Interpreting Trend Graphs

Below are examples of trend graphs that introduce various relationships among county, state, and national level data. Our trends are estimated using linear regression for all of the years of data included in the graph.

Decreasing (improving) trend – similar to state and national trends

This graph shows an improving trend in preventable hospital stays for Columbia County. The change appears linear, and is happening at a similar rate to the state and national trends. One can infer that the changes in the hospitalization rate are due to larger state or national changes that have taken place, rather than changes unique to local conditions. Changes such as these are unlikely to cause changes in ranks, since the county is progressing at the same rate as the state as a whole. 

 

Decreasing (improving) trend – not keeping up with state and national trends

This graph shows an improving trend in uninsured for Haskell County. The change appears relatively linear, but is happening at a slower rate than the state and national trends. This might mean the improvements happening in the state or nation are not reaching Haskell County, or that locally specific changes are preventing greater improvement. Changes like this can lead to worsening ranks.

 

Decreasing (improving) trend – improving faster than state and national trends

This graph shows an improving trend in preventable hospital stays for Niobrara County. The change appears linear, and is happening at a faster rate than the state and national trends. This might mean that locally specific changes are causing greater improvement. Changes like this can result in improved ranks.

 

Increasing (worsening) trend (with lots of uncertainty)-but not significantly different from state or national trend

This graph shows a worsening trend in premature death for Bell County. There is a good deal of uncertainty around the slope estimate for Bell County, most likely due to this being a smaller county with smaller numbers of deaths. It is hard to interpret these findings, other than being aware that that this area may be a worsening concern.

 

Increasing (worsening) trend – significantly different from state and national trends – regression to the mean

This graph shows a worsening trend in preventable hospital stays for Powhatan County. This trend is significantly worse than the state and national trends. However, this graph illustrates a statistical concern called “regression to the mean.“ Each year data is collected there will be counties with the best rate of preventable hospitalizations and the worst. For either of those types of counties, those rates are in part due to real underlying factors in the community and in part due to random chance. Being at the either end of the range, these counties have a greater likelihood of moving closer to the statistical average in the next year. This can suggest a trend where no underlying change has really occurred. In this graph, Powhatan County had rates far below the Virginia average for the first few years of measurement, but has been trending towards the state average since then. While some of this may be due to actual underlying changes, some is probably due to regression to the mean.

 

Increasing (worsening) trend – significantly different from state and national trends

This graph shows a worsening trend in preventable hospital stays for Red River Parish. This trend is significantly worse than the state and national trends. Unlike the illustration of regression to the mean, one can have greater confidence in the underlying factors affecting this trend. Red River Parish started with among the worst rates of preventable hospitalizations in Louisiana, and the rate has gotten worse (moving further from the mean), while the average rate in Lousiana has improved. This suggests that something in Red River Parish is causing continuing and worsening problems with preventable hospital stays.

 

Stable trend – similar to state and national trends

This graph shows a relatively stable trend in preventable hospital stays for Broward County. This trend is not significantly better or worse than the state and national trends. 

 

Stable trend – significantly different from state and national trends

This graph shows a relatively stable trend in preventable hospital stays for Mississippi County. However, this trend is significantly worse than the state and national trends. This might mean the improvements happening in the state or nation are not reaching Mississippi County, or that locally specific factors are preventing change. This type of situation might cause a change in ranks for the county, even with a stable trend.