Examining systemic differences in mortality after hip repair: a comparative analysis of 30- and 180-day adjusted mortality rates in five health systems

Figures

• 

  Figure 1.. Differences in hospitals median odds ratio across regions. (1.1) 30-day all-cause adjusted mortality. (1.2) 180-day all-cause adjusted mortality. Median odds ratio (MOR, bootstrapped confidence interval) by region. The MOR for each region is represented as a point range with error bars showing the bootstrapped confidence interval. Regions are sorted by MOR value from highest to lowest. Non-USA regions’ labels are bolded to facilitate differentiation. The red dashed vertical line is at value 1, which is the minimum possible MOR value. Extreme outliers are excluded—(1.1): Alaska, Delaware, Nevada, North Dakota, Oklahoma and Rhode Island (82 hospitals in total); and (1.2) Alaska and Nevada (24 hospitals in total).

• 

  Figure 2.. Predicted median for all-cause mortality rate after hip repair: within and across regions variation. (2.1) 30-day all-cause adjusted mortality. (2.2) 180-day all-cause adjusted mortality. Probability density function plot of the median predicted mortality rate per hospital in each region. Each density curve represents the distribution of the median values for all hospitals within a region and is coloured by the health system. Wide and overlapping curves show larger differences within regions than between regions. Extreme outliers are excluded—(2.1) Ten hospitals (seven below and three above, i.e. 0.45% of the total number of hospitals) which include 934 (0.17% of total surgeries) hip repair episodes. (2.2) Ten hospitals (eight below and two above, i.e. 0.45% of the total number of hospitals), which include 938 (0.18% of total surgeries) hip repair episodes.

• 

  Figure 3.. 30-day predicted mortality rates vs. 180-day predicted mortality rates. 30-day predicted mortality rates versus 180-day predicted mortality rates by hospital. Each dot represents a hospital and is coloured by the health system. Coloured dots flag those hospital providers outside the IQR of the median predicted value in both indicators. The size of the dot represents the number of surgeries performed in each hospital. The dashed vertical line marks the median predicted 30-day mortality rate, which is 0.04 per hip repair episode, and the dashed horizontal line marks the median predicted 180-day mortality rate, which is 0.14 per hip repair episode. Extreme outliers are excluded: seven hospitals in Alaska and one hospital in Canada.