COST COMPARISON BETWEEN PAPER-AND-PENCIL AND WEB BASED DATA COLLECTION FOR LARGE-SCALE LONGITUDINAL HEALTH OUTCOMES RESEARCH

Purpose:  To compare costs and outcomes between paper-and-pencil (P&P) and web-based (WB) survey methods, accounting for differences in volume, method mix, and response rates for the CaPSURE™ longitudinal prostate cancer database. 

Methods:  The workflow process for processing the CaPSURE™ semi-annual questionnaires for both P&P and WB questionnaires were defined for January 1 – June 30, 2006.  A time trial was conducted to assess staff processing time. Fixed costs for optical character recognition (OCR) software ownership and web hosting costs were determined. Variable costs were compared for staff wages associated with questionnaire throughput for pre-processing (distribution); processing (receiving, triage, editing, follow up, OCR processing); and post-processing (double data entry (DDE), filing, storage).  Outcomes were cost/survey, cost/response and cost/error-free response. 

Results:  4,541 (90.7%) of eligible participants opted for P&P and 467 (9.3%) for WB surveys.  Double data entry was performed on 20% of paper-and-pencil questionnaires and 10% of web-based surveys.  The response rate was 76% and 88%, and DDE error rate 14% and 9%, respectively.  Total cost to process this 90:10 mix was $75,210 or $15.02/survey (Table 1) with web-based costs higher. 

Table 1.  Total Costs for Processing One Actual Phase of CaPSURE™ Surveys
90:10 Mix	Paper	Web-based
	N=4,541	N=467
Price/survey	$      14.66	$     18.46
Price/response	$      19.32	$     21.03
Price/error-free-response	$      22.55	$     23.06
Total	$66,587.38	$8,622.95

Table 2 shows costs at 50:50 mix is only $58,049 or $11.61/survey, with web-based now saving.  

Table 2.  Total Costs for Processing One Hypothetical Phase of CaPSURE™ Surveys
50:50 Mix	Paper	Web-based
	N=2,500	N=2,500
Price/survey	$      15.20	$     8.02
Price/response	$      20.04	$     9.13
Price/error-free-response	$      23.38	$    10.01
Total	$38,007	$20,042

Conclusion: The efficiency of web-based surveys depends on improved response rates and error reductions. Web-based becomes cheaper as volume increases and it becomes a higher proportion of the survey mix. The cost-efficient choice is not to use paper or web-based but what mix of both.