4966 articles were identified from databases and five articles were screened from the reference lists. After removing duplicates and screening titles and abstracts, 20 studies were chosen for the full-text review. Ultimately, 13 articles were included in this systematic review [6, 21, 22,23,24,25,26,27,28,29,30,31,32]. Seven studies were excluded for the following reasons: one study was a systematic review [16], four studies were conference abstracts [33,34,35,36], in one study comparators were HD first and HD to PD group [37], and one study had a small sample size without comparators [38]. The detailed information relevant to the search results is presented in a study flow chart (See Fig. 1).

Fig. 1Characteristics of patients in the included studies

The characteristics of patients including sample size, mean age, BMI, time on HD, diabetes as a cause of ESRD, creatinine clearance, urine output, haemoglobin, albumin, and the time of follow-up are shown in Table 1. Of the 13 included studies, 12 studies are cohort studies, and one study is a case-control study [31]. The sample sizes of the PD first group and HD to PD group range from 37 to 9404 and 28 to 3757 patients, respectively, and include patients from Argentina, Canada, China, Turkey, Poland, France, Iran, Australia, New Zealand, and America. Additionally, the duration of HD varied from less than a month to 286 months, and the follow-up time varied from one month to 132 months. Each group of studies had varying degrees of missing data that is summarised in Table 1.

Table 1 The characteristics of patients in included studiesThe clinical outcomes of included articles

The summary of clinical outcomes (dropout events, median time on PD, technique survival rates, the reasons for PD dropout, mortality rates, transplantation rates, and technique failure (TF) definitions) are presented in Table 2. Reasons for HD switch to PD are reported in Table 3. Detailed explanations of the tables are as follows.

Table 2 The rate and reasons for patient’s dropout in PDTable 3 Reasons for HD switch to PDTechnique failure

TF was defined in six studies [6, 22,23,24, 28, 31]. TF definitions included a transfer to HD or death [22, 24]; a transfer to HD [6, 23, 28]; and a transfer to HD, transplant, or death [31]. The different definitions of TF are also displayed in Table 2. Based on all definitions of TF, the rate of technique survival at one year varied from 84% to 100. in the PD first group [22–23], and from 74 to 94% in HD to PD group [22, 31]. Four studies found that the PD first group had a better technique survival than the HD to PD group [6, 21, 26, 32], however, only two studies demonstrated a statistically significant difference between the two groups [6, 21]. According to our definition of PD dropout (transfer to HD), four studies were included in the meta-analysis [6, 25, 27, 28]. The meta-analysis found that there was a lower rate of PD dropout in the PD first group, but that there was no statistically significant difference in PD dropout between the two groups (OR: 0.81; 95%CI: 0.61, 1.09; I2 = 83%; P = 0.16) (Fig. 2).

Fig. 2

Forest plot of PD dropout for PD first group vs. HD to PD group

The median time on PD

Half studies reported the median time on PD, nevertheless, only three of them outlined the varying durations of PD in two distinct groups [6, 22, 31]. In comparison to patients switching from HD (from 36months to 51months), those who were PD first showed a longer duration on PD (from 48months to 57months) [6, 22].

The reasons for HD switch to PD

There are seven studies that illustrate the factors leading to the switch from HD to PD [21, 22, 25, 28, 29, 31, 32]. The primary causes for this switch include vascular access failure (ranging from 7 to 70%), patients’ preferences or social issues (ranging from 11 to 64%), and cardiovascular disease (varying from 1 to 45%).

The reasons for PD dropout

Eight studies reported reasons for PD dropout, however, only three studies presented the reasons and proportion of patients withdrawing from PD in the two groups separately [6, 27, 28]. Inadequate dialysis and peritonitis were the primary reasons for PD dropout in both groups. However, only one study found statistically significant differences in these two reasons between the two groups [6]. Furthermore, Liberek et al. [21] found a correlation between peritonitis and TF only in anuric patients. They discovered that anuric patients transferred from HD to PD have a higher rate of peritonitis than PD first group, resulting in more TF. Reasons for PD dropout also included catheter dysfunction or mechanical problems, ultra-filtration failure, malnutrition, cardiovascular events, patient preference, membrane failure, exit site and tunnel infections [27, 28].

Psychosocial reasons associated with PD dropout were mentioned in three studies. Lobbedez et al. [27] found a higher rate of patients dropping out of PD because of psychosocial reasons in the HD to PD group than PD first group (16% vs. 11%). This was explained by the authors as potentially being due to insecurity and anxiety associated with the change in dialysis modality [26, 27]. They also reported that psychosocial reasons impact the duration of PD for patients transferred from HD. Chidambaram et al. [23] found that patients have 31% higher rate of TF when their physicians are male gender. Suggested reasons for this were more patient-centred communication among female physicians. Dong et al. [24] reported a correlation between the type of medical insurance and the risk of PD dropout in China. They discovered patients in peritoneal dialysis had better technique outcomes whose medical insurance (employee medical insurance) had more reimbursement and wider coverage for the treatment. However, no study reported statistically significant differences between the groups.

Mortality and transplantation rates

Six studies reported the death rate in each group, and four studies found the HD to PD group had a higher rate of mortality than PD first group that was statistically significant (Table 2) [6, 25, 29, 32]. Four studies reporting the detailed number of deaths were included in the meta-analysis [25, 28, 31, 32]. The meta-analysis demonstrated that patients have a lower mortality rate in the PD first group compared to the HD to PD group (OR: 0.48; 95% CI: 0.25, 0.92; I2 = 73%; P = 0.03 (Fig. 3)). Furthermore, two studies indicated the number of patients who withdraw from PD for kidney transplantation in the two groups but only one reported a statistically significant difference between the groups (Table 2) [25, 28].

Fig. 3

Forest plot of mortality for PD first group vs. HD to PD group

Quality of included studies and meta-analysis

Two studies did not report the time of follow-up [22, 26]. Three studies did not present confidence intervals [22, 28, 30]. One study lacked detailed information in the methods [32]. No study performed adequate matching for confounding factors. No study included a blinded collection of outcome information.

The quality level of the meta-analysis associated with PD dropout was very low due to statistical heterogeneity and wide confidence intervals. The estimates are therefore very uncertain. The meta-analysis of mortality rate had a moderate quality level, indicating the true effects are likely to be close to the estimated effects, but that further research may impact this estimate (Fig. 4).

Fig. 4

The evidence certainty of meta-analysis