Identifying misclassification of injury intent: A latent class analysis of burn registry data
1
Centre for Mental Health and Society, Bangor University, Wrexham, United Kingdom
2
Institute of Population Health, University of Liverpool, Liverpool, United Kingdom
3
Injury Prevention Research Centre, Public Health Foundation of India, New Delhi, India
4
Department of Plastic Surgery and Burns, Mysore Medical College and Research Institute, Mysuru, India
5
Obesity Institute, Leeds Beckett University, Leeds, United Kingdom
6
Social Care and Society, University of Manchester, Manchester, United Kingdom
Popul. Med. 2026;8(Supplement Supplement 1):A3826
ABSTRACT
BACKGROUND:
Development of a method to address misclassification bias of injury intent data is a recognised research priority, hindered by the lack of a ‘gold standard’. Clinical differentiation of intent is challenging for certain injuries, such as burns, that have a high disease burden globally. This study explores the use of latent class analysis (LCA) to identify potentially misclassified burn injuries.
METHODS:
We conducted a two-part study. First, a rapid literature review of burn injury studies from India informed class labelling. Second, LCA was applied to data from 1930 patients included in an Indian hospital burn registry from 2016–2022. Documented intent was excluded as a LCA model variable. Posterior probabilities were subsequently examined by patient-reported intent to estimate misclassification.
RESULTS:
The optimal model had two classes: one labelled ‘intentional’ comprised more females, young adults, and large burns; and one labelled ‘unintentional’ comprised more males, children, and small burns. Median probabilistic assignment to the ‘intentional’ class was 0.989 (IQR 0.878–0.998) for individuals reporting their burn as suicidal, and 0.978 (IQR 0.519–0.992) for homicidal. Among those reporting accidental intent, 23.4% (n=264) were highly likely to belong to the ‘intentional class’. For cases with missing intent, 28.9% (n=61) were similarly misclassified.
CONCLUSIONS:
LCA provided a novel, gold-standard-free preliminary estimate of misclassification of burn injury intent. Findings suggests intentional injuries may be more than twice as common as documented. This has implications for surveillance, development of diagnostic tools, and informing policy for the most affected group – young women with large burns in South Asia. It represents a major advance in methods to reduce misclassification bias in intent data and is likely to be applicable to other injury types.
Development of a method to address misclassification bias of injury intent data is a recognised research priority, hindered by the lack of a ‘gold standard’. Clinical differentiation of intent is challenging for certain injuries, such as burns, that have a high disease burden globally. This study explores the use of latent class analysis (LCA) to identify potentially misclassified burn injuries.
METHODS:
We conducted a two-part study. First, a rapid literature review of burn injury studies from India informed class labelling. Second, LCA was applied to data from 1930 patients included in an Indian hospital burn registry from 2016–2022. Documented intent was excluded as a LCA model variable. Posterior probabilities were subsequently examined by patient-reported intent to estimate misclassification.
RESULTS:
The optimal model had two classes: one labelled ‘intentional’ comprised more females, young adults, and large burns; and one labelled ‘unintentional’ comprised more males, children, and small burns. Median probabilistic assignment to the ‘intentional’ class was 0.989 (IQR 0.878–0.998) for individuals reporting their burn as suicidal, and 0.978 (IQR 0.519–0.992) for homicidal. Among those reporting accidental intent, 23.4% (n=264) were highly likely to belong to the ‘intentional class’. For cases with missing intent, 28.9% (n=61) were similarly misclassified.
CONCLUSIONS:
LCA provided a novel, gold-standard-free preliminary estimate of misclassification of burn injury intent. Findings suggests intentional injuries may be more than twice as common as documented. This has implications for surveillance, development of diagnostic tools, and informing policy for the most affected group – young women with large burns in South Asia. It represents a major advance in methods to reduce misclassification bias in intent data and is likely to be applicable to other injury types.
| eISSN: | 2654-1459 |
We process personal data collected when visiting the website. The function of obtaining information about users and their behavior is carried out by voluntarily entered information in forms and saving cookies in end devices. Data, including cookies, are used to provide services, improve the user experience and to analyze the traffic in accordance with the Privacy policy. Data are also collected and processed by Google Analytics tool (more).
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
