Prevalence of multimorbidity and association with neighborhood deprivation
Multimorbidity is a phenomenon that affects 24% of the population aged 16 or over living in England and Wales; 76% of the population have been diagnosed with no or only one condition (Table 2). 18% of the population (representing 75% of the multimorbid population) have multiple physical conditions only, 1% have multiple mental conditions only, and 5% have a mixed condition profile. On average, the population has less than one of 35 listed health problems, with the average number of physical problems exceeding the average number of mental problems (0.84 versus 0.13, respectively).
We find that a similar share of the population is affected by multimorbidity when we use an alternative classification where the 35 health conditions are first grouped into types. According to this classification, 79% of the population is not classified as multimorbid. 13% were diagnosed with two and 8% with three or more of the nine types of conditions. We do not observe substantial differences in the sociodemographic characteristics of people defined as multimorbid on these different measures. From now on, we will focus our empirical analysis on the simplest counting measure.
Table 3 shows the share of the population with multimorbidity overall and broken down by mix of physical and mental health conditions, by neighborhood deprivation deciles.
The prevalence of multimorbidity is only higher in the most deprived (lower) decile of the neighborhood deprivation distribution compared to the least deprived (upper) decile of the neighborhood deprivation distribution. Multimorbidity from physical conditions alone is significantly more prevalent in the most deprived 10% of neighborhoods than in the least deprived 30% of neighborhoods. There is no association between neighborhood deprivation and multimorbidity of mental disorders alone or a mixture of physical and mental disorders.
Figure 1 shows the multimorbidity rates for each decile of the seven neighborhood deprivation domains. Panel A presents the areas we suspected to be closely related to IM and multimorbidity due to overlapping definitions. Panel B presents the respective figures for the domains where we did not suspect a strong link.
The results suggest that panel A domains track IMD rates closely across all deciles, particularly in the bottom 30% and top 20% of the distribution. The breakdown of the income domain most closely resembles that of the IMD. In contrast, none of the Panel B domains track IMD rates across deciles.
Socioeconomic and demographic correlates of multimorbidity and multivariate regressions of multimorbidity on neighborhood deprivation
Multimorbidity also varies by population characteristics (for detailed results, see Supplementary Table S2 online). While older people have higher multimorbidity rates in general, multimorbidity rates involving only mental health conditions are more common among younger age cohorts (i.e., 2-3% per year). compared to
Multimorbidity rates are highest among the white British population compared to their ethnic minority counterparts. Although there is generally no association between multimorbidity and social class of current occupation, multimorbidity rates are significantly higher among inactive people, whether they are retired (43%) or May they be healthy for a long time. illness or disability (65%). Additionally, the long-term ill and disabled have higher rates of multimorbidity involving only mental health conditions (7%).
Finally, there are some associations between multimorbidity and location, that is, the business and professional environment in which participants live. We find that multimorbidity rates are particularly high among those living in busy urban shopping centers, compared with those living in mixed cosmopolitan metropolitan and suburban areas, and those living in independent professional metropolitan service areas. Those living in commercial areas also have high rates of mixed multimorbidity compared to those living in predominantly residential suburban areas, “servant to society” areas, and those living in mixed metropolitan and cosmopolitan suburban areas. .
Next, we performed logistic regressions of two multimorbidity measures on overall neighborhood deprivation and neighborhood deprivation domains. The first measure of multimorbidity, presented in the upper half of Table 4, only considers physical conditions. In contrast, the second measure takes into account both physical and mental conditions (presented in the lower half of Table 4). Since the coefficients of the logistic regressions do not lend themselves to simple interpretation, we report relative marginal effects. Relative marginal effects (MR) express by how many percentage points the average probability would change if the explanatory characteristic changed by one unit, holding everything else constant. For categorical variables, MEs express how much the probability would change if we were to observe a discrete change outside the base category. The baseline predicted probability of multimorbidity, calculated at the mean of the explanatory variables, and the population means provide a reference point for knowing whether MEs are small or large.
The models predict a multimorbidity rate of 21% when only physical conditions are considered and when estimates are not adjusted for socioeconomic and demographic factors; the respective figure is 16% in the adjusted models. All other things being equal, moving from one neighborhood deprivation decile to the next reduces the probability of being multimorbid by 0.007 percentage points. The effect strengths for the health, income and employment domains and the education domain are in the same range and have equal statistical significance. Effect sizes in models fitted to individual characteristics are slightly larger, although this is only revealed at 3 decimal place precision. Interestingly, the effect of crime reaches statistical significance in these models.
Focusing on predictions of multimorbidity when physical and mental conditions are taken into account, the models predict a multimorbidity rate of 24% in unadjusted models and 21% in adjusted models. Trends in effect sizes are consistent with those observed for the measure of multimorbidity related to physical conditions only. Overall, the precision of the estimates is somewhat reduced, due to the greater overall socioeconomic and demographic heterogeneity of the multimorbid population with mental disorders.
Examine the extent of unreported or undiagnosed mental health problems and their association with neighborhood poverty
Participants surveyed online reported a greater number of mental disorders on average than those surveyed face-to-face, controlling for age (results not reported). We find no difference in the reporting rates of chronic physical problems, indicating that there may be a problem with under-reporting or undiagnosed mental problems.
Analyzes were repeated using a common rating of mental disorders collected from all participants to examine whether there is an association between not reporting diagnosed mental health problems and cases of psychological distress assessed by the GHQ. We report the results in Table 5.
We find that having a high GHQ score is associated with having more physical and mental health problems on average. Additionally, there is an association between not reporting mental disorders and having a high GHQ score, controlling for age. There is no association between not reporting chronic physical conditions and having a high GHQ score.
We also find that there is an association between not reporting mental disorders and living in more deprived neighborhoods. The panel nature of the survey data allowed us to confirm that these findings hold when looking at a subset of the sample that had a high GHQ score in the past year and the current year. . This group may be considered more likely to have a chronic mental health condition that may not yet have been medically diagnosed.