Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Treatment protocol for LRTI


Bandolier 101 asked whether treatment protocols delivered better healthcare, mainly from randomised trials. Mostly they did, but more examples would be better. Part of the problem is that assessment of treatment protocols in randomised trials is rare, and other study designs, particularly the before-after design, are used, and may be more appropriate. So we chose as an example a before-after design for treating community acquired lower respiratory tract infection (LRTI) from a hospital in Ulster [1].


The study was conducted in the medical wards of a single hospital in Antrim. All adult patients admitted with a primary diagnosis of LRTI during December 1994 to February 1995 formed the control group. Diagnoses were made on clinical grounds supplemented with X-rays in most cases. Patients received empirical treatment before development of a treatment protocol.

After development and institution of a treatment protocol in November 1995, all patients admitted with a primary diagnosis of LRTI from December 1995 to February 1996 formed the intervention group.

The treatment protocol consisted of measuring the severity of the condition according to age more than 60 years, respiratory rate above 30 breaths/minute, diastolic blood pressure below 60 mmHg, white cell count below 4 or above 20 billion cells/L, new confusion, new atrial fibrillation and multiple lobe involvement on X-ray. One point was given for the presence of each of these, and treatment instituted depending on severity:
Protocol construction was with involvement and support of all consultant physicians. Introduction involved presentations, seminars and ward discussions, involvement of new junior medical staff, distribution of written summaries of the protocol, posting the algorithm in all wards, encouragement of implementation by clinical pharmacists.

Details of patients and outcomes were collected on a customised data collection form. Treatment success was a major improvement or complete resolution of all signs and symptoms, and failure persistence or progression of signs and symptoms, or development of new clinical findings, or death from the primary diagnosis, or discontinuation of medicines because of adverse reaction.


There were 112 patients in the control group, and 115 in the treatment protocol group. Their mean age was about 68 years, with a mean onset of about five days at admission. Two thirds were moderate and one third severe on admission. There were no differences between the groups, and no patient was very severe on admission. Most patients (99%) had an X-ray. The only significant difference in laboratory testing was that 98% of patients on the protocol had a sputum cultured, while only 55% of controls had this test.

There were 35/112 treatment failures (31%) on control and 9 (8%) on the protocol. The reasons for the failures are shown in Figure 1. Protocol was better than control for every reason for failure. For every four patients on the protocol there was one fewer treatment failure than if the protocol had not been used (NNT 4.3, 95% CI 3.0 to 7.4).

Figure 1: Results in Antrim before and after protocol for LRTI

Control patients had a mean length of stay of 9.2 days. Those on the protocol had a mean length of stay of only 4.5 days. The overall average cost per control patient was £2,024 and £1,020 for a protocol patient, a saving of £1,000 per patient. Most savings came from lower bed costs and lower antimicrobial costs (£11 protocol vs £54 control).


Adopting a treatment protocol delivered better care at lower cost. Protocol construction and implementation was exemplary. This is a study worth reading, though individual hospitals may want to institute different regimens because of local differences.


  1. FA Al-Eidan et al. Use of a treatment protocol in the management of community-acquired lower respiratory tract infection. Journal of Antimicrobial Chemotherapy 2000 45: 387-394.
previous or next story in this issue