Mortality in men admitted to hospital with acute urinary retention: database analysis

Methodological limitations

We included only men who were admitted to hospital and did not consider those who were treated within primary care or solely in a hospital emergency department. Those patients might have been in a better overall condition than the patients who were admitted. According to a 2004 survey, however, only 10% of UK urologists would not admit men presenting with acute urinary retention.17 Also, a Dutch epidemiological study, the only population based study to date in which the incidence of acute urinary retention was estimated through direct access to a prospectively collected primary care database, reported an incidence of 2.2 per 1000 men per year.3 This is lower than the incidence of 3.1 per 1000 men per year that we reported previously based on hospital admissions in England.4 We therefore consider it unlikely that we missed a substantial number of men.

The Charlson score based on administrative data underestimates the prevalence of comorbid disease.18 19 Undetected comorbidity may account for our observation that men without comorbidity according to the Charlson score still had a higher mortality than the general population. None the less, we found that the Charlson score identified at least one comorbid condition in more than a third of men with acute urinary retention and that mortality was most increased in this group. A more accurate instrument could have shown an even more marked impact of comorbidity.

The reported mortality after acute urinary retention was based on the linkage between the HES database of all admissions to NHS hospitals in England and the mortality database of the Office for National Statistics. Most (96%) patients had a date of death that was based on robust linkage methods that relied on an exact match of at least two of the NHS number (a unique 10 digit code allocated to an individual to enable identification for NHS health care), date of birth, and postcode. For these patients, we found contradictory linkage results (that is, an admission date after a date of death) in only 0.3% of patients. In the 4% of patients remaining, we used a less robust linkage method (partial match of date of birth and exact agreement of sex and postcode) and found contradictory results in 54% of the patients. As explained before, we assumed that men with contradictory mortality results were alive until the end of the study period. Sensitivity analyses of alternative assumptions showed that the mortality for all men within one year after acute urinary retention always changed by less than 0.5 percentage point, and in turn the relative increases in mortality compared with the general population did not change appreciably.

Comparison with other studies

We specifically looked at longer term mortality after acute urinary retention. The only other study to date that provides some evidence on mortality after acute urinary retention was carried out in five healthcare regions in the UK in the mid-1990s in 3966 men who underwent prostatic surgery.20 That found that the 1242 men who underwent a prostatectomy after acute urinary retention had a higher mortality in the first 90 days (3.0%) compared with the 2724 men who had surgery for symptoms alone (0.7%). Other studies that have reported on outcomes after acute urinary retention tend to focus only on recurrence and the need for prostatic surgery.4 12

Acute urinary retention shares several characteristics with fractured neck of femur. Both are acute and serious age related events that almost always result in hospital admission but do not directly cause death. Two English studies have recently reported the one year mortality in men after fractured neck of femur to be just over 40%.21 22 Mortality clearly increased with age from 20% in men aged 65-69 to 50% in men aged 85-89.21 Comorbidity was found to be an important predictor of outcome, with mortality in the first year about 1.5 times higher in patients with at least one comorbid condition than in patients without comorbidity.22

The comparison with fractured neck of femur shows that acute urinary retention constitutes a health problem of a similar magnitude. The overall one year mortality was 20%, which is lower than that observed after fractured neck of femur. The incidence of primary acute urinary retention, however, is 3.1 per 1000 men per year, which is higher than the incidence of fractured neck of femur, estimated to be about 2.4 per 1000 men per year based on the HES database.

Explanations for the increased mortality

Though acute urinary retention is generally considered a urological emergency, serious consequences are rare with appropriate treatment. Our finding of a high mortality in the first year seems to contradict this. An obvious explanation for the increased mortality is that occurrence exposes men to risks associated with admission to hospital, catheter related and other nosocomial infections, and invasive procedures that often need a general anaesthetic.

While these factors do play a role, the high mortality seems to be linked mainly to comorbidity for several reasons. Firstly, the presence of comorbidity according to the Charlson score substantially increased mortality within patients of similar age and with the same type of acute urinary retention. Secondly, mortality was also higher in patients with precipitated rather than spontaneous retention. Precipitated retention occurs after a triggering event that is unrelated to the prostate and therefore a further indication of comorbidity. Thirdly, there seems to be a “dose-response” relation as the lowest mortality was found in patients with spontaneous retention and no comorbidity according to the Charlson score and the highest mortality in patients with precipitated retention and comorbidity. Lastly, a detailed inspection of the diagnostic fields in patients with spontaneous acute urinary retention and no comorbidity apparent showed that 58% of these patients had at least one diagnosis not related to the prostate and not captured by the Charlson score.

Of the patients who had comorbidity according to the Charlson score, irrespective of the type of acute urinary retention, about 30% had cardiovascular disease, 25% diabetes, 25% chronic pulmonary disease, and 15% a malignancy. The relatively high prevalence of cardiovascular disease and diabetes is consistent with the findings of recent aetiological studies linking hypertension and metabolic syndrome with the progression of benign prostatic hyperplasia.7 8

Clinical implications

The high mortality that we observed in men admitted to hospital with acute urinary retention shows that they are a vulnerable group of patients. These observations reinforce the importance to the clinician of adopting a multi-disciplinary approach in the assessment and management of such patients.23 Men with acute urinary retention should undergo a comprehensive investigation for comorbid disease. The extent to which mortality after acute urinary retention can be reduced, however, will depend on the nature and severity of the comorbidities involved and the effectiveness of the available treatments. For example, some patients will have comorbidity that is already optimally treated and for whom little further can be done.

Our observation that about half of the deaths after acute urinary retention occur within the first 90 days suggests that assessment of comorbidity must occur promptly. Indeed, perhaps screening for other morbidities should start in men presenting with lower urinary tract symptoms attributable to benign prostatic hyperplasia before they experience acute urinary retention.

Research implications

The two most important unanswered questions relate to the way in which comorbidities can be identified and the extent to which their treatment will reduce mortality after acute urinary retention. In our view, research building on administrative databases, such as the hospital episode statistics, enriched through linkage with additional clinical data seems to be the obvious next step to answer both questions. Firstly, administrative databases can provide near complete follow-up in terms of further treatment as well as death. Clinical data, perhaps collected prospectively in a subgroup of patients and linked to the administrative data at the individual patient level, will allow a further exploration of the impact of comorbidity according to its nature and severity. The results of this will inform the development of treatment strategies. Secondly, these strategies can then be evaluated with research designs that build on combining administrative data with data derived from clinical trials.