Avoiding missed opportunities for immunization in the Central African Republic: potential impact on vaccination coverage.

Avoiding missed opportunities for immunization in the Central African Republic: potential impact on vaccination coverage J.G. Kahn,1 A.H. Mokdad,2'3 M.S. Deming,4 J.-B. Roungou,5 A.M. Boby,5 J.-L. Excler,6 & R.J. Waldman7 Quantified in the study are the extent of missed opportunities for immunization and the potential increases in vaccination coverage and timeliness that could be achieved by using all health centre visits to administer childhood vaccinations in the Central African Republic. The data were collected during a national vaccination coverage survey of 642 children aged 12-23 months from three areas: rural, urban, and the capital, Bangui. Dates of all vaccination visits and other health centre visits were obtained from combined vaccination/health cards. Nationwide, 70% of all opportunities for valid measles vaccination were missed. Of these, 28% occurred at visits when at least one vaccine was given, while 72% occurred at other health centre visits. If there had been no missed opportunities to administer all vaccinations due when at least one vaccine was given, the coverage would have increased from 53% to 67% for the diphtheria-pertussis- tetanus series, from 54% to 70% for measles, and from 34% to 59% for all antigens. If there had been no missed opportunities at any visit, the corresponding increases would have been to 70%, 76%, and 65%. For measles, 46% of the potential increase depends on recognizing that an earlier dose of the vac- cine was invalid and on revaccinating. Days-at-risk for measles (after the age of 270 days) would have been reduced by a mean of 74 days per subject with a health card had no opportunities been missed. The method used serves as a valuable adjunct to evaluations of missed opportunities based on exit interviews at health facilities. It may be feasible and useful to incorporate it in vaccination coverage sur- veys in other countries where all health facility visits are recorded on similar home-based cards. Introduction In 1986, the Central African Republic initiated an accelerated immunization programme, which became fully operational in 1988. As part of the programme, a policy of vaccinating eligible children at all health facility contacts was adopted (referred to subsequent- ly as the "all-visits" policy). National surveys conducted in 1985 and in 1989 indicated that there I Institute for Health Policy Studies and Department of Epidemi- ology and Biostatistics, University of California, San Francisco, CA, USA. Requests for reprints should be sent to Dr Kahn at the following address: Institute for Health Policy Studies, 1388 Sufter St., 11th floor, San Francisco, CA 94109, USA. 2 Department of Epidemiology, Emory University, Atlanta, GA, USA. 3 Chronic Disease Prevention Branch, Nutrition Division, Cen- ters for Disease Control and Prevention, Atlanta, GA, USA. 4 International Heafth Program Office, Centers for Disease Con- trol and Prevention, Atlanta, GA, USA. 5 Directorate of Preventive Medicine, Ministry of Health, Central African Republic. 6 Pasteur Merieux Serums & Vaccins, Paris, France. 7World Health Organization, Geneva, Switzerland. Reprint No. 5567 had been a substantial increase in vaccination cover- age. For example, coverage of 1 2-23-month-olds with measles vaccine increased from 30% in 1985 to 57% in 1989.a However, in 1989 the Ministry of Health obtain- ed evidence that the "all-visits" policy was not being widely implemented. In 1990, the national vaccination coverage survey required interviewers to record all dates of health facility visits for curative care and growth monitoring. The Ministry of Health undertook this expanded survey for the following reasons: to estimate the maximum gain in vaccina- tion coverage that could be realized by using all opportunities for vaccination; and to establish base- line data for monitoring future changes in missed vaccination opportunities, thus allowing evaluation of new vaccination policies and health worker train- ing programmes. The present article describes the methods used and results obtained in the 1990 mis- sed opportunity analysis. a Ministry of Health, Central African Republic, and Interna- tional Health Program Office, Centers for Disease Control. Une enquete nationale sur la couverture vaccinale. Unpublished document, Ministry of Health, Central African Republic, 1991. Bulletin of the World Health Organization, 1995, 73 (1): 47-55 © World Health Organization 1995 47 J.G. Kahn et al. Methods A stratified national vaccination coverage survey of children aged 12-23 months was conducted in Cen- tral African Republic from 14 May to 31 May 1990. Included were children from villages and small towns (the "rural" stratum); from towns and cities with populations of 5000-40 000 (the "urban" stra- tum); and from Bangui (the national capital). With one exception the sampling procedures followed WHO guidelines.b The starting point in each cluster was selected randomly from houses enumerated in a wedge-shaped area (expanding from the centre of the cluster to its outside edge), rather than from houses located along a line between the centre and the edge. This method was considered to be more effective in equalizing the likelihood of choosing houses in the periphery and at the centre of the cluster. The dates of all childhood vaccinations (BCG, diphtheria-pertussis-tetanus (DPT) 1-3, poliovirus vaccine 0-3, measles, and yellow fever) and all other health facility visits (for growth monitoring and ill- ness episodes) were copied from the children's health booklets (referred to subsequently as "cards"). In the Central African Republic, one health card is used to record the dates of all health facility visits. The dates when vaccinations are administered are recorded in one section of the card, and those of health facility visits for all other reasons are noted in other sections. Maternal recollection of additional vaccination doses was also recorded, but we omitted such data since we could not verify the doses or specify the dates when they were administered. Only card-documented vaccine doses were included in the analysis. Adherence to the all-visits policy was expressed as the proportion of all opportunities to vaccinate that were not missed. Missed opportunities were fur- ther categorized as occurring during either vaccina- tion visits (including visits that combined vaccina- tions and other services) or other visits. For DPT3, the analyses were restricted to children who had received the prior doses (whether valid or not). For measles, adherence was calculated in two ways - including and excluding missed opportunities that followed an invalid measles dose (administered before a child was aged 270 days). The latter exclu- sion reflects the likelihood that health workers recog- nized the complete absence of measles vaccination more readily than the incorrect timing of a recorded dose. b Training for mid-level managers: facilitator guidelines for coverage surveys. Unpublished document WHO/EPI/MLM/FGC/ 88, 1988. Vaccination coverage was reported using three proportions, as described below. For each propor- tion, the denominator was taken to be all subjects, including those with and without health cards. * Valid coverage was defined as the percentage of subjects whose vaccination dates met the following minimum age and interval requirements of the Cen- tral African Republic's Expanded Programme on Immunization: DPT1/poliovirus vaccine 1 should be given at or after 6 weeks of age; measles vaccine, at or after 9 months of age (270 days was used in the analysis); and that the interval between the first and second, and the second and third doses of DPT and oral poliovirus vaccine be at least 4 weeks. If the DPT or poliovirus vaccine doses were non-valid, the subsequent doses were renumbered accordingly, e.g., reported dose 2 became valid dose 1 if reported dose 1 had been given too early. Vaccines given after a child was 1 year of age were accepted if otherwise valid. * Simultaneous coverage was defined as the per- centage of subjects who could have received speci- fied valid vaccinations if all needed vaccines had been given on the dates when at least one vaccina- tion was administered. For the analysis, these visits were termed "vaccination visits", although they could have been initiated for another reason. For example, using this measure all children who received any vaccination after 270 days of age would be considered as vaccinated for measles. Simultane- ous coverage was calculated first because it requires only minor behavioural changes by health workers, since the children were already receiving some vac- cination services and had their vaccination cards checked. * Finally, all-visits coverage was defined as the per- centage of subjects who could have received valid vaccinations if all documented health centre visits (including vaccination visits) had been used to vacci- nate eligible children. The difference between all visits and simultaneous coverage represents the addi- tional increase in coverage caused by taking advan- tage of non-vaccination health centre visits, after already avoiding missed opportunities at vaccination visits. For a child with missed opportunities for the same vaccine dose at both vaccination and non- vaccination health centre visits, the increase in coverage was credited to simultaneous vaccination. This method minimizes the coverage gain attributed to non-vaccination visits. A missed opportunity for vaccination was defined as a vaccination visit or other health centre visit by a child who did not receive a vaccination for which he or she was eligible. Health cards do not record the presence of contraindications to vaccina- 48 WHO Bulletin OMS. Vol 73 1995 Avoiding missed opportunities for immunization: impact on vaccination coverage tion. Immunization policy in the Central African Republic encourages the immunization of all age- eligible children, unless they are sick enough to war- rant hospitalization. In our analysis, eligibility was decided solely on the basis of age, since no informa- tion on hospitalizations was collected. Missed oppor- tunities for simultaneous vaccination would be unaf- fected, since by definition these children were considered well enough to vaccinate. Full coverage was defined as BCG, DPTI-3, poliovirus vaccine 1-3, and measles, in keeping with the Ministry of Health definition. Days-at-risk for measles was defined as the age in days at measles vaccination minus 270 days for children who had received a valid measles vaccina- tion. For children who had not received a valid measles vaccination, days-at-risk for measles was defined as their age in days on 31 May 1990 minus 270 days. Means for this variable include only subjects with health cards. Vaccination coverage at specific ages in months was calculated using only children who had reached these ages at the time of the survey. Coverage was calculated only up to 21 months of age because of diminishing sample size beyond this age. Data were analysed at the Centers for Disease Control and Prevention using Statistical Analysis System (SAS) version 5.18 (1) and PROC SESU- DAAN software (2). National results for all means and percentages were weighted by the proportion of the population of the Central African Republic living in each of the study strata, adjusted for slight differ- ences in area sample sizes. The proportions used were as follows: 0.578 (rural), 0.237 (urban), and 0.184 (Bangui).c Results Vaccination information was obtained for 642 chil- dren, 84.5% of whom (national weighted mean) had a health card (Table 1). Among children with a health card, the mean number of vaccination visits was 4.1 and the mean number of health facility visits for other reasons was 8.8. A total of 72% had at least one card-documented health facility visit for non-vaccination purposes; 36.7% had at least one such visit after 270 days of age (the minimum age for measles vaccination). c Bureau of the Census, Central African Republic. Prelimi- nary findings from 1989 national census. Unpublished data pre- pared for the Ministry of Health, Central African Republic. Adherence to the all-visits policy Nationwide, 61% of all opportunities for valid DPT1 and 62% of all opportunities for valid DPT3 vaccina- tion were missed (Table 2). For measles, 70% of all opportunities for valid vaccination were missed nationwide. These proportions were similar in the three geographical areas but were slightly higher in urban areas and Bangui. For DTP1, 19% of the missed opportunities nationwide occurred at visits when another vaccine was given, and 81% when no other vaccine was given (Table 3). The results for DPT3 were similar, although only 11% of missed opportunities occurred at the visits when another vaccine was given. A total of 28% of missed opportunities for measles vaccina- tion occurred at visits when another vaccine was given, while 72% were at non-vaccination visits. For all three vaccines, the rural area had the highest pro- portion of missed opportunities that occurred when another vaccine was given, but in all the study strata most missed opportunities occurred at non-vaccina- tion visits. Nationwide, 70% of missed opportunities for measles vaccination occurred when no previous mea- sles dose had been given, and 30% occurred when an earlier dose was invalid (Table 4). The proportion that occurred when no previous dose had been received was higher in Bangui (80%) than in rural and urban areas. Vaccination coverage: valid and potential The highest valid coverage levels for all antigens were found in Bangui, followed by the urban and then the rural area. DPT3, measles, and full vaccina- tion coverage levels are shown in Table 5 and Fig. 1. If vaccination visits had been used to administer all DPT doses that were due, national valid coverage for the full DPT series would have increased from 53% to 67%; avoiding missed opportunities at all health facility visits would have further increased valid coverage to 70%. Thus, once all missed opportuni- ties at vaccination visits were avoided, the more numerous missed opportunities at non-vaccination visits yielded little additional benefit. National valid vaccination coverage for measles would have in- creased from 54% to 70% by using vaccination visits, and to 76% by using all health facility visits. The biggest potential improvements in measles coverage from using all health facility visits were in rural areas (48% to 70%) and in urban areas (57% to 79%). The greatest potential increase of a coverage measure was for full vaccination coverage (34% to 65%); this increase was greatest in urban areas (38% to 75%). WHO Bulletin OMS. Vol 73 1995 49 J.G. Kahn et al. Table 1: Possession of a health card and frequency of vaccination and other health facility visits among the study sample of 12-23-month-olds, Central African Republic, 1990 No. in area: Rural Urban Bangui Nationala Total sample size 220 210 212 No. with health cards 174 (79.1)b 184 (87.6) 209 (98.5) (84.5) Frequency of health centre contacts: Mean vaccination visits per subject with card 3.5 4.6 5.1 4.1 Mean non-vaccination visits per subject with card 5.1 13.0 13.6 8.8 Total 8.6 17.6 18.7 12.9 No. with .1 non-vaccination health facility visits: At any age 92 (52.9)c 170 (92.4) 207 (99.0) (72.0) At .270 days of age 44 (25.3) 87 (47.3) 114 (54.5) (36.7) a Weighted data. b Figures in parentheses are percentages. c Percentage of those with cards. Potential gains in measles coverage reflected missed opportunities of several types. As noted above, most potential increases resulted from simul- taneous vaccination, leaving little potential benefit from non-vaccination visits. The main exception was measles coverage in Bangui and other urban areas: in such areas, vaccines other than measles were usually administered before the minimum age of 270 days for measles vaccine, providing few chances to administer this vaccine. Also, most increases (54%) represented vaccinations of children who had never been vaccinated, rather than replacing vaccinations that had been days of age). administered too early (prior to 270 Days-at-risk for measles If all health facility visits had been used as opportu- nities to vaccinate, days-at-risk for measles could potentially have been reduced nationally by a mean of 74 days per child with a health card (Table 6). The overall potential reduction was greatest in rural areas and lowest in Bangui, but the geographical pat- tem of findings varied by measles vaccination status. For all children who had a documented valid measles Table 2: Frequency of missed opportunities for vaccination, by antigen, among the study sample of 12-23-month-olds, Central African Republic, 1990 No. in area: Antigen Rural Urban Bangui National DPT1 No. of missed opportunities 223 (57)a 291 (64) 415 (68) (61) No. of valid vaccinations 167 (43) 164 (36) 191 (32) (39) Total 390 445 606 DPT3 No. of missed opportunities No. of valid vaccinations Total Measles No. of missed opportunities No. of valid vaccinations Total 143 (58) 104 (42) 247 228 (68) 105 (32) 333 302 (69) 137 (31) 439 348 (75) 119 (25) 467 311 (64) 173 (36) 484 319 (67) 154 (33) 473 (62) (38) (70) (30) a Figures in parentheses are the percentages of the total opportunities to vaccinate for each geographical area. WHO Bulletin OMS. Vol 73 199550 Avoiding missed opportunities for immunization: impact on vaccination coverage Table 3: Frequency of missed opportunities for vaccination, by antigen and whether another vaccine was given, among the study sample of 12-23-month-olds, Central African Republic, 1990 No. in area: Rural Urban Bangui Nationala Missed opportunity for DPT1 Another vaccine was given 47 (21)b 46 (16) 61 (15) (19) No other vaccine was given 176 (79) 245 (84) 354 (85) (81) Total 223 291 415 Missed opportunity for DPT3C Another vaccine was given 16 (11) 30 (10) 31 (10) (11) No other vaccine was given 127 (89) 272 (90) 280 (90) (89) Total 143 302 311 Missed opportunity for measles Another vaccine was given 86 (38) 40 (11) 51 (16) (28) No other vaccine was given 142 (62) 308 (89) 268 (84) (72) Total 228 348 319 a Weighted data. b Figures in parentheses are the percentages of the total missed opportunities for each geographical area. c Analysis for DPT3 is restricted to children who had received DPT2 (whether valid or not). vaccination, the mean potential reduction was 31 days; for the subset of children who had a valid measles vaccination but who could have been vacci- nated earlier (i.e., excluding those without an oppor- tunity for earlier vaccination), it was 70 days. For children lacking a valid measles vaccination, the mean reduction was 151 days; for those who had at least one missed opportunity for valid measles vacci- nation, it was 220 days. Measles vaccination coverage versus age The improvement in timeliness of measles vaccina- tion is shown in Fig. 2. Maximum vaccination levels were higher for Bangui than for urban areas, and much higher than for rural areas; in addition, the vaccinations were more timely, with more of the vaccines being given closer to the ideal age of 9 months. This is illustrated by the steep rise in the curve and by the early plateau for Bangui and, to a lesser extent, for urban areas. In rural areas, coverage increased much more slowly below 1 year of age, but continued to increase steadily over subsequent months. Valid coverage was lowest and rose the slowest. While potential coverage using all vaccination visits was much higher than valid coverage, potential coverage using all health centre visits was higher still. The elimination of missed opportunities at non- vaccination health centre visits provided the greatest additional gain in coverage in urban areas, but rela- tively little additional gain in rural areas. Discussion The analysis presented here quantifies the frequency of missed opportunities and the potential for vaccina- tion at every health facility visit to increase vaccina- tion coverage and to reduce days-at-risk for disease. The proportion of opportunities for measles vaccina- tion that were missed (70%) indicates that many children had many missed opportunities, even if they received other vaccinations and had no recorded evi- Table 4: Frequency of missed opportunities for measles vaccination, by vaccination history, among the study sample of 12-23-month-olds, Central African Republic, 1990 No. in area: Rural Urban Bangui Nationala Dose not received previously 155 (68)b 235 (68) 255 (80) (70) Earlier dose invalid 73 (32) 113 (32) 64 (20) (30) Total 228 348 319 a Weighted data. b Figures in parentheses are the percentages of the total missed opportunities for each geographical area. WHO Bulletin OMS. Vol 73 1995 51 J.G. Kahn et al. Table 5: Valid vaccination coverage and potential increase In such coverage by avoiding missed opportunities for simultaneous vaccination and all missed opportunities for vaccina- tion, among the study sample of 12-23-month-olds, Central African Republic, 1990 % in area: Rural Urban Bangui National Full DPT seriesa Valid doses noted on card 44 60 75 53 Potential coverage using: - vaccination visits only 57 75 93 67 - all health centre visits 59 78 95 70 Measles Valid dose noted on card 48 57 73 54 Potential coverage using - vaccination visits only 68 67 82 70 - all health centre visits 70 79 90 76 Potential increase: -maximum 22 22 17 22 - administering only absent dosesb 10 14 12 12 Maximum potential increase 15 18 20 17 Full coveragec Valid doses noted on card 27 38 54 34 Potential coverage using: - vaccination visits only 51 63 81 59 - all health centre visits 55 75 89 65 Maximum potential increase 28 37 35 31 a Potential coverage for DPT series based on taking advantage of missed opportunities for DPT1, DPT2, and DPT3. b Excludes replacement of invalid doses. c BCG, DPT1-2-3, poliovirus vaccine 1-2-3, and measles (eight doses in total). Fig. 1. a) "Valid" vaccination coverage, b) potential increase in such coverage by avoiding missed opportu- nities for simultaneous vaccination, and c) all opportu- nities for vaccination. National vaccination coverage sur- vey of 12-23-month-olds, Central African Republic, 1990. (R = rural; U = urban; B = Bangui; N = national). R U B N R U B N Measles All 8 doses Potential using Potential using all LI vaccinaton visits El health centre visits dence of prior measles vaccination. This suggests that the all-visits policy was not being widely imple- mented in the period preceding the survey. If all documented health centre visits had been used, national measles coverage would have been increas- ed by more than a third, from 54% to 76%, and days-at-risk for measles, reduced by more than 2 months (74 days). Missing no opportunities for simultaneous vaccination accounted for most of this benefit nationally, and especially in rural areas. The finding that most of the potential increase in vaccination coverage would be achieved using simul- taneous vaccination is a reflection of our method of analysis. Since we assumed that it requires less behavioural change on the part of health care staff to avoid missed opportunities at vaccination visits than at other health facility visits, we initially calculated the simultaneous coverage. Thus, for a child with four missed opportunities for measles vaccination at non-vaccination visits and one missed opportunity at a vaccination visit, the gain in coverage was attribut- ed to simultaneous vaccination. Since few children had missed opportunities only at non-vaccination WHO Bulletin OMS. Vol 73 1995 R U B N DPT 3 vValid doses QD i 52 Avoiding missed opportunities for Immunization: impact on vaccination coverage Table 6: Potential decrease in days-at-risk for measles by avoiding all missed opportunities for vaccination among the study sample of 12-23-month-olds, Central African Republic, 1990 Decrease in days-at-risk in area: Rural Urban Bangui National All subjects with cards 78.3 (174)a 72.7 (184) 62.8 (209) 74.0 Children with valid measles vaccination noted on card Children with opportunity 98.9 (40) 49.3 (59) 46.9 (94) 69.6 for earlier vaccination Mean for all children with 37.7 (105) 24.4 (119) 28.6 (154) 30.6 valid vaccination Children without valid measles vaccination noted on card Children with opportunity 209.7 (48) 227.6 (46) 242.3 (36) 219.7 for valid vaccinationb Mean for all children without 145.9 (69) 161.1 (65) 158.6 (55) 151.1 valid vaccination a Figures in parentheses are the total number for the particular subgroup. b Includes those never vaccinated (n=79) and those vaccinated before age 270 days (n=51); results for the two groups were very similar. visits, such visits would have yielded relatively little additional benefit after all missed opportunities at vaccination visits were avoided. Had we initially cal- culated the potential increases in coverage due to avoiding missed opportunities at non-vaccination visits, there would have been less gain associated with doing so at vaccination visits. Using this approach, the potential gain in measles vaccination coverage due to avoiding missed opportunities at vaccination visits would have been 8%, rather than the 16% shown in Table 5. For the full DPT series, the potential gain in coverage associated with vacci- nation visits would have been 2% rather than 14%. Although WHO recommends avoiding missed immunization opportunities and many ministries of health have adopted such a policy, exit interviews at health facilities show that missed opportunities con- tinue to occur frequently; up to 60% of children in some countries have at least one missed opportunity (3). One explanation for the continued missed oppor- tunities is that exit interviews fail to demonstrate how much coverage could be increased by avoiding missed opportunities. As a result, ministries of health may have been uncertain about what priority to attach to the policy. The results of our study have reduced this uncertainty by demonstrating major potential increases in vaccination coverage. One pre- vious study reported that in Mozambique (based on growth monitoring dates) and in Guinea (relying mostly on mothers' histories of visits), 8% and 19%, respectively, of incompletely vaccinated children had sufficient health centre contacts for them to have been fully and correctly vaccinated (4). Thus, the impact of avoiding missed opportunities is potential- ly greater in the Central African Republic than in these two countries. An analysis based on the records of all health facility visits may not be feasible in most settings, since it requires access to data often not recorded on documents kept at home. In the Central African Republic, however, the use of a single health docu- ment to record both vaccinations and other health care enabled us to consider all types of health facility visits. The practice of using different health cards for vaccination, preventive care, and curative care, which is common in other countries, might pose logistical barriers to collecting data on all health cen- tre visits; also, in some countries, there are no home- based records of curative visits. Furthermore, because of the high availability of health cards in the Central African Republic (85% nationally), the missed opportunity analysis could be applied to the majority of children in the survey sample. Even in the absence of information on all health centre visits, much of the potential increase in vaccination cover- age can be documented with data limited to vaccina- tion visits. The potential benefit from simultaneous vaccination can be assessed using COSAS 4.3 soft- ware for vaccination survey analysis, which permits an analysis similar to the method we used.d Our analysis did not attempt to determine the reasons why immunization opportunities were d Expanded Programme on Immunization. COSAS 4.3. Unpublished WHO document, November 1991. WHO Bulletin OMS. Vol 73 1995 53 J.G. Kahn et al. Fig. 2. Measles vaccination coverage, by age, for the study sample of 12-23-month-olds, Central African Republic, 1990. a) Rural area. b) Urban area. c) Bangui. 100 a)80 060 20 900 1 1 1 1 0 9 10 11 12 13 14 15 16 17 18 19 20 100 80 -b) &60 O40 0 0 20- 9 10 11 12 13 14 15 16 17 18 19 20 100 c) 80 M60- co > 40- 0 20 0~~~~~~~~~~~~ 9 10 11 12 13 14 15 16 17 18 19 20 Age in months Valid -a- Potential using + Potential using all vaccination visits health centre visits missed. Among the possibilities reported in other countries are interruption of the cold chain, stock shortages, staff absences, inefficient clinic operation, health worker screening and contraindication errors, cost, and matemal concerns about waiting time and vaccination during illness (3, 5-7). Ideally, commu- nity-based surveys to quantify the extent of missed opportunities and the potential impact of avoiding them should be combined with more focused assess- ments to determine why opportunities are missed and how they can be avoided. Facility-based exit inter- views and health worker supervision can help describe the reasons for missed vaccination opportu- nities (8).e Eliminating missed opportunities is criti- cal when resources are limited and health facility contacts are infrequent - providing few opportu- nities to achieve maximum coverage. Successful attainment of this goal will be easier for vaccination doses not previously administered; recognizing prior doses as invalid will be more challenging. In countries where combined vaccination/health cards are used, the method described in this study can provide a well-documented, population-based assessment of the frequency of missed opportunities and the potential gains in coverage achievable by avoiding them. Efforts to reduce missed opportu- nities should be monitored using this and other meth- ods to demonstrate the extent to which potential increases in coverage can be practically achieved, and to identify the best training, supervisory, and record-keeping techniques for achieving these increases. e Training for mid-level managers: identify missed opportunities. Unpublished document WHO/EPI/MLM/91.7, 1991. Acknowledgements We thank J. Naimoli, Technical Officer, and E. Kiteze, Expanded Programme on Immunization, Central African Republic, for crucial logistic support and suggestions; the survey teams in Central African Republic for their excel- lent work and assistance in the design of the survey ins- trument; the writing seminar at the Institute for Health Poli- cy Studies (UCSF) for critique of an earlier draft; and M. Harvey and D. Boyd for their suggestions on the days-at- risk analysis. This research was funded by the Africa Child Survival Initiative-Combatting Childhood Communicable Diseases Project (698-0421) of the United States Agency for Inter- national Development; the Ministry of Health, Central African Republic; and the United Nations Children's Fund. Dr Kahn received support from the Association of Schools of Public Health, the Agency for Health Care Policy and Research, and the Pew Charitable Trusts. Resume Eviter les occasions manquees de vaccination en R6publique centrafricaine: impact potentiel sur la couverture vaccinale Cette 6tude r6alisee en R6publique centrafricaine 6value l'ampleur des occasions manquees de vaccination, ainsi que l'amelioration potentielle de la couverture vaccinale et du respect du calen- 54 WHO Bulletin OMS. Vol 731995 Avoiding missed opportunities for immunization: impact on vaccination coverage drier vaccinal qui pourrait etre obtenue si l'on pro- fitait de chaque visite a un centre de sante pour effectuer les vaccinations necessaires. Les donnees ont 6te recueillies au cours d'une enquete nationale sur la couverture vacci- nale, selon une version modifiee des directives de l'OMS. L'echantillon se composait de 642 enfants ages de 12 a 23 mois, recrutes dans des zones rurales, des zones urbaines et dans la capitale, Bangui. Les dates de toutes les vaccinations et des autres visites aux centres de sante (pour les contr6les de croissance et les soins curatifs) ont ete obtenues d'apres les carnets de sant6 des enfants. La frequence des occasions manquees, la couverture vaccinale actuelle et la couverture vaccinale qu'il devrait etre possible d'atteindre ont ete calculees d'apres les ages et les intervalles requis pour les diverses doses de vaccin confor- m6ment a la politique nationale de vaccination. Les "jours a risque" pour la rougeole ont ete defi- nis comme le nombre de jours ecoul6s sans vac- cination antirougeoleuse valable, apres qu'un enfant ait atteint l'age de 270 jours. Au total, 84,5% (moyenne nationale pond6- ree) des enfants recrutes dans l'etude poss6- daient un carnet de sante. A l'6chelle du pays, 70% de toutes les occasions de vaccination anti- rougeoleuse ont Wt6 manquees, et se r6partis- saient comme suit: 28% ont ete manqu6es lors de visites comportant l'administration d'au moins un vaccin, et 72% lors d'autres visites au centre de sante. Si aucune occasion n'avait 6t6 man- quee pour administrer toutes les vaccinations dont 1'enfant avait besoin lors d'une visite au cours de laquelle au moins un vaccin avait ete administre, la couverture vaccinale serait passee de 53% a 67% pour la s6rie diphterie-tetanos-coqueluche, de 54% a 70% pour la rougeole, et de 34% a 59% pour 1'ensemble des antigenes. Si aucune occasion n'avait 6t6 manquee, c'est-a-dire lors de toutes les visites au centre de sant6, les chiffres seraient passes, respectivement, a 70%, 76% et 65%. Pour le vaccin antirougeoleux, 46% de I'augmentation potentielle supposerait que l'on reconnaisse la non validit6 d'une dose ant6rieure et que l'on procede a une revaccination. Le nombre de jours a risque pour la rougeole aurait ete reduit en moyenne de 74 jours par sujet pos- s6dant un carnet de sant6 si aucune occasion de vaccination n'avait 6te manquee. En fournissant des estimations bien documen- t6es, obtenues en population, sur la fr6quence des occasions manqu6es de vaccination et leur impact potentiel sur la couverture vaccinale, la m6thode utilisee ici constitue un appoint valable aux evaluations des occasions manqu6es fond6es sur des interrogatoires a la sortie des centres de sant6. Les resultats montrent qu'en Republique centrafricaine, environ les deux tiers de 1'en- semble des occasions de vacciner les enfants sont manqu6es, que l'on obtiendrait une am6liora- tion substantielle de la couverture vaccinale si ces occasions manquees 6taient evitees, que la majeure partie du gain potentiel r6sulterait des occasions non manquees d'administration simulta- n6e de plusieurs vaccins, et qu'environ la moiti6 du gain potentiel de couverture vaccinale antirou- geoleuse pourrait etre obtenue sans qu'il soit necessaire de reconnaitre les doses non valables. 11 serait possible, et utile, d'incorporer ces m6- thodes dans les enquetes de couverture vaccinale realis6es dans d'autres pays ou toutes les visites a un centre de sant6 sont enregistr6es sur un car- net de sant6 conserv6 a la maison. References 1. SAS user's guide, version 5.18. Cary, NC, SAS In- stitute Inc., 1985. 2. Shah BV. SESUDAAN: standard errors program for computing of standardized rates from sample survey data. Research Triangle Institute, NC, USA, 1981. 3. Hutchins SS et al. Studies of missed opportunities for immunization in developing and industrialized countries. Bulletin of the World Health Organization, 1993, 71: 549-560. 4. Cutts FT et al. Obstacles to achieving immunization for all 2000: missed immunization opportunities and inappropriately timed immunization. Journal of tropi- cal pediatrics, 37: 153-158. 5. Expanded Programme on Immunization. Missed opportunities for immunization, Mozambique. Week- ly epidemiological record, 1989, 64(5): 32-34. 6. Cutts FT et al. Determinants of vaccination in an urban population in Conakry, Guinea. International journal of epidemiology, 1991, 20: 1099-1106. 7. Nasseri K et al. Determinants of partial participation in the immunization programmes in Iran. Social sci- ence and medicine, 1990, 30: 379-383. 8. Bryce J et al. Assessing the quality of facility-based child survival services. Health policy and planning, 1992, 7: 155-163. WHO Bulletin OMS. Vol 73 1995 55