INTRODUCTION ( STREP THROAT STUDY)
Clinical diagnosis of infection due to Group A beta-haemolytic Streptococci, is most inexact.
The decision whether to prescribe penicillin for a sore throat, should ideally depend on this, however.
Rheumatic fever and acute post-Streptococcal glomerulonephritis, often follow mild Streptococcal sore throats.
This study was undertaken in order to try to identify clinical features helpful in distinguishing Streptococcal from viral sore throats.
METHODS USED (STREP THROAT STUDY)
Throat swabs were taken from patients complaining of sore throat, in whom the throat was reddened but no follicular exudates were present, as an aid in routine management.
The swabs were placed in Stuart’s transport medium and mailed to the Microbiological Diagnostic Unit at the University of Melbourne.
The clinical features were recorded on the patient’s medical record, early in the study. More features were included as the study proceeded, and a form was then used to record the data.
The results of the swabs were checked against the treatment given at the initial consultation, to allow me to contact patients with untreated Streptococcal infections. No other follow-up was attempted.
The swab results and the clinical features, were not compared during the study, in order to avoid forming opinions about which features were useful – which might have introduced bias into the recording of the clinical features.
Further details will be given in the discussion.
RESULTS OF THIS STREP THROAT STUDY
The following table contains the symptoms and signs which were found to be more common in the Streptococcal infections, in order of decreasing differences from non-Streptococcal infections.
The underlined factors have been compared to the others, in the significance tests. The values of p are the probabilities that the observed differences were due to chance. The totals are different in some cases, as I recorded more features as the study progressed.
The last item was scored by asking the patient to swallow, and then point to where swallowing hurt. The value of this test is increased if it is used only in the presence of tender glands.
There was one feature more common in the viral sore throats – conjunctivitis.
The age distributions of the Strep and non-Strep cases, were also significantly different. There were twice as many Strep infections under 20 years, as over this age. There were equal numbers below and above this age, in the other group.
The scales on the ordinate have been set so that the difference in numbers of Strep and non-Strep infections, is eliminated. The relative excess of Strep infections under the age of ten, is clearly seen.
There were no cases under the age when a complaint of sore throat could be verified. Early in the study, nineteen swabs on very small children with inflamed throats, revealed only one case with strep. Pyogenes. These were not included in the present study.
Some of the clinical features mentioned, were tabulated according to age group to see whether their value was confined to any particular age group. For example, the presence or absence of tenderness over the tonsillar glands, is set out in the following table...
Tenderness over the tonsillar glands is more useful if it is scored as unilateral or bilateral...
Further details will be given in the discussion. There were a number of features which were distributed equally between Streptococcal and viral sore throats, or where the difference was more likely to be due to chance (p = 0.17 or greater.)
The bacteriological methods used did not allow isolation of Haemophilus influenza, but did produce the following bacteria in moderate to heavy growth...
64 Group A beta-haemolytic Streptococci
13 Haemophilus parahaemolyticus
11 beta haemolytic Streptococci other than group A
5 Staph aureus
3 Escherichia coli
1 Klebsiella sp.
THE POPULATION THIS STREP THROAT STUDY IS BASED ON
I didn’t attempt to make this a consecutive series.
Cases were included in the study as time allowed, and some examples were collected from each epidemic of sore throat during the period of the study.
Sore throat was probably not the worst symptom, nor the presenting symptom, in some cases included. The study included 96 cases with mild throat discomfort and only 9 with severe pain.
I believe that this study included a representative sample of the total cases of sore throat presenting. The only conscious bias in selecting cases, was that at the end of the study I was more likely to swab someone if they had been previously swabbed. This was in an attempt to increase the number of intra-person comparisons of different throat infections. This did alter the apparent recurrence rate, as will be discussed later.
It is possible that I was unconsciously biased towards including or excluding certain age groups, so the age distribution should be treated with some reservation. The difference between the age distributions of the Streptococcal and viral sore throats, remains valid, however, as I was not able to pick them apart with any accuracy. The figures for penicillin treatment attest to this.
In any study carried out similarly (not consecutively,) it would be little trouble to record the ages of cases seen but not swabbed, in order to check for bias in the age distribution.
229 swabs were taken between September 1969 and April 1972.
There were approximately 12500 consultations in this period, so the study involved less than 2% of consultations.
2 THE METHODS USED IN THIS STREP THROAT STUDY
One swab was rubbed over the tonsils or tosillar fossae.
The swabs were received at the laboratory mostly two days after they were taken and virtually all were there within four days.
Discussion of the bacteriological methods will be found in appendix 1.
A single swab was taken from each case, so there will have been an error due to some Streptococci not being isolated. Three previously reported studies used repeated swabs to assess this error. The results in these series were as follows...
C.A.R.D. 1945 77% of 157 +ve on first swab
Keith 1946 86% of 118 +ve on first swab
Stillerman 1961 97% of 136 +ve on first swab
Clinical details were not recorded consistently until the duplicated forms were used. These made the job much easier and quicker, but a number of problems remained. The grading of severity of several features was rather subjective. It was difficult to decide whether or not some features were present in mild degree, and I tended to score a symptom as present (mild) if it were complained of, even if I couldn’t be sure on examination. This applied to such features as voice changes, halitosis, fever and coryza.
It is possible that different variations of the same symptom, may have different significance. The symptom of sore throat is given as an example.
A person may have continuous pain in the throat, or have pain only on swallowing or coughing. The sore throat may be present only at night and/or in the morning. The throat may have been sore some days ago at the start of the illness, but not now. The throat may be dry or scratchy as well as sore.
The following definitions and comments apply to the present study...
Temperature normal to 37.5, mild fever up to 38 degrees.
Growing pains – children with leg or joint pains, but not abdominal pains. Adults with polyarthralgia were included here.
Coryza included recent onset of running, blocked and sneezing nose or post nasal discharge. It was impossible to distinguish infections from allergy, but not many of these symptoms are likely to have been the start of a nasal allergy.
Oedema of the tonsils and uvula was seldom recorded, but probably present quite often. The parent sometimes knew how large a child’s tonsils were normally, and said they were larger than usual. Without this information, one could only tell that tonsils were swollen by examination after the infection subsided. Grading tonsils according to size, would include chronic hypertrophy with acute inflammatory swelling. This might still be of value.
The glands examined for tenderness and presence, were the “tonsillar glands” – upper cervical, in anterior triangle at the angle of the jaw. The same remarks would apply to the glands, as made above in relation to grading according to size. Patients or parents would be more likely to know whether glands were larger than normal, than in the case of the tonsils.
Statistical methods are discussed in appendix 2.
3. THE CRITERIA USED FOR DIAGNOSING STREP THROAT INFECTION
Moderate or heavy growth of group A beta haemolytic Streptococci, was the sole criterion of infection with this organism.
Cases with scanty growth of this organism, or growth of any other potential pathogen, were excluded from the analysis. Cases with Streptococci other than group A, were also excluded.
Cases where no potential pathogen was isolated, are referred to as “viral.” This is not strictly true, as mycoplasmas are responsible for a number and aetiological agents can be isolated in only half of sore throats (M.R.C., BMJ ’65, Glezen et al, J.A.M.A. ’67.)
The presence of bacteria obviously is not a satisfactory proof of infection, and the literature reveals differences of opinion, so this problem will be approached from basic principles.
This question is important because of likely errors due to inclusion of carrier states as infections, missing Strep infections where the organism is entirely within the tissues and failure to recognize mixed infections (Strep plus viral.)
Streptococcus pyogenes is a pathogen with a marked capacity to invade and spread within tissues, due to its fibrinolysin and hyaluronidase. In throat infections, the high incidence of lymphadenitis attests to this.
At least in the presence of tonsils, one can probably safely assume that Streptococcal infection will be more than a surface infection.
There being no specific test for detecting tissue invasion and damage by the Streptococcus, one has to rely on indirect evidence from the body’s reaction to the invader. The anti-streptolysin O titre is the most commonly used parameter. This is only of use in studies such as the present one, because it allows a diagnosis only after considerable delay, in the same way that the development of acute rheumatic fever or glomerulonephritis do.
A significant rise in the ASO titre has been reported to follow proven Streptococcal infections in 78 to 90% of cases in different series...
eg: 78% uncomplicated scarlatina untreated
85% food-born epidemic sore throat untreated
90% severe follicular tonsillitis
The rise in titre is largely prevented, however, by early eradication of Streptococci by penicillin. Stillerman found that of 75 patients treated within a week of onset and cured, 83% had no significant rise in ASO titre. This has been found by a number of workers.
It must be accepted therefore, that there is no way in which Streptococcall infection can be proven to be present. A blood test for some toxin produced by Strep. Pyogenes, could be very useful in a study such as this.
In the present study, some of the “Strep infections” must be viral infections in Streptococcal carries and some are probably mixed infections in persons previously Strep carriers.
These distinctions may be held to be unimportant, and diagnosis directed to identification of persons with Streptococci in their throats. Treatment of carriers will help to prevent further cases.
The opposite point of view is also reasonable. There is little risk in leaving carrier states untreated, and some possibility of interfering with the development of natural immunity.
This problem can be eliminated by studying a defined population and taking throat swabs regularly as well as at times of illness (Meyer et al, Honikman et al.)
4. THE RESULTS OF THIS AND OTHER STREP THROAT STUDIES
Some findings from previous studies are set out in the following table, for comparison with the findings of this study. The fractions in the table refer to percentages with each feature in Streptococcal/presumed viral infections.
Thus, support is found in the literature for four of the features found useful in my study and contradiction for one (the presence or absence of tonsils.) Two more are described in an earlier paper describing tonsillitis (Felty, 1923.) These are the coated tongue and fetid breath. Two papers included figures for conjunctivitis, and agreed with the present study in finding this more common in viral sore throats (Chapple and Stillerman.)
Coryza, moderate/severe pain and reddening, although not in the positive findings of this study, were in other studies.
The practical application of these results in the diagnosis of sore throats, will be discussed.
These findings must firstly be applied to the same population. This means sore throats without exudate and including people where the sore throat may not be the predominant complaint.
Perusal of the figures in the first table, shows that the greatest differences between Strep and viral cases were in the absence of the features listed. The viral infections more often lacked these features.
The simplest way to use these findings, is to disregard the differences in usefulness of these features, and treat any case lacking any three (say) of them as viral. This is identical to the APGAR system for scoring the condition of newborn babies – a point is lost for this and that feature and the lost points totaled.
For those who do not wish to introduce any arithmetic into their diagnosis, knowledge of which clinical features are relatively more common in Strep throats, should help improve accuracy.
The most accurate way of applying these results to diagnosis, involves giving appropriate weight to each clinical feature, according to its’ distribution between Strep and viral cases in study. This is explained further in appendix 2.
This last method, applied to the present study, diagnosed all the Strep throats at the expense of including 73% of the viral. The first method, similarly applied, diagnosed all of the Strep throats at the expense of including 81% of the viral illnesses.
5. FURTHER DISCUSSION OF INDIVIDUAL RESULTS
This sign is undoubtedly caused
in some people by cigarette smoking, but this study did not demonstrate this,
perhaps due to the small number of smokers (see later.)
The proportion of people in this study with tonsils removed (one in three,) is the same as in Rantz’s study and greater than in the studies of Chapple and Stillerman.
The proportion of the general population with tonsils removed is unknown. Over the four years prior to this study, the births averaged 47pa and tonsillectomies 19pa. This suggests that the present study contained a representative sample with respect to presence or absence of tonsils.
The explanation for the different findings of the three studies quoted, must be the selection of cases. Rantz studied servicemen with respiratory infections other than pneumonia, with a fever of 100 degrees F or more. Chapple studied throat and ear infections, considered severe enough to warrant antibiotics. Stillerman only included paediatric age group. All these studies included cases with follicular exudates on tonsils and some without sore throat (both excluded in my study.)
This question could be resolved in any future study, by recording part of the same questionnaire for an age matched population without sore throat. The next person presenting with the same age could be asked the pertinent questions.
The other studies are probably more relevant to the decision about tonsillectomy after one attack of rheumatic fever, as they dealt with all respiratory Strep infections.
The presence of diffuse reddening rather than the brightness of the reddening, was found to be the better index of Strep infection. The figures are repeated here for comparison...
Rantz previously described the modification of signs in the throat by previous tonsillectomy. This was seen in the present series, but didn’t reduce the value of diffuse reddening. Injected vessels were more often seen after previous tonsillectomy, and diffuse reddening more often in the presence of tonsils. This is illustrated by the following figures (viral cases only.)
A larger series would probably demonstrate that severe reddening is in favour of Strep infection, but this would always be diffuse reddening anyway.
There were 120 cases where the symptom of fever and the temperature on examination, were both recorded. A temperature of >37.3 degrees was recorded only once, when there was no history of feeling hot, cold or shivery.
In this study at least, there was nothing gained by taking the temperature as opposed to simply enquiring carefully for a history of fever.
POSITION OF PAIN IN THE THROAT
The result of this test was exactly the opposite to my previous information – Strep cases more often pointed to the midline. It is possible that the opposite may have been obtained if I had recorded only the position of pain in those cases with pain in the throat without swallowing.
I presume that the explanation of this sign, is the greater degree of inflammation in the throat in Strep infections. When adenitis is present, the discomfort from this is relatively increased (compared with in the throat itself,) in the viral cases. It is possible that some of the discomfort “in the glands” in these cases, is actually from the Eustachian tubes.
The simple presence of palpable glands, as recorded in the present study, was more associated with the age of the patient than the infection...
NASAL SYMPTOMS AND SIGNS
These didn’t prove to be significantly more common in the viral cases, but probably would be in a larger series. Three other studies found a greater proportion in the non-Strep cases.
The frequent association of coryza and Strep isolation was not so evident in the other studies. I found it difficult to identify and grade these features, and probably included symptoms which would have been disregarded in the other studies.
The present study cannot help in understanding this association. That by Chancellor ’65, has data on Strep plus virus double isolations which may be relevant. There were 34 virus isolations, 4 double isolations and 31 group A haemolytic Strep – so 13% of the Strep isolations were double. This study was of general practice patients and the age distribution of the Strep cases was identical to that of the present study.
Given the low rate of isolation of viruses, the proportion of Strep infections with viruses present could be substantially higher than 13%.
Another study bearing on this question, was that of Chapple et al. They were surprised to find that their Strep and non-Strep cases responded equally to penicillin and sulphonamide, and postulated that the viral cases may involve a pathogenic role for the normal bacterial flora. If this were so, why not a pathogenic role for a well known pathogen being carried by the patient before the virus infection?
PREVIOUS HISTORY OF SORE THROATS
One quarter of cases in this study had a sore throat in the previous 12 months. This is the same proportion as found by Pollard et al, in a study of schoolchildren in Perth. Of 63 children with sore throats, 16 (25%) had two attacks in the year of the study, which is equivalent to one index attack plus one previous attack in the twelve months.
The regular sore throat sufferers had two or more attacks in the previous twelve months. This is pointed to by the difference between Pollard’s unselected cases (8%) and the present series with 35%. The last figure is probably a little too high, due to selection of cases with a previous swab taken, at the end of the study. This only involved 14 swabs.
There were only six out of seventy four patients over the age of twenty, who admitted to smoking more than ten cigarettes a day.
It is well known that sometimes a person will cease getting attacks of tonsillitis after they have a mouthful of decayed teeth removed. This wasn’t evident in this study.
6. CARRIER STATE, SUB-CLINICAL STREP THROAT INFECTION
Carrier rates in recent Australian studies, are as follows...
RACGP CSL study 9%
Essendon school entry 3%
RACGP Prince of Whales 16%
Two other studies from UK and USA respectively, found...
MRC ’61-64 4.7%
Breese ’54 2%
Honikman et al, following a group of children closely, found a new Strep acquisition in association with 10% of clinical infections, and a rise in ASO titre in nearly 6% of infections.
Other studies of acute respiratory infections gave higher figures...
RCH Melbourne 21% of 586 children aged 4 to 14
RACGP CSL study 29% of 113
study 25% of 486
RACGP Prince of Wales 26% of 987
Present study 23% of 276
Two studies of hospitalized patients (Rantz and Keith) found 34% and 43% Strep isolations respectively.
Thus the proportion of Strep isolations increased with the severity of the infection, as was demonstrated within the present study.
Stuart swab kits are an effective method for ensuring prolongued survival of upper respiratory pathogens (Cooper, Stuart et al.)
The swabs were plated onto HBA aerobically and HBA with gentian violet 1:500,000 anaerobically.
Dr. J.R.L. Forsyth, of the Microbiological Diagnostic Unit, Melbourne University School of Microbiology, provided encouragement and information on the methods above and the Essendon school entry study data. His help was greatly appreciated.
Mr. Bill Finger, of the Commonwealth Serum Laboratories, did significance tests and examined the data and information supplied for flaws in methodology. He then did a discriminant analysis using the features found to be significantly associated with the isolation of group A beta haemolytic Streptococci. This all took a long time, (being long before readily available computers,) and was greatly appreciated.
The method used for the analysis, was as follows...
Six factors were chosen and the likelihood ratios worked out as above. In using this to make a diagnosis, the appropriate likelihood ratios for each of them, are multiplied together. This was done with a table in which all the multiplications had been carried out.
Return to main home page of this site
This study of sore throats was conducted during ordinary surgery sessions, in a rural general practice. The recording of clinical details and taking of a throat swab didn't involve much extra time and was done only when time allowed.
The advice and assistance of Dr. J R L Forsyth of Melbourne University and Mr. Bill Finger of CSL are gratefully acknowledged.
276 swabs were collected between September 1969 and April 1972
Symptoms and signs on examination were studied in relation to the swab results.