Healthcare Quarterly Vol.16 No.1 2013 85 84 Healthcare Quarterly Vol.16 No.1 2013

Comparative Models of Cervical Cancer Screening in Manitoba Linda DeRiviere et al. Linda DeRiviere et al. Comparative Models of Cervical Cancer Screening in Manitoba

1-Conventionalcytology (CC)

2 - LBCThinPrep

3 - LBCSurePath

4 - LBC forprimary

testing & HPVtriage

5 - HPV forprimary

testing & LBC

6 - HPV forprimary

testing & LBCimplementover 5 yrs

7 - HPV forprimary

testing & CC

8 - HPV forprimary

testing & CCimplementover 5 yrs

Chan

ge in

Cos

t per

Spe

cim

en (%

)

4035302520151050-5-10- 15

10.313.3

19.624.4

13.8 15.518.9

24.0 19.0

33.2

-10.9 -9.4 -5.6

7.2 4.70

FIGURE 3. Percentage change in cost per specimen from year one to year five

HPV = human papillomavirus; LBC = liquid-based cytology.

= % change in average total cost per specimen from year 1 to year 5: costs include consumables; wage costs and benefits; and fee tariffs for cytology, colposcopy and other related visits.

Costs are net of savings from a decrease in cases of cervical cancer. Costs exclude capital equipment.

= % change in laboratory cost per specimen from year 1 to year 5: wage costs, benefits and consumables only.

2-LBC, ThinPrep

3 - LBC SurePath

Labour costs

Colposcopy cost

Lengthening screening intervals

4 - LBC forprimary

testing & HPVtriage

5 - HPV forprimary

testing & LBC

6 - HPV forprimary

testing & LBCimplementover 5 yrs

7 - HPV forprimary

testing & CC

8 - HPV forprimary

testing & CCimplementover 5 yrs

Tota

l Cos

t Sav

ings

(in

Thou

sand

s of

Dol

lars

)

20000

18000

16000

14000

12000

10000

8000

6000

4000

2000

0

-2000

43.5 0 0 43.5 0 0 0-954.6 -1,303.1

1,702.7784.6

1,702

16,549.7

18,531.8

11,985.8

12,888.2

16,550.0

18,330.8

11,985.7

12,848.4

784.6

FIGURE 4. Total cost savings compared with conventional cytology over five years

CC = conventional cytology; HPV = human papillomavirus; LBC = liquid-based cytology.

discussed earlier, the fluctuating quantities of specimens each year also affected the percentage change in cost per specimen from year one to year five in model five. The total costs in year one were spread over 205,000 specimens compared with 95,000 in year five, thus affecting the percentage change in cost per specimen over the period. On the other hand, if we considered year four estimates (205,000 specimens) for model five, the average total cost per specimen increased by only 9.4% (10.3% for labour and consumables).

HPV DNA as primary screening, whether coordinated with liquid-based or conventional cytology as secondary screening, had the slowest pace of changing costs per specimen from year one to year five. In fact, some of the estimates decreased over the five-year period. Moreover, it was anticipated that the costs per specimen for HPV DNA testing would decline over time as further companies received regulatory approval for HPV genotyping tests. Cost savings were primarily attributed to labour reductions, longer screening intervals for HPV-tested women, fewer colposcopies over the long run and an expected decrease in cases of cervical cancer requiring hospital treatment. In sum, the models that adopted HPV DNA testing as primary screening had the lowest aggregate cost estimates, as well as costs per specimen.

As shown in Figure 4, the largest influence on marginal cost

reductions was the labour complement, which also fed into reduced costs from longer screening intervals. The HPV testing approaches were high throughput and cost-effective since they assumed that labour reductions of laboratory assistants, general duty, charge and senior technologists and cytopathologists were feasible. Indeed models six and eight were more realistic compared with models five and seven since the reduction of labour unfolded over a period of five years in the former. At an average cost of $104 per procedure, the rate of colposcopy referral and uptake was another variable for which reductions were expected after implementing the HPV testing platform. Both labour and colposcopy reductions were partly attributed to the lengthening of screening intervals for 73% of women who present for cervical screening (women aged 30 years or older).

Finally, labour costs aside, it was anticipated that the enhanced predictive value of HPV DNA testing would substan-tially reduce the incidence of cervical cancer and, subsequently, the marginal costs associated with hospital treatment. However, some cases would not be prevented due to under-screening or no screening at all in certain population groups. Figure 5 shows estimates of the costs per case averted, in terms of incidence of cervical cancer requiring hospitalization in Manitoba.

We conducted sensitivity analysis to address some of the

uncertainty in the point estimates of the newer technology models. Two variables, labour costs and colposcopy procedures, were at highest risk of deviating from the point estimate. If labour was not reduced according to the assumptions in models five and seven, the costs would be higher by $1.37 million in the first year and $7.0 million over five years, compared with conventional cytology in the baseline model. Similarly, the rate of colposcopy referral and uptake could vary by approximately 5,900 procedures in models two, three and four, resulting in a marginal cost of $613,600 per annum. For model two, this represented an increase of 7.5% in the first year budget and 7.1% over five years (model three: 6.3% and 6.9%; and model four: 7.1% and 6.6%, respectively). In models five and six, additional colposcopy procedures could increase total costs by $276,640 per annum, which represents a budget increase of 6.0% in model five and 5.0% in model six over five years.

The rate of repeat cytology due to a higher number of unsat-isfactory smears was another variable that may have resulted in costs that exceeded the point estimates. However, the five-year marginal cost impact for any applicable model totalled less than 0.5% of the point estimate reported in Figure 1. However, even with sensitivity analysis, the lowest cost models were still those that adopted HPV testing as the primary screening model (models five to eight).

Study LimitationsIn terms of the degree of external validity of the findings, it is difficult to generalize the applicability of this study’s results to different settings, patient groups, provinces or countries. The conclusions may or may not be limited to this study. However, at the risk of providing sweeping statements that endorse the widespread adoption of HPV testing, the current study’s findings are consistent with previous economic evaluations of cervical cancer screening platforms (Chuck 2010; Krahn et al. 2008; Kulasingam et al. 2009).

ConclusionThe traditional Pap test is outdated and costly. The case of cervical screening in Manitoba illustrates that there are efficien-cies to be gained by implementing the new high-throughput technology platforms, such as HPV testing. Moreover, HPV testing can be centralized in one facility and automated. The biggest cost savings resulted from reduced labour, reduced rate of colposcopy referrals and increased length of routine screening intervals. Hospital treatment costs for cervical cancer would be reduced by half by using HPV testing models (models five to eight) compared with conventional cytology. The adoption of HPV testing as a primary screening model for women aged 30 years or older represented the most cost-efficient strategy. LBC