Developing a Regional Distributed Data Network for Surveillance of Chronic Health Conditions: The Colorado Health Observation Regional Data Service.

Electronic health records (EHRs) provide an alternative to traditional public health surveillance surveys and administrative data for measuring the prevalence and impact of chronic health conditions in populations. As the infrastructure for secondary use of EHR data improves, many stakeholders are poised to benefit from data partnerships for regional access to information. Electronic health records can be transformed into a common data model that facilitates data sharing across multiple organizations and allows data to be used for surveillance. The Colorado Health Observation Regional Data Service, a regional distributed data network, has assembled diverse data partnerships, flexible infrastructure, and transparent governance practices to better understand the health of communities through EHR-based, public health surveillance. This article describes attributes of regional distributed data networks using EHR data and the history and design of Colorado Health Observation Regional Data Service as an emerging public health surveillance tool for chronic health conditions. Colorado Health Observation Regional Data Service and our experience may serve as a model for other regions interested in similar surveillance efforts. While benefits from EHR-based surveillance are described, a number of technology, partnership, and value proposition challenges remain.

The Colorado Collaborative for Nursing Research: Providing Trended Nurse-Sensitive Patient Outcome Data Across Entities.

Nurse leaders lack timely access to trended electronic health record (EHR) data to drive decision making. Robust nurse-sensitive patient outcome data are difficult to locate in EHRs and largely absent across entities. The Colorado Collaborative for Nursing Research is currently testing a federated data system to get nurse leaders the information they need, when they need it.

Haemophilus influenzae LicB contributes to lung damage in an aged mice co-infection model.

Phosphorylcholine (ChoP) decoration of lipopolysaccharides is an important virulence strategy adopted by Haemophilus influenzae to establish a niche on the mucosal surface and to promote adherence to the host cells. The incorporation of ChoP on the LPS surface involves the lic1 operon, which consists of the licA, licB, licC, and licD genes. Among which, licB is a choline transporter gene required for acquisition of choline from environmental sources. In this study, we investigated the pathogenesis of the licB gene in an aged mice infection model. Due to immediate clearance of H. influenzae upon infection in mice, we employed influenza A virus and H. influenzae co-infection model. Our data showed that in the co-infection model, the secondary bacterial infection with a very low H. influenzae concentration of 100 colony forming unit is lethal to the aged mice. Although we did not observe any differences in weight loss between parent and licB mutant strains during the course of infection, a significant reduction of lung tissue damage was observed in the licB mutant infected aged mice. These results suggest that the licB gene is a virulence factor during H. influenzae infection in the lung in aged mice, possibly due to the increased binding to the host cell receptor via ChoP expression on the bacterial surface. In addition, when aged mice and mature mice were compared in the challenge experiments, we did not observe any protective immunity in the co-infection model suggesting the detrimental effects of the secondary bacterial infection on the aged mice in contrast to obvious immune-protections observed in the mature mice. The results of our experiments also implied that the co-infection model with influenza A virus and H. influenzae may be employed as a model system to study H. influenzae pathogenesis in vivo in aged mice.

Federated Aggregate Cohort Estimator (FACE): an easy to deploy, vendor neutral, multi-institutional cohort query architecture.

Cross-institutional data sharing for cohort discovery is critical to enabling future research. While particularly useful in rare diseases, the ability to target enrollment and to determine if an institution has a sufficient number of patients is valuable in all research, particularly in the initiation of projects and collaborations. An optimal technology solution would work with any source database with minimal resource investment for deployment and would meet all necessary security and confidentiality requirements of participating organizations. We describe a platform-neutral reference implementation to meet these requirements: the Federated Aggregate Cohort Estimator (FACE). FACE was developed and implemented through a collaboration of The University of Alabama at Birmingham (UAB), The Ohio State University (OSU), the University of Massachusetts Medical School (UMMS), and the Denver Health and Hospital Authority (DHHA) a clinical affiliate of the Colorado Clinical and Translational Sciences Institute. The reference implementation of FACE federated diverse SQL data sources and an i2b2 instance to estimate combined research subject availability from three institutions. It used easily-deployed virtual machines and addressed privacy and security concerns for data sharing.

Responding to COVID-19 with integrative health and sheltering models for persons experiencing homelessness in Southern Ontario, Canada: protocol for a qualitative study exploring implementation and sustainability.

INTRODUCTION: COVID-19 has disproportionately impacted persons experiencing homelessness in Canada, who are at an increased risk of infection and severe outcomes. In response to the pandemic, several regions have adopted programmes that aim to address the intersecting nature of health and social challenges faced by persons facing homelessness. These programmes adopted during the pandemic may contribute to broader health and social impacts beyond limiting COVID-19 transmission, but the processes involved in developing and implementing these types of programmes and their sustainability after the pandemic are unknown. Our overall goal is to understand the processes of developing and implementing integrative health and sheltering initiatives in Ontario during COVID-19, as well as their sustainability post-pandemic. METHODS AND ANALYSIS: This study will use a multiple case study design-two cases over 1 year-enabling us to investigate how integrative health and sheltering approaches have been implemented in two mid-sized cities in Ontario, Canada. Each case will offer a unique narrative; through cross-case analysis, the cases will highlight programme operations, successes and challenges. Data will be collected using semi-structured interviews with programme staff and managers, and document analysis. Project partners will be brought together to further explore and interpret findings, along with co-creating a sustainability action plan and policy documents. ETHICS AND DISSEMINATION: Ethics clearance was obtained through the Western University Research Ethics Board and the University of Waterloo Office of Research Ethics. Findings will be disseminated through publications, conference presentations and lay summary reports.

A process to deduplicate individuals for regional chronic disease prevalence estimates using a distributed data network of electronic health records.

Introduction: Learning health systems can help estimate chronic disease prevalence through distributed data networks (DDNs). Concerns remain about bias introduced to DDN prevalence estimates when individuals seeking care across systems are counted multiple times. This paper describes a process to deduplicate individuals for DDN prevalence estimates. Methods: We operationalized a two-step deduplication process, leveraging health information exchange (HIE)-assigned network identifiers, within the Colorado Health Observation Regional Data Service (CHORDS) DDN. We generated prevalence estimates for type 1 and type 2 diabetes among pediatric patients (0-17 years) with at least one 2017 encounter in one of two geographically-proximate DDN partners. We assessed the extent of cross-system duplication and its effect on prevalence estimates. Results: We identified 218 437 unique pediatric patients seen across systems during 2017, including 7628 (3.5%) seen in both. We found no measurable difference in prevalence after deduplication. The number of cases we identified differed slightly by data reconciliation strategy. Concordance of linked patients' demographic attributes varied by attribute. Conclusions: We implemented an HIE-dependent, extensible process that deduplicates individuals for less biased prevalence estimates in a DDN. Our null pilot findings have limited generalizability. Overlap was small and likely insufficient to influence prevalence estimates. Other factors, including the number and size of partners, the matching algorithm, and the electronic phenotype may influence the degree of deduplication bias. Additional use cases may help improve understanding of duplication bias and reveal other principles and insights. This study informed how DDNs could support learning health systems' response to public health challenges and improve regional health.

Prevalence of self-reported postpartum depression specific to season and latitude of birth: evaluating the PRAMS data.

To determine whether season of infant birth or amount of daylight at time and location of birth is a risk factor for self-reported postpartum depression (PPD). The primary hypothesis was that the prevalence of PPD will peak during the darkest winter months. A cross-sectional analysis was conducted using the Pregnancy Risk Assessment Monitoring System (PRAMS) Phase 5 (2004-2006) data set (N = 67,079). Self-reported PPD was established using a modified version of the Patient Health Questionnaire-2 (PHQ-2) included in the PRAMS questionnaire. Logistic regression for complex survey design was used to determine odds ratios and 95% confidence intervals. No significant relationship was found between mild or moderate PPD and either season of birth or length of daylight at birth. By analyzing a large, multi-state sample, this study adds to the equivocal preexisting literature suggesting that there is no significant relationship between the season of birth or length of daylight at birth and PPD.

Evaluating Population Coverage in a Regional Distributed Data Network: Implications for Electronic Health Record-Based Public Health Surveillance.

OBJECTIVE: Electronic health records (EHRs) hold promise as a public health surveillance tool, but questions remain about how EHR patients compare with populations in health and demographic surveys. We compared population characteristics from a regional distributed data network (DDN), which securely and confidentially aggregates EHR data from multiple health care organizations in the same geographic region, with population characteristics from health and demographic surveys. METHODS: Ten health care organizations participating in a Colorado DDN contributed data for coverage estimation. We aggregated demographic and geographic data from 2017 for patients aged ≥18 residing in 7 counties. We used a cross-sectional design to compare DDN population size, by county, with the following survey-estimated populations: the county population, estimated by the American Community Survey (ACS); residents seeking any health care, estimated by the Colorado Health Access Survey; and residents seeking routine (eg, primary) health care, estimated by the Behavioral Risk Factor Surveillance System. We also compared data on the DDN and survey populations by sex, age group, race/ethnicity, and poverty level to assess surveillance system representativeness. RESULTS: The DDN population included 609 840 people in 7 counties, corresponding to 25% coverage of the general adult population. Population coverage ranged from 15% to 35% across counties. Demographic distributions generated by DDN and surveys were similar for many groups. Overall, the DDN and surveys assessing care-seeking populations had a higher proportion of women and older adults than the ACS population. The DDN included higher proportions of Hispanic people and people living in high-poverty neighborhoods compared with the surveys. CONCLUSION: The DDN population is not a random sample of the regional adult population; it is influenced by health care use patterns and organizations participating in the DDN. Strengths and limitations of DDNs complement those of survey-based approaches. The regional DDN is a promising public health surveillance tool.

Assessing the relationship between health information exchanges and public health agencies.

While health information exchange (HIE) continues to be promoted by both private and public sector supporters as potentially transformative for healthcare, there is concern as to how it will relate to and assist population health activities. This article reports the results of an on-line survey of regional health information organizations (RHIOs) and state and local health department representatives' beliefs and attitudes regarding collaboration in HIE. Factors that may encourage cooperation among public health agencies and RHIOs include broader geographic and multiorganization catchment areas, early laboratory involvement, early inclusion of public health as a RHIO partner, and efforts to resolve perceived barriers to cooperation. The exchange of electronic lab data was the most common type of data being exchanged and could serve as a model for how to accomplish HIE in general. Health departments (state and local) may serve as neutral conveners of RHIOs and could offer a bridge for acceptance by stakeholders, dispelling mistrust, offering data management expertise, and a capacity and tradition of ensuring privacy for confidential data.

The relationship between gender and clinical management after exercise stress testing.

BACKGROUND: Controversy remains regarding whether gender differences exist in clinical management after exercise treadmill testing (ETT). METHODS: We studied 7,506 patients (49.8% women) without documented coronary heart disease referred for ETT from July 2001 to June 2004 in a community-based setting. We assessed the relationship between gender and subsequent diagnostic testing (secondary stress testing or coronary angiography) within 6 months after ETT. Secondary outcomes included subsequent stress testing, coronary angiography, and new cardiology visits in the 6-month interval. Multivariable analyses assessed the relationship between gender and these outcomes adjusting for demographic, clinical, and stress test characteristics. In subsequent analyses, patients were stratified by Duke Treadmill Scores (Duke University, Durham, NC). RESULTS: Compared with men, women referred for ETT were older, had a higher prevalence of some cardiac risk factors, achieved lower peak workloads, and, more often, experienced chest pain or ST-segment changes. After accounting for differences in clinical and ETT parameters, gender was not associated with any subsequent diagnostic testing in the 6 months after ETT (OR 1.0, 95% CI 0.85-1.18). In secondary analyses, women were less likely to undergo angiography (OR 0.63, 95% CI 0.47-0.83) with a trend toward more subsequent stress testing. Stratified analyses revealed less subsequent testing in high-to-intermediate Duke Treadmill Score women compared with men (OR 0.61, 95% CI 0.48-0.79). Women and men were equally likely to die (hazards ratio 0.93, 95% CI 0.61-1.44) in the adjusted survival analysis. CONCLUSIONS: Overall, women and men equally underwent subsequent diagnostic testing after ETT. Although women were less likely to undergo angiography and higher-risk women were less likely to undergo subsequent testing, adverse events were not higher in women. Given these findings, assumptions regarding gender disparities in clinical management after ETT should be reevaluated in other settings.

High-resolution microbiome profiling uncovers Fusobacterium nucleatum, Lactobacillus gasseri/johnsonii, and Lactobacillus vaginalis associated to oral and oropharyngeal cancer in saliva from HPV positive and HPV negative patients treated with surgery and chemo-radiation.

Microbiome studies show altered microbiota in head and neck squamous cell carcinoma (HNSCC), both in terms of taxonomic composition and metabolic capacity. These studies utilized a traditional bioinformatics methodology, which allows for accurate taxonomic assignment down to the genus level, but cannot accurately resolve species level membership. We applied Resphera Insight, a high-resolution methodology for 16S rRNA taxonomic assignment that is able to provide species-level context in its assignments of 16S rRNA next generation sequencing (NGS) data. Resphera Insight applied to saliva samples from HNSCC patients and healthy controls led to the discovery that a subset of HNSCC saliva samples is significantly enriched with commensal species from the vaginal flora, including Lactobacillus gasseri/johnsonii (710x higher in saliva) and Lactobacillus vaginalis (52x higher in saliva). These species were not observed in normal saliva from Johns Hopkins patients, nor in 16S rRNA NGS saliva samples from the Human Microbiome Project (HMP). Interestingly, both species were only observed in saliva from Human Papilloma Virus (HPV) positive and HPV negative oropharyngeal cancer patients. We confirmed the representation of both species in HMP data obtained from mid-vagina (n=128) and vaginal introitus (n=121) samples. Resphera Insight also led to the discovery that Fusobacterium nucleatum, an oral cavity flora commensal bacterium linked to colon cancer, is enriched (600x higher) in saliva from a subset of HNSCC patients with advanced tumors stages. Together, these high-resolution analyses on 583 samples suggest a possible role for bacterial species in the therapeutic outcome of HPV positive and HPV negative HNSCC patients.

Injuries from slips and trips in construction.

Construction injuries preceded by a slip or trip were documented using data from the building of the Denver International Airport (Denver, Colorado, USA), the largest construction project in the world at the time. Slips and trips occurred at a rate of 5/200,000 h worked accounting for 18% of all injuries and 25% of workers' compensation payments, or more than $10 million. Slips contributed to the vast majority (85%) of same-level falls and over 30% of falls from height, as well as a significant number of musculoskeletal injures sustained after slipping or tripping but without falling. The injury burden would have been under-recognized in analyses of most coded compensation records. In contrast to other types of injuries, the most common contributing factors were environmental in nature including conditions of walking and working surfaces, terrain and weather. Due to the very dynamic nature of construction work, reducing slips and trips will require a focus on environmental and organizational solutions that evolve as the site changes and the construction project evolves.

Refining the focus of construction injury surveillance.

We conducted two studies of construction injury occurring at Denver International Airport (DL4), whose construction required 31 million work hours. Initially we conducted a retrospective cohort study that allowed estimation of injury and workers' compensation (WC) payment rates for strata such as size of employer and type of work; risk factors were also estimated. The second study examined written injury reports for 4,000 injuries at DIA. We modified Haddon's matrix to classify factors contributing to injury. We identified 108 factors within 4 broad categories: human, object, environment and organization. This approach provided information on rates at which each factor contributed to injury and the WC payment rates for each factor. A study shortcoming was that injury reports varied in completeness and quality. In a third ongoing study, to compensate for the shortcomings of injury reports, particularly to improve consistency and completeness of data, we designed a worker questionnaire completed immediately after injury, which included questions specific to hazards associated with each type of injury. Upon completion, the interviewer (a safety professional) uses the modified Haddon's matrix to note contributors to the injury and explain briefly the reasons for each notation. This requires the interviewer to consider a full set of possible factors and determine whether they contributed to injury. This process elicits richer data and places specific factors within the four higher-level categories. This process confer advantages on both contractors and researchers. Contractors can become immediately aware of contributing factors and ameliorate them quickly. The data can also be used in post hoc analysis of injury etiology. Moreover, the data are sufficiently flexible and complete that they can be coded into schemes describing sequences of events leading to injury, as well as those simply identifying factors contributing to injury. Haddon's matrix is invaluable in such analysis because it leads to a fuller understanding of the origins of the most proximate contributors to injury than would otherwise occur. Particularly for contractors and owners with significant safety infrastructures, this approach may be attractive, because it allows for more complete and quicker correction of specific hazards and, with systematic evaluation, recognition of more general safety concerns present on many construction sites.

16S rRNA amplicon sequencing identifies microbiota associated with oral cancer, human papilloma virus infection and surgical treatment.

Systemic inflammatory events and localized disease, mediated by the microbiome, may be measured in saliva as head and neck squamous cell carcinoma (HNSCC) diagnostic and prognostic biomonitors. We used a 16S rRNA V3-V5 marker gene approach to compare the saliva microbiome in DNA isolated from Oropharyngeal (OPSCC), Oral Cavity Squamous Cell Carcinoma (OCSCC) patients and normal epithelium controls, to characterize the HNSCC saliva microbiota and examine their abundance before and after surgical resection.The analyses identified a predominance of Firmicutes, Proteobacteria and Bacteroidetes, with less frequent presence of Actinobacteria and Fusobacteria before surgery. At lower taxonomic levels, the most abundant genera were Streptococcus, Prevotella, Haemophilus, Lactobacillus and Veillonella, with lower numbers of Citrobacter and Neisseraceae genus Kingella. HNSCC patients had a significant loss in richness and diversity of microbiota species (p<0.05) compared to the controls. Overall, the Operational Taxonomic Units network shows that the relative abundance of OTU's within genus Streptococcus, Dialister, and Veillonella can be used to discriminate tumor from control samples (p<0.05). Tumor samples lost Neisseria, Aggregatibacter (Proteobacteria), Haemophillus (Firmicutes) and Leptotrichia (Fusobacteria). Paired taxa within family Enterobacteriaceae, together with genus Oribacterium, distinguish OCSCC samples from OPSCC and normal samples (p<0.05). Similarly, only HPV positive samples have an abundance of genus Gemellaceae and Leuconostoc (p<0.05). Longitudinal analyses of samples taken before and after surgery, revealed a reduction in the alpha diversity measure after surgery, together with an increase of this measure in patients that recurred (p<0.05). These results suggest that microbiota may be used as HNSCC diagnostic and prognostic biomonitors.

Analysis of text from injury reports improves understanding of construction falls.

OBJECTIVE: We combined payroll data, coded workers' compensation (WC) data, and text descriptions of injuries from the construction of Denver International Airport to create a more comprehensive picture of falls from height (FFH) than is typically available from WC data. Text descriptions were coded to identify circumstances surrounding falls. Slips/trips preceded one third of FFH, often involving motor vehicles or heavy equipment. Another third involved movement or collapse of work surfaces, usually ladders or scaffolds. CONCLUSIONS: The significant contribution of motor vehicles and heavy equipment to FFH, particularly those preceded by slips/trips, was not apparent from coded data. Heavy equipment engineering modifications are called for and workers in street/roadway construction/site development need fall protection training. Text analyses allow exploration of factors not identified at the time of data collection and better understanding of the context in which injuries occur.

Sharing behavioral data through a grid infrastructure using data standards.

OBJECTIVE: In an effort to standardize behavioral measures and their data representation, the present study develops a methodology for incorporating measures found in the National Cancer Institute's (NCI) grid-enabled measures (GEM) portal, a repository for behavioral and social measures, into the cancer data standards registry and repository (caDSR). METHODS: The methodology consists of four parts for curating GEM measures into the caDSR: (1) develop unified modeling language (UML) models for behavioral measures; (2) create common data elements (CDE) for UML components; (3) bind CDE with concepts from the NCI thesaurus; and (4) register CDE in the caDSR. RESULTS: UML models have been developed for four GEM measures, which have been registered in the caDSR as CDE. New behavioral concepts related to these measures have been created and incorporated into the NCI thesaurus. Best practices for representing measures using UML models have been utilized in the practice (eg, caDSR). One dataset based on a GEM-curated measure is available for use by other systems and users connected to the grid. CONCLUSIONS: Behavioral and population science data can be standardized by using and extending current standards. A new branch of CDE for behavioral science was developed for the caDSR. It expands the caDSR domain coverage beyond the clinical and biological areas. In addition, missing terms and concepts specific to the behavioral measures addressed in this paper were added to the NCI thesaurus. A methodology was developed and refined for curation of behavioral and population science data.

Enduring effects of prenatal and infancy home visiting by nurses on maternal life course and government spending: follow-up of a randomized trial among children at age 12 years.

OBJECTIVE: To test, among an urban primarily African American sample, the effects of prenatal and infancy home visiting by nurses on mothers' fertility, partner relationships, and economic self-sufficiency and on government spending through age 12 years of their firstborn child. DESIGN: Randomized controlled trial. SETTING: Public system of obstetric and pediatric care in Memphis, Tennessee. PARTICIPANTS: A total of 594 urban primarily African American economically disadvantaged mothers (among 743 who registered during pregnancy). Intervention Prenatal and infancy home visiting by nurses. MAIN OUTCOME MEASURES: Mothers' cohabitation with and marriage to the child's biological father, intimate partner violence, duration (stability) of partner relationships, role impairment due to alcohol and other drug use, use and cost of welfare benefits, arrests, mastery, child foster care placements, and cumulative subsequent births. RESULTS: By the time the firstborn child was 12 years old, nurse-visited mothers compared with control subjects reported less role impairment owing to alcohol and other drug use (0.0% vs 2.5%, P = .04), longer partner relationships (59.58 vs 52.67 months, P = .02), and greater sense of mastery (101.04 vs 99.60, P = .005). During this 12-year period, government spent less per year on food stamps, Medicaid, and Aid to Families with Dependent Children and Temporary Assistance for Needy Families for nurse-visited than control families ($8772 vs $9797, P = .02); this represents $12 300 in discounted savings compared with a program cost of $11 511, both expressed in 2006 US dollars. No statistically significant program effects were noted on mothers' marriage, partnership with the child's biological father, intimate partner violence, alcohol and other drug use, arrests, incarceration, psychological distress, or reports of child foster care placements. CONCLUSION: The program improved maternal life course and reduced government spending among children through age 12 years.

"Safety is everyone's job:" the key to safety on a large university construction site.

PROBLEM: Construction risk management is challenging. METHOD: We combined data on injuries, costs, and hours worked, obtained through a Rolling Owner-Controlled Insurance Program (ROCIP), with data from focus groups, interviews, and field observations, to prospectively study injuries and hazard control on a large university construction project. RESULTS: Lost-time injury rates (1.0/200,000 hours worked) were considerably lower than reported for the industry, and there were no serious falls from height. Safety was considered in the awarding of contracts and project timeline development; hazard management was iterative. A top-down management commitment to safety was clearly communicated to, and embraced by, workers throughout the site. DISCUSSION AND IMPACT: A better understanding of how contracting relationships, workers' compensation, and liability insurance arrangements influence safety could shift risk management efforts from worker behaviors to a broader focus on how these programs and relationships affect incentives and disincentives for workplace safety and health.

Effects of nurse home visiting on maternal and child functioning: age-9 follow-up of a randomized trial.

OBJECTIVE: Our goal was to test the effect of prenatal and infancy home visits by nurses on mothers' fertility and children's functioning 7 years after the program ended at child age 2. METHODS: We conducted a randomized, controlled trial in a public system of obstetric and pediatric care. A total of 743 primarily black women <29 weeks' gestation, with previous live births and at least 2 sociodemographic risk characteristics (unmarried, <12 years of education, unemployed), were randomly assigned to receive nurse home visits or comparison services. Primary outcomes consisted of intervals between births of first and second children and number of children born per year; mothers' stability of relationships with partners and relationships with the biological father of the child; mothers' use of welfare, food stamps, and Medicaid; mothers' use of substances; mothers' arrests and incarcerations; and children's academic achievement, school conduct, and mental disorders. Secondary outcomes were the sequelae of subsequent pregnancies, women's employment, experience of domestic violence, and children's mortality. RESULTS: Nurse-visited women had longer intervals between births of first and second children, fewer cumulative subsequent births per year, and longer relationships with current partners. From birth through child age 9, nurse-visited women used welfare and food stamps for fewer months. Nurse-visited children born to mothers with low psychological resources, compared with control-group counterparts, had better grade-point averages and achievement test scores in math and reading in grades 1 through 3. Nurse-visited children, as a trend, were less likely to die from birth through age 9, an effect accounted for by deaths that were attributable to potentially preventable causes. CONCLUSIONS: By child age 9, the program reduced women's rates of subsequent births, increased the intervals between the births of first and second children, increased the stability of their relationships with partners, facilitated children's academic adjustment to elementary school, and seems to have reduced childhood mortality from preventable causes.

Methods for using narrative text from injury reports to identify factors contributing to construction injury.

BACKGROUND: Several methods exist for classifying injuries from written text, thereby identifying possible points of intervention. We describe an innovative method for such classification. METHODS: Using Haddon's matrix as a framework, two independent reviewers coded text from over 4,000 injury reports into a qualitative software package to identify factors contributing to injuries sustained during construction of Denver International Airport (DIA). We compared our classification scheme with three others. RESULTS: This process created a coded data set, an expanded version of Haddon's matrix adapted for construction injury, and coding rules for interpreting narrative text. Haddon's matrix provides a flexible theoretical framework for coding information about a spectrum of contributing factors. CONCLUSIONS: Narrative descriptions from injury reports can provide detail on circumstances surrounding injuries and identify factors contributing to injury. Forms guiding investigators to explicitly consider human, organizational, and environmental factors could foster more complete descriptions of factors contributing to construction injury.