Intended for healthcare professionals


Delay in the diagnosis of pulmonary tuberculosis, London, 1998-2000: analysis of surveillance data

BMJ 2003; 326 doi: (Published 26 April 2003) Cite this as: BMJ 2003;326:909
  1. Alison Rodger, specialist registrar (alison.rodger{at},
  2. Shabbar Jaffar, senior lecturera,
  3. Stuart Paynter, specialist registrarb,
  4. Andrew Hayward, senior lecturerc,
  5. Jacqui Carless, information officer (TB)d,
  6. Helen Maguire, consultant regional epidemiologistd
  1. a Infectious Disease Epidemiology Unit, London School of Hygiene and Tropical Medicine, London WC1E 7HT
  2. b European Centre on the Health of Societies in Transition, London School of Hygiene and Tropical Medicine
  3. c UCL Centre for Infectious Disease Epidemiology, Department of Primary Care and Population Sciences, Royal Free and University College Medical School, London NW3 2PF
  4. d Communicable Disease Surveillance Centre, London Regional Unit, London, W2 3QR
  1. Correspondence to: A Rodger
  • Accepted 4 December 2002

Noted cases of tuberculosis each year have doubled in London since 1987. In 2000, 12.9 cases per 100 000 population in England and Wales were recorded compared with 40.3 cases in London.1 A delay in the diagnosis of tuberculosis increases the risk of poor clinical outcome—including death and transmission of tuberculosis. 2 3 Understanding which factors influence this delay is crucial for controlling tuberculosis.

Only one small study has previously investigated delays in the diagnosis of pulmonary tuberculosis in the United Kingdom.4 Using surveillance data from London, we estimated the delays in diagnosis of tuberculosis and investigated the factors independently associated with delays.

Methods and results

We analysed surveillance data collected by doctors (1999-2000) and an anonymised national survey (1998) for cases of tuberculosis in London from 1998 to 2000. We calculated the delay in diagnosis as the number of days between the onset of symptoms and diagnosis or the start of treatment (which were on the same day in cases with both recorded). Delay was characterised as greater than the median or at or less than the median. We used unconditional logistic regression to investigate factors that were independently associated with delay.

Table 1.

Odds ratios for delay in the diagnosis of tuberculosisin patients testing positive for tuberculosis in smear tests, London, 1998 to 2000*

View this table:

A total of 1355 patients had a positive result in smear tests of pulmonary sputum; we give results for 853 (63%) about whom data on the time between onset of symptoms and diagnosis had been recorded. Patients with data and those without were similar for age, sex, and ethnic group.

The median age was 34 (interquartile range 26-51) years; 505/849 (60%) of patients were men. A total of 263/842 (31%) patients were white and 267/842 (32%) were black; 542/782 (69%) of patients were born outside the United Kingdom. Median delay was 49 (14-103) days. Univariate analysis showed that factors significantly associated with delay of longer than 49 days until diagnosis or treatment were age, birthplace (United Kingdom or overseas), sex, and ethnic group (table). The geometric mean delay in days were 72 (95% confidence interval 63 to 80) among white patients and 43 (39 to 45) among all other ethnic groups, 72 (66 to 77) among women and 61 (56 to 65) among men, and 64 (55 to 74) among those aged ≥40 years and 45 (40 to 51) among patients aged <40 years. Among patients not born in the United Kingdom, time since entry was significantly positively associated with delay being greater than the median (P=0.01). In multivariate analysis, delays were more likely for white patients (adjusted odds ratio 1.67 (1.2 to 2.5); P=0.01) and women (1.42 (1.1 to 1.9); P=0.01). Age and birth place were not independently associated with delay.


Delay between the onset of symptoms of pulmonary tuberculosis and diagnosis or treatment (median 49 days) was more common for white people and for women. This median delay is similar to findings in other large cities in industrialised nations.5 This might be because tuberculosis may be suspected and investigated more readily among men or black or Asian people.

Our study was limited by the amount of missing surveillance data. It was also impossible to determine the relative contribution of patient and healthcare provider to the total delay. Potential confounders—for example, coinfection with HIV or the accuracy of the data among patients whose first language was not English—were not taken account of.

Recent campaigns have tried to raise awareness of tuberculosis, particularly among ethnic minority groups. Our data suggest that campaigns also need to be targeted at white people, who comprise a third of cases.


We thank John Watson for access to the anonymised data from the national tuberculosis survey 1998.

Contributors: AR, SJ, SP, and AH conceived and designed the study. AR and SJ conducted the analysis. JC manages the database. AR drafted the paper and all authors revised drafts and approved the final version. AR is guarantor.


  • Funding Shabbar Jaffar is supported by a Medical Research Council strategic grant in epidemiology.

  • Competing interests None declared.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract