Data sources and linkage procedure

The WA Data Linkage System is an internationally recognized and validated data linkage system that facilitates person-level linkage of records in various state health-related population-based data sets. The Cervical Screening Registry (CSR) of WA is legislatively required to record all cervical test results (human papillomavirus [HPV], cytology, and histology) for WA residents. The CSR is an opt-off register with <0.05% of women requesting the removal of their demographic information and cervical test results. The CSR of WA has been operational since 1991 and utilizes many quality assurance processes to ensure data completeness. The Cancer Registry of WA has been operational since 1982 and records all cases of cervical cancer.

The WA Data Linkage System provided a deidentified extraction of linked data from the CSR of WA (2001 through 2012), Cancer Registry of WA (2001 through 2012), and WA Death Registry (2001 through 2012) for all women with biopsy-confirmed AIS. Each record in the linked data sets contained the encrypted patient identification and age. Data fields in the CSR of WA included cervical screening data, type of procedure (eg, cervical punch biopsy, LEEP, CKC biopsy, or hysterectomy), patient age at time of excisional treatment, margin status, and depth of the surgical specimen (measured macroscopic extent along the cervical canal). Data fields in the death records included: date of death.

Study data were obtained following approval from the Curtin University Human Research Ethics Committee (HREC) (project number: HR 86/2012), the WA Department of Health HREC (project number: 2012/49), and the University of Notre Dame Australia HREC (project number: 016031F).

Selection of patients

Patients aged ≥18 years who had histologically confirmed AIS recorded in the CSR of WA from 2001 through 2012 were identified. Patients were then excluded if:

• •
  

  AIS was identified on cervical screening or on punch biopsy prior to the excisional treatment.

  
  
• •
  

  They had a history of treatment for high-grade CIN or cervical carcinoma.

  
  
• •
  

  The AIS was initially treated with hysterectomy.

Cervical cytology results were classified according to the Australian Modified Bethesda System 2004. Patient age at the time of excisional treatment was classified as ≤30 or >30 years. Postcode of residence was used to assign a socioeconomic level using the Australian Bureau of Statistics 2006 Socioeconomic Indexes for Areas. Patients underwent excisional treatment procedures, CKC biopsy, LEEP, or hysterectomy on the recommendation of the treating specialist.

Histopathology findings

All histopathology reports were reviewed to confirm the type of excisional biopsy (CKC biopsy or LEEP), number of surgical specimens, depth of the specimen, presence of concurrent high-grade CIN, and resection margin status. The number of surgical specimens was classified as 1 or >1, the latter including both intentional 2-stage procedures (LEEP followed by “top-hat” endocervical sampling) as well as technically difficult procedures that resulted in multiple, fragmented, or incomplete specimens. Margin status was considered positive if any margin (ectocervical, endocervical, or deep/circumferential) was involved by CIN 2/3 and/or AIS; negative if all margins were histologically clear of in situ disease; and indeterminate if margins could not be assessed or were not documented. The extent of AIS was determined whenever possible from the initial histopathology reports in 1 of 2 ways. In some specimens, the maximal extent of AIS was specifically documented while in other cases the number of tissue blocks demonstrating AIS was recorded; in the latter situation, AIS extent was estimated assuming a standard block width of 2.5 mm multiplied by the total number of involved blocks. In the event that the lesion size and/or number of positive blocks was not reported, the specimens were recorded as: lesion size not reported. For those cases with reported AIS extent, the lesion size was categorized as: <1.5, ≥1.5-≤2.0, >2.0-≤2.5, >2.5-≤5, >5-≤8, or >8 mm.

Follow-up

Subsequent follow-up of patients potentially included cytological review, repeat CKC or LEEP biopsy, and/or hysterectomy. The follow-up period was defined as the date of the initial excisional procedure to the date of the last follow-up procedure (eg, cervical cytology, biopsy, or hysterectomy) or death.

Principal outcomes

The principal outcomes to be ascertained were: (1) persistence of CIN 2/3 or AIS, or a diagnosis of adenocarcinoma during the follow-up period (defined as <12 months after initial diagnosis); and (2) recurrence of CIN 2/3 or AIS, or a diagnosis of adenocarcinoma within the surveillance period (defined as ≥12 months after initial diagnosis).

Statistics

Kaplan-Meier graphs were constructed to investigate the survivorship function (time from initial excisional procedure until biopsy-confirmed high-grade cervical neoplasia persistence or recurrence), and log-rank tests were used to assess equality of the survivorship function. Proportional hazards models were constructed after simultaneously adjusting for multiple factors to investigate the relative rate (hazard ratio) of having, or subsequently developing, a high-grade cervical lesion after initial treatment. Models were constructed using purposeful selection of covariates including age at treatment, type of excisional procedure (CKC biopsy or LEEP), margin status, lesion size, and depth of excised tissue.

Statistical significance was determined as a P value <.05 and the 95% confidence intervals (CI) for hazard rate ratios were calculated. Violation of the proportional hazard assumptions was assessed and biologically plausible interaction terms between variables were tested. Software (Stata/IC 13.0; StataCorp LP, College Station, TX) was used for statistical analysis.

Data sources and linkage procedure

The WA Data Linkage System is an internationally recognized and validated data linkage system that facilitates person-level linkage of records in various state health-related population-based data sets. The Cervical Screening Registry (CSR) of WA is legislatively required to record all cervical test results (human papillomavirus [HPV], cytology, and histology) for WA residents. The CSR is an opt-off register with <0.05% of women requesting the removal of their demographic information and cervical test results. The CSR of WA has been operational since 1991 and utilizes many quality assurance processes to ensure data completeness. The Cancer Registry of WA has been operational since 1982 and records all cases of cervical cancer.

The WA Data Linkage System provided a deidentified extraction of linked data from the CSR of WA (2001 through 2012), Cancer Registry of WA (2001 through 2012), and WA Death Registry (2001 through 2012) for all women with biopsy-confirmed AIS. Each record in the linked data sets contained the encrypted patient identification and age. Data fields in the CSR of WA included cervical screening data, type of procedure (eg, cervical punch biopsy, LEEP, CKC biopsy, or hysterectomy), patient age at time of excisional treatment, margin status, and depth of the surgical specimen (measured macroscopic extent along the cervical canal). Data fields in the death records included: date of death.

Study data were obtained following approval from the Curtin University Human Research Ethics Committee (HREC) (project number: HR 86/2012), the WA Department of Health HREC (project number: 2012/49), and the University of Notre Dame Australia HREC (project number: 016031F).

Selection of patients

Patients aged ≥18 years who had histologically confirmed AIS recorded in the CSR of WA from 2001 through 2012 were identified. Patients were then excluded if:

• •
  

  AIS was identified on cervical screening or on punch biopsy prior to the excisional treatment.

  
  
• •
  

  They had a history of treatment for high-grade CIN or cervical carcinoma.

  
  
• •
  

  The AIS was initially treated with hysterectomy.

Cervical cytology results were classified according to the Australian Modified Bethesda System 2004. Patient age at the time of excisional treatment was classified as ≤30 or >30 years. Postcode of residence was used to assign a socioeconomic level using the Australian Bureau of Statistics 2006 Socioeconomic Indexes for Areas. Patients underwent excisional treatment procedures, CKC biopsy, LEEP, or hysterectomy on the recommendation of the treating specialist.

Histopathology findings

All histopathology reports were reviewed to confirm the type of excisional biopsy (CKC biopsy or LEEP), number of surgical specimens, depth of the specimen, presence of concurrent high-grade CIN, and resection margin status. The number of surgical specimens was classified as 1 or >1, the latter including both intentional 2-stage procedures (LEEP followed by “top-hat” endocervical sampling) as well as technically difficult procedures that resulted in multiple, fragmented, or incomplete specimens. Margin status was considered positive if any margin (ectocervical, endocervical, or deep/circumferential) was involved by CIN 2/3 and/or AIS; negative if all margins were histologically clear of in situ disease; and indeterminate if margins could not be assessed or were not documented. The extent of AIS was determined whenever possible from the initial histopathology reports in 1 of 2 ways. In some specimens, the maximal extent of AIS was specifically documented while in other cases the number of tissue blocks demonstrating AIS was recorded; in the latter situation, AIS extent was estimated assuming a standard block width of 2.5 mm multiplied by the total number of involved blocks. In the event that the lesion size and/or number of positive blocks was not reported, the specimens were recorded as: lesion size not reported. For those cases with reported AIS extent, the lesion size was categorized as: <1.5, ≥1.5-≤2.0, >2.0-≤2.5, >2.5-≤5, >5-≤8, or >8 mm.

Follow-up

Subsequent follow-up of patients potentially included cytological review, repeat CKC or LEEP biopsy, and/or hysterectomy. The follow-up period was defined as the date of the initial excisional procedure to the date of the last follow-up procedure (eg, cervical cytology, biopsy, or hysterectomy) or death.

Principal outcomes

The principal outcomes to be ascertained were: (1) persistence of CIN 2/3 or AIS, or a diagnosis of adenocarcinoma during the follow-up period (defined as <12 months after initial diagnosis); and (2) recurrence of CIN 2/3 or AIS, or a diagnosis of adenocarcinoma within the surveillance period (defined as ≥12 months after initial diagnosis).

Statistics

Kaplan-Meier graphs were constructed to investigate the survivorship function (time from initial excisional procedure until biopsy-confirmed high-grade cervical neoplasia persistence or recurrence), and log-rank tests were used to assess equality of the survivorship function. Proportional hazards models were constructed after simultaneously adjusting for multiple factors to investigate the relative rate (hazard ratio) of having, or subsequently developing, a high-grade cervical lesion after initial treatment. Models were constructed using purposeful selection of covariates including age at treatment, type of excisional procedure (CKC biopsy or LEEP), margin status, lesion size, and depth of excised tissue.

Statistical significance was determined as a P value <.05 and the 95% confidence intervals (CI) for hazard rate ratios were calculated. Violation of the proportional hazard assumptions was assessed and biologically plausible interaction terms between variables were tested. Software (Stata/IC 13.0; StataCorp LP, College Station, TX) was used for statistical analysis.