Hospital readmission is broadly defined as an episode in which a patient discharged from one hospital is readmitted to another hospital within a specified time interval. The initial hospitalization is called a “baseline admission” and the subsequent hospitalization is called a “readmission” [1] . Different timeframes of hospitalization are used, the most common of which are hospital readmissions at 30 days (HR30), 90 days (HR 90) and one year after the initial hospitalization. Readmissions can be planned, i.e. scheduled during or shortly after the index hospitalization. However, they are most often unplanned and, therefore, represent a significant problem for health systems.

Hospital readmissions are encountered in both medical and surgical services with high frequencies of 21.1% and 15.6%, respectively, for HR30 according to Kripalani and Theobald in the USA [2] . They are associated with poor patient quality of life and high hospital costs, with approximately $17.4 billion in annual Medicare expenditures alone, with 19.6% of patients being readmitted within 30 days of discharge and 34% when looking at 90-day hospitalizations [2] - [4] . In Canada, one in 11 patients is readmitted within 30 days of discharge, with an estimated annual expenditure of $2.1 billion [5] .

Factors such as age and chronic disease tend to increase hospital readmissions. For example, patients with chronic kidney disease, especially those with end-stage renal disease undergoing iterative hemodialysis, face a higher readmission rate compared to the general population and exceptionally high hospitalization costs [6] . In 2013, 34.9% of hemodialysis patients in Canada were readmitted within 30 days of discharge according to MATHEW [4] , and 70.1% of these readmissions were preventable [7] - [9] .

In sub-Saharan Africa, the situation may seem more critical due to poverty, inadequate technical facilities, poor health structures and the increasing frequency of chronic diseases. However, very little work has been done on hospital readmissions of patients with kidney disease.

The present study aims to describe the profile of readmitted patients and to identify the risk factors for readmission to the nephrology setting in our practice context.

We collected 1142 patients hospitalized during the study period. During this period 154 patients were readmitted, a prevalence of 13.48%. We excluded 41 patients for incomplete records such as epidemiologic, administrative and medical information about the patient.

Our study therefore involved 113 patients ( Figure 1 ).

Readmission was verified based on the patient’s ID and whether he or she had already been hospitalized and discharged from the department at least once.

We included 65 male patients (57.52%), a sex ratio of 1.35. The average age of our patients was 42.31 ± 13.62 years, with extremes of 15 and 77 years. The most affected age group was 35 - 45 years (33.62%). The majority of our patients were unemployed (24.78%) or worked in the informal sector as traders or artisans in 17.78% and 14.16% of cases respectively. Housewives represented 11.5% of patients.

The patients were known to be hypertensive in 63.72% of cases, HIV-infected in 15.04% and diabetic in 11.50%.

The average time to readmission was 28.23 ± 21.83 with extremes of 3 and 123 days and this readmission was early in 65.48%. The average length of stay was 11.41 ± 8.36 with extremes of 1 and 45 days for the last hospitalization. Patients were admitted for the second time in 88.49% of cases ( Table 1 ).

The main reason for readmission was uremic syndrome (68.14%).

The main clinical signs at the last hospitalization were hypertensive crisis (83.19%), conjunctival pallor (97.35%), and acute pulmonary oedema (38.94%).

Kidney disease was stage 5 in 95.6% of the cases at the last hospitalization. The hemoglobin level was below 8g/dl in 82.3% of cases at the time of the last hospitalization. Anemia was normocytic normochromic in 63.39% and microcytic hypochromic in 36.61% of the cases at the time of the last hospitalization. Thrombocytopenia was observed in 43.36% of cases at the time of the last hospitalization.

The diagnosis chosen was dialysis failure 82.30% followed by dialysis catheter infection 10.6%. Prior to readmission 97.35% of patients were treated with iterative dialysis, while 2.65% had stabilised renal function.

The main treatments received were dialysis (93.61%) of cases, blood transfusion (91.15%), antihypertensive (74.34%) and antibiotic therapy (61.85%).

Mortality was 31%, and in 69% of cases our patients were on iterative dialysis.

In univariate analysis, we looked for risk factors for death and factors associated with the risk of early readmission. Thus, creatinine levels were on average higher in the deceased group compared to the non deceased group with a statistically significant difference (p = 0.049). Similarly, hemoglobin levels were on average higher in the deceased group compared to the non deceased group with a statistically significant difference (p = 0.037). In addition, the proportion of patients with hypocalcemia was statistically significantly higher in the deceased group (p = 0.03). The same was true for the proportion of patients with obstructive nephropathy (p = 0.029) ( Table 2 ).

We also compared patients with early readmission to those with non-early readmission. The proportion of patients with a hemoglobin level ≤ 8 g/dl was statistically higher in the early readmission group (p = 0.047). The proportion of patients with comorbid hypertension was statistically lower in the early readmission group ( Table 3 ). Thus, hemoglobin level ≤ 8 g/dl appears to be a factor associated with early readmission and hypertension is a factor for non-early readmission.

The risk of hospital readmission is high in patients with chronic kidney disease. In our study, the overall prevalence was 13.48% and the prevalence of readmissions within 30 days of hospital discharge (HR30) was 65.5%. The frequency of HR30 readmissions is significantly higher than that observed in the West. In a study of readmission of patients with end-stage renal disease on hemodialysis, the HR30 prevalence was 35.2% according to Mathew and Rosen in the USA with a hospitalization rate of 1.73 per patient year, and 19% in the general population [4] [8] . According to Jencks and Williams, the highest rates of HR30 readmission are found in patients with heart failure at 26.9% and recent vascular surgery at 23.9% [3] .

The sex ratio of 1.35 in our study confirms the excess male morbidity observed in the nephrology department already reported by other authors, notably Sakandé in Burkina Faso who reported that 55.2% of these patients were men [15] . According to Pouteil-Noble and Villar [16] , this male predominance could be explained by the higher frequency of renal disease in men and the more rapid progression of these diseases towards CKD. In the West, the unfavourable influence of the male gender on hospital readmission was established both in the general population and in patients with renal disease [8] [17] [18] .

Readmission can occur at any age. In our series, our patients were young adults with a mean age of 42.31 and extremes of 15 and 77 years. In addition, 6.19% of the patients were 65 years or older. This contrasts with Western data where the mean age of patients readmitted for renal disease is over 65 years according to Jencks in the USA and 74 years according to Sood in Canada [3] [19] . In our work, the shorter life expectancy in Africa but also the late diagnosis of renal disease at the stage of complications whose management remains very costly would explain this observation to a large extent.

The socio-professional groups most affected in our series were the unemployed (24.78%) followed by traders (17.70%) and craftsmen (14.16%). In the majority of cases, these were patients with low socio-economic status and a precarious financial situation. The low socio-economic status of patients observed in our study confirms data from African studies [20] - [22] . Similarly, in the West, according to Mathew and al., patients living in low-income postal codes, particularly the unemployed, were at high risk of hospital readmission [4] [8] [23] . The low socio-economic level poses a problem of overall management of hospital readmission that is extremely costly, especially for hemodialysis patients [3] [7] .

The comorbidities observed in our study were dominated by high blood pressure in 63.72% of cases followed by HIV/AIDS infection (15%) and diabetes 11.5%. These rates are significantly lower than the Flythe rates found in hospitals at the University of North Carolina in renal readmitted patients, which are 75% for hypertension and 35% for diabetes, respectively [24] . These comorbidities are also reported by most western series [4] [25] [26] , and complications such as heart failure, stroke, myocardial infarction are true readmission factors [27] .

In almost all cases (86.73%), patients were admitted via the medical emergency department with impaired renal function as the main reason for admission (84.07%). However, uraemic syndrome was the main reason for readmission (68.14%).

The clinical manifestations observed in our study were polymorphic as in most reviews of readmission of patients with chronic kidney disease [4] [19] [24] [25] , and are almost identical during the different episodes of readmission. This is not the case in the Western series, due to the early management of patients in a dialysis-transplant programm in the West [21] as soon as creatinine clearance is less than 10 ml/min and also due to the effective existence of medical insurance especially for patients with chronic kidney disease [3] [25] [28] .

The main clinical manifestations observed in our study are cutaneous-mucosal pallor related to anemia in 91.15% of cases and high blood pressure (86.73%), which is the main cardiovascular and respiratory manifestation in our series. High blood pressure is widespread in the population of hemodialysis patients as described by most western series with a high rate of hospital readmission related to its many complications [18] [25] [27] . Furthermore, edema was noted in 55.75% of cases, and acute lung edema in 24.78% of cases compared to 14% according to Plantinga in the United States [29] . Other signs such as dehydration 24.78%, neurological manifestations 16.81%, fever 14.16%, and digestive manifestations (12.36%) were observed. The high frequency of these signs observed in our study would be largely related to insufficient supplementation treatment. Interventions to reduce readmissions related to these signs could have a substantial impact on readmissions as a whole.

In our study, more than 9 out of 10 patients had an ESRD according to the MDRD equation and only 5% had a stage 4 CRF. These are patients whose survival is strongly linked to treatment by renal supplementation with its risk cortege of re-occurrence hospitalization associated with dialysis and its complications [30] - [32] .

Anemia was observed in all patients. It was mainly normochromic normocytic in 56% and severe in 65% of cases. The same observation was made in France but with a lower prevalence of anemia [33] . The high frequency is partly related to the high cost of recombinant erythropoietin in the management of anemia in our hemodialysis patients, unlike in the West where the availability of cheaper EPO has considerably reduced readmissions [34] - [36] .

Thrombocytopenia was noted in 43.36% of cases during readmission and hypocalcemia in 69.02% of cases. This hypocalcemia results from the inability of the kidney to hydroxylate 25-hydroxyl D3, which is the active form of vitamin D. Vitamin D deficiency, which leads to decreased intestinal absorption of calcium, may explain the observed hypocalcemia [37] . According to POWE, post-hospital vitamin D intake reduces readmissions in ESRD patients by 4.5% [34] .

The cause of readmission was different from the reason for the baseline hospitalization in the majority of patients. Our patients were readmitted for dialysis failure (82.30%) followed by vascular access complications including dialysis catheter infection (10.60%) and deep vein thrombosis (4.42%). In the West, cardiovascular causes remain the leading cause of readmission of patients with end-stage renal disease and account for 40% of deaths [38] . According to Wetmore et al., they are dominated by congestive heart failure (35%) followed by arrhythmias (11%) and acute coronary syndromes (9%) [39] [40] . According to Doshi [25] in the United States, among MEDICARE patients undergoing iterative hemodialysis, the causes of readmission are dominated by infections, vascular access complications and sepsis [36] [41] [42] . The inaccessibility of dialysis for the majority of our patients, due to its high cost and the insufficient number of dialysis machines in our health structures, would be the main cause of readmission in our work.

Treatment in our study was mainly symptomatic. It consisted of treatment of anemia, hypertension, correction of biological abnormalities and dialysis. The treatment of anemia involved blood transfusion and the prescription of erythropoietin. Transfusion of packed red blood cells still plays an important role in the treatment of anemia in Africa in relation to the severity of chronic renal failure at the time of diagnosis. It represented 71% for Ouattara [43] , 73% for Ahmed [44] . In our study, 91.15% of the patients had received a transfusion of packed red blood cells. The prescription of erythropoietin was rare, as in African series, unlike in the West, due to its prohibitive cost [22] .

The treatment of hypertension involved therapeutic classes recommended in the literature, namely diuretics, conversion enzyme inhibitors and calcium channel blockers. According to Honkonen, the class and quantity of antihypertensive drugs does not affect the readmission rate [45] as it did in our work.

In Africa, very few ESRD patients benefit from replacement therapy. In our study, 82.30% of patients were readmitted for dialysis failure. The lack of dialysis facilities described in most African studies is the main reason [21] [22] .

In the West, in addition to symptomatic treatment, the emphasis is on communication and coordination between hospitals and dialysis centers for better exchange of patient data according to Wingard and al. An additional monthly visit by the nephrologist and the use of palliative care consultation can reduce hospital readmissions [46] - [50] .

In our study, the in-hospital mortality rate was 30.97%. Nine out of ten patients who died had ESRD on chronic hemodialysis. In Canada, according to Sood, the death rate was 7.5% and 22% in patients over 65 years of age with ESRD [19] . The high mortality rate noted in our study is linked to poverty and comorbidities, particularly heart disease, and to the inadequacy of the technical facilities in our health structures. In the West, patients with CKD have a shorter length of stay in hospital, averaging 4 days, and a longer average time to readmission [36] [39] . In our study, the average length of stay was 15.89 ± 11.5 days with extremes of 3 and 61 days. Hypocalcemia was associated with mortality risk (p = 0.030) as was obstructive nephropathy (p = 0.029), probably related to the cause of obstruction dominated by pelvic cancers.

Finally, severe anemia (Tx hemoglobin < 8 g/dl) was associated with the risk of early readmission (p = 0.037) and among the comorbidities, hypertension was associated with the risk of non-early readmission (p = 0.034) due to the fact that the blood pressure figures remained high and were linked to poor compliance with the treatment, the accessibility of antihypertensive treatment which often includes three or even four or five drugs.

Elsewhere, we did not find an association between the different factors studied (reason for admission, occupation, age, sex, clinical signs, stages of CKD, treatment) and the death of patients as well as readmissions.

Our work has some limitations: the absence of some information due to the retrospective nature of the work has influenced our comments. Indeed, it was impossible to specify the complete imaging data and the nature of self-medication.

The prevalence of readmission in our study was high (13, 48%). Readmitted patients were young adults, male, with low socio-economic status. The main causes of readmission were insufficient dialysis and complications of vascular access, including infection. Access to dialysis outside of hospitalization is not easy in our context. In fact, the cost of dialysis is enormous compared to the SMIG. Patients who are no longer hospitalized can no longer have regular dialysis, which leads to their return to hospitalization in the context of severe uremia. High blood pressure was a factor for non-early readmission. Hemoglobin level below 8 g/dl was a factor in early readmission.

Mortality was high (30, 97%). Complications of CKD such as hypocalcemia and anemia were associated with the risk of death. These results reveal the value of accessibility for all ESRD patients to dialysis while waiting for kidney transplantation and the search for strategies to reduce the risk of readmissions of patients with chronic kidney disease.

This study was authorized by the National Research Ethics Committee [CNER] in Côte d’Ivoire.

Both attested to the ethical nature of the study and gave their consent to its conduct.

All the experiments carried out during our study and the information collected were done in strict compliance with current guidelines and regulations (Declaration of Helsinki).

Participation in this study was entirely free and voluntary. Participants were free to accept or refuse to participate in our study. Those who accepted were also free, at any time, to end their participation, and, on simple verbal notice without giving an explanation.

The study was conducted in compliance with ethics and medical deontology, human dignity.

The participants were submitted to a confidential questionnaire, and in order to preserve their anonymity a unique number was assigned to them.

Participation in the study was not subject to financial compensation for the participants

The physical data of the study including the recordings of the interviews will be kept by kidney department of the teaching hospital of Treichville. For the duration of the study, the information concerning the study has been stored in a restricted and closed room.

The data of this study are available on request to Pr. YAO KOUAME HUBERT whose contact details are available in the chapter relating to the contact details of the authors.

This study was funded by the authors with their own funds.

KONAN Serge Didier wrote this article, conducted the study and entered the data. DIOPOH Sery Patrick carried out the statistical analysis of the data. YAO Kouamé Hubert directed this work from the protocol to the final draft and coordinated the study and publication of this article. KONAN Serge Didier submitted the article online and adapted it to the journal’s recommendations. OUATTARA Kolo Claude, KOUADIO Marie Dominque, YEO Donafologo Drissa followed the patients in the Department of Nephrology-Internal Medicine D, Teaching Hospital of Treichville. All the authors contributed to the writing and proofreading of this work.

To the Ivorian Society of Nephrology (SINeph) which encourages research in the field of kidney disease and its encouragement to young nephrologists.