Single assessment of risk predicted which elderly patients would fall

Editor—Oliver et al used a scientific approach to develop an evidence based risk assessment tool that predicts with high specificity which elderly inpatients will fall.¹ Between July and September 1996 we examined risk factors prospectively for 154 patients admitted to two acute and rehabilitation geriatric wards in a district general hospital. Our aim was to identify factors on admission that might have predictive value for the duration of the inpatient episode. A simple risk assessment form for falls was completed once by the admitting nurse.²

We recorded 53 falls by 29 patients. There was no association between a fall and the following observations: agitation, temporal or spatial disorientation, toileting difficulties, and mobility not requiring supervision. For several factors the difference in distribution between fallers and non-fallers seemed unlikely to have occurred by chance. These factors included a medical history of broken hip, stroke, or Parkinson’s disease; history of a fall within the past month; supervision needed for all mobility; and poor eyesight (all P<0.05).

We then considered that having more than two of these associated factors indicated a high risk of falling. In the lower risk group the probability of falling was 1 in 7, and in the higher risk group it was 1 in 2, a risk ratio of 3.31 (95% confidence interval 1.8 to 6.1). Calculations of specificity and sensitivity were 89.6% and 38% respectively (table).

In contrast to Oliver et al, we chose to examine factors that were present just on admission rather than assess the patients at weekly intervals. Elderly patients can undergo rapid changes in orientation and mobility after acute admission; we found that the factors with a predictive value for the duration of the patients’ stay were purely historical and unlikely to change quickly. Oliver et al’s repeated assessments allowed the impact of more changeable predictors to be registered.

In the population in a district general hospital Oliver et al found that if specificity of 87.6% was achieved, sensitivity dropped to 54.4% (42.8% to 65.7%). This raises the issue of whether repeated assessments in this population are worth while, as these figures are similar to our own. Ultimately the type of assessment needed will reflect the ward population, as a longer stay rehabilitation ward may be just as well served by a single assessment of risk based on more stable factors, whereas an acute ward would need a repeated review of predictors that can change more rapidly.

Table.

Number of risk factors noted for 154 patients admitted to geriatric wards who did and did not fall during their stay in hospital

No of risk factors	Faller	Non-faller	Total
3 or 4 (“at risk”)	11	13	24
0 to 2 (“no risk”)	18	112	130
All patients	29	125	154