This study shows reduced health care utilization for patients choosing HHD as their initial RRT compared with patients on IHD and PD. Hospital admission rate was 65 % higher for patients on IHD and 33 % higher for patients on PD compared with HHD. The number of days in hospital was 68% higher for PD and 18 % higher for IHD compared with HHD. Time to first admission was longer for patients with HHD, 0.7 years as compared with 0.4 years for PD and 0.3 years for IHD. Finally, technique survival was better for HHD than PD.
The advantage of HHD for all-cause admission rates compared with IHD, in the present study, are in line with a Swiss study which included patients commencing RRT between 1970 and1995(6). Of note is that these differences between HHD and IHD were not found in studies from the US, utilizing data mainly from the 21st century (13, 15). The only previous studies comparing HHD and PD were also conducted in the US during the 21st century and did not show an advantage for HHD in analyses restricted to incident patients (15) (5) (14). The selection of patients to a dialysis modality might differ between Europe and the US with disparities in comorbidities, age and socioeconomic factors, the organisation of health care praxis and variations in the prescriptions of dialysis, all contributing to the observed discrepancies.
The European part of the DOPPS study reported a hospital admission rate of around one for IHD between 1998 to 2000 compared with 1.7 to 2.8 in our study for all dialysis modalities(19). The USRDS reported an annual admission rate for all dialysis patients of 2.1 during 2005 which decreased to 1.7 for IHD and 1.6 for PD during 2014. A number of recent studies from the US, comparing HHD with IHD or PD during the 21st century, reported admission rates ranging from 0.7-1.8 for HHD; 1.1-1.7 for IHD and 0.7-1.9 for PD, all of which were lower than those found in the present study (5, 13-15). These differences prevail for the number of days of hospital care per year, with lower numbers in other studies.
Thus, our results differ compared with other studies as well as in comparison with the European DOPPS and the USRDS, irrespective of dialysis modality, both in regard to differences in actual number of hospital admissions per year and number of days in hospital. In the present study, patients were included from 1991, which is earlier than any other study. During 1991 to 2000 the overall prognosis for dialysis patients was worse, both in Sweden and internationally, than from 2000 onwards(1, 2). Moreover, during the nineties, Sweden had more hospital beds per capita than the US and the other European countries in the DOPPS study(20), which might have resulted in more frequent and longer admissions. There might also be methodological differences, as in the Swedish Inpatient Registry a new admission might be generated each time a patient is transferred to a new in-patient department during the same hospital stay.
Cardiovascular disease and infections are the most common causes of death in patients on dialysis (1) . In our study, patients on HHD had significantly fewer admissions with a cardiovascular diagnosis compared with IHD patients. This is in line with other studies, which reported advantages for HHD regarding admissions with a cardiovascular diagnosis in comparison to both prevalent and incident IHD(5, 15). Others have shown an advantage when comparing HHD with prevalent, but not with incident PD patients(5, 15). Regarding admissions with infections, we registered significantly fewer admissions for HHD patients compared with PD. Earlier studies have, in accordance with our results, reported an advantage for HHD compared with PD, but contrary to our results, a disadvantage for HHD compared with IHD (5, 13).
Although patients in this study were matched for comorbidity at start of RRT, progress of and subsequent development of comorbidity, was probably lower in patients with HHD, and most likely contributed to the lower utilization of health care. This is supported by a better survival for HHD patients, which has previously been reported both by our group and others (10, 12). Several studies have shown that the higher dialysis doses(22, 24, 25)and extensive patient education, (26, 28), which are associated with HHD, are related to better fluid balance and phosphate control, both important factors in the development of cardiovascular morbidity(31) (29). In the present study there was no significant advantage for HHD concerning hospital admission rates with a cardiovascular diagnosis in the intention to treat analyses compared with IHD or PD. A weakened impact of the initial RRT after renal transplantation, could explain this absence of a significant carry over effect on the admission rates. In the present study, admission rates with cardiovascular diagnoses were low, 0.02-0.06 per year, compared with 0.36-0.48 with follow up according to intention to treat in the studies by Weinhandl(5, 13). The patients in his studies were somewhat older, had higher proportions of diabetes as renal diagnoses and were not incident to RRT, factors that could well explain some of the difference in morbidity due to cardiovascular disease.
Another cause for this discrepancy might be due to how cardiovascular events are registered in the Swedish Inpatient Registry. The organization of the Swedish Inpatient Registry makes it impossible to discriminate between a cardiovascular event occurring during a hospital admission and a chronic cardiovascular comorbidity, which the patient had prior to admission and which has no direct impact on the cause of admission. For a chronic cardiovascular comorbidity to have an impact on the cause of hospital admission, it must be assigned the position of principal diagnosis. Thus, most cardiovascular ICD codes. registered as secondary diagnoses, could not be used when classifying cardiovascular admissions, which most probably results in an underestimation of the number of admissions with a cardiovascular disease in the present study.
A possible explanation for the lower admission rate in HHD as compared with PD could be related to the resilience of the treatment modality. In accordance with earlier studies, the technique survival, was superior for HHD compared with PD(5, 16, 32). In the present study, the two years technique survival was 93 % for HHD and 64 % for PD. Other studies from the US, Australia, New Zealand and Europe, have reported a two-year technique survival ranging from 75 to 96 % for incident HHD patients and 64 to 74 % for incident PD patients. Some of the differences between studies are related to methods and dialysis prescriptions. In one study from Canada, reporting a higher HHD technique survival, all patients had nocturnal HHD and some patients were completely dependent on caregivers for HHD treatment. In Sweden HHD is always self-care and administered by the patient in their own home. Possibly the setup with HHD administered by caregivers enabled a longer technique survival(33). In another study, in which all patients used a single low-dialysate flow dialysis device, the reported HHD technique survival was lower compared with our study(5).
There are limitations to the present study, mainly due to the retrospective design. Despite strict matching and statistical adjustment for renal diagnoses, there is still a risk of differences between groups, especially concerning socioeconomic factors and smoking. However, health care in Sweden is publicly funded and the praxis and access to different RRT are relatively homogenous for all citizens. It is important to point out, that the results from the present study cannot necessarily be extrapolated to patients with older age, more comorbidities, different socioeconomic status or who, for whatever reason, are unable to be compliant to an independent home-based dialysis regime or societies with other healthcare structures. However, our results showing a lower admission rate for cardiovascular disease in patients on HHD compared with IHD and for infectious diseases in HHD compared with PD strongly suggest an effect of the modality rather than only from patient selection. A modality effect is further supported by our finding that these differences decrease after transplantation as seen in the intention to treat analysis. In a previous single centre report, we have shown that patients on HHD have a better control of fluid balance and hypertension compared with IHD, which could contribute to the lower tendency for cardiovascular admissions(9).
This study also has important merits. The SRR contains data on all patients in RRT and is updated when patients change treatment modality. All renal units in Sweden report to the SRR. Moreover, it is compulsory for all the hospitals in the country to report to the Swedish in-patient registry. Recently, the accuracy of the reported diagnoses has been validated(34). Thus, the close to complete coverage of these registries enabled us to include virtually all Swedish patients starting HHD as initial RRT. This study also adds important knowledge compared with other recent studies,
that are solely from the US, as it adds a European perspective and reflects effects of the different dialysis modalities in an entire and homogenous population with long-term follow up.