Are hard-to-reach populations being reached with immunization services? Findings from the 2005 Papua New Guinea national immunization coverage survey

Page 1

AFs

SCCa

b

c

d

e

a

ARRAA

KIHP

1

vatm

0d

Vaccine 28 (2010) 4673–4679

Contents lists available at ScienceDirect

Vaccine

journa l homepage: www.e lsev ier .com/ locate /vacc ine

re hard-to-reach populations being reached with immunization services?indings from the 2005 Papua New Guinea national immunization coverageurvey

teven Toikilika, George Tugesb, Jamie Laganib, Elis Wafiwareb, Enoch Posanaib, Ben Coghlanc,hristopher Morganc, Rohan Sweeneyc, Nan Millera, Anatoly Abramovd, Anthony Stewartc,. John Clementse,∗

Health Services Support Program, Papua New GuineaNational Department of Health, Papua New GuineaBurnet Institute, Melbourne, AustraliaThe United Nations Children’s Fund (Port Moresby Office), Papua New GuineaSchool of Population Health, The University of Melbourne, Melbourne, Australia

r t i c l e i n f o

rticle history:eceived 5 June 2009eceived in revised form 29 March 2010ccepted 21 April 2010vailable online 6 May 2010

eywords:mmunization coverage surveyard-to-reach populationsapua New Guinea

a b s t r a c t

Objective: To measure immunization coverage among children aged 12–23 months in Papua New Guinea(PNG) and to assess if and why there are differences between hard-to-reach and more accessible com-munities.Methods: WHO cluster sampling methodology was employed to measure immunization coverage in PNG’sfour regions. Survey data were re-analyzed according to a local assessment of geographical accessibilityindicated by census unit type: urban, rural and hard-to-reach. Census units were designated as hard-to-reach if they were five or more kilometres from a health centre.Findings: Nationwide coverage for most antigens falls below the national target of 80% although thereare regional differences with Islands performing the best. Late doses are a major concern: just 4% werefully immunized with valid (“on time”) doses by 1 year of age. Coverage was lower in both rural andremote communities: at 6 months 48% of children from urban units had received three valid doses ofDTP-3 but only 16% in rural areas and 13% in hard-to-reach communities. Reasons for failure to immunizevaried: 21% of mothers said their child was not immunized because distance, travel conditions or cost oftransportation prevented access to local health centres; 27% cited a lack of knowledge or misconceptions

about immunization; while 29% believed it was because of an issue with the health system.Conclusions: Throughout PNG there is an urgent need to increase immunization coverage and to ensurethat children are immunized on time according to the schedule. Both coverage and timeliness of dosesare worse for children living in hard-to-reach and rural areas. Achieving national immunization targetsrequires improvements in health service delivery, including outreach, especially for remote and ruralcommunities, as well as greater community education and social mobilisation in support of immunization services.

. Introduction

Reaching hard-to-reach communities with immunization ser-

ices is a perennial problem for all countries. UNICEF has declared itpriority to ensure that children from the hardest to reach popula-

ions – those with limited or no geographical access, the urban poor,inorities and children in conflict situations – all have access to

∗ Corresponding author. Tel.: +61 3 9775 2931.E-mail address: [email protected] (C.J. Clements).

264-410X/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.oi:10.1016/j.vaccine.2010.04.063

© 2010 Elsevier Ltd. All rights reserved.

immunization [1]. In line with this priority, the World Health Orga-nization (WHO) is revitalizing outreach services to remote ruralcommunities with their “RED” strategy—Reaching Every District[2,3]. And European immunization managers have recently ear-marked hard-to-reach populations as their current area of focus [4].

In Papua New Guinea (PNG), immunization is provided byboth governmental and church-based organizations using a mix

of strategies including health facility-based (provincial hospital,urban clinic, rural health centres/sub-centres, aid posts) and out-reach services. Local conditions, however, make achieving uniformhigh coverage challenging. The political climate can be unpre-

Page 2

4674 S. Toikilik et al. / Vaccine 28 (2010) 4673–4679

Table 1National immunization schedule in 2005.

Vaccine Age

At birth 1 month 2 months 3 months 6 months 9 months 12 months

Tuberculosis (BCG)√

Hepatitis B√ √ √√ √ √ √

dcorwaimdtpamr

e(cb

2

taicdEE

2

ipIulO

wnsfSug(v±c

Oral polioDiphtheria–pertussis–tetanus

√ √MeaslesVitamin A

ictable, and episodic violence within and between communitiesan disrupt service provision. Communication is demanding withver 600 distinct languages spoken. Rugged terrain means that theoad system is not interlinked and is often impassable during theet season; many villages in the mountainous central highlands

re only accessible by foot; and large populations live on scatteredslands. Consequently, most medical equipment and supplies are

oved by expensive air transport and many rural communities faceifficulties in accessing services. Additionally, major decentraliza-ion of government responsibilities has taken place over the decadereceding this survey, often coinciding with insufficient financialllocation to health care in general [5]. In fact, several recent assess-ents have noted a decline in the quality, utilisation and reach of

ural services for child health [6,7].We report findings from the 2005 national immunization cov-

rage survey conducted by the National Department of HealthNDOH) including findings related to how and why immunizationoverage varies according to a community’s geographical accessi-ility.

. Method

During July 2005, the NDOH conducted a household surveyo document the proportion of children aged 12–23 months ofge who had been correctly immunized according to the nationalmmunization schedule (Table 1), and to identify the reasons whyhildren had not received their scheduled doses of vaccine. A stan-ard WHO two-stage cluster survey methodology [6] was used.thical clearance for the study was obtained from the Nationalthical Committee of PNG.

.1. Sampling strategy

We conducted a separate survey in each of PNG’s four admin-strative regions: the Highlands region (with 38% of the nationalopulation of 5.2 million), Momase (27%), Southern (21%), and

slands (14%). In the first stage of sampling, we selected 30 censusnits in each region with a probability proportional to their popu-

ation size from a complete list provided by the National Statisticsffice.

In the second stage, we randomly selected a starting householdithin each census unit using the standard World Health Orga-ization Expanded Program on Immunization (WHO/EPI) clusterampling random walk method [6]. Verbal consent was obtainedrom the head of household to survey the youngest eligible child.ubsequent households were selected on the basis of proximityntil seven infants had been sampled in each selected census unit,

iving a total sample size of 210 infants for each regional survey.This is the standard sample size for immunization coverage sur-eys using the WHO/EPI method and calculated with a precision of10%, a confidence level (Cl) of 95%, and assuming an immunization

overage level of 50%.)

√√ √√ √

2.2. Household data collection and data management procedures

Data collection was supervised by 12 survey supervisors whowere mid-level managers recruited from the NDOH in PortMoresby. They received training and pre-tested the method andquestionnaire in three sites in Southern Region. In each region,supervisors subsequently trained provincial supervisors and localinterviewers drawn from the health and education departmentsusing a standardized training protocol. All trainers and supervisorsspoke English and Pidgin, and interviewers spoke English, Pidginand local dialects.

Interviewers worked in groups of two to survey households:one interviewer conducted the interview while the other observedfor data collection errors. Responses were recorded on a standardform adapted from the WHO protocol. Immunization status wasrecorded as confirmed by card or reported by parents. For thoseinfants not fully immunized, mothers were asked to nominate thesingle main reason why they were not fully immunized. Surveyorsthen matched this with a reason from the standard WHO list. Allmothers were asked where they had delivered their child.

Supervisors checked all data forms and forwarded them to theNDOH in Port Moresby for collation. If errors were detected, thehead of the household was re-visited and re-interviewed. Datawere separately entered into Microsoft Excel in Melbourne and PortMoresby to check for discrepancies.

2.3. Post hoc classification of census units

After data collection, the national immunization programmemanager classified the surveyed census units into one of three cat-egories: urban, rural, and hard-to-reach. Urban census units werelocated within provincial capital towns; rural census units wereoutside the provincial capitals but located within five kilometres ofa health centre; hard-to-reach units were in a rural area and situ-ated more than five kilometres from the nearest health centre. Onezone was categorised as mixed rural, while another two zones couldnot be categorised by the national immunization manager. Thesethree clusters were excluded from analyses by census unit type.

2.4. Data analysis

Data were analyzed using EpiInfo 6.04 and EpiData 3.1. Regionaldata were weighted according to population size to derive nationalimmunization coverage levels. These were compared with histor-ical administrative reports and a demographic and health survey.Table 2 shows the definitions for “valid” doses of antigens that weexamined by region and census type. Throughout the paper, wereport only card-confirmed doses. Note that we have not reportedconfidence intervals for estimates by census unit type because

the post hoc categorisation of units meant only a small numberof clusters was sampled for some types, with a possible loss ofrepresentativeness.

We aggregated the reasons for failure to immunize children intofour distinct categories (1: lack of knowledge or misconceptions; 2:

Page 3

S. Toikilik et al. / Vaccine 28 (2010) 4673–4679 4675

Table 2Definitions of valid doses.

Antigen(s) Abbreviation Definition

Bacillus Calmette–Guérin BCG Birth to 90 days of age

Hepatitis BHepB0 Dose must be given within 24 h of birthHepB1–2 Three doses given by 6 months of age with at least 28 days between doses

Oral polioOPV-0 Dose must be given within 14 days of birthOPV-1–3 Four doses given by 6 months of age with at least 28 days between doses

Diphtheria–tetanus–pertussis DTP-1–3 Three doses given by 6 months of age with at least 28 days between doses

MeaslesM6 Dose 1 given between 6 months (180 days) and 9 months of age (270 days)M9 Dose 2 given 28 days after dose 1 and by 12 months of age

Table 3Distribution of infants by census unit type and region.

Region Type of census unit TotalNumber of infants (number of clusters)

Urban Rural Hard-to-reach Mixed Not classified

Islands 21 (3) 168 (24) 7 (1) – – 196 (28)Highlands 34 (5) 121 (18) 49 (7) – – 204 (30)

1)3)

2)

i4

3

(itat

a86m(r

3

m2da

TN

a

ab

Momase 56 (8) 111 (16) 7 (Southern 42 (6) 125 (18) 21 (

Total 153 (22) 525 (76) 84 (1

ssues with the health service; 3: transport and travel challenges;: family issues) and analyzed by census unit category.

. Results

We surveyed 783 children in 113 clusters across the four regionsTable 3). Data collection errors led to four clusters being missedn Momase, two in the Islands and one in Southern region, and lesshan seven infants being surveyed in three clusters in the Highlandsnd one each in Momase and Southern region. No families refusedo participate in the survey.

Over 93% of the children sampled had a valid immunization cardt the time of the survey (Islands 100%, Highlands 93.6%, Momase7.3%, Southern 92.6%). Census units in rural areas constituted7.3% (76/113) of all census units surveyed; those in urban areasade up 19.4% (22/113); and hard-to-reach communities 10.6%

12/113). Seven of the hard-to-reach areas were in the Highlandsegion.

.1. Immunization coverage

Survey data suggested that coverage had slightly improved forost antigens since the earliest available administrative report in

002 (Table 4), although these methods are difficult to compareirectly. Measles and BCG were important exceptions, however,nd coverage may actually have declined for these vaccines since

able 4ational immunization coverage: comparison of survey data with historical reports, 1995

Method Year BCG DTP-1 DTP-3

Demographic andhealth surveya

1995 91% 79% 47%

Administrative reportsfrom health facilities

2002 74% 81% 61%2003 76% 88% 68%2004 74% 81% 62%2005 73% 80% 61%

Cluster surveyb

(confidence intervals)2005 81% (75.5–87.1) 82% (76.2–87.5) 66% (58.8–73.4

Reported in WHO/UNICEF Review of National Immunization Coverage (August 2008) wged 12–23 months.Weighted national averages of card-confirmed doses.

7 (1) – 181 (26)– 14 (2) 202 (29)

7 (1) 14 (2) 783 (113)

the last community based survey in 1995. Only the point estimatesfor BCG and DTP-1 coverage for our 2005 survey exceeded thenational target of 80%, although the true level of coverage for all vac-cines may be higher than we report given we used a conservativemeasure (card-confirmed doses).

The pattern of immunization coverage across the four regionswas the same for all antigens: point estimates were highest inthe Islands followed by Southern region, Highlands and Momase(Fig. 1). Children living in urban areas had better coverage thanchildren in other census types, whereas there was little differencebetween rural and hard-to-reach populations.

3.2. Timeliness of doses

Late delivery of doses was a problem throughout the country.For example, while 66% (CI = 58.8–73.4%) of infants throughout PNGhad received three doses of DTP by 12 months of age, just 21.3%(CI = 16–26%) had received all three doses at the right (‘valid’) time.In fact, based on immunization cards only 29% of infants receivedall scheduled antigens by 12 months of age, and only 16% of thesechildren received valid doses (4.4% of the entire sample).

An examination of the timing of BCG doses provides an illustra-tion of similarities and differences between census units. Coverageof the birth dose of BCG was low across all units: only about onein five neonates was vaccinated within 24 h of delivery (Table 5)despite 56% of surveyed women delivering their baby in a health

–2005.

OPV-3 HBV-3 MCV6 MCV9

47% 57% Introduced in 1996 76%

47% 60% 61% 63%53% 67% 63% 53%49% 60% 60% 50%50% 63% Not reported 49%

) 63% (55.9–70.8) 68% (61.0–75.1) 72% (64.1–78.2) 55% (47.9–62.1)

ithout confidence intervals. Target population was for EPI coverage was children

Page 4

4676 S. Toikilik et al. / Vaccine 28 (2010) 4673–4679

Table 5Timing of card-confirmed BCG doses by type of census unit (unweighted).

Time after birth Urban (n = 132) Rural (n = 412) Hard-to-reach (n = 63) Total (n = 607)

% n % n % n % n

Day of birth 18.9% 25 21.6% 89 19.0% 12 20.8% 1262–7 days 31.8% 42 20.9% 86 11.1% 7 22.2% 1358–30 days 23.5% 31 15.0% 62 14.3% 9 16.8% 10231–90 days 12.1% 16 18.2% 75 36.5% 23 18.8% 114

0.6–2

fHemu5camo

91–180 days 8.3% 11 11.2%>180 days 5.3% 7 13.1%

Median no. days 6.5 days (3–10.8) 17 days (1

acility. The same was seen for the birth dose of hepatitis B (Fig. 2).owever, distance from a local health facility did appear to influ-nce the subsequent delay in receiving BCG vaccine if an infantissed it on the actual day of birth—by 30 days after birth 74.2% of

rban infants had received a dose of BCG vaccine compared with7% of rural and only 44.4% of hard-to-reach infants. This trend is

lear from the median time to delivery of BCG: 6.5 days in urbanreas, 17 days in rural villages and 32 days in hard-to-reach com-unities. Fig. 2 shows that this pattern holds true for all antigens

n the national schedule.

Fig. 1. Immunization coverage (card confirmed) by re

46 14.3% 9 10.9% 6654 4.8% 3 10.7% 64

6) 32 days (13–46) 15 days (10–21)

3.3. Reasons for failure to immunize by census unit type

Mothers of infants who were not fully immunized at the time ofthe survey (402) were asked to list the single most important rea-son for this (Table 6). Over 20% of all women said that they could notmake it to the local health centre because of distance, travel con-

ditions or cost of transportation. Unexpectedly, travel challengeswere less commonly identified as the most important barrier forwomen from hard-to-reach census units (11%) compared to thosefrom urban (17%) and rural (23%) areas. A lack of knowledge about

gion and census unit type for all children, 2005.

Page 5

S. Toikilik et al. / Vaccine 28 (2010) 4673–4679 4677

Fig. 2. Immunization coverage of “valid” doses of antigen by census unit type, 2005 (unweighted).

Table 6Principal reason given by mother for why a child had not been fully immunized, by census unit.

Reason for child immunization failure Census unit

Grouping Specific reason Hard-to-reach Rural Urban Total

Lack of knowledge or misconceptions Unaware of need for immunization 4 20 7 31Place/time of immunization unknown 6 20 4 30unaware of need for second/third dose 3 8 4 15No faith in immunization 5 10 2 17Wrong ideas about contraindications – 5 – 5Husband refused to allow immunization 1 2 – 3Fear of side effects – 4 1 5Rumours 1 1 – 2

Group total 20 70 18 108and column percentage 38% 24% 29% 26.9%

Issues with health services Vaccines not available 2 30 4 36Vaccinator did not show up 1 25 3 29Long waiting time – 9 2 11child ill, brought but not immunized 1 8 2 11Immunization clinic postponed 8 20 3 31

Group total 12 92 14 118and column percentage 23% 32% 22% 29.4%

Transport and travel challenges Place of immunization too far 3 46 9 58Weather/travel conditions 3 14 1 18Cost of transport too high – 5 1 6Security problem – 2 – 2

Group total 6 67 11 84and column percentage 11% 23% 17% 20.9%

Family concerns Time of immunization inconvenient 4 13 8 25Mother too busy 3 9 2 14Family problems incl. mother’s illness 5 4 2 11Child ill, not brought 2 – 1 3Could not leave older child – 1 – 1

Group total 14 27 13 54and column percentage 26% 9% 21% 13.4%

Other/unknown Other – 21 – 21Unknown 1 9 7 17

Group total 1 30 7 38and column percentage 2% 10% 11% 9.5%

Total 53 286 63 402

Page 6

4 cine 2

wta(aba

4

nttBcnc

bwticsttgWtfhsmob

tnivtp[ccmBfi

difdehdbf

aatrae

678 S. Toikilik et al. / Vac

here to access immunization services, or misconceptions abouthe need and safety of immunization were more frequently citeds the most important barrier to immunization for all mothers38% from hard-to-reach areas; 29% urban; 24% rural settings). Inlmost one in three instances, a child had not been fully immunizedecause of a reported issue with the health system. This varied fromlow of 22% in urban areas to a high of 32% in rural regions.

. Discussion

This survey highlights a number of challenges faced by theational immunization programme in PNG, challenges that appearo be having a greater impact on immunization coverage andimeliness of doses in rural and remote parts of the country.oth rural and hard-to-reach sites suffer similar levels of pooroverage—suggesting that the major point of inequity in immu-ization services is between urban settings and the rest of theountry.

National immunization coverage by 1 year of age remainedelow national targets for most vaccines, although some regionsere performing better than others. While most of the popula-

ion in the Islands region lived in rural areas distant from the mainsland of PNG, shorter distances and less hostile terrain meant thatommunities tended to be better linked to health services on eachmall island with resulting high immunization coverage. In con-rast, the coverage in Momase was the lowest in the country, evenhough it had just one cluster designated as hard-to-reach, sug-esting issues in the organization and management of services.ith higher numbers of remote communities and similarly difficult

opography to Momase, Southern and Highlands regions outper-ormed the northern coastal region. On the other hand, the Islandsad the highest proportion of children who failed to receive theecond dose of measles vaccine at 9 months after receiving the 6-onth dose. Clearly, there are lessons to be shared between regions

n the organization of immunization services and overcoming localarriers.

Administrative data over the years immediately precedinghis survey suggested that there had been few improvements inational coverage since 2002—a year when government and donor

nvestments saw a revitalization of cold chain infrastructure andaccine distribution [8]. More recent administrative data followinghis survey, reported to the United Nations, shows that this trendersists [9]. The earlier Demographic and Health Survey of 19957] is arguably the more appropriate comparator (albeit with noonfidence intervals reported) for our survey. This suggests thatoverage in 2005, as compared with 1995, seemed to improve forost vaccines, but there were apparent declines in coverage for

CG and the 9-month dose of measles. Interestingly, these are therst and last doses on the national schedule.

BCG coverage analysis provides some insights into why birthoses and timely coverage of vaccines are problematic. The dispar-

ty between the proportion of women who gave birth in a healthacility (56%) and the proportion of infants who received the birthose of BCG within 24 h of delivery (21%) (with similar low cov-rage for hepatitis B vaccine) suggests that vaccine availability orealth services organization inhibits scheduled delivery of birth-ose vaccines. These findings imply that there is a need to improveoth collaboration with maternal health services and proceduresor vaccine supply.

Coverage levels of birth doses are better than those deliveredt 2 months of age, which in turn are better than those delivered

t 6 or 9 months of age. There are a variety of potential explana-ions for why coverage of the 9-month dose of the measles vaccineemains low in this survey. However the most likely reason, in ourssessment and that of other commentators, is the fact that par-nts access preventive services less as children get older, and that

8 (2010) 4673–4679

routine health services have great difficulty in conducting regularoutreach in most parts of PNG [10–13]. This problem is nationwidebut worse outside the main towns. This may relate to difficultiesin recording and tracking child populations, lower health serviceutilisation when compared to urban sites or a lack of health sys-tem responsiveness. The latter includes factors such as: a paucityof staff for outreach services; fixed facilities that provide vaccinesonly on certain days and not on demand; and failure to immunizeopportunistically children attending health facilities for other rea-sons. In contrast, there has been an improvement in PNG’s control ofmeasles infection through the use of regular Supplementary Immu-nization Activities that are designed to work flexibly from districtto district in a fashion that supports routine services [12].

Added to the issues of overall coverage levels is the problemof late doses: only 1 child in 25 was immunized according to thenational schedule by 12 months of age. As a result, the majorityof infants in PNG are vulnerable to vaccine-preventable infectionsduring the period of their life when they most need protection.These findings contrast with a recent study in a Highland valleycommunity in 2005 that recorded high coverage (89%) only becauseit included children aged up to 5 years [14]. With the addition tothe infant immunization schedule of new, more expensive vac-cines such as Haemophilus influenzae b (Hib), the issue of timelinessbecomes even more important.

Our survey indicates that less-than-ideal coverage stemmedfrom a mixed picture of health service issues and lack of healthcare access and utilisation—and this is consistent with caregivers’reports of why their children had not been immunized accord-ing to the national schedule. For almost 80% of women, threemain categories of reasons were offered: parental lack of knowl-edge or misconceptions; issues with health services; and problemswith transportation. These reasons varied between types of censusunits. Health authorities had anticipated that difficulties in reach-ing health services would be by far the most important barrier toimmunization, but this was not the case for any census unit type. Infact, in the hard-to-reach areas relatively few mothers cited travelchallenges as the principal reason for immunization failure; lackof knowledge and misconceptions about immunization were moreimportant. Perhaps these women, accustomed to their remote loca-tion, viewed other issues as more pertinent barriers. Improvingtransport and service access may therefore not improve immuniza-tion coverage without concurrent community education. A recentqualitative evaluation of immunization services had already indi-cated a serious need for greater community education and socialmobilisation in support of immunization services [8], and two ear-lier studies in urban PNG also noted mothers’ lack of understandingabout immunization services and schedules [15,16]. Then again, itshould not be forgotten that 17% of urban residents cited travelchallenges as their prime barrier, most saying they lived too farfrom the health centre. This may imply a degree of social as well asgeographical isolation in these settings.

Health service issues were, nevertheless, deemed important bymothers. Nearly one in three mothers said that a problem withthe health service was the major barrier to immunizing their child,confirming other calls for an urgent review of the quality of servicedelivery—Samiak’s 2005 study of 185 children attending 4 healthcentres in Momase and Southern regions found that health staffhad missed an opportunity to immunize 29% of their sample [17]. Aquality service also means delivering potent vaccines—the integrityof the cold chain needs consideration for delivery of vaccines toremote communities [18].

4.1. Limitations

Errors in data collection meant that seven of the intended 120clusters were missed and there were less than the desired num-

Page 7

cine 2

bte

fiTicfatoit

lrfshifniM

5

teotlccetiehhtrantP

A

mi

e

[

[

[

[

[

[

[

S. Toikilik et al. / Vac

er of children in another seven clusters. Even with standardizedraining and practical exercises, a high level of field supervision isssential to maintain quality data collection.

Our survey was not designed to measure coverage across dif-erent types of census units, so there are relatively few clustersn some of these groupings and our analysis was retrospective.his prevents an analysis of census units within regions, and find-ngs on coverage in rural, urban and hard-to-reach areas should beonsidered indicative rather than conclusive. Additionally, distancerom a health facility is not necessarily a reliable indicator of howccessible health services are to the local population. Time taken toravel to a health facility, mode of transport available, and presencer absence of outreach immunization services are arguably betterndicators but not captured in the standard WHO tools we used inhis study.

As the printed questionnaire was not translated into localanguages, misclassification may have occurred particularly witheporting of reasons for failure to immunize. The reasons given forailure to immunize are responses from mothers, restricted to aingle choice from a prescribed list, without cross-checking withealth staff or health facility records. Further quantitative and qual-

tative analyses would be useful for exploring why young childrenail to get immunized. An examination of socially marginalised,ot just geographically hard-to-reach populations, may also be

mportant in rapidly growing urban areas such as the capital, Portoresby.

. Conclusions

Immunization coverage surveys are useful for mapping progressoward national targets, calibrating administrative reports of cov-rage and for exposing areas for improvement. Reaching all sectionsf the community with immunization services is not easy in any set-ing, let alone in Papua New Guinea where geography, climate andimited infrastructure challenge health care providers and healthare seekers alike. While coverage levels are admirable given theseonstraints, PNG needs to not only increase its immunization cov-rage, but to ensure that infants are immunized on time accordingo the schedule, particularly in remote and rural locations. Timings becoming a more pressing concern with the introduction of new,xpensive vaccines. Most populations outside the urban centresave lower vaccine coverage. The reasons why populations remainard-to-reach differ across census unit types and more (qualita-ive) information is required to develop tailored solutions to locallyelevant physical and social barriers to immunization. Addressingll these shortcomings will require action at local, provincial andational levels including an examination of health services andhe support, education and mobilisation of all communities withinNG.

cknowledgements

We would like to thank the many staff from the National Depart-ents of Health and Education who spent long hours collecting data

n difficult field conditions.The National Department of Health also gratefully acknowl-

dges the support in conducting this survey of the following

[

[

8 (2010) 4673–4679 4679

partners: The World Health Organization (WHO Port MoresbyOffice); The United Nations Children’s Fund (UNICEF Port MoresbyOffice); The Australian Agency for International Development(AUSAid); and the Centre for International Health, The Macfar-lane Burnet Institute for Medical Research and Public Health Ltd,Melbourne, Australia. The authors also gratefully acknowledge thecontribution to this work of the Victorian Operational Infrastruc-ture Support Program received by the Burnet Institute.

Conflict of interest: The involvement of NDOH staff in trainingenumerators is a potential conflict of interest. Mid-level staff fromthe National Department of Health were involved in training theenumerators; junior health staff were involved in household selec-tion and data collection. Note, however, that no NDOH staff wereinvolved in cluster selection or data analysis. NDOH has endorsedthe findings of the survey.

References

[1] UNICEF. Vaccinating the hard-to-reach. http://www.unicef.org/immunization/index hardtoreach.html [viewed 2 January 2008].

[2] The RED (Reaching Every District) strategy. http://www.who.int/immunization delivery/systems policy/red/en/ [viewed 2 March 2010].

[3] Immunization in Practice. A practical guide for health staff. Geneva: WorldHealth Organization; 2004. WHO/IVB/04/06.

[4] WHO Programme Managers Meeting: reaching the hard-to-reach.http://www.euro.who.int/vaccine/20070914 5 [viewed 2 January 2008].

[5] NDOH Planning and Administration. Final Annual Sector Review (ASR) report;2006.

[6] Expanded Programme of Immunization. Immunization coverage clustersurvey—reference manual. Geneva: World Health Organization; 2004.WHO/IVB/04.23. http://whqlibdoc.who.int/hq/2005/WHO IVB 04.23.pdf[viewed 2 March 2010].

[7] WHO/UNICEF Review of National Immunization Coverage. Port Moresby, PNG:World Health Organization; August 2008.

[8] International Development Support Services Pty Ltd. (IDSS). Papua New GuineaWomen’s and Children’s Health Project Activity Completion Report. AusAID andthe PNG National Department of Health; 2004.

[9] World Health Organization. Papua New Guinea Reported Immuniza-tion Coverage; WHO vaccine-preventable diseases: monitoring system2009 global summary. http://www.who.int/immunization monitoring/en/globalsummary/timeseries/tscoveragebycountry.cfm?C=PNG [viewed 15March 2010].

10] Duke T. The crisis of measles and the need to expand the ways of delivering theEPI in PNG. PNG Med J 2003;46:1–7.

11] Kurubi J, Vince J, Ripa P, Tefuarani N, Riddell M, Duke T. Immune response tomeasles vaccine in 6 month old infants in Papua New Guinea. Trop Med IntHealth 2009;14:1–7.

12] Morgan C, Hinton R, Pongua G, Aeno H, Auwun A, Clements CJ, et al. Betteruse of public health campaigns for child survival: the impact and operations ofPapua New Guinea’s Supplementary Immunisation Activity 2003–05. ResearchMonograph, Child Health and Nutrition Research Initiative; 2009.

13] Clements CJ, Morgan C, Sakamoto C. Quality assessment—findings of an in-depth evaluation of the immunization programme in Papua New Guinea. PNGMed J 2006;49(1–2):5–13.

14] Vengiau G, Phuanukoonon S, Siba P. Coverage of EPI vaccination inrural community in the Highlands: base in the demographic survey from2005 in Asaro Valley, PNG. In: Presented at 2007 Medical Symposium.2007.

15] Bukenya GB, Freeman PA. Possible reasons for non-completion of immuniza-tion in an urban settlement of Papua New Guinea. PNG Med J 1991;34:22–5.

16] Freeman PA, Thomasen JA, Bukenya GB. Factors affecting the use of immu-nization among urban settlement dwellers in Papua New Guinea. PNG Med J1992;35(3):179–85.

17] Samiak L, Teiptoror L. Factors contributing to missed opportunities to vaccinatechildren in health centres in Papua New Guinea. In: PNG medical conference.2005.

18] Wirkas T, Toikilik S, Miller N, Morgan C, Clements CJ. A vaccine cold chainfreezing study in PNG highlights technology needs for hot climate countries.Vaccine 2007;25:691–7.