Acute admission risk stratification of New Zealand primary care patients using demographic, multimorbidity, service usage and modifiable variables
Chris Van Houtte 1 2 * , Chris Gellen 1 , Dipan Ranchhod 11 Tū Ora Compass Health, Level 4/22–28 Willeston Street, Wellington Central, Wellington 6011, New Zealand.
2 Present address: Plant and Food Research, 23 Batchelar Road, Palmerston, North 4410, New Zealand.
Journal of Primary Health Care 14(2) 116-123 https://doi.org/10.1071/HC21174
Published: 21 June 2022
© 2022 The Author(s) (or their employer(s)). Published by CSIRO Publishing on behalf of The Royal New Zealand College of General Practitioners. This is an open access article distributed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND)
Abstract
Introduction: Risk stratification tools in primary care may help practices better identify high-risk patients and plan for their treatment. Patients of all ages can be at high risk of acute hospital admissions.
Aim: We aim to improve existing risk stratification tools by using larger datasets, and accounting for practice-level variations in hospitalisation rates and read-code quality.
Methods: This work has derived an acute admission risk stratification tool in the Wellington, Kāpiti Coast and Wairarapa regions of New Zealand. An open cohort, starting 1 March 2017 and finishing 1 November 2021, contains 319 943 patients. An accelerated failure time survival regression model is used to model acute admission risk. Candidate models are tested on holdout data using six different test metrics.
Results: Patient risk is most affected by demographic, and the frequency of recent healthcare system usage. Morbidity categories have less predictive capability, but may still be useful from a practical perspective. The preferred model has an area under the receiver operating characteristic curve (AUROC) of 0.75 for a 6-month forecast period.
Discussion: The model is straightforward to apply to other datasets. Although most of the highest-risk patients will be well-known to their primary care practices already, the model helps to identify the patients who are high risk but not regularly attendees of the practice, and may benefit from proactive care planning.
Keywords: algorithmic, care planning, Health Care Home, hospitalisation, morbidity, primary health care, risk assessment, risk stratification.
References
[1] Hefford M. From good to great: the potential of the HCH model to improve primary health care quality in New Zealand. J Prim Health Care 2017; 9 230–233.| From good to great: the potential of the HCH model to improve primary health care quality in New Zealand.Crossref | GoogleScholarGoogle Scholar | 29530177PubMed |
[2] Allaudeen N, Schnipper J, Orav E, et al. Inability of providers to predict unplanned readmissions. J Gen Intern Med 2011; 26 771–776.
| Inability of providers to predict unplanned readmissions.Crossref | GoogleScholarGoogle Scholar | 21399994PubMed |
[3] Miller W, Nguyen K, Vangala S, et al. Clinicians can independently predict 30-day hospital readmissions as well as the LACE index. BMC Health Serv Res 2018; 18 32
| Clinicians can independently predict 30-day hospital readmissions as well as the LACE index.Crossref | GoogleScholarGoogle Scholar | 29357864PubMed |
[4] Nguyen OK, Washington C, Clark C, et al. Man vs machine: comparing physician vs electronic health record-based model predictions for 30-day hospital readmissions. J Gen Intern Med 2021; 36 2555–2562.
| Man vs machine: comparing physician vs electronic health record-based model predictions for 30-day hospital readmissions.Crossref | GoogleScholarGoogle Scholar | 33443694PubMed |
[5] Donnan P, Dorward D, Mutch B, et al. Development and validation of a model for predicting emergency admissions over the next year (PEONY). Arch Intern Med 2008; 168 1416–1422.
| Development and validation of a model for predicting emergency admissions over the next year (PEONY).Crossref | GoogleScholarGoogle Scholar | 18625922PubMed |
[6] Hippisley-Cox J, Coupland C. Predicting risk of emergency admission to hospital using primary care data: derivation and validation of the QAdmissions score. BMJ Open 2013; 3 e003482
| Predicting risk of emergency admission to hospital using primary care data: derivation and validation of the QAdmissions score.Crossref | GoogleScholarGoogle Scholar | 23959760PubMed |
[7] Tomlin A, Lloyd H, Tilyard M. Risk stratification of New Zealand general practice patients for emergency admissions in the next year: adapting the PEONY model for New Zealand. J Prim Health Care 2016; 8 227–237.
| Risk stratification of New Zealand general practice patients for emergency admissions in the next year: adapting the PEONY model for New Zealand.Crossref | GoogleScholarGoogle Scholar | 29530206PubMed |
[8] Ehrenberg N, Terris A, Marshall K. A rapid review of the HCH Model in Capital and Coast District Health Board. Wellington: International Foundation for Integrated Care; 2020.
[9] Charlson M, Pompei P, Ales K, et al. A new method of classifying prognositic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40 373–383.
| A new method of classifying prognositic comorbidity in longitudinal studies: development and validation.Crossref | GoogleScholarGoogle Scholar | 3558716PubMed |
[10] Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care 1998; 36 8–27.
| Comorbidity measures for use with administrative data.Crossref | GoogleScholarGoogle Scholar | 9431328PubMed |
[11] Stanley J, Semper K, Millar E, et al. Epidemiology of multimorbidity in New Zealand: a cross-sectional study using national-level hospital and pharmaceutical data. BMJ Open 2018; 8 e021689
| Epidemiology of multimorbidity in New Zealand: a cross-sectional study using national-level hospital and pharmaceutical data.Crossref | GoogleScholarGoogle Scholar | 29794103PubMed |
[12] Violan C, Foguet-Boreu Q, Flores-Mateo G, et al. Prevalence, determinants and patterns of multimorbidity in primary care: a systematic review of observational studies. PLoS One 2014; 9 e102149
| Prevalence, determinants and patterns of multimorbidity in primary care: a systematic review of observational studies.Crossref | GoogleScholarGoogle Scholar | 25048354PubMed |
[13] Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005; 43 1130–1139.
| Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.Crossref | GoogleScholarGoogle Scholar | 16224307PubMed |
[14] Khan N, Perera R, Harper S, et al. Adaption and validation of the Charlson Index for Read/OXMIS coded databases. BMC Fam Pract 2010; 11 1
| Adaption and validation of the Charlson Index for Read/OXMIS coded databases.Crossref | GoogleScholarGoogle Scholar | 20051110PubMed |
[15] Metcalfe D, Masters J, Delmestri A, et al. Coding algorithms for defining Charlson and Elixhauser co-morbidities in Read-coded databases. BMC Med Res Methodol 2019; 19 115
| Coding algorithms for defining Charlson and Elixhauser co-morbidities in Read-coded databases.Crossref | GoogleScholarGoogle Scholar | 31170931PubMed |
[16] Jordan K, Porcheret M, Croft P. Quality of morbidity coding in general practice computerized medical records: a systematic review. Fam Pract 2004; 21 396–412.
| Quality of morbidity coding in general practice computerized medical records: a systematic review.Crossref | GoogleScholarGoogle Scholar | 15249528PubMed |
[17] Crooks C, West J, Card T. A comparison of the recording of comorbidity in primary and secondary care by using the Charlson Index to predict short-term and long-term survival in a routine linked data cohort. BMJ Open 2015; 5 e007974
| A comparison of the recording of comorbidity in primary and secondary care by using the Charlson Index to predict short-term and long-term survival in a routine linked data cohort.Crossref | GoogleScholarGoogle Scholar | 26048212PubMed |
[18] Wei L. The accelerated failure time model: a useful alternative to the Cox regression model in survival analysis. Stat Med 1992; 11 1871–1879.
| The accelerated failure time model: a useful alternative to the Cox regression model in survival analysis.Crossref | GoogleScholarGoogle Scholar | 1480879PubMed |
[19] Khanna S, Rolls D, Boyle J, et al. A risk stratification tool for hospitalisation in Australia using primary care data. Sci Rep 2019; 9 5011
| A risk stratification tool for hospitalisation in Australia using primary care data.Crossref | GoogleScholarGoogle Scholar | 30899054PubMed |
