Performance of missing data approaches under nonignorable missing data conditions

Pohl, Steffi; Becker, Benjamin J.

doi:10.5964/meth.2805

Cited by 16 publications

(26 citation statements)

References 18 publications

(48 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Costs are based on activity and researchers must determine reasons for low or no activity. Where there was a priori knowledge of missing dialysis attendance for DxMoC4 and 5 (nonignorable missing values [ 53 ]) we made every effort to gather this information from other sources and link the data sets. We did not interpolate data nor impute costs as we could not be certain activity had occurred.…”

Section: Limitationsmentioning

confidence: 99%

Do remote dialysis services really cost more? An economic analysis of hospital and dialysis modality costs associated with dialysis services in urban, rural and remote settings

Gorham

Howard

Cunningham

et al. 2021

BMC Health Serv Res

View full text Add to dashboard Cite

Background Rates of end-stage kidney disease in Australia are highest in the Northern Territory (NT), with the burden of disease heaviest in remote areas. However, the high cost of delivering dialysis services in remote areas has resulted in centralisation, requiring many people to relocate for treatment. Patients argue that treatment closer to home improves health outcomes and reduces downstream healthcare use. Existing dialysis cost studies have not compared total health care costs associated with treatment in different locations. Objective To estimate and compare, from a payer perspective, the observed health service costs (all cause hospital admissions, emergency department presentations and maintenance dialysis) associated with different dialysis models in urban, rural and remote locations. Methods Using cost weights attributed to diagnostic codes in the NT Department of Health’s hospital admission data set (2008–2014), we calculated the mean (SD) total annual health service costs by dialysis model for 995 dialysis patients. Generalized linear modeling with bootstrapping tested the marginal cost differences between different explanatory variables to estimate ‘best casemix’/‘worst casemix’ cost scenarios. Results The mean annual patient hospital expenditure was highest for urban models at $97 928 (SD $21 261) and $43 440 (SD $5 048) and lowest for remote at $19 584 (SD $4 394). When combined with the observed maintenance dialysis costs, expenditure was the highest for urban models at $148 510 (SD $19 774). The incremental cost increase of dialysing in an urban area, compared with a rural area, for a relocated person from a remote area, was $5 648 more and increased further for those from remote and very remote areas to $10 785 and $15 118 respectively. Conclusions This study demonstrates that dialysis treatment in urban areas for relocated people has health and cost implications that maybe greater than the cost of remote service delivery. The study emphasises the importance of considering all health service costs and cost consequences of service delivery models. Key points for decision makers Relocation for dialysis treatment has serious health and economic consequences. Relocated people have low dialysis attendance and high hospital costs in urban areas. While remote dialysis service models are more expensive than urban models, the comparative cost differences are significantly reduced when all health service costs are included. The delivery of equitable and accessible dialysis service models requires a holistic approach that incorporates the needs of the patient; hence dialysis cost studies must consider the full range of cost impacts beyond the dialysis treatments alone.

show abstract

Section: Limitationsmentioning

confidence: 99%

Do remote dialysis services really cost more? An economic analysis of hospital and dialysis modality costs associated with dialysis services in urban, rural and remote settings

Gorham

Howard

Cunningham

et al. 2021

BMC Health Serv Res

View full text Add to dashboard Cite

show abstract

“…The use of sufficiently complex imputation models, such as the Gaussian copula model (Hollenbach et al, 2018), mixture models (Murray and Reiter, 2016), or latent class models (Vermunt et al, 2008;Si and Reiter, 2013) are advantageous to minimize possible distributional misspecifications for MAR data. Appropriate imputation models can also treat specific deviations from MAR (missing not at random; MNAR; Harel and Schafer, 2009;Jung et al, 2011;Kano and Takai, 2011;Zhang and Reiser, 2015;Bartolucci et al, 2018;Kuha et al, 2018;Pohl and Becker, 2020).…”

Section: Discussionmentioning

confidence: 99%

Why Ordinal Variables Can (Almost) Always Be Treated as Continuous Variables: Clarifying Assumptions of Robust Continuous and Ordinal Factor Analysis Estimation Methods

Robitzsch

2020

Front. Educ.

189

View full text Add to dashboard Cite

“…Concerning the missing-data-generating mechanism, we assumed a linear relationship between the proficiency to be measured and the omission propensity. Note that nonignorability may also occur when this relationship is not linear or when item omission depends directly on the true item response (Pohl & Becker, in press; Robitzsch, 2016). A special case of the latter missing data mechanism can lead to the extreme situation in which missing values occur only on items that would otherwise be incorrect.…”

Section: Discussionmentioning

confidence: 99%

When Nonresponse Mechanisms Change: Effects on Trends and Group Comparisons in International Large-Scale Assessments

Sachse

Mahler

Pohl

2019

Educational and Psychological Measurement

Self Cite

View full text Add to dashboard Cite

Mechanisms causing item nonresponses in large-scale assessments are often said to be nonignorable. Parameter estimates can be biased if nonignorable missing data mechanisms are not adequately modeled. In trend analyses, it is plausible for the missing data mechanism and the percentage of missing values to change over time. In this article, we investigated (a) the extent to which the missing data mechanism and the percentage of missing values changed over time in real large-scale assessment data, (b) how different approaches for dealing with missing data performed under such conditions, and (c) the practical implications for trend estimates. These issues are highly relevant because the conclusions hold for all kinds of group mean differences in large-scale assessments. In a reanalysis of PISA (Programme for International Student Assessment) data from 35 OECD countries, we found that missing data mechanisms and numbers of missing values varied considerably across time points, countries, and domains. In a simulation study, we generated data in which we allowed the missing data mechanism and the amount of missing data to change over time. We showed that the trend estimates were biased if differences in the missing-data mechanisms were not taken into account, in our case, when omissions were scored as wrong, when omissions were ignored, or when model-based approaches assuming a constant missing data mechanism over time were used. The results suggest that the most accurate estimates can be obtained from the application of multiple group models for nonignorable missing values when the amounts of missing data and the missing data mechanisms changed over time. In an empirical example, we furthermore showed that the large decline in PISA reading literacy in Ireland in 2009 was reduced when we estimated trends using missing data treatments that accounted for changes in missing data mechanisms.

show abstract

Performance of missing data approaches under nonignorable missing data conditions

Cited by 16 publications

References 18 publications

Do remote dialysis services really cost more? An economic analysis of hospital and dialysis modality costs associated with dialysis services in urban, rural and remote settings

Do remote dialysis services really cost more? An economic analysis of hospital and dialysis modality costs associated with dialysis services in urban, rural and remote settings

Why Ordinal Variables Can (Almost) Always Be Treated as Continuous Variables: Clarifying Assumptions of Robust Continuous and Ordinal Factor Analysis Estimation Methods

When Nonresponse Mechanisms Change: Effects on Trends and Group Comparisons in International Large-Scale Assessments

Contact Info

Product

Resources

About