Diminished antibiotic sensitivity of Neisseria gonorrhoeae in urban and rural areas in Kenya.

Bull. Org. mond. Sante 1971, 45, 707-717Bull. Wid Hith Org. Diminished Antibiotic Sensitivity of Neisseria gonorrhoeae in Urban and Rural Areas in Kenya * A. R. VERHAGEN,1 M. VAN DER HAM,2 A. L. HEIMANS,3 0. KRANENDONK4 & A. N. MAINA2 A total of 1 703 cases of suspected gonorrhoea were investigated in eight urban and rural clinics in Kenya; sensitivity tests were performed in 736 positive cases. A majority of the strains had relative resistance to two drugs in common use-namely, penicillin and streptomycin ;for tetracycline a slightly decreased sensitivity was oftenfound, and resistance to these antibiotics was usually linked. It was found that relative resistance to antibiotics is already common in rural areas, in which the majority of the African population live, and that resistance to penicillin is even more common in rural areas than in towns. Evidence that this may result from the continuation of inadequate treatment, especially in developing tropical countries, was obtained. Such countries are becoming an immense reservoir of strains with a high level ofrelative resistance to penicillin. Ways ofremedying this situation are discussed. Relative resistance (RR) of the gonococcus to penicillin and other antibiotics was first discovered in European countries (Guthe, 1961). An even more alarming development was reported later from several developing and tropical countries where a large majority of the strains showed relative resistance to penicillin (Willcox, 1970). Patients investigated in countries of the latter category usually came from selected social groups that were rather unrepresentative of the general population. Almost all investigations were made in capital cities and in harbour areas, often among special elite groups or among foreign soldiers and " hostesses " catering for them. Such groups probably differ strongly from most gonorrhoea patients, who still live in a more or less traditional, rural environment. In some respects, such as ease of * From the Medical Research Centre, Nairobi, Kenyal and the Department of Tropical Hygiene, Royal Tropica, Institute, Amsterdam, Netherlands. 1 Senior lecturer in dermato-venereology, University of Nairobi, Kenya, and consultant at the Kenyatta National Hospital. ' Senior laboratory technician. 3Consultant dermato-venereologist, St Franciscus Hos- pital, Rotterdam, Netherlands. ' Director and Bacteriologist, Institute of Tropical Hygiene, Royal Tropical Institute, Amsterdam, Netherlands. access to treatment and therapeutic adequacy, or in sexual habits, the minority groups investigated resemble the population of developed countries more closely than they resemble the majority of their countrymen. One might indeed advance the view that little is known about the largest reservoir of gonococcal infections found in tropical countries (WHO Chronicle, 1965). The purpose of this investigation was to obtain data from a more representative cross-section of the population of a developing tropical country, includ- ing that in rural areas, and to acquire better informa- tion on the sensitivity of the gonococcus to anti- biotics. METHODS AND MATERIALS General The investigation was carried out in several urban and rural areas of Kenya, a country of 11 million in- habitants situated on the equator. A group of gonorrhoea strains isolated in the Netherlands was reinvestigated, by means of the same bacteriological methods, for comparison. Sampling in Kenya The patients were investigated during the last months of 1968 and during 1969 in government and 2762 - 707 - 708 A. R. VERHAGEN AND OTHERS city-council clinics. These clinics are visited by all sections of the population, with the probable excep- tion of a small minority of wealthy Kenyans and expatriates. Timboni clinic in Mombasa was an exception; here, prostitutes were investigated exclu- sively, during a legally authorized examination for venereal disease. Timboni and Nairobi have special VD clinics (that in Nairobi is for the general popula- tion); the other clinics were outpatient departments of general hospitals. A special team of physicians and laboratory assistants stayed in each hospital for several weeks, the intention being to study every patient in whom gonorrhoea had been provisionally diagnosed, or was at least suspected, by general medical personnel. However, not all such patients were referred to the special team (see Discussion). The eight clinics are located in densely populated parts of the country between the Indian Ocean and Lake Nyanza (Victoria), which are 800 km apart. Nairobi (population of 500 000) is the largest city and the capital; Mombasa, which has two clinics, is the largest port in East Africa (population of 250 000); Kisumu (population of 35 000) and Kericho (10 000) are provincial towns. Kitui (3 000), Machakos (6 000), and Nyeri (10 000) hardly have an urban character and are administrative and trading centres for large rural areas; over 80% of the cases investi- gated in these three clinics came from virtually every part of the surrounding areas, which comprise large districts with 250 000-750 000 inhabitants. Bacteriological investigation Specimens were taken from the urethra and the endocervix; if indicated, prostatic secretions were obtained by massage. Other possible sites were dis- regarded. Two smears were made from every patient; these were stained by Gram's method (Reyn, 1965) and assessed independently; the discharge was also inoculated on the medium described by Thayer & Martin (1966) and incubated at 37°C in a C02-rich environment; the culture was then taken by car or aircraft to the main laboratory in Nairobi. A previ- ous trial had indicated that, at least in warm coun- tries, this method was better than the use of special culture media for transport, this finding being in agreement with that of Robinson, Hicks & Davidson (1970). If a positive result was obtained, a pure subculture was freeze-dried and sent to the Royal Tropical Institute, Amsterdam. The freeze-dried strains were re-grown on the same medium and the gonococci were identified by means of Gram's stain, a positive oxidase reaction, the absence of growth on plain nonenriched agar at room tempera- ture (20°C), and the usual fermentation tests with glucose, maltose, and sucrose. All these criteria were established for every strain identified as Neisseria gonorrhoeae. Sensitivity to antibiotics was tested by the plate- dilution method. For penicillin, tetracycline, and streptomycin, the Thayer & Martin medium without the usual antibiotics (vancomycin, colistin, and nystatin) was used, and for sulfadimidine the medium of Mueller & Hinton (Difco). With every batch of medium prepared from one stock solution, three WHO reference strains (III, V, VII; Dr A. Reyn, Statens Seruminstitut, Copenhagen) and several Netherlands strains were tested concurrently. The sensitivity of all strains to penicillin and tetracycline was tested but a sample of 418 strains was investi- gated for sensitivity to streptomycin and sulfa- dimidine. The sample included strains from each town, but was otherwise random. Strains from the Netherlands These strains have been described fully by one of us.' They were isolated exclusively in women during 1964-65, mainly at the University Dermato-Venereo- logy Clinic, Amsterdam. Reproducibility and reliability of methods In addition to control tests using the three WHO reference strains, about 10% of all tests with each drug were duplicated. For the three antibiotics, " one-step " differences in the minimum inhibitory concentration (MIC) values were numerous, both in the WHO strains and in duplicate tests of Kenyan strains; for penicillin and tetracycline they were found in almost a quarter of all tests. However, " two-step " or larger dif- ferences with penicillin and tetracycline, or discre- pancies in the tests with streptomycin (with respect to this antibiotic the strains were divided only into sensitive and resistant), were rare; such discrepancies may be defined as errors. With the WHO strains, a total of 112 control tests were performed for penicil- lin, 100 for tetracycline, and 57 for streptomycin; duplicated tests numbered 117, 85, and 37 for these three antibiotics, respectively. Less than 5% of the sensitivity determinations deviated 2 or more " steps " from the known value of the reference 1 Heimans, A. L. (1967) Over diagnostiek en behandeling van gonorroe bij vrouwen (thesis, Amsterdam). ANTIBIOTIC SENSITIVITY OF N. GONORRHOEAE IN KENYA strains or the value of the test strains for any anti- biotic, as determined by duplicate tests. Batches of media in which more than one WHO strain gave inconsistent results were discarded. In the penicillin tests, most of the small number of two- step, and larger, discrepancies remained within the respective groups of sensitive or RR strains; only 3 of a total of 229 controls (duplicate and WHO strains) crossed the boundary (MIC=0.1 IU/mI) be- tween these groups. Tests with sulfadimidine were less easily repro- ducible, as other studies have shown. In 53 tests with the three WHO reference strains and in 40 duplicate tests one-step and two-step differences were found in over 50%, and larger discrepancies were noted in about 10%. In conclusion, little significance can be attached to minor differences in MIC for the three antibiotics, such as those between 0.01 and 0.02 IU/ml for penicillin, but the margin of error in the overall distribution of sensitive and RR strains is small. However, the sensitivity tests for sulfadimidine provide approximate results only. Data processing All data were processed by computer at the University Computer Centre, Nairobi. RESULTS Table 1 provides a survey of the entire group investigated, with the exception of babies with neonatal conjunctivitis. It includes all patients referred by general medical personnel except those in whom the diagnosis was obviously incorrect. The positive diagnosis made in 902 cases was based either on a positive culture or on two independently assessed positive smears when the culture could be classified as failed, e.g., if it was massively con- taminated. Of these positive strains, 18.4% were lost, mainly because it proved difficult to obtain pure cultures in hot lowland climates with a long trans- port time to Nairobi (e.g., from Kisumu and Mombasa). All other 801 cases were classified as i" negative" or " doubtful "; doubtful cases-for example, those with two positive smears and a negative but good culture-numbered 26. No other Neisseriae or Mimiae were encountered in Kenya; the specific tests performed in the Nether- lands confirmed the classification of gonorrhoea made in Kenya on the appearance of the colony and the Gram strain in almost every case (a yeast was found twice). We consequently feel justified in assuming that the great majority of the 166 strains (Table 1) lost during transport and subculturing were also gonococci. Table 2 shows the distribution of sensitivity to penicillin of Kenyan samples compared with strains from the Netherlands. The hospitals and spe- cialized clinics where the work was carried out were centres for referral from health centres; a minority (1 3 %) of the patients had already been treated else- where, almost always with penicillin (see Discussion). Relative resistance is usually defined as an MIC >0.1 IU/ml and this classification has been adopted here. The treated cases obviously represented a selected group, since the proportion of RR strains was much higher (86.5 %). Statistical analysis was therefore performed separately for untreated and treated patients. The following scoring system was used: 0, MIC=0.01-0.05 IU/ml; I, MIC = 0.1 and 0.2 IU/ml; 2, MIC=0.5 IU/ml; 3, MIC=1 and 2 IU/ml. For the entire group the average score wassignificantly higher for the treated cases (2.15) than for the untreated cases (1.53), P <0.001; this result was also found in each individual clinic. In a comparison of the clinics, differences in scores were insignificant for the small group of treated cases. For untreated Table 1. Suspected cases of gonorrhoea, bacteriological confirmation, and sensitivity tests Total of Negative Positive Tested for Not tested forclinical suspects o diagnosis sensitivity sensitivityinvestigated diagnosis men 1 020 271 749 618 131 (17.5%) women 683 530 153 118 35 (22.9 %) total 1 703 801 902 736 166 (18.4%) 709 A. R. VERHAGEN AND OTHERS Table 2. Sensitivity of gonococcal strains to penicillin * Entire group Untreated patients Treated patients MIC of penicillin (lU/ml) Srosm 0.01, Per- Per- 0.02, 012 0.5 1 2 Total centage Total RR centage Total RR0.05 02RR RR Nairobi 117 17 69 71 5 279 58.1 252 142 56.3 27 20 Mombasa (general) 50 6 54 33 2 145 65.5 123 74 60.2 22 21 Mombasa-Timboni 5 1 2 2 1 11 (54.5) 9 4 (44.4) 2 2 Kisumu 16 2 13 34 1 66 75.8 55 40 72.7 11 10 Kericho 18 0 7 24 2 51 64.7 41 25 61.0 10 8 Machakos 16 1 30 31 0 78 79.5 62 48 77.4 16 14 Kitui 31 5 14 39 3 92 66.3 86 55 64.0 6 6 Nyeri 4 1 1 7 1 14 (71.4) 13 9 (69.2) 1 1 total 257 33 190 241 15 736 65.1 641 397 61.9 95 82 Netherlands 76 15 16 4 0 111 31.5 Percentages in parentheses are based on less than 20 tests. patients, however, a significant difference between the average scores at the seven centres was found (0.001 <P <0.005). It was evident that these local averages fell into two groups: the two large towns (Nairobi and Mombasa) on the one hand (total average=1.35), and the five up-country clinics on the other hand (total average =1.79). Each in- dividual up-country clinic scored considerably higher than either of the two large towns. Up- country, the proportion of RR strains was 72 %, and of strains with an MIC>1 IU/ml 47.5%; in the towns, the proportions were 60.5% and 260%, respectively. It may be assumed that untreated cases provide the most representative, non-selected sample of strains as they circulate in the country. Strains with relative resistance to penicillin are accordingly more com- mon up-country. A moderate degree of relative resistance (MIC=0.1 or 0.2 IU/ml) was rare every- where in Kenya and much more common in the Netherlands strains tested simultaneously. Table 3 gives the same results for tetracycline. In a similar analysis (score 0-4 for each MIC value), the average score was again higher for the treated cases than for the untreated cases (2.67 against 2.1 1; P <0.01), but it was remarkable that, unlike the penicillin results, there was no significant difference between urban and up-country clinics (P >0.25). Strains with an MIC > 2 Hg, usually considered as relatively resistant to this antibiotic, formed a small minority everywhere, but an MIC of 1 jig is common; for reasons given in the Discussion, we have termed this " slightly decreased sensitivity ". Resistance to streptomycin is shown in the follow- ing tabulation (1) for the entire sample of 418 strains from Kenya and 50 strains from the Netherlands. Kenya Netherlands Sensitive (MIC = 10, 20 or 50 ,uglinl) 195 (47%) 43 (86%) Resistant (MIC > 200 jsg/ml) 223 (53%) 7 (14%) Resistance to sulfadimidine is given in the following tabulation (2), also for the 418 strains from Kenya and 50 strains from the Netherlands. MIC (gg/ml) 3 or 6 12 25 50 100 200 500, 1 000, or over 1 000 Kenya 44 69 63 88 75 45 Netherlands 6 6 9 12 11 3 34 3 No regional analysis was performed since many of the up-country groups were too small. In all places, about half, or a small majority, of the strains were resistant to streptomycin. Further analysis of the resutls are not given in detail but one remarkable finding must be mentioned. In untreated cases, strains from women (average of 710 ANTIBIOTIC SENSITIVITY OF N. GONORRHOEAE IN KENYA 711 Table 3. Sensitivity of gonococcal strains to tetracycline * Entire group Untreated patients Treated patients Source of sample MIC of tetracycline (,ug/ml) Per- Per- Total centage centage Total MIC Per- 0,1 0,2 0,5 1 20rRR pg/ml pg/ml centage Nairobi 22 65 68 100 24 279 8.6 44.4 252 108 42.9 27 16 Mombasa (general) 8 41 16 74 6 145 4.1 55.2 123 62 50.4 22 18 Mombasa-Timboni 0 5 3 2 1 11 (9.1) (27.3) 9 2 (22.2) 2 1 Kisumu 2 12 18 27 7 66 10.6 51.5 55 25 45.5 11 9 Kericho 7 7 13 19 5 51 9.8 47.1 41 16 39.0 10 8 Machakos 4 18 6 46 4 78 5.1 64.1 62 37 59.7 16 13 Kitui 5 27 17 40 3 92 3.3 4.7 86 38 44.2 6 5 Nyeri 1 7 2 4 0 14 (0.0) (28.6) 13 3 (23.1) 1 1 total 49 182 143 312 50 736 6.8 49.2 641 291 45.4 95 71 Netherlands 17 55 21 16 2 111 1.8 16.2 * Percentages in parentheses are based on less than 20 tests. 1.81) scored significantly higher than those from men (average of 1.49) in resistance to penicillin (P= 0.04); the situation with regard to other antibiotics was similar. This finding was not explained by geogra- phical origin; on the contrary, more than half of the women had been investigated in Nairobi, the town with the lowest overall relative resistance. Only in Nairobi was the smaller group of women's strains large enough to be compared with that from men, and the difference in score between the sexes (1.72 against 1.22) in this single clinic was again significant (P= 0.02). Table 4 and the following tabulation (3) show the correlation between sensitivity to penicillin on the one hand and to tetracycline and streptomycin on the other. MIC penicillin Streptomycin (lUlmi) Sensitive Resistant Total 0.01, 0.02, 0.05 39 10 49 0.1,0.2 5 6 11 0.5 10 19 36 178 46 2 0j 2) 2 total 58 88 146 It appears from Table 4 that strains fully sensitive to penicillin never showed relative resistance to tetra- cycline (MIC > 2 ytg/ml); on the other hand, many Table 4. Correlation between sensitivity of gonococcal strains to penicillin and tetracycline MIC of MIC of tetracycline (<g/ml) l penicillin (lU/ml) 0.1 | | 0.5 |1 1 >1 Tota 0.01, 0.02, 0.05 44 140 66 7 0 257 0,1, 0,2 5 14 7 5 2 33 0.5 0 10 47 124 9 190 1 0 18 22 169 32 241 2 0 0 1 7 7 15 total 49 182 143 312 50 736 strains with relative resistance to penicillin are still fully sensitive to tetracycline (MIC <0.5 ,ug/ml). Statistical analysis showed that sensitivity to both antibiotics was strongly correlated (P < <0.001; r (product moment correlation coefficient) = 0.7273). The correlation between sensitivity to penicillin and streptomycin given in tabulation 3, although for a smaller random sample than that shown in tabula- tion 1 1 was significant (P < <0.01; r = 0.594). 1 Most of the streptomycin test results were obtained after the computer processing had been carried out. 712 A. R. VERHAGEN AND OTHERS The pattern found in other countries has been confirmed; in a majority of strains, resistance to both antibiotics is linked but isolated resistance to either one occurs and, contrary to the relationship with tetracycline, isolated resistance to streptomycin in penicillin-sensitive strains is not rare. The following tabulation (4) shows the correlation between all three antibiotics: Streptomycin Penicillin Sensitive Resistant sensitive 38 10 RR 17 21 tetracycline total 55 31 sensitive sensitive 1 0 total 3 57 tetracyclne Resistance to streptomycin appears to be more closely correlated with decreased sensitivity to tetra- cycline (MIC =1 jug/ml) than with relative resistance to penicillin; 95% of the strains with an MIC for tetracycline of >1 jug/ml are also resistant to streptomycin. Relative resistance, survival of RR strains, and " sequential selection " by treatment The high proportion of RR strains, especially in developing countries (Willcox, 1970), demands an explanation. It has been assumed that once the first RR strains appear, inadequate treatment fails in patients infected with these strains, so that from a bacteriological point of view there is selective sur- vival of RR strains; frequent repetition of this event causes sequential selection of RR strains in the gonococcal population of an area. The highest pro- portion of RR strains would then be found in areas, such as developing countries, where inadequate treatment regimes have been continued; absolute proof for this hypothesis is not available but we obtained rather impressive circumstantial evidence. Tabulations 5 and 6 give some data from the case histories. Total numbers are slightly lower than the number of patients (736) in whom tests for RR could be done since no satisfactory history could be elicited from a small minority of patients. The following tabulation (5) shows that 84 patients who were interviewed had already received one or more injections (probably of penicillin) when they were first referred to the gonorrhoea team; this treat- ment had usually been given in smaller institutions such as health centres. Distribution of response to test MIC ofpenicillin All Those interviewed for sensitivity interviewed with history Percenltage test (IlUmi) of treatment 0.01,0.02, 0.05 254 10 3.9 0.1, 0.2 32 3 9.4 0.5 185 25 13.5 1 239 42 17.6 2 15 4 26.7 total 725 84 11.6 It is clear that the proportion of such patients increases stepwise in connexion with relative re- sistance to penicillin. The selective survival of RR strains after treatment with injections (which can be assumed to have been penicillin) is thus demonstrated (P = 0.01; r = 0.192). The following tabulation (6) shows that in men there is indeed a prolonged survival of RR strains, probably caused partly by therapeutic failures. The proportion of patients having complaints for longer than 1 week increases consistently with the increase in RR (P <0.01; r = 0.232). In the smaller female group there was no such correlation, probably be- cause many women are hardly aware of the infection, and least of all its beginning. MIC ofpenicillin Complaints Percentage (IU/mi) Total (> 1 week) 0.01-0.05 213 38 17.8 0.1-0.5 185 54 29.2 1-2 205 89 43.4 total 603 181 30.0 The correlation between sensitivity to antibiotics and the effect of treatment, established independently by different persons, was studied in a random group of 85 men with positive cultures in the Nairobi clinic. The patient received the usual treatment in this clinic (1.2 megaunit of PAM) 1 immediately after the investigation and the results of the therapy were checked after I week. When it is realized that this was the only speci- alized VD clinic under medical supervision, the results given in Table 5 are rather appalling. The selective survival of RR strains is apparent; this is illustrated even better by the subsequent fate of the patients infected with these strains. The patients were routinely requested to come back after 1 week and failures (35 in our sample-see Table 5) were treated with I g of streptomycin and 4 g of sulfa- methoxypyridazine. After another week, only 3 pa- tients were cured, 5 defaulted, and 27 were still l'Procaine penicillin G in oil with aluminium mono- stearate. ANTIBIOTIC SENSITIVITY OF N. GONORRHOEAE IN KENYA Table 5. Sensitivity and results of treatment with 1.2 megaunit of PAM MIC of Result of test Total No. penicillin (lU/mi) Cured Defaulted Failed treated 0.01-0.05 19 12 1 (3.1 %) 32 0.2 or 0.5 8 5 10 (43.5 %) 23 1 or 2 4 2 24 (80.0 %) 30 total 31 (36.5 %) 19 (22.4 %) 35 (41.2 %) 85 infected. The infected group was then treated orally with tetracycline (3.5 or 7 g) and after a further week there were 10 cures, 11 defaulters, and 6 failures. The gonococcal strains in the latter group had invariably shown an MIC to tetracycline of I or 2 ,ug/ml at the first attendance. In spite of that, the usual treatment scheme had never been changed in this clinic because failures were ascribed to irrespon- sible promiscuity and subsequent reinfection. For a majority, the impressive correlation with sensitivity tests suggests another explanation. At least 17 of 30 strains in the category with the highest MIC for penicillin (1 or 2 IU/mI) survived until treat- ment with tetracycline was begun; that is, for 2 weeks after the first attendance, not counting defaulting patients. The inadequacy of treatment with respect to a final cure does not mean that symptoms have not subsided temporarily; thus, another contact may have been infected with the RR strain during such a clinical remission. Apart from those in this sample, 28 patients were referred back to the team with persistent infections during the short periods when the team worked in the clinics. In two cases, a large difference in MICs at the first and second attendances indicated a new infection, but it was striking that all other 26 cases had RR strains, of which 18 had an MIC for penicil- lin of 1 IU/mi. Outside Nairobi, the usual treatment given to these patients had been 600 000 IU of pro- caine penicillin for several days. Neonatal conjunctivitis is not included in the results described above. The team investigated 10 suspected cases but only 3 were caused by N. gonor- rhoeae. In developing countries it is too often assumed that the gonococcus is the only organism causing this disease. DISCUSSION Method of sampling Recent investigations in developing countries in which a very high proportion of RR strains has been reported raise the question of whether patient selec- tion might have influenced the results-in other words, whether the strains investigated were a random sample from the clinics and whether the patients in such clinics were representative of the country or area. This question is difficult to answer but usually it has scarcely been considered. As far as the Kenyan investigation is concerned, the follow- ing points deserve discussion. (1) The sample was mainly selected by auxiliary medical personnel who had to refer patients to the gonorrhoea team because it was impracticable to supervise selection personally in clinics with up to 700 outpatients a day. Consequently, criteria for selection varied from one clinic to another; in some clinics, mainly male cases with a rather obvious clinical diagnosis were referred (Mombasa, Kericho), while at other clinics many patients were included in whom the diagnosis was at best suspect, and these were mainly women (Nairobi, Machakos, Kitui). This variation in the male: female ratio may have had a slight influence on the proportion ofRR strains, but the female group remained such a small minority everywhere that it is not conspicuous when the results of these clinics are compared. (2) With regard to the loss of 18.5% of positive strains, the data from some successful periods with only 2% losses could be compared with less satis- factory periods in the same clinics; there was no evidence that these losses had influenced the propor- tion ofRR strains. Most of the losses were the result 713 A. R. VERHAGEN AND OTHERS of accidents that could be expected to affect the survival of all strains equally (transport, etc.). (3) Some patients had already been treated unsuc- cessfully elsewhere but their results have been analysed separately; the case history was elicited from every patient by a specially trained assistant according to a routine questionnaire. In conclusion, the strains investigated may be considered as representative for the patients attending the clinics in respect of the proportion ofRR strains. Whether such strains are representative for the gonoccocal population of an area is a question that has perhaps never been answered satisfactorily because, for obvious social reasons, gonorrhoea can hardly be studied in random samples. We have tried to approximate this ideal by studying rather large and geographically widespread samples in general clinics. Discussion of results No significance must be attached to the male: female ratio, which is about 5: 1 for positive cases (Table 1), since women were simply not referred in some clinics. The fact that there were nevertheless many more women than men with negative results demonstrates how difficult it is to spot gonorrhoea clinically in female patients, as has often been pointed out (Allen, 1970; Caterall, 1970; Cave et al., 1969; Ellner, 1969; Johnson et al., 1970; Lucas et al., 1967; Rees & Annels, 1969; Schmale, Martin & Domescik, 1969). The sensitivity to penicillin of the Kenyan strains showed some peculiarities when compared with the results of similar investigations; since these charac- teristics were not found in the Netherlands strains tested simultaneously it is improbable that the find- ings could be attributed to the methods used. Not only was relative resistance found in two- thirds of the strains, but a more noteworthy finding was that it was present at a high level in the majority (MIC = 0.5 or 1 IU/ml). In most countries, and in the strains from the Netherlands, the majority of strains with relative resistance have been in the range MIC = 0.1-0.5 IU/ml. High-level RR has been found more frequently in recent years, e.g., in Scan- dinavia (Gundersen, Odegaard & Gjessing, 1969; Juhlin, 1965; Nielsen, 1970; Rantasalo, Salot & Wallenius, 1964). However, it is mostly in develop- ing countries, such as Kenya, that an alarming situa- tion, with more than half of all strains isolated having high-level RR, has been described (Johnson et al., 1970; Keys, Halverson & Clarke, 1969; Reyn, 1969). This frequent high-level RR is combined with a paucity of strains with moderate RR (MIC=0.1 or 0.2 IU/ml), which results in a two-peaked curve. Although a bimodal distribution has been found in numerous investigations, it was never as clearcut as in the Kenyan group (Amies, 1970; Chacko & Yogeswari, 1966; Craddock-Watson, Shooter & Nicol, 1958; Curtis & Wilkinson, 1958; Gjessing & Odegaard, 1964; Gundersen, Odegaard & Gjessing, 1969; Holmes, Johnson & Floyd, 1967; Juhlin, 1965; King, 1958; Nicol, Ridley & Symonds, 1968; Nielsen, 1970; Reyn, Korner & Bentzon, 1958; Smith & Levey, 1967; Storck & Schwarz-Speck, 1969). In contrast to the multitude of strains with an MIC=1 IU/ml those with an MIC=2 IU/ml are fortunately rare, and a strain with higher resistance was never encountered. This is of great importance since it means that treatment with penicillin in high doses, preferably in combination with probenecid, is still possible. In other countries, strains with MICs higher than 2 IU/ml have been reported (Krook & Juhlin, 1965; Reyn, 1969; Reyn, Bentzon & Eriks- son, 1963; Storck & Schwarz-Speck, 1969), but there are few indications that during the last decade the development of relative resistance to penicillin has passed the 0.5-1 IU/ml MIC level to any significant extent. This single favourable finding in an otherwise unpleasant development appears to be true also in developing countries, and the expected " second step" towards resistance levels that would make penicillin treatment impracticable has not yet occurred. The situation with regard to the sensitivity of gonococci to tetracycline (Table 3) might appear to be more favourable if the usual classification, which implies that only strains with an MIC higher than I Mg/ml should be considered as relative resistance, is accepted. This definition appears to be based mainly on the theoretical consideration that a corresponding blood level is easily attained with usual doses of tetracycline. The MIC level at which clinical thera- peutic failures with doses such as 250 mg at 6-hourly intervals may occur has not been sufficiently in- vestigated in gonorrhoea; in our experience this may be at MIC = 1 ,ug/ml. Data for these " slightly less sensitive" strains, as Arya & Phillips (1970) called them, are therefore included in Table 3. In Kenya, tetracycline has been used more often in towns be- cause it is not available in the smaller institutions up- country. This might explain why, no statistically significant differences between town and country 714 ANTIBIOTIC SENSITIVITY OF N. GONORRHOEAE IN KENYA 715 were found for this antibiotic by contrast with penicillin. It is not surprising that resistance to streptomycin and relative resistance to sulfonamides are also common since the use of these drugs has con- tinued in Kenya, although there is no place for them in the treatment of gonorrhoea. Perhaps the most important, and alarming, finding of this investigation is the fact that in Kenya, both the highest proportion of RR strains and the highest level of relative resistance to penicillin was found in remote rural areas and in clinics where 80% of the patients were farmers, most of whom lived far from any town, hospital, or health centre. In the tropics, groups of strains with over 90% of relative resistance had previously been found mostly in urban prostitutes and their clients, especially in and around foreign military bases and harbours (Ho & Chang, 1967; Holmes, Johnson & Floyd, 1967; Johnson et al., 1970; Keys, Halverson & Clarke, 1969; Reyn, 1969; Smith & Levey, 1967). Sequential selection of resistant strains in this environment is obviously very rapid because anti- biotics are readily available and are often misused as a prophylaxis and because promiscuity is high, resulting in a rapid circulation of strains. Such groups might, however, constitute an international " closed-circuit " and the gonococcal strains might not be representative of those circulating in the static population at large; this was found in sailors in the Netherlands.1 However, we found evidence that the reservoir of RR strains in a tropical country includes the average rural population; and if this is confirmed elsewhere, the reservoir is immense. Regional differ- ences in the proportion of RR strains have been described elsewhere (Chalmers & Cornelius, 1970; Krook & Juhlin, 1965; Medical Research Council, 1961) and a sample from a single clinic should not be considered as representative of any country. Our data on relative resistance to the three anti- biotics are in accordance with the results obtained in Kampala, the capital city of Uganda (Arya & Phillips, 1970; Phillips et al., 1969). Different results for relative resistance to antibiotics between strains from untreated men and women has to our knowledge never been described and seems hard to explain. One would expect that the distribution of relative resistalace in strains from men would reflect that of the strains from the promiscuous women who are the source of the infection. In female secondary con- tacts, however, a further selection of RR strains 1 See footnote 1, page 708. before the patients are seen at the clinic is not excluded. In this group, infections may be discovered only after an unknown, but possibly very long, in- terval. This is especially true in Africa, for instance, during antenatal care; the patients may have had specific treatment for another disease, and thera- peutic failure may not have been diagnosed and treated. Public health aspects The outlook for individual treatment and for more systematic campaigns in the developing countries is certainly not good. A high incidence of gonorrhoea and its complications, and a high proportion of RR strains, has now been reported from Kenya. The infection has also been studied extensively in Uganda (Arya & Bennett, 1967, 1968; Arya & Phillips, 1970; Bennett, 1962; Kagwa-Nyanzi, 1970; Kibukamu- soke, 1965a, 1965b; Somers, 1964). Unpublished communications from other African countries in- dicate that the situation may be the same elsewhere. In South-East Asia and in the western Pacific area, relative resistance to antibiotics has been proved to be a widespread problem (Chacko & Yogeswari, 1966; Ho & Chang, 1967; Holmes, Johnson & Floyd, 1967; Johnson et al., 1970; Keys, 1969 ; Maurer & Schneider, 1969; Nelson, 1969; Smith & Levey, 1967; Willcox, 1970). The origin of this development is indicated by tabulations 5 and 6 and Table 5, which show that when populations of sensitive and RR strains co- exist, the latter have a better chance to survive and spread, notably in developing countries with in- adequate medical facilities. The situation in Kenya, as illustrated by an enquiry held in 15 hospitals and clinics, may serve as an example. In 12 of the clinics, routine prescription still consisted of only 600 000 units of procaine penicillin repeated daily for 5 or 6 days, although a large proportion of failures was reported everywhere. Second-line treatment consisted usually of streptomycin (10 clinics) and sulfonamides (4 clinics). Problems of diagnosis in women, which cause difficulties even in countries with very good health services, are even greater in developing countries. It was found in Kenya that personnel in one out- patient clinic had only 13 seconds per patient in which to make the diagnosis (Vogel et al., 1970), and laboratories have to deal with many more important diseases. Moreover, most research done in recent years with a wide range of new antibiotics for gonorrhoea cannot generally be applied in many of 716 A. R. VERHAGEN AND OTHERS the developing countries because the drugs are simply too expensive. It is suggested that the problem in developing countries with regard to gonorrhoea is that although there are adequate medical facilities to bring about the selection of a large proportion of RR strains, the facilities, even more than in the better developed countries, will be inadequate to stop or reverse this trend, except in small, isolated communities, such as those in Greenland (Olsen & Lomholt, 1969). As short-term radical changes in diagnostic and laboratory services might be an unrealistic target, although much is being done for their gradual improvement, the only measure that might offer satisfactory results is the provision of adequate guidance and information regarding the correct dosages of the older and cheaper antibiotics. In the East African countries, the present routine treatment of 3 megaunits of procaine penicillin spread over a week could be given together with probenecid on a single day, and the results would probably be much better. ACKNOWLEDGEMENTS The authors thank Mrs E. van Wijngaarden-Ooms, Miss M. J. de Wit, and Mr P. Mutua, laboratory tech- nicians; Dr L. C. Vogel and Dr A. S. Muller, epidemio- logists; Mr W. Gemert and Mr A. M. van Tilborgh, statisticians, and all medical officers and medical assis- tants in Kenya who helped with this investigation. They especially thank Dr L. J. d'Costa, head of the VD clinic in Nairobi, and Dr J. Likimani, Director of Medical Services, Kenya, for permission to publish this paper. RESUME DIMINUTION DE LA SENSIBILITE AUX ANTIBIOTIQUES CHEZ NEISSERIA GONORRHOEAE DANS DES RIEGIONS URBAINES ET RURALES DU KENYA On a procede a des examens bacteriologiques chez 1703 personnes frequentant des etablissements de soins en milieu urbain et rural au Kenya et suspectees d'etre atteintes de blennorragie. Le diagnostic d'infection gonococcique a ete pose dans 902 cas et on a recherche la sensibilite aux antibiotiques de 736 des souches de N. gonorrhoeae isolees. Parmi ces souches, 65% faisaient preuve d'une resis- tance relative a la penicilline, 53,5% a la streptomycine et 7% seulement a la tetracycline; cependant, pour 42% du total des souches, la concentration inhibitrice minimale (CIM) de la tetracycline etait de I pg/mi, valeur proche de celle correspondant a une resistance relative. En general, on notait une resistance associee a I'egard des trois antibiotiques. La resistance relative des souches aux antibiotiques etait tres frequente dans les regions rurales et les endroits eloignes des centres; en ce qui concerne la penicilline, son incidence etait significativement plus elevee en milieu rural (72%) que dans les villes (60,5 %). On a compar6 ces rdsultats a ceux obtenus avec des souches de N. gonorrhoeae originaires des Pays-Bas etudiees simultanement et aux donnees recueillies dans d'autres pays. I1 en ressort que les valeurs de la CIM sont elevees (0,5-2 UI/ml) pour la tres grande majorite des souches kenyennes relativement resistantes a la penicilline. Le fait que ce haut degre de resistance relative est parti- culierement frequent dans les zones rurales, oui vivent la plupart des habitants, donne a craindre que l'Afrique tropicale ne constitue un jour un immense reservoir de souches resistantes. Cette evolution semble resulter de l'administration persistante de doses insuffisantes d'anti- biotiques, la proportion des souches relativement resis- tantes etant beaucoup plus elevee parmi les malades traites sans succes. Les auteurs examinent les moyens de remedier a cette situation. REFERENCES Allen, E. S. (1970) Brit. J. vener. Dis., 46, 334-335 Amies, C. R. (1969) Brit. J. vener. Dis., 45, 216-222 Arya, 0. P. & Bennett, F. J. (1967) Brit. J. vener. Dis., 43, 275-279 Arya, 0. P. & Bennett, F. J. (1968) Brit. J. !ener. Dis., 44, 160-165 Arya, 0. P. & Phillips, 1. (1970) Brit. J. vener Dis., 46, 149-152 ANTIBIOTIC SENSITIVITY OF N. GONORRHOEAE IN KENYA 717 Bennett, F J. (1962) E. Afr. med. J., 39, 332-342 Caterall, R. D. (1970) Brit. J. vener. Dis., 46, 122-124 Cave, V. G., Bloomfield, R. D., Hurdle, E. S., Gordon, E. W. & Hammock, D. (1969) J. Amer. med. Ass., 210, 309-311 Chacko, C. W. & Yogeswari, L. (1966) Indian J. med. Res., 54, 823-838 Chalmers, E. & Cornelius, C. E., III (1970) Brit. J. vener. Dis., 46, 330-333 Craddock-Watson, J. E., Shooter, R. A. & Nicol, C. S. (1958) Brit. med. J., 1, 1091-1092 Curtis, F. R. & Wilkinson, A. E. (1958) Brit. J. vener. Dis., 34, 70-81 Ellner, P. D. (1969) Amer. J. clin. Path., 52, 174-175 (ijessing, H. C. & Odegaard, K. (1964) Acta derm.- venereol. (Stockh.), 44, 132-136 Gundersen, T., Odegaard, K. & Gjessing, H. C. (1969) Brit. J. vener. Dis., 45, 235-237 Guthe, T. (1961) Bull. Wld Hlth Org., 24, 297-306 Ho, T. & Chang, S. (1967) Far East med. J., 3, 20-324 Holmes, K. K., Johnson, D. W. & Floyd, T. M. (1967) J. Amer. mzed. Ass., 202, 461-466 Johnson, D. W., Kvale, P. A., Afable, V. I., Stewart, S. D., Haverson, W. C. & Holmes, K. K. (1970) New Entai. J. Mfed., 283, 1-6 Juhlin, I. (1965) Acta derm.-venereol. (Stockh.), 45, 223- 230 Kagwa-Nyanzi, J. A. (1970) E. Afr. med. J., 47, 159-162 Keys, T. F., Halverson, C. W. & Clarke, E. J. (1969) J. Amer. med. Ass., 210, 857-861 Kibukamusoke, J. W. (1965a) Trans. roy. Soc. trop. Med. Hyg., 59, 642-648 Kibukamusoke, J. W. (1965b) Brit. J. vener. Dis., 41, 135-136 King, A. J. (1958) Lancet, 1, 651-657 Krook, G. & Juhlin, I. (1965) Acta derm.-venereol. (Stockh.), 45, 242-253 Lucas, J. B., Price, E. V., Thayer, J. D. & Schroeter, A. (1967) New Engl. J. Med., 276, 1454-1459 Maurer, L. H. & Schneider, T. J. (1969) J. Amer. med. Ass., 207, 946-948 Medical Research Council (1961) Lancet, 2, 226-230 Nelson, N. (1969) Publ. Hlth Rep. (Wash.), 84, 980-984 Nicol, C. S., Ridley, M. & Symonds, M. A. E. (1968) Brit. J. vener. Dis., 44, 315-317 Nielsen, R. (1970) Brit. J. vener. Dis., 46, 153-155 Olsen, G. A. & Lomholt, G. (1969) Brit. J. vener. Dis., 45, 144-148 Phillips, I., Fernandes, R., Pirani, A. A. & Wagaine, D. (1969) E. Afr. med. J., 46, 38-45 Rantasalo, I., Salo, 0. P. & Wallenius, J. 0. (1964) Brit. J. vener. Dis., 40, 273-277 Rees, E. & Annels, E. H. (1969) Brit. J. vener. Dis., 45, 205-215 Reyn, A. (1969) Bull. Wid Hlth Org., 40, 257-262 Reyn, A., Bentzon, M. W. & Eriksson, H. (1963) Acta path. microbiol. scand., 57, 235-245 Reyn, A., Korner, B. & Bentzon, M. W. (1958) Brit. J. vener. Dis., 34, 227-239 Robinson, M. H., Hicks, C. & Davidson, G. (1970) Publ. Hith Rep. (Wash.), 85, 390-392 Schmale, J. D., Martin, J. E. & Domescik, G. (1969) J. Amer. med. Ass., 210, 312-314 Smith, D. D. & Levey, J. M. (1967) Med. J. Aust., 1, 849-850 Somers, K. (1964) Brit. J. vener. Dis., 40, 194-195 Storck, H. & Schwarz-Speck, M. (1969) Dermatologica (Basel), 139, 254-259 Thayer, J. D. & Martin, J. E. (1966) Publ. Hith Rep. (Wash.), 81, 559-562 Vogel, L. C., Dissevelt, A. G., Quadros, F. C. & Gemert, W. (1970) Bull. int. Epidem. Ass., 20, 171-178 WHO Chronicle, 1965, 19, 7 Willcox, R. R. (1970) Brit. J. vener. Dis., 46, 217-242