Validity study of the severity index, a simple measure of urinary incontinence in women

BMJ 2001; 322 doi: (Published 05 May 2001) Cite this as: BMJ 2001;322:1096
  1. Janet Hanley (Janet.Hanley{at}, research managera,
  2. Ann Capewell, consultant physician in medicine of the elderlyb,
  3. Suzanne Hagen, research adviser/statisticianc
  1. a Lothian Primary Care NHS Trust, Edenhall Hospital, Musselburgh EH21 7TZ
  2. b Royal Victoria Hospital, Edinburgh EH4 2DN
  3. c Nursing Research Initiative for Scotland, Faculty of Health, Glasgow Caledonian University, Glasgow G4 0BA
  1. Correspondence to: J Hanley
  • Accepted 23 January 2001

Lack of a universally accepted, easily applied, outcome measure is one reason why urinary incontinence in women is poorly evaluated and treated.1 The severity index, developed by Sandvik et al, is short and simple enough for use in almost any context.2 We evaluated the reliability, validity, and sensitivity to change of the severity index in a wide range of women in Scotland.

Methods and results

The severity index comprises the following two questions. How often do you experience urine leakage (0=never, 1=less than once a month, 2=one or several times a month, 3=one or several times a week, 4=every day and/or night)? How much urine do you lose each time (1=drops or little, 2=more)? The total score is the score for the first question multiplied by the score for the second question (0=dry, 1-2=slight, 3-4=moderate, 6-8=severe). We added the category “dry” for women whose urinary incontinence was cured. A version of the index splitting the “severe” category into severe and very severe has recently been published.3

To investigate test-retest reliability, the revised index was administered to study participants twice, three days apart. To assess validity, 48 hour urinary diaries and measures of urine leakage over 48 hours (calculated by weighing of pads) were collected. To test sensitivity to change, participants being treated for their urinary incontinence completed the measures again 12 weeks after either surgery or first attendance at a non-surgical continence clinic. Age, physical disability (Barthel index),4 and cognitive disability (abbreviated mental test) were also recorded.5

Data were collected from women with stable incontinence not undergoing treatment, identified via community nurses; women undergoing initial assessment and non-surgical treatment at a continence clinic; and women awaiting surgical treatment (colposuspension) for stress incontinence. Those included were medically stable and either were cognitively intact (abbreviated mental test 8) or had a carer to help them.

Overall, 237 women were recruited (table). Community patients were significantly older (F2,234=113.6, P<0.001), more cognitively impaired (Kruskal-Wallis χ2=34.1, P<0.001), and more physically disabled (χ2=88.8, P<0.001) than women in the two treatment groups. They also experienced greater urine leakage (χ2=31.6, P<0.001) and more episodes of incontinence (χ2=26.8, P<0.001).

Baseline characteristics, measures of incontinence, and number (percentage) in each severity index category by patient group

View this table:

Most women (202; 88%) recorded the same severity index response category on days 1 and 4 (κ=0.78, P<0.001). Test-retest reliability for each question was also good (κ=0.69 for question 1 and 0.83 for question 2, P<0.001 for both).

Higher severity index categories were associated with increasing urine leakage (r=0.36, P<0.001). Median urine leakage over 48 hours was 32 g for those in the “slight” category, 29 g for “moderate,” and 143 g for “severe” (χ2=14.9, P<0.001; mean ranks 41.8, 50.2, and 80.7 respectively). There was a similar association between severity index category and number of episodes of incontinence (r=0.55, P<0.001). The median number of episodes of urinary incontinence over 48 hours was 0 for slight, 1 for moderate, and 6 for severe (χ2=67.3, P<0.001; mean ranks 52.4, 79.0, and 141.1). These significant associations suggest that the severity index is measuring what it is intended to measure, the severity of the physical condition.

Sixty per cent of women in the two treatment groups moved to a lower severity index category after intervention. The surgery group had a significantly greater improvement in severity index than the clinic group (Mann-Whitney U test 442.5, P<0.001). Change in severity index category was significantly related to both change in amount of urine leakage (χ2=8.4, P=0.015) and number of episodes of incontinence (χ2=24.1, P<0.001). The severity index thus detected changes in these measures of urinary incontinence associated with treatment.


The severity index is a short, simple, valid, reliable, and sensitive measure of urinary incontinence in women. It can therefore be recommended for routine use.


Contributors: JH contributed to the idea for and design of the study and conducted the research in Lothian and the initial drafting and revision of the paper. AC contributed to the idea for and design of the study and revision of the paper and negotiated access to clinical services in Lothian. SH was the research group leader, contributed to the idea for and design of the study and revision of the paper, conducted the research in Ayrshire, and analysed the data. Violet Millar negotiated access to clinical services in Ayrshire. Jean Donaldson collected data in Lothian. Louisa Sheward collected data in Ayrshire. JH, AC, and SH will act as guarantors for this paper.


  • Funding Research grant from the Scottish Office Department of Health Chief Scientist Office.

  • Competing interests None declared.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.