Delay in commencement of palliative care service episodes provided to Indigenous and non-Indigenous patients: cross-sectional analysis of an Australian multi-jurisdictional dataset

Woods, John A.; Johnson, Claire E.; Ngo, Hanh T.; Katzenellenbogen, Judith M.; Murray, Kevin; Thompson, Sandra C.

doi:10.1186/s12904-018-0380-7

Table 2 Association between being identified as Indigenous and delay in commencement of palliative care episodes – sensitivity analysis comparing estimates based on complete case and multiple imputation data

From: Delay in commencement of palliative care service episodes provided to Indigenous and non-Indigenous patients: cross-sectional analysis of an Australian multi-jurisdictional dataset

Model	Indicative N^a	Complete case analyses^b			Multiple imputation analyses^b
Model	Indicative N^a	OR^c	95% CI^d	P value	OR^c	95% CI^d	P value
(a) Crude (univariate) estimates
Total sample	73,657/84,238	0.99	0.72–1.36	0.96	1.07	0.80–1.42	0.66
(b) Adjusted estimates, main effects^e plus interaction terms^f
Total sample	73,657/84,238	1.32	0.98–1.77	0.07	1.41	1.07–1.86	0.02
Restriction by setting (adjusted model)^g
Hospital	41,283/47,803	1.35	0.80–2.25	0.26	1.49	0.96–2.32	0.08
Community	32,374/36,435	1.32	0.91–1.91	0.15	1.29	0.91–1.85	0.16
Restriction by episode sequence (adjusted model)^g
First episode of care in PCOC records	50,229/57,325	1.42	1.02–1.96	0.04	1.53	1.14–2.06	0.005
Second or subsequent episode of care in PCOC records	23,428/26,913	1.03	0.55–1.94	0.93	1.08	0.61–1.90	0.79
Restriction by age group (adjusted model)^g
< 65 years at episode start	19,737/~ 22,767	1.52	0.98–2.36	0.06	1.66	1.14–2.41	0.009
≥ 65 years at episode start	53,920/~ 61,471	1.14	0.75–1.74	0.54	1.09	0.71–1.67	0.69
Restriction by remoteness of residence (adjusted model)^g
Major city	55,592/~ 63,850	1.09	0.72–1.65	0.70	1.16	0.77–1.73	0.48
Inner Regional	13,514/~ 15,069	2.15	1.24–3.73	0.005	1.97	1.19–3.28	0.009
Outer Regional/Remote/Very Remote	4551/~ 5319	1.18	0.57–2.44	0.66	1.72	0.97–3.05	0.06
Restriction by calendar period (adjusted model)^g
Jul 2013 – Jun 2014	36,850/42,416	1.28	0.84–1.95	0.24	1.38	0.94–2.02	0.10
Jul 2014 – Jun 2015	36,807/41,822	1.34	0.89–2.01	0.16	1.40	0.95–2.06	0.09
Restriction by principal diagnosis category (adjusted model)^g
Cancer	59,455/~ 67,680	1.34	0.97–1.85	0.08	1.41	1.04–1.91	0.03
Other	14,202/~ 16,558	1.23	0.60–2.50	0.58	1.42	0.78–2.59	0.25

^aNumbers shown are those from Complete Case analyses/Multiple imputation analyses. For the multiple imputation analyses, the number displayed for total sample is the actual number of episodes in the analyses; for stratified subgroups of variables with missing values, the distribution of numbers of episodes across subgroups varies slightly between imputed datasets because of the imputation process. In such cases (marked ~), indicative numbers are shown from dataset #1 of m = 15 imputed datasets. (There were zero missing values for setting, episode sequence and calendar period [Table 1])
^bAll estimates based on logistic regression models accounting for unbalanced clustering of episodes among individual patients
^cOR: Odds ratio for episodes in Indigenous versus non-Indigenous patients
^dCI: Confidence interval
^eCovariates in multiple logistic regression: age in years (continuous), sex, principal diagnosis category (cancer/other), episode setting (community/ hospital [overnight & same-day admissions plus outpatients]), remoteness of residence at baseline, social disadvantage of area of residence at baseline, episode of care sequence (based on lookback to inception of service’s participation in PCOC collection), jurisdiction, calendar month, clinical status at episode start (four domains of PCPSS modelled separately), functional status at episode start (RUG-ADL total score modelled by tertile). Data breakdown by individual jurisdiction not presented
^fInteractions included in analyses of the total sample were: settingXsequence, calendar-periodXsetting, jurisdictionXsequence, jurisdictionXsetting, jurisdictionXseifa, sequenceXremoteness, seifaXremoteness. An applicable subset of these interactions was selected for each restricted analysis
^gSubgroup estimates are adjusted for all potential confounding factors (listed in [e,f] above) other than the covariate (main effect and interaction terms if applicable) by which the analysis is restricted. In the analyses restricted by age group, age in years as a continuous variable remained in the model

Back to article page

ISSN: 1472-684X

Contact us

Submission enquiries: bmcpalliativecare@biomedcentral.com
General enquiries: ORSupport@springernature.com

BMC Palliative Care

Contact us