Chapter 8 Discussion
8.3 Comparison with other studies
Primary care supply
Two previous national studies have examined variation in primary care supply between large administrative areas of England. Gravelle and Sutton163 found substantial and persistent between area variation in physician supply between 1975 and 1995. Goddard and colleagues71 extended this time series by adding the years 1996 to 2006and found that variation between administrative areas increased between 1995 and 2006. Our results agree with these previous studies showing large and widening pro-rich inequalities up until 2006/7, after which we see this trend reverse with inequalities narrowing over the remainder of our study period, by the end of which we observe pro-poor inequality in need adjusted primary care supply. While the previous studies examined overall variation between large and
socioeconomically diverse administrative areas, our study adds value by looking specifically at socioeconomic-related inequality between small areas. We are able accurately to attribute GP supply to small areas, based on the location of patients registered to each GP practice, and so can paint a much more fine-grained picture of the socioeconomic distribution of the
primary care workforce than has previously been possible.
Primary care quality
One previous national study examined trends in socioeconomic inequality in primary care process quality from the UK pay-for-performance programme.162 This study only covers the first three years of our eight year study period (2004/5 to 2006/7) but agrees with our findings of reductions in socioeconomic inequality. We find that this reduction in inequality continued but slowed down thereafter and levelled off from 2010/11 to 2011/12.
Waiting time
Most previous studies of inequality in waiting time have used disease-specific indicators focusing on particular specialties or procedures, rather than general indicators covering the whole range of hospital activity which are then adjusted for disease-specific differences in waiting times. To our knowledge, the only other previous study using a general indicator was a cross sectional study using individual level data on men aged 67 and over from Norway in 2004/5, which found very little evidence of socioeconomic differences in waiting time after adjusting for all primary and secondary diagnoses, severity and hospital supply.164 By
contrast, previous disease-specific studies have generally found pro-rich inequality in waiting time for publicly funded inpatient hospital treatment in a range of high income countries with universal health systems.165, 166 Previous disease-specific studies have also found a trend of reducing socioeconomic inequality in the English NHS during the 2000s, for a handful of common non-emergency hospital procedures such as hip replacement, knee replacement cataract, heart bypass and coronary angioplasty.166, 167 Our findings using a general waiting time indicator are thus diametrically opposed to previous findings using disease-specific waiting time indicators. Our indicator did adjust for differences in waiting times between specialties, though not for within-specialty differences between procedures or disease
categories. The disease-specific studies are more reliable, since they focus cleanly on a fairly homogeneous procedure and some of them also include controls for waiting time
prioritisation by severity (the number and type of diagnoses) and for cross sectional
differences in hospital supply (hospital fixed effects). However, the disease-specific studies are more vulnerable to selection bias because they have only examined a selected handful of specific hospital procedures which may not be representative of waiting time differentials across all areas of hospital activity. Both types of study are also subject to selection bias relating to the decision to seek privately funded care, which is partly motivated by the desire to gain a shorter waiting time than publicly funded NHS care.
Preventable hospitalisation
One previous national study examined socioeconomic inequality in preventable
hospitalisation in England covering years 2001/2 to 2012/13.168 This study found similar trends to those we observe, showing a gradual decrease in the rate of chronic ambulatory care sensitive emergency admissions for the average patient and substantial and persistent
socioeconomic inequalities in ambulatory care sensitive emergency admissions over the period.
Repeat hospitalisation
To our knowledge, no previous study has examined socioeconomic inequality in repeat emergency hospitalisation within the same year, or time trends therein. However, many disease-specific studies of 30-day emergency re-admission rates have used socioeconomic status as a control variable in regressions performed for purposes other than measuring socioeconomic inequality. These studies have consistently found substantial and significant cross sectional associations between socioeconomic status and 30-day emergency re-
admission following both emergency and non-emergency hospitalisation.169, 170 The selection of an appropriate duration for this indicator illustrates a tension between capturing the quality of co-ordinated care across different primary and acute care providers over a long time period, versus pinpointing precisely at which point on the patient pathway inequality arises – i.e. which primary or acute service provider is responsible for generating inequality at what point in time.
Dying in hospital
Previous cross sectional studies have found socioeconomic inequalities in dying in hospital, and interpreted this as an indicator of differences in the quality of end-of-life care.171 To our knowledge, however, no previous study has examined trends in these socioeconomic
inequalities over time.
Amenable mortality
One previous national study examined socioeconomic trends in amenable mortality172 in England from 2001/2 to 2011/12. However, this study was conducted at a large area level (324 local authorities) which may potentially mask changing patterns of inequality within these large areas, and it excluded mortality in people aged over 75. This study found both average levels and absolute measures of inequality in amenable mortality to have fallen over this period. Our finer grained analysis looking at much smaller areas (32,482 LSOAs) and including amenable mortality in those over 75 years of age confirms this basic pattern, though reveals a widening of relative inequality that was not apparent in the previous study. Furthermore, our inclusion of this older section of the population results in a higher overall rate of amenable mortality and the more detailed level of analysis we employ reveals wider socioeconomic inequalities.
Overall mortality
Numerous studies have found socioeconomic inequality in overall mortality and life
expectancy in low, middle and high income countries.32, 173, 174 Previous studies have found reductions in absolute socioeconomic inequality in overall mortality in England in the
2000s175 and reductions in both absolute and relative inequality in life expectancy.176 This is all in line with our findings. The reason that relative inequality in life expectancy and
mortality moved in different directions during the 2000s is that the means of these two variables were moving in different directions. Mortality is a shortfall measure (more is worse) which is falling over time, whereas life expectancy is an attainment measure (more is better) which is rising over time.177