Quality of life after injury: an Individual Patient Data (IPD) meta-analysis

There is a need for information on the Health Related Quality of Life (HRQoL) of injury populations. HRQoL measures, like the EQ-5D, have seen a dramatic increase in use over the last ten years. Despite this, general information on HRQoL pathways to recovery are missing from the literature, as many injury studies focus on specific injury types within exclusive populations. This study: (1) identified studies using the EQ-5D in injury populations, (2) collected individual level patient data, and (3) produced population level predictions of HRQoL, as defined by the EQ-5D, across injury subtypes. A systematic search of literature concerning injury outcome, published between January 1990 and November 2008 (inclusive), was completed. Studies were excluded if the EQ-5D was not used, a degenerative disease was required for study inclusion or injury was a secondary outcome not required for the study. Researchers were invited to contribute Individual Patient Data (IPD) to the meta-analysis. Injuries collected were grouped into 39 categories based on the Eurocost injury classifications. Multilevel mixed effects models were used to produce predictions across both the dimensions and the Visual Analogue Score (VAS) of the EQ-5D measure. From a potential pool of 13,785 persons across 39 study datasets, IPD from 10,496 injured persons was sourced from 18 discrete datasets. Predictions were fitted to 25 of the 39 injury categories, covering a wide spectrum of injury types. Point predictions were produced for the dimensions and VAS at 3, 30, 120 and 360 days after injury. Figures were also produced of recovery pathways within the EQ-5D dimensions, where relationships with the facet of the EQ-5D and time were statistically significant. Across the majority of injuries pain and discomfort (PD) was the most severely impaired dimension at both 30 and 360 days post injury. Recovery for PD was gradual, with fractures of the femur, lower leg and pelvis showing marked improvements in mobility and usual activities at 360 days but not PD. Regardless of anatomical location of injury a similar pattern was seen between fractures, with high initial impairment and rapid recoveries, and strains, where impairment at 30 days was significantly less but recovery was gradual. VAS scores showed a great variation in subjective assessments of general health states after an injury. The dimensions of cognitive function and anxiety and depression showed a limited relationship with most injury categories. The HRQoL predictions produced here will aid clinical researchers in comparing their populations to norm values, as well as aiding policy makers in obtaining a greater understanding of the burden of injury types. This will aid in both justifying prevention measures, as well as identifying areas for prioritisation of resources