Trial emulation and survival analysis for disease incidence registers: A case study on the causal effect of pre‐emptive kidney transplantation

Abstract

When drawing causal inference from observed data, failure time outcomes present additional challenges of censoring often combined with other missing data patterns. In this article, we follow incident cases of end-stage renal disease to examine the effect on all-cause mortality of starting treatment with transplant, so-called pre-emptive kidney transplantation, vs starting with dialysis possibly followed by delayed transplantation. The question is relatively simple: which start-off treatment is expected to bring the best survival for a target population? To address it, we emulate a target trial drawing on the long term Swedish Renal Registry, where a growing common set of baseline covariates was measured nationwide. Several lessons are learned which pertain to long term disease registers more generally. With characteristics of cases and versions of treatment evolving over time, informative censoring is already introduced in unadjusted Kaplan-Meier curves. This leads to misrepresented survival chances in observed treatment groups. The resulting biased treatment association may be aggravated upon implementing IPW for treatment. Aware of additional challenges, we further recall how similar studies to date have selected patients into treatment groups based on events occurring post treatment initiation. Our study reveals the dramatic impact of resulting immortal time bias combined with other typical features of long-term incident disease registers, including missing covariates during the early phases of the register. We discuss feasible ways of accommodating these features when targeting relevant estimands, and demonstrate how more than one causal question can be answered relying on the no unmeasured baseline confounders assumption.

Causal inference in survival analysis under deterministic missingness of confounders in register data

“…We would then conclude that the survival probability difference between receiving a PKT and receiving dialysis is 0.232 (95%CI: 0.204-0.261) at t 0 = 5 and 0.273 (95%CI: 0.238-0.308) at t 0 = 10. These results are similar to those obtained by Olarte Parra et al 6…”

“…For the sake of clarity, we will ignore it in the further developments. For more details on this data set, we refer to Olarte Parra et al 6 Due to changing transplant eligibility criteria over time, the subpopulation consisting of patients who received a PKT exhibits slight changes over time. In particular, PKT patients who entered the register at earlier times tended to be younger and healthier, with fewer comorbidities (see Figure 1).…”

“…For the sake of clarity, we will ignore it in the further developments. For more details on this data set, we refer to Olarte Parra et al 6 …”

“…Of these patients, 5142 entered the register before 1998 (26%), and have missing comorbidities information, 1097 (0.06%) received a PKT and 12 073 (62%) died during the study period. For a more detailed description of the data set, we refer to Olarte Parra et al 6 An important difference between the simulated data sets presented in the previous section and the real data set is that in the real data set, patients who start dialysis may still receive a transplant at a later date. This simplification in the simulation was done for clarity in presentation, and may result in different survival patterns in the real data than those in the simulation setting, especially in the dialysis arm, as we explain below.…”

Reporting of Observational Studies Explicitly Aiming to Emulate Randomized Trials

“…We retrieved 3133 unique records, of which 200 were included in the review (Figure 1). All reasons for excluding records after full-text review are given in eAppendix 6 in Supplement 1.…”