mass index and risk of preterm birth: a collaboration using large routine health datasets

BackgroundPreterm birth (PTB) is a leading cause of child morbidity and mortality. Evidence suggests an increased risk with both maternal underweight and obesity, with some studies suggesting underweight might be a greater factor in spontaneous PTB (SPTB) and that the relationship might vary by parity. Previous studies have largely explored established body mass index (BMI) categories. Our aim was to compare associations of maternal pre-pregnancy BMI with any PTB, SPTB and medically indicated PTB (MPTB) among nulliparous and parous women across populations with differing characteristics, and to identify the optimal BMI with lowest risk for these outcomes.MethodsWe used three UK datasets, two USA datasets and one each from South Australia, Norway and Denmark, together including just under 29 million pregnancies resulting in a live birth or stillbirth after 24 completed weeks gestation. Fractional polynomial multivariable logistic regression was used to examine the relationship of maternal BMI with any PTB, SPTB and MPTB, among nulliparous and parous women separately. The results were combined using a random effects meta-analysis. The estimated BMI at which risk was lowest was calculated via differentiation and a 95% confidence interval (CI) obtained using bootstrapping.ResultsWe found non-linear associations between BMI and all three outcomes, across all datasets. The adjusted risk of any PTB and MPTB was elevated at both low and high BMIs, whereas the risk of SPTB was increased at lower levels of BMI but remained low or increased only slightly with higher BMI. In the meta-analysed data, the lowest risk of any PTB was at a BMI of 22.5 kg/m2 (95% CI 21.5, 23.5) among nulliparous women and 25.9 kg/m2 (95% CI 24.1, 31.7) among multiparous women, with values of 20.4 kg/m2 (20.0, 21.1) and 22.2 kg/m2 (21.1, 24.3), respectively, for MPTB; for SPTB, the risk remained roughly largely constant above a BMI of around 25–30 kg/m2 regardless of parity.ConclusionsConsistency of findings across different populations, despite differences between them in terms of the time period covered, the BMI distribution, missing data and control for key confounders, suggests that severe under- and overweight may play a role in PTB risk.

Can routine data be used to estimate the mental health service use of children and young people living on Gypsy and Traveller sites in Wales? A feasibility study.

Gypsies and Travellers have poorer physical and mental health than the general population, but little is known about mental health service use by Gypsy and Traveller children and young people. Finding this group in routine electronic health data is challenging, due to limited recording of ethnicity. We assessed the feasibility of using geographical markers combined with linked routine datasets to estimate the mental health service use of children and young people living on Traveller sites. Welsh Government supplied a list of Traveller site postcodes included in Caravan Counts between 2012 and 2020. Using spatial filtering with data from the Adolescent Mental Health Data Platform (ADP) at Swansea University's SAIL Databank, we created a cohort of Traveller site residents aged 11-25 years old, 2010-2019. ADP algorithms were used to describe health service use, and to estimate incidence and prevalence of common mental disorders (CMD) and self-harm. Our study found a subgroup of young Gypsies and Travellers (n = 802). We found no significant differences between our cohort and the general population for rates of CMD or self-harm. The rate of non-attendance for psychiatric outpatient follow-up appointments was significantly higher in our cohort. Rates were higher (but not statistically significant) among Gypsies and Travellers for measures suggesting less well-managed care, including emergency department attendance and prescribed CMD medication without follow-up. The small size of the cohort resulted in imprecise estimates with wide confidence intervals, compared with those for the general population. Gypsies and Travellers are under-represented in routine health datasets, even using geographical markers, which find only those resident in authorised traveller sites. Routine data is increasingly relied upon for needs assessment and service planning, which has policy and practice implications for this underserved group. To address health inequalities effort is required to ensure that health datasets accurately capture ethnicity.

Use of Routine Health Datasets to Assess the Appropriateness of Diagnostic Tests in the Follow-Up of Breast Cancer Patients: A Population-Based Study on 3930 Patients.

PurposeClinical practice guidelines (CPGs) recommend against intensive follow-up in asymptomatic women with breast cancer (BC). The present study assessed the adherence to CPGs of diagnostic tests ordering during BC follow-up by exploring routinely collected health data through an algorithm developed to distinguish patients according to their status at follow-up.Patients and MethodsA retrospective population-based cohort study was performed monitoring the diagnostic tests ordered during 5 years of follow-up in all BC cases incident in 2013 in the Veneto Region, Italy. Data were extracted from the Veneto Tumour Registry, the Hospital Discharge Records and the Outpatients’ Records of Diagnostic and Therapeutic Procedures. The algorithm was developed using information on infusion of anticancer agents, imaging exams ordered, and death.ResultsThe algorithm classified patients by status at follow-up in four groups: (i) probably no-evidence-of-disease (NED), (ii) suspicious signs of relapse not confirmed, (iii) increased risk of relapse and (iv) advanced disease at presentation or progressive disease. A total of 3930 consecutive incident cases were followed-up for 5 years, corresponding to 17,184 person-years, 15,345 of which pertaining to NED cases. In NED cases, 32,900 tumour markers and 15,858 imaging exams were ordered. Liver ultrasonography and chest radiography were most frequently ordered.ConclusionIn contrast with recommendations of CPGs, a substantial overordering of tumour markers and imaging exams occurred in NED BC patients. The developed algorithm can be repeatedly applied to routine health datasets for regular monitoring of the adherence to CPGs and of the impact of interventions to improve appropriateness.

Using linked English cancer registration data to assess variation in diagnostic pathway length

BackgroundFor the majority of cancer patients, the length of the diagnostic pathway is unknown. Only those on a specific pathway are monitored. Quantifying pathway length is important in understating which patients may benefit most from an expedited pathway.
 AimThis study aimed to develop a novel methodology using linked cancer registration data to calculate the Secondary Care Diagnostic Interval (SDCI): the period from first interaction with secondary care to receiving a cancer diagnosis. Subsequently variation in SCDIs across twenty-five cancer sites were analysed.
 MethodsEnglish cancer registrations (2014-15) from Public Health England’s National Cancer Registration and Analysis Service were linked to three routine health datasets: Hospital Episode Statistics, Diagnostic Imaging Dataset and Cancer Waiting Times.
 The SCDI was calculated as the difference between the earliest event relating to the secondary care diagnostic process (comprised of referral into/secondary care appointment or diagnostic procedure in the 6 months before diagnosis) and the diagnosis date.
 SCDI length was stratified by stage at diagnosis, route to diagnosis (e.g. emergency presentation, GP referral) and patient characteristics.
 ResultsThe median SCDI (days) was shortest for acute lymphoblastic leukaemia (ALL) (2) and longest for kidney (45). In general, SCDIs decreased with later stage (for example stage 1-4 colorectal: 35 to 20, lung: 75 to 25). Patients diagnosed after routine GP referrals had longest SCDIs compared to urgent referrals, for all sites except ALL. Geographical variation in SCDIs exists by Cancer Alliance, with the greatest differences compared to the England average observed for prostate cancer (median national SCDI: 28, Cancer Alliance range: 21 - 42).
 ConclusionSubstantial variation exists in SCDIs by cancer site, stage and presentation route. Many patients are experiencing SCDIs over 28 days, even those diagnosed via urgent GP referrals. Understanding variation can support creation of targeted initiatives to expedite diagnostic pathways where appropriate.

Regional multidisciplinary team intervention programme to improve colorectal cancer outcomes: study protocol for the Yorkshire Cancer Research Bowel Cancer Improvement Programme (YCR BCIP)

IntroductionAlthough colorectal cancer outcomes in England are improving, they remain poorer than many comparable countries. Yorkshire Cancer Research has, therefore, established a Bowel Cancer Improvement Programme (YCR BCIP) to improve...

PCN254 CHARACTERISING INDIVIDUALS WITH NON-METASTATIC CASTRATION-RESISTANT PROSTATE CANCER USING ROUTINE LINKED HEALTH DATASETS: A FEASIBILITY STUDY

Non-metastatic castration-resistant prostate cancer (nmCRPC) is a disease state in which cancer is localized but resistant to testosterone suppression therapy (Androgen Deprivation Therapy (ADT) or bilateral orchidectomy). There is relatively little real world data focusing on this disease state either for characterising patients or describing their treatment pathways. This study set out to assess the feasibility of using linked routine health datasets to identify individuals with nmCRPC and to estimate their nmCRPC progression date. The study datasets were routine health data for Wales hosted by the SAIL Databank, Swansea University, Wales. Records of patients diagnosed with prostate cancer between 2000 and 2015 were identified by clinical coding (ICD-10 and Read). An algorithm was designed to specify the features of nmCRPC based on presence of codes for testosterone suppression therapy, Prostate Specific Antigen (PSA) test results and codes indicating absence of metastatic disease. Only records that contained sufficient PSA test results to determine a rise during treatment (indicating progression to castration resistance) were included. The algorithm was tested with three different criteria for PSA test levels and intervals, and two thresholds of stringency for determining cases. The datasets produced a cohort of 38,021 individuals diagnosed with prostate cancer. Of these 14,860 received ADT while non-metastatic and 6,101 had at least three recorded PSA test results during ADT. The number of nmCRPC cases identified was between 1,281 and 1,534 (non-stringent criteria) and between 439 and 509 (stringent criteria). Case identification is dependent on accurate, complete coding. The results of this study suggest that it is feasible to use linked routine health datasets to create cohorts of individuals with nmCRPC, including an estimation of date of progression to nmCRPC. Various processes were investigated to improve data quality, particularly in the identification of metastases, resulting in cohorts of sufficient size to enable statistical analysis.

Incidence of Medication-Treated Depression and Anxiety Associated with Long-Term Cancer, Cardiovascular Disease, Diabetes and Osteoarthritis in Community-Dwelling Women and Men

Background: Physical chronic conditions may increase the risk of mental disorders but the findings are inconsistent between studies and whether the associations differed between genders is unclear. This study aimed to examine whether long-term cancer, cardiovascular disease, diabetes and osteoarthritis were associated with higher risks of incident depression and anxiety and these associations in women compared with men. Methods: We included 115,094 participants (54.3% women) aged 45-64 years from the 45 and Up Study who were free of depression, anxiety, and Parkinson's disease at baseline (2006-2009). The incidence of depression and anxiety was identified using claim databases during follow-up until December 2016. Cox regression models were used to examine the association of cancer, cardiovascular disease, diabetes, and osteoarthritis at baseline with incident depression and anxiety. Findings: During a mean eight-year follow-up (958,785 person-year), the cumulative incidence of depression and anxiety was 12·5% and 5·9% in the healthy population. The hazard ratios ([HRs] (95% CI) VS healthy population) for incident depression associated with long-term cancer, cardiovascular disease, diabetes, and osteoarthritis were 1·19 (95% CI: 1·13-1·25), 1·08 (1·00-1·16)), 1·18 (1·09-1·28), and 1·94 (1·80-2·10), respectively. The corresponding HRs (95% CIs) for incident anxiety were 1·11 (1·03-1·20), 1·26 (1·14-1·39), 1·10 (0·98-1·24), and 2·01 (1·80-2·23), respectively. The positive association between cancer and incident depression was more evident in men (HR (95% CI): 1·24 (1·13-1·35) than in women (1·14 (1·07- 1·21). Long-term diabetes was an independent risk factor for incident anxiety in men (1·21 (1·02-1·44) but not in women (1·09 (0·93-1·28)). Other associations were significant in men and women. Interpretation: Long-term osteoarthritis, cardiovascular disease, and cancer were independent risk factors for incident depression and anxiety in both genders with osteoarthritis having the highest relative risk. Depression screen deserves more concern in men with long-term cancer and anxiety screen in men with long-term diabetes. Funding Statement: Professor Mingguang He receives support from the University of Melbourne at Research Accelerator Program and the CERA Foundation. The Centre for Eye Research Australia (CERA) receives Operational Infrastructure Support from the Victorian State Government. The specific project is funded by Australia China Research Accelerator Program at CERA. Professor Mingguang He is also supported by the Fundamental Research Funds of the State Key Laboratory in Ophthalmology, National Natural Science Foundation of China (81420108008). The sponsor or funding organization had no role in the design or conduct of this research. Declaration of Interests: The authors declare no competing interests. Ethics Approval Statement: The study protocol was approved by the Royal Victorian Eye and Ear Hospital Human Research Ethics Committee. Participants provided consent to follow-up and link their data to routine health datasets.

Leading Determinants of Healthy Ageing in Community-Dwelling Middle-Aged Men and Women: A Cohort Study

Background: Identifying leading determinants of healthy ageing may provide evidence for action priorities, which is imperative for public health with an expanding aged population worldwide. Methods: We included 52,036 participants aged 45-64 years from the 45 and Up Study who were free of 13 predefined chronic conditions at baseline (2006-2009). Healthy ageing was defined as the absence of developing any of these chronic conditions during follow-up until 2016. We used machine learning methods to evaluate the importance of potential predictors and analyzed the association between the number of leading modifiable healthy factors and healthy ageing. Findings: Healthy ageing was found in about half of both men and women during a mean nine-year follow-up. The five most common leading predictors were body mass index (6·4-9·5% of total variance), self-rated health (5·2-8·2%), self-rated quality of life (4·1-6·8%), red meat intake (4·5-6·5%), and chicken intake (4·5-5·9%) in both genders. Modifiable behavioral factors including body mass index, diets, smoking, alcohol consumption and physical activity, contributed to 37·2-40·3% of total variance. Participants having six or more modifiable healthy factors were 1·63-8·76 times more likely to remain healthy ageing and had 0·60-2·49 more healthy ageing years (out of nine-year follow-up) than those having two or fewer. Non-behavioral factors including low levels of education, income and relative socioeconomic disadvantage, were leading risk factors for healthy ageing. Interpretation: Body mass index, diets, smoking, alcohol consumption and physical activity are key factors for healthy ageing promotion. Individuals with low socioeconomic status are more in need of care. Funding: Professor Mingguang He receives support from the University of Melbourne at Research Accelerator Program and the Centre for Eye Research Australia Foundation. The Centre for Eye Research Australia receives Operational Infrastructure Support from the Victorian State Government. The specific project is funded by Australia China Research Accelerator Program at Centre for Eye Research Australia. Professor Mingguang He is also supported by the Fundamental Research Funds of the State Key Laboratory in Ophthalmology, National Natural Science Foundation of China (81420108008). The sponsor or funding organization had no role in the design or conduct of this research. Declaration of Interest: We declare no competing interests. Ethical Approval: The 45 and Up study has ethical approval from the UNSW Human Research Ethics Committee. The study protocol was approved by the Royal Victorian Eye and Ear Hospital Human Research Ethics Committee. Participants provided consent to follow-up and link their data to routine health datasets.

Long term cardiac outcomes and anthracycline exposure in breast cancer survivors: A retrospective cohort data linkage and competing risks analysis.

553 Background: Anthracycline chemotherapy is used frequently in adjuvant breast cancer treatment but there is clinical trial evidence of cardiac toxicity. Attempts to quantify this risk in routine care have been limited by follow up time & confounding factors. The aim of this project was to exploit rich Scottish NHS datasets for this purpose. Methods: Patients treated surgically for stage I-III invasive breast cancer between 2000 & 2010 were identified from a local cancer database (Edinburgh Cancer Centre). Outcomes were captured by linkage to the Scottish Morbidity & death Records. Follow-up was until March 2016. The primary outcome was a cardiac event or cardiac death, identified from coding. Analysis used the Latouche approach (estimating cause-specific hazards & sub-distribution hazards) for the primary outcome & the competing risks of death from breast cancer & death from other causes. Results were adjusted for age, deprivation (SIMD), co-morbidity (Charlson), year of diagnosis, side of radiotherapy, cancer stage, grade, ER & HER 2 status. Results: 4080 patients were identified, 1658 received an anthracycline containing regime, 297 received non-anthracycline chemotherapy & 2125 received no chemotherapy. A total of 33946 women years were analysed. During median follow up of 8.2 years there were 448 cardiac events & 559 breast cancer deaths. After adjustments there was no association between anthracycline use & cardiac outcomes (HR 0.9, 95% CI 0.67-1.21). There was an increased risk of breast cancer death (HR 1.66, 95% CI 1.28-2.16).Age & Charlson score were associated with an increased cardiac risk. Stage & grade were statistically associated with breast cancer death. Conclusions: No increased risk of cardiac events was seen in women treated with anthracyclines, they can be safely used in carefully selected patients. This data suggests selection may be over-cautious, a lower threshold for treatment may lead to improved breast cancer outcomes. Rich routine health datasets & appropriate analysis methods make outcomes monitoring feasible in Scotland. Oncologists are skilled at assessing both cardiac risk factors & breast cancer risk.

The future of population registers: linking routine health datasets to assess a population's current glycaemic status for quality improvement

ObjectivesTo determine the diabetes screening levels and known glycaemic status of all individuals by age, gender and ethnicity within a defined geographic location in a timely and consistent way to...

Use of country of birth as an indicator of refugee background in health datasets.

BackgroundRoutine public health databases contain a wealth of data useful for research among vulnerable or isolated groups, who may be under-represented in traditional medical research. Identifying specific vulnerable populations, such as resettled refugees, can be particularly challenging; often country of birth is the sole indicator of whether an individual has a refugee background. The objective of this article was to review strengths and weaknesses of different methodological approaches to identifying resettled refugees and comparison groups from routine health datasets and to propose the application of additional methodological rigour in future research.DiscussionMethodological approaches to selecting refugee and comparison groups from existing routine health datasets vary widely and are often explained in insufficient detail. Linked data systems or datasets from specialized refugee health services can accurately select resettled refugee and asylum seeker groups but have limited availability and can be selective. In contrast, country of birth is commonly collected in routine health datasets but a robust method for selecting humanitarian source countries based solely on this information is required. The authors recommend use of national immigration data to objectively identify countries of birth with high proportions of humanitarian entrants, matched by time period to the study dataset. When available, additional migration indicators may help to better understand migration as a health determinant. Methodologically, if multiple countries of birth are combined, the proportion of the sample represented by each country of birth should be included, with sub-analysis of individual countries of birth potentially providing further insights, if population size allows. United Nations-defined world regions provide an objective framework for combining countries of birth when necessary. A comparison group of economic migrants from the same world region may be appropriate if the resettlement country is particularly diverse ethnically or the refugee group differs in many ways to those born in the resettlement country.SummaryRoutine health datasets are valuable resources for public health research; however rigorous methods for using country of birth to identify resettled refugees would optimize usefulness of these resources.

The need for a revised staging system of metastatic (M) colorectal cancer (CRC): Evidence from a national perspective on survival following surgically treated (HPX) liver metastases

4099 Background: AJCC V.6 (2002) places all patients with MCRC beyond the lymph node basin of the primary tumor in a homogenous Stage 4. Patients with inoperable hepatic MCRC can be made operable with curative intent with chemotherapy yet remaining in Stage 4. Via the linkage of routine health datasets across England this population-based study sought to determine the impact of HPX for MCRC on stage-matched survival at initial presentation. Methods: All patients between 1998–2001 undergoing surgery for CRC in England were identified via the national-linked cancer registry HES dataset. All care episodes in the 3 years following initial colorectal surgery were examined to determine the frequency of subsequent HPX. Kaplan-Meier curves and log- rank tests were used to examine 5-year survival following HPX for MCRC compared to all Stage 3 and Stage 4 at presentation. Survival was calculated from the date of resection of each patient's primary colorectal tumor. Results: 68,307 individuals were identified as undergoing surgery for primary CRC over the study period. 20,298 were Stage 3 at presentation. 1,483 (2.2%) subsequently underwent HPX <3 years of their colorectal operation. 55 patients died within 30 days of HPX (mortality rate: 3.7%). Crude 5-year survival of patients who underwent HPX was 41.6% (95%CI 39.0–44.1%) from time of initial colectomy. This survival rate was significantly better than that for both Stage 3 (38.6% (95%CI 37.9%-39.2%, P<0.01) and 4 (6.1% 95%CI 5.3–6.9%, P<0.01) overall. Conclusions: 5-year survival following HPX for MCRC is better than that seen overall for all Stage 3 patients (with MCRC confined to the regional draining lymph node basin) following initial colectomy. Our data support the hypothesis that all MCRC that is potentially resectable with curative intent should be stratified within Stage 3, and Stage 4 should only contain those MCRC patients for whom surgery is not an option. If further evidence emerges to support this theory then a revision of the current staging system will be required. [Table: see text]

The Extent of Collection of Information on Migrant and Asylum Seeker Status in Routine Health and Social Care Data Sources in England

While efforts to mainstream collection of ethnicity in routine health datasets have gathered pace since the Race Relations (Amendment) Act 2000, relatively little attention has focused on the capture of migrants, and data on asylum seekers and refugees are even more sparse. There is negligible coverage in the key datasets for primary and secondary care, and only a few of the new contract datasets to support the National Service Frameworks accord importance to the migrant population. Some of the communicable disease data collections record country of birth, but its incompleteness is a drawback. Given the growing size of the non‐UK‐born population and the accumulating evidence on the health and health care needs of the migrant population, country of birth merits a place alongside other access variables such as age, sex, ethnicity, socioeconomic status and place of residence in routine data collection. While record linkage studies offer some scope, and there is currently a focus on obtaining improved migration statistics in the forthcoming census and other data flows, health and social care sources remain a neglected area.