Improved survival from diagnosis of AIDS in adult cases in the United Kingdom and bias due to reporting delays.

OBJECTIVE: To measure developments in survival patterns among United Kingdom adult AIDS cases. DESIGN: A follow-up survey of cases reported voluntarily to the national surveillance schemes was undertaken to obtain up-to-date information on vital status. METHODS: All reporting clinicians who had a current AIDS patient not known to have died whose AIDS-defining illness was diagnosed before the end of September 1990 were contacted. A total of 3984 cases were included in the analysis. RESULTS: An extra third of deaths other than those reported through routine channels were ascertained by follow-up. Median survival for patients diagnosed before and after the end of 1986 increased from 15 to 18 months for men who had sex with men presenting with Kaposi's sarcoma, from 10 to 19 months for other men who had sex with men and from 7 to 16 months for all others. Improvement in survival was greatest in the first 3 months. One-third of patients have been surviving 2 years or more. Factors observed with independent effects on improved survival are recent diagnosis, younger age and larger cumulative AIDS case load of reporting centre. HIV encephalopathy and other central nervous system symptoms may be associated with poorer survival. CONCLUSIONS: Survival patterns have been changing and generally improving. Average survival for very recent cohorts tends to be underestimated because longer survival has been observed in patients for whom there is a longer delay between AIDS diagnosis and report to the Communicable Disease Surveillance Centre. Information on mortality is improved by active follow-up.

Statistical analysis of case-control studies of communicable diseases.

At the Communicable Disease Surveillance Centre most outbreak investigations are carried out by questioning cases and unaffected 'controls' to look for associations with possible sources of infection. Illness rates are compared in those exposed with those not exposed to a possible risk factor, using statistical techniques appropriate to the survey design. Significance testing to obtain evidence of the source of transmission is made as quickly as possible so that action may be taken. Microbiological corroboration is sought wherever possible. Unlike chronic disease epidemiology the estimation of odds ratios and relative risk is seldom of primary importance. Examples are given of the analysis of three types of study. Firstly where the whole population is interviewed and then where cases are matched 1:1 and 1:M with controls, including an example with missing data, ie variable numbers of controls.

Excess deaths attributable to influenza in England and Wales: age at death and certified cause.

The multiple regression statistical method has already been used to estimate excess deaths attributable to influenza in England and Wales by winter period. Now we report further studies of deaths by age group and certified cause of death. During the ten winters since the influenza A/Hong Kong (H3N2) virus first arrived (1968/69 to 1977/78) there have been about 120,000 excess deaths. Of these about 82% were estimated to be in those aged 65+ years, 17% in the 40-64 year age group and 1% in younger adults. Sixty-seven per cent were certified as due to respiratory disease and 31% due to circulatory system disease. Respiratory deaths increased in all age groups during an epidemic, but of the deaths certified as due to circulatory disease, cerebrovascular deaths were mostly in the 65+ age group and ischaemic heart disease deaths in the 40-64 year age group. In this 40-64 year age group there was evidence that the effects of cold weather and epidemic influenza were multiplicative rather than additive. During the worst influenza winter of 1969/70 respiratory deaths increased by approximately 55% and circulatory system deaths by 4%. Deaths in the elderly increased by 10%, in those aged 40-60 years by 8% and in younger adults by 4%. There was no evidence that excess deaths are followed by a deficit during the following year.

Correlations between microbial parameters from water samples: expectations and reality.

Data which are collected in order to estimate the correlation between parameters must be analysed with caution. Classical statistics of correlation are often inappropriate. The "r" statistic is very easily distorted by non-Normal data. Non-parametric statistics can be helpful. The interpretation and usefulness of the estimates of correlation will depend on the study plan. If water samples come from disparate sources (e.g. upstream or downstream from sewage outlets) then parameters A and B may occur in their highest and lowest numbers according to how close the samples were to contamination sources thus correlating closely. However, if all samples come from sources with similar pollution levels then plots of A and B will show considerable scatter and apparently little correlation. So what is the relationship between A and B? An example of "perfect" correlation, as demonstrated by replicate counts of a single parameter from split samples, gave an r value of only 0.63 (p = 0.62) due to random variation in numbers of organisms between the two halves of the sample. Thus large amounts of data are needed for studying true correlation because relationships between parameters are embedded in the natural variation. This also illustrated that Standards for a single parameter can be "passed" or "failed" by two halves of the same sample. Study design is clearly of fundamental importance. Consideration must be given to the appropriate way of asking questions about correlation between different parameters.

Monitoring infectious diseases using routine microbiology data. II. An example of regression analysis used to study infectious gastroenteritis.

Routine data used to study infectious diseases may contain biases which obscure trends. A 16-year series (up to 1968) of routine laboratory data was used to study patterns of incidence of infective gastroenteritis for which no laboratory diagnosis could be made. An artificial pattern was detected. This arose because GPs tended to refer a greater proportion of their patients during dysentery epidemics. Multiple regression analysis was used to separate out this effect so that the underlying trends could be observed. The seasonal pattern of undiagnosed cases showed an autumn peak. There were also early-winter epidemics of disease with little or no excretion of red blood or pus cells in the diagnostic faeces specimen. Some of the winter communicable disease among older children and adults appeared to be associated with signs of a temporary fat malabsorption in pre-school age cases. Undiagnosed cases in older children and adults were not related to the E. coli serotypes causing disease in infants during this period. The statistical method applied increased the usefulness of these routine data. Although this series of laboratory records is now more than a decade old the results of the analysis can be compared with new observations as more is learned about the epidemiology of previously unrecognized pathogens, especially rotaviruses.

Most probable numbers of organisms: revised tables for the multiple tube method.

Estimation of numbers of organisms is often made using dilution series, for example when examining water samples for coliform organisms. In this paper the most probable numbers (MPNs) are calculated for a 15-tube series consisting of five replicates at three consecutive tenfold dilutions. Exact conditional probabilities are computed to replace previous approximations. When growth is observed in several of the tubes it is not realistic to select a single MPN. Instead a most probable range (MPR) should be reported. But using an MPR creates problems when comparison has to be made with a legislated, single-valued Standard. It is suggested that the wording of the Standards should be expressed differently when the multiple tube method is used.

Monitoring infectious diseases using routine microbiology data. I. Study of gastroenteritis in an urban area.

Sources of information for monitoring infectious disease are routine data, special surveys and ad hoc investigations. In practice much use is necessarily made of routine notifications and laboratory records although this reporting is often incomplete and may therefore be biased. In a retrospective study of a 16-year series (up to 1968) of routine records concerning the diagnosis of gastroenteritis at one Public Health Laboratory we found it possible to identify biases. During school outbreaks of dysentery, laboratory investigation of diarrhoea increased appreciably and such response to publicity affects the use of routine data in surveillance. Although the patients examined were probably representative diagnostically, their selection may not have reflected the age incidence of disease. Valid geographical comparisons within the urban area were not feasible because medical practitioners differed in their use of laboratory facilities and in their habits of notification. Nevertheless, as far as can be established retrospectively, these data did reflect time trends in disease incidence and so had value for monitoring purposes. Several of the biases defined are likely to apply to other sets of routine data. A further communication will describe a statistical method of correcting for quantifiable bias.

Dysentery in general practice: a study of cases and their contacts in Enfield and an epidemiological comparison with salmonellosis.

During a laboratory study of diarrhoea in a large urban area during the years 1953-68, 5319 Sonne dysentery infections in 2620 households and six cases of Flexner dysentery in five households were confirmed. The annual incidence per 10,000 population fluctuated between 2 and 31 with an average of 12.Shigella sonnei was found in 10% of new cases of diarrhoea investigated. Plural infections were found in 58% of the households studied and 30% of all contacts examined were shown to be infected. Among these contacts the infection rate was higher for children (42%) than adults (20%), and higher for adult females (24%) than males (16%). The duration of infection was under 2 weeks in nearly half the index cases followed up and longer than 2 months in only 3%. Intermittent excretion was observed in 15%. Repeated Sonne dysentery infection and illness was observed after intervals as short as 5 months. Antibiotic sensitivity was recorded after 1955 and colicine typing after 1960. Changes in antibiotic sensitivity were noted. Children of primary-school age were most susceptible to Sonne dysentery, but symptoms in ill cases were as severe in adults as in children. During years of high incidence cases were concentrated within a small number of primary schools.Sh. sonnei infections were compared with salmonella infections studied concurrently. Sonne dysentery was a winter disease and most cases were under 11 years old, and case-to-case infection was the usual means of spread. Salmonellosis was prevalent in summer and affected nearly as many adults as children, often persons sharing an article of food. It was more prolonged and severe than Sonne dysentery, but less infectious. In all these respects S. typhimurium occupied an intermediate place between Sonne dysentery and ;other salmonellas'.

Diarrhoea in general practice: a sixteen-year report of investigations in a microbiology laboratory, with epidemiological assessment.

Results are presented of the laboratory examination of faeces specimens from 20,273 patients with acute diarrhoea. These were household index cases seen in general practice in a London borough during the years 1953-68. An annual average of about 2 per cent of households in the area were affected, but there was considerable fluctuation with year and season. Half the patients were children although only one-fifth of the population at risk was under 15 years of age. The greatest incidence of diarrhoea was among children under 5 years old. Male children, but female adults predominated. Specimens were sent for laboratory diagnosis at the discretion of the general practitioner. The laboratory found some abnormality in nearly a third and there were indications that transmissible infection was involved in about one-fifth of patients. The most common diagnosis was Sonne dysentery (9 per cent) which came in epidemic waves and made its greatest impact among young school children. Microscopy was useful, and giardiasis was diagnosed in 1-4 per cent of index patients. Other parasites were less commonly found. Fatty globules characteristic of an infectious condition we have called 'fatty diarrhoea' were frequently observed by microscopy in stools from young children and occasionally from older persons. Blood or pus cells were seen in less than half the shigella and salmonella infections and in a much smaller proportion of the remainder. A test for occult blood performed on specimens from all patients of 40 years or older was positive, in the absence of visible red cells, in a tenth of these cases. Other studies on the bacteriology of diarrhoea in general practice are referred to and some epidemiological comparisons made. The possible place of unidentified infective agents in the aetiology of undiagnosed diarrhoeas and of 'fatty diarrhoea' is discussed.

Sonne dysentery in day schools and nurseries: an eighteen-year study in Edmonton.

A study of Sonne dysentery infections in 19 primary and 11 secondary day-schools and 4 day-nurseries has been made over a period of 18 years in an urban area. Measures were taken throughout to try to prevent and control outbreaks. Sonne dysentery was not endemic in the school population and, even at times of high incidence, epidemics were localized within a few of the primary schools, usually in the spring or autumn terms. Incidence rates of Sonne dysentery were highest in primary schools with large infant departments and in nurseries. Secondary schools entirely escaped outbreaks.The co-operation of head teachers was engaged for all precautionary measures in schools. A policy of immediate exclusion of suspected and infected children was useful in preventing and controlling school epidemics of dysentery. Toilet hygiene was often poor in schools with outbreaks, and this was found to be a profitable field for applying control measures. Infected kitchen workers were only occasionally involved.Recommendations are given in the light of this study, and some reference is made to the more difficult problem posed by dysentery outbreaks in day-nurseries, where temporary closure may be the best policy. It is important that responsibility for infectious disease control in schools be clearly delineated in the reorganized health services of 1974. Teachers can play an important part in limiting infection.

Influenza surveillance in England and Wales using routine statistics. Development of 'cusum' graphs to compare 12 previous winters and to monitor the 1980/81 winter.

Surveillance of influenza in England and Wales is made by monitoring weekly data. Principal indices are deaths, sickness-benefit claims (SBC), laboratory reports and observations from general practitioners (GPs). The 12 winter 1968/9 to 1979/80 have been studied to see which indices best described size and timing of influenza epidemics. A method of plotting the data (called cusums) is suggested which makes it easier to see the effect of small epidemics. Cusums for GP statistics and respiratory deaths were found to be the most helpful indices for describing both size and timing of the epidemics, followed by total deaths and SBC, which were less specific to influenza, and influenza deaths, which lagged behind other indices. Deaths certified as pneumonia have been increasing over these years, whereas bronchitis deaths have been decreasing and these indices should not be used separately for monitoring. The laboratory reporting system is important. It confirms the presence of influenza virus in the community and indicates prevalent strains. Because it is a voluntary system with no defined population base the reports are not reliable numerically for estimating relative size of epidemics or for developing cusums. Cusum plots were unanimous in describing the winter of 1980/1 as one of little influenza activity.

Quality control of the isolation rate of pathogens in medical microbiology laboratories.

Two statistical analyses are suggested to compare the success rates in isolating and identifying pathogenic organisms from specimens achieved by different laboratories participating in a quality control scheme. An example is given in which the analyses are applied to 25 laboratories that received 30 simulated specimens.

Some problems in the prediction of future numbers of cases of the acquired immunodeficiency syndrome in the UK.

The outbreak of the acquired immunodeficiency syndrome in the UK has great implications for health services, and predictions of future numbers of cases have been requested. There is not enough information to make a good estimate of the future epidemic curve, but the trend in new cases can be extrapolated and leads to a predicted number of new cases in the range 460 to 7300 in 1988, with a possible 700 patients from earlier years still alive and needing care at the beginning of 1988.

Excess morbidity and mortality associated with influenza in England and Wales.

By means of a stepwise regression analysis, an estimate was made of the number of deaths and new sickness benefit claims associated with influenza in the winters of 1976--79. In the moderately widespread outbreak of 1976---76 there were 22250 estimated excess deaths, of which 19590 were in the age group 65 years and over. In the same winter there were 908000 estimated excess sickness benefit claims. The smaller outbreaks of the winters of 1976--79 were associated with between 4000 and 6000 excess deaths. An estimate of the effect of winter temperatures on mortality in the cold winter of 1978--79 gave a figure of 4500 deaths in excess of the number in the previous winter.

External quality assessment in water microbiology: statistical analysis of performance.

A UK-based scheme of water microbiology assessment requires participants to record counts of relevant organisms. Not every sample will contain the target number of organisms because of natural variation and therefore a range of results is acceptable. Results which are tail-end (i.e. at the extreme low or high end of this range) could occasionally be reported by any individual laboratory by chance. Several tail-end results might imply a laboratory problem. Statistical assessment is done in two stages. A non-parametric test of the distribution of tail-end counts amongst laboratories is performed (Cochran's Q) and, if they are not random, then observed and expected frequencies of tail-end counts are compared to identify participants who may have reported excessive numbers of low or high results. Analyses so far have shown that laboratories find high counts no more frequently than would be expected by chance, but that significant clusters of low counts can be detected among participants. These findings have been observed both in short-term and in long-term assessments, thus allowing detection of new episodes of poor performance and intermittent problems. The analysis relies on an objective definition of tail-end results. Working definitions are presented which should identify poor performance in terms of microbiological significance, and which allow fair comparison between membrane-filtration and multiple-tube techniques. Smaller differences between laboratories, which may be statistically significant, will not be detected. Different definitions of poor performance could be incorporated into future assessments.

Surveillance of outbreaks of waterborne infectious disease: categorizing levels of evidence.

Public health surveillance requires the monitoring of waterborne disease, but sensitive and specific detection of relevant incidents is difficult. The Communicable Disease Surveillance Centre receives information from various sources about clusters of cases of illness in England and Wales. The reporter may suspect that water consumption or recreational water exposure is the route of infection, or subsequent investigation may raise the hypothesis that water is associated with illness. It is difficult to prove beyond reasonable doubt that such a hypothesis is correct. Water samples from the time of exposure are seldom available, some organisms are difficult to detect and almost everyone has some exposure to water. Therefore, we have developed a method of categorizing the degree of evidence used to implicate water. The categories take into account the epidemiology, microbiology and water quality information. Thus outbreaks are classified as being associated with water either 'strongly', 'probably' or 'possibly'. This system allows a broad database for monitoring possible effects of water and is not confined to the few outbreaks which have been intensively investigated or have positive environmental microbiology. Thus, for reported incidents, the sensitivity of classifying it as water associated should be high but this may be at the expense of specificity, especially with the 'possible' association.

Note: development of an external quality assurance scheme for the detection and enumeration of Pseudomonas aeruginosa from water and comparison of results using modified King's A broth and a commercial agar.

Two trials of the isolation and enumeration of a given strain of Pseudomonas aeruginosa from water are reported. In each trial participants received concentrated samples from two batches, one with low and one with high counts, to be diluted to 500 ml in sterile distilled or deionized water and examined for Ps. aeruginosa by membrane filtration. Membranes were incubated at 37 degrees C for 48 h on pads soaked in modified King's A broth (MKAB) and Unipath Pseudomonas Agar plus CFC supplement (PCFC). The first trial involved eight Public Health Laboratories (PHL) and the organizers provided media from single batches. The second trial, involving 50 PHL, examined the feasibility of a large scale external quality assessment (EQA) distribution. Participants were invited to use the same two media and their usual medium if different. Average counts were close to expected and the spread of results was comparable to that observed from the EQA scheme for indicator organisms. From the results of the two trials a better isolation of the strain of Ps. aeruginosa under consideration was noted with PCFC compared with MKAB.

Hepatitis in dialysis units in the United Kingdom.

A prospective study of hepatitis in 20 dialysis units in the United Kingdom showed that the number of units with outbreaks increased from one in 1968 to three in 1969 and the overall attack rate rose from 1.7 to 5.3% in patients and from 0.5 to 1.3% in staff.In July 1969 a pilot study of patients' sera showed that all of the outbreaks were associated with Australia (Au) antigen. Au antigen was not detected in the unaffected units or in one unit in which there were several patients with abnormal liver function.In January 1970 systematic prospective testing for Au antigen was begun. Au antigen was detected in single patients in five more of the 20 units. The transfer of the affected patients to isolation for dialysis seemed to prevent the spread of infection in the units. The incidence of hepatitis in 1970 was 5.6% in patients and 0.4% in staff.