Cross-classified Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy (MAIHDA) to evaluate hospital performance: the case of hospital differences in patient survival after acute myocardial infarction

A cross-sector approach to explore socio-ecological associations with treatment engagement behaviours in Northern Ghana

BACKGROUND

Cancer presents a growing global burden, not least in African countries such as Ghana where high cancer treatment dropouts has been identified due to numerous social, cultural and financial reasons. There is little understanding regarding patterns of treatment access behaviour, especially in Northern Ghana, which this study was designed to explore.

METHODS

Through cross-sector collaboration, we extracted and clinically validated cancer patient records available in the Tamale Teaching Hospital. These were analysed descriptively and through multi-variate logistic regression. A treatment mapping process was also applied to highlight challenges in data collection. Multiple imputation with chained equations was conducted for high levels of missing data. Sensitivity analysis was applied to assess the impact of missing data.

RESULTS

Treatment drop-out was high even when uncertainty due to missing data was accounted for, and only 27% of patients completely engaged with treatment. High drop-out was found for all cancers including those covered by the Ghana National Health Insurance scheme. Multi-variate logistic regression revealed that social, health condition and systemic factors influence treatment engagement until completion. High missing data was observed for liver, ovarian, colorectal, gastric, bladder, oesophageal and head and neck and skin cancers, and soft tissue sarcomas, which limited model fitting.

CONCLUSION

Treatment drop-out is a critical issue in Northern Ghana. There was high missing data due to the dynamic, complex and decentralised treatment pathway. Future studies are needed to understand the complex challenges in data recording.

POLICY SUMMARY

Treatment drop out is a pertinent issue that policy makers should look to address. Further discussion with stakeholders involved in cancer treatment and data collection is required to better understand challenges to routine data collection in the local setting. This will allow policy to be designed to cater for the impact of multiple intersecting health and social factors on treatment completion.

Between-hospital variation in indicators of quality of care: a systematic review

BACKGROUND

Efforts to mitigate unwarranted variation in the quality of care require insight into the 'level' (eg, patient, physician, ward, hospital) at which observed variation exists. This systematic literature review aims to synthesise the results of studies that quantify the extent to which hospitals contribute to variation in quality indicator scores.

METHODS

Embase, Medline, Web of Science, Cochrane and Google Scholar were systematically searched from 2010 to November 2023. We included studies that reported a measure of between-hospital variation in quality indicator scores relative to total variation, typically expressed as a variance partition coefficient (VPC). The results were analysed by disease category and quality indicator type.

RESULTS

In total, 8373 studies were reviewed, of which 44 met the inclusion criteria. Casemix adjusted variation was studied for multiple disease categories using 144 indicators, divided over 5 types: intermediate clinical outcomes (n=81), final clinical outcomes (n=35), processes (n=10), patient-reported experiences (n=15) and patient-reported outcomes (n=3). In addition to an analysis of between-hospital variation, eight studies also reported physician-level variation (n=54 estimates). In general, variation that could be attributed to hospitals was limited (median VPC=3%, IQR=1%-9%). Between-hospital variation was highest for process indicators (17.4%, 10.8%-33.5%) and lowest for final clinical outcomes (1.4%, 0.6%-4.2%) and patient-reported outcomes (1.0%, 0.9%-1.5%). No clear pattern could be identified in the degree of between-hospital variation by disease category. Furthermore, the studies exhibited limited attention to the reliability of observed differences in indicator scores.

CONCLUSION

Hospital-level variation in quality indicator scores is generally small relative to residual variation. However, meaningful variation between hospitals does exist for multiple indicators, especially for care processes which can be directly influenced by hospital policy. Quality improvement strategies are likely to generate more impact if preceded by level-specific and indicator-specific analyses of variation, and when absolute variation is also considered.

PROSPERO REGISTRATION NUMBER

CRD42022315850.

What does the MAIHDA method explain?

Multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA) is a new approach to quantitative intersectional modelling. Along with an outcome of interest, MAIHDA entails the use of two sets of independent variables. These include group demographics such as race, gender, and poverty status as well as strata which are constructs such as Black female poor, Black female wealthy, and White female poor. These constructs represent the combination of the demographic variables. To operationalize the approach, an initial random intercepts model with strata as a level 2 context is specified. Then, another model is specified that includes the strata as well as the demographic variables as level 1 fixed effects. As such, it is argued that MAIHDA uniquely identifies the additive and intersectional effects for any given outcome. In this paper we show that MAIHDA falls short of this promise: the strata are an individual-level composite variable not a level 2 context. Rather than being analogous to neighborhoods as contexts, strata are analogous to socio-economic status which is a combination of individual-level demographic variables, albeit often presented as a group-level characteristic. The result is that the demographic variables are inserted in both level 2 and 1. This duplication across the levels in MAIHDA means that there is a built-in collinearity across the levels and that the models are mis-specified and, therefore, redundant. We conclude that single-level models with the demographic variables and interactions or with the strata as fixed effects are still the more accurate models for quantitative intersectional analyses.

Mapping socio-geographical disparities in the occurrence of teenage maternity in Colombia using multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA)

BACKGROUND

The prevalence of teenage pregnancy in Colombia is higher than the worldwide average. The identification of socio-geographical disparities might help to prioritize public health interventions.

AIM

To describe variation in the probability of teenage maternity across geopolitical departments and socio-geographical intersectional strata in Colombia.

METHODS

A cross-sectional study based on live birth certificates in Colombia. Teenage maternity was defined as a woman giving birth aged 19 or younger. Multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA) was applied using multilevel Poisson and logistic regression. Two different approaches were used: (1) intersectional: using strata defined by the combination of health insurance, region, area of residency, and ethnicity as the second level (2) geographical: using geopolitical departments as the second level. Null, partial, and full models were obtained. General contextual effect (GCE) based on the variance partition coefficient (VPC) was considered as the measure of disparity. Proportional change in variance (PCV) was used to identify the contribution of each variable to the between-strata variation and to identify whether this variation, if any, was due to additive or interaction effects. Residuals were used to identify strata with potential higher-order interactions.

RESULTS

The prevalence of teenage mothers in Colombia was 18.30% (95% CI 18.20-18.40). The highest prevalence was observed in Vichada, 25.65% (95% CI: 23.71-27.78), and in the stratum containing mothers with Subsidized/Unaffiliated healthcare insurance, Mestizo, Rural area in the Caribbean region, 29.08% (95% CI 28.55-29.61). The VPC from the null model was 1.70% and 9.16% using the geographical and socio-geographical intersectional approaches, respectively. The higher PCV for the intersectional model was attributed to health insurance. Positive and negative interactions of effects were observed.

CONCLUSION

Disparities were observed between intersectional socio-geographical strata but not between geo-political departments. Our results indicate that if resources for prevention are limited, using an intersectional socio-geographical approach would be more effective than focusing on geopolitical departments especially when focusing resources on those groups which show the highest prevalence. MAIHDA could potentially be applied to many other health outcomes where resource decisions must be made.

Using geospatial trajectories to explore how the COVID-19 pandemic affects the associations between environmental attributes and runnability of park trails

The worldwide coronavirus disease 2019 (COVID-19) pandemic and the associated social distancing measures have produced alterations in park visits of individuals, as well as their park-based physical activity (e.g. running exercise). Although studies on the influence of the COVID-19 pandemic on changes in running activity patterns are becoming an emerging focus, less is known about how these changes are related to the environmental attributes of parks before and after the pandemic, the knowledge of which is essential to planning green infrastructure that better supports physical activities. Therefore, we employed a volunteered geographic information approach to investigate the runnability of park trails in Shenzhen, utilizing self-tracking routes from Strava, in order to uncover the associations between trail characteristics and park features with the running intensity before and after the pandemic. Multilevel regression model analyses revealed that trail network connectivity was the only environmental attribute indicating consistent and positive associations with running intensity. Blue space density was positively correlated with running intensity in urban parks but indicated no significant association in forest parks before the pandemic. In the pre-pandemic era, population density was positively related to running intensity in urban and forest parks. However, after the pandemic, the associations between running behaviours and population density remained positive in forest parks but turned insignificant in urban parks. The outbreak of the pandemic also altered the influence of other park features (e.g. park shape and trail density) on running intensity. The evidence-based knowledge provides planners with significant insights into pandemic-resilient park planning for the post-COVID era.

Geographical and sociodemographic differences in statin dispensation after acute myocardial infarction in Sweden: a register-based prospective cohort study applying analysis of individual heterogeneity and discriminatory accuracy (AIHDA) for basic comparisons of healthcare quality

BACKGROUND

In Sweden, as in many other countries, official monitoring of healthcare quality is mostly focused on geographical disparities in relation to a desirable benchmark. However, current evaluations could be improved by considering: (1) The intersection of other relevant axes of inequity like age, sex, income and migration status; and (2) The existence of individual heterogeneity around averages. Therefore, using an established quality indicator (ie, dispensation of statins after acute myocardial infarction, AMI), we valuate both geographical and sociodemographic inequalities and illustrate how the analysis of individual heterogeneity and discriminatory accuracy (AIHDA) enhances such evaluations.

POPULATION AND METHODS

We applied AIHDA and calculated the area under the receiver operating characteristics curve (AUC) of regional and sociodemographic differences in the statin dispensations of 35 044 patients from 21 Swedish regions and 24 sociodemographic strata who were discharged from the hospital with an AMI diagnosis between January 2011 and December 2013. Following the Swedish National Board of Health and Welfare, we used a benchmark value of 90%.

RESULTS

Dispensation of stains after AMI in Sweden did not reach the desired target of 90%. Regional differences were absent/very small (AUC=0.537) while sociodemographic differences were small (AUC=0.618). Women, especially those with immigrant background and older than 65 years, have the lowest proportions of statin dispensations after AMI.

CONCLUSIONS

As the AUC statistics are small, interventions trying to achieve the benchmark value should be universal. However, special emphasis should nevertheless be directed towards women, especially older women with immigrant backgrounds.

Neighbourhood deprivation and intersectional inequalities in biomarkers of healthy ageing in England

While social and spatial determinants of biomarkers have been reported, no previous study has examined both together within an intersectional perspective. We present a novel extension of quantitative intersectional analyses using cross-classified multilevel models to explore how intersectional positions and neighbourhood deprivation are associated with biomarkers, using baseline UK Biobank data (collected from 2006 to 2010). Our results suggest intersectional inequalities in biomarkers of healthy ageing are mostly established by age 40-49, but different intersections show different relationships with deprivation. Our study suggests that certain biosocial pathways are more strongly implicated in how neighbourhoods and intersectional positions affect healthy ageing than others.